A2 Power transformers

and reactors

Power Transformer Audible

Sound Requirements
TECHNICAL BROCHURES
September 2024 - Reference 940
TECHNICAL BROCHURE

Power Transformer Audible

Sound Requirements
WG A2.54
Members

Christoph PLOETNER, Convenor DE Emanuel ALMEIDA, Secretary PT

Ali AL-ABADI DE Heribert BRUNE DE
Frank CORNELIUS DE Janine DICKINSON GB
Jan DONCUK CZ Gao FEI CN
Max GILLET FR Werner GOETTE DE
Shingo KANO JP Kim JUHYUN KR
Gabor NADOR HU Kyu-ho LEE KR
Bart SIMONS NL Miha PIRNAT SI
Frank TRAUTMANN DE Peter TARMAN SI
Kohei YAMAGUCHI JP Mark WARREN CA
Jiwoo YOO KR Jae-cheol YANG KR
Selim YUREKTEN TR Han YU CN

Corresponding Members

Mohinder PANNU AU Yuriy ODARENKO AU

Copyright © 2024
“All rights to this Technical Brochure are retained by CIGRE. It is strictly prohibited to reproduce or provide this publication in any
form or by any means to any third party. No part of this publication may be reproduced or utilized without permission from CIGRE”.

Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any responsibility, as to the
accuracy or exhaustiveness of the information. All implied warranties and conditions are excluded to the maximum extent permitted
by law”.

WG XX.XXpany network provided access is restricted to their own employees. No part of this publication may be
reproduced or utilized without permission from CIGRE”.

Disclaimer notice
“CIGRE gives no warranty or assurance about the contents of this publication, nor does it accept any
responsibility, as to the accuracy or exhaustiveness of the information. All implied warranties and conditions are
excluded to the maximum extent permitted by ISBNlaw”. : 978-2-85873-645-4
 TB 940 – Power Transformer Audible Sound Requirements

Executive summary
Christoph Ploetner (DE), Emanuel Almeida (PT)

This brochure was developed by CIGRE WG A2.54 between years 2016 and 2023 and covers all aspects of audible
sound relevant for new power transformer purchases to be considered in the process from specification until
installation. Strict distinction is made between the individual sound level components ‘no-load sound level’, ‘load
sound level’ and ‘cooling system sound level’ but the combination of the components as required in practice is also
explained in detail. Easy to use graphs together with a minimum of mathematical formulations enable the application
of the results for any engineer. While the main focus of the work was laid on three-phase liquid-immersed power
transformers with a power rating above 3 MVA, distribution transformers of different types (dry-type, gas-insulated,
and liquid-immersed with conventional and amorphous cores) were also considered and some information on sound
level differences between three-phase transformers, single-phase transformers and transformer banks is given as
well. Because information on the sound development of cooling system components and the resulting sound levels
is only rarely publicly available, this aspect is given a larger room in the brochure. Installation site aspects including
sound level legislation requirements, the substation design process concerning sound level guarantees and sound
mitigation possibilities are furthermore discussed. Although the brochure deals exclusively with total A-weighted
sound levels (such comprise the relevant audible frequency range in one level), useful information on frequency
selective sound levels if measured as A-weighted or as non-weighted levels is provided in an appendix. Frequency
selective sound levels (frequency spectra) are of importance for research purposes but also for special
investigations such as in case of acoustic and/or vibration queries. The brochure’s content may also be used as
reference material for acoustic queries on existing transformer installations and also in sales business for a first
approach on the sound levels to expect of new transformer installations.
The provided content of the brochure is entirely based on physical approaches but at the same time high effort was
laid on good readability and easy understanding. A large part of the brochure’s content is newly developed
knowledge and should raise the interest of the industry.
The brochure is also intended to serve as an educational document. Together with IEC 60076-10 / IEC 60076-10-
1 it will serve as an excellent and comprehensive introduction into the field of power transformer acoustics.

3
 TB 940 – Power Transformer Audible Sound Requirements

Table of contents

Executive summary ................................................................................................................ 3

Figures .................................................................................................................................... 7

1. Introduction................................................................................................................. 10
1.1 General ..................................................................................................................................................... 10
1.2 Scope ........................................................................................................................................................ 10
1.3 Range of application ............................................................................................................................... 11
1.4 Clarification of basic acoustic terms ..................................................................................................... 11
1.4.1 SOUND versus NOISE ................................................................................................................... 11
1.4.2 SOUND LEVEL ............................................................................................................................... 11
1.5 Remarks ................................................................................................................................................... 12
1.6 Acknowledgements ................................................................................................................................. 12

2. Background................................................................................................................. 13
2.1 Physics of sound ..................................................................................................................................... 13
2.1.1 Acoustics, sound and noise ............................................................................................................ 13
2.1.2 Sound generation ............................................................................................................................ 13
2.1.3 Sound propagation .......................................................................................................................... 13
2.1.4 Sound transmission......................................................................................................................... 14
2.1.5 Sound pressure, p ........................................................................................................................... 15
2.1.6 Particle velocity, 𝒖 ........................................................................................................................... 15
2.1.7 Sound intensity, 𝑰 ............................................................................................................................ 15
2.1.8 Sound power, W .............................................................................................................................. 15
2.1.9 Sound pressure and sound power levels ........................................................................................ 15
2.1.10 Sound level measurement .............................................................................................................. 16
2.2 Sound development in power transformers .......................................................................................... 16
2.2.1 No-load sound ................................................................................................................................. 16
2.2.2 Load sound ..................................................................................................................................... 18
2.2.3 Cooling system sound ..................................................................................................................... 20
2.3 Transformer sound level control ............................................................................................................ 40
2.3.1 Design and manufacturing towards the intended sound level ......................................................... 40
2.3.2 Transformer sound level versus installation sound/noise ................................................................ 41

3. Sound levels of liquid-immersed transformers ....................................................... 42

3.1 Concepts and boundaries for the derivation of typical sound level ranges ...................................... 42
3.2 Typical ranges of no-load sound power level ....................................................................................... 43
3.3 Typical ranges of load sound power level ............................................................................................. 47
3.4 Typical ranges of cooling system sound power level .......................................................................... 52
3.4.1 Introduction ..................................................................................................................................... 52
3.4.2 Description of case studies ............................................................................................................. 52
3.4.3 Case study 1: 400 MVA Network Transformer ................................................................................ 53
3.4.4 Case study 2: 63 MVA Network Transformer .................................................................................. 55
3.4.5 Cooling system sound power levels ................................................................................................ 55
3.5 Combination of sound levels .................................................................................................................. 57
3.6 Sound power levels of transformers for 50 Hz vs 60 Hz power frequency ......................................... 58
3.6.1 Findings from the collected database.............................................................................................. 58
3.6.2 Findings from sound level testing activities of CIGRE WG A2.54 membership ............................... 60
3.6.3 Interpretation of observed sound level differences as a function of power frequency ..................... 62
3.7 Sound power levels of single-phase units forming transformer banks .............................................. 65

4
 TB 940 – Power Transformer Audible Sound Requirements

4. Sound levels of other transformer types ................................................................. 68

4.1 Dry-type transformers ............................................................................................................................. 68
4.1.1 Introduction ..................................................................................................................................... 68
4.1.2 Technologies ................................................................................................................................... 68
4.1.3 Enclosures ...................................................................................................................................... 69
4.1.4 Applications ..................................................................................................................................... 69
4.1.5 Sound / sound levels ....................................................................................................................... 69
4.2 Gas-insulated transformers .................................................................................................................... 70
4.2.1 General information......................................................................................................................... 70
4.2.2 Application range of gas-insulated transformers ............................................................................. 71
4.2.3 Structural differences between liquid-immersed and gas-insulated transformers ........................... 71
4.2.4 Sound level characteristic of gas-insulated transformers ................................................................ 72
4.3 Transformers with cores made from amorphous steel ........................................................................ 73
4.3.1 General information......................................................................................................................... 73
4.3.2 Losses............................................................................................................................................. 73
4.3.3 Sound levels ................................................................................................................................... 74
4.3.4 No-load losses and sound power levels of liquid-immersed distribution transformers .................... 75
4.3.5 Summary ......................................................................................................................................... 75

5. Sound level specification and legislation ................................................................ 76

5.1 Introduction .............................................................................................................................................. 76
5.2 Survey / questionnaire ............................................................................................................................ 76
5.2.1 Reasons for specifying maximum transformer sound levels ........................................................... 76
5.2.2 Examples of noise limits in different countries across the world ..................................................... 77
5.2.3 Sound Level specification for new transformers .............................................................................. 78
5.2.4 On site / off site limits ...................................................................................................................... 78
5.2.5 Sound mitigation ............................................................................................................................. 78
5.2.6 Tenders ........................................................................................................................................... 78
5.2.7 Compliance with noise limits ........................................................................................................... 78
5.2.8 Summary of questionnaire findings ................................................................................................. 79
5.3 Noise and substation design .................................................................................................................. 79
5.3.1 Noise study ..................................................................................................................................... 79
5.3.2 Noise modelling .............................................................................................................................. 80
5.3.3 Complaints ...................................................................................................................................... 81
5.4 Impact of sound level specification on costs, losses, transportation ................................................ 82
5.4.1 Analysis........................................................................................................................................... 82
5.4.2 Results ............................................................................................................................................ 82
5.5 Parameters for specification purposes ................................................................................................. 83
5.5.1 Sound power ................................................................................................................................... 83
5.5.2 Sound level test parameters ........................................................................................................... 83
5.5.3 Combined sound power levels ........................................................................................................ 84
5.6 Achievable sound power levels for specification purposes ................................................................ 84
5.6.1 Example 1: Separate winding transformer ...................................................................................... 84
5.6.2 Example 2: Auto-transformer .......................................................................................................... 89
5.6.3 Example 3: Application of sound mitigation ..................................................................................... 90
5.7 Sound Levels and Transformer purchasing .......................................................................................... 91
5.7.1 Bulk purchase ................................................................................................................................. 91
5.7.2 Special purchase............................................................................................................................. 91
5.8 Tender process ........................................................................................................................................ 91
5.9 On-site noise measurements after transformer installation ................................................................ 92

6. Sound mitigation techniques .................................................................................... 93

6.1 Overview................................................................................................................................................... 93
6.2 In tank solutions ...................................................................................................................................... 93
6.3 Cooling system solutions ....................................................................................................................... 94
6.4 Design of sound barriers ........................................................................................................................ 95
6.4.1 Physical phenomena ....................................................................................................................... 95

5
 TB 940 – Power Transformer Audible Sound Requirements

6.4.2 Determination of total insertion loss ................................................................................................ 97

6.5 Factory-installed sound barrier solutions ............................................................................................. 99
6.6 Site-installed sound barrier solutions ................................................................................................. 103
6.7 Sound mitigation technologies not in regular use ............................................................................. 106

7. Summary and further work ...................................................................................... 108

APPENDIX A. Typical sound power level ranges of liquid-immersed power

transformers – Essential information at a glance for everyday use ............................. 110

APPENDIX B. References .................................................................................................. 114

APPENDIX C. Sound level specification using frequency bands .................................. 116

C.1. General ................................................................................................................................................... 116
C.2. Octave bands ......................................................................................................................................... 116
C.3. Frequency weighting filters .................................................................................................................. 118
C.4. Case studies: Measured sound levels ................................................................................................. 119
C.5. Conclusions ........................................................................................................................................... 123

APPENDIX D. Extension of the formulation for load sound power levels .................... 124

APPENDIX E. Cavitation .................................................................................................... 125

E.1. Cavitation – the phenomena ................................................................................................................. 125
E.2. Cavitation from a mechanical point of view ........................................................................................ 125
E.3. Cavitation from a chemical point of view ............................................................................................ 125
E.4. Cavitation from an acoustic point of view ........................................................................................... 125
E.5. Applications of cavitation ..................................................................................................................... 126
E.6. Cavitation and transformer cooling ..................................................................................................... 126

APPENDIX F. Survey on transformer sound level specification ................................... 127

6
 TB 940 – Power Transformer Audible Sound Requirements

Figures
Figure 1 – Spatial and temporal representation of a sound wave .......................................................................... 13
Figure 2 – Processes on sound wave during transmission .................................................................................... 14
Figure 3 – Scheme of no-load sound generation for a transformer with tank ........................................................ 17
Figure 4 – Curves showing relative change in lamination length during complete cycles of applied 50 Hz a.c.
induction up to peak flux densities Bmax (Source: IEC 60076-10-1:2016) ............................................................ 17
Figure 5 – Induction (smooth line) and relative change in lamination length (dotted line) as a function of time due
to applied 50 Hz a.c. induction at 1.8 T (Source: IEC 60076-10-1:2016)............................................................... 18
Figure 6 – Typical no-load sound level spectrum .................................................................................................. 18
Figure 7 – Cross-section of a transformer illustrating the generation process of load sound ................................ 19
Figure 8 – Typical Load sound spectrum (Source: IEC 60076-10-1:2016) ............................................................ 19
Figure 9 – Typical sound pressure level spectrum of transformer cooling fans ..................................................... 20
Figure 10 – Typical transformer oil pumps ............................................................................................................ 22
Figure 11 – Typical pump installation with air and water coolers ........................................................................... 22
Figure 12 – Typical pump installations with radiators ............................................................................................ 22
Figure 13 – In-Line Propeller Pump (left side) and Angle Pump (right side) .......................................................... 23
Figure 14 – Example of sound mitigation canopy over pump ................................................................................ 23
Figure 15 – Acoustic map using beamforming method .......................................................................................... 24
Figure 16 – Sound power level versus number of fans & diameter (example: vertical blowing EC fans) .............. 25
Figure 17 – Performance curves of an axial fan (ηsF – static efficiency, psF – static pressure, ............................ 26
Figure 18 – The impeller can be starved of air when the inlet to the fan is obstructed .......................................... 26
Figure 19 – Sound power level at different fan diameters with horizontal and vertical blowing ............................. 27
Figure 20 – Left: vertical blowing; Right: horizontal blowing .................................................................................. 27
Figure 21 – Left: vertical blowing; Center: horizontal blowing; Right: schematic diagram ..................................... 28
Figure 22 – Dependencies of Transformer Oil-Air Cooler Design.......................................................................... 29
Figure 23 – Sound power level as a function of air velocity [m/s] .......................................................................... 32
Figure 24 – Sound power level as a function of cooling surface [m2] .................................................................... 32
Figure 25 – Sound power level as a function of power consumption per cooler [kW] ............................................ 32
Figure 26 – AC / EC Fan service at different cooling capacity ............................................................................... 34
Figure 27 – Sound power level depending on cooler capacity (EC- vs. AC-Fans) ................................................ 35
Figure 28 – Power consumption depending on cooler capacity (EC- vs. AC-Fans) .............................................. 35
Figure 29 – Possible orientations of OFAF / ODAF coolers .................................................................................. 36
Figure 30 – Sound level measurement on coolers - concept ................................................................................. 36
Figure 31 – Sound level measurement on coolers – example measurement ........................................................ 37
Figure 32 – Possible directions of cooler air flow................................................................................................... 37
Figure 33 – Example OFWF (left side) and ONWF (right side) coolers ................................................................. 39
Figure 34 – Example OFWF (double tube safety technology, section view) coolers ............................................. 39
Figure 35 – Measured (red) and calculated (green) no-load sound power levels over core mass – calculation with
non-optimized (left) and optimized (right) parameter set ....................................................................................... 44
Figure 36 – Calculated range (envelope) of typical no-load sound power levels over core mass (left) and over unit
building power (right) with calculated data points .................................................................................................. 45
Figure 37 – Typical ranges of no-load sound power levels for the specification of 3-phase 50 Hz transformers ... 46
Figure 38 – Typical ranges of no-load sound power levels for the specification of 3-phase 60 Hz transformers ... 46
Figure 39 – Load sound power levels of the 50 Hz transformers and derived average functions - ‘Reiplinger’
formulation (left), improved model (right) ............................................................................................................... 49
Figure 40 – Typical ranges of load sound power levels for the specification of 3-phase 50 Hz transformers. ....... 51
Figure 41 – Typical ranges of load sound power levels for the specification of 3-phase 60 Hz transformers. ....... 51
Figure 42 – Example arrangements of cooling systems investigated within case 1 studies .................................. 53
Figure 43 – Example arrangements of cooling systems investigated within case 2 studies .................................. 53
Figure 44 – Typical ranges of cooling system sound power levels for the specification of 3-phase 50 Hz
transformers .......................................................................................................................................................... 56
Figure 45 – Typical ranges of cooling system sound power levels for the specification of 3-phase 60 Hz
transformers .......................................................................................................................................................... 56
Figure 46 – Transformer sound power level components and relevant combinations over the entire range of
transformer operation (300 MVA example transformer) ........................................................................................ 58
Figure 47 – Measured 3~ transformer sound power levels – left side no-load, right side load condition ............... 59
Figure 48 – No-load and load sound level difference between 50 Hz and 60 Hz power transformers ................... 59
Figure 49 – No-load sound level difference of 14 sample transformers tested at 50 Hz and at 60 Hz .................. 60
Figure 50 – No-load sound level measurements at multiple frequencies with constant (rated) induction. Average
sound level slope: 0.4 dB(A)/Hz ............................................................................................................................ 61
Figure 51 – Load sound level difference of 14 sample transformers when excited at 50 Hz and 60 Hz at constant
(rated) current ........................................................................................................................................................ 61
Figure 52 – Load sound level measurements at multiple frequencies with constant (rated) current...................... 62
Figure 53 – Tank vibration pattern measured in load (left) / no-load condition (right) at acoustically dominating
frequencies of 100 Hz / 300 Hz ............................................................................................................................. 64

7
 TB 940 – Power Transformer Audible Sound Requirements

Figure 54 – Measured surface velocity patterns of the short tank wall of a large power transformer left side at
50 Hz and right side at 60 Hz excitation frequency ................................................................................................ 65
Figure 55 – Measured 1-phase 60 Hz no-load sound power levels over assigned 1-phase unit building power –
displayed in enveloping curves for 3-phase 60 Hz transformers ........................................................................... 66
Figure 56 – Measured 1-phase 60 Hz load sound power levels over assigned 1-phase reactive power – displayed
in enveloping curves for 3-phase 60 Hz transformers ........................................................................................... 66
Figure 57 – Typical ranges of no-load sound power levels for 3-phase dry-type transformers .............................. 70
Figure 58 – Applicable range of gas-insulated transformers ................................................................................. 71
Figure 59 – Design concept of gas-insulated transformers ................................................................................... 72
Figure 60 – Comparison of no-load sound power levels of gas-insulated and liquid-immersed transformers ....... 73
Figure 61 – Hysteresis loops of core steel types ................................................................................................... 73
Figure 62 – Measured sound level vs induction – a case study............................................................................. 74
Figure 63 – Worlwide coverage of questionnaire responses ................................................................................. 76
Figure 64 – Process of installing a new transformer on a substation ..................................................................... 79
Figure 65 – Noise model of a substation showing the existing noise climate with two transformers in operation
[source - CEPS noise study] .................................................................................................................................. 80
Figure 66 – Noise model of existing substation with two transformers operating .................................................. 81
Figure 67 – Noise model of existing substation with two transformers operating, plus two additional proposed
transformers .......................................................................................................................................................... 81
Figure 68 - Noise model of existing substation with two transformers operating, plus two additional proposed
transformers and a 20 m long, 10 m high barrier ................................................................................................... 81
Figure 69 – Noise model of existing substation with two transformers operating, plus two additional proposed
transformers both in noise enclosures ................................................................................................................... 81
Figure 70 – Typical sound power level ranges for 3-phase 50 Hz transformers at no-load ................................... 85
Figure 71 – Typical sound power level ranges for 3-phase 50 Hz transformers at load ........................................ 86
Figure 72 – Typical sound power level ranges for 3-phase 50 Hz cooling systems............................................... 87

App Figure C.1 – Amplitude versus Frequency of an octave band ...................................................................... 117
App Figure C.2 – Typical transformer load sound level measurement (frequency spectrum) ............................. 119
App Figure C.3 – Typical transformer load sound level measurement in 1/1 Octave bands ............................... 120
App Figure C.4 – Typical transformer load sound level measurement in 1/3 octave bands ................................ 120
App Figure C.5 – Typical transformer no-load sound level measurement (frequency spectrum) ........................ 120
App Figure C.6 – Typical transformer no-load sound level measurement in 1/1 Octave bands .......................... 121
App Figure C.7 – Typical transformer no-load sound level measurement in 1/3 octave bands ........................... 121
App Figure C.8 – Typical transformer operational sound level (frequency spectrum), calculated ....................... 122
App Figure C.9 – Typical transformer operational sound level in 1/1 Octave bands, calculated ......................... 122
App Figure C.10 – Typical transformer operational sound level in 1/3 octave bands, calculated ........................ 123
App Figure F.1 – Covering email sent to utilities ................................................................................................. 127

Tables
Table 1 – Sound emission sources of the different cooling systems ..................................................................... 24
Table 2 – Five cooler designs with AC-fans for a given cooler specification with same materials ......................... 31
Table 3 – Comparison of AC / EC fan technology in terms of sound level and power consumption...................... 33
Table 4 – Comparison of cooler performance with AC- and with EC-fans for different cooling requirements ........ 35
Table 5 – Derived standard deviations and displacements from exact average curve .......................................... 49
Table 6 – Transformer cooling modes ................................................................................................................... 52
Table 7 – Case study 1 requirements for the cooling system ................................................................................ 53
Table 8 – Summary of case study 1 findings for ODAF cooling mode ................................................................... 54
Table 9 – Summary of case study 1 findings for ODAN cooling mode .................................................................. 54
Table 10 – Case study 2 requirements for the cooling system .............................................................................. 55
Table 11 – Summary of case study 2 findings for ONAF cooling mode ................................................................. 55
Table 12 – A-Weighting for typical transformer sound harmonics ......................................................................... 63
Table 13 – Common dry-type technologies ........................................................................................................... 68
Table 14 – Enclosures for dry-type transformers ................................................................................................... 69
Table 15 – Design features of liquid-immersed and gas-insulated transformers ................................................... 72
Table 16 – Parameters of selected core steel types .............................................................................................. 74
Table 17 – Loss and sound power level requirements for liquid-immersed transformers ...................................... 75
Table 18 – Examples of noise limits across the world ........................................................................................... 77
Table 19 – Noise reduction versus range of noise as per CIGRE A2.54 recommendation - 400MVA................... 82
Table 20 – Noise reduction versus range of noise as per CIGRE A2.54 recommendation - 63MVA..................... 82
Table 21 – Sound power levels for specification – Example 1 ............................................................................... 88
Table 22 – Sound power levels for specification – Example 2 ............................................................................... 90
Table 23 – In-tank sound mitigation solutions........................................................................................................ 94
Table 24 – Cooling system sound mitigation solutions .......................................................................................... 95
Table 25 – Sound panels system performance and associated sound level mitigation ......................................... 98
Table 26 – Sound enclosure system performance and associated sound level mitigation .................................... 99

8
 TB 940 – Power Transformer Audible Sound Requirements

Table 27 – Transformers-installed sound barrier solutions .................................................................................. 100

Table 28 – Mitigation techniques included in tank design .................................................................................... 102
Table 29 – Site-installed sound mitigation solutions ............................................................................................ 104
Table 30 – List of sound mitigation technologies not in regular use .................................................................... 107

App Table C.1 – 1/1 (full) and 1/3 (one-third) octave bands’ center, lower limit and upper limit frequencies ....... 118
App Table C.2 – A-Weighting for 1/1 octave bands ............................................................................................. 118
App Table D.1 – Derived standard deviations and displacements from average curves ..................................... 124
App Table F.1 – CIGRE WG A2.54 questionnaire –- power transformers audible sound requirements .............. 128

Equations
Equation 1 ............................................................................................................................................................. 14
Equation 2 ............................................................................................................................................................. 14
Equation 3 ............................................................................................................................................................. 14
Equation 4 ............................................................................................................................................................. 15
Equation 5 ............................................................................................................................................................. 15
Equation 6 ............................................................................................................................................................. 15
Equation 7 ............................................................................................................................................................. 15
Equation 8 ............................................................................................................................................................. 16
Equation 9 ............................................................................................................................................................. 20
Equation 10 ........................................................................................................................................................... 26
Equation 11 ........................................................................................................................................................... 43
Equation 12 ........................................................................................................................................................... 45
Equation 13 ........................................................................................................................................................... 48
Equation 14 ........................................................................................................................................................... 48
Equation 15 ........................................................................................................................................................... 48
Equation 16 ........................................................................................................................................................... 49
Equation 17 ........................................................................................................................................................... 49
Equation 18 ........................................................................................................................................................... 49
Equation 19 ........................................................................................................................................................... 57
Equation 20 ........................................................................................................................................................... 57
Equation 21 ........................................................................................................................................................... 62
Equation 22 ........................................................................................................................................................... 62
Equation 23 ........................................................................................................................................................... 63
Equation 24 ........................................................................................................................................................... 63
Equation 25 ........................................................................................................................................................... 64
Equation 26 ........................................................................................................................................................... 66
Equation 27 ........................................................................................................................................................... 67
Equation 28 ........................................................................................................................................................... 86
Equation 29 ........................................................................................................................................................... 86
Equation 30 ........................................................................................................................................................... 88
Equation 31 ........................................................................................................................................................... 96
Equation 32 ........................................................................................................................................................... 96
Equation 33 ........................................................................................................................................................... 96

App Equation C.1 ................................................................................................................................................ 116

App Equation C.2 ................................................................................................................................................ 117
App Equation C.3 ................................................................................................................................................ 117
App Equation C.4 ................................................................................................................................................ 117
App Equation C.5 ................................................................................................................................................ 117
App Equation C.6 ................................................................................................................................................ 117
App Equation C.7 ................................................................................................................................................ 121
App Equation C.8 ................................................................................................................................................ 121
App Equation D.1 ................................................................................................................................................ 124
App Equation D.2 ................................................................................................................................................ 124
App Equation D.3 ................................................................................................................................................ 124

9
 TB 940 – Power Transformer Audible Sound Requirements

1. Introduction
Christoph Ploetner (DE) and Emanuel Almeida (PT)

1.1 General
Environmental requirements have steadily increased across society in recent years. In terms of power transformers
and reactors, one environmental concern that stands out from others is the audible sound generated and emitted
into the surroundings, affecting residents living nearby as well as substation staff. Permissible sound pressure levels
for substation and residence boundaries are given by lawful authorities and are set so as to avoid negative effects
for people. Industry entities (utilities) intend to prevent their personnel from long term hearing issues. Today’s sound
level specifications reveal on average more stringent requirements than in the past. This is owing to the increased
awareness of the effects of sound on humans, the demand for high living quality, but is often also the result of newly
built residential districts encroaching upon existing substations, that influence lawful authorities’ general noise
legislation [B1]
As power transformers (and reactors) are usually the most significant sound sources in a substation, it is necessary
to consider their acoustic characteristics with care, particularly for new purchases. This however is often challenging
because validated information for transformer sound level specification is almost not available. Existing information
[B2][B3][B4] is widely outdated, simply because technology developed over the years. As a result, transformer
sound level specifications are found to be repeatedly technically unreasonable.
Looking to the ways utilities handle sound level aspects when purchasing transformers, it can be observed that their
practice for sound level specification is not at all harmonized but frequently influenced by vague factors instead of
physical facts. One of the observed soft ways to specify sound levels is to look into the history and specify sound
levels as identified for similar old units in possession, even if the transformer parameter of the new purchase is
different. Other utilities have one or a few standardized sound levels for specification purposes of certain transformer
classes in their files that are unchanged for a long time, even for decades. Others in turn simply use the limit for the
sound pressure level at the substation boundary (fence) as received from authorities and put that level into the
transformer specification, not considering its applicability to the new purchase at all. This frequently leads to
unnecessarily high, but recently often also to too small, specified sound levels in order to technically fulfill them
without external sound mitigation means – such as sound panels or sound enclosures. It is therefore an urgent
need to develop figures and guidelines for a technically realistic transformer sound power level specification that
can be used easily by utilities for the technical transformer procurement process but also serve for vendors as rough
information while quoting.

1.2 Scope
A study providing typical ranges of sound power levels for oil-immersed shunt reactors was completed by CIGRE
WG A2.48 in year 2015 [B5]. It is also a part of CIGRE Brochure 655 [B6]. As the results turned out to be practically
useful for shunt reactor specification but also for manufacturer quotation purposes, the idea to initiate a WG that
likewise looks into the more challenging area of power transformers arose in CIGRE SC A2. In September 2015,
SC A2 finally decided to close this gap of knowledge by forming a WG with the task to develop and provide
information and guidelines for the sound level specification of power transformers. CIGRE WG A2.54 was
subsequently formed and started its work in March 2016. It is the first CIGRE body dealing with a subject related to
transformer sound after more than two decades. The last two CIGRE publications date from 1992 and 1996
[B7][B8]. Ever since, a lot of work has been done in the field of transformer acoustics related to sound control by
design (manufacturers), sound level measurement (IEC [B9][B10] IEEE [B11]) and sound mitigation (utilities,
manufacturers); however, no activities on sound level specification have taken place.
WG A2.54 split up the given task into the following areas with all WG members being assigned to work on one of
the areas:
▪ Basic acoustics and sound development in transformers
▪ Derivation of typical sound power levels of transformers
▪ Sound level specification and legislation
▪ Sound mitigation
WG results presented in this technical brochure reflect this structure with the following chapters:
Chapter 1 – Introduction – Scope, Range of application, Clarification of basic acoustic terms, Remarks.

10
 TB 940 – Power Transformer Audible Sound Requirements

Chapter 2 – Background knowledge – Physics of sound, Sound level development in power transformers 1 ,
Transformer sound level control.
Chapter 3 – Sound levels of liquid-immersed power transformers – Concepts and boundaries for the derivation
of typical sound level ranges, Typical ranges of no-load sound power level, Typical ranges of load sound power
level, Typical ranges of cooling system sound power level, Combination of sound levels, Sound power levels of
transformers for 50 Hz vs 60 Hz power frequency, Sound power levels of single-phase units forming transformer
banks.
Chapter 4 – Sound levels of other transformer types – Dry-type transformers, Gas-insulated transformers,
Transformers with cores made from amorphous steel.
Chapter 5 – Sound level specification and legislation – Survey / questionnaire, Noise and substation design,
Impact of sound level specification on costs, losses, transportation, Parameters for specification purposes,
Achievable sound power levels for specification purposes (case studies), Sound Levels and Transformer
purchasing, Tender process, On-site noise measurements after transformer installation.
Chapter 6 – Sound mitigation techniques – Overview, In-tank solutions, Cooling system solutions, Design of
sound barriers, Factory-installed sound barrier solutions, Site-installed sound barrier solutions, Sound mitigation
technologies not in regular use.

1.3 Range of application

The findings and results provided in the brochure are applicable
▪ for three-phase transformers with rated three-phase power of 3 MVA and above,
▪ partly for single-phase transformers with rated single-phase power of 1 MVA and above,
▪ for transformers used in 50 Hz and 60 Hz network applications with basically sinusoidal terminal quantities
(voltage, current),
▪ for transformers with core-form and shell-form construction,
▪ for transformers without any external sound mitigation measure such as sound panels or sound enclosure.
The findings and results provided in the brochure are NOT applicable
▪ for transformers connected to power electronic devices that are subject to significant voltage and/or current
harmonics influencing the transformer sound level, such as industrial, SVC and HVDC transformers,
▪ for transformers with rated power frequencies other than 50 Hz or 60 Hz, such as for 16.7 Hz and 25 Hz
applications,
▪ for transformers operated under specific network situations, such as DC currents flowing in the neutral
and/or line terminals,
▪ for transformer service beyond rated specification such as over-excitation and overloading, unless a
specific service condition is properly reflected and follows the brochure’s intention / context,
▪ for ‘exotic’ transformer design solutions.

1.4 Clarification of basic acoustic terms

1.4.1 SOUND versus NOISE
Throughout this document, the term SOUND is used to describe the radiation of acoustic energy by a source, in
this case from transformers including accessories such as cooling systems. NOISE is used to describe an immission
of acoustic energy that is undesired by a recipient. Transformer humming is therefore classified as SOUND rather
than NOISE. The term NOISE is used in this brochure mainly in chapter 5 when the acoustic performance at
installation sites such as in and around substations is discussed. NOISE however may also be used habitually when
the term SOUND would actually be preferred and vice versa.

1.4.2 SOUND LEVEL

Although the term SOUND LEVEL is frequently used in the brochure it must be recognized that it is a generic term
that can stand for three different precise terms: sound pressure level, sound intensity level, sound power level. If
the term is used, the context does either not require to distinguish because the relevant clause is generic, or it is
fully clear which of the precise terms is meant.

1 A limited quantity of literature is available within the industry on transformer cooling system sound when
compared to transformer no-load sound and load sound. This is reason why section 2.2.3 ‘Cooling system sound’
is providing extensive information.

11
 TB 940 – Power Transformer Audible Sound Requirements

1.5 Remarks
▪ It is important to understand that WG A2.54 did not deal with sound level control by design – this is the
task and proprietary matter of manufacturers and not up for public exchange – but based the entire work
on the sound level performance of existing/installed units in a fully anonymous manner. The brochure
therefore does not give any insights on how to improve the sound level performance of transformers
beyond commonly known techniques. Instead, it provides guidance based on average observations of
existing installations.
▪ Throughout the brochure, sound levels will be given as A-weighted levels because this notation is globally
standardized and most used in practice.
▪ Throughout the brochure, sound levels comprise the entire frequency range of interest. Tonal/tonality
considerations are not made, although transformer no-load sound and load sound consist naturally of
individual tones.

1.6 Acknowledgements
Fulfilling the given task to a CIGRE WG requires involvement and contributions from all WG members. A number
of members made contributions beyond the normal expectation by performing specific transformer tests, data
analysis and preparing the brochure. To acknowledge such contributions, their names are mentioned right under
the title of the relevant chapter.

12
 TB 940 – Power Transformer Audible Sound Requirements

2. Background
Ali Al-Abadi (DE), Miha Pirnat (SI), Christoph Ploetner (DE), Peter Tarman (SI), Gabor Nador (HU), Kim Ju Hyun
(KR), Mark Warren (GB), Werner Goette (DE) and Janine Dickinson (GB)

2.1 Physics of sound

2.1.1 Acoustics, sound and noise
Acoustics is the science of sound that deals with sound generation, sound emission, sound propagation and sound
reduction. Sound is an audible oscillation that originates from the vibration of a mechanical structure and propagates
in waveform through an elastic or viscous transmission medium and is perceived by a receiver. Sound energy is
not transferred by particle transmission but by particle oscillation that is passed on from particle to particle. The
term noise is understood throughout this document as an undesired sound imission.

2.1.2 Sound generation

Sound originates when a structural object vibrates. These vibrations are caused due to internal or external forces
acting in or on the structural object.
In power transformers and reactors there are three kinds of forces acting on structural components that cause them
to vibrate:
▪ Maxwell forces (magnetic forces) acting on ferromagnetic materials where magnetic flux enters, such as
on magnetic shunts and tank walls but also between stacked sheets in core corners and in shunt reactor
gapped cores where flux overcomes gaps.
▪ Lorentz forces (electromagnetic forces) acting on diamagnetic and paramagnetic material components
carrying current in a magnetic field, for instance forces in windings and forces on copper or aluminum
shields.
▪ Magnetostriction forces acting in components made from electrical steel carrying magnetic flux, such as
core and shunts.
The vibrating elements disturb / excite the molecules of the surrounding medium and therewith create sound. The
generated sound then propagates through mediums in waveform from emitter to receiver.

2.1.3 Sound propagation

Sound propagates in the form of waves and requires an elastic medium for propagation, and therefore cannot travel
through vacuum. Sound waves are like mechanical waves and are distinguished from electromagnetic waves. The
sound wave is a mechanical oscillation in pressure, due to applied compression and elongation stresses on the
medium’s particles. When sound waves disturb particles in a medium, these particles then displace other
surrounding particles and pass the energy therein on. Particles in the sound wave move outward forming a wave
pattern. The sound wave is characterized by its frequency f, wavelength λ and amplitude A, and can be represented
in spatial or temporal domains as shown in Figure 1.

Figure 1 – Spatial and temporal representation of a sound wave

A disturbance in an acoustic medium will propagate through the medium as a wave with a finite velocity. The velocity
of the propagation of disturbance through the medium is called acoustic velocity or speed of sound c. The speed of
sound is related to the change in pressure dp and change in density dρ of the transfer medium and can be expressed
by Hook's Law as,

13
 TB 940 – Power Transformer Audible Sound Requirements

1⁄
𝑐 = (𝛽⁄𝜌) 2

Equation 1
where
𝑑𝑝
𝛽= 𝜌
𝑑𝜌
Equation 2
is the Bulk Modulus of Elasticity. Hence, sound travels faster through media with higher elasticity and/or lower
density.
In a perfect gas e.g. air, the heat transfer associated to the acoustic disturbance in a small region can be neglected
(adiabatic). Furthermore, it can be assumed isentropic, and so the ideal gas law can be used with the speed of
sound calculated as,
𝑛𝑝 1 1
𝑐 = ( ⁄𝜌) ⁄2 = (𝑛𝑅𝑇) ⁄2
Equation 3
with
n = specific heat ratio (adiabatic index) (for air = 1.4)
R = gas constant (for air = 287 J/kg K)
T = absolute temperature
The speed of sound depends on medium properties. The wavelength which is a spatial description of the
propagation is calculated as λ = c / f.
The sound wave carries and transfers the sound energy emitted by a vibrating source into the medium as it travels.
The sound energy through the medium loses its intensity as it moves away from the source. Sound energy is
associated directly to loudness of sound.

2.1.4 Sound transmission

The transmission of sound waves happens across and in between mediums. Medium properties are therefore
important for the transmission process. Transmission of sound waves starts from the source, propagates through
the mediums and reaches the receiver without any frequency loss.
During the transmission process, sound waves pass through the medium and strike structural or fluid obstacles (a
change in medium). As a result, the sound wave may be exposed to reflection, refraction, diffraction, absorption
and/or scattering within an obstacle, where part of the wave is transmitted through to become a sound wave of a
different pattern and energy, Figure 2.

