Proceedings ASME Turbo Expo 2005

International Gas Turbine and Aeroengine Congress

June 6-9, 2005 Reno, Nevada

This is a preprint or reprint of a paper presented at a

conference. Since revisions may be made prior to formal
publication, this version is made available with the
understanding that it will not be cited or reproduced without
the permission of the author.
GT2005-69132

THE EFFECT OF ULTRAPOLISH ON A TRANSONIC AXIAL ROTOR

William B. Roberts Patricia S. Prahst
Airfoil Technologies International and AP Solutions
Flow Application Research Solon, Ohio, USA
Freemont California, USA

Scott Thorp Anthony J. Strazisar

NASA-Glenn Research Center NASA-Glenn Research Center
Cleveland, Ohio, USA Cleveland, Ohio, USA

ABSTRACT Figure 1, taken from Suder et.al. [1], shows the impact of
Back-to-back testing has been done using NASA fan rotor 67 in surface roughness of 2.5-3.0 µm on the design speed
the Glenn Research Center W8 Axial Compressor Test Facility. performance of a transonic compressor rotor. The six
The rotor was baseline tested with a normal industrial RMS configurations shown in this figure differ in the location of
surface finish of 0.5-0.6 µm (20-24 microinches) at 60, 80 and roughness on the blade surface. One key finding from this
100% of design speed. At design speed the tip relative Mach work was that surface roughness on the leading edge of the
number was 1.38. The blades were then removed from the blade, which exists for configurations G, H, and I, causes the
facility and ultrapolished to a surface finish of 0.125 µm (5 most severe performance penalty. Suder’s work leads to the
microinch) or less and retested. question of whether or not a gain in efficiency could be
achieved if axial fan and compressor blade surfaces are
At 100% speed near the design point, the ultrapolished blades polished to finishes less than 0.5-0.6 µm.
showed approximately 0.3 - 0.5% increase in adiabatic
efficiency. The difference was greater near maximum flow. According to Koch and Smith [2], the blade surface is
Due to increased relative measurement error at 60 and 80% considered hydraulically smooth for equivalent sand roughness
speed, the performance difference between the normal and Reynolds number less than 90. In theory, a better (lower RMS)
ultrapolished blades was indeterminate at these speeds. finish than hydraulically smooth offers no benefit to
performance. The standard industry surface finish of 0.5-0.6 µm
is typically near or less than a roughness Reynolds number of
INTRODUCTION 90 at high altitude cruise operating conditions where most fuel
As fuel costs continue to increase, the industries that use gas burn occurs. At the test conditions used in this work, a surface
turbine engines have a pressing need to improve in-service finish of 0.125 µm (5.0 microinches) corresponds to a
performance. This need is especially true in the airline roughness Reynolds number of approximately 20. Therefore,
industry, for which fuel burn and performance retention in high there is some justified skepticism as to whether ultrapolish,
by-pass ratio turbofan engines is a major concern. Any defined as a surface finish of 0.125 µm or better, can improve
approach to improve the efficiency of present equipment will efficiency. However, occasional airline trials with ultrapolished
be considered if cost effective. blading have indicated that ultrapolishing is beneficial in that
lower fuel burn is observed during acceptance testing after
Previous research by Suder et.al. [1] has shown that surface engine refurbishment. To verify and quantify the impact of
finish of axial fan and compressor blading is very important. ultrapolished surface finish on blade performance, back-to-back
This earlier research indicates that there is a significant testing was done in the NASA Glenn Research Center’s W8
performance penalty if the surface finish degrades from an Axial Compressor Test Facility using NASA fan rotor 67.
industry-standard RMS finish of 0.5-0.6 µm (20-24 Baseline testing at a nominal surface finish of 0.5 µm (20
microinches) to a finish of 2.5-3.0 µm (100-125 microinches). microinches) was followed by testing with the blades

1 Copyright © 2005 by ASME

 surface finish is 0.6 µm (24 microinches) or better, the airfoil
surface tolerance is ± 0.04 mm, and the running tip clearance is
approximately 0.5 mm.

TEST FACILITY
A schematic diagram of the NASA Glenn Axial Flow
Compressor Test Facility is shown in Figure 3. The drive
system consists of a 7000 hp electric motor with a variable-
frequency power supply. Motor speed is controllable from 400
to 3600 rpm. The motor is coupled to a 5.25 gear ratio speed
increaser gear box that in turn drives the rotor. The facility is
sized for a maximum airflow of 45 kg/sec with atmospheric air
as the working fluid.

