1. DEVELOPMENT OF NEW SOFTWARE
TO ANALYSE AND PREDICT THE
MODULE PERFORMANCE
Susana Iglesias Puente
2. OUTLINE
Performed tasks
• Indoor measurements
• Outdoor measurements
• Data treatment software
Conclusions
Suggested Software improvements
3. INDOOR MEASUREMENTS
PASAN LAPSS Laboratory
The module is characterised at each point on a matrix of
Pmax (W) as a function of Tmod and irradiance
Final result: empirical equation to estimate Pmax
result:
a + b ⋅ ln Irr + c ⋅ T AMB
Pmax =
1 + d ⋅ ln Irr + e ⋅ (ln Irr ) + f ⋅ T AMB
2
5. OUTDOOR MEASUREMENTS
Tracker:
Tracker: avoid the
effects of the angle of
incidence.
incidence.
Irradiance measured by
two different kinds of
devices:
devices: Pyranometer
and ESTI sensor.
sensor.
6. OUTDOOR MEASUREMENTS
Rack: in-plane
in-
measurements.
Result:
Result: text files
storing the different
variables involved in
module performance.
performance.
8. DATA TREATMENT SOFTWARE
Main task: development of specific software to
treat the data from the outdoor measurements.
Employed software: Matlab.
9. OUTDOOR MEASUREMENTS: data treatment
MEASUREMENTS:
The text files contain data for several years,
therefore they are large.
The data treatment using a
spreadsheet is impractical.
Solution: creation of special software to treat
the outdoor data systematically.
10. DATA TREATMENT SOFTWARE
Main objectives:
objectives:
• Obtain the values of the measured and estimated
energy produced by the module
• Obtain the energy coming from the sun (irradiation)
• To be able to calculate these at different time
intervals, e.g. day, month, year, etc.
etc.
• Compare measured and estimated energies, and
other output results numerically and graphically
11. DATA TREATMENT SOFTWARE
Software developed to treat the data:
• Solar_data_treatment
• Data_writing
• Data_plotting
• Eq_fit_params
• NOCT_estimation
• Montly_sum
• Month_teller
12. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file.
file.
• Obtain the parameters of the empirical equation.
equation.
• Estimate Tmod from Tamb and Irradiance.
Irradiance.
• Estimate Pmax values with the empirical equation.
equation.
• Integrate Pmax over the day and store the results in
3-D arrays.
arrays.
• Obtain the energy for every month and every year.year.
• Calculate the BIAS error and the module efficiency.
efficiency.
13. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file.
file.
• Obtain the parameters of the empirical equation.
equation.
• Estimate Tmod from Tamb and irradiance.
irradiance.
• Estimate Pmax values with the empirical equation.
equation.
• Integrate Pmax over the day and store the results in
3-D arrays.
arrays.
• Obtaining the energy for every month and every year.
year.
• Calculate the BIAS error and the module efficiency.
efficiency.
15. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file. file.
• Obtain the parameters of the empirical equation. equation.
• Estimation of Tmod with the empirical equation.
equation.
• Estimation of Pmax a + b ⋅ ln Irr + c ⋅ T AMB
Used function: Eq_fit_params the empirical equation.
values with equation.
Pmax =
• Integration of Pmax over theln Irr )andf storing the results
1 + d ⋅ ln Irr + e ⋅ ( day + ⋅ T AMB
2
in 3-D arrays.
arrays.
• Obtaining the energy for every month and every year. year.
• Calculation of the BIAS error and the module efficiency.
efficiency.
16. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file.file.
• Obtain the parameters of the empirical equation. equation.
• Estimate Tmod from Tamb and Irradiance. Irradiance.
• Estimate Pmax values with the empirical equation. equation.
NOCT − day
Used function:MODover the 20 ⋅ Irrand
Integrate PmaxNOCT_estimation store the results in
•
T = + T AMB
3-D arrays.
arrays. 800
•
Nominal Operatingfor every month and every year.
Obtain the energy Cell Temperature year.
• Calculate the BIAS error used in empirical eqn, efficiency.
- Necessary because Tmod is
and the module efficiency.
not Tamb
17. NOCT − 20
TMOD = ⋅ G + T AMB
800
DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file. file.
• Obtain the parameters of the empirical equation.
equation.
• Estimate Tmod with the empirical equation.equation.
• Estimate Pmax values with the empirical
equation.
• Integrate Pmax over the day and store the results in
- ESTI irrad (& measured Tmod)
3-DPyran irrad (& measured Tmod)
- arrays.
arrays.
