More detail please for the Calculation Method for P90 and assorted statistical fits

I am looking for a comprehensive description of how SAM calculates the P90 and other confidence fits. I have read what papers I could find regarding this (Dobos, Gilman and Kasberg) but they do not provide sufficient detail regarding the calculation method, most specifically in the temporal span used. The paper says that it is based on hourly data but the P90 values are then annual. Because PPA's are hourly and monthly and energy variance is also hour dependent, I would like to ascertain in more detail how SAM calculated P90 values. Glad to do this with telephone call (please?)

The P50/P90 calculations involve two timeframes. The first is the simulation timestep, which by default is one hour. The second is the number of weather files in the set you use for your analysis, which determines the number of years involved in the analysis.

The "default weather database file" contains file sets for locations in the U.S. For some locations, the file set has 15 files representing 15 years worth of data, and for other locations the set contains 30 years' worth of data.

For a given location, SAM runs an hourly simulation for each year for each of the years in the file set to calculate that year's total electricity output in kWh and other metrics. For example, for a location with 30 weather files, SAM would calculate 30 annual electricity output values.

After running the simulations, SAM then makes the statistical calculations on those annual values. For the location with 30 files, SAM would report the minimum, maximum, P50 and P90 annual output values of the 30 values.

I'm sorry that we do not have the resources to provide telephone support, but please don't hesitate to post follow-up questions if you have them.

Best regards,
Paul.

Thanks Paul. SAM can report P90 etc information for revenue as well, how does the system apply hourly and monthly PPA pricing with the P90 + calculations?

I'm glad you asked that question -- I was going to mention it in my first reply, but thought it might be too confusing to include with the description of annual output metrics.

For the financial metrics, SAM does a complete cash flow calculation over the analysis period that you specify on the Financing page for each year in the P50/P90 weather file set.

For a location with 30 years of weather data and an analysis with a 25-year analysis period, SAM calculates revenue, PPA price, LCOE and other metrics over the 25-year period for each of the 30 years. The P50/P90 values are then the minimum, maximum, P50, P90 values of each metric over the 30-year period.


Thanks,
Paul.

Thanks Paul. So to make sure I understand. When is the hourly PPA pricing data applied for a Reveue P90 analysis. Is the P90 energy data for a specific month, day, hour multipled by the appropriate PPA price for that hour, and then the annual revenue is the sum of all the P90 houlry revenue slices or is the PPA price applied at a less granular level. I am sorry to be do dense. Thank you. (a phone call might be faster: if you want to call me my number is 704-778-1909)

The PPA Price is a single value that applies to all hours of the year. If you include time-of-delivery factors in your analysis, then the revenue in a given hour with a TOD factor other than 1 is the product of the electricity output, PPA price, and TOD factor for that hour. The annual revenue is the sum of the hourly values. (To further confuse things, the hourly output is for Year 1 of the Financing page's analysis period, which means that the "year-to-year decline" value on the Performance Adjustment page does not affect P50/P90 results).

SAM repeats this calculation for each of the years in the P50/P90 analysis (determined by the number of weather files, not the analysis period on the Financing page), which means that the PPA price for the year 1 weather file might be different than the PPA price for the year 2 weather file because of the difference in solar resource between the two years.

Best regards,
Paul.