Figure 1. Pressure rise characteristics of a transonic

compressor rotor as a function of the extent of surface
roughness, from Suder [1].
ultrapolished to 0.125 µm (5 microinches) or better at 60, 80
and 100% of design speed.

TEST ROTOR
NASA Rotor 67 is shown in Figure 2. It is a low-aspect-ratio
design and the first-stage rotor of a two-stage fan. A complete
description of the aerodynamic design of the full two-stage fan
is given in [3] and [4].
Figure 3. Schematic diagram of the NASA-Glenn Axial
Compressor Test Facility.
Air is drawn into the facility from an inlet located on the roof of
the building. The air first passes through a 10 µm filter to
remove large particles. An alternate dry-air source can also be
used to supply air to the facility. This source was used in the
present work to eliminate humidity variations between testing
of the baseline and ultrapolished blading. The inlet air passes
through a flow measuring station consisting of a thin-plate
orifice, through inlet throttling valves, and into a settling
chamber. The air is accelerated into the compressor test section
through a nozzle, passes through the test rotor, and then passes
through a sleeve valve into a collector before it is exhausted
back into the atmosphere. The airflow is controlled through the
sleeve valve.

INSTRUMENTATION AND MEASUREMENT

TECHNIQUES
Massflow is measured using a calibrated orifice located far
upstream of the compressor. The orifice measurements are
Figure 2. NASA fan rotor 67. corrected to standard-day conditions based on settling chamber
The rotor design pressure ratio is 1.63 at a mass flow of 33.35 temperature and pressure. Radial distributions of total and static
kg/sec. The design rotational speed is 16,043 rpm, which pressure and flow angle are measured using rakes at stations 1
yields a tip speed of 429 m/sec and an inlet tip relative Mach and 4 shown in Figure 4. Stage 67 was operated in a rotor-only
number of 1.38. The rotor has 22 blades and an aspect ratio of mode without the stator installed in the present work. Station 4,
1.56 (based on average span/root axial chord). The rotor which is normally used to survey the stator outlet flow, is
solidity varies from 2.11 at the hub to 1.29 at the tip. The inlet therefore the first available aerodynamic survey station
and exit tip diameters are 51.14 and 48.5 cm respectively, and downstream of the rotor. The rake measurements are corrected
the inlet and exit hub/tip radius ratios are 0.375 and 0.478 for streamline slope based on a calibration of each probe used
respectively. A fillet radius of 1.78 mm is used at the airfoil- and on the design streamline slope. All rake measurements are
hub juncture. The square root of the mean square of the airfoil corrected to sea-level standard-day conditions in the settling

2 Copyright © 2005 by ASME

 1.74
Stn 1 Stn 4 1.68
100% speed

Pressure Ratio
1.62

1.56

1.50
flow
1.44
32 33 34 35 36

1.40
80% speed
-20 -10 0 10 20 1.38

Pressure Ratio
Baseline
X (cm) 1.36 Ultrapolish
1.34
Figure 4. Location of aerodynamic survey measurement
stations. 1.32

chamber. Radial distributions of total pressure are energy 1.30

averaged by converting them to their enthalpy equivalents and 26 27 28 29 30
then mass averaging them across the annulus. Rotor efficiency
is determined using the pressure rise measured across the rotor 1.20
and torque measured by a torque meter. Measurement
1.19
uncertainties are: massflow, ±0.11 kg/s; flow angle, ±0.5