Figure 2 – Processes on sound wave during transmission

Depending on the transmitting medium, the wave transmission is distinguished into:
▪ Airborne sound, where sound waves are transmitted from a source to a receiver through air (or any fluid
medium). Airborne sound waves transfer only longitudinally (compression waves) because particles of the
medium through which the sound is conveyed vibrate in the direction that the sound wave moves.
▪ Structure-borne sound, that is caused by vibrations of a structure which transmits through solid materials
/ structures. In solid bodies, various wave types can occur and depend on shape elasticity, inner bounding
forces, and resistance against deformation. Typical wave types are: compression, elongation, transverse

14
 TB 940 – Power Transformer Audible Sound Requirements

and bending waves. Their occurrence depends also on the structural geometry in relation to the sound
wave characteristics.

Following sections 2.1.5 - 2.2.2 are based on IEC 60076-10:2016 and IEC 60076-10-1:2016.

2.1.5 Sound pressure, p

The root-mean-square (r.m.s.) of instantaneous sound pressure over a given time interval at a specific location is
called the sound pressure. It is measured in pascal, Pa. Sound pressure is a scalar quantity, meaning that it is
characterized by magnitude only. The lowest sound pressure that a healthy human ear can detect is strongly
dependent on frequency; and at 1 kHz it has a magnitude of 20 μPa. The threshold of pain corresponds to a sound
pressure of more than a million times higher, 20 Pa.

⃗
2.1.6 Particle velocity, 𝒖
The root-mean-square (r.m.s.) of instantaneous particle velocity over a given time interval at a specific location is
called particle velocity. It is measured in meters per second, m/s. This quantity describes the oscillation velocity of
the particles of the medium in which the sound waves are propagating. It is characterized by magnitude and
direction and is therefore a vector quantity.

2.1.7 Sound intensity, 𝑰

The time-averaged product of the instantaneous sound pressure p and instantaneous particle velocity 𝑢
⃗ at a specific
location is called sound intensity:

1 ⬚
𝐼 = ⃗ (𝑡)) 𝑑𝑡
∫ (𝑝(𝑡) × 𝑢
𝑇 𝑇
Equation 4
It is measured in watts per square meter, W/m². Sound intensity describes the sound power flow per unit area and
is a vector quantity with magnitude and direction. The normal sound intensity is the sound power flow per unit area
measured in a direction normal, i.e. at 90º to the specified unit area. The direction of the sound power flow is
determined by the phase angle of the particle velocity at the specific location.

2.1.8 Sound power, W

Sound power is the rate of acoustic energy radiated from a sound source. It is stated in watts. A sound source
radiates power into the surrounding air resulting in a sound field. Sound power characterizes the emission of the
sound source. Sound pressure and particle velocity characterize the sound at a specific location; thus the sound
power W is related to acoustic intensity 𝐼 and enveloping surface 𝐴, i.e. it is the scalar product of 𝐴 and 𝐼 :

𝑊 =𝐴∙𝐼
Equation 5
The sound pressure which is heard or measured with a microphone is dependent on the distance from the source
and the properties of the acoustic environment. Therefore, the sound power of a source cannot be quantified by
simply measuring sound pressure or intensity alone. The determination of sound power requires an integration of
sound pressure or sound intensity over the entire enveloping surface. Sound power is more or less independent of
the environment and is therefore a unique descriptor of the sound source.

2.1.9 Sound pressure and sound power levels

To avoid the use of large numbers, the decibel scale (dB) is used in acoustics. The reference level p0 for sound
pressure for the logarithmic scale is 20 μPa and the 20 Pa threshold of pain corresponds to 120 dB. Sound pressure
level Lp is then defined as:
𝑝
𝐿𝑝 = 20Log10 ( ) dB
𝑝0
Equation 6
The reference level I0 for sound intensity for the logarithmic scale is 1 × 10−12 W/m² and the normal sound intensity
level 𝐿𝑖 defined as
|𝐼𝑛 |
𝐿𝑖 = 10Log10 ( ) dB
𝐼0
Equation 7
with 𝐼𝑛 being perpendicular directed to the measurement surface.

15
 TB 940 – Power Transformer Audible Sound Requirements

An additional and very useful aspect of the decibel scale is that it gives a better approximation to the human
perception of loudness than the linear pascal scale as the ear responds to sound logarithmically. In the field of
acoustics, it is generally accepted that
▪ 1 dB change in level is imperceptible;
▪ 3 dB change in level is perceptible;
▪ 10 dB change in level is perceived to be twice as loud.
Human hearing is frequency dependent. The sensitivity peaks at about 1 kHz and reduces at lower and higher
frequencies. An internationally standardized filter termed ‘A-weighting’ ensures that sound measurements reflect
the human perception of sound over the whole frequency range of hearing.
Analogously, the defined decibel scale for the sound power level LW is:
𝑊
𝐿𝑊 = 10𝐿𝑜𝑔10 ( )
𝑊0
Equation 8
with a commonly used reference power 𝑊0 of 1 × 10−12 W.

2.1.10 Sound level measurement

Acoustic parameters measured by a sound level measurement device are either sound pressure level or sound
intensity level, or both simultaneously. Both parameters describe the acoustic activity of a sound field and are
quantitatively fully aligned by the dB scale. This means, in case of an ideal acoustic field (a free field), where the
phase angle between the sound pressure wave and the sound velocity wave is zero, measured sound pressure
level and sound intensity level return identical dB values. If the acoustic field is however disturbed – for instance
due to reflections occurring in the test environment – then the phase angle between pressure and velocity is no
longer zero and this results in a sound intensity level measured somewhat smaller than the sound pressure level.
While the sound intensity level measurement accounts for the unintended disturbance, the sound pressure level
falsely contains sound contributions from the disturbance and is therefore too large. Without going into details, by
respecting the rules given in [B9] and [B11] for transformer sound level measurements, the returned sound intensity
level approaches the always unknown true sound level in almost all cases more accurately than the sound pressure
level and is therefore the preferred sound level parameter to be measured.
If the sound probe is kept stationary at a specific location in the sound field, the measured sound level applies to
that one location (point) in the sound field. In order to characterize a transformer acoustically, sound levels at many
points, distributed around the transformer according to defined rules, have to be measured and the levels of the
individual points averaged. The averaged sound level is then the one sound level characterizing the device. Note
that modern sound measurement devices automatically perform the averaging and for higher accuracy also allow
the probe to be moved continuously around the test object while averaging.
By adding the so-called surface measure (in dB), derived from the enveloping surface and thereby standing for the
dimension of the test object, to the measured sound level, i.e. to either the sound pressure level or to the sound
intensity level, the sound power level of the transformer is derived and is the final and preferred unit of measurement
for the acoustic characterization of transformers. In [B12], useful hints are given for the execution of reliable sound
level measurements.

2.2 Sound development in power transformers

2.2.1 No-load sound
When a transformer is energized, a magnetic flux density is established in the core. Due to the subsequent
magnetostriction forces in core steel but also due to Maxwell forces mainly occurring in core corners as a result of
flux crossing the stacked core sheets, the transformer core is vibrating and is thereby the no-load sound source.
Note that core vibrations are also impacted by the extended core structure with all its eigenfrequencies which may
or may not be excited by the core forces. From the core, sound waves are transmitted via structural elements and
via the surrounding fluid (liquid, air or gas) to the tank (if applicable) and further into the surrounding air, Figure 3.

16
 TB 940 – Power Transformer Audible Sound Requirements

Figure 3 – Scheme of no-load sound generation for a transformer with tank

Magnetostriction is the change in dimension observed in ferromagnetic materials when they are subjected to a
change in magnetic flux density (induction). In electrical core steel this dimensional change is in the range of 0,1
μm to 10 μm per meter length (μm/m) at typical induction levels. Figure 4 shows magnetostriction versus flux density
for one type of core lamination measured at five different flux densities. Each loop describes one 50 Hz cycle with
flux density Bmax.

3,0

2,0

1,0
Strain m/m

Bmax = 1,9 T
Bmax = 1,8 T
0,0 Bmax = 1,6 T
Bmax = 1,4 T
–1,0 Bmax = 1,2 T

–2,0

–3,0
–2,0 –1,0 0,0 1,0 2,0
Flux density T IEC 1411/05

Figure 4 – Curves showing relative change in lamination length during complete cycles of applied
50 Hz a.c. induction up to peak flux densities Bmax (Source: IEC 60076-10-1:2016)
The strain does not depend on the sign of the flux density, only on its magnitude and orientation relative to certain
crystallographic axes of the material. When excited by a sinusoidal flux, the fundamental frequency of the
dimensional change will therefore be twice the exciting frequency. The effect is highly non-linear, especially at
induction levels near saturation. This non-linearity will result in a significant harmonic content of the strain and this
causes the vibration spectrum of the core. Figure 5 shows the magnetostriction for a sinusoidal induction with Bmax
= 1,8 T and a frequency of 50 Hz. It has a periodicity of double the exciting frequency with peaks at 5 ms and 15
ms which are indistinguishable.

17
 TB 940 – Power Transformer Audible Sound Requirements

3,0 3,0

2,0 2,0

1,0 1,0

Strain m/m
Flux density T

0,0 0,0

–1,0 –1,0

–2,0 –2,0

–3,0 –3,0
0 10 20 30 40

Time ms

Figure 5 – Induction (smooth line) and relative change in lamination length (dotted line) as a function of
time due to applied 50 Hz a.c. induction at 1.8 T (Source: IEC 60076-10-1:2016)
The sound emitted by transformer cores depends on the velocity of the vibrations, i.e. the rate of change of the
magnetostriction (dotted line in Figure 5). This results in an amplification of the harmonics (distortion) in relation to
the fundamental which is at double the exciting frequency. Several even multiples of the exciting frequency will be
seen in the spectrum; in such cases the fundamental component at double the exciting frequency is seldom the
dominant frequency component in the A-weighted sound level spectrum as can be seen in Figure 6.

70
A-Weighted Sound Pressure Level

60

50
[dB(A)]

40

30

20

10

0
0 100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
Figure 6 – Typical no-load sound level spectrum
A d.c. bias in magnetization can significantly affect the sound level of a transformer due to significant increase in
vibration amplitudes. A transformer undergoing sound tests shall therefore be energized up until the point where
temporary effects of inrush currents and remanence have decayed and the sound levels have stabilized, before
testing is carried out.

2.2.2 Load sound

Load sound of a transformer is the sound which a transformer produces when load current is flowing in the
transformer windings. The usually dominating source elements of load sound are the windings. Other parts
contributing to load sound are stray flux control elements and structural parts, illustrated in Figure 7. At the right
side in Figure 7, the current, leakage field and resulting forces are indicated. At the left and bottom side, elements’
vibrations and transfer paths towards the tank are illustrated. At the transformer air side, tank vibrations emit the
audible sound into the surrounding air.

18
 TB 940 – Power Transformer Audible Sound Requirements

Tank
Air Liquid

Top yoke
Vibrations Force
Top clamping

Audible sound

Flux control element

HV Winding

LV Winding
Current

Core leg
Bottom clamping
Liquid path
Bottom Leakage field
vibrations
yoke
Structurally borne
path vibrations

Figure 7 – Cross-section of a transformer illustrating the generation process of load sound

Load currents in transformer windings generate a magnetic field at excitation frequency that passes through the
winding bloc, called stray or leakage field. As a consequence, the electromagnetic forces (Lorentz forces) act on
the winding cables in both axial and radial direction with local magnitudes being dictated by the stray field
components. Since the magnetic field is proportional to the load current, the resulting force is proportional to the
load current squared, at twice the excitation frequency. The magnitude of the vibrations depends on the elastic
properties of the conductor, the electrical insulation and the proximity of mechanical eigenfrequencies (natural
frequencies) to the vibration frequency of the complete winding set (e.g. windings, insulation, insulating medium,
geometrical parameters, clamping structure and pressure). The elastic moduli of insulation materials are almost
linear in a well designed and produced winding for displacement ranges occurring under normal operating currents.
Conductor metals have very linear elastic moduli. Harmonic vibrations are therefore normally minimal and the
fundamental vibration frequency at double the excitation frequency dominates the vibration and thereby the load
sound spectrum (see Figure 8). The magnitude of the vibrations is proportional to the winding forces.

70
A-Weighted Sound Pressure Level

60

50
[dB(A)]

40

30

20

10

0
0 100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]

Figure 8 – Typical Load sound spectrum (Source: IEC 60076-10-1:2016)

Further parts contributing to load sound are stray field control elements, intended to prevent overheating of metal
parts such as tank walls, tie bars and other structural parts. In principle three solutions are available:
▪ Magnetic shunts to guide the stray field which are made of laminated electrical steel sheets;
▪ Aluminum or copper shields repel the stray field by eddy current loops inside the shield;
▪ By using non-magnetic stainless steel for tank parts (for instance turrets, flanges, inserts),
▪ By sizing the tank such that stray flux control is not necessary.
Magnetic shunts are made from laminated electrical steel sheets, which guides the stray flux. Shunt vibrations are
caused by Maxwell forces when stray flux is entering the shunts as well as by magnetostriction forces when stray
flux is passing through the shunts.

19
 TB 940 – Power Transformer Audible Sound Requirements

Aluminum or copper shields repel the stray field away from the shield by eddy currents circulating in the shields.
Vibrations produced by these shields are caused by electromagnetic forces experienced by the shields.
Deflections and vibrational velocities of windings, shields and shunts (due to maxwell forces) are proportional to the
excitation force which is proportional to the square of the load current. The sound power radiated from a vibrating
body is proportional to the square of the vibration velocity. Consequently, the sound power generated by windings
varies with the fourth power of the load current.

2.2.3 Cooling system sound

2.2.3.1 Cooling devices
There are two active (sound producing) devices regularly used for transformer cooling purposes, namely air fans
and oil pumps.
2.2.3.1.1 Fans
In many cases, cooling fans are one of the major contributors to the overall transformer sound. The resulting cooling
fan sound originates from the aerodynamic fan sound itself, from air flow generated sound in the cooling system
(radiators, oil-air coolers) and from fan vibrations. Fan sound is of broadband type, mainly in the range of 300 Hz
to 3 kHz, see Figure 9 as example (fan diameter 450 mm, 8 impeller blades, rotational speed 460 rpm).

1/3 Octave Band Spectrum - axial fan

50
45
40
35
30
LpA (dB(A))

25
20
15
10
5
0
-5
-10
-15
-20
28

100

315
40
50
63
80

125
160
200
250

400
500
630
800
1000

3150

10000
1250
1600
2000
2500

4000
5000
6300
8000

12500
17783
f (Hz)

Figure 9 – Typical sound pressure level spectrum of transformer cooling fans

For the aerodynamic sound (and the heat transfer), the air velocity is of great importance. Any air velocity decrease
inside fans and cooling system results in a lower sound level, but results also in less effective cooling.
Fan sound is dependent on:
a) Aerodynamic design of the fan (shape of blades, housing and other parts that are inside the air stream).
The fan design is complex. Priorities are focused on the design of blades that create the pressure difference and
therefore the flow across the fan. For good efficiency and low sound, angle change and twist are given to the blade
at various positions outward from the hub to tip.
b) Rotational speed.
If the rotational speed is significantly reduced, often the fan is no longer the dominating source of sound. The sound
level of a fan can be estimated over a wide range of speed using the basic relation
𝑛2
𝐿𝑤2 ≅ 𝐿𝑤1 + 𝑘 log
𝑛1
Equation 9
with
𝐿𝑤1 sound power level at rotational speed 𝑛1 (in dB);
𝐿𝑤2 sound power level at rotational speed 𝑛2 (in dB);
k scaling constant between 50 and 55.
Making use of the EC (Electronically Commutated) motor technology for fans in new project designs or when
replacing old fans provides additional opportunities. EC fans are fans with a brushless DC motor and are

20
 TB 940 – Power Transformer Audible Sound Requirements

significantly more efficient than fans with AC motors. EC fans use electronic circuitry to control the armature current
for precise control of motor speed. They allow continuous airflow control between zero and design maximum and
facilitate the possibility to adapt the cooling capacity to the actual transformer loading.
Sound level performance and energy consumption of the cooling system can be improved significantly by controlling
the rotational speed of the fans.
Furthermore, different modes of operation are possible, e.g. the operation of the transformer at a constant
temperature independent of transformer loading and losses, operation regarding lowest noise emission or power
consumption, or even transformer precooling operation. A further positive effect of a controlled continuous fan
operation is the prolongation of lifetime by operating the fans continuously instead of in on-off mode.
NOTE: The consequence of rotational speed reduction is a lower cooling airflow.
c) Driving motor.
Usually, the motor represents a very small part of the total sound generated by the fan. Sometimes the motor sound
level can be increased if frequency regulation is applied.
d) Fan vibrations.
The fan impeller and the rotor of the driving motor are usually well balanced before fan assembly to minimize
vibrations. However, soft connections between fan and radiators / oil-air coolers are desirable. Also an independent
mounting of the fans from the radiators/coolers may be considered.
Comparing fan sound levels as guaranteed by various manufacturers is practically almost impossible for the
following reasons:
i. manufacturers have different facilities and methodologies for sound level measurements
ii. different definitions for the sound level are used
a. sound pressure level at 1 m, 1.5 m, 2 m or 3 m
b. sound power level
c. sound level measured on either suction side or pressure side or both sides (total sound level)
iii. measurements are made along different measurement paths
a. on the circumference of a circle in a plane through the axis of the fan
b. on the circumference of a circle in a plane perpendicular to the axis of the fan or
c. over a hemisphere on suction or pressure side of the fan
iv. various standards define different test methods
a. ISO 3745 (measurement over a sphere with free field condition) [B13]
b. ANSI/AMCA 300-08 (measurement based on reference sound source substitution) [B14]
c. etc.
As a consequence, users of fans, i.e. cooler and transformer manufacturers, can get comparable and reliable data
only by testing the fans of potential suppliers under the same precisely defined conditions. Alternatively, they should
use extra safety margins in their design until getting enough experience with fans of a new supplier.
2.2.3.1.2 Pumps
2.2.3.1.2.1 General
Transformer oil pumps are used where forced circulation of oil is required, in tandem with the use of radiators, water
coolers or air coolers.
Transformer oil pumps are electro-mechanical devices and therefore do generate sound. In most cases the pump
sound is not audible as it is significantly lower than the sound generated by any fans used on radiators or coolers.
Where fans are using variable speed control however, when the pumps are running at constant speed it may be
that the pump sound becomes the dominant source if the fans are running at a reduced speed.
There are recommendations for sound mitigation provided later in this brochure together with guidance on pump
installation to assist in avoidance of unusual sound associated with flow irregularities.

21
 TB 940 – Power Transformer Audible Sound Requirements

Figure 10 – Typical transformer oil pumps

Transformer oil pumps are typically installed in conjunction with radiators, OFAF/ODAF coolers or water coolers as
shown in the following examples.

Figure 11 – Typical pump installation with air and water coolers

Figure 12 – Typical pump installations with radiators

Pumps are manufactured in both an in-line and angle format and use both centrifugal and axial type impellers or
even axial type propellers (designed for minimized flow resistance in case of non-operation).

22
 TB 940 – Power Transformer Audible Sound Requirements

Figure 13 – In-Line Propeller Pump (left side) and Angle Pump (right side)
2.2.3.1.2.2 Sound level expectations
A guidance figure for the typical sound power level of a transformer pump is in the range 65 dB(A) to 75 dB(A).
When compared to a typical sound power level of a single fan, being in the range of 75 dB(A) to 95 dB(A), the pump
sound is in most cases insignificant.
Most often the sound emission of the pump is negligible, because of the difference of more than 10 dB(A) between
the sound power level of the pump and cooler. Considering the possibility to adopt variable speed control of fans,
for both cooler and radiator applications, it needs to be considered that pump sound may become the dominant
sound source for the cooling system at low loading conditions.
What are typical installation mistakes causing sound level issues of pumps?
▪ Wrong rotation direction by rotating field mismatch (e.g. wiring hook-up errors)
▪ High pressure drops along piping/cooler on the suction side of the pump resulting in turbulences or even
soft cavitation in the pump inlet region. Reasons for this effect may be valves which are not completely
opened or also a weak design of the cooling circuit. For more information on soft cavitation see
APPENDIX E.
Note: In case of trouble, as a first step compare electrical data of the pump with that of other pumps in the
system and also available hydraulic data.
What is the major contributor to “pump sound” – the pump or the installation?
In almost all cases, the audible sound associated with a transformer oil pump is attributable to shortcomings in the
installation or piping arrangements.
The importance of these points to minimize audible sound cannot be over stated.
The following are some of the key points to be born in mind:
▪ flow velocity
▪ reduction/extension of pipe diameter (abrupt change of cross-sections)
▪ abrupt deflections of the oil flow
▪ amount, place and shape of measuring devices, valves, deflections…
What sound mitigation measures can be recommended?
It is possible to add a canopy over the pump, such as the example illustrated in Figure 14, taking care to ensure
that adequate access to the pump is maintained for service.

Figure 14 – Example of sound mitigation canopy over pump

2.2.3.1.2.3 Installation guidance notes

i. Use adequate pipe velocities. Higher velocities such as 2,5 m/s or more increase the tendency for noise
emission.

23
 TB 940 – Power Transformer Audible Sound Requirements

ii. Provide a positive feed flow pre-pressure and not a negative pressTable 1 – Sound emission sources of
the different cooling systemsure upstream of the pump intake. If this is not maintained, turbulences or in
extreme cases even soft cavitation may take place.
iii. Minimize the number of restrictions on the suction piping, i.e. elbows, valves, flow gauges etc. Place those
at the pressure side of the pump if possible.
iv. Allow a minimum length of five pipe diameters of straight pipe between the pump suction and the
restrictions outlined above.
v. Never use piping on the suction side of the pump that is smaller than the nominal pump suction size.
vi. When pipe size reduction is required on horizontal runs, use an eccentric reducer with the eccentric side
down to avoid air pockets.
vii. Ensure there is no mis-alignment of any piping/flange connections during installation.
viii. In case of use of collectors, consider the design operation conditions and rated oil flow at each single
position when determining the dimensions of the pipe work.

2.2.3.2 Cooling systems

Available power transformer cooling systems (cooling methods) are well-known to the entire transformer
community. They are listed in Table 1 together with the sound producing equipment arranged in decreasing order
with the principal sound power emission being largest for the methods given in first line.
Table 1 – Sound emission sources of the different cooling systems
Equipment water oil
cooling pump pump fan
system sound sound sound
OFAF, ODAF - X X
ONAF - - X
OFWF, ODWF X X -
OFAN, ODAN - X -
ONWF X - -
ONAN - - -
Note: Water pumps may be located in the cooling water circuit remote from the transformer.
2.2.3.2.1 Cooling with radiators: ONAN, ONAF, OFAN, ODAN, OFAF, ODAF
2.2.3.2.1.1 General
Radiator cooling without rotating components like fans or pumps (ONAN) provides one of the most reliable cooling
solutions and offers the lowest possible sound emission.
The space demand for radiator cooling may become a limiting factor. The cooling efficiency can be increased and
the footprint of the radiator system reduced significantly by forcing an air flow between radiator plates by fans (ONAF
cooling) or/and circulating the oil in the transformer-radiator system by pumps (ODAN, OFAF or ODAF cooling).
While the pump sound is usually negligible, fans shall be considered as significant sources of sound.

Figure 15 – Acoustic map using beamforming method

NOTE: Acoustic beamforming is a method for sound source identification where an array of microphones is used
to capture the sound field. It is well-suited for large objects, where the array of microphones is placed relatively far

24
 TB 940 – Power Transformer Audible Sound Requirements

from the sources, outside the turbulent flow. Beamforming is valid for medium to high frequencies (500 Hz to
20 kHz). For source identification at lower frequencies (down to 100 Hz), ‘Optimized Nearfield Acoustical
Holography (SONAH)’ is a better method.
2.2.3.2.1.2 Design aspects
The transformer designer can play with a limited number of design parameters to fulfil the required cooling sound
level limit:
a) Fan diameter.
If a fan with larger diameter is used then it covers a larger cooling area and therefore runs at lower
rotational speed and lower air velocity for the same cooling effect. Therefore, fans with larger diameter can
contribute for achieving a lower sound level.
b) Number of fans.
Increasing the number of fans and running them at a lower speed reduces the cooling sound level.
As a case study, the required airflow for a 63 MVA ONAF cooled transformer with fixed radiator configuration was
kept constant, while the number and speed of vertical blowing fans (EC fans) were varied such to keep the top oil
temperature rise constant. The exercise was executed for fans with three different diameters with the total sound
power level being calculated for each individual combination. Results are shown in Figure 16.

Figure 16 – Sound power level versus number of fans & diameter (example: vertical blowing EC fans)
The different fan configurations can ensure the same cooling effect with a maximum of 28 dB(A) sound power level
difference!
c) Operating point compared to the point of maximum fan efficiency.
Any airflow other than the optimum flow (at maximum fan efficiency) increases the sound level. The fan operating
point depends on the pressure losses through the radiators. Air velocity, radiator geometry, number (at horizontal
blowing) or height (at vertical blowing) of radiators behind the fan, fan distance from the radiator are important
parameters that influence the pressure losses and therefore the airflow reduction of the fan. In many cases, the
radiator fans will operate in average at a higher airflow than the optimal one, due to lower pressure losses that exist
through typical radiators. Following Figure 17 shows an example for a fan running at the optimal operating point.

25
 TB 940 – Power Transformer Audible Sound Requirements

Figure 17 – Performance curves of an axial fan (ηsF – static efficiency, psF – static pressure,
LwA – sound power level, based on suction side measurement, qv - airflow).
d) Fan inlet and outlet obstructions of flow.
Any unnecessary inlet or outlet obstructions of flow should be avoided. For example, if the distance to firewall is too
small, the inlet flow is not uniform any more (Figure 10). Consequences could be higher sound level, reduction of
airflow or also recirculation of hot air (and consequently a reduction in cooling).

Figure 18 – The impeller can be starved of air when the inlet to the fan is obstructed
Obstructions such as radiators and guard grilles may not be avoided due to functionality and safety reasons.
Obstructions can also result in increased tone noise at blade pass frequency (BPF):
𝐵𝑃𝐹 = 𝑁𝑏 𝑛
Equation 10
where
BPF is blade pass frequency in Hz;
Nb is the number of fan blades;
n revolutions of impeller in s-1.

e) Position of fans (vertical versus horizontal).

Case study
The required airflow to maintain a constant top oil temperature rise for a 63 MVA ONAF cooled transformer with
fixed radiator configuration was calculated. According to the available space (which was different for horizontal and
vertical blowing), the possible number of fans were estimated for a few different fan diameters. By varying the speed
of the EC fans from same supplier, the required air flow was adjusted and the total sound power level calculated
for each arrangement (Figure 19).

26
 TB 940 – Power Transformer Audible Sound Requirements

Figure 19 – Sound power level at different fan diameters with horizontal and vertical blowing
The case study shows that horizontal blowing seems to be quieter. Some tests confirm the conclusion (Test 1) but
some do not (Test 2). Cooling efficiency and sound level depend very much on radiator geometry and fan
arrangement.
Test 1
(the height of radiators is equal to the horizontal length of the radiator arrangement)

Figure 20 – Left: vertical blowing; Right: horizontal blowing

Main parameters
Rated power: 64 MVA
Cooling method: ONAF
Supplied losses: 266.5 kW
Number of radiators: 2x4=8
Number of plates per radiator: 3 + 21
Header pipe distance: 3 x 1900mm, 21 x 2500mm
Distance between the
centerlines of the radiators: 600 mm
Number of fans: 2x2=4
Diameter of fans: 1000 mm
Fan speed: 500 rpm
Air volume per fan: 20780 m3/h | 5.77 m3/s
The whole length of the
4 radiators per side: 2320 mm
All available cooling area is covered with fans in both cases
Test results:

27
 TB 940 – Power Transformer Audible Sound Requirements

Average oil temp. rise (K) Top oil temp. rise (K)
Vertical blowing 32.4 47.6
Horizontal blowing 28.8 43.1
Horizontal blowing gives 4.5 K less top oil rise. So, the air velocity can be reduced resulting in lower noise
level at the same oil temperature rise.

Test 2
(the height of the radiators is larger than the horizontal length of the radiator arrangement)

Figure 21 – Left: vertical blowing; Center: horizontal blowing; Right: schematic diagram
Main parameters
Heat source: Immersion electric heater
Input heat: 90 kW
Number of radiators: 4
Number of plates per radiator: 34
Header pipe distance: 3000 mm
Number of fans: 3
Diameter of fans: 665 mm
Fan speed: 1150 rpm, 920 rpm
Air volume per fan: 13320 m3/h | 3.7 m3/s (1150 rpm), 10440 m3/h | 2.9 m3/s (920 rpm)
The whole length of the
4 radiators per side: 2320 mm
Test results:
Average oil temp. rise (K) Top oil temp. rise (K)

Vertical blowing (10440 m3/h) 23.4 30.9

Horizontal blowing (13320 m3/h) 23.9 31.4

The result shows that vertical blowing needs less air volume than horizontal blowing to give similar oil
temperature rise. This gives the opposite result to Test 1, however it depends on the fan and radiator types
used.
The following conclusion is possible. For a special case (Test 1) where the height of radiators is equal to the
horizontal length of the radiator arrangement and where the fans cover almost all cooling area (either from side or
from below) the horizontal blowing is more efficient than the vertical one.
At most of the transformer designs, the height of the radiators and the horizontal length of the radiator arrangement
are significantly different. So, there are many factors having an influence on cooling efficiency.
f) Turbulence in radiators.
Turbulence in radiators is contributing to the total sound power level, too. Avoiding any unnecessary obstacles or
sharp corners in the air flow path through the radiators is essential.

28
 TB 940 – Power Transformer Audible Sound Requirements

As a trend on the transformer market, low sound and low loss levels are usually required together. To dissipate the
reduced losses, radiator cooling without fans (ONAN or ODAN) becomes more often the preferred solution. This
way, fan sound is eliminated.
2.2.3.2.2 Cooling with oil-air coolers: OFAF, ODAF
2.2.3.2.2.1 General
The design of transformer oil-air coolers is more and more influenced by limitations for the sound emission as well
as the necessity to increase energy efficiency. Main parameters determining the sound level are the rotational
speed of the fans and corresponding air velocity, the cooling surface and the requested cooling capacity. The actual
cooler design (dimensions, volume) can be adapted to the given sound level constraints. That means, a given
cooling capacity can be achieved either with a smaller cooler and a higher fan speed or a larger cooler combined
with lower fan speed. The latter case would result in a lower sound level and in addition to a reduced power
consumption.
2.2.3.2.2.2 Design of oil-air coolers
The design of transformer oil-air coolers is depending on the specific project conditions. The requested cooling
capacity is equivalent to the expected transformer losses. The environmental conditions including ambient air
temperature determine the cooling fluid temperature. Environmental requirements are given in [B15]. These
conditions have an influence on the material selection, corrosion protection solutions to be used, structure of the
cooling surface (shape of fins, use of turbulators, etc.). Furthermore, other constraints like specified reserve
capacity, available space and requested mounting situation may be important.
The general task of an oil-air cooler design is to find an optimum between three main factors. The cooling capacity
can be determined as the product of the overall heat transfer coefficient, the heat exchanger surface and the
effective temperature difference between insulation fluid and air, in Figure 22 described as driving temperature
force.
The heat transfer coefficient is mainly determined by the velocities of the fluids. Based on the limits of pressure
drop, energy consumption and sound, the permissible range can be deduced.
The temperature difference is determined by the design air temperature and the permissible top oil rise. The design
top oil rise is mainly the result of calculations of the transformer designer considering the heat transfer processes
inside the transformer winding and the expected transformer life time.
Each of the constraints can be the leading design condition, together with constraints on sound and physical
dimensions.

Figure 22 – Dependencies of Transformer Oil-Air Cooler Design

A sound level limitation by the customer specification restricts the air velocities and the heat transfer coefficient, the
compensation can be achieved by an increased temperature difference of the fluids or an increased heat exchanger
surface.
Because the maximum oil temperature is typically also limited, thereby fixing the temperature difference, the cooler
design freedom is often limited to the heat transfer surface.

29
 TB 940 – Power Transformer Audible Sound Requirements

2.2.3.2.2.3 Sound level expectation

Transformer Oil-Air Coolers using fans with constant rotational speed (AC technology)
The following case study shows the wide range and flexibility of potential adaption to specific requirements. For the
evaluation of the sound level dependency parameters, different coolers were designed and their behavior
compared. The layout represents an example for a typical specification.
AC-fans have a constant rotational speed in operation. The following assumptions were applied:
▪ Capacity: 300 kW
▪ Temperature Rise: 45 K
▪ Max. Oil Temperature: 85 °C
▪ Oil Flow: 90 m³/h (25 l/s)
▪ Design Ambient Temperature: 40 °C
The study is based on the variation of following parameters:
▪ Cooler size
▪ Amount and size of fans (type, diameter)
▪ Fan rotational speed
These parameters determine the air flow quantity, the air outlet temperature, the air velocity and the sound emission.
The main parameters and results are collected in Table 2, see also [B16].

30
 TB 940 – Power Transformer Audible Sound Requirements

Table 2 – Five cooler designs with AC-fans for a given cooler specification with same materials
COOLER SPECIFICATION
Cooling capacity kW 300
Oil temperature rise ºK 45
3
Oil volume flow m /h 90
Oil inlet temperature ºC 85
Oil outlet temperature ºC 78
Voltage V 400
Frequency Hz 50
Cooler versions # 1 2 3 4 5
Cooler dimensions – Height mm 5750 5150 5150 3960 3060
Cooler dimensions - Width mm 1502 1610 1236 1236 1502
Cooler dimensions - Depth mm 910 930 830 830 910
Cooler weight kg 2700 2300 1800 1550 1350
Oil volume for one cooler l 285 235 180 170 140
Cooling surface m2 555 440 330 295 226
Air volume flow m3/s 9.6 11.1 12.6 12.9 17
Air velocity m/s 1.26 1.54 2.33 3.2 4.63
Air inlet temperature ºC 40
Air outlet temperature ºC 70 65.7 62.4 61.8 56.3
Fan type / fan diameter mm 910 1000 710 710 910
Rotational speed of fan rpm 365 390 840 1100 1265
Number of fans 4 3 4 3 2
Working power of one fan kW 0.2 0.35 0.6 1.6 4.5
Working power of one cooler kW 0.8 1.05 2.4 4.8 9.0
Sound power level for one cooler dB(A) 61 69 80 89 98

The following graphs illustrate the data contained in above Table 2. The sound power level depends on the air
velocity. The air velocity itself is a function of air flow quantity in relation to the bundle size. Thus, an increasing
cooling surface area provides a lower sound power level. Finally, the correlation between power consumption and
sound power level per cooler is shown.

31
 TB 940 – Power Transformer Audible Sound Requirements

Figure 23 – Sound power level as a function of air velocity [m/s]

Figure 24 – Sound power level as a function of cooling surface [m2]

Figure 25 – Sound power level as a function of power consumption per cooler [kW]

Transformer Oil-Air Coolers using fans with EC technology (variable rotational speed)
Using fans with EC-motors opens several new opportunities for the operation of transformer oil-air coolers. The
following physical relations have to be considered:
Doubling the speed of a motor increases its power input by a factor of 8 so that it is very inefficient to operate fans
at speeds greater than required to achieve the desired cooler performance. For example, in a partial load condition
where only 50% of the rated airflow is required, a comparison of AC fan versus EC fan operation shows benefits in
sound level reduction and auxiliary power reduction as illustrated in Table 3.

32
 TB 940 – Power Transformer Audible Sound Requirements

Table 3 – Comparison of AC / EC fan technology in terms of sound level and power consumption

2.2.3.2.2.4 Sound level design study --- oil-air coolers with AC fans versus EC fans
The focus of this study is the impact of variability of the rotational speed onto the sound power level.
Today the vast majority of all fans of transformer oil-air coolers have AC-motors. The scope and the type of
installations and control strategies usually constrain the control of single fans to on-off mode or an alternating
operation of complete coolers. In this solution, fans are operated only at full speed.
Using EC-fans, different control modes are possible, e.g. the continuous reduction of fan rotational speed to reduce
the sound level and the power consumption in case of transformer loading below nominal loading.
For the design study, it is assumed that the required cooling capacity is reduced to 75%, 50% and 25% of the rated
cooling capacity and is equivalent to 87%, 70% and 50% transformer loading respectively.
In case AC-fans are used, single fans are stepwise shut off fan by fan according to the reduced transformer loading.
EC-fans have the advantage to allow continuous cooling adaption to the transformer loading. The base for this
study is case no. 3 of the AC-cooler study presented before in Table 2 of this chapter with a cooling performance
of 300 kW, equipped with four fans and a resulting sound power level of 80 dB(A). The previously assigned
constraints remain unchanged for the AC-fans.
The comparison shows that at service with 75% cooling capacity the sound power level if using EC-Fans is
decreased by 8 dB(A) compared to 1 dB(A) if using AC-Fans. In case of a requested cooling capacity of 50%, a

33
 TB 940 – Power Transformer Audible Sound Requirements

reduction of 17 dB(A) against a reduction by 3 dB(A) is achieved. The advantage of EC-fans is 14 dB(A). This
advantage increases up to 22 dB(A) when the loading is only 25% of the original value. A cooler with one AC-Fan
in operation means a sound power level of 74 dB(A). For the same cooling capacity and if using EC-Fans, 52 dB(A)
are achieved.
Cooling Capacity: 100% Cooling Capacity: 75%

Cooling Capacity: 50% Cooling Capacity: 25%

Figure 26 – AC / EC Fan service at different cooling capacity

34
 TB 940 – Power Transformer Audible Sound Requirements

Table 4 – Comparison of cooler performance with AC- and with EC-fans for different cooling requirements
Cooling capacity 100% 75% 50% 25%
AC- EC- AC- EC- AC- EC-
AC-Fans EC-Fans
Fans Fans Fans Fans Fans Fans
number of running fans 4 4 3 4 2 4 1 4
cooling capacity [kW] 300 300 225 225 150 150 75 75
air inlet temperature [°C] 40,0 40,0 40,0 40,0 40,0 40,0 40,0 40,0
oil inlet temperature [°C] 85,0 85,0 85,0 85,0 85,0 85,0 85,0 85,0
oil outlet temperature [°C] 78,0 78,0 79,8 79,8 81,5 81,5 83,4 83,4
rotational speed [%] 100 100 100 65 100 40 100 20
rotational speed [1/min] 840 860 840 555 840 340 840 170
power consumption of all
2,4 2,2 1,8 0,8 1,2 0,3 0,6 0,1
running fans [kW]
sound power level for one
80 80 79 72 77 63 74 52
cooler [dB(A)]
+/- 0 dB(A) - 7 dB(A) - 14 dB(A) - 22 dB(A)

Taking into account that transformer loading does not follow steps but is a continuous process the advantage
becomes more evident. While AC-Fans can be shut off only stepwise when reaching the capacity limit, e.g. 75%,
EC-Fans can be adapted continuously.

