RER Renewable Energy Research  français
   Overview       Expertise       Publications       Resources        Photos    
    All       By Topic       Articles       Reports       Non-technical       Presentations    

PVToolbox Simulation Output Compared with Monitored Data
from PV Hybrid Test Bench

Michael M.D. Ross
RER Renewable Energy Research

Dave Turcotte and Marc-André Fry
CANMET Energy Technology Center-Varennes (Natural Resources Canada)

Full Text Article
Link to SESCI (conference)


Research conducted on behalf of the Photovoltaics and Hybrid Systems Program at the CETC-Varennes (Natural Resources Canada) with partial funding from the Panel on Energy Research and Development (PERD).


Ross, Michael M. D., Dave Turcotte, and Marc-André Fry. "PVToolbox Simulation Output Compared with Monitored Data from PV Hybrid Test Bench". Proceedings of the 30th Annual Conference of the Solar Energy Society of Canada, Burnaby, B.C., August 20 to 24, 2005.


The CETC-Varennes has developed two advanced tools for the investigation of photovoltaic hybrid system behaviour. First, PVToolbox is a library of component models facilitating the simulation of hybrid systems in the Matlab® Simulink® environment. Second, a configurable test bench consisting of engine-driven generator ("genset"), array, battery, and control and monitoring equipment has been built on the CETC-Varennes premises.

Data from the hybrid test bench has been very useful in validating the PVToolbox and identifying those areas where it is accurate and those areas where it is not. Thus far, the accuracy of PVToolbox has been quite good. While the goal during development of PVToolbox was that it predict major energy flows (e.g., photovoltaic array output accepted by the system and electrical energy provided by the genset) with an error of 10% or lower, in some recent tests errors of around 1 to 2% -roughly the accuracy of the measuring devices-has been observed. Not only were average energy flows accurate, but the timing of genset starting and stopping was closely modelled by the simulation.

This study shows that while the simulation tool can be quite accurate, it will not, in general, achieve this 1 to 2% error level. When variable loads are used in place of constant current loads, or when voltage thresholds instead of a time criterion determines the end of genset operation, simulated behaviour does not coincide with reality so well: errors in energy flows of 5 to 8% are more typical.

Two aspects of the simulation tool merit further attention. First, the battery model is less accurate under conditions of changing current. The reasons for this are explored; it is expected that the limitations associated with this battery model also apply to many other published models. It is proposed, as a consequence, that the current accepted by a battery under constant voltage charging be used as a test of any battery model. Second, the algorithm for dividing solar irradiance on the horizontal into beam and diffuse components is somewhat inaccurate. This shortcoming has been explored elsewhere.