Pressure Ratio
degrees; total pressure, ±0.01 N/cm2; torque ±11.5 cm-kg (10 60% speed
1.18
in-lb).
1.17
TEST RESULTS 1.16
The compressor characteristics for Rotor 67 with the baseline
and ultrapolish surface finish are shown for 60, 80, and 100% 1.15
of design speed in Figure 5. Performance of the baseline 19 20 21 22 23
configuration was not measured at flow rates near stall because Massflow (kg/sec)
of a concern on maintaining rotor balance during a stall. The
performance was measured all the way to stall for the Figure 5. Pressure rise characteristics for baseline and
ultrapolished configurations at 80% and 100% speed. ultrapolished surface finishes.
capability. The pressure ratio measured in the present work is
The error bars in Figure 5 indicate estimated uncertainties in consistent with that measured and predicted using 3D Navier-
the calculated quantities based on a propagation of error Stokes simulations for this rotor by Pierzga and Wood [5].
analysis using the measurement uncertainties. The
measurement uncertainties at 60% and 80% speed are large due The main purpose of this research is to determine the difference
to the decreased relative instrument sensitivity that results from in adiabatic efficiency between the baseline and ultrapolished
the lower pressure rise and torque levels at these speeds. The blades. Figure 6 shows the torque-based efficiency plotted
differences in performance curves for 60% and 80% speed are versus corrected mass flow for baseline and ultrapolished
small between the baseline and polished blades. blades at 60, 80 and 100% speed with error bars added. The
actual rotor efficiency is higher than that shown here because
At 100% speed performance differences between the baseline this data has not been corrected for the tare torque of the rotor
and ultrapolished rotor are discernible and are larger than the drive system.
measurement uncertainties. At design speed the results would
seem to imply that the maximum flow capacity of the DISCUSSION
ultrapolished rotor is higher than that of the baseline rotor. We The results in Figure 6 indicate that for 100% speed there is a
are not certain if this is true since the rotor was not yet choked noticeable difference between the baseline and ultrapolished
at the lowest backpressure measured for either surface finish. adiabatic efficiency, with the ultrapolished blades showing a
We do note however that the trend toward higher flow capacity small but significant increase. Near the maximum efficiency
as the surface finish improves is consistent with the trend operating condition, the difference is approximately one-half of
observed by Suder et al [1], as shown in Figure 1. one percent (0.5%).
We also note that the pressure ratio at the design operating Although the results in Figure 6 indicate that the difference in
condition is significantly higher than the design value of 1.63. efficiency is greater at higher mass flow conditions, transonic
This rotor was designed in the mid 1970’s, before the fans in service operate near maximum efficiency. Therefore,
availability of accurate turbomachinery performance prediction the improvement in performance seen by an engine in service

3 Copyright © 2005 by ASME

 0.88
100% speed
0.87

0.86

0.85
Baseline
0.84
Ultrapolish
0.83

0.82 Blade leading edge shape before and after enhanced

32 33 34 35 36 refurbishment.
0.94
Torque-based Efficiency

80% speed
0.93

0.92

0.91

0.9

0.89

0.88
26 27 28 29 30
0.94 Comparison of thrust-specific fuel consumption
60% speed before and after enhanced refurbishment.
0.93

0.92 Figure 7. Impact of leading edge shape on the thrust

specific fuel consumption of a high bypass ratio turbofan
0.91 engine, from Roberts [6].
0.9 To check the sensitivity of Rotor 67 performance to leading
0.89 edge shape, a quasi three-dimensional computational analysis
was performed on the blade geometry at 100 percent design
0.88
speed. The 70 percent span location (from the hub) was chosen
19 20 21 22 23
for this analysis since it is well outside the endwall region but
Massflow (kg/sec)

Figure 6. Adiabatic efficiency characteristics for baseline

and ultrapolished surface finishes.
would likely be on the order of 0.5% rather than the larger
values shown for higher mass flow. The larger efficiency
differences observed at high mass flow rates are likely due to
changes in the oblique rotor shock strength around the baseline
and ultrapolished leading edge.

References [1] and [6] indicate that the leading edge of axial
blading can have a significant impact on aerodynamic
performance, with rough, oversized, or blunt leading edges
significantly decreasing performance. Figure 7, taken from
Roberts [6], shows the impact of fan blade leading edge shape
on thrust specific fuel consumption of a high bypass ratio
turbofan engine.

Figure 8 shows a photographic magnification of the leading

edge of Rotor 67 before and after ultrapolish. Although the
leading edge of the ultrapolished blade is not perfectly smooth,
the surface finish is more uniform than the baseline leading
edge. The ultrapolish process does remove a very small amount
of material from the surface of the blade. Therefore, if the
ultrapolish process improved the leading edge of the blades by
smoothing or decreasing the thickness, all or part of the
efficiency increase shown in Figure 6 could be from leading Figure 8. Magnified photograph of rotor 67 leading edge
edge shape change. before and after ultrapolishing.