Figure 27 – Sound power level depending on cooler capacity (EC- vs. AC-Fans)

3
Power consumption of all

2,5
working fans [kW]

2
1,5 AC-Fans

1 EC-Fans

0,5
0
0 25 50 75 100
Cooling capacity [%]

Figure 28 – Power consumption depending on cooler capacity (EC- vs. AC-Fans)

In the design study, 40°C as fixed ambient air temperature has been considered. Generally, the opportunity to
reduce the cooling capacity can also be taken if the air temperature is below the design temperature. For realistic

35
 TB 940 – Power Transformer Audible Sound Requirements

service conditions, most of the time the air temperature is below design temperature. Consequently, a lower cooling
capacity is required for almost the entire operation time and the EC-fans can be operated with an adapted rotational
speed. That means, a reduction of the rotational speed is nearly always possible.
2.2.3.2.2.5 Installation and arrangement conditions
Impact of cooler orientation
Depending on the installation requirements and preferences, OFAF/ODAF coolers may be installed in a number of
different orientations. The most common examples are displayed below.

Figure 29 – Possible orientations of OFAF / ODAF coolers

The sound power level of the cooler depends on the properties of the fans. The sound pressure level depends on
the kind of measurement prescription and method and furthermore on the place and distance of the fans related to
the measurement path.
Whilst the sound power level of the coolers is not dependent upon the chosen orientation, there is a significant
impact on the sound pressure level profile when measured at different aspects around the cooler.
At a distance of 2 meters from the cooler, the differences may be significant, although this difference will reduce
rapidly as distance from the sound source increases.
In the following example, eight points around a test unit are to be measured at each plane of measurement, with
one plane of measurement between each fan. The sketch shows the plan view of a vertically mounted cooler with
the eight positions noted, measured at each plane of measurement. Typical test distance from cooler to sound level
meter is 2 m

Figure 30 – Sound level measurement on coolers - concept

In this example the sound pressure level varied from a maximum of 75.3dB(A) at measurement point 3 (directly in
front of the fan) to a minimum of 67.4dB(A) to the side of the cooler.
Measurement details are displayed below.

36
 TB 940 – Power Transformer Audible Sound Requirements

Figure 31 – Sound level measurement on coolers – example measurement

Some consideration of the impact of cooler orientation upon how sound is measured is worthwhile.
Typically, at least a 5dB(A) difference in sound pressure level will be observed, when comparing the measurements
taken directly in the cooler airflow (in front of the fans) or measuring from the side, when the measurement distance
is set at 2 m.
The difference will decrease as measurement distance is increased, but the cooler sound level can be a significant
factor that impacts the operators that may be working close to the transformer.
Direction of air flow
Forced draft
The air is conducted from the fan to the heat exchanger bundle. The fans are operating with cool (ambient
temperature) air, subject to the environmental and installation conditions which must avoid the recirculation of
heated air.
In comparison to induced draft design the mass flow per fan is higher and therefore the efficiency of the fan is also
increased. A lower air flow velocity may be used with a correspondingly lower sound.
As a general recommendation, the air should be directed away from the transformer tank wall to the outside in order
to get a free flow of air, without additional pressure drop caused by the tank and associated accessories, and to
reduce the risk of recirculation. This does, however, mean that forced draft with fans on the tank side of the coolers
has the disadvantage of placement of the fan very close to the transformer with limited access to the fans for
maintenance, whilst forced draft with fans on the outside of the coolers has the potential disadvantages of restricted
airflow and recirculation.

Forced Draft Induced Draft

Air Air

OiI OiI

Figure 32 – Possible directions of cooler air flow

37
 TB 940 – Power Transformer Audible Sound Requirements

Induced draft
Air is conducted through the heat exchanger bundle to the fan. The fan is operating with hot air. This means less
mass flow compared to operating with cold air. The operation temperature of the fans is increased and should be
taken into account when selecting motor characteristics and bearings.
Some users prefer induced draft arrangement as the fans and motors are generally on the outside of the cooler,
making access for maintenance easier.
As the fans are on the outside of the cooler the airflow can be more easily directed (e.g. upwards over surrounding
firewalls)
2.2.3.2.2.6 Other arrangement considerations
Coolers placed close to the transformer, often along one long side
Most often the coolers are placed close to the transformer with the air direction arranged to lead the heat away from
the transformer to the outside. Consideration must be given to ensure adequate space between the back face of
the coolers and the transformer tank, to ensure adequate airflow. How the coolers are placed along different sides
will significantly impact the sound level measurement profile around the transformer.
Coolers on a separate place
This arrangement is used in special situations, e.g. in case of special sound limitations, the transformer may be
completely enclosed and the coolers may be mounted on a frame some distance from the transformer in order to
direct sound in a preferred direction.
Surroundings, e.g. fire walls
This arrangement is used in case of special safety requirements. It is important to plan a sufficient arrangement for
the air flow at the inlet side, e.g. between transformer and cooler and on the outlet side (between cooler and fire
wall) to warrant a removal of heated air from the transformer cell. The height of the fire wall should also be
considered so as not to restrict air flow. Preferably, the coolers are placed on the open side of the transformer with
no firewall obstructing the airflow, to ensure maximum cooling efficiency.
Wind direction
The prevailing wind direction shall be used to support the provision of cool environmental air. In case of a wind
direction in opposite to the air flow the operation of the fans may be influenced or even be disturbed. This can cause
problems with cooling capacity and fan lifetime.
Application guidelines for sufficient air flow
▪ Availability of cold air and removal of heated air:
o cross-sectional areas on the suction side of the fan shall be sufficient
o cross-sectional areas on the discharge side of the fan shall be sufficient
▪ Separation of suction and discharge side; if possible, use of chimney effect
▪ All possible service conditions shall be considered, e.g. the effects of solar radiation and wind on removal
of heated air
▪ Air flow rates to be considered. For an OFAF example cooler system, 500 kW at an air temperature
difference of 20 K with a reference air inlet temperature of 40°C means an air flow of around 22 m³/s. The
installation conditions shall enable the access of this volume of fresh air.
▪ Continuous and unobstructed air flow from outside, preferably perpendicular to the heat exchanger surface

2.2.3.2.3 Cooling with oil-water coolers: OFWF, ODWF and ONWF

2.2.3.2.3.1 General
Transformer oil can be effectively cooled by use of water where this is readily available. Water coolers can be used
in both forced oil arrangements (OFWF/ODWF) or sometimes also in a natural oil flow arrangement (ONWF).

38
 TB 940 – Power Transformer Audible Sound Requirements

Figure 33 – Example OFWF (left side) and ONWF (right side) coolers

Figure 34 – Example OFWF (double tube safety technology, section view) coolers
Water cooling is the preferred choice in a number of circumstances, including the following:
▪ Transformer is installed indoors (restricting availability of free-flowing ambient air), e.g. furnace
transformers
▪ Transformer is installed in chemical or petroleum plants, where water cooling can be used to reclaim heat
for secondary water processes.
▪ Transformer is installed in an urban environment with strict sound limitations
▪ Transformers installed underground
▪ An off-shore installation (e.g. an off-shore HVDC installation) where the corrosive environment is too
severe for placement of fans near to the transformer.
▪ The need to redirect sound away from the transformer and dissipate elsewhere (e.g. rooftop air-blast
cooler). Here the heated water is used in a closed-circuit with air cooling positioned outside of the building
away from sensitive areas.
Since water cooling is much more effective compared to air cooling, water coolers should be considered for
conditions with space limitations, high air temperature, a corrosive atmosphere and/or good availability of water.
2.2.3.2.3.2 Sound expectation
With water cooling, there is no air movement generated in the vicinity of the transformer and therefore minimal
sound is produced.
Heat can be transferred to another location either for recovery for local heating, use in a secondary process or
dissipation through air coolers positioned in less sensitive areas such as rooftop sites.
Water coolers themselves generate no discernible sound.
In case of OFWF/ODWF cooling, the associated oil pump produces some sound (see section 2.2.3.1.2).
The water pump used in the cooling circuit has associated sound but is usually located remote from the transformer.
2.2.3.2.3.3 Installation and arrangement considerations
▪ Availability and check of water flow shall be ensured, often a central water supply for many cooling devices
which are installed in parallel is installed. Interaction between the different users has to be considered.
▪ Benefit of monitoring of water flow amount and pressure drop should be checked.
▪ Potential pollution has to be considered especially in case non-circuit water is used.
▪ Around the water coolers sufficient place shall be considered for maintenance purposes.
▪ Provision of measures against the risk of freezing must be made.

39
 TB 940 – Power Transformer Audible Sound Requirements

2.3 Transformer sound level control

2.3.1 Design and manufacturing towards the intended sound level
As explained in section 1.5, the scope of this brochure is not to discuss design aspects for controlling the transformer
sound level performance. This is the task and proprietary matter of manufacturers and not up for public exchange.
After more than 130 years of transformer making, however, there exists quite some public knowledge and
understanding through publications and other forms of interactions in the industry, that can be discussed and
summarized publicly. As this is deemed to be of interest, it is provided in this section. The three measurable sound
components will separately be discussed but also a few general aspects introduced:
No-load sound level
As discussed before, the source of the no-load sound is the vibrating core. It is obvious that core dimensions / core
mass impact the no-load sound power level, simply because the larger a vibrating body, the larger the sound power
level. All manufacturers have design rules for the sound level impact of core dimensions / core mass in conjunction
with the core type. It is, however, important to recognize that the possibilities to vary core dimensions / core mass
with the target to address the no-load sound level are limited because the core size correlates at first order
consideration with the power rating of the transformer.
The second well-known parameter impacting the core vibrations and thereby the no-load sound level is the peak
induction (peak flux density) in the core. Its impact is in full control by manufacturers and is the number one control
parameter for the no-load sound level.
A further parameter impacting the no-load sound level is the proximity of natural eigenfrequencies of the core to
frequencies excited by core vibrations, i.e. at twice the power frequency and its first harmonics. If one or more
frequencies are close or even matching, this will increase the no-load sound level. As the calculation of core natural
frequencies is not fully certain, it is difficult to exclude such situations for sure by design. The insertion of damping
elements in the core is a measure to minimize vibration amplifications due to resonances and often replaces the
eigenfrequencies’ calculation.
The core stacking concept also impacts the no-load sound level as it influences the Maxwell forces in core corners
where flux is crossing between core sheets.
Although the sound level impact of different steel grades is well-known to manufacturers, sometimes this can also
be supplier dependent. There exists however no reliable possibility to verify the acoustic performance of core
laminations. In rare cases, a lower-than-expected acoustic performance of a delivered core steel batch can lead to
transformer no-load sound level issues.
Core clamping structure including applied pressure also has an influence on the no-load sound level.
To minimize the vibration transfer from core to tank, necessary connections between both elements are made
preferably in a vibration isolated manner. Placing the transformer core via anti-vibration pads on the tank bottom is
common practice.
Load sound level
Main sources of the load sound level are usually the winding vibrations, although vibrations of flux control elements
and structural parts may contribute or in certain situations even dominate the load sound level.
Winding dimensions / winding mass obviously increase with the power rating of the transformer and thereby impact
the load sound power level as per the same principle as explained for the core. Winding dimensions / winding mass
are however rarely used as design parameters but instead correlation is made to the rated transformer power – as
introduced by Reiplinger in 1988 [B4], or to the reactive power stored in the stray field (leakage field) – as introduced
later in this brochure.
Further correlations for load sound level control with respect to winding, flux control elements and structural design
are not publicly known, manufacturers have, however, in-house design rules and practices to control the load sound
design more precisely than the rough estimate of correlating the load sound level to unit rated power / leakage field
power.
It is generally known that the winding arrangement, the selected winding types including design features, as well
as the applied winding pressure impact the load sound level.
Another design criterion of importance for the load sound level is to avoid proximity of natural winding
eigenfrequencies with the excited vibration frequency of the windings, that is twice the power frequency for load
sound.
Cooling system sound level
The cooling system sound level can be controlled by design in a wide range as outlined in section 2.2.3 and is in
most cases not a limiting factor for the transformer sound level design. This is due to the availability of fans, including
EC fans, and pumps exhibiting very low sound power levels. In cases where the cooling system sound level would
be limiting, radiators with natural air flow can be selected for the cost of a larger footprint of the cooling system.

40
 TB 940 – Power Transformer Audible Sound Requirements

Tank design impact

The tank as a regular transformer component impacts the sound level always in the same principal manner. Design
differences between tanks are in most cases acoustically not very significant and are generally not considered in
the sound level design process. Rarely occurring tank resonance issues are usually resolved pragmatically by minor
tank modifications.
Manufacturing impact
High manufacturing quality is a mandatory condition for achieving the designed no-load and load sound level.
Defined manufacturing tolerances for core, winding and winding bloc making, including winding clamping / pressing,
must specifically be maintained.
Sound level measurement (FAT)
As sound level measurements are sensitive with mistakes being easily made, precautions should be taken. At first,
correct device settings must be assured. Secondly, measurement failures can be best excluded by strictly following
the measurement procedures as outlined in IEC and IEEE sound level measurement standards [B9][B10][B11]. By
doing repeatability tests, the measurement accuracy will be practically best proven and must not vary by more than
±1.5 dB.
Statistical evaluation
Sound level prediction (design) is subject to uncertainty. It is therefore important to statistically evaluate design
sound levels and measured sound levels regularly. This will enable a continuous improvement of the sound level
design accuracy.

2.3.2 Transformer sound level versus installation sound/noise

In a substation, all sound emitting installations (devices) contribute to the substation noise. As the installations in
substations depend on their electrical tasks, significant differences exist between substations of different type and
thereby in the radiated substation noise. Industrial and power plant substations usually have many different
installations, e.g. motors, manufacturing equipment, turbines, generators, coolers etc. with fully different sound
producing patterns (level and spectrum) than such of the installed transformers. It may therefore happen that the
sound emission from such equipment dominates the substation noise, with the transformers’ sound contribution
being less, or not relevant. On network substations, the sound emission of reactors, regulators and corona
(particularly in EHV substations) contributes to the overall noise beside the transformers’ contribution. A further
sound source to consider are switchgear devices when operating because such levels may be very high, although
occurring only occasional. Despite such situations, the noise emission of substations is in many cases, or maybe
even in most cases, dictated by the transformer installations.
The sound level generated by transformers when installed in a substation can be very different from the sound level
measured in the factory’s test bay. While the test arrangement and test conditions in the test bay are well controlled
and standardized, the transformer’s sound level on site is affected by the installation and operating conditions.
Specific supply and loading conditions, such as asymmetries in voltages and loads, variation in operating frequency,
voltage and current harmonics, DC components, etc. affect the sound level generated by the transformer if installed
at site. Additionally, sound reflections on surrounding objects like firewalls, bund walls and buildings, often increase
the sound level locally in the substation. This may also lead to an increased sound emission in certain directions
into the far field and at the same time reduce the sound radiation in other directions. The use of acoustically
absorbent coverings on critical surfaces can help to reduce sound reflections. Walls and buildings around the
transformer installation can also be used purposely as acoustic reflectors or acoustic barriers, depending on
▪ distance between wall/building and transformer,
▪ dimensions of wall/building,
▪ material of wall/building,
▪ relative position of wall/building to the transformer and
▪ position of the building/wall relative to the nearest noise sensitive premises.
For more details see [B10]. It is also noted here that a new Task Force with designation “CIGRE TF A2.01 - Power
transformers sound levels on site” started working in 2022 to explore this subject in more detail.
In conclusion, to clarify the noise/sound emission performance of a substation, it is mandatory to do this case-by
case, taking into consideration all sound emitting installations but also all civil installations within the substation plus
potentially also affected premises around the substation. Relying on transformer sound levels as tested in
manufacturers test bay is clearly not sufficient, it is, however, a key information necessary to perform the substation
noise analysis because transformer sound levels are often relatively high.

41
 TB 940 – Power Transformer Audible Sound Requirements

3. Sound levels of liquid-immersed transformers

Christoph Ploetner (DE), Bart Simons (NL), Frank Trautmann (DE) and Miha Pirnat (SI)

3.1 Concepts and boundaries for the derivation of typical sound level
ranges
The vast majority of transformers for energy transfer which produce perceivable and potentially annoying sound
levels for humans, are of 3-phase liquid immersed type with a power rating above 3 MVA and are used for all kinds
of energy transmission and distribution tasks. In terms of sound level control, this class of transformers is therefore
of specific importance for the industry and special focus was laid upon it while investigating the acoustic
performance, i.e. the level of emitted sound power.
The target of this chapter is to outline the development process and present the derived figures of typical sound
power level ranges for 3-phase liquid-immersed transformers as well as to inform on some side findings such as
the acoustic differences between 50 Hz and 60 Hz transformers and to give some hints on the acoustic differences
of 1-phase transformers and transformer banks when comparing with regular 3-phase transformers.
The main purpose of the developed figures for typical sound power level ranges is to guide and support transformer
users while specifying sound levels of new transformer purchases. Besides that, the figures are also useful for an
easy and quick evaluation of the acoustic performance of existing transformers. And they may also be used by
manufacturers as a first and rough indication of the possible sound power level range in the quoting and negotiating
process of transformers. This does, however, require understanding of the in-house sound level design capabilities
to ensure that quoted sound levels can finally also be met.
The development process of the presented figures is not explained in detail here; the interested reader can find
such information in publications issued before by CIGRE WG A2.54 in [B17]2 [B18][B19][B20][B21].
As outlined in chapter 2, transformer sound generation is linked to a number of different sources. Best to
determine/measure the acoustic performance of transformers, sound tests in three distinct conditions must be
executed, namely at no-load service, with load current flowing and while the auxiliary equipment for cooling (if any)
is in operation. The investigations done here were split accordingly and the derived results are separately displayed
in the form of graphical figures for the three distinct conditions.
‘Sound Power Level’ is the selected property used throughout the brochure when dealing with transformer sound
emission. It is the characteristic acoustic property to describe the sound emission of a device. A ‘device’ in this
chapter’s context is either the transformer tank, the cooling system or the combination of both. The sound power
level of a device at a specific operating condition has a certain magnitude and its accurate determination by sound
level measurements requires careful handling because the determination process is not trivial. As long as the sound
level measurements are done under well-defined FAT conditions and strictly follow latest requirements (methods,
procedures, corrections) of international transformer sound level measurement standards [B9] and [B11], the
determined sound power levels can be assumed to be sufficiently accurate.
The selected methodology for the development of typical sound power level ranges of transformers for no-load
sound (section 3.2) and load sound (section 3.3) is based on generic models describing the predominant physical
characteristics of the sound generation process but with unknown scaling factors. By applying regression methods
to a database containing a large number of datasets, each formed by certain design data and the FAT measured
sound power level of existing transformers, the scaling factors are determined. The now scaled model is finally used
to derive the typical sound power level ranges by re-calculating the sound levels of the entire population of the
database.
For the derivation of cooling system sound power levels, a number of cooling system case studies covering the
different cooling modes over the transformer power range of interest were performed and the typical sound power
level ranges derived by interpolation (see section 3.4).
Besides the generic sound development principles, the sound level of a transformer is also impacted by its design
concept and certain design details. To assure acceptance of the derived typical sound power level ranges while
utilizing statistical methods, it was mandatory to involve a representative number of transformer manufacturers
utilizing different transformer design concepts and contributing to the established database. Representatives from
13 countries and 19 different companies were active (standing in for 14 transformer and 2 cooler manufacturers as
well as 4 utilities). Due to the number of contributors representing many different design concepts, and also with
respect to the number of collected datasets (see below), the derived figures are considered universally
representative.
The number and type of parameters collected per transformer were kept at a minimum and are all of basic nature.
To each dataset of specification/design data, the determined no-load / load sound power level from FAT was added,

2Figures 9 and 10 presented in [B17] were preliminary and deviate slightly from final figures provided in this
brochure.

42
 TB 940 – Power Transformer Audible Sound Requirements

including information on the measurement method (pressure or intensity), and in case of an intensity measurement
the P-I-index. With this information, the adjustment of the provided sound power levels to the same basis according
[B9] and [B11] was possible so as to have comparable values in the database. Altogether, 1049 datasets for the
no-load sound power level analysis and 1275 for the load sound power level analysis were available and carefully
audited one-by-one. Any datasets judged to be questionable were marked and not used further; this however
applied only to a few, without statistical impact.
When using the derived figures presented in the following sections 3.2 to 3.4 for the estimation of typical sound
power level ranges, it is important to note the following:
1. The required input data are all design parameter specified by the transformer user/purchaser. They are
also given as standard data on the transformer nameplate.
2. The derived output quantity ‘typical sound power level range’ refers to sound power levels determined
(measured) according to latest versions of international sound level measurement standards [B9] and
[B11].
The following conditions and boundaries apply to the derived figures:
▪ prerequisites outlined in section 1.3,
▪ valid for all liquid-immersed transformers larger than 3 MVA with cores made from grain-oriented steel, no
distinction is made between different liquids,
▪ no running fans / pumps during no-load sound level and load sound level test,
▪ no external sound mitigation measures applied (panels or enclosures).

3.2 Typical ranges of no-load sound power level

The derivation of the typical no-load sound power level ranges of transformers is based on a model describing the
predominant physical relations of the no-load sound generation process. Again, the model does not consider design
details specific to individual design concepts but primarily describes the generic physical phenomena determining
the no-load sound generation. Seven specification/design parameters were selected for each transformer, forming
an individual dataset in the database together with the no-load sound power level determined during FAT. The
design parameters used as model input are:
▪ Core type discrete parameter comprising 3-limb and 5-limb/shell type class
▪ Core mass continuous parameter
▪ Core steel grade discrete parameter comprising a few classes: HiB, CGO/RGO
▪ Core induction continuous parameter
▪ Power frequency discrete parameter comprising 50 Hz and 60 Hz class
▪ Transformer rated power continuous parameter
▪ Transformer auto factor continuous parameter
while the model output is the
▪ Transformer no-load sound power level.
Sound level contributions by continuous parameters are modeled as continuous functions for the entire parameter
range. Discrete parameters are either handled by classification, i.e. several parameter sets are derived, each
covering an individual class or by discrete sound level contributors.
The development of the model is described in detail in [B17]; in the following a summary is given.
In a first step, the model describes the transformer no-load sound power level as the sum of sound power levels
assigned to a number of individual contributors:
𝐿𝑤 = 𝐿𝑚 + 𝐿𝑏 + 𝐿𝑔 + 𝐿𝑜
Equation 11
Lw transformer no-load sound power level in dB(A)
Lm no-load sound power level due to core mass in dB(A)
Lb no-load sound power level due to core induction in dB(A)
Lg no-load sound power level adder due to steel grade in dB(A); Lg = 0 set for grade HiB
Lo no-load sound power level due to other factors in dB(A) – it can be positive or negative
Equation 11 covers three contributors out of the mentioned parameter input list above: ‘core mass’, ‘core induction’,
‘core steel grade’. Not considered in Equation 11 are listed input parameters ‘core type’ and ‘power frequency’. This
is because it was decided to define two classes for transformer ‘core types’ and two classes for ‘power frequency’.
Such four classes were all treated separately in order to maintain good accuracy:
▪ 50 Hz – 3-limb cores
▪ 50 Hz – 5-limb cores + shell type
▪ 60 Hz – 3-limb cores

43
 TB 940 – Power Transformer Audible Sound Requirements

▪ 60 Hz – 5-limb cores + shell type

Also not included in Equation 11 are contributors due to ‘rated power’ and ‘auto factor’. Both input parameters will
be introduced in a later step of the model setup (see below).
Introduced in Equation 11 and not mentioned in the input parameter list is Lo – the sound level contribution due to
other factors: All sound level affecting parameters other than the before discussed parameters are excluded from
analytical calculations – simply because it is not possible, or it cannot be handled beyond one specific design
concept. The most important factors contributing to Lo are:
▪ specific design aspects/impacts;
▪ material and manufacturing tolerances;
▪ temperature impacts;
▪ uncertainties related to the sound level measurement.
The no-load sound power level model handles the impact of all such factors with one equivalent parameter, defined
as Lo.
So far, the no-load sound power level is physically described by the model but is not yet scaled. This is done in a
next step, where the model parameters are numerically scaled by statistical regression methods using the before
collected database. Regression target is to minimize the average deviation between calculated and measured
sound power level. In Figure 35 (taken from [B17]), measured and calculated sound power levels for the database
entries are shown as a function of core mass exemplarily for class ‘50 Hz – 3-limb cores’. The sound power level
calculation (green points) on the left side was done with a non-optimized parameter set and on the right side with
the derived optimized set. It can be seen that the application of the optimized parameter set results in a zero
deviation between the average curves, i.e. Lo being zero in average over all datapoints. This further means that the
developed model returns in average the same sound power level as measured. The standard deviation of Lo (the
difference between measured and calculated sound power levels) for this class of transformers was found to be 4.0
dB(A) – a fairly large number, if considering no-load sound power level being a guaranteed value.

Figure 35 – Measured (red) and calculated (green) no-load sound power levels over core mass –
calculation with non-optimized (left) and optimized (right) parameter set
A few further comments are made here:
1. The model parameter included in the calculation besides core mass (abscissa parameter) is core induction
only and is handled independently from core mass. In case of a utilized core steel grade other than HiB, a
manufacturer specific correction has been applied for relevant database entries and must be applied while
using the model.
2. The sound power levels calculated with the optimized parameter set represent the sound level expectation
values for the transformers in the database. The sound power level spread around the average green
curve relates to the varying core induction of the individual transformers in the database and was found to
be between 1.3 and 1.8 T (= typical core induction design range).
3. The differences between calculated (green) and measured (red) sound level points in right-side graph
stand for the sound level contribution due to other factors, represented in the model by Lo. The significant
standard deviation mentioned before is clearly visible.
4. Model extension by adding further sound level contributors to Equation 11 and re-scaling all model
parameters may result in an increased model accuracy. Smaller absolute values for Lo will be the
consequence, or in other words, the standard deviation of Lo will be reduced. It is however obvious that
such a refinement can only be done on manufacturer level because information on specific design details
etc. would be necessary. It is known that good processes and tools in use by individual transformer
manufacturers usually enable a lower standard deviation for the no-load sound power level than was found
here.
The developed model as outlined so far is useful for pin-pointing individual transformer no-load sound power levels,
it requires however input quantities which are not available at the time of specification and are normally not known
to the user/purchaser. As the development of rules for sound power level specification purposes was set as a target
to WG A2.54, so far derived results are not final.

44
 TB 940 – Power Transformer Audible Sound Requirements

By displaying (relating) the calculated no-load sound power levels of the database not over the core mass as done
so far, but instead over a correlating transformer nameplate parameter that is known at the time of transformer
specification, a figure for the typical no-load sound power level range depending on this nameplate parameter may
be derived.
An obvious parameter to consider in this respect is transformer rated power because it is correlated to the core
mass. A refined analysis suggests however, not to consider the transformer rated power but rather its building
power. The unit building power of a transformer is equal to the rated power as long as there is no wired electric
power flow from one voltage system to the other. This, however, is not always the case: in autotransformers, part
of the transformer’s total power flow from one voltage system to the other occurs via a wired connection. This results
in a smaller transformer active part with a lower rated power than transformer nameplate power rating and is called
unit building power. Although the relation between core mass and unit building power is not absolutely direct, it can
be considered direct within a range of scatter that is acceptable for statistical considerations included herein. In
Figure 36, the calculated no-load sound power levels of the database entries of the same transformer class as used
before (50 Hz -- 3-limb cores) are presented over core mass (left) and unit building power (right) – together with
derived average and enveloping functions. Unit building power is calculated as the product of collected database
parameters ‘Transformer rated power’ and ‘Transformer auto factor’. While ‘Transformer rated power’ is a
specification parameter, ‘Transformer auto factor’ is calculated for autotransformers from system voltages UHV and
ULV – per definition here at rated tap position – as follows:
(𝑈𝐻𝑉 − 𝑈𝐿𝑉 )
𝑈𝐻𝑉
Equation 12

Figure 36 – Calculated range (envelope) of typical no-load sound power levels over core mass (left)
and over unit building power (right) with calculated data points
Following remarks are made with respect to Figure 36:
1. The number of datapoints and the sound power levels (ordinate values) of datapoints shown in both graphs
are identical. Datapoints are, however, organized differently along the abscissa axis (x-axis) in the two
graphs to match the relevant abscissa parameters ‘core mass’ resp. ‘unit building power’.
2. In both representations, almost all datapoints are covered (enclosed) by the enveloping functions. The
spread of about 20 dB – representing the varying core induction of transformer designs – is not affected
by introduced abscissa parameter ‘unit building power’. This also applies to the average and enveloping
functions. Using unit building power instead of core mass as the major input parameter for the specification
of typical ranges of no-load sound power levels is thereby justified.
3. The enveloping functions shown in Figure 36 define the typical ranges of no-load sound power levels for
the analysed transformer class. Therein, the lower and upper boundaries (the blue dashed lines) represent
transformers with a core induction of about 1.3 T resp. 1.8 T, while the average curve (the red dashed line)
stands for units with about 1.6 T core induction.
The graph shown on the right side of Figure 36 represents the target result for the selected transformer class “50
Hz -- 3-limb cores”: the typical range of no-load sound power levels as function of nameplate derivable parameter
‘unit building power’.
Next, the previously described procedure was repeated for the three remaining transformer classes to likewise
derive typical ranges of no-load sound power levels as a function of unit building power. The found standard
deviations of Lo are in the same range as for class ’50 Hz – 3-limb cores’ and are all given here at a glance:
▪ 50 Hz – 3-limb cores 4.0 dB(A)
▪ 50 Hz – 5-limb + shell type cores 3.5 dB(A)
▪ 60 Hz – 3-limb cores 4.1 dB(A)
▪ 60 Hz – 5-limb + shell type cores 3.9 dB(A)

45
 TB 940 – Power Transformer Audible Sound Requirements

Then, as a last step, an attempt was made to combine the results of the individual transformer classes and form,
finally, just two classes. Without further explanation of the process here, this attempt was successful by combining
the results for all transformers with the same power frequency into one class. Details are described in [B17].
The following two figures represent the ultimate result – Figure 37 the typical no-load sound power level range for
3-phase 50 Hz transformers and Figure 38 the typical no-load sound power level range for 3-phase 60 Hz
transformers.

Figure 37 – Typical ranges of no-load sound power levels for the specification of 3-phase 50 Hz
transformers

Figure 38 – Typical ranges of no-load sound power levels for the specification of 3-phase 60 Hz
transformers
The derived figures are considered universally valid, primarily for specification purposes, with the following
application conditions and remarks:
I. Application conditions
A. Ranges of sound power levels are intended for specification purposes. To account for practically
unavoidable uncertainties to the design value in light of parameter Lo, a 2 dB margin is included,
i.e. the curves are shifted up by 2 dB if compared with derived curves.
B. Ranges of sound power levels indicate typical sound power levels representing most transformer
designs, however not all.
C. Ranges of sound power levels apply for sound power levels determined as per latest IEC / IEEE
standard requirements.

46
 TB 940 – Power Transformer Audible Sound Requirements

D. It is noted that a lower sound power level threshold of 50 dB(A) applies for the determination in
normal test facilities with a sound pressure disturbance level of not more than 35 dB(A). This
must be considered if sound power level specifications/measurements fall below the threshold.
E. Ranges of sound power levels are valid for core steel grade HiB. The use of other steel grades
typically, but not invariably, results in higher sound power levels. This impact must be considered
by manufacturers in the design process. It also means sound levels beyond the upper typical limit
are possible and do not indicate an issue.
F. Ranges of sound power levels are valid for transformers without any external sound mitigation
measures such as panels or enclosures.
G. Ranges of sound power levels are valid for no-load condition without sound contribution of
running auxiliaries (such as fans and/or pumps) for cooling.
H. The horizontal axis in MVA is related to the unit building power at rated tap position and rated
transformer load.
II. Remarks
A. Exceeding the upper limit of the sound power level (typical natural maximum) does not indicate
any marginal or pure quality. Design optimization due to specification requirements can cause
the sound power level to be above the upper limit.
B. The specified transformer no-load sound power level has a large impact to the unit dimension,
mass and cost. For those reasons, it is recommended to avoid specifying the sound power level
at any lower level than indeed required.
C. No-load sound power level specifications at or above the typical average curve (red) do typically
not impose design limitations.
D. Design boundaries tighten, the closer a no-load sound power level specification approaches the
‘minimum achievable’ curve.
E. For units with a large power rating, transport limitations become possible (likely), the closer the
no-load sound power level specification approaches to the ‘minimum achievable’ curve.
F. No-load sound power level specifications below the ‘minimum achievable’ curve usually require
the installation of sound panels or of an enclosure.
Some guidance on how to do a suitable sound level selection/specification in conjunction with the load and cooling
system sound power levels is given in sections 3.5 and 5.4 - 5.6.

3.3 Typical ranges of load sound power level

The derivation of the typical load sound power level ranges of transformers is based on a model that is correlating
the governing physical parameter of the load sound generation process with the transformer load sound power
level. The model does not consider any design details specific to individual design concepts. Five
specification/design parameters were selected for each transformer, forming an individual dataset in the database
together with the load sound power level determined during FAT. The design parameters used as model inputs are:
▪ Transformer rated power3 continuous parameter
▪ Short-circuit impedance continuous parameter
▪ Power frequency discrete parameter comprising 50 Hz and 60 Hz classes
▪ Core type discrete parameter comprising two classes
▪ Transformer auto factor continuous parameter (collected but not used)
while the model output is the
▪ Transformer load sound power level.
Sound level contributions by continuous parameters are modelled as continuous functions for the entire parameter
range. Discrete parameters are handled by classification, i.e. several parameter sets are derived, each covering an
individual class.
The development of the model is described in detail in [B19]; in the following a summary is given.
The developed model is an improved model of the ‘Reiplinger’ load sound level equation [B4]. The ‘Reiplinger’
formulas are given in IEC 60076-10:2016 for three-phase 50 Hz and 60 Hz transformers as

3‘rated power’ and ‘transformer rated power‘ are equivalent terms to ‘apparent power’ and ‘transformer apparent
power’.

47
 TB 940 – Power Transformer Audible Sound Requirements

𝑆𝑟
𝐿𝑊 𝑅𝑒𝑖𝑝𝑙𝑖𝑛𝑔𝑒𝑟 3𝑝ℎ 50𝐻𝑧 (𝐼𝐸𝐶) = 39 + 18 log10
𝑆0
Equation 13
𝑆𝑟
𝐿𝑊 𝑅𝑒𝑖𝑝𝑙𝑖𝑔𝑒𝑟 3𝑝ℎ 60𝐻𝑧 (𝐼𝐸𝐶) = 44 + 18 log10
𝑆0
Equation 14
Lw transformer load sound power level in dB(A)
Sr transformer rated power in MVA
S0 base power 1 MVA
The basic idea behind the ‘Reiplinger’ concept is the correlation of an increasing load sound power level with
growing transformer rated power. The scaling of the equation was at the time done graphically by plotting a relatively
small number of load sound level measurements, all made at the same manufacturer/test environment, versus the
transformer rated power. The equation consists of two terms:
▪ a constant sound power level accounting for the acoustic test environment or in other words an equivalent
sound power level standing in for the background noise,
▪ a scaled logarithmic term representing the relative sound power level contribution of a transformer with
rated power Sr relative to a 1 MVA transformer, that is assumed to not contribute anything (0 dB).
The improved model proposed herein follows the same concept with some additional ideas:
1. The driving force for all vibrations of the transformer in load condition is the stray (leakage) field, caused
by winding currents. With reference to section 2.2.2, vibrating elements (sound sources) in load condition
are windings, stray field control elements and inactive metal parts with the windings being the dominating
source.
2. As a quantitative distinction between the vibrating source elements is not feasible by a simple approach,
the forces experienced by the windings are considered the driving parameter for the transformer load
sound. A parameter proportional to the winding forces is the transformer reactive power – calculated as
the product of rated power and short-circuit impedance – and is used as the model input parameter instead
of transformer rated power. The theoretical background is outlined in [B19].
3. Instead of using only a few datasets for the statistical scaling process, a large database, unique in the
industry and containing 1275 datasets as described above, collected by 14 independent transformer
manufacturers from all over the world, was formed and used for the scaling process.
The improved generic model for the load sound level is thereby given as:
𝑆𝑟 𝑢𝑘
𝐿𝑊 𝐶𝑖𝑔𝑟𝑒 = 𝑎 + 𝑏 log10 ( )
𝑄0
Equation 15
uk transformer short-circuit impedance in pu
Q0 base reactive power 1 MVA
Parameters Sr and uk are fundamental specification parameters and are known for each transformer. Not contained
in Equation 15 are collected parameters ‘power frequency’ and ‘core type’. Those are discrete parameters and
treated by establishing four separate transformer classes.
▪ 50 Hz -- 3-limb cores
▪ 50 Hz -- 5-limb cores + shell type
▪ 60 Hz -- 3-limb cores
▪ 60 Hz -- 5-limb cores + shell type
As for the no-load sound level analysis, it was finally possible to combine the two classes of ‘core type’ without
losing significant accuracy. Note, that for a detailed analysis at manufacturer level, it may be worthwhile to separate
‘core type’; and this applies for the no-load sound level analysis too. Finally, two classes of units remained for the
derivation of typical load sound power level ranges – the 50 Hz and the 60 Hz transformer classes.
Parameters a and b for the improved load sound power level model are next determined by linear regression using
the least square error method on the relevant collected datasets out of the database (1085 sets for the 50 Hz class,
190 sets for the 60 Hz class). After determining the model parameters, a statistical analysis has been performed to
determine the accuracy of the model. Following graphs (taken from [B19]) show the results exclusively for the 50
Hz transformer class. On the left side, the results using the ‘Reiplinger’ formulation are shown while on the right
side the results of the improved model are presented. Note the different abscissa parameter of the graphs; rated or
apparent power on the left and reactive power on the right.

