On the design and monitoring of photovoltaic systems for rural homes

It is estimated that 1.6 billion people today live without access to electricity. Most of these people live in remote rural areas in developing countries. One economic solution to this problem is the deployment of small domestic photovoltaic (PV) systems called solar home systems (SHS). In order to improve the performance and reduce the life cycle cost of these systems, accurate monitoring data of real SHSs is required. To this end, two SHSs typical of those found in the field were designed and installed, one in a rural area of the Eastern Cape of South Africa and the other in the laboratory. Monitoring systems were designed to record energy ows in the system and important environmental parameters. A novel technique was developed to correct for measurement errors occurring during the utilization of pulse width modulation charge control techniques. These errors were found to be as large as 47.6 percent. Simulations show that correction techniques produce measurement errors that are up to 20 times smaller than uncorrected values, depending upon the operating conditions. As a tool to aid in the analysis of monitoring data, a PV performance model was developed. The model, used to predict the maximum power point (MPP) power of a PV array, was able to predict MPP energy production to within 0.2 percent over the course of three days. Monitoring data from the laboratory system shows that the largest sources of energy loss are charge control, module under performance relative to manufacturer specifications and operation of the PV array away from MPP. These accounted for losses of approximately 18-27 percent, 15 percent and 8-11 percent of rated PV energy under standard test conditions, respectively. Energy consumed by loads on the systems was less than 50 percent of rated PV energy for both the remote and laboratory systems. Performance ratios (PR) for the laboratory system ranged from 0.38 to 0.49 for the three monitoring periods. The remote system produced a PR of 0.46. In both systems the PV arrays appear to have been oversized. This was due to overestimation of the energy requirements of the loads on the systems. In the laboratory system, the loads consisting of three compact fluorescent lamps and one incandescent lamp, were used to simulate a typical SHS load pro le and collectively consumed only 85 percent of their rated power. The 8 predicted load profile for the remote system proved to be signi cantly overestimated. The results of the monitoring project demonstrate the importance of acquiring an accurate estimation of the energy demand from loads on the system. Overestimations result in over-sized arrays and energy lost to charge control while under-sized systems risk damaging system batteries and load shedding. Significant under-performance of the PV module used in the laboratory system, underlines the importance of measuring module IV curves and verifying manufacturer specifications before system deployment. It was also found that signi cant PV array performance gains could be obtained by the use of maximum power point tracking charge controllers. Increased PV array performance leads to smaller arrays and reduced system cos