- ESTI irrad (& estimated Tmod using NOCT)
• Obtain the energy for everyusing NOCT) every year.
- Pyran irrad (& estimated Tmod month and year.
• Calculate the BIAS error and the module efficiency.
efficiency.
18. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file.
file.
• Obtain the parameters of the empirical equation.
equation.
• Estimate Tmod with the empirical equation.
equation.
• Estimate Pmax values with the empirical equation.
• Integrate Pmax and irradiance over the day and
store the results in 3-D arrays.
• Obtain the energy for every month and every year.
year.
Used function: Monthly_sum
• Calculate the BIAS error and the module efficiency.
efficiency.
19. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Data storage:
One 12x31 matrix containing the energy for every day
for each year.
These matrices are stored in the same variable to form
a 3-D array (tensor) for a number of years.
20. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file.
file.
• Obtain the parameters of the empirical equation.
equation.
• Estimate Tmod with the empirical equation.
equation.
• Estimate Pmax values with the empirical equation.
equation.
• Integrate Pmax over the day and store the results in
3-D arrays.
arrays.
• Obtain the energy for every month and every year. year.
• Calculate the BIAS error and the module efficiency.
efficiency.
21. DATA TREATMENT SOFTWARE: Solar_data_treatment
SOFTWARE:
Actions carried out by the program:
program:
• Import the data from the text file.
file.
• Obtaining the parameters of the empirical equation.
equation.
• Estimation of Tmod with the empirical equation.
equation.
• Estimation of Pmax values with the empirical equation.
equation.
• Integration of Pmax and irradiance over the day and
storing the results in 3-D arrays.
arrays.
• Obtaining the energy for every month and every year.
year.
• Calculation of BIAS error and module efficiency.
22. DATA TREATMENT SOFTWARE: Data_writing
SOFTWARE:
Actions carried out by the program:
Creating M-files to store the calculated variables.
These data can be easily imported to a
spreadsheet (e.g. Excel) for further analysis.
23. DATA TREATMENT SOFTWARE: Data_plotting
SOFTWARE:
Actions carried out by the program:
Plotting the different variables of interest to study the
module performance.
Bar graphs were chosen instead of scatter/line
graphs.
Month_teller gives the month name that is being
plotted.
24. DATA_PLOTTING:
DATA_PLOTTING: Energy for ai01 in 2003
Module surface area=0.49 m2
Year 2003 NORMALISED MONTHLY ESTIMATES 2003
9000 9000
8000 8000
7000 7000
6000 6000
Energy (W·h)
Energy (W·h)
5000 5000
4000 4000
3000 3000
Mean monthly energy
2000 2000 Measured energy
Measured energy
Empirical ESTI energy
Empirical ESTI energy
Empirical ESTI & Tmod energy
1000 Empirical ESTI & Tmod energy 1000
0 0
1 2 3 4 5 6 7 8 9 10 11 12 2 4 6 8 10 12
Months Months
Main differences: Corrected values: divided by the
Jun, Aug, Oct, Nov, Dec number of days actually measured
per month
25. DATA_PLOTTING:
DATA_PLOTTING: Energy for ai01, Jan 2003
JANUARY 2003
400
Measured energy
• Days without
350
Empirical ESTI energy
Empirical ESTI & Tmod energy measurements
300
• Measurements
250
not carried out
Energy (W·h)
200 the same
150 amount of
100 hours every
50
day.