48
 TB 940 – Power Transformer Audible Sound Requirements

Figure 39 – Load sound power levels of the 50 Hz transformers and derived average functions -
‘Reiplinger’ formulation (left), improved model (right)
The functions in the left-side graph show the ‘Reiplinger’ equation (beige) as given in Equation 13 and the function
derived by applying a regression to the collected dataset using the ‘Reiplinger’ model (black). While the parameters
are rounded to full numbers there, the ‘Average’ (blue) curve is showing the exact (not rounded) function.
𝑆𝑟
𝐿𝑊 𝑅𝑒𝑖𝑝𝑙𝑖𝑛𝑔𝑒𝑟 3𝑝ℎ 50𝐻𝑧 𝑓𝑜𝑟 𝑐𝑜𝑙𝑙𝑒𝑐𝑡𝑒𝑑 𝑑𝑎𝑡𝑎𝑏𝑎𝑠𝑒 = 33 + 21 log10
𝑆0
Equation 16
The function shown in the right-side graph is the derived result rounded to full numbers when applying the
regression to the collected dataset using the improved model.
𝑆𝑟 𝑢𝑘
𝐿𝑊 𝐶𝑖𝑔𝑟𝑒 3𝑝ℎ 50𝐻𝑧 = 53 + 19 log10 ( )
𝑄0
Equation 17
The similarly derived function for the 60 Hz transformer class results in
𝑆𝑟 𝑢𝑘
𝐿𝑊 𝐶𝑖𝑔𝑟𝑒 3𝑝ℎ 60𝐻𝑧 = 64 + 15 log10 ( )
𝑄0
Equation 18
Table 5 – Derived standard deviations and displacements from exact average curve
Standard deviation Displacement
Reiplinger equation -- 50 Hz -- [dB(A)] 5.3 -0.6
4
Reiplinger model -- 50 Hz -- [dB(A)] 5.1 -0.4
CIGRE A2.54 model -- 50 Hz -- [dB(A)] 4.7 0.5
CIGRE A2.54 model -- 60 Hz -- [dB(A)] 4.8 0.1
At a first glance, the improvement offered by the new model is not too significant. It is, however, important to note
that the deviation of individual datapoints from the derived function in the new model is more realistic because of
the improved model’s physical character. This is of importance for manufacturers to better understand and improve
their in-house load sound level estimation.
All functions/curves derived and discussed so far consider the average load sound power level only. This is because
the model contains only one input parameter and it consequently can return only one dependency and is the
average load sound power level, while in reality many more factors impact the transformer load sound level. The
description of the actual load sound power level by the model therefore requires the addition of the sound level
contributor due to all other factors. Following the same concept as for the analysis of the no-load sound power level,
let us call it again 𝐿𝑜 .
𝐿𝑜 load sound power level due to other factors in dB(A) – it can be positive or negative
The meaning of 𝐿𝑜 is simple: It is the difference between the calculated load sound level (using the model functions
provided above) and the actual (measured) load sound level of the transformer. As 𝐿𝑜 is unknown and not used in

4 Reiplinger equation with adapted parameters to fit the collected dataset.

49
 TB 940 – Power Transformer Audible Sound Requirements

practice, it is not added here to the functions given by Equation 13 to Equation 18 but it is important to be aware of
it.
As for the no-load sound power level estimation, 𝐿𝑜 comprises all factors not described by the model functions that
impact the load sound level of a transformer. Such are related to
▪ specific design aspects/impacts;
▪ material and manufacturing tolerances;
▪ temperature impacts;
▪ uncertainties related to the sound level measurement;
▪ tapping position, if applicable.
The parameter of tapping position has not been used in the analysis due of the manifold variants and design aspects
of tapping applications, and also because of the different definitions existing for tapping ranges. It is simply not
feasible to handle such aspects on the large number of database entries prepared by the many different
manufacturers. Tap position impacts are thereby considered a manufacturer relevant feature requiring their
attention.
The developed and improved model returns a standard deviation of 4,7 dB(A). This means that the measured load
sound power level of approximately 95 % of the transformers is within ±10 dB(A) of the estimated average load
sound power level. The spread of 20 dB(A) between the noisiest and the quietest transformer at the same reactive
power is accounted for by 𝐿𝑜 , i.e. to sound level contributions by factors not described in the model. This large
spread indicates the challenge of predicting the load sound power level of a transformer during the design stage,
because for the load sound level an independent distinct control parameter covering at least a significant part of
the spread does not exist (such as the core induction for no-load sound level control). The load sound level spread
is consequently induced by many different factors. To improve the accuracy of the model, those factors and their
impact have to be understood and the relevant parameters added to the model. As such parameters are mainly
design parameters and are only known by manufacturers, they are beyond the scope of WG A2.54 and cannot be
discussed here.
The following two figures represent the ultimate result of the development process – Figure 40 the typical load
sound power level range for 3-phase 50 Hz transformers and Figure 41 the typical load sound power level range
for 3-phase 60 Hz transformers.
The three curves presented in each figure give the typical range of load sound power levels for three-phase
transformers.
The curve in the middle is the typical average curve described by Equation 17 for 50 Hz transformers and Equation
18 for 60 Hz transformers. The average curve indicates that 50 % of the transformer population produces a higher
and 50 % a lower sound level than such given by the curve.
The top curve is the typical natural upper limit curve, which is the average curve plus 10 dB. All transformers
produced should be able to comply with this upper limit without making use of sound mitigation measures, resulting
in a (for sound purposes) lower-cost design.
The bottom curve is the typical natural lower limit curve, which is the average curve minus 10 dB. Almost all
transformers are noisier than this lower limit. A load sound level specification at this limit will likely lead to an increase
in costs and in most cases require external sound mitigation measures to be applied.
Specifying below the typical average curve results in a high probability that the sound level cannot be met without
special design provisions and/or external sound mitigation measures. In cases where the load sound power level
has to be specified below the average curve, possible solutions should be discussed at first with the transformer
manufacturers but potentially also with suppliers of sound mitigation solutions to come up with a technically and
economically reliable solution.
CIGRE WG A2.54 therefore recommends only specifying load sound power levels above the average curve, as a
normal practice.

50
 TB 940 – Power Transformer Audible Sound Requirements

Figure 40 – Typical ranges of load sound power levels for the specification of 3-phase 50 Hz transformers.

Figure 41 – Typical ranges of load sound power levels for the specification of 3-phase 60 Hz transformers.

The derived figures are considered universally valid for primarily specification purposes with the following
application conditions and remarks:
I. Application conditions
A. Ranges of sound power levels are intended for specification purposes.
B. Ranges of sound power levels indicate typical sound power levels representing most transformer
designs, however not all.
C. Ranges of sound power levels apply for sound power levels determined according to latest IEC /
IEEE standard requirements.
D. A lower sound power level threshold of 50 dB(A) applies for the determination in normal test
facilities with a sound pressure disturbance level of not more than 35 dB(A). This must be
considered if sound power level specifications fall below the threshold.
E. Ranges of sound power levels are valid for transformers without any external sound mitigation
measures such as panels or enclosures.
F. Ranges of sound power levels are valid for load condition without sound contribution of running
auxiliaries (fans, pumps) for cooling.
G. Load sound power level ranges can be derived for all loading conditions of the transformer as
long as the reactive power Q in MVAr as input parameter is correctly determined for the loading
condition in consideration. Note that short-circuit impedance voltage varies proportionally with
loading.

51
 TB 940 – Power Transformer Audible Sound Requirements

II. Remarks
A. Exceeding the upper limit of the sound power level (typical natural maximum) does not indicate
any marginal or pure quality. Design optimization due to specification requirements can cause
the sound power level to be above the upper limit.
B. The specified transformer load sound power level can impact the unit design and cost
significantly. For that reason, it is recommended to specify the load sound power level not lower
than indeed required.
C. Load sound power level specifications at or above the typical average curve (red) do not typically
require the application of external sound mitigation measures.
D. Load sound power level specifications below the ‘typical average’ (red curve) typically require
special design provisions or external sound mitigation measures.
Some guidance on how to do a suitable sound level selection/specification in conjunction with the no-load and
cooling system sound power levels is given in sections 3.5 and 5.4 to 5.6.

3.4 Typical ranges of cooling system sound power level

3.4.1 Introduction
The third major component contributing to the total sound power level of transformers, besides core and windings,
is the cooling system. This section describes the derivation of typical ranges of sound power levels for most common
power transformer cooling systems, for the purpose of evaluating or specifying sound levels for transformer
installations. The presented results were derived by performing two case studies.
The following types of cooling system are considered: Radiator cooling either tank mounted or realized as separate
stand-alone cooling banks and air coolers. Water coolers are excluded in the evaluation here because of the limited
range and number of applications and also because their sound level is usually relatively negligible when comparing
to the other two transformer sound level components. More design related details of the cooling
equipment/components, their performance and operation are provided in section 2.2.3 of this brochure.
IEC 60076-2 defines the following cooling modes depending on the fire point of the internal insulating liquid with
designation O and K and on the external cooling medium with designation A (air) and W (water), see Table 6.
Table 6 – Transformer cooling modes

ONAF / ONAN Radiators/Radiator cooling banks with natural liquid circulation and with
or without fans.
KNAF / KNAN

ODAF / ODAN Radiators/Radiator cooling banks or air coolers with forced and directed
KDAF / KDAN liquid circulation and with or without fans.

OFAF / OFAN Radiators/Radiator cooling banks or air coolers with forced liquid
KFAF / KFAN circulation and with or without fans.

Water cooler with forced and directed or with only forced liquid
ODWF / OFWF
circulation and with forced water flow (not further considered here).
KDWF / KFWF

3.4.2 Description of case studies

Two case studies with the target to derive the typical ranges of cooler system sound power levels have been
performed by WG A2.54 members – representing transformer and compact cooler manufacturers. The transformer
power rating for the two case studies was selected so as to best cover the power range of interest. As a precondition
for the studies, the loss of life was set ≤ 1. This means the temperature limits as per IEC 60076 of 60 K top liquid
rise, 65/70 K average winding rise and 78 K winding hot-spot rise were set as the thermal design base for the two
case studies. As internal cooling medium, the use of oil was defined.
As a first case, a 400 MVA system transformer equipped with a separate radiator bank or with air coolers was
defined to cover the upper power range. To derive the typical minimum and maximum sound power levels, the two
cooling modes ODAN and ODAF have been studied. For the ODAF mode, the two possible cooling systems –
radiator bank and air coolers – were investigated, for the ODAN mode only the radiator bank. As the ONAN cooling
mode does not contribute any sound and is not very common in the upper power range, a detailed analysis in terms
of the required footprint has been omitted for ONAN cooling mode while it was done for the other modes.
As a second case, a 63 MVA system transformer was defined to cover the lower power range. This transformer is
equipped with tank mounted radiators that were designed for cooling mode ONAN and for cooling mode ONAF.
Although the ONAN cooling mode does not contribute any sound, the ONAN design is needed to determine the

52
 TB 940 – Power Transformer Audible Sound Requirements

required space/footprint for the cooling system. Comparing the sound levels of the cooling system without relating
them to their space/footprint would not have practical value.

Case 1 example with ODAN Case 1 example with ODAF radiator Case 1 example with ODAF air
radiator bank bank cooler
Figure 42 – Example arrangements of cooling systems investigated within case 1 studies

Case 2 example with ONAF Case 2 example with ONAN

radiators radiators
Figure 43 – Example arrangements of cooling systems investigated within case 2 studies
The analysed results from the case studies are therefore always twofold: sound power level and footprint. The
reason for this is the varying footprint for the cooling system, in dependence on the target sound power level and
the type of cooling system. It is noted that the footprint information provided in this section always considers the
cooling system only. The necessary spacing between the transformer tank and the cooling system, as well as
between the cooling system and surrounding (fire) walls – to ensure proper service of the cooling system by
sufficient and properly directed airflow – is not included in the footprint information. As this spacing is of importance,
the following rule of thumb information is given here for all cooling systems comprising fans: a minimum distance
of 1.5 times the fan diameter is to be maintained between cooling system and any wall or ground.
A reduced clearance may be acceptable specifically for coolers but would require a detailed assessment of the
situation by the cooler manufacturer considering the total number of coolers and their arrangement, the number of
fans per cooler and the air flow per fan. The minimum distance may also be decreased if the cooler airflow is
manipulated by updraft louvres, directing the hot air for example above surrounding walls and thereby away from
the transformer vault.

3.4.3 Case study 1: 400 MVA Network Transformer

The parameters given in Table 7 were set as the requirement for the cooling system design study to be executed
by the different manufacturer/cooler suppliers. For the ODAF cooling mode, there was no sound level restriction
imposed in order to design the cooling system as compact and cost-efficient as possible. For the ODAN mode, the
cooling system (radiator bank) is intended to be of low noise. Due to the absence of fans, the low noise requirement
is naturally imposed because cooling fans are generally dominating the sound level when comparing with pumps.
Table 7 – Case study 1 requirements for the cooling system
kind of cooling system separate radiator bank / air cooler
cooling mode ODAN ODAF
sound level requirement low noise no limitation
oil volume flow rate m³/h 200 200
total losses to dissipate by cooling system kW 1300 1300

53
 TB 940 – Power Transformer Audible Sound Requirements

top oil temperature rise of transformer K 45 45

ambient temperature °C 20 20
number of radiator batteries /circuits 2 2
fan arrangement - vertical
auxiliary frequency Hz 50/60 50/60/EC
oil type Nynas Libra Nynas Libra
Within this first case study, in total 11 designs of transformer manufacturers and 2 parameter sets of cooler suppliers
have been derived and used for the analysis. The result in terms of cooling system sound power level and footprint
are provided in the following.

3.4.3.1 ODAF cooling mode

The average sound power level of the designs using radiator banks was found to be 97 dB(A). Within the evaluated
data sets, a total difference of 18 dB(A) between maximum and minimum sound power level was found. The spread
is mainly caused by the individual design guidelines of the respective manufacturers as well as by the different
number of fans and fan types used in the design.
The average footprint for the ODAF cooling system with 2 radiator banks is approximately 23.4 m² and found to be
about 5.4 times larger compared to the consumed footprint by the cooling system using air coolers. The acoustic
behaviour of the air cooler solutions was found to be comparable with such of radiator banks – an interesting and
important finding.
Table 8 – Summary of case study 1 findings for ODAF cooling mode
highest sound power level of radiator banks 106 dB(A)
average sound power level of radiator banks 97 dB(A)
lowest sound power level of radiator banks 88 dB(A)
average sound power level of air coolers 98 dB(A)
average footprint of radiator banks 23.4 m²
average footprint of air coolers 4.3 m²

3.4.3.2 ODAN cooling mode

The average sound power level of the designs using radiator banks (11 studies performed) was found to be
74 dB(A). Compared to the ODAF cooling mode the average value is reduced by more than 20 dB(A). Within the
evaluated data sets, a total difference of 15 dB(A) between maximum and minimum sound power level was found
for the ODAN cooling mode and was therewith slightly smaller than the spread for the ODAF solution. The sound
power level spread of pumps seems to be slightly smaller than such of fans.
The average footprint for the ODAN cooling system with 2 radiator banks is approximately 57 m². The footprint ratio
between the ODAN and the ODAF radiator banks is 2.4. This means, the designer must consider a 2.4 times larger
footprint area required for the ODAN cooling system. The benefit of a lower sound level goes along with the
disadvantage of a significantly increased footprint for the cooling system.
For completeness, a comparison was made for the ODAN radiator banks solution with an air cooler solution
equipped with EC fans of low noise type. While the two ODAN radiator banks require a 11.9 times larger footprint
area than the air coolers, the minimum possible sound power level for the air coolers (when using EC fans) is about
5 dB(A) higher than the average sound power level of the ODAN cooling mode. A difference of 8 dB(A) applies for
the average sound power level difference.
Table 9 – Summary of case study 1 findings for ODAN cooling mode
highest sound power level of radiator banks 83 dB(A)
average sound power level of radiator banks 74 dB(A)
lowest sound power level of radiator banks 68 dB(A)
average sound power level of air coolers 82 dB(A)
average footprint of radiator banks 57 m²
average footprint of air coolers 4.8 m²

54
 TB 940 – Power Transformer Audible Sound Requirements

3.4.4 Case study 2: 63 MVA Network Transformer

Parameters given in Table 10 were set as the requirement for the cooling system design study to be executed by
the different transformer manufacturers. Two types of cooling system were designed and compared in the same
manner as for the first example. While the ONAN cooling mode does not contribute to the sound level, the ONAF
mode with radiators mounted directly to the tank and equipped with fans offers the most compact solution, however
for the price of a potentially significant sound level contribution while cooling the transformer.
Table 10 – Case study 2 requirements for the cooling system
kind of cooling system mounted radiators
cooling mode ONAN ONAF
sound level requirement low noise no limitation
total losses to dissipate by cooling system kW 350 350
top oil temperature rise of transformer K 53 53
ambient temperature °C 20 20
fan arrangement - vertical
auxiliary frequency Hz - 50/60/EC
height of tank mm 3300 3300
oil type Nynas Libra Nynas Libra
distance middle radiator to middle winding mm 600 600
Within this second case study, in total 9 designs of transformer manufacturers have been executed and used for
the analysis. Both cooling modes are standard solutions for this kind of transformer.

3.4.4.1 ONAF cooling mode

The average sound power level of the designs was found to be 88 dB(A). Within the evaluated data sets, a total
difference of 19 dB(A) between maximum and minimum sound power level was found. This spread is practically
the same as found for the ODAF mode in the first case study.
The average footprint requirement for the radiators utilizing ONAF cooling mode is 6.7 m².
Table 11 – Summary of case study 2 findings for ONAF cooling mode
highest sound power level 98 dB(A)
average sound power level 88 dB(A)
lowest sound power level 79 dB(A)
average footprint of radiators 6.7 m²

3.4.4.2 ONAN cooling mode

There is no sound contribution for the ONAN cooling mode. Audible sound generated due to oil flow has no impact
on the overall transformer sound power level.
The average footprint derived from the data sets for the ONAN cooling mode is 13.4 m². The ONAN mode requires
a two times larger footprint compared to the ONAF mode. The footprint ratio between ONAN and ONAF cooling
system is somewhat smaller than found in the first case study but nevertheless practically comparable.

3.4.5 Cooling system sound power levels

The performed case studies revealed a possible sound level spread by design of about 15-20 dB(A) within one
specific cooling mode. When considering all available cooling modes other than ONAN (no sound contribution), the
possible theoretical design spread is about 35-40 dB(A) – a huge figure. This immediately allows the conclusion
that the sound power level of the cooling system can be adjusted as required and is practically never a limiting
factor for new power transformer purchases – of course as long as there are no footprint (or cost) restrictions.
Footprint limitations together with cost considerations (costs for transportation, cooling system, cooling system
losses, maintenance) will almost always reduce the likely sound power level range.
In terms of footprint requirement, it was found that for lower sound level specifications the footprint demand will
increase.
Conclusions from the case studies are summarised as follows:

55
 TB 940 – Power Transformer Audible Sound Requirements

▪ the footprint required for cooling mode ODAN versus that required for cooling mode ODAF (using
radiators/radiator banks) is about double
▪ the footprint required for cooling mode ONAN versus that required for cooling mode ONAF is also about
double
▪ ODAF cooling mode: in case air coolers are used instead of radiators/radiator banks, the footprint is
reduced by a factor of 5
▪ ODAN cooling mode: in case air coolers are used instead of radiators/radiator banks, the footprint is
reduced by a factor of >10
Typical ranges of cooling system sound power levels as a function of transformer rated power were developed from
the sound level results of performed case studies. By interpolation of the results as well as reasonable extrapolation,
the curves shown in Figure 44 and Figure 45 were derived. As footprint considerations are of importance for the
specification of new purchases as well and are not included in below figures, the previously outlined footprint
aspects must also be considered in the specification process.

Figure 44 – Typical ranges of cooling system sound power levels for the specification of 3-phase 50 Hz
transformers

Figure 45 – Typical ranges of cooling system sound power levels for the specification of 3-phase 60 Hz
transformers
Due to the higher fan and pump speed for 60 Hz applications but also due to higher losses for 60 Hz transformers,
an increased upper limit of sound power level for the cooling system must be considered. Comparisons between
50 Hz and 60 Hz applications and further evaluations within the WG resulted in upshifted 50 Hz curves for 60 Hz
applications as shown in Figure 45.

56
 TB 940 – Power Transformer Audible Sound Requirements

The use of air coolers will generally enable any selected sound level between the upper and lower curves to be
realized. With the assumption of the same transformer losses and the adaption of the coolers as well as the use of
EC fans, the difference in the sound power level ranges obtainable at 50 Hz and 60 Hz is removed.

3.5 Combination of sound levels

In sections 3.1 to 3.4, the transformer sound power level components LW_no-load, LW_load and LW_cooling system have been
considered individually. For a transformer in service, the three sound level components are acting together and
have therefore to be superimposed to determine the total sound power level of the transformer. The derived
superimposed value is the input quantity for all types of acoustic studies of the transformer’s (the substation’s)
surrounding environment and is therefore of high importance for the end user. Needless to say, other sound sources
in the substation have likewise to be considered.
The superimposition of the three transformer sound power level components is done according to the following
equation:
𝑳𝒘 −𝑳 𝑳𝒘 −𝑳 𝑳𝒘
𝒏𝒐−𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒏𝒐−𝒍𝒐𝒂𝒅 𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒍𝒐𝒂𝒅 𝒄𝒐𝒐𝒍𝒊𝒏𝒈
[ 𝟏𝟎
] [ 𝟏𝟎
] [ 𝟏𝟎
]
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 + 𝟏𝟎 + 𝟏𝟎 )

Equation 19
For the no-load and load sound power level it is noted that mitigation components have been added in the equation.
Such apply only in the case that external sound mitigation measures, such as panels or an enclosure, are applied
to the tank. In the normal case without external sound mitigation, such components are set to zero.
In sections 3.1 to 3.4, the sound power levels of the transformer at rated conditions, i.e. at rated voltage, at rated
power (current) and with the cooling system running as specified for rated power, have been studied. If applying
selected sound power levels from the derived typical ranges at rated conditions for the three components, Equation
19 returns the total sound power level of the transformer for this condition, i.e. for rated service. The operation at
this or a similar operation point is a service condition that usually occurs only rarely, because it reflects the full and
not the typical power flow through the transformer.
For that reason, it is helpful, and recommended, to not only consider the transformer’s rated service, but also the
entire range of operation conditions. This provides a practically more realistic view on the transformer sound
emission. To derive such sound levels, a few pre-considerations have to be made for the individual sound power
level components:
No-load sound power level
The transformer no-load sound power level is directly linked to the network’s system voltage. As power grids are
operated at constant voltages, the transformer’s no-load sound power level is more or less constant for all operation
conditions. As there is, however, a tolerance band defined for the system voltages, a variation of ±2 dB (as per
experience) can practically be assumed due to varying system voltages, with the higher value occurring during light
network loading (at night).
Load sound power level
The load sound power level is increasing as a function of transformer loading according to the so called ’40 log
rule’. If assuming a known sound power level Lwr at rated load Sr, the sound power level at any other load S is
calculated as
𝑆
𝐿𝑊 = 𝐿𝑊𝑟 + 40 log10
𝑆𝑟
Equation 20
Note that this rule (formula) is using the transformer apparent power as input and not the transformer’s reactive
power as used for the derivation of the load sound power level in section 3.3.
Cooling system sound power level
The transformer cooling system, if configured other than exclusively for mode ONAN, is operated in dependency of
the transformer loading, either in steps (1, 2 or 3 steps typically), or in a more modern way continuously controlled
with EC fans (see section 2.2.3.2.2). Considering the classic control here, up to three different sound power levels
for the three cooling stages apply for specified loading ranges.
In the following example, the transformer’s sound power level components together with two types of sound power
level combinations (no-load + load and total) are derived using Equation 20 and are graphically presented for the
entire range of transformer operation:
From a transformer specification sheet for a double wound (full) transformer purchase, it is read:
300 MVA, 400 ±16% / 110 / 30 kV, 20 %, 50 Hz,

57
 TB 940 – Power Transformer Audible Sound Requirements

Cooling: 60 % load ONAN – 100 % load ODAN – beyond 100 % load ODAF (max load 120 %)
From the diagrams providing the ‘typical sound power level ranges’ – given in sections 3.2 to 3.4 – the following
sound power levels at rated service are selected:
No-load sound power level: At light transformer loading up to 40 % a slightly increased system voltage is assumed
which leads to an increased no-load sound power level of 84 dB(A). As loading increases and voltage decreases
assumingly to rated value, the no-load sound level decreases to 82 dB(A).
Load sound power level @ rated loading: 88 dB(A)
Cooling system sound power level:
▪ First cooling stage up to 60% of rated load: ONAN --- 0 dB(A)
▪ Second cooling stage beyond 60% and up to 100% of rated load: ODAN --- 72 dB(A)
▪ Third cooling stage beyond 100% of rated load: ODAF --- 87 dB(A)
Figure 46 represents the example transformer’s acoustic behaviour (sound power level) for the entire operation
range as specified.

Figure 46 – Transformer sound power level components and relevant combinations over the entire range
of transformer operation (300 MVA example transformer)
It is noted that the presentation as shown in Figure 46 can likewise be prepared using as input the sound power
level components as measured/determined during FAT and would serve as an ‘acoustic footprint’ of the transformer
over the entire range of operation. Such a figure can then be used as base input for the handling of all acoustic site
aspects that may occur.

3.6 Sound power levels of transformers for 50 Hz vs 60 Hz power

frequency
Differences in the emitted transformer sound power level as a function of power frequency are well known and
handled by manufactures in the design process. Common practice is to consider a fixed dB value (a positive or
negative adder) in the sound level design if deviating from the principal power frequency a manufacturer is working
with. A more detailed distinction is normally not made. This situation is, for instance, also reflected in Appendix B.5
of IEEE C57.12.90 [B11], where the sound level conversion between 50 Hz and 60 Hz excitation for transformers
is discussed.

3.6.1 Findings from the collected database

While analysing the 50 Hz and 60 Hz sound level data collected by WG A2.54, the somewhat surprising observation
was made that the sound level difference is not a constant (fixed) value but shows clearly a decreasing dependency
as a function of increasing power rating or more generally, of the transformer size. Figure 47 shows the observation
for both, the transformer no-load and load sound power level, based on the collected database entries.

58
 TB 940 – Power Transformer Audible Sound Requirements

Figure 47 – Measured 3~ transformer sound power levels – left side no-load, right side load condition
If discussing the acoustic performance of transformers as a function of power frequency, two situations must be
distinguished:
1. A given transformer designed for one power frequency which is sound level tested at both frequencies, at
the same core induction (no-load sound level) and same winding current (load sound level).
2. Two transformers, one designed for 50 Hz and the other designed for 60 Hz with same core mass and
core induction (no-load sound level) and same reactive power (load sound level).
The dependency shown in Figure 47 coincides with situation #2 and indicates obviously a dimensional impact: the
larger the unit is, the smaller is the sound power level difference between 50 Hz and 60 Hz transformers. This is
valid for no-load and load sound levels.
To verify and characterise the observation, the following was done:
Measured no-load sound power level datapoints of the two classes ’50 Hz 3-limb cores’ and ’60 Hz 3-limb cores’
were re-calculated using the developed calculation model (section 3.2) while setting the core induction to 1.63 T for
all units (1.63 T is the average core induction over all collected datapoints). By doing so, all impacts due to the
induction design value are removed and impacts of whatever other factors contributing to the observation are
thereby emphasized. In terms of load sound, the average sound power level difference between 50 Hz versus 60
Hz of 3-phase power transformers is simply derived by subtraction of Equation 18 and Equation 17 (section 3.3),
i.e. as a function of reactive power – the driving parameter for the load sound power level. Results are graphically
shown in Figure 48.

Figure 48 – No-load and load sound level difference between 50 Hz and 60 Hz power transformers
The sound level difference between 50 Hz and 60 Hz transformers is with Figure 48 confirmed to be strongly
dependent of the transformer size (power rating) and applies to both sound components, no-load and load sound
power level5. A difference of up to 10 dB(A) or even more for small units is a lot and means that small 60 Hz power
transformers radiate significantly more sound than comparable 50 Hz units. At the other end of the scale, very large
60 Hz power transformers are acoustically (on average) only to a minor extent affected by the larger power
frequency. It is again noted that the derived curves are of an average nature and sound control of individual
transformer designs to overcome this principal characteristic is always – at least within limits – possible.
The large number of datasets involved in the derivation of such findings clearly suggest a ruling physical
phenomenon behind, even though it is not yet clarified.

5 50 Hz vs 60 Hz sound level differences were first published by CIGRE WG A2.54 in [B17]. Figures 11 and 12 in
that reference were preliminary with the graphs representing 5-limb units being not correct. Instead, the trend
shown by graphs in Figure 48 in this document here are valid for all core types.

59
 TB 940 – Power Transformer Audible Sound Requirements

3.6.2 Findings from sound level testing activities of CIGRE WG A2.54 membership
The findings presented so far were derived from sound level tests of transformers designed for 50 Hz or for 60 Hz
(situation #2). The question, if the same or a similar finding applies also to above mentioned situation #1, i.e. to
transformers designed for one out of the two power frequencies and tested at both frequencies, requires further
investigation, although one would expect a similar behaviour.
In order to verify, or better to clarify this, tests were undertaken by the WG membership and are presented next.

3.6.2.1 No-load sound level tests

The no-load sound level of 14 transformers from different manufacturers were tested at three defined core
inductions (1.3 T, 1.5 T, 1.7 T) at 50 Hz and at 60 Hz power frequency. In Figure 49, a graphical representation of
the results (sound level differences) is given. Exact details (values) of the individual transformers including
interpretation details are given in [B20].
Figure 49 indicates a large spread of about 8 dB(A) for the averaged sound level differences (black dots).
Surprisingly, it is more or less independent of core mass (transformer size), i.e. the anticipated strongly reducing
sound level difference with increasing core mass as seen in Figure 48 (left graph) is not visible, although the average
curve indicates the tendency at least somewhat. The sound level difference of all tested units is in average about 4
dB(A) and is practically unchanged if analysing the sound level differences measured at the three induction levels
individually (coloured dots in Figure 49).

Figure 49 – No-load sound level difference of 14 sample transformers tested at 50 Hz and at 60 Hz

To further verify the findings, no-load sound level measurements at multiple frequency steps on four transformers
were conducted while the induction was kept constant at its rated value, see Figure 50.
Trendlines have been casted over the measured data points in order to obtain a slope function. An average value
for the slope of 0.4 dB(A)/Hz was found, which means 4 dB(A) per 10 Hz change in supply frequency. This is fully
aligned with the difference found above.

60
 TB 940 – Power Transformer Audible Sound Requirements

Figure 50 – No-load sound level measurements at multiple frequencies with constant (rated) induction.
Average sound level slope: 0.4 dB(A)/Hz

3.6.2.2 Load sound level tests

The same concept as previously used for the exploration of the no-load sound level difference was followed for the
load sound level. 14 transformers were tested at multiple frequencies, now excited with a constant (rated) current.
Some of the units are 50 Hz designs while the others are 60 Hz designs. Details of the individual transformers
including interpretation details of the test results are given in [B21].
In Figure 51, the main result is presented, i.e. the sound level difference of transformers tested at 50 Hz and at 60
Hz excitation with same current6. The overall spread of the measurements is with roughly 8 dB(A) high once again.
The average sound level difference derived from the graph is 5.9 dB(A). The large dependency of the sound level
difference from the transformer size (power rating) as derived and seen in Figure 48 (right graph) is also not visible
here, nevertheless, the lowering tendency of the sound level difference with increasing reactive power also indicates
a dimensional impact, but if so, to a lower extent than seen in Figure 48 for transformers, designed for 50 Hz and
for 60 Hz.

Figure 51 – Load sound level difference of 14 sample transformers when excited at 50 Hz and 60 Hz at
constant (rated) current
In Figure 52, the frequency dependency of the sound level for 4 selected transformers out of the 14 tested units are
shown, if excited at multiple frequencies, together with the slope lines (trendlines).

6 Note that the reactive power of a transformer changes by 20% when switching the frequency between 50 Hz to
60 Hz and keeping the current constant. This has not been considered in Figure 51 and datapoints are shown for
reactive power at principal frequency. The introduced uncertainty by doing so is acceptable for the effects
investigated here.

61
 TB 940 – Power Transformer Audible Sound Requirements

Figure 52 – Load sound level measurements at multiple frequencies with constant (rated) current.
The average sound level slope of the four slope lines in Figure 52 is with about 0.52 dB(A)/Hz or 5.2 dB(A) difference
between 50 and 60 Hz measurements and is well aligned with the previously derived average sound level difference
of 5.9 dB(A).

3.6.3 Interpretation of observed sound level differences as a function of power

frequency
The previously presented findings of sound level differences between 50 Hz and 60 Hz transformer designs, as
well as the frequency dependency of individual transformers, require interpretation. Such findings and
dependencies were to a good extent unexpected.

3.6.3.1 Theoretical considerations

In general, radiated sound power of a sound source (here transformers) is determined as [B22]:
𝑊 = 𝜌 𝑐 𝑆 𝜔2 𝑥 2 𝜎
Equation 21
where W is the radiated sound power, ρ the fluid density, c the speed of sound, S the sound radiating surface area,
ω the angular velocity, x the average magnitude (r.m.s.) of the surface displacement and σ the radiation efficiency.
The observed differences in the sound power levels between 50 Hz and 60 Hz for both situations (#1 and #2) are
therefore attributed to parameter changes in Equation 21 and are discussed and quantified (as far as possible)
here. Not contained in Equation 21 but also discussed here is the difference due to A-weighting. Note that A-
weighting is an acoustic filter to equalize the loudness for humans over the audible frequency range and is not a
physical parameter.
A. Surface velocity ν
Surface velocity is the product of angular velocity ω and average displacement x,
𝜈 = 𝜔𝑥
Equation 22
and is typically used in acoustics because radiated sound power is proportional to velocity square.
The comparison of no-load sound levels at different excitation frequencies always assumes an unchanged core
induction. With this boundary condition, the displacement magnitude of the vibrating core surface is unchanged
(constant) because magnetostriction is a direct dependency of core induction and is not dependent on frequency
(for the frequency range discussed here). The surface velocity of the core however varies with the excitation
frequency due to Equation 22. With radiated sound power being proportional to surface velocity square, the
difference of the radiated sound power level between two frequencies can be written and evaluated for 60 Hz vs.
50 Hz as:

62
 TB 940 – Power Transformer Audible Sound Requirements

𝑓1 60
𝛥𝐿𝑊 ~ 𝛥𝐿𝑝,𝜈 = 20 log ( ) = 20 log ( ) = 20 log(1.2) = 1.6 𝑑𝐵
𝑓2 50
Equation 23
Conclusion: The no-load sound power level of transformers excited at 60 Hz is increased by 1.6 dB. This is due to
the higher frequency, and thereby higher surface velocity, as long as the core induction is kept constant and other
impacts affecting the average displacement x (for instance resonances) are excluded.
The comparison of load sound power levels at different excitation frequencies always assumes a constant current
in contributing windings. To achieve this at different frequencies, the excitation voltage has to be varied
proportionally and means that the transformer power (voltage times current) varies proportionally to the frequency
change. The constant winding current results in constant forces and unchanged surface displacement magnitudes.
The surface velocity varies with the excitation frequency due to Equation 22. With radiated sound power being
proportional to surface velocity square, the difference of the radiated sound power level between two frequencies
can be written and evaluated for 60 Hz vs. 50 Hz also as per Equation 23.
Conclusion: The load sound power level of transformers loaded at 60 Hz is increased by 1.6 dB. This is due to the
higher frequency and thereby higher surface velocity, as long as the winding currents are kept constant and other
impacts affecting the average displacement x (for instance resonances) are excluded.
B. Weighting
All gathered and considered sound level measurement data are A-weighted. Because the weighting values are
frequency dependent, the average A-weighting is different for 50 Hz and 60 Hz excitation/transformers. Individual
weighting values are given in Table 12 for the first six main sound harmonics.
Table 12 – A-Weighting for typical transformer sound harmonics
Excitation Freq. 50Hz Excitation Freq. 60Hz
Harmonic A-weight Harmonic A-weight Difference
[Hz] [dB] [Hz] [dB] [dB]
100 -19.1 120 -16.7 2.4
200 -10.8 240 -9.1 1.7
300 -7.1 360 -5.6 1.5
400 -4.8 480 -3.5 1.3
500 -3.2 600 -2.2 1.0
600 -2.2 720 -1.3 0.9

Transformer no-load sound levels are determined by the first three to six harmonics as shown in Table 12,
depending on a number of design parameters such as core induction and selected core steel grade. The average
weighting difference between 50 Hz and 60 Hz excitation/transformers for the first three harmonics is 1.9 dB and
for the first six harmonics 1.5 dB. Hence the contribution of the A-weighting on the difference between 50 Hz and
60 Hz no-load sound levels is in the range of 1.5 to 1.9 dB, depending on the dominance of the individual harmonics.
Transformer load sound levels are typically determined by the first harmonic, in some cases additionally by the
second and maybe the third harmonic. The average weighting difference between 50 Hz and 60 Hz transformers
for the first three harmonics is 1.9 dB and for the typical case 2.4 dB. Hence the contribution of the A-weighting on
the difference between 50 Hz and 60 Hz load sound levels is approximately 2.4 dB.
Adding the two contributors evaluated above (surface velocity and A-weighting) to the sound level difference
between 50Hz and 60 Hz excitation/transformers, we find:
▪ No-load sound levels: about 3 to 3.5 dB(A) increase for 60 Hz excitation/transformers
▪ Load sound levels: about 4 dB(A) increase for 60 Hz excitation/transformers
This means a remaining gap of up to about 6 dB (for power transformers at the lower end of rated power) is to be
explained, based on findings provided at the beginning of this section.
C. Radiation efficiency σ
Radiation efficiency σ can formally be described by re-organizing Equation 21 as
𝑊
𝜎=
𝜌 𝑐 𝑆 𝜈2
Equation 24
It practically describes the part of the sound power generated by surface vibrations that is radiated into the acoustic
far field of the surrounding medium.
To determine radiation efficiency, the accurate prediction of a) the radiated sound power W that is emitted into the
far field and of b) the average surface velocity ν of the entire vibrating body is necessary according Equation 24.

