How is the energy production of a solar system forecast?

Solar production estimates are based on various factors. Some can be controlled and modeled with a high degree of certainty and others are closer to estimates about the future.

The forecast is the amount of energy that we expect to be generated by the solar plant after considering the weather conditions that the system is exposed to (irradiance, temperature, etc), the configuration of the plant and how it was installed (number of modules, how they are connected, what angle they are facing, etc) and any constraints on production such as not being able to export into the grid.

Solar production estimates take all of these factors into account, and hence are complex models with varying levels of uncertainty built in. Some factors can be controlled and modelled with a high degree of certainty and others are closer to estimates about the future. Because these are just estimates, we need to state a confidence level for each estimate.

There are many ways to describe the risks associated with a solar array. Generally, you could put them into two categories: “construction risk” and “operating risk.”

Construction risk is any source of risk that happens before the Commercial Operation Date, when the system is not operating. This can include site risk, site control, interconnection risk, EPC and construction risk, and more. For the most part, construction risks are about understanding and controlling the cost and time required to build the system. There are some factors, during the engineering and installation phases, that can have larger impacts on the potential production of the array than others.

Operating risks are the risks associated with running the facility and generating revenue from the production of energy. These can still include some site and equipment failure or warranty risks, but, assuming those are controlled for, the major risk after a solar system has been constructed is how much power it will produce.

The value of energy produced by a solar array is a function of two items:

  • how much energy is produced and
  • the value of that energy.

The value of that energy can be based on a number of factors: the kWh rate it’s offsetting, any net-metering laws that are in place, the negotiated lease rate, potential demand charge reductions (kVA reductions), any production-based incentives, and more.

We cannot predict with 100% certainty the amount of solar radiation that will hit an array over any given period of time, but to understand and communicate the potential solar resource we use P50 and P90 production levels of an array.

While these production estimates rely to some degree on system design and siting, the main variable is weather

The definition of P50 and P90

In P50 and P90, the P stands for probability.

P50 means there is a 50% chance in any given year that production will be at least a specific amount. If an array has a P50 production level of 500 kWh in year 1, it means that in Year 1 there is a 50% chance that production will be AT LEAST 500 kWh.

P90 production means that there is a 90% chance that in any given year production will be at least the specific amount. This means that there is only a 10% chance that production will be lower than the stated amount. If any array has a P90 production level of 400 kWh for year 1, it means that in year 1 there is a 90% chance that production will be AT LEAST 400 kWh.

Because the variability of solar production, and thus the difference between P50 and P90, is largely based on the variability of weather, extensive weather analysis is performed to calculate these values. For any given location, a minimum of a 20-year data set of the weather experienced in that location is analysed to create a typical meteorological year (TMY). This TMY is generated by averaging the weather conditions of each specific hour of the year, over the 20-year period. The TMY therefore represents an hourly indication of the typical weather experienced for that location over the last 20 years.

TMYs based on a 20-year dataset used to be very reliable, however with the advent of global warming, we are seeing much more extreme weather conditions, and a larger variance between TMY forecasts and reality.

Areas that have less sporadic weather changes have closer P50 and P90 values.. Lower variability means a lower standard deviation across the distribution of solar irradiation values.

The difference between P50 and P90 production levels in areas with moderately variable weather can have large impacts on the assumed production for an array.

What’s critical to understand about P50 and P90?

  • P50 and P90 production levels can only accurately be determined with complete 20-year data sets and software models. The 20-years datasets can also be expensive to procure.
  • P90 is more conservative, so investors will focus on this amount. P50 is less conservative, so developers tend to focus on this.
  • The greater the variability of weather in a specific area, the greater the variability between P50 and P90 yield estimates.

At Sun Exchange our forecasts are based on P50 production levels, however alternative contingencies are built into the model such as 1% grid unavailability, and 1.4% production loss due to loadshedding. This 1.4% is based on 5 loadshedding events of 2 hours per month, resulting in 10 hours per month, or 120 hours per year: 120/8760 = 1.37%.