0
5 10 15 20 25 30
Days
26. DATA_PLOTTING:
DATA_PLOTTING: Energy for ai01, May 2003
MAY 2003
400
350
300
250
Energy (W·h)
200
150
100
Measured energy
Empirical ESTI energy
50 Empirical ESTI & Tmod energy
0
5 10 15 20 25 30
Days
27. DATA_PLOTTING:
DATA_PLOTTING: Efficiency for ai01 in 2003
Year 2003 AMBIENT AND MODULE TEMPERATURE FOR AI01 IN 2003
50
Mean efficiency ESTI
14 Measured energy <> ESTI irrad
45
Estimated ESTI energy <> ESTI irrad
Estimated ESTI & Tmod energy <> ESTI irrad
12 40
35
10
Tem perature (ºC)
Efficiency (%)
30
Energy produced by the module (W ⋅ h )
8
Efficiency (% ) =
25
⋅ 100
6 (
Energy coming from the sun W ⋅ h / 20 2 ⋅ module surface area m 2
m ) ( )
15
4
10 Ambient temperature
2 Measured module temperature
5 Estimated module temperature
0 0
2 4 6 8 10 12 1 2 3 4 5 6 7 8 9 10 11
Months Months
The efficiency is lower in summer time
when the module temperature is higher
28. DATA_PLOTTING:
DATA_PLOTTING: Efficiency for ai01, Jan 2003
JANUARY 2003
Measured energy <> ESTI irrad
14 Estimated ESTI energy <> ESTI irrad
Estimated ESTI & Tmod energy <> ESTI irrad
12
10
Efficiency (%)
8
6
4
2
0
5 10 15 20 25 30
Days
29. DATA_PLOTTING:
DATA_PLOTTING: Efficiency for ai01, May 2003
MAY 2003
Measured energy <> ESTI irrad
14 Estimated ESTI energy <> ESTI irrad
Estimated ESTI & Tmod energy <> ESTI irrad
12
10
Efficiency (%)
8
6
4
2
0
5 10 15 20 25 30
Days
30. OVERVIEW ON NUMERICAL RESULTS
Energy comparison for AI01 (polycrystalline)
Annual energy (W·h) Relative error (%)
Measured value 70345 —
Estimate ESTI 70471 0.18
Estimate Pyran 70740 0.53
Estimate
70890 0.77*
ESTI & Tmod
Estimate
71162 1.16*
Pyran & Tmod
*including December 2003 NOCT estimation with bad Tamb data
31. OVERVIEW ON NUMERICAL RESULTS
Energy comparison for LE02 (monocrystalline)
Relative
Energy, 7 months (W.h)
error (%)
Measured value 29624 —
Estimate ESTI 29401 -0.75
Estimate Pyran 29219 -1.37
Estimate
29395 -0.77
ESTI & Tmod
Estimate
29213 -1.39
Pyran & Tmod
32. ENERGY PREDICTION ON PV-GIS WEB SITE
PV-
Solar irradiation map Energywe use monthlyon
Can prediction based
(of T. Huld & M. Suri) empirical model of c-Si module
averages for energy
rating?
T. Huld new calculations:
Monthly averages based on
our meteo tower data (2003 &
2004).
Assumption: in a month the
energy is the same for every
day.
Calculate expected
instantaneous values from
sun position & airmass.
33. ENERGY PREDICTION ON PV-GIS WEB SITE
PV-
Measured Estimate Relative Estimate Relative
2003
energy (Wh) PV-GIS (Wh) error (%) Pyran (Wh) error (%)
Jan 4500 4928 9.5 4648 3.3
Feb 6007 6650 10.7 6214 3.4
Mar 7831 6032 -23.0 8018 2.4
Apr 6751 6949 2.9 6854 1.5
May 8659 8644 -0.2 8593 -0.8
Jun
“PV-
“PV-GIS type” prediction is good for a long
7006 8033 14.7 6925 -1.2
Jul period of time but not for single months
8202 8344 1.7 8256 -0.8
Aug 7471 8056 7.8 7354 -1.6
Sep 5554 6983 21.2 5513 -0.7
Oct 3546 4229 19.3 3553 0.2
Nov 1328 1964 47.9 1339 0.8
Dec 3513 2835 -14.7 3594 2.3
TOTAL 70367 73398 4.3 70740 0.5
34. CONCLUSIONS (Software)
Systematic treatment of the outdoor measurement data.
data.
Nevertheless, the program is flexible as it can be easily
modified by adding new functions.
functions.
The program can function correctly with missing data.
data.
The results are obtained in far less time than employing
a spreadsheet, and different data sets of different
lengths and from different modules can be easily
analysed.
analysed.
At the same time, the results are more reliable.
reliable.
35. Suggested Software improvements
Check the number of hours during which the
measurements were done for every day.
day.
If ∆t > 6 min, the integration of Pmax is not precise.
precise.
More parameters should be plotted, e.g. irradiance,
BIAS error, mean values, etc.
etc.
There should be taken into account that the empirical
equation to estimate Pmax can change depending on
the module.
module.
36. CONCLUSIONS (Predictions)
The empirical equation from solar simulator gives good
predictions compared with long term outdoor
measurements.
measurements.
2 Crystalline (mono and poly) modules have been
analysed.
analysed.
Comparisons with estimates based on average
irradiance and temperature data (i.e. PV-GIS) are very
(i. PV-
encouraging – proves the validity of using monthly
averages for Energy Rating purposes
37. Thanks to the RE Unit for giving me the
opportunity of participating in their projects
Thanks to Thomas Huld and all my other
colleagues for their assistance
Special thanks to my supervisor
Robert Kenny