63
 TB 940 – Power Transformer Audible Sound Requirements

Both parameters are not easy to determine – neither by simulations nor by measurements – and in case of power
transformers they are practically impossible to determine with sufficient accuracy. Simulations are usually unable
to cope accurately with the distributed sound source inside the tank (i.e. the active part with its vibrating elements)
while measurements suffer from the complex and extensive tank geometry with access limitations for the
measurement equipment (test room space, access limits to horizontal tank surfaces, tank accessories, high voltage
life parts…). To give an impression of the subject’s complexity in terms of tank geometry and vibration patterns,
Figure 53 taken from [B12] shows exemplarily the tank wall vibration patterns of a large power transformer.

Figure 53 – Tank vibration pattern measured in load (left) / no-load condition (right) at acoustically
dominating frequencies of 100 Hz / 300 Hz
It is noted that such patterns are not fully stable but easily vary as a result of more or less minor condition
modifications such as tap changer setting, temperature, loading etc. They normally vary also between identical units
under the same conditions due to tolerances involved. Nevertheless, vibration measurements on parts of the tank
(usually individual walls) are possible with acceptable accuracy so as to reveal qualitative insights which can be
useful for special investigations and partly also for verification purposes of simulations. Estimation of radiation
efficiency for an entire transformer still remains almost impossible with reasonable effort. This is the reason why
radiation efficiency is not a parameter used quantitatively in the transformer design process but is rather used as a
“filling-in parameter” to explain remaining (not quantifiable) sound level differences.
Although it is not really possible to estimate radiation efficiency as an absolute value, a relative difference between
50 Hz and 60 Hz excitation for a part of a large power transformer was derived exemplarily and can at least indicate
the order of impact. The velocity square of one tank wall determined by laser scanning in parallel to 50 Hz and 60
Hz load sound level measurements on three identical power transformers returned a 3.7 dB average difference.
The average sound level difference was found to be 4.1 dB (not A-weighted). Writing Equation 24 in logarithm form,
the difference in radiation efficiency between 50 Hz and 60 Hz excitation can be calculated as
𝛥𝜎 = 𝛥𝑊 − 𝛥𝜈 2 = 4.1 𝑑𝐵 − 3.7 𝑑𝐵 = 0.4 𝑑𝐵
Equation 25
Although uncertainties are involved in the derivation of this value, it can be assumed that the change in radiation
efficiency for large power transformers excited at 50 Hz and 60 Hz is not significant.
Unfortunately, no comparable investigations on transformers with low rated power were available. A few further
explanations of impacts to radiation efficiency are given in [B20][B21].
Following effects may explain the remaining discrepancies at least partly.
D. Resonances (linked to surface velocity ν)
Structural resonances of any of the main transformer parts, such as core, windings, tank and in part also
accessories can substantially increase the sound power levels beyond expectation and is caused by increased
deflection magnitudes of the vibrating (resonating) elements that finally increases the tank surface velocity ν. This
can be observed in Figure 50 and Figure 52, where transformer 4 and transformer 9 respectively have substantial
sound level peaks at or close to 50 Hz. It is however anticipated that in a large database some transformers
experience resonance while others do not, and therefore, the effect of resonances to the average functions as
derived by the statistical analysis of the collected database contributes somewhat but is not determining.
E. Wavelengths of sound in oil and air relative to tank dimensions (linked to radiation efficiency σ)
With the sound wavelength being defined as λ=c/f, the ratio of 1.2 follows for the acoustic wavelengths of 100 Hz
versus 120 Hz and its harmonics. Considering the fundamental acoustic waves, the following approximated
wavelengths result
▪ in air: 100 Hz – 3.4 m; 120 Hz – 2.8 m
▪ in oil: 100 Hz – 12 m; 120 Hz – 10 m
Consideration of the wavelengths of the fundamental harmonics in relation to the transformer tank dimensions, may
explain the observed differences between transformers produced for or excited at 50 Hz and 60 Hz beyond the
previously derived findings. For small liquid-immersed transformers with tanks of, let us say, 2 meters edge length

64
 TB 940 – Power Transformer Audible Sound Requirements

(much less than half the wavelength in oil), tank vibrations follow practically the active part vibrations. This will
translate into a kind of global three-phase piston vibration pattern with high radiation efficiency. On the air side,
where the tank bending wavelength is close to the sound wavelength, minor differences between tank bending
wavelength and sound wavelength in air play an important role for the sound radiation. The shorter the length of
the sound wave, the higher the radiated sound power perpendicular (normal) to the tank wall, i.e. into the far field
(due to a higher radiation efficiency) and this results in the large sound power level difference observed for small
transformers. Note that tank edge lengths can be assumed to vary only a little in average between 50 Hz and 60 Hz
units.
On the other hand, for large transformers with tanks of edge lengths of several meters up to sound wavelength in
oil or even more, tank vibration patterns are formed by complex interactions and have fairly small bending
wavelengths, see Figure 54.

Figure 54 – Measured surface velocity patterns of the short tank wall of a large power transformer
left side at 50 Hz and right side at 60 Hz excitation frequency
As the vibration patterns as shown in Figure 54. are obviously not too different for 50 Hz and 60 Hz excitation,
and the sound wavelengths in air of both frequencies are much larger than the bending wavelength of the tank
pattern, the radiation efficiency is very low and a dramatic difference in sound radiation (radiation efficiency)
between the two frequencies cannot be expected on large power transformers (such as derived in the example
calculation above). This results in the smaller sound power level difference observed for large power transformers
close to the theoretically derived base difference of 3-4 dB.
More studies are necessary to explore the dimensional effect and further impact parameters in more detail. With
the explanations given here, CIGRE WG A2.54 just wants to raise awareness of the topic and trigger further
research activities.

3.7 Sound power levels of single-phase units forming transformer banks

The installation of single-phase power transformers in the power system is common for transmission tasks of large
or very large power and/or if very high system voltages apply. In such situations, transformer transportation and/or
the spacing between phases (such as bushing clearance) are limiting factors. As an example, transformers with a
highest voltage of equipment beyond 550 kV are practically exclusively executed as banks formed by single-phase
units. Furthermore, units with a highest voltage of equipment of 550 kV are already widely executed as banks.
Additionally, transformers for HVDC applications are for the same purposes in the majority designed as single-
phase units.
Therefore, CIGRE WG A2.54 also collected data for single-phase power transformers according to the same
principles as described in sections 3.1 to 3.3 with the intention to derive similar typical ranges of sound power levels
that could serve as a basis for the derivation of typical ranges of sound power levels for transformer banks. The
number of received data was however much smaller than such for three-phase units and naturally limited to large
power ratings. Moreover, the vast majority of collected data refers to 60 Hz transformers such that relevant
information for 50 Hz units is not possible to provide. In addition, the different core types used for single-phase
transformers would reduce the accuracy of a detailed analysis further and fair quantitative provisions are
consequently not possible. As a result, it was decided to present the received and validated sound power level
measurement data for single-phase 60 Hz transformers only for information, but to not do further processing of
them. The development of enveloping figures as done for three-phase units would not be reliable. Despite missing
data for 50 Hz single-phase transformers, a conversion of 60 Hz sound levels to 50 Hz sound levels is possible with
reference to the findings and information given earlier in section 3.6. The following conversion is suggested for both,
no-load and load sound levels:

65
 TB 940 – Power Transformer Audible Sound Requirements

𝐿𝑊 1𝑝ℎ 50𝐻𝑧 = 𝐿𝑊 1𝑝ℎ 60𝐻𝑧 − 3 𝑑𝐵

Equation 26
The following graphs show the received 60 Hz single-phase sound power level data points for no-load (Figure 55)
and load (Figure 56) service at their assigned 1-phase power, displayed however in the enveloping figures
developed and presented in sections 3.2 and 3.3 for three-phase 60 Hz transformers7.

Figure 55 – Measured 1-phase 60 Hz no-load sound power levels over assigned 1-phase unit building
power – displayed in enveloping curves for 3-phase 60 Hz transformers

Figure 56 – Measured 1-phase 60 Hz load sound power levels over assigned 1-phase reactive power –
displayed in enveloping curves for 3-phase 60 Hz transformers
The sound power level relation between single-phase transformers of a transformer bank and the transformer bank
was derived based on the assumption of three identical sound sources being superimposed without interference.

7 In [B18], typical ranges of no-load sound power levels of single-phase transformers and transformer banks were
presented and discussed. Although such information is correct, the methodology outlined there was not further
followed by WG A2.54 and herein given information developed instead.

66
 TB 940 – Power Transformer Audible Sound Requirements

This assumption is sufficiently accurate for large units and was proven by the following investigation: Two no-load
sound level tests were performed in a large test lab, the first on three 60 Hz, 400 MVA, single-phase autotransformer
(216 MVA building power) with the individual results being averaged and the second on the corresponding 1200
MVA bank. The increased sound level of 4.7 dB(A) that was measured for the bank confirmed the theoretical
assumption of 4.8 dB(A) fully.
To confirm the sound level difference between single-phase transformers of a transformer bank and the transformer
bank, the same units tested before for no-load service were now likewise tested at load service. The found sound
level difference for load service was with approximately 7 dB(A) larger than the theoretical expectation of 4.8 dB(A)
and is attributed to the tight/difficult test environment for the bank test that affects specifically the load sound level
test due to the dominant first sound harmonic at load service. Important to recognize from both tests is that a
reduced sound level difference compared to the theoretically assumption of 4.8 dB(A) due to sound interferences
between the individual poles cannot be expected. For specification purposes a sound level difference of 5 dB(A) is
therefore suggested to apply with Equation 27 presenting the relevant formulation:
𝐿𝑊 3𝑝ℎ 𝑏𝑎𝑛𝑘 = 𝐿𝑊 1𝑝ℎ + 5 𝑑𝐵
Equation 27
Note that this finding / equation is important because transformer banks are almost never tested as banks but the
individual poles only. For the substation sound level guarantee the bank sound power level is however relevant.
The usage of presented sound level data in Figure 55 and Figure 56. together with Equation 26 and
Equation 27 for specification purposes of single-phase power transformers and the assigned bank should be
as follows:
▪ Select reasonable sound levels based on derived building power and reactive power of new transformer
purchase using Figure 55 and Figure 56. It is recommended to not select/specify sound levels
below red average lines because of uncertainties involved in single-phase applications.
▪ For 50 Hz applications, convert the selected sound levels using Equation 26.
▪ To derive the sound level of the assigned transformer bank, apply Equation 27.
▪ For your information, compare derived transformer bank sound levels with sound levels of an equivalent
three-phase transformer selected from relevant figures in sections 3.2 and 3.3.

Example
Given: three-phase autotransformer bank Sr = 1000 MVA, 60 Hz [50 Hz], 24%, auto-factor 0.53
Derived single-phase unit: Sr = 333 MVA, Sb = 333 MVA*0.53 = 176 MVA, Q = 333 MVA *0.24 = 80 MVAr

Single-phase transformer sound power levels selected from Figure 55 and Figure 56:
no-load sound power level = 90 dB(A) @ 50 Hz - Equation 26 [87 dB(A)]
load sound power level = 96 dB(A) @ 50 Hz - Equation 26 [93 dB(A)]
bank sound levels calculated acc. Equation 27
no-load sound power level = 95 dB(A) [92 dB(A)]
load sound power level = 101 dB(A) [98 dB(A)]
three-phase transformer sound levels selected from section 3.2. and 3.3
no-load sound power level = 97 dB(A) @ Sb = 530 MVA [93 dB(A)] @ Sb = 530 MVA
load sound power level = 102 dB(A) @ Q = 240 MVAr [101 dB(A)] @ Q = 240 MVAr
Note: Sound levels of three-phase transformers are usually 1-3 dB higher than the assigned bank sound levels.

67
 TB 940 – Power Transformer Audible Sound Requirements

4. Sound levels of other transformer types

Frank Cornelius (DE), Kohei Yamaguchi (JP), Shingo Kano (JP) and Selim Yurekten (TR)

4.1 Dry-type transformers

4.1.1 Introduction
Dry-type transformers use a solid insulation medium and ambient air for unit insulation and cooling. The transformer
core is typically protected from corrosion by a suitable painting system and clamped with a steel frame, while the
winding conductors are sealed or coated within a suitable solid insulation material. The core and coils can be
designed and manufactured in different ways and technologies to fulfill the electrical, mechanical and thermal
requirements for the different applications. Due to the properties of the insulation system (solid / solidified material
and air) including required safety margins for moisture and dust, dry-type transformers are larger compared to
liquid-insulated transformers of same rating and have higher losses.
The majority of dry-type transformers are installed and used for industrial applications, indoors and outdoors.
Special solutions using a single unit, or only a few units of the same design, are frequently requested. Dry-type
transformers for industrial solutions operate normally – at least at one voltage level – on power electronic converters.
Such converters can be understood as non-linear sources or loads producing non-sinusoidal currents and/or
voltages. The connected transformer is therefore exposed to harmonic spectra that must be handled adequately.
Dry-type transformers for distribution applications (“dry-type distribution transformers”) operate at sinusoidal
quantities, i.e. source and load are categorized as linear. They are typically operated indoors. For outdoor
installations, an enclosure with adequate protection against solid particle and liquid ingress is used.
The typical power range of dry-type transformers is up to 40 MVA and the highest voltage of equipment can be
typically up to 72.5 kV.

4.1.2 Technologies
A large variety of different dry-type transformer technologies exist in the market. The difference is mostly in the coil
design and manufacturing technology. Most common technologies are
▪ Vacuum Cast Coils (VCC)
▪ Open Wound (OW)
▪ Vacuum Pressure Impregnated (VPI)
▪ Glass fiber/epoxy imbedded coils
Table 13 – Common dry-type technologies

VCC (Vacuum Cast Coils) OW (Open wound) transformers Transformers with glass fiber/
transformers: Dry-type / VPI (Vacuum pressure epoxy imbedded coils: Dry-type
transformers with vacuum cast impregnated transformers: Dry- transformers with glass fiber
coils. type transformers with open reinforced and impregnated
wound and optional vacuum coils manufactured in one
pressure impregnated coils. process step.
Coils are imbedded in epoxy resin.
The coil shape is given by a casting
mold. Coils are openly wound and Coils are imbedded in a glass
impregnated in an epoxy bath in fiber/epoxy matrix.
the second step.

68
 TB 940 – Power Transformer Audible Sound Requirements

4.1.3 Enclosures
For the purpose of protection against environmental impacts such as pollution and weather conditions but also for
health and safety reasons, dry-type transformers are frequently located and operated inside enclosures. There are
many different variants of enclosures available, with their design mainly driven by the required protection code and
cooling requirements.
Table 14 – Enclosures for dry-type transformers
IP 23 IP 31 IP44
Enclosure for natural air Enclosure for natural air cooling Enclosure with external cooler AFWF (Air
cooling AN AN Forced Water Forced)

It is noted that beside the IP code system defined in IEC 60529 [B23] other protection systems, such as defined in
NEMA Standard 250 [B24] are in common use.
The enclosure parameters also have a significant impact on the aggregate’s (transformer plus enclosure) sound
emission and therewith to the sound level applicable for guarantee purposes. Although the tendency can be
observed, that a higher IP code of the enclosure results in a larger sound level reduction, this statement is not
always valid. A case-by-case consideration remains unavoidable. Typical no-load sound level reductions for the
aggregate compared to the bare transformer are in the range of 2-4 dB(A). Higher levels of sound reduction are
possible, but it is also noted that in case of an unfortunate dynamic situation (sound/vibration excitation matching a
natural enclosure resonance) the opposite can occur, and a sound level increase is possible for the aggregate.

4.1.4 Applications
Besides applications for distribution purposes, dry-type transformers are used in many special and niche
applications. For these various applications and installation spots, transformer and enclosure design differ widely.
The typical main reason for the application of dry-type transformers instead of smaller and lower loss liquid-
immersed transformers is the dramatically reduced risk in case of a transformer failure (explosion) regarding health
& safety aspects and environmental concerns.
Typical applications for dry-type transformers are:
▪ power distribution
▪ rolling-stock (railway transformers)
▪ solar and wind generation
▪ marine
▪ industrial solutions (drives, mining, pulp & paper, oil & gas, cement…)
▪ utilities (water, electrical)
▪ inside buildings
▪ nuclear facilities

4.1.5 Sound / sound levels

For most dry-type transformers, sound requirements are not in focus. This is reflected in “IEC 60076-11 Power
Transformers – Part 11: Dry-type transformers” [B25] where the sound level test is defined as a special test and
therewith only required on customer request.
If the sound level of a dry-type transformer is of interest, it typically concerns the no-load sound level only. The stiff
structure of the windings results in comparably low winding vibrations with the load sound level as a result being
significantly lower than the no-load sound level. Only at very low no-load sound levels, i.e. for designs with low core
induction, the load sound level may contribute to the total sound level. It is noted that load sound levels for dry-type
transformers have not been investigated by WG A2.54.
No-load sound levels of dry-type and liquid-immersed transformers are not comparable: The slightly increased core
mass and dimensions at the same power rating contribute to a higher sound level of dry-type transformers, the
major sound level increase impact for dry-type transformers is however related to the absence of the liquid-filled

69
 TB 940 – Power Transformer Audible Sound Requirements

tank. As a rule of thumb, no-load sound levels of dry-type transformers are 10 dB(A) or more above such of liquid-
immersed transformers with the same power rating, keeping core steel grade and induction equivalent.
The physics of sound generation in core and winding is identical for dry-type and liquid-immersed transformers.
Due to the higher stiffness of the entire dry-type structure, structural resonances are typically occurring at higher
frequencies than for comparable liquid-immersed transformers with also higher resonance amplification due to a
lower damping – mainly because of the missing liquid. Therefore, the control of mechanical resonances is of
importance.
The following graph given in Figure 57 was derived from sound level test data collected from a few manufacturers
of dry-type transformers but also from sound level tests specifically performed on dry-type transformers by WG
A2.54 members as well as from knowledge / information available to the WG.

Figure 57 – Typical ranges of no-load sound power levels for 3-phase dry-type transformers
Notes / hints for the application of Figure 57:
▪ Curves are based on CGO / RGO steel grades. The use of HiB based steel grades provide potential for a
sound level reduction.
▪ The typical range of sound power levels between the “minimum achievable” and the “typical natural
maximum” boundary curves is mainly controlled by varying the core induction and steel grade.
▪ Due to insufficient sound level data for 60 Hz transformers, typical ranges of sound power levels could not
be derived. Instead, a best possible rule is provided by A2.54, taking all available information and
knowledge into consideration.
▪ As the available sound level data for transformers above 10 MVA was limited, curves are shown in dotted
format. Although this is intended as a warning, there is no physical reason why this range should not be
valid.

4.2 Gas-insulated transformers

4.2.1 General information
In order to handle safety requirements for power transformers in a better way and protect them against dangerous
fires, gas-insulated transformers were developed. So far, gas-insulated transformers use SF6 (sulfur hexafluoride)
gas as insulating medium. The development took place in the latter half of the 1980s – mainly in Asian-Pacific
countries – and hence the majority of installations is found there. The technology is used whenever disaster
prevention and environmental protection have to be considered. In particular for indoor transformer installations,
including underground substations for power distribution, but also power transmission in urban areas.
Compared to mineral oil and other insulating liquids, SF6 has the following features:
1. SF6 gas is nonflammable, so disaster prevention can be improved. In addition, disaster preventing
equipment such as fire extinguishing systems can be omitted.

70
 TB 940 – Power Transformer Audible Sound Requirements

2. The height of the transformer can be lowered because it is unnecessary to provide a conservator attached
on top of the tank, necessary for liquid-immersed transformers to absorb liquid expansion due to
temperature changes. This offers a higher flexibility for the installation of gas-insulated transformers.
3. The gas-insulated transformer and the gas-insulated switchgear can be connected directly via the gas
bus. This makes it possible to rationalize the layout of the entire substation.
4. The density of SF6 gas is much smaller than such of insulating liquids and this enables high flexibility for
the arrangement of the cooling plant away from the transformer (cooling plant or instance at upper floors
while the transformer is underground).
5. SF6 is an extremely strong greenhouse gas. Its GWP (Global Warming Potential) is more than 23500 times
that of CO2.

4.2.2 Application range of gas-insulated transformers

Figure 58 shows the applicable range of gas-insulated transformers (2018 basis). Higher insulation voltage requires
higher SF6 gas pressure and higher transformer capacity requires forced gas cooling.

Figure 58 – Applicable range of gas-insulated transformers

4.2.3 Structural differences between liquid-immersed and gas-insulated transformers

Figure 59 illustrates a configuration of a gas-insulated transformer and Table 15 shows design differences between
liquid-immersed and gas-insulated transformers. The remarkable difference is the tank design: the tank of gas-
insulated transformers is designed as a pressure vessel. The tank strength is therefore higher and the tank wall
thickness approximately 1.5 times the wall thickness of liquid-immersed transformers. Due to this strong and thicker
tank, tank vibrations will be reduced – positively affecting the transformer sound level.

71
 TB 940 – Power Transformer Audible Sound Requirements

Figure 59 – Design concept of gas-insulated transformers

Table 15 – Design features of liquid-immersed and gas-insulated transformers
Liquid-immersed transformer Gas-insulated transformer
Tank design Oil tightness and vacuum-proof Pressure vessel design
Thickness of tank 100% Approx. 150%
Conservator Required Not-required
Firefighting system Required Not-required
Fire wall Required Not-required
Oil pit Required Not-required
Transportation volume 75% (without liquid) 100%
Unit height Low High

4.2.4 Sound level characteristic of gas-insulated transformers

A comparison of no-load sound power levels of liquid-immersed and gas-insulated transformers with equivalent
rating and similar flux density is shown in Figure 60. Gas-insulated transformers exhibit a lower sound level than
their equivalent liquid-immersed transformers. The higher tank strength of gas-insulated transformers usually
results in reduced tank vibrations and therefore a lower no-load sound level. Based on Figure 60 the no-load sound
power level of gas-insulated transformers is approximately 1-2 dB(A) lower than that of liquid-immersed
transformers.
As the power rating of SF6 transformers analyzed within this work is in the lower MVA range (see Figure 58), the
load sound level is practically not relevant. It is important to note that detailed information from units with larger
power rating could not be found within this work. Information on sound levels of SF6 units beyond a rating of 10 MVA
can therefore not be given here.

72
 TB 940 – Power Transformer Audible Sound Requirements

Figure 60 – Comparison of no-load sound power levels of gas-insulated and liquid-immersed

transformers
Based on the analysis done and exchanges with experts, as a whole it seems that the differences in sound levels
between liquid-immersed and gas-insulated transformers are not very significant.

4.3 Transformers with cores made from amorphous steel

4.3.1 General information
Amorphous cores are commercially mainly used for liquid-immersed distribution transformers but also for dry-type
applications. The maximum rated power of transformers with amorphous cores is about 15 MVA. Currently, there
are approximately 5 million distribution transformers worldwide with amorphous cores in operation – in India 2
million, in China 1,5 million and in Japan, the USA, Canada, Brazil and South Korea together about 1,5 million units.
In other regions, such as in Europe, the technology is practically not utilized. One reason for this is the availability
of the material.
Although the production process is more complex, the material costs are higher and the saturation induction of
amorphous materials are remarkably lower – resulting in larger cores and therewith transformers – amorphous
cores are used for the reason of significantly reduced core losses compared to transformers with cores made from
grain-oriented silicon steel (specifically such with conventional grades as for instance grade M3). Considering all
effects, transformers with amorphous cores have, overall, a somewhat lower total cost of ownership (TCO). It should
however be noted that the transformer failure rate seems to be higher, based on informal discussions – details are
however not known. It is also worth mentioning that the advantage in core losses of amorphous steels has been
reduced with the development of silicon steel domain refined HiB grades with thicknesses at or below 0.20 mm.

4.3.2 Losses
Hysteresis losses and eddy current losses are the two major components of losses in transformer cores. Both
components contribute to the reduced losses for amorphous materials. Amorphous metals can be more easily
magnetized than conventional (regular) grain-oriented silicon steel (here indicated as CRGO). This is indicated by
the smaller hysteresis loop that means lower hysteresis losses, see Figure 61.

Figure 61 – Hysteresis loops of core steel types

The eddy current losses are also lower in amorphous metals due to a combination of the approximately ten times
lower sheet thickness and the high electrical resistivity (more than 2 times higher) compared to CRGO silicon steel.

73
 TB 940 – Power Transformer Audible Sound Requirements

4.3.3 Sound levels

The audible sound level for distribution transformers is primarily caused by the core with its magnetostriction being
the root cause. As the magnetostriction is smaller in amorphous metals than for silicon steel, the no-load sound
level should theoretically be lower for amorphous core transformers. In practice however, this is not the case. The
opposite can be observed instead: Amorphous core transformers generate a higher sound level in the range of 5-
12 dB(A), depending upon the amorphous grade but also due to the following other effects:
1. Amorphous metal is very sensitive to mechanical stresses and is also brittle. Both characteristics require
a more careful handling in all subsequent production steps for amorphous core transformers than for
transformers with standard core steel. Unintended stresses introduced during production affect no-load
losses and audible sound level significantly.
2. Larger gaps exist between the stacked sheets for amorphous cores. These gaps affect the flux distribution
inside the core and cause out-of-plane forces, adversely affecting mainly the audible sound emission and
to a lower extent also core losses.
3. As amorphous cores are larger than cores made from standard silicon steel for transformers with the same
power rating (up to 50% more core mass), the audible sound level is enlarged.
The audible sound level dependency of amorphous cores is comparable to that of cores made from silicon steel. A
published experimental study shows that an increase of the core induction of 0.1 T results in a sound level increase
of 3-5 dB(A) [B26]. The finding of working group A2.54 for silicon steel transformers was 2-5 dB(A), with the lower
value applicable for low induction levels and the larger value applicable for large induction levels.

Figure 62 – Measured sound level vs induction – a case study

There are again and again attempts to reduce/mitigate the sound emission of amorphous core transformers.
Nothing, however, that is generally applicable, moreover easy and reliable to use, and, at the same time being cost
efficient, has been found and introduced to the industry so far.
As can be seen from Figure 62, there are two types of amorphous alloys used industry wide – both produced by
Metglas® Inc [B27]. Common abbreviations for them are SA1 and HB1. While the HB1 material offers moderate
5-10% lower no-load losses than SA1 material, the sound level difference is a minimum of 5 dB(A) lower for HB1.
Due to this behavior, overall, HB1 material is the preferred material for transformer applications with stringent sound
emission requirements.
Table 16 – Parameters of selected core steel types
Material Saturation flux Coercive force Specific losses Sheet thickness
density Bs [T] Hc [A/m] P13/50 [W/kg] [mm]
GO HiB steel 2.03 5.0 0.44 0.23
domain refined
Amorphous steel 1.56 2.0 0.1 0.025
SA1
Amorphous steel 1.64 1.5 0.085 0.025
HB1

74
 TB 940 – Power Transformer Audible Sound Requirements

4.3.4 No-load losses and sound power levels of liquid-immersed distribution

transformers
Losses and the no-load sound power levels for 50 Hz transformers with highest voltage for equipment not exceeding
36 kV are specified in EN 50708-2-1 [B28] and EN 50708-2-6 [B29]. Table 17 represents those values for liquid-
immersed transformers plus values added by CIGRE A2.54 in last column.
Table 17 – Loss and sound power level requirements for liquid-immersed transformers
EN 50708-2-1 EN 50708-2-6* CIGRE A2.54
Cores with conventional core steel amorphous cores amorphous cores
Rated Power TIER1 TIER1 TIER2 TIER2 LWA P0 LWA**
Pk P0 Pk P0
kVA W W W W dB(A) W dB(A)
≤ 25 900 70 600 63 37 35 -
50 1 100 90 750 81 39 45 -
100 1 750 145 1 250 130 41 75 -
160 2 350 210 1 750 189 44 105 -
250 3 250 300 2 350 270 47 150 61
315 3 900 360 2 800 324 49 180 63
400 4 600 430 3 250 387 50 220 64
500 5 500 510 3 900 459 51 260 65
630 6 500 600 4 600 540 52 300 66
800 8 400 650 6 000 585 53 330 67
1000 10 500 770 7 600 693 55 390 68
1250 11 000 950 9 500 855 56 480 69
1600 14 000 1 200 12 000 1 080 58 600 70
2000 18 000 1 450 15 000 1 305 60 730 72
2500 22 000 1 750 18 500 1 575 63 880 73
3150 27 500 2 200 23 000 1 980 64 1100 74
EN 50708-2-6*: Transformer no-load loss data P0 for amorphous cores as given in the standard are copied here in
the penultimate column. Corresponding no-load sound power levels are not given there. Proposed no-load sound
power level requirements in the last column LWA** are given by CIGRE A2.54 based on a group survey on current
industry practice. If required, a few dB(A) lower sound power level may be specified for new purchases (up to 5
dB(A)) as this can normally be designed for by reducing the nominal core induction. This of course will have a cost
impact.

4.3.5 Summary
Taking into consideration all facts outlined above, the technical and economic impact between distribution
transformers using silicon steel grades and such using amorphous steel, is limited. This also explains why the
amorphous core technology has not spread out all over the world but is still used in specific regions/countries only.
In case energy costs would further rise, it may be anticipated that the amorphous technology would increase its
market share.

75
 TB 940 – Power Transformer Audible Sound Requirements

5. Sound level specification and legislation

Janine Dickinson (GB) and Jan Dončuk (CZ)

5.1 Introduction
There is a lack of reliable and up-to-date guidance for the industry on how to specify sound power levels for
transformers, additionally there is insufficient guidance on the range of typical and achievable sound power levels
for transformers of different rated power. Consequently, sound level specifications for new transformers are often
unreasonable. Sound power levels are frequently specified unnecessarily high, but more often they are specified
too low, such that it is impossible to achieve without external sound mitigation measures, such as sound panels or
even full enclosures.
The guidance in this document aims to inform the industry, by providing useful guidelines for the sound level
specification of power transformers.

5.2 Survey / questionnaire

To understand how utilities currently specify sound levels for new transformer purchases and to identify any
common themes across the world, CIGRE WG A2.54 produced a questionnaire about utilities practice and
experience of transformer sound level specification.
The questionnaire, which comprised a series of 20 multi-choice and free text questions, was sent to utilities across
the world. 97 completed questionnaires were received and over 40 countries were represented in the responses,
as shown in Figure 63.

Figure 63 – Worlwide coverage of questionnaire responses

A copy of the questionnaire can be found in APPENDIX F, along with a summary of the results. The key findings of
the questionnaire are outlined hereafter.

5.2.1 Reasons for specifying maximum transformer sound levels

The need for utilities to specify sound limits for the transformers they purchase is driven by the requirement to
adhere to the noise limits laid down in their country of operation. Of the respondents, 93% of utilities reported some
form of noise limit they aren’t allowed to exceed and 98% specify a maximum sound level when purchasing a new
transformer. Utilities reported various reasons for specifying maximum sound levels for transformers, including:
▪ Environmental legislation
▪ Health and safety legislation
▪ Local policy
▪ Company policy
▪ Nearby residents

76
 TB 940 – Power Transformer Audible Sound Requirements

There is no noise limit that applies the world over. Noise limits vary from country to country and also within individual
countries. The survey revealed huge variation in the level of noise that is acceptable across the world and many
factors which affect the acceptability of those noise levels.
Noise limits vary for several reasons and depend on factors such as:
▪ Existing ambient noise levels in the area
▪ The nature of the surrounding land use e.g. residential, leisure, parks, commercial, industrial
▪ Tonality of the noise
▪ The time of day or night
▪ The location where the noise limit applies e.g. substation boundary, residential areas
▪ The level of noise deemed appropriate or acceptable in individual countries
▪ The type of transformer
In some countries, acceptable noise levels are set as absolute levels, whilst in others, they are set relative to existing
noise levels and in certain cases, penalties apply where the noise is tonal (a typical characteristic of transformer
noise).

5.2.2 Examples of noise limits in different countries across the world

Noise limits are generally lower at night when people are trying to sleep and lower in residential areas where people
live than in commercial or industrial areas where people work.
▪ Examples of absolute noise limits (sound pressure levels)
 30 dB(A) indoors at night with windows closed
 70 dB(A) in an industrial area
▪ Examples of relative noise limits (sound pressure levels)
 The noise must be 10 dB(A) below the existing background noise level
 Noise levels 3 dB(A) above background levels permitted during the daytime
Table 18 shows examples of noise limits (sound pressure levels) that apply in different locations across the world,
demonstrating the vast range of noise limits, locations and times of day during which they apply.
Table 18 – Examples of noise limits across the world

Noise Limit Location Time of Day Legislation/policy

Inside residential houses with

30 dB(A) Night Legislation
windows closed

35 dB(A) Residential area Day Legislation

Legislation
35 dB(A) Hospital Night Local Policy
Health and Safety

Legislation
40 dB(A) Residential area Night
Local Policy

Legislation
45 dB(A) Hospital Night
Local Policy

Legislation
50 dB(A) Schools Day
Local Policy

55 dB(A) Residential area Day Company Policy

60 dB(A) Commercial area Day & Night Legislation

77
 TB 940 – Power Transformer Audible Sound Requirements

Legislation
70 dB(A) Industrial area Day & Night
Local Policy

10 dB(A) below
Residential area Day & Night National Guidance
existing noise

1 dB(A) increase Residential area Night Company Policy

3 dB(A) above Legislation

Substation Boundary Day
background Local Policy

2 dB(A) above Legislation

Substation Boundary Night
background Local Policy

5.2.3 Sound Level specification for new transformers

When purchasing a transformer, 51% of utilities specify sound level limits in terms of sound power and 49% specify
sound level limits in terms of sound pressure; an almost equal split. There are also variations in the mode of sound
generation to which those sound level limits apply, such as:
▪ No-load sound level limits
▪ Load sound level limits
▪ Cooling system sound level limits
▪ Total sound level limits (e.g. no-load + load + cooling system or no-load + load)
In some cases, rather than to specify maximum sound levels for transformer purchases, utilities specify noise limits
that relate to the whole substation, for example, the installation of the transformer must not exceed a certain sound
pressure level at the substation boundary.

5.2.4 On site / off site limits

When determining the sound limit for a transformer purchase, 65% of utilities specify general sound level limits that
are not site specific.
In 44% of cases, noise limits apply on the substation site, whilst in 56% of cases the limits apply off site. Of the
cases where noise limits apply off site, in 56% of cases, the limits apply to residential areas.

5.2.5 Sound mitigation

To determine whether sound mitigation is required, utilities need to know the level of sound produced by the
transformer. Designed / guaranteed sound levels are verified during the Factory Acceptance Test with sound level
tests that shall follow standardized requirements as provided for instance in IEC 60076-10, [B9] or in IEEE
C57.12.90 [B11], utilizing either the sound pressure method or the sound intensity method. The survey revealed for
transformer sound level testing performed within the Factory Acceptance Test, 58% of utilities specify the sound
pressure test method, 14% specify the sound intensity test method and 28% do not specify a test method.
Utilities try to purchase quieter transformers to avoid the need to install mitigation. In the vast majority of cases
utilities try to mitigate noise at the source, rather than at the affected location. Typical forms of mitigation include
sound enclosures, tank mounted panels and sound barriers.

5.2.6 Tenders
Utilities use sound power and sound pressure levels for tender specification purposes. In many cases, where the
manufacturer cannot meet the required sound limit, they are no longer considered in the tender and in some cases,
they face financial penalties. Sometimes utilities specify low sound transformers to meet noise limit requirements
and utilities tend to look for the best sound level to cost ratio.

5.2.7 Compliance with noise limits

In 58% of cases utilities carry out sound level measurements either before or after a new transformer is installed,
whilst in 42% of cases no monitoring is undertaken at all. In 36% of cases, sound level monitoring is undertaken
both before and after installation to check compliance with noise limits.

78
 TB 940 – Power Transformer Audible Sound Requirements

5.2.8 Summary of questionnaire findings

In summary, the findings of the questionnaire demonstrate there are various reasons why utilities specify sound
level limits for transformers they purchase. There is no single noise limit that applies the world over. Noise limits
vary significantly from one country to another in terms of acceptable levels and the location at which those
acceptable levels apply.

5.3 Noise and substation design

When a operator decides to install a new transformer (whether a replacement on the same footprint, or a new
transformer in a new location), it is necessary to consider the noise impact as part of the design process. This is to
ensure any noise limits laid down in the country of operation are met when the transformer is in operation. These
noise limits may be outlined in specific legislation, regulations and company policy. Examples of noise limits across
the world are shown in Table 18.
Noise can affect the location and design of the transformer installation and the design of the substation as a whole.
The process for installing a new transformer, including consideration of the noise impact and need for sound
mitigation is outlined in Figure 64.

