This project has received funding from the European Union’s Horizon 2020

research and innovation programme under grant agreement No 792059

Oct. 29th 2020

PV System: Operation & Maintenance
(14:30-16:30)
 PV System: Operation & Maintenance, ENEL GREEN POWER
Matteo Poletto, Tech. Supp O&M Solar Competence Center

Index
1. Introduction to EGP

2. Maintenance: some definitions

3. Maintenance: the costs

4. Predictive maintenance: some hints

5. Innovative solutions for PV maintenance

GoPV Project | 1st TRAINING COURSES
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 REN Operation & Maintenance global presence
2020 August Italy Germany
Plants: 600 Plants: 1 Technology MW Plants
USA & Canada Capacity: 14.034 MW Capacity:22 MW*/ 35 MWh

Plants: 88 Romania
Hydro 27.914 761
Capacity: 6.250 MW Iberia
Plants: 12
Plants: 270
Capacity: 534 MW Geo 881 38
Capacity: 7.477 MW
Mexico Greece-Bulgaria
Wind 13.309 265
Plants: 19 Panama Plants: 53
Capacity: 2.823 MW Capacity: 503 MW
Plants: 8 Solar 4.990 108
Morocco
Guatemala Capacity: 362 MW
Plants: 1
Plants: 5 Colombia Capacity: 131 MW
Biomass 61 11
Capacity: 164 MW India
Plants: 13
Capacity: 3.183 MW Plants: 3 Storage* 22 1
Costa Rica Capacity: 172 MW

Plants: 3 Peru Brazil

Zambia 47.155 1.184
Capacity: 81 MW
Plants: 10 Plants: 1
Plants: 49
Capacity: 1.104 MW Capacity: 34 MW
Capacity: 3.308 MW
Australia
Chile Argentina South Africa
Plants: 2
Plants: 36 Plants: 2 Plants: 8 Capacity: 309 MW
Capacity: 4.723 MW Capacity: 1.328 MW Capacity: 635 MW

1.184 plants 47,1 GW installed 109,5 TWh produced(2019) 23 Countries 4.400 O&M people
2
October 26-29th 2020
 MAINTENANCE: some definitions

There are different kind of maintenance:

• SCHEDULED maintenance: actions and/or testing and/or measurements to ensure optimal operating
conditions of equipment and of the entire PV plant and to prevent defects and
failures

• CORRECTIVE maintenance: actions and/or techniques taken to correct failures, breakdowns, malfunctions,
anomalies or damages detected during inspections, or through monitoring,
alarming, or reporting or any other source

• PREDICTIVE maintenance: Set of activities that detect changes in the physical condition of equipment in
order to predict and anticipate failures

• EXTRAORDINARY maintenance: actions and/or works performed in case of major unpredictable faults,
such as serial defects, force majeure events, natural events, etc

GoPV Project | 1st TRAINING COURSES

4
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 MAINTENANCE: some definitions

The maintenance activities covers all the parts of a PV plant

GoPV Project | 1st TRAINING COURSES

5
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 MAINTENANCE: some definitions

The maintenance (Scheduled and Corrective) activities may be:

• HARDWARE maintenance: action carried out directly on field by operators with specialized or not
specialized tools such as screwdrivers etc.

• SOFTWARE maintenance: actions carried out remotely or on field with computers and communication
instruments

The Scheduled activities are:

• Inspections → Visual, Mechanical, Electrical, Software
• Measurements and Tests → Electrical, Mechanical, Software
• Substitution of components

GoPV Project | 1st TRAINING COURSES

6
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 Benchmark Costs for O&M Global Service
O&M Service Provider in EGP Solar Fleet

• Full-scope contracts
are no longer the
norm.

Note: Global Service is an awarded service to an external contractor dealing with one or more maintenance services
 Benchmark Costs for Solar Inverter Maintenance
Inverter Scheduled Maintenance costs in EGP Countries

• The O&M contracts are related to the Inverter centralized into the CU.
Benchmark Costs for Module Cleaning
OPEX costs in EGP Countries

• Yearly fees depend

from n° of cleanings
(*) and manpower costs
in the Country
(especially for manual
cleaning).

