© Copyright 2013, First Solar, Inc.
2
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Isis-2: Energy Model Integrated with Business Systems
Parametric Generator
Cost Model
Isis Energy Model
Financial Model
Optimized Plant
Design & Layout
Sales Tool
Fleet Performance
Database
Expected Performance
Reporting
Energy Prediction Map
3
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Key Differentiating Features
Spectral Shift Native implementation – traditionally backed into with other gain/loss factors
Soiling Ramped model with rain-triggered and manually-triggered cleanings
Lifetime analysis DC-side degradation in voltage & current Multi-year performance estimates/performance analysis
Module temperature Transient model taking into account all heat fluxes
Inverter Redefined as state engine with zones User-selectable maximum power setpoint with temperature & elevation derate Efficiency curves at many voltages
Plant architecture Block-by-block breakdown with independent module characteristics, DC:AC loading factors, etc. with staggered installation & energization schedule for key financial analysis
Time Scale Sub-hourly modeling to better avoid modeling artifacts due to weather averaging (inverter clipping) Improved power plant analysis
Application Multi-user web application with shared components library Secure database of simulations results Integrated with other business systems
4
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Simulation
5
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Representation of a Power Plant
Preliminary?
As-Built?
Contractual?
Re-Usable Components
6
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Soiling Model with Ramp Rates & Cleaning Events
Direct Monitoring of Energy Lost due to Soiling on First Solar Modules in California, 38th PVCS
7
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
CdTe Spectral Shift
Changes in Cadmium Telluride Photovoltaic System Performance due to Spectrum, 38th PVSC
H2O Absorption Bands
Monthly and hourly
calculation of the CdTe
spectral response using
precipitable water or
relative humidity
8
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Shading & Incidence Angle Modifier Losses
Increased Energy Production of First Solar Horizontal Single-Axis Tracking PV Systems without Backtracking, 39th PVSC
0
100
200
300
400
500
AC
Pow
er
(kW
)
Median System
System C
System D
100
200
300
400
500
AC
Pow
er
(kW
)
Median System
System C
System D
0
100
200
300
400
500
AC
Pow
er
(kW
)
Median System
System C
System D
9
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Dynamic Thermal Model
A Time Dependent Model for Utility Scale PV Module Temperature, 40th PVCS
20
30
40
50
60
70M
odule
Tem
pera
ture
, C
els
ius
Hourly Avg. Module Cell Temperatures; Desert Southwest during the Summer
Measured
PVsyst Default Model
Updated PVsyst Model
FSLR Model
Reduces irradiance-weighted Tmod error at hot climate
sites from 3.2 °C (RMSE) to 1.5-2.0 °C, overall RMSE
reduction by 51%, MBE by 30%
Measured Default Static Model Updated Static Model Dynamic Model
10
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Non-Linear Temperature Coefficient
𝑖 = 𝐼ph − 𝐼0 𝑒𝑉d
𝑛cs ∙𝛾∙𝑉th − 1 −𝑉d
𝑅sh
− 𝐼ph ∙ 𝑏1
𝑛cs𝑉bi − 𝑉𝑑
𝑉d = 𝑣 + 𝑖 ∙ 𝑅s
11
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Array – Inverter Interaction
Once module temperature is computed, the
1-diode coefficients are temperature corrected, and the array MPP is solved
However, the inverter behavior is dynamic as well:
• Efficiency is a function of V & P
• Max capacity is a function of temperature and elevation
What inverter “zone” are we operating at?
12
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Parametric Analysis for Power Plant Design Optimization
13
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Benchmarking Goals
To quantify system-level prediction accuracy by using a well-
understood subset of extremely high-confidence data sets
• Deep look at performance to prediction
• Error analysis of energy prediction Reporting
• Feedback for energy model development by re-benchmarking
• Test-bed for advanced analytic methods
Continuous Improvement
• Demonstrate that Isis will hit the P50 for an ensemble of plant performance analyses in different climates & configurations
Acceptance
14
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Monitoring Points
Evaluating accuracy and error step-by-step through the model
with a final goal of understanding WHY prediction error occurs at
the energy meter
Energy Meter
AC Power
Inverter Efficiency
DC Power
DC Voltage
DC Current
Module Surface Temperature
Plane of Array Irradiance
Gigawattsbytes of powerdata reduced into
meaningful, easy-to- interpret results
15
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
-5 -4 -3 -2 -1 0 1 2 3 4 50
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
Model Error (%)
Num
ber
of
Syste
ms
Average Error = 0.43%
StDev Error = 2.53%
Span Error = 7.75%
Prediction Accuracy at the Energy Meter
Isis-2 overpredicted energy by 0.43% on average with a standard
deviation of 2.53%
-5 -4 -3 -2 -1 0 1 2 3 4 50
1
2
3
4
5Energy Meter
Model Error (%)
Num
ber
of
Syste
ms
Average Error = 0.06%
StDev Error = 1.96%
Span Error = 6.72%
Introduction of advanced lifetime model of voltage and current reduced
the error to 0.06%
10 sites | > 375 MW of PV modules | 15 system-years
3
2
1
No
. Sys
tem
s
16
© C
op
yrig
ht
20
13
, Fi
rst
Sola
r, I
nc.
Look for First Solar at the 40th IEEE PVSC in Denver!
• “Evaluation of GHI to POA Models at Locations across the United States”
— Joint effort with SNL to quantify accuracy of irradiance decomposition/transposition models
• “A Time Dependent Model for Utility Scale PV Module Temperature”
— Updated module temperature module reduces RMSE by 51% and MBE by 30%
• “Measuring Soiling Losses at Utility-scale PV Power Plants”
— Continuing collaboration with Atonometrics to measure soiling losses
• “Performance Characterization of Cadmium Telluride Modules Validated by Utility Scale and Test Systems”
• “Self-Reported Field Efficiency of Utility-Scale Inverters”
• “Spectral Mismatch Considerations in Multi-irradiance Characterization of PV Modules”
• “Evaluation of a CdTe Spectrally Matched c-Si PV Reference Cell for Outdoor Applications”
• “Regional Atmosphere-Solar PV Interactions”