Summary
Batteries have an assumed 15-year lifetime, and are assumed to be fully retired at the end of 15 years. However, many aspects of a battery system have lifetimes greater than 15 years, and many batteries are coupled with generation facilities that will have lifetimes greater than 15 years (nearly half of the storage in the queue is in hybrids). We think it would be more appropriate to model batteries as a 30-year asset with some refurbishment cost in year 15 rather than as a 15-year asset. PV+battery hybrids in the Annual Technology Baseline (ATB) already use this approach.
Proposed Changes
Vignesh Ramasamy did some bottom-up calculations to estimate that a refurbished battery would cost 71-83% as much as a new build (where the range depends on if 80% or 100% of the battery packs are replaced in year 15). He sent me his bottom-up modeling spreadsheet we can use to get specific values as needed.
We would want the refurbishment cost to take into account the cost of storage in a future year, so I think the best approach would be:
[ ] Take a close look at Vignesh's work and any battery replacement literature to choose a refurbishment cost and add that refurbishment cost number (e.g., 75% of a new build cost) to scalars.csv
[ ] Decide whether to add the refurbishment cost as an upfront capex cost or as a FOM cost (ATB uses the FOM approach for hybrids)
[ ] Perform the refurbishment cost calculation in plantcostprep.py. It will need to read in some financial parameters to do discounting properly, but I think it still fits better in plantcostprep.py than in calc_financial_inputs.py.
[ ] Ensure that the refurbishment cost flows into the objective function properly
[ ] Change the economic and physical lifetime of batteries to 30 years
Additional context
Depending on whether the FOM or capex approach is taken, it might be worth thinking about how this refurbishment cost shows up the system cost and retail rate calculations.
Summary
Batteries have an assumed 15-year lifetime, and are assumed to be fully retired at the end of 15 years. However, many aspects of a battery system have lifetimes greater than 15 years, and many batteries are coupled with generation facilities that will have lifetimes greater than 15 years (nearly half of the storage in the queue is in hybrids). We think it would be more appropriate to model batteries as a 30-year asset with some refurbishment cost in year 15 rather than as a 15-year asset. PV+battery hybrids in the Annual Technology Baseline (ATB) already use this approach.
Proposed Changes
Vignesh Ramasamy did some bottom-up calculations to estimate that a refurbished battery would cost 71-83% as much as a new build (where the range depends on if 80% or 100% of the battery packs are replaced in year 15). He sent me his bottom-up modeling spreadsheet we can use to get specific values as needed.
We would want the refurbishment cost to take into account the cost of storage in a future year, so I think the best approach would be:
[ ] Take a close look at Vignesh's work and any battery replacement literature to choose a refurbishment cost and add that refurbishment cost number (e.g., 75% of a new build cost) to
scalars.csv[ ] Decide whether to add the refurbishment cost as an upfront capex cost or as a FOM cost (ATB uses the FOM approach for hybrids)
[ ] Perform the refurbishment cost calculation in
plantcostprep.py. It will need to read in some financial parameters to do discounting properly, but I think it still fits better inplantcostprep.pythan incalc_financial_inputs.py.[ ] Ensure that the refurbishment cost flows into the objective function properly
[ ] Change the economic and physical lifetime of batteries to 30 years
Additional context
Depending on whether the FOM or capex approach is taken, it might be worth thinking about how this refurbishment cost shows up the system cost and retail rate calculations.