• Local content
obligations affect the
service awarding.

(*) fee includes also grass cutting

Benchmark Costs for Grass Cutting
OPEX costs in EGP Countries

• Yearly fees depend from n° of grass cuttings and

manpower costs in the Country (especially for
manual cleaning).
 PREDICTIVE MAINTENANCE

Benefits of Predictive Maintenance

Scheduled
Condition

maintenance ➢ Avoid severe damage of equipment and propagation

to other components

➢ In some cases completely prevent the occurrence of

Incipient damage
failure detection Predictive
by predictive maintenance
➢ Optimized maintenance planning
Corrective
maintenance
➢ Lower repair costs
Failure
undetected
➢ Reduced devices downtime and relative revenue loss,
Time
particularly for inverters

GoPV Project | 1st TRAINING COURSES

11
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 IPF Curve
Installation, Point of Defect, Failure

LEADING LAGGING

GoPV Project | 1st TRAINING COURSES

TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS
28/10/2020
October 26-29th 2020
 Instrumentations and sensors on PV Field

Solar PV Field is almost totally composed from pure electronic components, only newest plant has one mechanical
movement actuated by trackers who follow the sun. Most important sensors are electrical and environmental types.

Conversion Unit IDC A Analog Input 1 Current (A)

VDC V Analog DC Input Voltage (V)
IAC A Analog Grid current (Average of phase 1,2,3)
INVERTER
VAC V Analog Grid Voltage (Average of phase r,s,t)
• DC and AC Current Sensors
PAC kW Analog Inverter AC Power (Average of phase r,s,t)
• NTC/PT100 sensors
T_IGBT °C Analog IGBT Temperature Measure (°C)
• Voltage AC and DC Transductors
FAN_SPEED rpm Analog fan_speed (RPM)

ENVIRONMENTAL SENSORS
• PT 100 GTI W/m2 Analog Irradiance Tilted from pyranometer (or solar cell)

• Pyranometers (Secondary Standard class) TEMP_M °C Analog PV Module Temperature

GTI W/m2 Analog Irradiance Tilted from pyranometer (or solar cell)
Meteo Station GHI W/m2 Analog Irradiance Horizontal from pyranometer (or solar cell)
DHI W/m2 Analog Diffused Horizontal Irradiance
• Pyranometer (Secondary Standard class) TEMP_AMB °C Analog Enviroment Temperature
TEMP_M °C Analog PV Module Temperature
• PT100
WS m/s Analog Wind Speed
• Humidity Sensor WD ° Analog Wind Direction
• Rain Gouge RAIN mm Analog Pluviometer
• Wind direction and intensity sensor SNOW mm Analog Snow
RH kg/m3 Analog Relative Humidity
• Soiling sensor PRESSURE Pa Analog Atmospheric Pressure
Tilt ° Analog Tilt of the relative Sensor
Azimuth ° Analog Azimuth of the relative Sensor
Tracker
for each one
ROLL_POSITION ang.deg. (°) Analog Actual Position of Solar String
• Position transducer TARGET_POSITION ang.deg. (°) Analog Target Position of Solar String
• Wind sensors WS m/s Analog Wind Speed

October 26-29th 2020

 Operational Efficiency process strongly focused on
inverter performances
4.000 Reference Power Curve for single MPPT of each Inverter created through fully in-house developed algorithms
based on neural network approach and statistical analysis on historical data and updated on yearly bases

Up to 1-minute signals Lost production calculation at single MPPT level

gathering from each plant Reference power curve vs 15minute actual point

Inverter measurements
from gathering systems (WMR/SCADA)
No Lost Production:
Actual Point over the
• Voltage, Current, Power Lower acceptance limit
• Both DC and AC side measurement
Producible =
expected output
Environmental measurements

AC Power
using reference
from field sensors power curve (μ)
• Total irradiance in the plane of the array
• Ambient air temperature in a radiation shield Lost
• Module temperature of the panel back-side Production

