Questions about ES Model

[Kirktj]

First Impression questions:

1. What’s the logic for when the battery discharges vs charges?
   a. It looks like there are some peaks where the battery isn’t running.
   b. Reading the description page it looks like this focused on arbitrage- is that he only value stream measured?
2. How is the battery’s SoC managed?
   a. Are there limits on the DoD?
3. How is the value from arbitrage calculated? I see that’s there a demand charge input, but what else feeds into it?
4. Any plans to provide example numbers (ex. The discharge, charge
5. Want: total number of battery cycles and total number of cycles to 80% or more

[dpinney]

Good questions.

1a. Demand charges are calculated per month, so the algorithm attempts to shave the highest peaks in the month. We're pretty sure the charge/discharge cycle is optimal. Example of peaks you see that aren't shaved?
1b. We only do arbitrage currently from a distribution utility perspective. What other value streams would you like to see?
2. SoC is calculated as we iterate through the time series to calculate max peak shaving potential.
2a. The DoD is limited by the battery capacity--can't go below 0% naturally. The are also rate limits on charge and discharge.
3. Arbitrage amount is total peak reduction per month times demand charge.
4. What are you asking here?
5. Yes, we should calculate cycles. However, what depth of discharge defines a cycle and is that battery-dependent? I'm talking with the NREL SAM team who've made an attempt at answering this.

@delysis might have other details.

[Kirktj]

1. Here's the one that jumped out to me:
   .
   Where can I find the algorithm?

1b. From the ES toolkit: "Although each case is unique, if a plant used for electric energy time-shift is in the right location and if it is discharged at the right times, it could also serve the following applications: electric supply capacity, T&D upgrade deferral, transmission congestion relief, electric service reliability, electric service power quality, and ancillary services."

2a/5. I didn't phrase this well. Depending on what battery you use the cycle life will vary depending on the DoD of each cycle. Ideally the algorithm managing the charge/discharge cycle would factor in the SoC if necessary and model could predict the battery's useful life based on cycle life, not just financial projection length since some load profiles will result in more cycles than others.

1. I left out some words. Any plans to provide example batteries? Example, the charge and discharge rate for a specific GE battery? Or a generalized Li-ion battery?
2. The model is the ES model but it looks like it's just focused on battery ES. Is that accurate?

[dpinney]

Thanks for the follow up.

1a. We need to see the timescale for that graph. Without the timescale the behavior could be correct or not--it's impossible to know. The algorithm is in the source code at /omf/omf/models/energyStorage.py.
1b. We're considering these other value streams. How to monetize them and stack them is a huge research question.
2a. What qualifies as a cycle for a given battery is a huge research question.

1. (4?) Yes, we'd like to do specific batteries.
2. (6?) No, the model applies to any form of energy storage.

[Kirktj]

Looking at it my guess is that because it's in the same month as the other peak it's not the highest?

[dpinney]

Yes, correct.

[dpinney]

I'm going to close this and create a generic energyStorage issue where we can track other things.