Research Papers

Monitoring Solar Home Systems With Pulse Width Modulation Charge Control

[+] Author and Article Information
Nathaniel J. Williams, E. Ernest van Dyk, Frederik J. Vorster

Centre for Energy Research, Department of Physics, Nelson Mandela Metropolitan University, Port Elizabeth 6031, South Africanathaniel.williams@nmmu.ac.za

J. Sol. Energy Eng 133(2), 021006 (Mar 23, 2011) (7 pages) doi:10.1115/1.4003586 History: Received June 14, 2010; Revised December 04, 2010; Published March 23, 2011; Online March 23, 2011

With the high cost of grid extension and approximately 1.6 billion people still living without electrical services, the solar home system is an important technology in the alleviation of rural energy poverty across the developing world. The performance monitoring and analysis of these systems provide insights leading to improvements in system design and implementation in order to ensure high quality and robust energy supply in remote locations. Most small solar home systems now use charge controllers using pulse width modulation (PWM) to regulate the charge current to the battery. A rapid variation in current and voltage resulting from PWM creates monitoring challenges, which, if not carefully considered in the design of the monitoring system, can result in the erroneous measurement of photovoltaic (PV) power. In order to characterize and clarify the measurement process during PWM, a mathematical model was developed to reproduce and simulate measured data. The effects of matched scan and PWM frequency were studied with the model, and an algorithm was devised to select appropriate scan rates to ensure that a representative sample of measurements is acquired. Furthermore, estimation methods were developed to correct for measurement errors due to factors such as nonzero “short circuit” voltage and current/voltage peak mismatches. A more sophisticated algorithm is then discussed to more accurately measure PV power using highly programmable data loggers. The results produced by the various methods are compared and reveal a significant error in the measurement of PV power without corrective action. Estimation methods prove to be effective in certain cases but are susceptible to error during conditions of variable irradiance. The effect of the measurement error has been found to depend strongly on the duty cycle of PWM as well as the relationship between scan rate and PWM frequency. The energy measurement error over 1 day depends on insolation and system conditions as well as on system design. On a sunny day, under a daily load of about 20 A h, the net error in PV energy is found to be 1%, whereas a system with a high initial battery state of charge under similar conditions and no load produced an error of 47.6%. This study shows the importance of data logger selection and programming in monitoring accurately the energy provided by solar home systems. When appropriately considered, measurement errors can be avoided or reduced without investment in more expensive measurement equipment.

Copyright © 2011 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.



Grahic Jump Location
Figure 1

Simplified charge control and dc measurement circuit. During PWM, the PWM shunt (a MOSFET), represented in the figure as a switch, is closed in the short circuit state and open in the charging state. dc measurements are take by the data logger across shunts and voltage dividers as indicated.

Grahic Jump Location
Figure 2

During PWM, PV voltage peaks correspond to PV current valleys, creating two square waves of identical periods and duty cycles

Grahic Jump Location
Figure 3

Data logger scans at the beginning of PWM. Mismatch of current and voltage peaks is particularly prevalent at very high and very low duty cycles. In this example at high duty cycle, several mismatched peaks can be seen as well as partial peaks integrated over pulse edges.

Grahic Jump Location
Figure 4

Series of scans of PV current and voltage under PWM made by the data logger. The combination of duty cycle, scan rate, and PWM frequency cause distinct patterns in measurements, demonstrating that the relationship between scan rate and PWM frequency can produce a nonrepresentative sample of measurements. In the upper graph, the scan rate was 920 ms, the PWM frequency was 9.25 Hz, and the duty cycle was 44%. The lower graph has a scan rate of 930 ms, PWM frequency of 9.25 Hz, and duty cycle of 43%.

Grahic Jump Location
Figure 5

Using duty cycle, scan rate, and PWM frequency data in the examples in Fig. 4, the developed model successfully reproduces the patterns observed in measured data

Grahic Jump Location
Figure 6

By placing the pulse edges over one record interval on a 1 s scan interval, we can see that with a frequency of 9.2505 Hz, 37 broad bands are formed in the upper half of the figure, leaving gaps on the scan interval, which, with small enough duty cycle, will be left empty and result in a constant low measurement of power. At 9.25 Hz, seen in the lower half of the figure, the bands shrink to single pointed nodes. This results because 37 is divisible by both 1 and 9.25.

Grahic Jump Location
Figure 7

With a scan rate of 1 s and PWM frequency of approximately 9.25 Hz, there is a risk of getting stuck, making measurements in the PWM short circuit state as evidenced here. It can be seen that maximum voltage (solid ) and average voltage (dashed) correspond in a short circuit state for a period of about 25 min around 8:15 and then again during two shorter intervals around 8:55 and 9:35, indicating that no peaks were measured. Similarly, the average current (dotted) and minimum current (dot-dashed) also correspond at an elevated level on these intervals, indicating measurement at short circuit. Because the frequency of the PWM is not exactly constant, it is possible to drift in and out of this state.

Grahic Jump Location
Figure 8

Scan rate optimization algorithm

Grahic Jump Location
Figure 9

On a typical mostly sunny day, the battery will reach its gassing voltage and begin PWM between 12:00 and 13:00 depending on the daily load and previous state of charge. Pest1 (dotted) can be seen deviating from the true power during periods of varying insolation. Pest2 (long-dashed) also has problems with high irradiance variability, significantly underestimating power during steady decline in irradiance in the late afternoon. Measured power (solid) and the corrected power (dashed) follow similar trends, with deviation increasing with decreasing duty cycle.

Grahic Jump Location
Figure 10

Using the same data as in Fig. 9, we can see that the error in measured PV power depends highly on duty cycle




Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In