Solar Log Book LP: Actual Point under the

to detect and classify unavailability events Lower acceptance limit
• Unavailability events log book
• Created on daily basis from SCADA alarms
• Editable by supervisor with undetected alarms Irradiance

Dedicated algorithms developed in-house to detect underperformances for each inverter GoPV Project | 1st TRAINING COURSES
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS
with specific focus on lost production caused by soiling and grid limitation 14
October 26-29th 2020
 Solar Big Data Predictive: technical overview
Two approaches in parallel

Predictive models developed at Inverter level aimed to provide warnings for runtime operations. Two approaches have been investigated:

1. Event Detection (Diagnosis Model): predict inverter deviations from nominal behaviour

2. Advanced Fault Recognition: predict a specific class of faults (e.g. ground fault, bulk cap fail, ..)

Inverter signals 1.
Event
Detection
model

Worldwide Monitoring Events – operator

2.
Room Data dashboards
Fault
Recognition
Inverter signals model
GoPV Project | 1st TRAINING
15 COURSES
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 Solar Big data Predictive
Plant 1 (Romania) business case

Advanced modeling for inverter anomalies early detection

Deviation of expected vs real AC power

➢ The draining-off of the thermal paste due to

heat caused overheating and breakage of the
IGBT (Insulated Gate Bipolar Transistor) of the
inverter

➢ Predictive model of expected vs real AC power

could have allowed inverter fault early
detection and repair, avoiding 2 months of
lost production

~55k€ of potential avoided lost revenues thanks to predictive model implementationGoPV Project | 1st TRAINING
16 COURSES
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS
16
October 26-29th 2020
 Innovative Solutions in the Solar Technology
PV Field key features

Forecast & Nowcast production Drones (for security and inspections) Plant Managment System
Data Modeling Cleaning Machine Remote Assistance
Cloud Computing Grass Cutting Machine (Smart Glasses)
Exoskeleton

GoPV Project | 1st TRAINING COURSES

TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS
28/10/2020 17
October 26-29th 2020
 Solar Digitalization
Data Model
The Data Model is a Cross Technology project, started from Hydro Tech, suitable to achieve a new structure of
Objective information (tags) gathered from existing, and new, power plants with an object oriented language. This new way to
structure data is made follow two main innovative international standard regulations.

1st STEP

The first part of the activity is to convert the original Tag list of
each Inverter Manufacturer to the new EGP Data Model. The
choosen model is streghtly related to the Hydro DM, compliant
with RDS-PP and IEC61850. After deeper analisys we knew that
there aren’t any IEC Standard law for Solar Tech; but an
Alliance of many inverter supplier built a Technical
Documentation called sunSPEC who define also Data
Modelling.
Physical
object
2nd STEP
Apply the Data Modelled Tag to the as-built tag list for each
plant of all over the world. The activity will carry out a tag
renaming in the Pi Archive.
ISO/TS 81346: The same plant component (Object) can be
Object-data considered in a different way depending on the point-of-view
(Aspect) of the observer; designation semantics remain
unchanged in all the aspects GoPV Project | 1st TRAINING COURSES
28/10/2020
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS
Functional diagram
October 26-29th 2020
 Solar Digitalization
Cloud Computing* - Zone Monitoring: «EYE»

The Conversion Unit supplier doesn’t give us a well performing Zone Monitoring Algorithm, so thanks to O&M
Objective experience in graphical analysis of current behavior, we have choosen the idea to build an internal algorithm.

