Few can question the increasing role of photovoltaic (PV) technology as a scalable, robust means of renewable energy. Yet PV cells are but one element, albeit a critical one, in an overall PV system. Another essential subsystem performs what's called Maximum Power Point Tracking (MPPT).
The concept of MPPT is simple -- to automatically vary a PV array's load conditions so it can produce its maximum output power. This is necessary because a PV cell has non-linear current-voltage qualities. The power delivered by an array increases, to a point, as the current draw rises.
The maximum power point (MPP) is at the knee of the curve. Any additional current drawn from the array results in a rapid drop-off of cell voltage, thus reducing the array power output. The aim of an MPPT subsystem is to determine just where that point is, and to regulate current accordingly. Aging effects and external factors such as temperature and partial shading of an array make locating and tracking the maximum power point a bit more challenging. In most cases there is only one maxima. However, rapid changes in irradiance (moving clouds) or temperature (rain, etc.) may briefly introduce multiple local maxima. Some MPPT techniques address the issue of “phantom” maxima better than others.
One indication of how important MPPT can be to a system's overall economics and conversion efficiency is the number of papers, patents and algorithms describing various approaches. At least 200 papers on MPPT have been published since the late 1960s, and dozens of patents cover various techniques. These techniques differ widely in complexity, tracking accuracy, response time, costs, efficiency and target implementation (i.e., analog circuitry, simple digital circuitry, microcontrollers, DSPs or ASICs). I'll review the general concepts behind five MPPT systems in use today — by no means an exhaustive list. Here we assume a switching dc/dc converter or dc/ac inverter follows the array in all cases.
The first method is known as perturb and observe (P&O). In a P&O system, a small perturbation in array current is introduced at a regular interval and the resultant power is measured. This is usually done by slightly varying the duty cycle of the switching converter (the load) driven by the PV array. Changing the duty cycle changes the load current as well, effecting a small perturbation. Two sensors, one for PV voltage the other for PV current, are commonly used to determine if the perturbation resulted in an increase or decrease of instantaneous power.
In operation, the perturbation will oscillate about the MPP. Some P&O implementations vary the perturbation step size as the system converges, to reduce hunting and increase average efficiency. This can have drawbacks, however, as the small step sizes lengthens system response time to rapid changes in irradiance from moving shadows, etc. Variations on the P&O theme involve using fuzzy logic to adaptively determine step size and microcontrollers to average successive voltage and current readings and thus avoid phantom maxima.
Another widely used MPPT technique is known as incremental conduction, or IncCond. At MPP, the slope of the curve is zero. IncCond exploits the fact that the power curve slope reveals whether the array current has yet to reach MPP (positive slope, dP/dV > 0), is at MPP (dP/dV = 0), or has exceeded MPP and must be reduced (negative slope, dP/dV 0).
Because dP= d(IV), dP/dV = I+V dI/dV. Conductance is I/V. Thus if the system is at MPP, incremental conductance (ΔI/ΔV) =-I/V.
Referring to a simplified flowchart of an IncCond system, we are changing the array current by altering the converter duty cycle (α), similar to the P&O method. If the power converter is a boost configuration, then lengthening its duty cycle will raise inductor current and thus present more current to the array. With each measurement iteration, we adjust the duty cycle by a step size ± Δα, so the next duty cycle becomes αn= αn+1 ± Δ?α. Typically this would be implemented using individual sensors for both current and voltage plus a microcontroller or DSP with integrated a/d converters.
The switching action of a dc/dc converter or dc/ac inverter always results in some input current ripple; this ripple will be apparent in the PV array voltage and current. An interesting MPPT technique called ripple correlation control (RCC) harnesses that imposed ripple current and offers the advantages of quick response time and potentially less expensive analog implementation. RCC correlates the time derivative of power with the time derivative of current or voltage to drive the power gradient to zero, reaching MPP. Essentially, RCC can be viewed as a method of implementing P&O without the need of external perturbation. RCC is useful in certain types of mobile systems which may encounter rapidly changing irradiance, such as solar-powered vehicles that periodically dart under tree shadows.
A PV cell's open circuit voltage will vary under irradiance and temperature conditions with approximate similarity to an array under load. This is the principle behind fractional open circuit voltage (FOCV) and pilot cell methods. It can be realized as simply as
The proportionality constant k depends on the qualities of the particular PV cells being used. In the fractional open-circuit voltage scheme, the array is momentarily disconnected from the converter at regular intervals and the open circuit voltage is measured. Of course this results in a temporary loss of power. An alternative is to use one or more pilot cells which are selected to have the same qualities as the cells in the array. In this case the main array is always connected to the converter and the pilot cells are continuously available for VOpen Circuit measurement. FOCV and pilot cell systems are not true MPPT, however, because they can only approximate MPP. But they are certainly inexpensive to implement and are adequate in certain applications.
These MPPT techniques (and many others) vary greatly in implementation cost, complexity, performance and design time. P&O and IncCond are well understood and self-tuning, RCC is a good choice when fast convergence to MPP is important, and fractional OC or pilot cells may be economical for fixed outdoor lighting which does not need precision control. There are also some variants, like two-stage IncCond and P&O systems, which offer fast tracking in the first stage, then shift to fine tracking in the second stage as the MPP approaches.