Battery control strategies - dumb vs smart - and how it effects battery economics

Two questions we hear often at in the context of battery storage investments, particularly smaller-scale projects located within the distribution network are:

1. How cheap do batteries need to be to be “in the money” for any given project?

2. How much does the battery control strategy matter? On this point there’s often a view that using the battery to maximise self-consumption of solar is the most sensible thing to do, but is it really?

The questions are interrelated & in this post I’m having a crack at answering both at the same time using a common behind-the-meter use-case for the purpose.

I used Gridcog to build a simulation of a small BTM BESS co-located with solar & load, the kind of setup you might see for commercial energy user.

The baseline scenario just includes our site load plus a pre-existing solar system priced against the businesses' assumed energy supply arrangements. The alternative scenarios model the impact of adding battery storage with a view to answering questions 1 & 2 above.To test the impact of different control strategies I've run near-identical versions of the site where the only thing that varies is the way the BESS is controlled. We’re using three strategies:

☀ Solar self-consumption: battery charges off available solar that would otherwise be exported to grid & discharges into site load.

⏱ Simple heuristic or rules-based approach: an average curve for energy supply costs is created for each day of the week & fed to the battery. The battery then operates to reduce these costs

🤓 Full optimisation: the battery is continuously fed a forecast of site load, solar yield & wholesale prices, as well as the structure of the network tariff, & operates to reduce these costs.

To test the impact of different battery capex costs I've run each variation against a range of battery capex costs starting at AUD$1000/kWh all the way down to $100/kWh. What we’re looking to find is the capex cost that delivers us a breakeven project for each control strategy.

🏗 Model setup:
- Our site is located in Sydney, Australia
- Site load consumes 670MWh a year with peak demand of 227kW
- Solar is 300kWp & generates 407MWh pa of which 113MWh is exported back to the grid
- For energy supply costs the site is exposed to the wholesale market using the central case forecast from Endgame Economics, plus a 5% mark-up as a proxy for retailer costs.
- For DNUoS, site is connected to Ausgrid on network tariff EA305
- For certificates we have LRECs, SRECs & ESCs.
- Battery is 100kWh/50kW with 90% depth-of-discharge, 85% round-trip-efficiency & 2% degradation per year
- Model duration is 7 years, from 2024 to 2030. Duration based on max payback period a business customer might accept

📈 Model highlights:
- Control strategy has a huge impact on the commercial outcome
- Solar self-consumption is only economic with a battery cost of $190/kWh
- Optimisation delivers more the 2x the value of a simple, rules-based approach