4 Copyright © 2005 by ASME

 1
Baseline blade

0.8 1.5

Surface Mach Number

0.6 Ultrapolished blade
Y (mm)

Baseline blade
1.0
0.4
Ultrapolished blade
0.2

0.5
0
0 0.5 1 1.5 2 2.5 3
X (mm)
0
Figure 9. Measured blade leading edge profile before and 0 1
Normalized chord, x/c
after ultrapolishing.
Figure 10. Comparison of predicted surface Mach number
well within the transonic region, i.e., relative inlet Mach
distribution at 70% span from the hub at design speed
number at design speed equal to 1.15. A composite blade shape
for the baseline and ultrapolish blades was derived by using blade geometry as measured before (baseline) and
averaging the measurements acquired from the 22 individual after ultrapolishing.
blades by a coordinate measuring machine at 70 percent span
for pre- and post-polished cases. These composite blade
profiles are shown in Figure 9, where no significant difference number distributions are essentially the same. The two main
in leading edge shape can be seen. The coordinate measuring performance parameters, total pressure loss coefficient, ω , and
machine accuracy is on the order of 0.005 mm, which is smaller flow turning, θ, are slightly different. The difference between
than the data point symbol size in Figure 9. Note that the computed loss coefficients is ∆ω = 0.11798 - 0.11624 = .00174.
ultrapolished blade is actually thicker than the baseline blade From aerothermodynamic analysis and industrial experience it
near the leading edge because the ultrapolish process removed a is known that a 1% change in blade element loss coefficient
small amount of material from the nose of the blade, thus over a high pressure compressor leads to approximately 1%
effectively blunting the nose of the airfoil. change in adiabatic efficiency. Therefore, the efficiency
increase predicted by the quasi three-dimensional analysis due
From Figure 5, it can be seen that there are two data points at to airfoil leading edge changes after ultrapolishing is 0.00174
design speed for the baseline and ultrapolished cases that have or 0.174%. This is about one fourth of the measured
the nearly same corrected mass flow of 33.56 kg/sec (74 improvement in efficiency. The predicted difference in flow
lb/sec). This means that inlet-air-angles and inlet-relative Mach deflection before and after ultrapolishing is 0.3°, which is well
numbers are the same for both cases. However, there is a within the manufacturing tolerance of ±1° for Rotor 67 blading.
measured difference of 0.65% in efficiency between these Therefore, we conclude that the efficiency improvement after
points (see Figure 6) due to either surface finish or leading edge ultrapolishing shown in Figure 6 at design speed is due to the
shape. surface finish change.

A numerical analysis of the flow over the pre- and post-polish

composite profiles was performed for this mass flow rate using
the quasi three-dimensional Navier-Stokes solver developed by
Chima [7]. This code has previously been used to predict Parameter Baseline Ultrapolish
stream tube loss coefficient for the transonic fan of a large M1 1.15 1.15
production high-bypass turbofan engine. The computation
using the code predicted measurement within ∆ω = 0.0003 [8]. β 1 (deg) 65.6 65.6

β 2 (deg) 57.9 57.7

The relative inlet-air-angle and Mach number at 70% span was
deduced from the measured inlet total and static pressure, θ (deg) 7.6 7.9
plenum temperature and rotational speed. The outlet static to ω 0.11798 0.11624
inlet total pressure ratio was adjusted to minimize the massflow ∆m& (%) 0.1784 0.03264
residual, ∆m & . The results are shown in Figure 10 and Table 1.
Table 1. Summary of Rotor 67 blade element performance
It can be seen that the inlet and outlet Mach numbers, flow predictions for the baseline and ultrapolished geometry at
angles, total pressure loss coefficients and surface Mach 70% of span from the hub at 100% speed

5 Copyright © 2005 by ASME

 Change in adiabatic efficiency, % CONCLUSIONS
NASA Rotor 67 has been tested with an industry-standard
surface finish of 0.5 µm (20 microinches), ultrapolished to a
VERY SMOOTH ~ 0.125 µm 0.125 µm (5 microinch) finish, and retested. An approximate
0.4 increase in efficiency of 0.5% was measured for the
ultrapolished rotor near maximum efficiency at design speed.

0.0 BASELINE ~ 0.5 µm An increase in efficiency on the order of 0.5% across the fan
and compressor reduces the exhaust gas temperatures (EGT) by
-0.4 VERY ROUGH ~ 5°-8°C, resulting in increased time in service. A typical
2.5 - 3.0 µm medium-size turbofan engine such as the PW2037 or the
CFM56 burns between 1.5 and 2 million gallons of fuel per
-0.8 year in regular airline service [9]. At present day fuel prices it
would take 3-4 months to pay back the cost of ultrapolishing if
the process yields a 0.5% reduction in fuel burned. After the
-1.2 payback period, the fuel cost savings would be $7,500-$10,000
per engine per year (at a fuel cost of one dollar/gallon).

