PVWatts with almost no PV and batteries yields large net savings?

Hi Paul,
I'm very confused by the results of these simulations in both the 2020.1.14 beta and the 2018.11.11 SAM programs.  We were trying to figure out the effect of adding batteries, and began being suspicious that our solar PV was no large enough to charge the batteries during the day.  So I took the limit and made the solar array 0.001 kW and left the batteries at 250 kWh, and even though batteries inherently lose energy when charged and discharged.  (I was using peak leveling, and we have one rate as a function of time.) I found that SAM says the system saves $20,000 every year!  I'm enclosing the SAM files for you to fiddle with if you like.  I did put the value of lost load to zero in the 2020 beta.  What is going on?  Can you help me debug this?  The 2018 version even has a negative levelized cost of energy.
Thanks,
Rob 

Hi Rob,

In SAM 2018.11.11, I used the default weather file for Phoenix, Arizona, which resulted in a net annual electricity bill savings of about $5,000.

To see how SAM calculated those savings, I used the following "Monthly Data" variables on the Data Tables tab to show the four components of the electricity bill for the case with the system and without (you can export this data to a spreadsheet to do calculations by clicking "Save as CSV" or in Windows, "Send to Excel"):

Demand charge (TOU) with system
Demand charge (TOU) without system

Demand charge (flat) with system
Demand charge (flat) without system

Energy charge with system
Energy charge without system

Fixed charge with system
Fixed charge without system

The total bill with and without the system is the sum of the four components of the electricity bill, and the savings is the difference between the total bill without and with the system: Savings = Bill without System - Bill with System.

It looks like you double-counted the demand charge by modeling it as both a TOU rate and a flat rate. For a flat demand rate that is fixed from month to month, you can use the "Demand Rates by Time-of-use and/or Tiers" option on the Electricity Rates page to define a rate in $/kW for a single period and tier, and set the "Demand Rates by Month with Optional Tiers" rates to zero.

SAM is dispatching the battery to reduce the monthly demand charge, and you can see that reduction in the monthly by comparing the "with" and "without" values of the flat and TOU demand charges. (In this case, those savings are doubled because of the double-counting.)

The batteries have the effect of increasing the energy charge, because the system has to buy electricity to charge the battery. The battery requires more electricity from the grid than it delivers to the load because of the roundtrip efficiency losses. You can see that on the Time Series tab by plotting "Electricity to battery from grid" and "Electricity from battery to load." The monthly demand savings are more than the monthly energy charge increase, so there is a monthly net savings.

When I fixed the demand charge double-counting issue, the savings go down to about $1,000 and the NPV is negative, indicating that the savings are insufficient to cover the project costs.

The LCOE is always going to be negative for a system that only has storage because the model treats battery consumption over the year as negative energy.

Best regards,
Paul.