Figure 64 – Process of installing a new transformer on a substation

5.3.1 Noise study

A noise study is an important part of the substation design process. It is carried out to determine the maximum
allowable sound power level of a transformer, or level of sound mitigation necessary to meet any specific noise
limits. The level of sound mitigation required and maximum allowable sound power level of the transformer depends
on the specific noise limit and where that noise limit applies. It is important to determine the noise impact and any
requirement for sound mitigation as part of the design phase.
It is best practice to assess the noise impact of the proposed transformer as part of the substation as a whole,
rather than the individual transformer in isolation. Occasionally, a noise assessment of the individual transformer is
appropriate due to relevant noise limits.
The noise study requires several inputs to enable the best technical solution to be proposed during the design
phase. It is necessary to understand existing sound levels from equipment on site, the location of this equipment
and the location at which the noise limit applies. It may also be necessary to understand different modes of operation
of the equipment, if this affects the level of sound produced. Understanding existing noise levels enables overall
resulting noise levels to be predicted. It is highly recommended therefore to undertake sound level measurements
before the installation of a transformer.

79
 TB 940 – Power Transformer Audible Sound Requirements

5.3.2 Noise modelling

The existing noise climate is typically based on the 3D layout of the equipment within the substation, its
surroundings, local topography and existing sound sources (e.g. transformers in operation). The noise levels from
an existing substation can be modelled using noise modelling software packages. These software packages can
also be used to predict noise levels at various locations if a new transformer were to be installed on a site.
Noise modelling software programs require the sound power level of a transformer to be used as an input parameter
rather than sound pressure levels. Utilities however, frequently still specify sound pressure levels when procuring
transformers. This practice is confusing, as a sound pressure level must be specified at a set distance for it to have
any meaning and cannot be used as an input parameter into noise modelling software packages.
An example of a noise model produced to determine the existing noise climate on a substation site is shown in
Figure 65.

Figure 65 – Noise model of a substation showing the existing noise climate with two transformers in
operation [source - CEPS noise study]
If noise limits cannot be met without mitigation, the effect of installing various forms of mitigation can be predicted
using noise modelling software as shown in Figure 66 through Figure 69.

80
 TB 940 – Power Transformer Audible Sound Requirements

Figure 66 – Noise model of existing substation Figure 67 – Noise model of existing substation
with two transformers operating with two transformers operating, plus two
additional proposed transformers

Figure 68 - Noise model of existing substation Figure 69 – Noise model of existing substation
with two transformers operating, plus two with two transformers operating, plus two
additional proposed transformers and a 20 m additional proposed transformers both in noise
long, 10 m high barrier enclosures

Some forms of mitigation are only within the control of the manufacturer, as they apply specifically to the design of
the individual transformer. Other forms of mitigation are within the control of the operator, as they can be applied
on the substation site, such as sound enclosures and barriers. These forms of mitigation can be modelled by the
operator using noise modelling software packages as shown above. More information on mitigation is provided in
chapter 6.
When designing sound mitigation, the transformer arrangement, such as the cooling system, accessibility for
maintenance (OLTC, bushings and control cabinet) and safety rules must be taken into account.
In some cases, operators may need to obtain planning permission to build a new substation or to install a new
transformer. Often this requires utilities to predict the resulting noise level and demonstrate they will comply with
any noise limits laid down in the country of operation, for example, through the use of noise modelling software.
The planning permission may require noise levels to be measured to demonstrate the accuracy of the noise model
and to demonstrate any noise limits are not exceeded.

5.3.3 Complaints
If complaints are received from local residents during the operation of the substation, official noise measurements
are likely to be undertaken and these will inform any design requirements for noise mitigation. Complaints are often
challenging, complex and time consuming to investigate and resolve and may involve multiple parties with differing
priorities that require careful management.

81
 TB 940 – Power Transformer Audible Sound Requirements

5.4 Impact of sound level specification on costs, losses, transportation

5.4.1 Analysis
It is evident, that a reduction in transformer sound level will increase the total transformer cost. An indication is
given here on the cost increase based on a design study of two typical example transformers – a medium and a
large power transformer.
Large power transformer: 400 / 400 / 75 MVA station transformer, 400 kV +/- 16% // 145 kV // 30 kV, YNyn0d, 17%
impedance
Medium power transformer: 63 MVA, 120 kV // 66 kV, YNyn0(+d), 12% impedance
For both example transformers, a reduction of the no-load and load sound levels was proposed by modifying the
active part and tank design. The cooling system sound level was not addressed. A reduction of 5 dB for the no-load
sound level and of 4 dB for the load sound level were set as targets for both units. The reduction ranges for the
sound levels are given in Table 19 and Table 20. Also shown in Table 19 and Table 20 are the boundary sound
levels indicting the typical range of sound levels as derived by CIGRE WG A2.54 with the curves shown in
APPENDIX A.
In these examples, it is important to note the sound level reduction ranges relative to the typical range of sound
levels: For the load sound level, the reduction range is between the MAX and AVG boundaries for both example
transformers. The no-load sound level reduction for the 400 MVA unit occurs more or less symmetrically around
the AVG sound level while it is between the AVG and MIN boundaries for the 63 MVA unit.
Table 19 – Noise reduction versus range of noise as per CIGRE A2.54 recommendation - 400MVA
No-load sound Load sound
Calculated sound level reduction [dB] 92 dB(A) → 87 dB(A) = 5 95 dB(A) → 91 dB(A) = 4
Boundaries of typical range of sound levels MAX 95 – AVG 89 – MIN 83 MAX 98 – AVG 88
(APPENDIX A) [dB(A)]

Table 20 – Noise reduction versus range of noise as per CIGRE A2.54 recommendation - 63MVA
No-load sound Load sound
Calculated sound level reduction [dB] 73 dB(A) → 68 dB(A) = 5 77 dB(A) → 73 dB(A) = 4
Boundaries of typical range of sound MAX 86 – AVG 77 – MIN 67 MAX 80 – AVG 70
levels (APPENDIX A) [dB(A)]

5.4.2 Results
In both examples, the increase of total costs for the selected no-load sound level reduction is approximately 3%
and is also 3% for the selected load sound level reduction.
The increase of the total transformer costs is mainly driven by the additional material utilized (core and winding
material). The change of the core material grade/supplier may influence the costs for the core. Also manufacturing
costs will increase slightly.
As a rule of thumb, an increase of approximately 1% of the total transformer cost can be assumed for one decibel
noise reduction of the no-load sound level and an additional 1% for one decibel noise reduction of the load sound
level. This rule applies where the target sound level for the reduction is at least 1 dB above the AVG boundary in
the case of the load sound level and the MIN boundary in the case of the no-load sound level.
The no-load sound level reduction for both units was achieved by lowering the core induction. As a side effect, no-
load losses decreased by approximately 30%.
For the 400 MVA transformer, a reduction of the load losses of approximately 12% was incorporated in the design
change to reduce the load sound level.
In case a valuable loss capitalization is specified, the increased costs to achieve the reduction of the transformer
sound level may be paid back by the loss capitalization of the transformer.
As long as the target sound levels for a sound level reduction remain between the Max and AVG boundaries as per
CIGRE A2.54 curves (APPENDIX A), limitations or extra costs regarding transportation (specifically relevant for
large transformers) should not be expected.
In cases where the intended sound level reduction is targeting levels in the range between the AVG and MIN
boundaries, as per CIGRE A2.54 curves, transportation costs and manufacturing feasibility require attention.
Individual clarifications between bidder and purchaser in the case of particularly large units with low sound level

82
 TB 940 – Power Transformer Audible Sound Requirements

specifications may be necessary. In certain cases, the bidder may be forced to apply special design methods and
measures to achieve the maximum possible sound level reduction and transportation feasibility at the same time.
In such cases, the cost increase can be much more than calculated with the rule of thumb described above. Where
the specified sound levels are close to, at, or even below the MIN boundary, transformer costs can exponentially
rise and the proposed technical solution may not be feasible.

5.5 Parameters for specification purposes

When procuring a transformer, the purchaser should be clear about the test specification and parameters under
which the manufacturer should carry out the tests.

5.5.1 Sound power

IEC 60076-10 [B9] specifies that the acoustic performance of a transformer shall be indicated by its A-weighted
sound power level. Sound power is a universal, common denominator, which is irrespective of measurement
distance and independent of the environment. Sound power should be used for specification purposes.
The sound power level of a transformer is required to enable its operational noise impact to be predicted in its
intended location on a site. It is used as an input parameter into noise modelling software programs to enable a
prediction of the noise impact.
When procuring a transformer, the required / maximum permissible sound power level should be specified by the
purchaser, along with the applicable standard of measurement (e.g. IEC 60076-10).
Utilities frequently specify the sound pressure level instead of the sound power level when procuring a transformer.
This practice is confusing as a sound pressure level must be specified at a set distance for it to have at least some
meaning. But even if the measurement distance is known, the test result has no or only limited value as input for
noise propagation simulations.
A further observation is utilities frequently specifying a maximum sound pressure level at the substation boundary.
This is meaningless for a transformer manufacturer, as key information to enable them to calculate the maximum
allowable sound power level is missing. Specifying sound pressure levels can lead to misunderstanding and
inaccuracies and should not be used for specification purposes.

5.5.2 Sound level test parameters

The purchaser should be clear about the test specification/parameters under which the manufacturer should carry
out the tests, as they will affect the resulting sound power level.
It is best practice to specify maximum permissible sound power levels for all three individual components that make
up the total sound level of the transformer, these being:
▪ No-load sound
▪ Load sound
▪ Cooling system sound
By requesting the sound power level of each component individually (sound power level at no-load excitation, sound
power level due to load current (load sound power level) and sound power level of the cooling system, the noise
impact of each component can be fully understood.
No-load sound
The no-load sound power level may be an important consideration at night when the transformer is energised but
only lightly loaded.
Load sound
When combined with the no-load sound power level, the load sound power level gives an understanding of the
sound power level of the unit under load conditions, for example, during the day when the transformer is more
heavily or fully loaded.
The purchaser can specify different loadings and voltages under which the sound power level shall be determined
e.g. 50% loading and 105% of rated voltage. This aids the understanding of the sound power level of the transformer
under the relevant operating conditions.
Cooling system sound
The cooling system sound power level may be particularly important if the transformer is to be housed within a
sound enclosure, as once enclosed, the cooling system may become the dominant source of noise. Cooling system
sound may also be an important consideration as it may add to the total overall sound from the transformer.
When determining the sound power level with cooling system running, the number of running fans and the fan
speed of the cooling system selected for the sound level measurement shall be such necessary to operate the
transformer at its rated power under the most onerous external cooling medium conditions.
Cooling system sound may not be present at all times and therefore, depending upon the nature of the noise limit,
it may be possible to factor in the duration over which the cooling system will operate to enable a slightly higher

83
 TB 940 – Power Transformer Audible Sound Requirements

cooling system sound power limit. For example, if the site noise limit is set as an LAeq over 8 hours, but the cooling
system will only operate for 1 hour out of the 8, the contribution of cooling system sound averaged over an 8-hour
period would allow a higher cooling system sound power level to be specified than if the cooling system operated
for the full 8 hours. Purchasers may need to work with the supplier to determine realistic designs and sound power
limits. The information presented in 2.2.3 “Cooling system sound” can be used as a guide for specification purposes.
Section 11.1.2 of IEC 60076-10:2016 [B9] specifies a number of combined options for transformer energisation
during sound measurements. Two of these options involve the cooling system being in operation during both no-
load excitation and load current conditions. Attempting to determine the no-load sound and load sound power levels
with the cooling system in operation is considered bad practice, as it is difficult or may also be impossible to separate
the sound level contributions from the cooling system and transformer. It is best practice to specify sound power
levels for all three individual components separately.
It may also be relevant to specify the tapping position as a test parameter. For example, for transformers designed
to operate at variable flux, the sound power at no-load excitation is strongly impacted by the tapping position. The
tapping position for the noise measurement has therefore to be agreed between manufacturer and purchaser during
tender stage.

5.5.3 Combined sound power levels

Whilst it is best practice to specify maximum sound power levels for the three individual components, it may be
useful to specify combined sound power levels from the three components. This will depend on the noise limits in
the country of operation and the design and operation of the transformer. Two forms of combined sound power
levels are most commonly used:
▪ Transformer tank sound level
▪ Total transformer sound level
Transformer tank sound (no-load + load)
When modelling noise propagation from a substation, it may be useful to know the sound power level of the
transformer tank without cooling system operating. The sound power level limit for the transformer tank can be
specified and is the combination of the no-load and load sound power levels, separately measured and
logarithmically added. The procedure is outlined for example in chapter 13 of IEC 60076-10:2016 [B9] and also in
section 3.5 of this brochure.
Total sound (no-load + load + cooling system)
When modelling the noise propagation from a substation, the total sound level (no-load + load + cooling system
sound) may be the most important to consider, as this will give a worst case (highest sound level) assessment. This
feeds into the substation design, including any form of noise mitigation required. Transformer sound power level
limits may therefore be specified as a total limit, i.e. as a combination of no-load, load and cooling system sound
power levels.
If limits are specified for the three individual components separately, then the corresponding total sound power level
will be determined by logarithmic addition of the no-load, load and cooling system sound power levels. If the cooling
system sound level is more than 10 dB(A) below the tank sound power level, it will not contribute to the total sound
level and can be neglected.
In some circumstances it may also be considered good practice to specify the no-load sound power level and total
sound power level (comprising all three components). This provides the manufacturer with some design freedom
for the load sound level and cooling system sound level without compromising the transformer total sound level
requirements.

5.6 Achievable sound power levels for specification purposes

Using the graphs given in APPENDIX A, achievable no-load, load, cooling system and total sound power levels for
transformers with rated power larger than 3 MVA can be determined. When specifying the sound power level of a
transformer, the purchaser should use the charts to determine achievable sound power levels. Specifying sound
power levels outside of the given ranges may lead to additional costs, for example in cases where a ‘low sound’
special purchase transformer is required. The following three examples aim to provide guidance for the transformer
sound level specification by using the developed graphs.
All required data to determine the typical range of sound power levels of a transformer are known to the purchaser
at the time of specification. It is noted, that they are also given on any transformer nameplate. Normally not known
in a specification process is the need for external sound mitigating measures and the level of sound reduction
required. Although some hints are given in the last example here below, the exact estimate is typically a process in
close cooperation between the end user, transformer manufacturer and sound mitigation supplier to find the most
economical and best technical solution.

5.6.1 Example 1: Separate winding transformer

Transformer nameplate data

84
 TB 940 – Power Transformer Audible Sound Requirements

Rated power: 400 MVA

Rated voltage: 400±8*1.2% / 120 / 30 kV (HV/LV/TV)
Connection symbol: YNyn0d5
Number of phases: 3
Frequency: 50 Hz
Impedance voltage: 20.0 % at 400 MVA at nominal tap position HV-LV
Cooling: ONAN / ODAN / ODAF at 200 / 300 / 400 MVA
Sound mitigation: No sound mitigating measures
From the nameplate data it can be learned that the transformer is a three-winding transformer with 400 MVA top
rating of the main windings. Such windings are the high voltage winding of 400±8*1.2% kV and the low voltage
winding of 120 kV. The transformer additionally has a tertiary winding of 30 kV. For determining the sound power
level, the tertiary winding is not relevant. From the connection symbol it can be seen that the unit is a separate
winding transformer, for definitions of the connection symbol see IEC 60076-1.
No-load sound:
The typical range for the no-load sound power level is based on the building power at top rating. The building power
Sb of a separate winding transformer equals the rated power 𝑆𝑟 , so 𝑆𝑏 = 𝑆𝑟 = 400 𝑀𝑉𝐴. The minimum, average and
maximum no-load sound power levels at the corresponding building power for this transformer are derived from
Figure 70. Background information on the curves can be found in section 3.2.

Figure 70 – Typical sound power level ranges for 3-phase 50 Hz transformers at no-load
No-load Minimum Average Maximum
Sound power level 𝐿𝑤 [dB(A)] 83 89 95
Note: In case of a 3-limb unit, the minimum sound power level is 84 dB(A)

A transformer will operate during its lifetime continuously at or around rated voltage. This means that the no-load
sound power level of a transformer is almost constant and independent from loading.
Load Sound:
The typical range for the load sound power level is based on the reactive power Q of the transformer: 𝑄 = 𝑆𝑟 ⋅ 𝑢𝑘 =
400 𝑀𝑉𝐴 ⋅ 20 % = 80 MVAr. The minimum, average and maximum load sound power levels at the corresponding
reactive power for this transformer are deducted from Figure 71. Background information on the curves can be
found in section 3.2.

85
 TB 940 – Power Transformer Audible Sound Requirements

Figure 71 – Typical sound power level ranges for 3-phase 50 Hz transformers at load
Load Minimum Average Maximum
Sound power level 𝐿𝑤 [dB(A)] 79 89 99

The load sound power level of a transformer depends on the loading and therefore is not constant but varies as
explained in section 3.3.
The load sound power level of a transformer is normally required at rated power (top rating). When the load sound
power level needs instead or additionally to be known at a loading power other than the rated power (hereafter
indicated with index ‘2’), there are two ways to calculate it:
Method 1:
𝐼2 𝑆2
𝐿𝑤 2 = 𝐿𝑤 + 40 log10 𝑜𝑟 𝐿𝑤 2 = 𝐿𝑤 + 40 log10
𝑟 𝐼𝑟 𝑟 𝑆𝑟
Equation 28
e.g. the average load sound power level at 200 MVA loading instead of rated power will result in a sound power
level of:
𝑆2 200
𝐿𝑤 2 = 𝐿𝑤 + 40 log10 = 89 + 40 log10 = 89 − 12 = 77 𝑑𝐵(𝐴)
𝑟 𝑆𝑟 400
Equation 29
Method 2:
The load sound power level is directly read from the graph as shown in Figure 71 for the reactive power 𝑄2 at the
loading power 𝑆2 . Because the short-circuit impedance voltage in pu is proportional to the transformer loading, the
short-circuit impedance voltage changes with the same factor as the loading changes:
𝑆2
𝑆2 = 𝑆𝑟 ⋅ 𝑘 and 𝑢𝑘2 = 𝑢𝑘 ⋅ 𝑘 with 𝑘 = the loading factor in [pu].
𝑆𝑟

The reactive power will then be 𝑄2 = 𝑆𝑟 ⋅ 𝑢𝑘 ⋅ 𝑘 2 with 𝑢𝑘 and 𝑘 expressed in [pu].

For 200 MVA loading instead of the rated power of 400 MVA the reactive power becomes

200 2
𝑄2 = 𝑆𝑟 ⋅ 𝑢𝑘 ⋅ 𝑘 2 = 400 ⋅ 20 % ⋅ ( ) = 20 𝑀𝑉𝐴𝑟
400
From Figure 71 the load sound power level at 20 MVAr for the typical average line is found to be about 77 - 78 dB(A)
– practically equal to the sound power level found with Method 1.
Cooling system sound
The typical range for the cooling system sound power level is based on rated power Sr of the transformer, i.e. at
transformer top rating. This is because the cooling system including its sound producing components (fans and

86
 TB 940 – Power Transformer Audible Sound Requirements

pumps) are designed for this service condition. For service conditions other than top rating 400 MVA (ODAF), as
defined in the example with 200 MVA (ONAN) and 300 MVA (ODAN), specific considerations are necessary.
At the ODAF cooling stage the insulation liquid inside the transformer is circulated by means of pumps and the air
is forced through the coolers by fans. In general, it can be stated that fans are dominating the sound level if
compared to the pumps’ sound level.
The minimum, average and maximum cooling system sound power level at the corresponding rated power for this
transformer can be found in Figure 72 (green line). Background information can be found in section 3.4.

Figure 72 – Typical sound power level ranges for 3-phase 50 Hz cooling systems
Cooling ODAF (400 MVA) Minimum Average Maximum
Sound power level 𝐿𝑤 [dB(A)] 88 97 106

At ODAN cooling stage, the insulation liquid inside the transformer is circulated by means of pumps and the sound
produced by them has to be considered at this service condition. As the pumps are designed for rated power, the
pumps’ sound level is based on 400 MVA and not on ODAN defined service at 300 MVA (yellow dashed line in
Figure 72).
Cooling ODAN (300 MVA) Minimum Average Maximum
Sound power level 𝐿𝑤 [dB(A)] 68 74 78
Note: Sound level at ODAN is based on 400 MVA as pumps are designed for rated transformer power

At ONAN service, the cooling system does not produce any sound since there is only natural convection of both,
the air and the insulation liquid. Therefore, the contribution of the cooling system is 0 dB(A).
Cooling ONAN (200MVA) Minimum Average Maximum
Sound power level 𝐿𝑤 [dB(A)] 0 0 0

Combination of sound levels

During transformer service, all three sound level components may contribute to the total transformer sound level,
depending on the specific service condition. It is therefore necessary to combine the contributions from the
individual components for each defined service condition. Usually always required is the sound power level at rated
transformer service, i.e. the combination of all three components if the unit is running at rated voltage and rated
power and with the cooling system as designed for top rating switched on.
At the first step, it is necessary to select for each sound component a sound level that falls in the CIGRE derived
typical ranges of sound power levels. While selecting, it should be considered that lower sound level specifications
result generally in higher costs and may beyond a certain lower threshold result in the need for external mitigation
means, usually not a preferred solution. This is specifically true for the load sound power level – as indicated in the
relevant figures. Carefully studying the actual need for the sound power level and not selecting something lower
than required, is the right approach for this first step.

87
 TB 940 – Power Transformer Audible Sound Requirements

As a second step, the selected sound levels have to be combined by logarithmic summation, using the following
formula:
𝑳𝒘 −𝑳 𝑳𝒘 −𝑳 𝑳𝒘
𝒏𝒐−𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒏𝒐−𝒍𝒐𝒂𝒅 𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒍𝒐𝒂𝒅 𝒄𝒐𝒐𝒍𝒊𝒏𝒈
[ ] [ ] [ ]
𝟏𝟎
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 )

Equation 30
Note: 𝐿𝑚𝑖𝑡𝑖𝑔𝑎𝑡𝑖𝑜𝑛𝑛𝑜−𝑙𝑜𝑎𝑑 and 𝐿𝑚𝑖𝑡𝑖𝑔𝑎𝑡𝑖𝑜𝑛𝑙𝑜𝑎𝑑 are initially always set to zero to study the ‘bare’ transformer sound level emission
only. An example for the application of mitigation is given below in a third example.

In a third step, the sound power levels of the individual components may be reasonably varied and the combined
sound power level re-calculated to see the impact. Doing this iteratively will lead to a well justified specification of
either individual component sound power levels or a combined sound power level for the specific service condition.
It is noted that specifying combined sound power levels provides some more design freedom for the manufacturer
and should be considered in the specification process.
By repeating this process for all defined service conditions, a set of sound level specifications can be derived.
In the example case here, sound power level guarantees for three service conditions are specified:
1- Rated service: rated voltage, 100 % load, ODAF cooling stage
2- Typical service: rated voltage, 75% load, ODAN cooling stage
3- Nighttime service: rated voltage, 50% load, ONAN cooling stage
Obviously, the no-load sound level is identical for all three service conditions, while the load sound level and cooling
system sound level vary with loading. In Table 21 , the ‘Average’ sound power levels as derived from the curves,
as well as a selected reasonable alternative set of sound power levels called ‘Alternative’ are given for the individual
sound components together with the calculated combinations of the sound level components for defined service
conditions.
Table 21 – Sound power levels for specification – Example 1
Sound power level 𝐿𝑤 Average [dB(A)] Alternative [dB(A)]
Rated voltage 89 86
400 MVA load (100 %) 89 92
300 MVA load (75 %) 84 87
200 MVA load (50 %) 77 80
ODAF cooling stage 97 90
ODAN cooling stage 74 77
ODAN cooling stage 0 0
Combined rated service 98.2 → 98 94.7 → 95
Combined typical service 90.3 → 90 89.8 → 90
Combined nighttime service 89.3 → 89 86.2 → 86

The calculations for the defined service conditions are shown here for column ‘Average’. Input sound power levels
are those derived from the red average curves in above graphs:
𝟖𝟗−𝟎 𝟖𝟗−𝟎 𝟗𝟕
[ ] [ ] [ ]
Rated service: = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟗𝟖. 𝟐 𝒅𝑩(𝑨)
𝟖𝟗−𝟎 𝟖𝟒−𝟎 𝟕𝟒
[ ] [ ] [ ]
Typical service: = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟗𝟎. 𝟑 𝒅𝑩(𝑨)
𝟖𝟗−𝟎 𝟕𝟕−𝟎
[ ] [ ]
Night time service = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟖𝟗. 𝟑𝒅𝑩(𝑨)

Using the ‘Average’ sound power levels in Table 21 for the transformer sound level specification is one possibility,
for instance where typical or no specific sound level requirements apply. In cases where no sound level
requirements apply at all, sound levels between the red average curve and the upper limit curve may be selected
for the individual components. This may save some costs on new purchases. An alternative example is provided in
column ‘Alternative’ with the target to have a lower combined sound power level than for the ‘Average’ case. In the

88
 TB 940 – Power Transformer Audible Sound Requirements

alternative example, the load sound level is increased by 3 dB (applies to all three service conditions) while the no-
load sound level is reduced by 3 dB. The cooling system sound level is reduced by 7 dB for ODAF and increased
by 3 dB for ODAN stage.
Using the ‘Alternative’ sound power levels in Table 21 for the transformer sound level specification is one example
way to reduce the combined service sound levels without applying external sound mitigation measures. An
increased load sound level is generally welcome by manufacturers, as it is the most difficult sound level component
to control. The increased sound level for the ODAN cooling stage in the ‘Alternative’ example was selected as it
has a negligible impact to the combined sound level but provides more freedom when selecting the pumps. Also,
the no-load and cooling system sound level at ODAF are reduced; both can be controlled more easily than the load
sound. The achieved sound level differences between the two scenarios are significant for rated and nighttime
transformer service while the sound level at typical transformer service remains unchanged.
Although well-known, it is remarked once more here that sound levels should always be given as full numbers
without decimals. This is because of the logarithmic nature of sound levels, the insensitivity of the human hearing
for minor changes of sound levels and also because of natural uncertainties of sound level measurements.

5.6.2 Example 2: Auto-transformer

Transformer nameplate data
Rated power: 400 MVA
Rated voltage: 400±8*1.2% / 120 / 30 kV (HV/LV/TV)
Connection symbol: YNa0d5
Number of phases: 3
Frequency: 50 Hz
Impedance voltage: 20.0% at 400 MVA at nominal tap position HV-LV
Cooling: ONAN / ODAN / ODAF at 200 / 300 / 400 MVA
Sound mitigation: No sound mitigating measures
Example 2 is a similar transformer as Example 1 but the connection symbol has changed. In this case the
transformer is an auto-transformer instead of a separate winding transformer. The load sound and cooler sound
levels will be the same for an auto transformer and a separate winding transformer of same power rating, which is
the situation here. The no-load sound level however will change. An auto-transformer has a smaller core compared
to a separate winding transformer with equal rated power. This effect is taken into account by the building power.
Where for a separate winding transformer the building power equals the rated power 𝑆𝑏 = 𝑆𝑟 , for an auto
transformer this is not the case.
The building power of an auto-transformer is determined as per Equation 12 as follows:
𝑈𝐻𝑉 − 𝑈𝐿𝑉 400 𝑘𝑉 − 120 𝑘𝑉
𝑆𝑏 = 𝑆𝑟 ⋅ = 400 𝑀𝑉𝐴 ⋅ = 280 𝑀𝑉𝐴
𝑈𝐻𝑉 400 𝑘𝑉
Note: The no-load sound power level of auto-transformers is by definition correlated to the building power as appearing at rated
voltage.

By reading the no-load sound power level from the graph in Figure 70 for the auto-transformer with rated power of
400 MVA and corresponding building power of 280 MVA, the following no-load sound power levels are found:
No-load Minimum Average Maximum
Sound power level 𝐿𝑤 [dB(A)] 79 87 94

The determination of the total sound power level from the sound level components is done by the same procedure
as for the separate winding transformer discussed in Example 1. However, the modifications selected in column
'Alternative’ for the sound level components are different than such of Example 1, see Table 22. The target to
reduce the combined service sound levels for the three service conditions while respecting the provided ranges for
sound power levels outlined in the graphs above is however unchanged and fulfilled with the selected component
sound power levels.

89
 TB 940 – Power Transformer Audible Sound Requirements

Table 22 – Sound power levels for specification – Example 2

Sound power level 𝐿𝑤 Average [dB(A)] Alternative [dB(A)]
Rated voltage 87 84
400 MVA load (100 %) 89 89
300 MVA load (75 %) 84 84
200 MVA load (50 %) 77 77
ODAF cooling stage 97 88
ODAN cooling stage 74 77
ODAN cooling stage 0 0
Combined rated service 98.0 → 99 92.2 → 93
Combined typical service 89.0 → 90 87.4 → 88
Combined nighttime service 87.5 → 88 84.8 → 85

5.6.3 Example 3: Application of sound mitigation

To include external sound mitigation (panels, enclosures) into the sound level analysis as explored before, the
following information is required:
1. No-load sound level (without mitigation),
2. Load sound level (without mitigation),
3. Cooling system sound level (without mitigation),
4. Information on what surfaces are covered by sound mitigation elements,
5. Information on sound reduction of applied sound mitigation elements.
Points 1 to 3 can be found with use of APPENDIX A and are explained in examples 1 and 2.
Point 4: Usually the tank only is covered by sound mitigation elements, meaning that the no-load and load sound
levels are influenced by the sound mitigation measure. The cooling system is usually located outside the sound
mitigation measure and their sound level therefore not influenced. This is considered in the formula below where
only for the no-load sound and load sound components sound mitigation terms are included.
𝑳𝒘 −𝑳 𝑳𝒘 −𝑳 𝑳𝒘
𝒏𝒐−𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒏𝒐−𝒍𝒐𝒂𝒅 𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒍𝒐𝒂𝒅 𝒄𝒐𝒐𝒍𝒊𝒏𝒈
[ ] [ ] [ ]
𝟏𝟎
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 )

Where the cooling system is located inside the sound mitigation measure, a separate term for sound mitigation of
the cooling system has to be added to the equation above, similar to the no-load and load terms.
Point 5: The sound level reduction of sound mitigation elements is frequency dependent. For typical sound
mitigation solutions, as a rule of thumb it can be said for the frequencies of interest: the higher the frequency the
higher the sound reduction. This means that for no-load, load and if applicable cooling system sound, different
levels of reduction have to be used. Some guidance on this is given in chapter 6 and in APPENDIX C. The exact
levels of sound reduction however can only be found in close cooperation between transformer manufacturer and
supplier of the sound mitigating measure.
The auto-transformer discussed in example 2 is used hereafter as use case for external sound mitigation. While for
the no-load and load components the “Average” sound power levels are used as a starting point, the “Alternative”
sound power level is applied for the cooling system sound level – values as outlined in Table 22 above:
Rated service No-load Load Cooling system ODAF
Sound power level 𝐿𝑤 [dB(A)] 87 89 88
Sound mitigation is applied by a sound enclosure around the transformer, i.e. covering tank walls and tank cover.
The following (practically realistic) reduction for no-load and for load sound is assumed for this sound enclosure:
Sound enclosure No-load Load Cooling system ODAF
Sound reduction 𝐿 [dB(A)] 10 8 0 (outside sound mitigation)

90
 TB 940 – Power Transformer Audible Sound Requirements

The total sound level of this transformer at top rating without sound enclosure:
𝟖𝟕−𝟎 𝟖𝟗−𝟎 𝟖𝟖
[ ] [ ] [ ]
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟗𝟐. 𝟖𝒅𝑩(𝑨)

The total sound level of this transformer at top rating with sound enclosure:
𝟖𝟕−𝟏𝟎 𝟖𝟗−𝟖 𝟖𝟖
[ ] [ ] [ ]
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟖𝟗. 𝟏𝒅𝑩(𝑨)

A reduction in the total sound level of only 3 dB is achieved, although the sound enclosure provides 8-10 dB
reduction. The reason for this is the cooling system sound level remaining dominant, although it was chosen to be
the lowest possible value according to Figure 72. An option to avoid the dominance of the cooling system sound
could be to change the type of cooling system to ODAN. This would result in a larger footprint for the cooling system
however, the sound level of the cooling system will be drastically reduced. Here selected from Figure 72 was
74 dB(A) for the cooling system.
Rated service No-load Load Cooling system ODAN
Sound power level 𝐿𝑤 [dB(A)] 87 89 74

The total sound level of this transformer at top rating without sound enclosure:
𝟖𝟕−𝟎 𝟖𝟗−𝟎 𝟕𝟒
[ ] [ ] [ ]
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟗𝟏. 𝟐𝒅𝑩(𝑨)

The total sound level of this transformer at top rating with sound enclosure:
𝟖𝟕−𝟏𝟎 𝟖𝟗−𝟖 𝟕𝟒
[ ] [ ] [ ]
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 ) = 𝟖𝟑. 𝟎𝒅𝑩(𝑨)

The use case shows at first the importance to carefully evaluate the contribution of each individual sound component
(no-load, load and cooling system) to the total sound level. The question to ask is always “Which sound component
will dominate the total sound level?”. By reducing that component, the total sound level of the transformer will drop
by the largest amount. By following this principle consequently, it was possible here to realistically reduce the total
sound power level of example transformer 2 as a possible specification parameter from originally 98 dB(A) down to
83 dB(A). The technical and economic consequences were however not discussed here but are also significant
(larger footprint for cooling system, full sound enclosure for tank).

5.7 Sound Levels and Transformer purchasing

There are generally two alternatives when purchasing transformers:
▪ Bulk purchase (framework contract)
▪ Special (one-off) purchase

5.7.1 Bulk purchase

A large number of transformers are purchased under a bulk purchase contract, consequently the sound power level
of a transformer cannot be designed for a specific site. As such a general sound power level limit is applied to the
bulk purchase contract. The noise impact of installing a transformer from the bulk purchase on a specific site can
then be predicted using the maximum sound power level outlined in the contract. A determination can then be made
as to whether sound mitigation is required to meet any noise limits laid down in the country of operation. Purchasing
in bulk via a framework contract typically results in a lower unit cost.

5.7.2 Special purchase

A single transformer is procured through a special purchase or ‘one-off’ purchase, this allows the maximum sound
power level of the transformer to be designed specifically for the site in question. The need for noise mitigation may
therefore be avoided. However, special purchase transformers are typically significantly more expensive than
buying transformers under a bulk contract. Consequently, whilst the cost of installing sound mitigation may be
avoided, the cost of the transformer may outweigh the cost of any sound mitigation. Special purchase transformers
are typically installed in very noise sensitive areas and where very strict noise limits are in place that cannot be met
using a bulk purchase transformer.

5.8 Tender process

There is no general rule for the method of specifying maximum sound power levels in tender documentations. Clear
commercial conditions should be described in the tender documentation, which should include details of the sound

91
 TB 940 – Power Transformer Audible Sound Requirements

evaluation process with the sound power limit(s) clearly defined, along with the specific test
specification/parameters for the factory acceptance test.
The tender documentation should include options for different scenarios, penalties or negotiation, in case the
maximum allowable sound power level is exceeded.
In some cases, utilities specify the right to reject the transformer if the sound power level exceeds the maximum
allowable level outlined in the tender documentation. As an alternative, utilities should consider using clauses
specifying that the manufacturer shall be liable for the costs of mitigation, should the sound power level(s) of the
transformer exceed the level(s) outlined in the tender documentation. In very complex situations, where mitigation
is not possible, for example due to space constraints, or where even with mitigation, noise limits are exceeded,
other penalties or the right to rejection shall still apply. The right to rejection shall be a last resort where there are
no other feasible alternatives and noise limits will be exceeded.

5.9 On-site noise measurements after transformer installation

The noise restrictions outlined in legislation, regulations and policy dictate whether sound level measurements must
be undertaken before, after or both before and after installation of the transformer. It is however considered best
practice to undertake measurements both before and after installation of a transformer.
Undertaking sound level measurements before installation of a transformer provides an understanding of the
existing noise levels from a site and is considered best practice. This information may then be used to inform a
noise assessment to enable the maximum allowable sound power level of a transformer to be determined.
It is also considered best practice to undertake measurements after the installation and energisation of the
transformer to ensure noise limits are being complied with, to understand the change in noise and resulting impact.
Utilities may be required to do this to demonstrate to relevant authorities and enforcement bodies that noise limits
are being complied with.
Utilities may also choose to undertake sound level measurements on site to confirm the accuracy of the sound
power level determined by the manufacturer during the Factory Acceptance Test. Utilities must exercise caution in
these situations, as sound power levels determined in a laboratory test environment are likely to differ from sound
power levels determined on site, where factors such as transformer loading, tap position, site layout, site
characteristics and the influence of other noise sources may affect the measurements.
When determining the sound power level of a transformer on site, it is best practice to undertake sound intensity
measurements. The sound intensity method is, within certain limits, insensitive to steady-state background noise
and reflections. This helps to reduce the influence of noise from other noise sources in the surrounding area on the
measurement of interest. It is therefore a more accurate measurement technique. The method for determining the
sound power level of a transformer using sound intensity measurements is described in detail in IEC 60076-10-1
[B9].

92
 TB 940 – Power Transformer Audible Sound Requirements

6. Sound mitigation techniques

Max Gillet (FR)

6.1 Overview
The audible sound of power transformers is caused by three independent main components, whatever technologies
and designs are used, see section 2.2. The overall transformer sound level can be calculated as the logarithmic
sum of the three components because their acoustic fields are not coherent, see section 3.5. The sound dominating
component out of the three components no-load sound, load sound and cooling system sound varies from one
design to another and is driven by many different parameters, not necessarily linked to the parameters directly
influencing the sound components. As can be seen in Figure 37 and Figure 40, for smaller power transformers
typical no-load sound levels are higher than typical load sound levels, while the trend is inverse for large power
transformers. Cooling system sound is generally considered as being less restrictive than the aforementioned
components. This is because low sound or even zero sound emitting options exist (see section 2.2.3 and 3.4) and
can be selected, if necessary.
For transformer manufacturers and operators, it is essential to have sound mitigation solutions acting on each
sound component separately available. Mitigation solutions can be classified in three groups:
▪ mitigation at the source, acting directly on the vibration/sound generation process,
▪ mitigation at the transfer path between vibration/sound source and adjacent environment by modification
of the acoustic impedance,
▪ mitigation at the radiation surface by blocking and/or absorbing energy of the sound waves.
In the following sections 6.2 and 6.3, information on mitigation solutions at the source and on the transfer path will
be discussed, while solutions belonging to the barrier family will be outlined in section 6.4. It is important to note
that the application of sound mitigation techniques has almost always a certain thermal impact to the transformer.
Due to this, sound mitigation techniques and external cooling system design must be addressed at the same time.

