TY - JOUR
T1 - A fully transient novel thermal model for in-field photovoltaic modules using developed explicit and implicit finite difference schemes
AU - Aly, Shahzada Pamir
AU - Barth, Nicolas
AU - Figgis, Benjamin W.
AU - Ahzi, Said
N1 - Publisher Copyright:
© 2018 Elsevier B.V.
PY - 2018/7
Y1 - 2018/7
N2 - Three fully transient numerical thermal models have been developed for photovoltaic (PV) modules in MATLAB environment, using 2-D finite difference (FD) method. One of the thermal FD model is based on explicit time scheme, while the other two are based on implicit time schemes. Out of the two implicit FD models, one has been modeled using preexisting toolboxes of MATLAB, while the other has been modeled using a self-developed novel method. All the three FD models are based on energy balance of different control volumes, which as a whole constitute the complete solid domain of the PV panel. The models have been tested against a variety of experimental data, ranging from sunny clear days, sunny cloudy days, rainy overcast days and consecutive sunny clear days. All the three models are found to agree very well with experimental results, i.e. the errors between the modeled and experimental data ranges between 0.2–0.7 °C. The main difference is between their computational speeds. In terms of average execution time per iteration for transient analyses, the self-developed novel implicit method (referred to as implicit BB throughout the work) was about 1200 times slower than the explicit method. However, overall, the implicit BB method took less time for the entire transient analyses, as it requires less number of iterations due to its tolerance to adapt longer time steps for each iteration. In other words, the explicit method although slightly more accurate, took approximately 900 times more CPU time to simulate the same time span test compared to implicit BB method. Thus, the self-developed novel method (implicit BB) has been recommended for these types of thermal models.
AB - Three fully transient numerical thermal models have been developed for photovoltaic (PV) modules in MATLAB environment, using 2-D finite difference (FD) method. One of the thermal FD model is based on explicit time scheme, while the other two are based on implicit time schemes. Out of the two implicit FD models, one has been modeled using preexisting toolboxes of MATLAB, while the other has been modeled using a self-developed novel method. All the three FD models are based on energy balance of different control volumes, which as a whole constitute the complete solid domain of the PV panel. The models have been tested against a variety of experimental data, ranging from sunny clear days, sunny cloudy days, rainy overcast days and consecutive sunny clear days. All the three models are found to agree very well with experimental results, i.e. the errors between the modeled and experimental data ranges between 0.2–0.7 °C. The main difference is between their computational speeds. In terms of average execution time per iteration for transient analyses, the self-developed novel implicit method (referred to as implicit BB throughout the work) was about 1200 times slower than the explicit method. However, overall, the implicit BB method took less time for the entire transient analyses, as it requires less number of iterations due to its tolerance to adapt longer time steps for each iteration. In other words, the explicit method although slightly more accurate, took approximately 900 times more CPU time to simulate the same time span test compared to implicit BB method. Thus, the self-developed novel method (implicit BB) has been recommended for these types of thermal models.
KW - 2D finite difference
KW - Explicit time scheme
KW - Implicit time scheme
KW - PV modules
KW - Thermal model
KW - Transient model
UR - http://www.scopus.com/inward/record.url?scp=85040511899&partnerID=8YFLogxK
U2 - 10.1016/j.jocs.2017.12.013
DO - 10.1016/j.jocs.2017.12.013
M3 - Article
AN - SCOPUS:85040511899
SN - 1877-7503
VL - 27
SP - 357
EP - 369
JO - Journal of Computational Science
JF - Journal of Computational Science
ER -