Monitoring systems form an essential component of any photovoltaic system, however our experience shows that they have often been overlooked or not specified in order to keep costs down. Solar inverters are under fairly constant strain and have to tolerate a wide range of electrical loads, so as systems get older the inverter is the component most likely to develop a fault. Since it is such an essential component, it is important that any issue is identified and rectified as soon as possible.
All NWT’s installations are supplied with monitoring technology as standard. This allows you to keep track of how much electricity has been generated by your system throughout the year, and also to analyse the data in order to establish when the generation has occurred. This in turn gives the opportunity to begin to match energy use with solar generation, for instance by operating energy-intensive equipment in the middle of the day.
Making a financial saving by omitting monitoring systems from photovoltaic systems at the installation phase can prove to be very costly if issues are not identified. Because Feed-in Tariffs are paid over a fixed period, revenue lost through reduced productivity is irrecoverable. The longer it takes before issues are rectified, the more revenue will have been lost permanently.
Case Study – Business Park
In general terms it is difficult to make direct output comparisons between PV systems even when they consist of the same equipment. Factors such as roof pitch, orientation, cable lengths, number of modules & inverters (and their manufacturers) will all have an effect on the amount of electricity generated. Added to these attributes, the weather in any given year will vary, making comparisons problematic.
If there is a major issue with an installation where no monitoring system is in place then this is likely to become apparent through a warning light on an inverter or when an abnormally low metering reading is noted. The problem with this approach is that, given the variables mentioned above, there is an intrinsic variation in the amount of electricity generated by any given generator. This in turn makes it possible to assume that a meter reading is appreciably lower than the last because there has been poorer weather.
Figure 1 shows how the overall generation can vary even for the same installation from year to year. This inherent variability can make it hard to identify problems.
Fig. 1 – The amount of electricity generated by any given PV system will vary year-by-year, which can often mask underlying faults since they are attributed to bad weather.
If in fact there is an underlying issue with the system, then this issue may not be identified until much later on (because it is conceivable that weather variations were the cause of low readings). By this time the generator may have been performing far below its potential for some time, and resultant lost income can not be recovered.
Comparing identical installations over time
Occasionally there is an opportunity to make a direct comparison because identical systems have been installed at the same location, at the same time.
One recent NWT maintenance visit was to a business park with a total of twenty-one individual PV systems, some of them with identical specifications.
Looking at the total amount of electricity generated by five identical installations, situated side-by-side at the same orientation with the same photovoltaic system installed upon them, the owner had established that there were substantial differences. Figure 2 shows the total generation figures for each of the units.
Fig. 2 – The cumulative amount of electricity generated by each of the systems
Analysing the disparities
There were clearly issues with several of the installations however, although metering data was captured in detail, none of them had a monitoring system in place. In this case the amount of data available made it possible to build a detailed representation of the lifetime generation achieve by each system.
Because of their immediate proximity and identical equipment, it is reasonable to assume that they should all be generating the same amount of electricity. Since they were all installed at the same time, December 2011, it is possible to analyse the monthly automatic meter readings to see whether this has been the case.
The chart below shows the meter readings over time for each of these five identical PV generators.
Actual readings from five identical installations
The chart below compares the total annual generation between months in a series of years, starting from December 2012. Assuming perfect conditions, we would expect to see five identical lines showing approximately horizontal lines. In reality the data clearly show when systems began to experience issues as shown in Figure 3.
Fig. 3 – Total annual rate of generation for each system, Dec 2012 to Oct 2016.
Even at a glance, the chart shows:
- Unit 7 never reached the same level of generation as the rest of the systems after June 2012. It was also the first unit with an inverter that developed a serious issue approximately April 2013
- Next was Unit 6, which developed a significant issue in November 2013. Unit 6 generated almost double the amount of electricity in its first 2 years than it has in the almost 3 years since.
- Unit 4 had an issue around the same time as Unit 6, in November 2013. It then deteriorated substantially again in Feb 2015.
- Unit 8 appears to have been operating at the proper level until June 2015 then developed a fault.
Identifying faults early
What difference would monitoring have made?
There is a distinction here between metering and monitoring. The owner of the PV generators had access to fairly sophisticated records of the monthly meter readings for each of the five units, so in theory the issues should have been identified sooner.
Our experience shows that there is generally ambiguous responsibility for scrutinising meter readings and that in many cases no records are kept at all. Through the use of monitoring systems it is possible to establish at a glance whether a PV generator is operating as intended, as shown in Figure 4.
Fig. 4 – Example dashboard from SolarEdge monitoring of a 50kWp PV generator.