6.2 In tank solutions

As mentioned previously, transformer manufacturers have in their sound control toolkit a set of solutions applicable
to mitigate vibration/sound at the source and on the transfer path between source and radiating structure. It is
important to note that the potential mitigation ranges presented in Table 23 refer to the typical average curves for
transformer sound power levels for no-load and load sound as derived and presented in sections 3.2 to 3.4. In case
of low-cost designs, i.e. transformer sound power levels above the typical average curves used as reference for
sound mitigation, the performance achievable with in-tank solutions will be somewhat increased, particularly for no-
load sound. Table 23 summarizes the conventional in-tank sound mitigation techniques that can be applied for all
types of liquid-immersed power transformers.

93
 TB 940 – Power Transformer Audible Sound Requirements

Table 23 – In-tank sound mitigation solutions

Potential Potential
Factory range for range for
Mitigation installed no-load load Impact on
Advantages Drawbacks
technique / Site sound sound losses
installed mitigation mitigation
(dB(A)) (dB(A))
Increase of cost,
Reduction of core total mass, and
footprint Decrease
flux density and/or
FI [1-7] 0 of no-load
increase of core Potential impact loss
mass to
transportation
Use of low-noise
Very limited Possible
electrical steel
impact on Potential increase of
instead of FI [1-3] 0
transformer increase of cost no-load
conventional Hi-B
design loss
grades
Active part – tank Efficient
Precautions to
mechanical mainly for
FI [1-4] [0-1] be taken for None
decoupling [B30], no-load
transport
[B31] sound level
no-load
Impact to mass,
Allows load and load
Optimization of dimensions and
FI 0 [1-3] sound loss may
winding design therewith to
reduction increase or
cost
decrease

All mentioned techniques can be applied at the same time and their respective no-load and load sound level
reduction performances can be added as they are independent. Table 23 illustrates manufacturers’ possibilities and
limitations to mitigate sound levels by design. It is however also important to note that sound mitigation solutions
directly implemented at the source are relatively expensive, except the active part-tank decoupling. The presented
solutions may therefore not be the best options to reduce power transformers’ sound emissions at optimized costs.

6.3 Cooling system solutions

As outlined in section 2.2.3, several physical phenomena affect the emitted sound by fans, pumps and coolers.
Specifically fans can be designed for a wide range of sound power levels by adjustment of the influencing
parameters such as rotational speed, fan diameter, aerodynamic blade design, distance to obstructions,
arrangement relative to tank and radiator etc. Table 24 summarizes the findings from section 2.2.3 and presents
sound level mitigation figures in reference to cooling system solutions.

94
 TB 940 – Power Transformer Audible Sound Requirements

Table 24 – Cooling system sound mitigation solutions

Potential range
Factory
for cooling
installed
Mitigation technique system sound Advantages Drawbacks
/ Site
mitigation
installed
(dB(A))
● Reach lower losses
Use low noise ● Allows to reach ● Higher costs
equipment instead of FI [1-10] sound levels below the
standard devices ● Increase cooler size
transformer tank
sound level
● Allows to reach
lower losses

Use of EC technology ● Same performance

instead of AC for 50 Hz and 60Hz Effective mainly for partial
technology for FI [0-22] application loading (not relevant for
continuous speed ● Increase lifetime of GSU transformers)
adjustment [B16] fan motors
● Reduced
maintenance needed
● Increased cost and
ONAN cooling mode Noise footprint
instead of FI completely ● Not regularly applicable
ONAF/OFAF/ODAF suppressed for large power transformer
applications
● Higher effort for
Maintenance
Sound enclosure for ● Can be used as
FI / SI [1-5] ● Practically useful for
pumps retrofit action
OFAN and ODAN cooling
modes only
● Allows to reach very ● Higher effort for
low sound levels when Maintenance
combined with low-
Silencers for fans FI / SI [1-3] ● Possible adverse impact
noise fans
to cooling performance in
● Can be used as case of insufficient free
retrofit solution space for air suction

6.4 Design of sound barriers

6.4.1 Physical phenomena
Sound barriers are a well-known solution to mitigate the sound produced by power transformers. When correctly
designed, significant sound mitigation performance can be achieved, up to 40 dB(A) overall reduction is possible
for very sophisticated sound enclosures. The relatively high cost, the impact on the transformer cooling
performance, higher maintenance effort as well as potential modifications of civil work make the use of sound barrier
(enclosure) solutions not the preferred option for the industry. When extreme low sound levels are required however,
for example due to a transformer implementation in an urban area substation, sound barrier solutions offer an
efficient possibility to reach the required sound levels. Before describing the different possible arrangements, the
fundamental physical phenomena of sound barriers are introduced in order to quantify the characteristic parameter
“Insertion Loss” (IL). IL is defined as sound level difference at a specified location with and without insertion of a
sound control device in the sound path between source and specified location. The insertion loss is the final
parameter describing the effectiveness of a sound barrier solution by integrating all contributing and adverse effects.
It is generally presented in octave or 1/3-octave bands. The estimation of the insertion loss is complex and considers

95
 TB 940 – Power Transformer Audible Sound Requirements

the barrier’s transmission loss, damping (absorption) effects, coincidence frequencies, acoustic leaks, diffraction,
mechanical coupling and resonance phenomena.
The characteristic property of a sound barrier is its transmission loss (TL), also called sound reduction index, and
is the measure for its airborne sound isolating capability. It is physically defined by the ratio of transmitted versus
incident sound intensity. TL can be approximated by the general mass-law equation:
2
𝐾
𝐶𝐵 2 𝑀𝐵 ∙ 𝜔 − ( 𝐵⁄𝜔)
𝑇𝐿 = 10 ∙ 𝑙𝑜𝑔 {(1 + ∙ 𝑐𝑜𝑠𝜃) + ( ∙ 𝑐𝑜𝑠𝜃) }
2𝜌0 ∙ 𝑐 2𝜌0 ∙ 𝑐

Equation 31
where
CB Viscous damping of barrier
ρ0 Air density (1.21 kg/m3 at 20°C)
c Sound speed in air (343 m/s at 20°C)
θ Angle of wave incident to panel surface (oblique-incident wave)
ω Angular frequency of the incident wave
MB Per surface mass of barrier
KB Per surface stiffness of barrier
From Equation 31 it can be demonstrated that per surface stiffness KB, viscous damping CB and per surface mass
MB mainly determine the barrier’s TL performance. Assuming that transformers’ typical dominant frequencies are
at twice the power frequency and its first harmonics – namely 100 Hz to 600 Hz frequencies, and assuming that the
barriers are typically implemented close to the radiating object and the incident waves are travelling mainly in normal
direction to the barrier surface, the mass of the barrier can be considered as the key parameter to control the TL.
A further phenomenon impacting the TL performance of barrier solutions is the possible appearance of resonances.
When a barrier’s main vibration modes coincide with the incident wave frequencies, amplification phenomena can
strongly affect the TL. The complete avoidance of resonances for transformer sound barrier applications is difficult
to achieve due to a number of waves with different frequencies being typically emitted by the transformer. The
resonance impact however can be reduced by using damped barrier structures.
It is important to mention that the sound barrier design must be such to avoid the barrier’s coincidence frequency
being at one of the transformer sound frequencies that it is intended to mitigate. The barrier’s coincidence frequency
is the frequency at which the barrier’s bending wavelength matches the wavelength of the airborne sound radiated
by the transformer. In case the coincidence frequency is excited, the barrier becomes acoustically almost
transparent and the sound energy stored in the space covered by the barrier is passed through without sound
attenuation.
When closed barrier solutions surround transformers, an important part of the acoustic energy radiated by the tank
remains confined between tank and barriers, causing multiple reflections of acoustic waves on tank walls and the
barrier’s inner side. The sound level in the air gap is consequently amplified by the reverberated field, leading to a
global IL reduction. The reverberated acoustic power WR and the corresponding reverberated sound pressure pR
inside the air gap can be approximated, using Equation 32 and Equation 33:

(pR 2 ) WR
=
𝑊𝑅 = 𝑊 ∙ (1 − 𝛼̅) ρ0∙ c A
Equation 32 Equation 33
where
W Acoustic power radiated by tank walls
𝛼̅ Mean barrier system absorption coefficient
A Equivalent absorption area (with A = S·𝛼̅)
In accordance with Equation 32, it can be deduced that materials with high absorption implemented inside the air
gap limit the sound increase due to the reverberated field, or in the best case cancel it for 𝛼̅ = 1. For typical
transformer applications, depending on the barrier design, the absorption technique with its corresponding
absorption coefficient and the transformer sound excitation spectrum, the reverberated acoustic field increases the
overall sound power to mitigate by the barrier between 2 and 10 dB.
One further factor limiting the sound barriers’ or enclosures’ performance is the presence of acoustic leaks.
Openings in sound barriers can be required to allow crossings for radiator connections, bushings, conservators,
busbars, headers, valves, access to manholes, control cabinets, etc. Crossing elements may introduce leaks,
leading to a significant decrease of the global IL. For instance, the theoretical maximum of sound attenuation that

96
 TB 940 – Power Transformer Audible Sound Requirements

can be achieved by a barrier solution with 5% of its surface opened, is limited to 16 dB, whatever the initial TL value
is. In order to design and manufacture very efficient sound barrier solutions, it is mandatory to eliminate all holes,
as small as they might be.
When sound panels or walls are implemented around transformers, sound diffraction occurs at any edge of these
elements, specifically at the top edges on four-sided barrier solutions. Diffraction is defined as the bending of sound
waves when they reach obstacles, it is frequency dependent. The associated acoustic diffracted field generates
noise behind the sound barrier in the “shadow” zone, which is added to the noise transmitted through the barrier.
As low frequencies are dominant in power transformers’ sound spectra, the diffraction phenomenon should be
addressed with attention. Practical experience shows that the total insertion loss can’t be higher than 15 dB when
diffracted fields are present. The solution to minimize or even eliminate this contribution is to use a full sound
enclosure, or at least to shape and arrange the barriers appropriately.
Another critical aspect of efficient sound barrier design is the mechanical coupling between source and screening
element, in other words the transmission of vibration energy from the tank to the barriers through structural
elements. If the decoupling is weak, the sound reduction obtained for the airborne sound previously discussed can
be short-circuited and the barriers behave as a loudspeaker, amplifying the sound waves instead of acting as sound
isolation device. This is the reason why sound panels fixed rigidly on tank walls have a relatively poor efficiency,
regardless of the initial barrier’s transmission loss performance. In order to limit the influence of structural
transmission of vibrations, it is mandatory to ensure that a significant mechanical impedance mismatch exists
between transformer and barrier. This can be realized for instance by using resilient mounts when barriers are
attached to the transformer, or even better by implementing barriers fully separated from the transformer and even
decoupled from the transformer foundation.
Finally, in case the transformer to barrier spacing allows personnel to access, the presence of standing waves shall
be avoided by controlling the length of the airgap that shall not correspond to the wavelengths of the transformers’
main sound frequencies. Theoretically, standing waves are not supposed to propagate sound energy beyond
barriers, but it is nevertheless preferable to avoid them, which shall also protect maintenance operators during
transformer inspections.

6.4.2 Determination of total insertion loss

The total insertion loss of a sound barrier solution can be estimated by integrating all previously introduced
phenomena. In this section, the impact of the individual effects applicable to sound barrier applications for typical
power transformers will be outlined on two examples. Actually, the overall sound reduction performance can only
be determined by considering the sound level spectrum of the source (transformer) together with the barrier’s
performance at the individual frequencies. Knowledge of the transformer sound level spectrum is crucial, therefore,
for the selection and the design of the sound barrier / enclosure. Otherwise there is risk to undersize or oversize
the sound mitigation solution.
Typical no-load sound and load sound spectra for the first six harmonics of a 50 Hz network transformer (namely
100, 200, 300, 400, 500 and 600 Hz) are defined and used as input in Table 25 and Table 26 for the evaluation of
the two examples of sound barrier solutions. The selection of the lower frequencies up to 600 Hz only is made
because such are the main sound contributors of power transformers and determine the design of the barrier
solution. Sound emissions beyond 600 Hz for typical power transformers operated with sinusoidal quantities and
standard conditions are negligible for the overall transformer sound power level. Moreover, sound barrier solutions
become naturally more efficient for higher frequencies and such frequency components are therefore sufficiently
mitigated naturally.
Table 25 illustrates the sound mitigation for the individual harmonics plus the overall level achieved by a sound
panel system covering all 4 sides of a transformer. In this example, the panels are made from heavy material, are
resiliently suspended from the transformer tank and are equipped with absorbing material on the inside. The panel
height exceeds the tank height by 0.2 m and the presence of some sound leaks for radiator connections is assumed.

97
 TB 940 – Power Transformer Audible Sound Requirements

Table 25 – Sound panels system performance and associated sound level mitigation

Harmonic sound power levels (dB(A)) Overall

sound
4-sided sound barriers resiliently
power
tank-mounted
100 Hz 200 Hz 300 Hz 400 Hz 500 Hz 600 Hz level
(dB(A))
Transformer initial no-load
62 70 66 63 57 62 73
Transformer sound level

sound power level

Transformer initial load sound
70 54 47 44 45 43 70
power level
Overall transformer sound
power level (no-load + load)
71 70 66 63 57 62 75
before application of sound
barrier
Barrier’s transmission loss 27 33 37 43 47 53 -
Sound barrier characteristics

Barrier’s resonance -6 -5 -4 -3 -3 -2 -
Reverberation field -7 -4 -3 -2 -2 -2 -
Sound leaks -1 -6 -11 -12 -15 -17 -
Mechanical coupling -5 -4 -4 -3 -2 -2 -
Transmitted acoustic loss 8 14 15 23 25 30 -
Diffracted acoustic loss 9 11 12 13 14 15 -
Total insertion loss IL 5 9 10 13 14 15 -
Results

Overall sound power level after

66 61 56 50 43 47 68
installation of sound barrier

As outlined in the previous section, the transmitted acoustic loss of sound waves passing through the barrier is
determined by summing up the individual contributions of the barrier’s main characteristics for each frequency
individually. Then the total insertion loss IL is calculated for each frequency by combining the transmitted acoustic
loss and the acoustic loss while the barriers are acting as pure diffraction elements (diffracted acoustic loss). At the
end, the overall sound power levels for each frequency are obtained by subtracting the total insertion loss from the
initial overall transformer sound power level. Summing up such levels logarithmically gives the overall sound power
level after barrier installation.
Table 26 illustrates the sound mitigation for the individual harmonics plus the overall levels achieved by a sound
enclosure system manufactured with the same sound panels as in the first example (Table 25), i.e. the barriers
have identical transmission loss and absorption coefficients. However, the barriers in this example are self-
supported and installed on a concrete platform about 2 m away from the transformer (for worker access) in a
mechanically decoupled manner from ground. The transformer is further totally enclosed with sound barriers (sides
and top) and the enclosure is assumed to have no sound leakages and no diffraction.

98
 TB 940 – Power Transformer Audible Sound Requirements

Table 26 – Sound enclosure system performance and associated sound level mitigation

Harmonic sound power levels (dB(A)) Overall

sound
Walk-in sound enclosure made of
power
modular sound panels, including roof
100 Hz 200 Hz 300 Hz 400 Hz 500 Hz 600 Hz level
(dB(A))
Transformer initial no-load
62 70 66 63 57 62 73
Transformer sound level

sound power level

Transformer initial load sound
70 54 47 44 45 43 70
power level
Overall transformer sound
power level (no-load + load)
71 70 66 63 57 62 75
before application of sound
enclosure
Barrier’s transmission loss 27 33 37 43 47 53 -
Sound barrier characteristics

Barriers’ resonance -2 -2 -1 -1 -1 -1 -
Reverberation field -6 -3 -2 -2 -1 -1 -
Sound leaks 0 0 0 0 0 0 -
Mechanical coupling -1 -1 0 0 0 0 -
Transmitted acoustic loss 18 27 34 40 45 51 -
Diffracted acoustic loss - - - - - - -
Total insertion loss IL 18 27 34 40 45 51 -
Overall transformer sound
Results

power level after installation of 53 43 32 23 12 11 53

sound enclosure

All figures presented in Table 25 and Table 26 are hypothetical and aim to describe typical sound barrier and sound
enclosure performances using calculation methods available in literature. They enable the comparison of the
contributions by the individual phenomena for the two example applications with the same sound source. It is shown
that for the same sound barrier type but using different mounting technologies the overall sound reduction widely
varies, namely from 7 dB(A) for sound barriers attached to the tank walls to 22 dB(A) for a full enclosure. The
examples emphasize that a reliable design of a sound barrier system requires:
▪ knowledge of the transformer sound power levels (or sound pressure levels at a given distance) in terms
of sound spectra,
▪ knowledge of the sound barrier’s transmission loss value for the individual source frequencies,
▪ consideration of all effects listed in the two previous tables for the individual source frequencies.

6.5 Factory-installed sound barrier solutions

Sound barrier mitigation technologies applied to power transformers can be split in 2 categories: tank-mounted and
ground-supported solutions. Sound barrier systems attached directly on tank walls and/or the transformer cover
that are not connected with / supported by the transformer foundation exhibit many different variants. The
differences are related to the fixation type and rigidity, location of the panels, treatment of the tank cover, presence
or absence of airgaps between transformer and barrier etc. The main types of tank-mounted barrier techniques are
introduced and presented in Table 27.

99
 TB 940 – Power Transformer Audible Sound Requirements

Table 27 – Transformers-installed sound barrier solutions

Potential Potential
range for range for
Mitigation no-load load
Advantages Drawbacks Sketch
technique sound sound
mitigatio mitigatio
n (dB(A)) n (dB(A))
● Limited
efficiency
● Prevents
tank
inspections
and certain
maintenance
operations
● Decrease of
thermal tank
dissipation
● Ageing
(possibly
faster than
the
transformer)
Up to Up to ● Acting on
5 dB(A) 3 dB(A) tank sound
for walls for walls ● No radiation
only only impact on only. May
footprint require
Sound
barriers ● Minor separate
Up to Up to
rigidly tank impact on sound control
7 dB(A) 5 dB(A)
mounted design of cooling
for walls for walls
[B32] equipment.
and and ● No
partial partial impact on ● Acoustic
tank tank civil work performance
cover cover may be
treatment treatment reduced in far
field if only
tank walls are
treated.
Special
agreement
required for
sound
measurement
between
manufacturer
and purchaser
because IEC
sound
standards
exclude this
solution (see
IEC 60076-10-
1:2016, §6.5).

100
 TB 940 – Power Transformer Audible Sound Requirements

● Prevents
tank
inspections
and certain
maintenance
operations
● Requires
specific
adaptations
for radiators
● Ageing
(possibly
faster than
the
transformer)
● Acting on
Up to Up to ● Limited
tank sound
7 dB(A) 5 dB(A) impact on
radiation
for walls for walls footprint
only. May
only only ● Cost- require
Sound effective separate
barriers Up to Up to relatively to sound control
well 9 dB(A) 7 dB(A) potential of cooling
decoupled for walls for walls range of equipment.
from tank and and sound
● Acoustic
partial partial mitigation
performance
tank tank ● No may be
cover cover impact on reduced in far
treatment treatment civil work field if only
tank walls are
treated.
Special
agreement
required for
sound
measurement
between
manufacturer
and purchaser
because IEC
sound
standards
exclude this
solution (see
IEC 60076-10-
1:2016, §6.5).

In addition to the common tank-installed sound barrier solutions described in Table 27, further techniques that are
presented in Table 28 are frequently applied for tank-mounted sound mitigation. These methods are used when a
lower performance of sound level reduction is required. They are part of the transformer mechanical design because
many adaptations are necessary for radiator mounting, control cabinets etc. The lifetime of such solutions is
expected to be comparable to the transformer’s lifetime.

101
 TB 940 – Power Transformer Audible Sound Requirements

Table 28 – Mitigation techniques included in tank design

Potential Potential
range for range for
Mitigation no-load load
Advantages Drawbacks Sketch
technique sound sound
mitigatio mitigatio
n (dB(A)) n (dB(A))
● No
● Sound
impact on
reduction
transforme
limited
r active
Filling part design ● Acting on
hollow nor tank sound
stiffeners [0-2] [0-1] footprint radiation only.
(U-profile) and civil May require
with sand work separate
sound control
● Very
of cooling
limited cost
equipment.
increase
● Acting on
tank sound
radiation only.
● No May require
impact on separate
Sound transforme sound control
panels r active of cooling
fixed [1-4] [1-3] part design equipment
between nor ● Prevents
stiffeners footprint certain
and civil maintenance
work operations,
particularly
detection of
oil leaks.
● Complex
mechanical
design
● Special
precautions
for
transportation
may apply
Double ● Effective ● Prevents
tank [1-6] [1-4] sound certain
design reduction maintenance
operations,
particularly
detection of
oil leaks.
● Increases
transformer
weight and
footprint

102
 TB 940 – Power Transformer Audible Sound Requirements

Techniques outlined in Table 28 are generally factory-installed, although final assembly may be partly or completely
carried out on-site due to transport constraints. Tank-mounted solutions are advantageous for installation sites with
footprint constraints. They are also interesting because they do not require specific site arrangements for radiators,
oil-air coolers or any other cooling means.
In conclusion, tank-mounted sound mitigation solutions are suitable and cost-effective when long-term overall noise
reduction below 10 dB is sought.

6.6 Site-installed sound barrier solutions

The second category of sound barrier solutions comprises site-installed mitigation systems, usually ground-
supported and not mounted to the transformer. As for tank-mounted techniques, this group of solutions also allows
many different variants to apply. In Table 29, the five most typical applications are discussed.

103
 TB 940 – Power Transformer Audible Sound Requirements

Table 29 – Site-installed sound mitigation solutions

Potential Potential
range for range for
Mitigation no-load load
Advantages Drawbacks Sketch
technique sound sound
mitigation mitigation
(dB(A)) (dB(A))
● Not
reliable for 4
● Used when sided sound
sound reduction
mitigation is
required in ● Sound
certain emission
directions reinforced in
Firewalls / (maximum 3) not treated
free standing directions
● Combines
sound Up to Up to ● Efficiency
sound
barriers / 8 dB(A) 6 dB(A) limited due
mitigation
earthworks / to diffraction
in in with fire
embankments at barrier
screened screened protection
on 1, 2 or 3 directions directions top (works
● Does not well when
sides of the
require barrier
transformer
transformer height >>
design transformer
modifications height)
● Applicable ● Significant
as retrofit civil work
solution required
● High cost

● Can be ● Aging of
applied as a blanket
corrective faster than
solution for for
short- and transformer
Up to Up to medium- ● Acting on
7 dB(A) 5 dB(A) term tank sound
for walls for walls application radiation
only only (up to 20 only. May
Sound years) require
blankets [B1] separate
Up to Up to ● Applicable
as cost- sound
15 dB(A) 12 dB(A)
effective control of
for walls for walls
retrofit cooling
and top and top
solution equipment

● Easy to ● Not
implement, compatible
no impact on with
civil work automatic
fire

104
 TB 940 – Power Transformer Audible Sound Requirements

extinguishing
systems
● Requires
removal for
tank
inspections
● Decrease
tank thermal
dissipation
● Ageing
(possibly
faster than
the
transformer)
● Acting on
tank sound
radiation
only. May
require
separate
● Efficient
Up to Up to sound
solution
10 dB(A) 8 dB(A) control of
for walls for walls ● Limited cooling
Closely fitting only only footprint equipment
self- increase
● Requires
supported
● Does not some civil
enclosure Up to Up to
require work
13 dB(A) 9 dB(A)
transformer
for walls for walls ● Prevents
design
and top and top tank
modifications
inspection
and certain
maintenance
operations
● Not
compatible
with
automatic
fire
extinguishing
systems
● Increase of
Walk-in footprint
sound ● Efficient
solution ● Requires
enclosure
Up to Up to management
made of ● Allows
25 dB(A) 20 dB(A) of cooling
modular transformer equipment
sound panels inspection
including roof ● Requires
civil work

105
 TB 940 – Power Transformer Audible Sound Requirements

● High cost
● Very ● Increase of
efficient footprint
Walk-in
solution
sound ● Requires
enclosure Up to Up to ● Allows management
(made from 40 dB(A) 35 dB(A) transformer of cooling
bricks / inspection equipment
concrete)
● Life-long ● Requires
solution significant
civils work

Site-installed sound mitigation solutions belong generally in the transformer user’s scope of responsibility, though
it can be supplied by the transformer manufacturer. Other than for the sound blanket solution, significant site
modifications are required, such as civil work, transformer terminal (busbars, bushings) adaptations and a reliable
installation of the transformer cooling system.
Firewall installations must be made such to avoid adverse effects to the cooling performance. One important aspect
amongst others is to avoid hot air recirculation. The installation of updraft louvres on fans may be considered to
control the hot air stream. Although radiators and coolers may be installed inside the enclosure, such solutions
require high effort for the enclosure ventilation system including the use of silencers. Whether to install the cooling
system inside or outside the enclosure needs to be studied case by case to find the most suitable overall solution
in terms of sound emission, mechanical arrangement and transformer access for maintenance. For all enclosure-
type solutions, it is recommended to place the power transformer on properly designed decoupling pads, so as to
avoid transmission of transformer vibrations through the foundation to the enclosure and lose sound mitigation
efficiency.
In conclusion, site sound barrier solutions are suitable in following cases:
▪ Mitigation of sound required in preferential direction(s) by installation of firewalls, earthworks,
embankments, sound barriers away from the transformer
▪ The sound mitigation objective is to reduce the sound level by more than 10 dB(A), preferably 15 dB(A)
and have at the same time a maintenance friendly solution with full access to the transformer (separate
sound enclosure).
▪ Mitigation of sound is required by an easy to apply retrofit solution, not necessarily lasting for the
transformer lifetime (sound blankets)

6.7 Sound mitigation technologies not in regular use

For more than 50 years, power transformer sound mitigation solutions have been developed by manufacturers as
well as utilities to protect people living in the vicinity of substations from annoying sound emissions. Among these
technologies, there are some that have been abandoned from regular use due to inefficiency or efficiency only
under well-defined service conditions, limited robustness against harsh environmental conditions and also due to
high costs. Table 30 gives a short list of sound mitigation technologies that have been used in the past or are
currently only rarely in use by the industry.

106
 TB 940 – Power Transformer Audible Sound Requirements

Table 30 – List of sound mitigation technologies not in regular use

Factory
Mitigation installed Acting
Advantages Drawbacks Remarks
technique / Site on
installed
● Limited sound reduction
Used for
as only locally effective Mainly
Tuned shunt
● Easy to implement and to load
vibration ● Efficiency temperature reactor
dismount sound if
absorbers dependent applicatio
tuned
(spring- FI ● Can be efficient in case of ns (due to
● Allows sound reduction for
mass a particular dominant higher
for only one frequency 100 Hz
damper frequency levels of
(tuned frequency), poor or
systems) tank
efficiency for no-load 120 Hz
vibration)
sound
Masses
● Easy to implement
attached ● Limited sound reduction
on tank ● Effective in case of tank particularly when main Tank
Rarely
walls for FI resonance vibration modes are well resonan
used
detuning below excitation -ces
● Can be used as retrofit
tank frequencies
solution
resonances
● No proven sound
Tank
reduction and may even Not used
thickness FI -
increase the sound at all
increase
emission
● Effective sound
reduction, particular at low ● Significant cost
frequencies ● Inefficient to mitigate
● Can be implemented as high frequency sound
Active A few
retrofit solution (above 500 Hz)
vibration No-load pilot
and sound SI ● Global or directional ● Sensitive electronic and load applica-
cancellation action involved sound tions in
technologies use only
● Self-adapting to ● Vulnerable for
transformer sound environmental impacts
variations (overall and for ● Potential ageing issues
spectra)

107
 TB 940 – Power Transformer Audible Sound Requirements

7. Summary and further work

Christoph Ploetner (DE) and Emanuel Almeida (PT)

Between years 2016 and 2023, CIGRE WG A2.54 ”Power transformer audible sound requirements” studied aspects
required for the specification of realistic sound power levels for new power transformer purchases. The executed
studies are all based on common knowledge existing throughout the transformer industry, relevant literature and
on WG established databases of FAT measured sound power levels between years 2008 and 2018. Specific
proprietary knowledge of manufacturers to control the sound level performance by design was entirely excluded as
resource for WG activities. Most of the results are provided as graphical presentations clearly arranged and directly
applicable in practice. While mathematical formulations are purposely kept at a minimum, examples and case
studies are used to support an easy application of the results.
The main content/outcome of the individual brochure chapters is summarized as follows:
Chapter 1 – Introduction
The motivation and history for the WG setup are outlined and the scope of work including transformer types and
application range considered, are defined. Explanation und usage of the terms ‘Noise’ and ‘Sound’ is given and
also the generic meaning of the term ‘sound level’ explained.
Chapter 2 – Background knowledge
After having briefly outlined the very basics of the physics of sound, the fundamental sound development processes
of power transformers are described and the three independent measurable sound components ‘no-load sound’,
‘load sound’ and ‘cooling system sound’ are introduced. While many publications exist on the two first components,
this is not the case for the ‘cooling system sound’ and comprehensive information is therefore provided here.
Devices discussed are ‘air fans’ and ‘pumps for the cooling liquid’ but also all typical ‘cooling systems’ with
information presented on design aspects, possible arrangements, sound level expectations and measurement
specifics, as well as a few case studies. At the chapter’s end, an introduction to ‘sound level control’ is given,
considering the sound components individually. A discussion of other impacts to the sound level takes place.
Chapter 3 – Sound levels of liquid-immersed power transformers
Typical ranges of sound power levels of the three individual sound level components are introduced for three-phase
liquid-immersed power transformers with a rated power beyond 3 MVA. The derivation process of the figures and
their practical application is clearly described. The combination of the sound level components as required in
practice is outlined and a proposal made for an acoustic fingerprint figure of transformers – potentially to be included
in FAT reports. The unexpected finding of a not constant but dimensional dependency of the sound level difference
between 50 Hz and 60 Hz power transformers was explored and is explained. Finally, the sound level relation of
single-phase transformers and transformer banks with three-phase transformers is given.
Chapter 4 – Sound levels of other transformer types
Specific aspects of no-load sound as well as some rough information on assigned no-load sound levels are
discussed for dry-type transformers, gas-insulated transformers and transformers with cores made from amorphous
steel. First order findings on the sound level difference if compared with liquid-immersed transformers of the same
power rating are as follows:
• Dry-type transformers: in average about 10 dB(A) higher sound levels
• Gas-insulated transformers: sound levels are comparable
• Transformers with amorphous cores: in average about 14 dB(A) higher sound levels
Because of the smaller market share of such transformer types and limited knowledge spread on them within the
transformer community, some further information on the technologies and applications is also given.
Chapter 5 – Sound level specification and legislation
This chapter deals with all aspects in the sound level specification process that a transformer purchaser is assumed
to consider. Legislative requirements and utilities’ practices on transformer sound level specifications were studied
by a world-wide survey with a good return rate and gave valuable insights. Based on the survey results, on findings
and results from chapter 3 as well as on WG membership experience, best practice parameter for the sound level
specification of power transformers were identified and are described. The derivation process for the permissible
sound levels of new transformer purchases to be installed in a specific substation is outlined – including substation
design optimization with the target to find the overall cost minimum. In two case studies, the specification process
for realistic sound levels of new transformer purchases based on earlier derived figures of typical ranges of
transformer sound power levels is explained in detail and can be followed in practice. Some information on the
tendering process in terms of sound level specification as well as on sound level measurements in the substation
after new transformer installations is finally given.

108
 TB 940 – Power Transformer Audible Sound Requirements

Chapter 6 - Sound mitigation techniques

A comprehensive overview on possibilities to mitigate transformer sound levels is provided. Distinction is made
between in-tank solutions, solutions for the cooling system and sound barrier solutions. For sound barriers, the
physics behind the design is explained and then applied on two case studies – one on a 4-sided panel system and
one on a complete enclosure. Possibilities for factory installed tank-attached barrier solutions are provided but also
hints for site-installed barrier solutions not in contact with the tank are given.
Appendixes
Complementary information on specific associated aspects is provided in 6 appendixes.
This brochure intends to close the gap of missing information on transformer sound level specification aspects.
Beyond that, it provides information on topics related to transformer acoustics so far not yet fully understood and
clarified.
The brochure is also intended and can be used as a document for education on transformer acoustics. Together
with IEC 60076-10 / IEC 60076-10-1 it will serve as an excellent and comprehensive introduction into the field of
power transformer acoustics.
There are of course aspects remaining in the transformer acoustics area that require more clarification. One area
for further work is the acoustic performance of transformers when installed at site. Impacts of varying grid
parameters, foundation design, substation design, meteorological conditions/fluctuations… result in more or less
significant acoustic uncertainties. CIGRE TF A2.01 “Power transformer sound levels on site”, established in year
2022, has the task to look into those aspects. Another topic still requiring clarification is the acoustic impact of
harmonics applicable to transformers connected to modern voltage source converters.

109
 TB 940 – Power Transformer Audible Sound Requirements

APPENDIX A. Typical sound power level ranges of

liquid-immersed power transformers – Essential
information at a glance for everyday use
Bart Simons (NL), Christoph Ploetner (DE) and Frank Trautmann (DE)

The intention of this appendix is to provide at a glance the most essential information required for the specification
of transformer sound power levels. The information is condensed to three pages without this cover page. The first
two pages display the developed diagrams representing the typical ranges of transformer sound power level
components (no-load, load, cooling system) for 50 Hz and for 60 Hz liquid-immersed power transformers. The third
caption/instruction page provides guidelines for the usage of the diagrams and gives instruction for the derivation
of a transformer’s total sound power level from selected component sound levels. The relevant pages are intended
as reference for the everyday work.

110
 TB 940 – Power Transformer Audible Sound Requirements

Typical sound power level ranges for 3-phase 50 Hz transformers

no-load
load
cooling system

111
 TB 940 – Power Transformer Audible Sound Requirements

Typical sound power level ranges for 3-phase 60 Hz transformers

no-load
load
cooling system

112
 TB 940 – Power Transformer Audible Sound Requirements

Caption
Input on abscissa: Building power 𝑆𝑏
• For separate winding transformers: 𝑆𝑏 = 𝑆𝑟
𝑈𝐻𝑉 −𝑈𝐿𝑉
• For auto transformers: 𝑆𝑏 = 𝑆𝑟 ⋅
𝑈𝐻𝑉
no-load

𝑆𝑟 - rated power in MVA

𝑈𝑥𝑥 - line-to-line voltage in kV

• Sound level control is mainly achieved by variation of core induction – at a first

approach the three curves represent: top ~1.8 T, center ~1.6 T, bottom ~1.3 T
• Although the entire sound power level range can technically be achieved, the
sound level impacts first costs and may impact transportation for large units
Input on abscissa: Reactive power 𝑄
• 𝑄 = 𝑆𝑟 ⋅ 𝑢𝑘 = 𝑆𝑟 ⋅ 𝑧𝑘
𝑆𝑟 - rated power in MVA
𝑢𝑘 - short-circuit impedance voltage (short circuit impedance) in per unit

• Top curve: Typical natural upper limit curve. All transformers produced should be able to
comply with this upper limit without making use of sound mitigation measures. Low-cost
load

design for sound purposes.

