PR as performance guarantee

The Performance Ratio indicator (PR) is widely used as a performance guarantee of a PV plant. The PR is calculated with the forecast simulations, and compared to the PR measured on the field.

PR measurement

The PR calculated by PVsyst is referenced to the transposed irradiance GlobInc, directly from GlobHor without horizon loss.

The PR evaluated by the measurement should be normalized to a reference irradiance normally measured on-site. This is usually defined as

\(PR_{meas} = EGrid_{meas} / (GlobP_{meas} * PNomPV)\)

However there are several ways of determining the irradiance GlobP_meas:

• a simple solution is to use the hourly data from a weather data provider. These data should obviously be simultaneous with the plant operation. The synchonization should be carefully checked. The available data are the global horizontal, we have to evaluate the diffuse, and transpose this in the collector plane for getting the GlobP_meas.
• a better solution is to measure the irradiance in the horizontal plane, using a solarimeter (preferably a pyranometer). Again, we have to perform a transposition in the collector plane, this requires a calculation in hourly values (provided by PVsyst). A measurement of the diffuse would improve the transposition accuracy.
• the solarimeter is often directly mounted in the plane of array (POA measurement). This avoids the use of PVsyst for the calculation of the transposition: you can directly obtain the measured PR without using a transposition. The solarimeter should be carefully positioned on a place without mutual shadings, and without masking the albedo in front of the plant (for example on top of the first shed).
• with tracking systems, many people put a solarimeter on the tracking axis. In this situation, the solarimeter is affected by the mutual shadings of neighbour sheds, on the diffuse and especially the albedo components. The interpretation of the GlobP_meas (and the PR_meas) will require a correction due to shadings, as calculated by PVsyst. A solution for avoiding this would be to put the solarimeter on a prolongation of the axis, outside of the trackng array.
• For bi-facial systems, the irradiance on the rear side of the PV modules is required by the PRbifi calculation. But there is no consensus about how to measure it: the irradiance on the back side is highly inhomogeneous, depending on the position in the shed.

NB: The PVsyst simulation requires the measurement of the ambient temperature as well. This should be included in the data, along with the irradiance. It may possibly use the measured PV modules temperature.

Accuracy

In any case, the accuracy of the PR_meas is directly related to the accuracy of the GlobP_meas.

External data are usually satellite data, with an accuracy of 10 to 20% on hourly values (although the long-term data are better). Even if this is a weather station, the hourly accuracy is not better unless the station is situated at less than 10 km.

When using a solarimeter, an excellent calibration is required. It is sometimes necessary to periodically recalibrate the solarimeter. And the positioning should be carefully chosen, without shades. The horizontality of the device is essential. The accuracy of a good solarimeter measurement is of the order of 1-2%. The transposition model may be considered as sufficiently accurate (less that 1% on monthly periods).

The import of POA measurements (GlobP_meas) in PVsyst produces a reconstruction of the GlobHor and DiffHor variables, necessary for the shading calculations. These restitute exactly the original GlobP_meas when transposed during the simulation, so that we are not dependent on possible uncertainties of the transposition model.

The measurement on the tracker's axis is really a problem, as it is subject to mutual shadings which are evaluated by a complex calculation in PVsyst. Moreover, nothing ensures that the measurement at the axis level is fully representative of the irradiance on the full tracker.

PR comparison

Therefore the comparison between measured and forecat PR is difficult. When you specify a warranty based on a PR value, it is very important to specify the way this PR will be established on-site.

Many effects should be considered when comparing forecast and measured performance ratios.

Method of comparison

There are 2 ways of comparing the forecast and measured data:

• Either you keep the guaranteed PR value as such, and you try to evaluate the PR from the measured data without use of PVsyst (or a limited use for the evaluation of some corrections). In many cases the comparison may be done on accumulated results, avoiding the necessity of recalculating the system in hourly values.
• Or you use the available measured Global_meas for a complete resimulation of the system, and therefore you get a new evaluation of the reference PR based on measured input data. This should be very close to the forecast result. Then you compare the production values, usually well measured, with E_Grid.

The second solution provides a better flexibility in case of malfunctions of the system, as you can compare the simulated and measured results for a set of situations, cutting the doubtful data.

Effect of the Horizon

The horizon shading loss is usually included in the forecast PR by PVsyst, as the basic irradiance from data providers is usually given for a free horizon (satellite data). In the PVsyst simulation, the PR is referenced on GlobInc before horizon loss.

Now if you measure the irradiance on-site, the horizon effect is naturally included in the data. The PR will be different (higher) than the forecast PR. For a correct comparison, you should use a forecast PR based on the value GlobHrz instead of GlobInc.

Trackers: measurement on the axis

If you measure the Global_meas on the tracker axis, your measurement will be underevaluated due to diffuse shading loss.

Moreover, and more important, if you are on a hill with uneven terrain, you cannot avoid the "real" mutual shadings (on beam component), which cannot be registered by the solarimeter on the axis. This is a significant bias if you want to identify the Glob_meas to the value calculated by PVsyst. For the correction mentioned below, you have to consider only the loss on the diffuse and albedo, named ShdDLss and ShdALss in the simulation variables.

Therefore for comparing the PR to the forecast one, the only way is to evaluate the shading losses (ShdDLss + ShdALss) [kWh/m/²] provided by the simulation. You have to add these shading losses to the measured values for getting a reference irradiance.

If you perform a global comparison, you can add these shading losses in monthly or yearly values for the evaluation of the PR_meas.

But if you want to resimulate your system on the basis of the measured irradiance, you have to perform a first simulation, by registrating the shading losses in hourly values (CSV houly file). Then you add these values to your original hourly Global_meas data (in EXCEL), and reimport these corrected data in PVsyst for the definitive simulation.

Effect of the ageing

You should be aware that the PVsystem is subject to the degradaton due to ageing. Remember that there is already a little degradation during the first year of operation. At the end of the year the degradation is already one time the degradation rate parameter.

The performance ratio diminishes in the same way as the system performance. This is namely shown in the ageing tool. You should take this into account when defining the guaranteed PR value, i.e. define a degressive value.

P50 - P90

The PR is an indicator of the global system losses. It is not dependent on the yearly irradiaton: therefore the PR is independent of the fact that the considered year is a good or a bad year.