0.51 1.02 1.52 2.54 3.05 µm ACKNOWLEDGEMENTS

20 40 60 100 120 µinch The authors would like to thank Mr. Rick Brokopp, Mr. Helmi
Abulaban, and Mr. John Jones for supporting the installation
RMS surface roughness and testing of Rotor 67 in the NASA-Glenn Axial Compressor
Test Facility, and Dr. Randy Chriss for performing the
measurement error analysis. The first author would like to
Figure 11. Approximate variation of adiabatic efficiency thank Airfoil Technologies International for their financial
with surface finish for transonic rotors. support. The financial support of the NASA Ultra-Efficient
Suder et al [1] measured the efficiency change of a transonic Engine Program for this research is also gratefully
core compressor rotor when surface roughness was increased acknowledged.
above the industry-standard surface finish level of 0.5-0.6 µm.
In the present work we have investigated the efficiency change REFERENCES
of a transonic fan rotor when the surface roughness level was 1 Suder, K.L., Chima, R.V., Strazisar, A.J., and Roberts,
decreased below industry-standard levels. We have found that W.B., 1995, “The Effect of Adding Roughness and
these results are in line with one another as shown in Figure 11, Thickness to a Transonic Axial Compressor Rotor,”
which combines the efficiency changes measured by Suder ASME Journal of Turbomachinery, Vol. 117, pp. 491-
with those measured in the present work. This semi-logarithmic 505.
plot shows the trend in adiabatic efficiency change with surface
finish for transonic rotors. 2 Koch, C.C., and Smith, L.H., 1976, “Loss Sources and
Magnitudes in Axial Flow Compressors,” ASME
As blades foul in service, the efficiency of the compression Journal of Engineering for Power, Vol. 98, No. 3, pp.
system eventually drops off. The ability of a rotor to retain the 411-424.
ultrapolish efficiency gain in service is therefore a valid
concern. We assessed this issue by checking the performance 3 Cunnan, W.S., Stevens, W., and Urasek, D.C., 1978,
of this rotor after the completion of a test program that followed “Design and Performance of a 427 Meter-per-
the present work. The rotor was operated for 13 months in the Second-Tip Speed Two-Stage Fan Having a 2.40
follow-on program during which it accumulated 66 hours Pressure Ratio.” NASA TP-1314.
(~4000 minutes) of operation at design speed. Throughout this
follow-on program, the rotor was supplied with atmospheric 4 Urasek, D.C., Gorrell, E.T., and Cunnan, W.S., 1979,
air, filtered to 10 µm. We found no measurable difference in “Performance of Two-Stage Fan Having Low-Aspect-
efficiency compared to the initial ultrapolished performance. Ratio, First-Stage Rotor Blading,” NASA TP-1493.

A commercial aircraft engine operates at full-speed only during 5 Pierzga, M.J., and Wood, J.R, 1985, “Investigation of
the takeoff climb and during thrust reversal at landing. the Three Dimensional Flow Field Within a Transonic
Furthermore, there are substantially fewer atmospheric particles Fan Rotor: Experiment and Analysis,’ ASME Journal
that can foul blades above 10,000 feet. A commercial engine of Engineering for Gas Turbines and Power, Vol. 107,
therefore spends only 4-5 minutes per flight at design speed at No. 2, pp. 436-449.
altitudes where atmospheric particulates are a concern. The
4000 minutes of operation accumulated by the rotor before the 6 Roberts, W.B., 1995, “Advanced Turbofan Blade
performance was rechecked therefore represent 800-1000 Refurbishment Technique,” Technical Brief, ASME
cycles. Journal of Turbomachinery, Vol. 117, pp. 666-667.

6 Copyright © 2005 by ASME

7 Chima, R.V., 1987, “Explicit Multigrid Algorithm for
Quasi-Three-Dimensional Viscous Flows in
Turbomachinery,” AIAA Journal of Propulsion and
Power, Vol. 3, No. 5, pp. 397-405.

8 Roberts, W.B., 1994, “The Effect of Recontouring GE

CF6-80C2 Fan Blades,” FAR Technical Note prepared
for Sermatech International and KLM Royal Dutch
Airlines.

9 Delta Airlines, 2004, Private communication with Mr.

James Justice, Performance Engineering.

NOMENCLATURE
s/c space/chord ratio
M1 inlet relative Mach number
X distance along the chord line
Y distance normal to the chord line
PR total pressure ratio
PS pressure surface
SS suction surface
TSFC thrust-specific fuel consumption
∆m& massflow residual, quasi three-dimensional calculation
β1 inlet relative air angle
β2 exit relative air angle
θ airflow turning through the blading
ω total pressure loss coefficient

7 Copyright © 2005 by ASME