• Center curve: Typical average curve. Specifying below the typical average curve results in a
high probability that the sound level cannot be met without special design provisions
and/or external sound mitigation measures. It is therefore recommended to only specify at
or above the typical average curve.
• Bottom curve: Typical natural lower limit curve. Almost all transformers expose a higher
than this lower sound level limit. Specifying at this limit will most likely lead to an increase
in costs and in most cases require external sound mitigation measures to be applied. High-
cost design for sound purposes.
Input on abscissa: Rated power 𝑆𝑟
cooling system

𝑆𝑟 - rated power in MVA

• Sound level predominantly determined by fans
• Provided sound level ranges technically always achievable
• Generic relation: The lower the sound level, the larger the cooling system with the
cooling system footprint typically getting larger.
• In case of ONAN: cooling system sound power level is 0 dB
total sound power level

𝑳𝒘 −𝑳 𝑳𝒘 −𝑳 𝑳𝒘
𝒏𝒐−𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒏𝒐−𝒍𝒐𝒂𝒅 𝒍𝒐𝒂𝒅 𝒎𝒊𝒕𝒊𝒈𝒂𝒕𝒊𝒐𝒏𝒍𝒐𝒂𝒅 𝒄𝒐𝒐𝒍𝒊𝒏𝒈
[ ] [ ] [ ]
𝟏𝟎
𝑳𝒘 = 𝟏𝟎 𝐥𝐨𝐠 𝟏𝟎 (𝟏𝟎 𝟏𝟎 + 𝟏𝟎 𝟏𝟎 + 𝟏𝟎 )

1. Determine building power of the transformer (see ‘no-load’ above)

2. Read no-load sound level from graph
3. If external sound mitigation applies, subtract reduction level for no-load
4. Determine reactive power (see ‘load’ above)
5. Read load sound level from graph
6. If external sound mitigation applies subtract reduction level for load
7. Determine type of transformer cooling system
8. Read cooling system sound level from graph (for ONAN, sound level is 0 dB)
9. Calculate total sound power level of the transformer with above equation

113
 TB 940 – Power Transformer Audible Sound Requirements

APPENDIX B. References
[B1] Rajotte, C. et al. “Transformer audible noise – Evolution of substation practices in a large utility”,
CIGRE Study Committee A2 Colloquium, Cracow, Poland 2017
[B2] NEMA No. TR1-2013 “Transformers, Step Voltage Regulators, Reactors”
[B3] VDI 3739 “Characteristic noise emission values of technical sound sources – Transformers”, 1999
[B4] Reiplinger, E. “Study of noise emitted by power transformers based on today’s viewpoint”, CIGRE
Main Session, Paris, 1988
[B5] Ploetner, C. “Sound levels of oil-immersed shunt reactors – Development, prediction, specification”,
CIGRE SC A2 Colloquium, Shanghai, September 2015
[B6] CIGRE WG A2.48 “Technology and utilization of oil-immersed shunt reactors”, CIGRE Technical
Brochure 655, May 2016
[B7] CIGRE WG 12.12 “Transformer noise: Determination of sound power level using the sound intensity
measurement method”, ELECTRA 144, October 1992
[B8] CIGRE JWG 12/14.10 “HVDC converter transformer noise considerations”, ELECTRA 167, August
1996
[B9] IEC 60076-10:2016 ”Power Transformers – Part 10: Determination of sound levels”
[B10] IEC 60076-10-1:2016 “Power Transformers – Part 10-1: Determination of sound levels - Application
Guide”
[B11] IEEE C57.12.90 “IEEE Standard Test Code for Liquid-Immersed Distribution, Power, and Regulating
Transformers”
[B12] Swiatkowski, M. et al. “Uncertainty of the determined transformer sound power level in respect to the
applied measurement conditions”, CIGRE Study Committee A2 Colloquium, Cracow, Poland, 2017
[B13] ISO 3745 “Determination of sound power levels and sound energy levels of noise sources using
sound pressure — Precision methods for anechoic rooms and hemi-anechoic rooms”
[B14] ANSI/AMCA 300-08 “Reverberant Room Method for Sound Testing of Fans”
[B15] IEC 60721 “Classification of environmental conditions - Part 3-4: Classification of groups of
environmental parameters and their severities - Stationary use at non-weather protected locations”
[B16] Goette, W. et al. “Design influence to the noise emission and power consumption of OFAF/ODAF
coolers for power transformers”, CIGRE Study Committee A2 Colloquium, Cracow, Poland, 2017
[B17] Ploetner, C. “No-load sound power levels for specification purposes derived from more than 1000
measurements – a representative figure for three-phase transformers”, CIGRE Study Committee A2
Colloquium, Cracow, Poland, 2017
[B18] Ploetner, C. “Power transformer audible sound requirements”, Interim Report WG A2.54, ELECTRA
302, February 2019
[B19] Simons, B. “Load sound power levels for specification purposes of three-phase 50 Hz and 60 Hz
liquid-filled power transformers”, Progress Report WG A2.54, ELECTRA 310, June 2020
[B20] Pirnat, M. “Difference between 50 Hz and 60 Hz transformer no-load noise levels”, ICTRAM,
October 2019, Opatija, Croatia
[B21] Pirnat, M.; Gillet, M. “Difference between 50 Hz and 60 Hz transformer load noise levels”, CIGRE SC
A2, B2 & D1 International Tutorials & Colloquium, November 2019, New Delhi, India
[B22] Fahy, F.; Gardonio, P "Sound and structural vibration: Radiation, Transmission and Response"
(Elsevier, 2007)
[B23] IEC 60529 “Degrees of protection provided by enclosures (IP CODE)”
[B24] NEMA Standard 250, “Enclosures for Electrical Equipment (1000 Volts Maximum)”
[B25] IEC 60076-11:2018 “Power Transformers – Part 11: Dry-type transformers”
[B26] Hsu, C.-H.; Chang, Y.-H. “Impacts of Fe-based amorphous HB1 core transformers on energy
efficiency and environment protection”, Proceedings of the 8th WSEAS International Conference on
Instrumentation, Measurement, Circuits and Systems, Hangzhou, China, May 2009.
[B27] Hitachi Metals – Metglas® Inc. “Amorphous Alloys for Transformer Cores”, Technical documentation
of Metglas® 2605SA1 and Metglas® 2605HB1M.

114
 TB 940 – Power Transformer Audible Sound Requirements

[B28] EN 50708 “Power transformers – Additional European requirements: Part 2-1 Medium power
transformer – General requirements”
[B29] EN 50708 “Power transformers – Additional European requirements: Part 2-6 Medium power
transformer – Non-conventional magnetic steel technology”
[B30] IEEE C57.136-2000 (R2005) “IEEE Guide for Sound Level Abatement and Determination for Liquid-
Immersed Power Transformers and Shunt Reactors rated over 500kVA”
[B31] Herrera, E. et al. “Simple and effective measures for noise reduction in power transformers”, 2012
CIGRE Canada Conference
[B32] Gillet, M. et al. “Upon the importance of mastering all physical phenomena for the design of efficient
sound barriers for electrical apparatuses”, CIGRE Study Committee A2 Colloquium, Cracow, Poland
2017
[B33] Gülich, J.F. “Kreiselpumpen: Handbuch für Entwicklung, Anlagenplanung und Betrieb“, Springer-
Verlag GmbH, 2014
[B34] Young F.R. “Cavitation”, Imperial College Press, London, 1999
[B35] Alhelfi, A.; Sunden, B. “The cavitation phenomenon: a literature survey, Advanced Computational
Methods and Experiments” in Heat Transfer XIII, p351-362, WIT Transactions on Engineering
Sciences, Vol 83, 2014
[B36] Binama, M. et al. ”Cavitation effects in centrifugal pumps – a review”, Int. J. Eng. Res. Appl., pp52-
63, 2016
[B37] Darian, L.A. “Cavitation process diagnostics in high-voltage oil-filled electrical equipment”, CJSC
Technical inspection UES, Russia, CIGRE 2014

115
 TB 940 – Power Transformer Audible Sound Requirements

APPENDIX C. Sound level specification using

frequency bands
Ali Al-Abadi (DE) and Christoph Ploetner (DE)

C.1. General
This brochure deals predominantly (more or less exclusively) with total sound levels. The term ‘total sound level’ is
used to describe three different physical aspects of sound and all apply to the brochure’s usage of the term. At first,
a total sound level may describe the sound emission of a device as a whole, i.e. it is the one representative
parameter for sound radiated from the entire surface of the device at a specific service condition. Measurements
taken on individual points around the radiating surface are averaged to obtain the one representative total sound
level. The second meaning of ‘total sound level’ results from the combination (superposition) of the three individually
measured sound power level components ‘no-load’, ‘load’ and ‘cooling system’ and therefore provides the emitted
total sound power level of a transformer in service. The third meaning of ‘total sound level’ is that it comprises all
audible frequency components between 20 Hz and 20 kHz (practically often measured between 50 Hz and a few
kHz up to 10 kHz) of a sound level measurement. Measured, analyzed and discussed sound levels in this brochure
are single values given in dB(A), acoustically representing the device entirely in terms of its spatial extension and
frequency components. Dealing with total sound levels as per this definition was the preferred approach to achieve
the expected targets within the given work scope to WG A2.54.
For other targets it is often necessary to look at the sound emission of specific individual locations and/or analyze
the frequency content of a (measured) sound level to see individual contributing frequency components. For the
latter, two technologies are regularly available:
1. Transformation of the sound signal from time domain into frequency domain, usually by a Fast Fourier
Transformation (FFT). The returned result is a frequency spectrum with constant steps over the entire
frequency range. A resolution of smaller than 10 Hz, for instance 2 Hz or 4 Hz, is usually applied.
2. Filtering the sound signal frequency selective by a digital filter algorithm and representing it in a number
of frequency bands. The bandwidth is absolutely not constant but instead has a constant percent
bandwidth related to the center frequency of each band. The largest applicable bandwidth is the width of
a full octave with the bands designated as 1/1 octave bands. Fractions of octave bands commonly in use
are 1/3 octave bands and also 1/12 octave bands with the latter being called narrow bands.
Both methods are normally implemented in modern sound level meters and can equivalently be used. The main
difference between the technologies for the application is the consumed processing time with the frequency
transformation technique obviously taking more time due to its linear frequency resolution. Further valuable
information on both methods is given in clause 5.4. of IEC 60076-10-1:2016 [B10].
It is noted that IEC 60076-10:2016 [B9] is setting the 1/3 octave band measurement as standard resolution for
transformer sound level measurements because it provides sufficient information in nearly all cases – based on
years of long experience. This includes cases where the resolution of a sound issue becomes necessary. In
addition, the required processing time for 1/3 octave band measurements is sufficiently small and enables sound
pressure and sound intensity level measurements where the probe needs to be moved, such as for the ‘walk-
around’ measurement procedure, without any restriction.

C.2. Octave bands

Band measurements divide the audible spectrum into smaller sections called octaves or fraction of octaves, which
allow identifying sound level components within individual frequency ranges. The band is called octave when the
upper band frequency is twice the lower band frequency. Each band covers a specific range of frequencies and
excludes all others. Octave bands are characterized as follows:
If 𝑓𝑙 is the lower cutoff frequency and 𝑓𝑢 is the upper cutoff frequency, the ratio of band limits is given by
𝑓𝑢
= 2𝑘 ,
𝑓𝑙
App Equation C.1
where 𝑘 = 1 for full octave bands and 𝑘 = 1⁄3 for one-third octave bands.
A full octave has a center frequency 𝑓𝑐 that is √2 times the lower cutoff frequency and has an upper cutoff frequency
that is twice the lower cutoff frequency. Generalized it yields,

116
 TB 940 – Power Transformer Audible Sound Requirements

𝑓𝑐
𝑓𝑙 = 𝑘
(√2)
App Equation C.2
𝑘
𝑓𝑢 = (√2) 𝑓𝑐
App Equation C.3
where the center frequency is calculated as,
𝑓𝑐 = 103 ∙ (2𝑛∙𝑘 )
App Equation C.4
with 𝑛 being the octave number in reference to 𝑓𝑐 = 103 Hz (for the full octaves 𝑛{−6, 4}, and for 1/3 octave bands
𝑛{−18, 13}). The band width 𝑏𝑤 is then calculated as
𝑏𝑤 = 𝑓𝑢 − 𝑓𝑙 .
App Equation C.5
The percent fractional bandwidth per band is constant:
𝑓𝑢 − 𝑓𝑙
𝑏𝑤 % = 100 ∙ ( ).
𝑓𝑐
App Equation C.6
The characterization of an octave band is shown in App Figure C.1.

App Figure C.1 – Amplitude versus Frequency of an octave

band
Octave band measurements are used when the frequency composition of a sound field is needed to be determined.
While the 1/1 octave analysis is for instance used for hearing protection and in environmental noise issues, 1/3
octave bands provide more in-depth insight on sound level patterns across the frequency composition and have
been found to be very useful for the transformer sound analysis. 1/3 octave bands occur by splitting 1/1 octave
bands into three parts. Octave bands covering the important range of generated sound of power transformers are
listed in App Table C.1.

117
 TB 940 – Power Transformer Audible Sound Requirements

App Table C.1 – 1/1 (full) and 1/3 (one-third) octave bands’ center, lower limit and upper limit frequencies

1/1 Octave Bands 1/3 Octave Bands

Lower Band Center Upper Band Lower Band Center Upper Band
Limit 𝒇𝒍 [Hz] Frequency Limit 𝑓𝑢 [Hz] Limit 𝑓𝑙 [Hz] Frequency Limit 𝑓𝑢 [Hz]
𝑓𝑐 [Hz] 44 𝑓𝑐 50
[Hz] 56
44 63 88 56 63 70
70 80 88
88 100 111
88 125 177 111 125 140
140 160 177
177 200 223
177 250 354 223 250 281
281 315 354
354 400 445
354 500 707 445 500 561
561 630 707
707 800 891
707 1000 1414 891 1000 1122
1122 1250 1414
1414 1600 1782
1414 2000 2828 1782 2000 2245
2245 2500 2828
2828 3150 3564
2828 4000 5657 3564 4000 4490
4490 5000 5657
5657 6300 7127
5657 8000 11314 7127 8000 8980
8980 10000 11314

C.3. Frequency weighting filters

Within the audible sound range (20 Hz – 20 kHz), the human ear is perceiving the sound in a not constant manner.
To equalize the hearing intensity for the human ear, therefore, so-called weighting filters are introduced.
Frequency weighting filters are used to adjust the sound signal as per a specified function with the intent to present
the sound for the human ear in a comparable manner. Different weighting functions are used because the sound
meter responds differently to different types of sound.
A-Weighting
The human ear is relatively insensitive to very low and very high audible frequencies, but more sensitive to
frequencies between 1 and 5 kHz. The A-weighted frequency filter is designed to adjust stationary sound signals
over the entire audile frequency range accordingly. It is the most common filter used for measurements of
environmental and industrial sound / noise and therewith also the best possible filter for transformer sound. It is
applied worldwide as the standard filter in the transformer industry. The mathematical description of the A-weighting
function and further details on A-weighting are given in clause 5.2 of IEC 60076-10-1:2016 [B10].
If a sound meter is set up for a frequency selective band measurement together with A-weighting, then the A-
weighing is, by definition, applied to the center frequencies of the individual bands, exemplarily shown for 1/1 octave
bands in App Table C.2
App Table C.2 – A-Weighting for 1/1 octave bands
Frequency (Hz) 63 125 250 500 1000 2000 4000 8000 16000
A-Weighting filter (dB) -26.2 -16.2 -8.7 -3.2 0 1.2 1.0 -1.1 -6.7

Z-Weighting
Although the term weighting is used, Z-weighting means Zero-weighting, i.e., the measured signal (sound level) is
returned without any filtering. It therefore represents the sound / sound level over the entire frequency range 10 Hz

118
 TB 940 – Power Transformer Audible Sound Requirements

to 20 kHz as generated by the source. Z-weighting is preferably used for the detailed analysis in research and
development when frequency selective measurements of normally high resolution (FFT, narrow-band, 1/3 octave
band) exclusively apply.
Weighting in FAT certificates
A-weighting is applied worldwide as the standard filter for transformer sound levels. A-weighted sound levels shall
therefore be reported by manufacturers in FAT reports for FFT, narrow-band, 1/3 octave band and total sound
levels comprising the entire measured frequency range. Mixing A-weighted and Z-weighted sound levels in the
same FAT report is confusing / difficult to understand for non-specialists and should be avoided. In the rare case
that octave bands are required, such should be reported Z-weighted to exclude systematic errors introduced due
to A-weighting for octave bands; see last section C.5 of this APPENDIX C.

C.4. Case studies: Measured sound levels

Load sound level
The sound generated by power transformers can be measured and analyzed in different ways. App Figure C.2
shows a typical load sound level measurement of a power transformer as frequency spectrum analysis (FFT), both,
in Z- and A-weighting. It shows the dominating sound level peak of the 50 Hz transformer at twice the power
frequency (100 Hz) plus the higher harmonics – all as expected.

80
70
Z-Weighting
60 A-Weighting
Lp [dB], LpA [dB(A)]

50
40
30
20
10
0
0 100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
App Figure C.2 – Typical transformer load sound level measurement (frequency spectrum)
The octave bands are not aligned with the occurring transformer harmonics. Octave bands therefore capture a
different number of harmonics at different frequency ranges. While the 125 Hz octave band captures the first
(fundamental) harmonic at 100 Hz exclusively, the 250 Hz octave band covers the two harmonics of 200 Hz and
300 Hz. In the higher frequency ranges, more than two harmonics are covered per octave band.
The importance of using octave bands appears here: The higher harmonics are lower than the first harmonic, and
hence, the summation of the higher harmonics in the higher octave bands will show a comparison value to the first
harmonic in the 125 Hz octave band. Furthermore, the higher octaves’ load sound level may deliver information
about resonances that could coincide with higher harmonics.
The representation of the measured load sound level in 1/1 Octave bands for Z- and A-weighting is shown in App
Figure C.3.

119
 TB 940 – Power Transformer Audible Sound Requirements

1/1 Octave Bands

80
70
Z-Weighting

Lp [dB], LpA [dB(A)]

60 A-Weighting
50
40
30
20
10
0
63 125 250 500 1000 2000 4000 8000
Center frequencies of the 1/1 Octaves [Hz]

App Figure C.3 – Typical transformer load sound level measurement in 1/1 Octave bands
More detailed information of the load sound level can be shown with the representation in 1/3 octave bands, App
Figure C.4

1/3 Octave Bands

80
70
Z-Weighting
Lp [dB], LpA [dB(A)]

60
A-Weighting
50
40
30
20
10
0
50

500

1000
1250
1600
2000
2500
3150
4000
5000
6300
8000
63
80
100
125
160
200
250
315
400

630
800

10000

Center frequencies of the 1/3 Octaves [Hz]

App Figure C.4 – Typical transformer load sound level measurement in 1/3 octave bands
No-load sound level
A typical no-load sound level measurement as frequency spectrum analysis (FFT) is shown in App Figure C.5. The
higher harmonics are dominating due to the magnetostriction effect of the core.

80
70 Z-Weighting
60 A-Weighting
Lp [dB], LpA [dB(A)]

50
40
30
20
10
0
0 100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]
App Figure C.5 – Typical transformer no-load sound level measurement (frequency spectrum)
App Figure C.6 shows the no-load sound level presented in App Figure C.5 as frequency spectrum in 1/1 octave
bands. The domination of the third and fourth octaves (specifically if A-weighted) is obvious. The third octave band

120
 TB 940 – Power Transformer Audible Sound Requirements

at 250 Hz covers the 200 & 300 Hz harmonics, whereas the fourth octave band at 500 Hz covers the 400, 500, 600
and 700 Hz harmonics.

1/1 Octave Bands

80
70 Z-Weighting

Lp [dB], LpA [dB(A)]

60 A-Weighting
50
40
30
20
10
0
63 125 250 500 1000 2000 4000 8000
Center frequencies of the 1/1 Octaves [Hz]

App Figure C.6 – Typical transformer no-load sound level measurement in 1/1 Octave bands
More detailed information of the no-load sound level can be shown with the representation in 1/3 octave bands,
App Figure C.7.

1/3 Octave Bands

80
70
Z-Weighting
Lp [dB], LpA [dB(A)]

60
A-Weighting
50
40
30
20
10
0
315

4000
100
125
160
200
250

400
500
630
800
1000
1250
1600
2000
2500
3150

5000
6300
8000
50
63
80

10000

Center frequencies of the 1/3 Octaves [Hz]

App Figure C.7 – Typical transformer no-load sound level measurement in 1/3 octave bands
Sound level in transformer operation
During transformer operation at nominal service condition, the generated service sound level 𝑆𝐿𝑜𝑝 is the
combination of no-load and load sound level. It is calculated as logarithmic summation of load sound level
𝑆𝐿𝑙𝑜𝑎𝑑 and no-load sound level 𝑆𝐿𝑛𝑜 𝑙𝑜𝑎𝑑 and applies for individual bands, independent of band width and weighting
function,
𝑆𝐿𝑙𝑜𝑎𝑑 𝑆𝐿𝑛𝑜 𝑙𝑜𝑎𝑑
𝑆𝐿𝑜𝑝 = 10 ∙ 𝑙𝑜𝑔10 (10( 10
)
+ 10( 10
)
)
App Equation C.7
In case the summation is executed for Z-weighted levels, A-weighting can be applied afterwards for the individual
bands,
𝑆𝐿(𝐴)𝑜𝑝 = 𝑆𝐿(𝑍)𝑜𝑝 + 𝐴. 𝑊𝑒𝑖𝑔ℎ𝑡𝑖𝑛𝑔 𝑓𝑖𝑙𝑡𝑒𝑟
App Equation C.8
For the frequency spectrum analysis (FFT), the result of the logarithmic summation is shown in App Figure C.8.

121
 TB 940 – Power Transformer Audible Sound Requirements

100
90
80 Z-Weighting

Lp [dB], LpA [dB(A)]

70 A-Weighting
60
50
40
30
20
10
0
0 100 200 300 400 500 600 700 800 900 1000
Frequency [Hz]

App Figure C.8 – Typical transformer operational sound level (frequency spectrum), calculated
The calculation result of the operational sound levels 𝑆𝐿(𝑍)𝑜𝑝 and 𝑆𝐿(𝐴)𝑜𝑝 for 1/1 octave bands is shown in App
Figure C.9. A-weightings for the individual octave bands are as given in App Table C.2 – A-Weighting for 1/1 octave
bands. The operational sound level 𝑆𝐿(𝑍)𝑜𝑝 in this example is clearly dominated by the 125 Hz octave band,
whereas 𝑆𝐿(𝐴)𝑜𝑝 shows an almost comparable sound level contribution of the second, the third and the fourth
octave bands.

1/1 Octave Bands

80
70
Z-Weighting
Lp [dB], LpA [dB(A)]

60
A-Weighting
50
40
30
20
10
0
63 125 250 500 1000 2000 4000 8000
Center frequencies of the 1/1 Octaves [Hz]

App Figure C.9 – Typical transformer operational sound level in 1/1 Octave bands, calculated
More detailed information of the operational sound level can be shown with the representation in 1/3 octave bands,
App Figure C.10.

122
 TB 940 – Power Transformer Audible Sound Requirements

1/3 Octave Bands

80
70
Z-Weighting

Lp [dB], LpA [dB(A)]

60
A-Weighting
50
40
30
20
10
0

4000
50

100
125
160
200
250
315
400
500
630
800
1000
1250
1600
2000
2500
3150

5000
6300
8000
63
80

10000
Center frequencies of the 1/3 Octaves [Hz]

App Figure C.10 – Typical transformer operational sound level in 1/3 octave bands, calculated
It is noted that the methodology outlined in the aforementioned case studies can be straightforwardly extended for
the inclusion of the cooling system sound level (despite cooling system sound being usually of broadband type).
The methodology is applicable for all three types of sound level measurement, i.e. 1/1 octave bands, 1/3 octave
bands and frequency spectrum (narrow-band).

C.5. Conclusions
From the case studies above, it becomes clear that frequency spectrum measurements return the largest amount
of information. Due to the high resolution, the application of A-weighting is fully precise – each frequency component
over the entire frequency range is weighted individually as per the weighting function. Deriving band quantities (1/3
octaves, 1/1 octaves) from the measured frequency spectrum – as done here – is therefore providing exact sound
level results for both, Z-weighting and A-weighting.
For band measurements, this is not the case because A-weighting applies to the center frequencies of the bands
while the characteristic transformer frequencies do not necessarily match the center frequencies and also because
certain bands cover more than one characteristic transformer frequency. A systematic error is therefore introduced
for A-weighted band measurements and it is immediately clear that a larger bandwidth implies a larger error. The
effect however is of certain significance only for 1/1 octave band measurements. For 1/3 octave band
measurements the effect is negligible because concerned A-weighted band sound levels as well as the total A-
weighted sound level are impacted by only a fraction of a dB.
Z-weighted band sound level measurements are for all types of measurement (1/1 octave bands, 1/3 octave bands
and frequency spectrum (narrow-band)) free of systematic errors, representing the sound energy rate contained in
each frequency band / component correctly.

123
 TB 940 – Power Transformer Audible Sound Requirements

APPENDIX D. Extension of the formulation for load

sound power levels
Ali Al-Abadi (DE)

In section 3.3, typical ranges of load sound power levels were derived with target to provide guidance for transformer
users while specifying power transformer load sound levels for new purchases. The developed formulation relates
the transformer load sound power level to the reactive power of the transformer 𝑄 = 𝑆𝑟 𝑢𝑘 , i.e., to a single physical
quantity only. In order to further precise the formulation, other selected quantities/parameters (terms) may be added
to the formulation and included in the overall statistical optimization.
As an example, a parameter is added here to represent the transformer dimension (size) more explicit. A parameter,
known at the time of specification doing this, is the transformer (unit) building power 𝑆𝑏 , also used as input
parameter for the derivation of typical ranges of no-load sound power levels. For more information on 𝑆𝑏 including
its derivation see section 3.2. As 𝑆𝑏 basically relates to the core dimension and therewith also at first order to the
tank size, while 𝑆𝑟 𝑢𝑘 relates to the winding vibrations, by introducing a split of the two quantities in the formulation,
a slight improvement in the prediction accuracy can be expected and was finally found when applying it to the
collected database. Following generic App Equation D.1 shows the extended formulation Equation 15 from section
3.3:
𝑆𝑟 𝑢𝑘 𝑆𝑏
𝐿𝑊 𝐶𝑖𝑔𝑟𝑒 _𝑒𝑥𝑡 = 𝑎 + 𝑏 log10 ( ) + 𝑐 log10 ( )
𝑄0 𝑆0
App Equation D.1
Sr transformer rated power
uk transformer short-circuit impedance in p.u.
Sb transformer building power
Q0 base reactive power 1 MVA
S0 base power 1 MVA
Suitably scaling (optimizing) parameters a, b and c in the formulation by utilizing the collected database entries for
50 Hz and 60 Hz transformers separately, to follow the concept of section 3.3., leads to following scaled equations:
𝑆𝑟 𝑢𝑘 𝑆𝑏
𝐿𝑊 𝐶𝑖𝑔𝑟𝑒 3𝑝ℎ 50𝐻𝑧 _𝑒𝑥𝑡 = 56 + 8 log10 ( ) + 8 log10 ( )
𝑄0 𝑆0
App Equation D.2
𝑆𝑟 𝑢𝑘 𝑆𝑏
𝐿𝑊 𝐶𝑖𝑔𝑟𝑒 3𝑝ℎ 60𝐻𝑧𝑒𝑥𝑡 = 55 + 9 log10 ( ) + 8 log10 ( )
𝑄0 𝑆0
App Equation D.3
App Table D.1 – Derived standard deviations and displacements from average curvepresents the standard
deviations and displacements from exact average curve for the CIGRE A2.54 developed load sound model (section
3.3) and the extended model:
App Table D.1 – Derived standard deviations and displacements from average curves
Std. deviation Displacement Std. deviation Displacement
50 Hz 50 Hz 60 Hz 60 Hz
CIGRE A2.54 model [dB(A)] 4.7 0.5 4.8 0.1
Extended model [dB(A)] 4.6 0.4 4.7 0.1

As expected, the extended formulation provides a slight improvement versus the formulation introduced in section
3.3 in terms of standard deviation and enables therefore a somewhat better pin-pointing of the load sound power
level of individual transformers. This can be useful specifically for transformer manufacturers while designing
transformers. In terms of the average functions and typical ranges of load sound power levels as derived in section
3.3 and presented in Figure 40 and Figure 41 for mainly transformer specification purposes, differences between
the two models do not occur.

124
 TB 940 – Power Transformer Audible Sound Requirements

APPENDIX E. Cavitation
Werner Goette (DE)

E.1. Cavitation – the phenomena

The term cavitation is deduced from the Latin verb ‘cavitare’, which means ‘to hollow out’. There are two types of
cavitation that are closely linked but must be distinguished.
Hard cavitation (transient cavitation)
is the formation and disintegration of vapor-filled cavities (bubbles). The formation of the cavities occurs due to an
imposed local decrease of the static pressure below the vapor pressure. The consumed extra volume by the vapor
bubbles results in an increase of the pressure that causes the bubbles to collapse spontaneously with the vapor
being disintegrated. Both, formation and disintegration occur at the same time. The sudden implosion of the bubbles
may cause high local flow velocities, pressure peaks, noise, vibrations and high temperatures. Hard cavitation is
predominately occurring on pure liquids.
Soft cavitation
is the formation and dissolution of gas (vapor) bubbles. If the solution equilibrium between vapor and gas changes
because of a pressure change, the solubility of the gas in the liquid changes. Now, due to an imposed local decrease
of the static pressure, the gas (vapor) solubility is reduced (Henry's law) and gas bubbles are formed. The extra
volume consumed by the bubbles increases the pressure and the reverse process takes place, i.e. free gases are
dissolved into the liquid. Both, formation and dissolution occur at the same time. In contrast to the sudden bubble
collapse occurring at hard cavitation, dissolving of the gases into the liquid is here determined by diffusion which is
slower and smoother.
Cavitation is a very diverse and complex phenomenon with many facets and areas of application. Hence, it is a
common subject in scientific literature (e.g. [B33][B34][B35][B36]).
Cavitation occurs most frequently with fast-moving objects but also by abrupt changes in the cross-sectional area
(Bernoulli's law), e.g. on impellers of pumps and fittings which impose a local static pressure decrease. More
general, it occurs at locations where the pressure decreases below the vapor pressure due to acceleration
processes.

E.2. Cavitation from a mechanical point of view

Implosion of vapor bubbles implies extreme changes in the volume ratios in the liquid. This can lead to extreme
local temperatures and pressures. Water vapor, e.g. at 20 °C, consumes about 50,000 times more space than liquid
water (molar volume of water vapor to molar volume of water under normal conditions is about a factor of 1,200,
the vapor pressure of 23.4 mbar at 20 °C results in a factor of about 40).
As a result of imploding vapor bubbles, microjets are created which induce very high pressures near the wall and
can lead to microcracks in the material. Depending on intensity and duration, material is gradually removed. This
can lead to failure of affected material due to material fatigue.
Vibration sensors can be used to monitor the condition of machines that are at risk of cavitation. They timely provide
alarms before critical conditions occur.

E.3. Cavitation from a chemical point of view

The release of energy during the implosion of the vapor bubbles can lead to high temperatures. If the substances
are not pure but mixed, the components are usually in equilibrium. The reactivity of mixtures of substances depends
i.a. on composition and temperature, so that chemical reactions must be expected when high local temperatures
persist. Fast changing conditions mean that these reactions are not reversed, or only to a limited extent.

E.4. Cavitation from an acoustic point of view

The implosion of vapor bubbles creates shock waves that may be accompanied of noises of significant volume.
This effect, also known as “water hammer”, can cause considerable damage.
Noise levels can be measured during operation to detect the onset of cavitation states before the phenomenon can
be visually perceived (condition monitoring).
Acoustic cavitation describes the bubble formation and implosion, i.e. the bubble dynamics in liquids exposed to
ultrasonic fields. This mechanism is used for various applications.

125
 TB 940 – Power Transformer Audible Sound Requirements

E.5. Applications of cavitation

Cavitation occurs when power ultrasound (20 kHz - 100 kHz) enters fluids. In the process, transient bubbles are
generated in the fluid, which implode as cavitation bubbles and release high amounts of energy at certain points.
This energy can be used e.g. for:
▪ Emulsification and dispersion processes in pharmaceutical synthesis, chemical process engineering or
food process engineering
▪ Removal of fouling layers in pipes and devices e.g. heat exchangers
▪ Detachment and destruction of biofilms
▪ Digestion of biological cells (algae cells)
▪ Generation of radicals to initiate chemical reactions (e.g. emulsion polymerization)
▪ Sonoluminescence emission or cleaning and erosion activity
▪ Extracorporeal shock wave therapy (smashing kidney stones)
▪ Ultrasound irradiation to break down fat deposits (dissolving fat cells through cavitation) in plastic and
aesthetic medicine

E.6. Cavitation and transformer cooling

For pumps and cooling circuits of transformers, cavitation is a phenomenon that should be prevented by appropriate
design of the cooling system and operating parameters. Cavitation is discussed in connection with the assessment
of transformers in [B37].
The vapor pressures of the substances used as insulating fluids are in the mbar range. A pressure reduction to this
level is not very realistic when operating cooling circuits, but can also not fully be excluded. Transient (hard)
cavitation should therefore not or only very rarely occur in these systems. If cavitation effects occur, they are most
likely caused by dissolved gases. The main factors influencing the occurrence of soft cavitation are the geometrical
arrangement, operating or flow conditions, and liquid characteristics (especially temperature and gas saturation
level).
When designing the piping of cooling circuits, it is therefore very important to consider a positive inlet head at the
suction side, a suitable arrangement of the components and a proper design of the piping, especially on the suction
side. Section 2.2.3.1.2 of this brochure is dealing with transformer oil pumps and provides further recommendations
in this respect.

126
 TB 940 – Power Transformer Audible Sound Requirements

APPENDIX F. Survey on transformer sound level

specification
Janine Dickinson (GB) and Jan Dončuk (CZ)

To understand how utilities currently specify sound levels for new transformer purchases and to identify any
common themes across the world, CIGRE WG A2.54 produced a questionnaire about utilities practice and
experience of transformer sound level specification. The intention of this appendix is to provide at a glance the most
essential outcome from the survey.
The questionnaire, which comprised a series of 20 multi-choice and free text questions, was sent to utilities across
the world. 97 completed questionnaires were received and over 40 countries were represented in the responses.
A copy of the covering email sent to utilities requesting completion of the questionnaire is provided in App Figure
F.1, the questionnaire is outlined in App Table F.1 and the responses to the questionnaire are finally summarised.

Dear xxx,
CIGRE WG A2.54 (Power transformer audible sound requirements) is working on the production of an information pack to
provide guidance on typical transformer noise level ranges.
Please help us understand what drives the requirements for the noise levels you specify for the transformers you purchase.
We would be very grateful if you would complete the questionnaire attached. This will assist with our research and make
the information pack more informative for the industry.
This email has been sent to a group of people within your company (please see the email distribution list). We require only
one response to the questionnaire so please coordinate a single response between you. If there is a more appropriate person
within your company to complete the questionnaire, please forward this email to them or alternatively please provide their
contact details and we will contact them directly.
The attached questionnaire is composed of 20 brief questions focused on the Utility experience of transformer noise limits,
requirements and mitigation techniques.
If you would prefer to receive a paper copy or complete the questionnaire over the phone, please email or call us.
It would be very useful for our research and to support our work if you are also able to provide copies of your sound level
specifications/regulation documentation. Any documentation you are able to provide will be treated confidentially and not
distributed further.
Thank you in advance, your help is greatly appreciated.
Best regards,
CIGRE WG A2.54

App Figure F.1 – Covering email sent to utilities

127
 TB 940 – Power Transformer Audible Sound Requirements

App Table F.1 – CIGRE WG A2.54 questionnaire –- power transformers audible sound requirements

1. What is your company name (optional) and in which territories (country, city, state, region
etc.) do you operate?
2. What type of transformers do you operate (tick all that apply)?
a. Transmission network transformer
b. Autotransformers
c. Distribution network transformer
d. GSU (generator step-up unit)
e. Other (please specify)
3. Do you operate transformers indoors or outdoors?
a. Indoors
b. Outdoors
c. Both
4. Do you specify maximum noise levels for the transformers you purchase (tick all that apply)?
a. Yes – Maximum ‘no-load’ noise limits
b. Yes – Maximum ‘load’ noise limits
c. Yes – Maximum overall noise limits (load/no-load/cooler)
d. Yes – Maximum ‘no-load’ with cooler noise limits
e. Yes – Maximum noise limits for transformer cooler
f. No (go to Q7)
g. Other (please specify)

5. How do you specify your noise limits?

a. Site by site or case by case
b. General noise limits
6. Do you specify noise limits in terms of:
a. Sound pressure
b. Sound power
7. Which test method do you specify for the transformer Factory Acceptance Test?
a. Sound pressure
b. Sound intensity
c. No method specified
8. Are there any requirements relating to noise (e.g. maximum noise limits for substations)
that you must comply with?
a. Yes
b. No (go to Q15)
9. What are these requirements? (free text)
10. How do they affect the maximum noise levels you specify for transformers you purchase?
(free text)
11. Are there maximum noise levels you are not allowed to exceed?
a. Yes
b. No (go to Q15)
12. What are these maximum noise levels (e.g. 50 dB(A) LAeqT)?

128
 TB 940 – Power Transformer Audible Sound Requirements

13. At what time of day do they apply?

a. Day
b. Night
c. Day and Night
d. Other (please specify)
14. Where do these maximum noise levels apply?
a. Substation boundary
b. Transformer
c. Residential façade
d. Residential garden
e. Boundary of residential areas
f. Park
g. Commercial area
h. Industrial area
i. Other (please specify)
15. What is the reasoning/basis for the transformer noise limits you specify (tick all that apply)?
a. Legislation
b. Local policy
c. Nearby residents
d. Company policy
e. Health and Safety
f. Noise study results
g. Expected costs of noise mitigation
h. Other (please specify)

16. Which factors influence the tendering/procurement process when purchasing a transformer
(list all that apply in terms of priority, with the first being of the highest priority)?
• Cost of transformer
• Cost of noise mitigation (e. g. in case of exceeding specified noise limit)
• Noise levels from existing substation equipment
• The best losses to cost ratio
• The best noise level to cost ratio
• Penalties for exceeding noise limits specified in tender
• Other (please specify)

17. Do you consider the acceptability of the noise limit of the individual transformer or the
substation as a whole?
a. Individual transformer
b. Substation as a whole
c. Both
d. Other (please specify)

129
 TB 940 – Power Transformer Audible Sound Requirements

18. Do you carry out noise measurements on/around the substation site when a new
transformer is being installed (tick all that apply)?
a. Yes, only before installation of the transformer
b. Yes, only after installation of the transformer
c. Yes, before and after installation of the transformer
d. No
19. Do you purchase cheaper, louder transformers and then apply noise mitigation or do you
buy more expensive, quieter transformers to avoid the need to apply noise mitigation?
a. Louder transformers and then apply noise mitigation
b. Quieter transformers to avoid noise mitigation
c. Both
d. Other (please specify)
20. What techniques, products or methods do you use to mitigate noise from transformers (tick
all that apply)?
a. Noise enclosures
b. Noise absorbing barrier (as part of the transformer fire wall)
c. Noise barriers – on substation site
d. Noise barriers – close to houses
e. Double glazing at residential property
f. Transformer tank mounted sound reduction panels (or equivalents)
g. None
h. Other (please specify)

The graphs below show on the ordinate axis the number of individual responses. As some questions allowed
respondents to choose more than one option, the number of responses to individual questions does not equal the
total number of questionnaires completed.

130
TB 940 – Power Transformer Audible Sound Requirements

Summary of Questionnaire Responses

131
TB 940 – Power Transformer Audible Sound Requirements

132
TB 940 – Power Transformer Audible Sound Requirements

133
TB 940 – Power Transformer Audible Sound Requirements

134
TB 940 – Power Transformer Audible Sound Requirements

135
TB 940 – Power Transformer Audible Sound Requirements

136
TB 940 – Power Transformer Audible Sound Requirements

137
 ISBN : 978-2-85873-645-4

TECHNICAL BROCHURES
©2024 - CIGtRE
Reference 940 - September 2024