Problem: O&M Local team cannot

find underperformance strings due String Boxes Currents
to passive string boxes.
180
160

Centralization of Zone Monitoring, this 140

adaptive algorithm is standard and How to translate an
underperforming string 120
unique
boxes due to tracker100
Easy and friendly analysis on the unavailability in an
data algorithm or code? 80
60
Home made without additional costs
using company software (SAS) and 40
without external consulting (saving
20
CAPEX/OPEX)
0
Serie1 Serie2 Serie3 Serie4 Serie5
*Cloud computing :The term is generally used to describe data centers available to many users over the Internet.
GoPV Project | 1st TRAINING COURSES
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 Solar ROBOTIZATION
Actual PV panels Cleaning
Chile Romania, Greece Mexico South Africa Peru Brazil

BPMetalmeccanica Truck
(Carrera Pinto, Finis Terrae)

SunBrush Tractor
Roboklin25 Tractor SunBrush Tractor
(Pampa Norte)

Manual Robot (dry, under test) Manual dry and wet SunBrush Tractor (dry) Alexil semi robotic
Manual
(Diego de Almagro, Chanares, (under test)
La Silla, Lalackama I and II) October 26-29th 2020
 Solar ROBOTIZATION
Robot for module cleaning - Activities’ status

• Solution for fully automatic cleaning of photovoltaic modules

• Dry cleaning for sustainability solar power plant
Goal • Reduce up to 90% of soiling impact
• Increase Annual Energy yield (kWh)

• Use Case 1: Test for technical KPI evaluation and cleaning performance
of first prototypes:
→ Done in Passo Martino Lab -Catania
• Use Case 2: Test for golden sample approval:
Activities → Done in Passo Martino Lab -Catania
• Use Case 3: Test in real field: using a sub-field (1 MW) inside Pampa
Norte, Chile, 80 MW photovoltaic power plant:
→ Planned to be confermed.

3 distinct prototypes developed and tested according with GPG technical

specification:
• Excellent results in cleaning performance for all the samples;
First Results • WashPanel: difficulties when 2 consecutive trackers are not aligned;
• Crel: high Robot weight;
• Reiwa: compliant with all the examinated KPIs.

• Analysing the result of the test in real field, GPG'll evaluate the real
Next Steps benefits of this solution and its application during the design of new
PV plants and to be applied for some existing pv plants .

October 26-29th 2020

 Solar ROBOTIZATION
Drones and Solar thermal analysis with AI (Roboost)
Aerial thermography is the practice of assessing and monitoring photovoltaic (PV) system condition using data captured via an
aircraft equipped with a thermal camera.

Case1 Analysis of the images Increase Safety of

EGP’s drone O&M activities on site
EGP’s pilot Thermographic and through a AI Software
RGB images
More hours on high-
added-value activities
Case2
Fly drone by Operational
Ext. Contractor O&M Costs reduction

Lost Production
decrease
Case3
EGP
autonomous
Progressive discounts until 40% of 4th inspection
drone Level I Level II Level III

For each inspected MW* Module,String-level

Diode & hot spots Individual cell
600€ costs saving outages GoPV Project | 1st TRAINING COURSES
* Purchased drones + Framework Agreement Raptor Maps
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

Solar ROBOTIZATION
Drones and Solar thermal analysis with AI (Roboost)

October 26-29th 2020

 Solar - Pills
Other initiatives
Smart Glasses COMAU MATE exoskeletal device for shoulder support
Global Framework Agreement for Smart
Glasses devices and Augmented Reality Support the operator during the panel positioning,
solutions closed with Brochesia company fixing on Tracker structure and PV panel’s cleaning

Functionality activities, in which the shoulder joint reaches large

elevation angles reducing muscular fatigue and
• Remote assistance & «Hands-free»
maintenance support articulation overload.

• Training «on the job»

• Checklists & Safety procedures

Autonomous Grass Cutting

Project ongoing in Greece with the

goal to build autonomous
customized grass cutting systems
with related management software
algorithms.

GoPV Project | 1st TRAINING COURSES

TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020

 This project has received funding from the European Union’s Horizon 2020
research and innovation programme under grant agreement No 792059

Thank you for your attention!

Follow us
 BIBLIOGRAPHY

• Operation & Maintenance - Best Practices Guidelines V4.0 (Solar

Power Europe);

GoPV Project | 1st TRAINING COURSES

26
TECHNICAL FOCUS ON FUTURE SOLAR PV SYSTEMS

October 26-29th 2020