Solar Microgrid for Foundry: How to Size the Battery Storage for Continuous Melting
A solar microgrid for a foundry is a serious engineering project. The battery energy storage system (BESS) is the most expensive component after the PV array, and the sizing is critical - too small and the foundry cannot operate through the night or the cloudy spells, too large and the capital cost is wasted. Here is how the BESS sizing actually gets done.
Start with the load profile.
A foundry's load profile depends on the production schedule, the furnace size, the alloy being melted, and the casting cycle. A typical small foundry (1 ton induction furnace) operates 8 to 16 hours per day, with the following profile:
- Morning startup: 0.5 to 1 hour at 30 to 50 percent of full power
- First melt: 1 to 2 hours at 80 to 100 percent of full power
- Pouring and casting: 1 to 2 hours at 20 to 40 percent of full power
- Second melt: 1 to 2 hours at 80 to 100 percent
- Pouring and casting: 1 to 2 hours at 20 to 40 percent
- Cool down and shutdown: 0.5 to 1 hour at 10 to 20 percent
Over a 12-hour shift, the energy consumption is 8 to 15 MWh, depending on the production rate and the furnace efficiency. The peak demand is 800 to 1500 kW (during the melt), and the average demand is 700 to 1200 kW.
For a 24/7 foundry, the load is continuous but the per-hour demand is similar. The energy consumption is 16 to 36 MWh per day.
The solar generation profile is the next variable.
The solar generation depends on the location, the season, the weather, and the array size. A 2 MW PV array in a sunny location (US Southwest, Middle East) produces:
- Peak power (solar noon, summer): 1800 to 2000 kW
- Peak power (solar noon, winter): 1200 to 1500 kW
- Average over daylight hours: 1000 to 1400 kW (summer), 600 to 900 kW (winter)
- Energy per day (summer): 12 to 18 MWh
- Energy per day (winter): 6 to 12 MWh
The solar generation drops to zero at night. The solar generation varies during the day with the sun angle and the weather. The variation is the reason for the BESS.
The BESS sizing has to handle three scenarios.
Scenario 1: Night-time operation. The PV array produces nothing at night. If the foundry operates at night, the BESS has to supply all the energy. The energy required is the night-time load times the night-time hours. For a 12-hour night and a 1000 kW average load, the BESS has to supply 12 MWh.
Scenario 2: Cloudy day operation. The PV output drops to 10 to 30 percent of rated during heavy cloud cover. The BESS has to supplement the PV to maintain the load. The energy required is the load minus the PV output, times the duration of the cloud cover. For a 4-hour cloud cover event and a 1000 kW average load with the PV at 20 percent of rated (400 kW from a 2 MW array), the BESS has to supply 600 kW x 4 hours = 2.4 MWh.
Scenario 3: Peak demand. The induction furnace has high peak demand during the melt. The PV array and the BESS have to supply the peak together. If the peak demand is 1500 kW and the PV is providing 1000 kW, the BESS has to supply 500 kW for the duration of the peak (typically 1 to 2 hours). The energy required is 0.5 to 1.0 MWh.
The BESS sizing is a balance of these three scenarios.
A BESS sized for scenario 1 (12 MWh for a 12-hour night) is a large investment - roughly $3 to $6 million in 2026 for a 12 MWh LFP system. The payback depends on the avoided grid electricity cost.
A BESS sized for scenario 2 (2 to 4 MWh for cloudy day bridging) is more modest - $600,000 to $1.2 million. The payback is faster because the BESS is used more frequently.
A BESS sized for scenario 3 (0.5 to 1 MWh for peak shaving) is small - $150,000 to $300,000. The payback is fastest because the BESS is used every day during the peak demand.
Most solar microgrid foundries use a BESS that is a compromise between scenarios 1 and 2 - sized to bridge the night and the cloudy days, but not to provide weeks of autonomy. The BESS is recharged from the PV array during the day.
A common sizing approach is the "autonomy days" method.
The autonomy days method asks: how many days can the BESS supply the average load without any PV input? A 1-day autonomy BESS is sized to supply the average load for 24 hours. A 2-day autonomy BESS is sized to supply the average load for 48 hours. A 3-day autonomy BESS is sized for 72 hours.
For a 1000 kW average load:
- 1-day autonomy: 24 MWh
- 2-day autonomy: 48 MWh
- 3-day autonomy: 72 MWh
The 1-day autonomy is the minimum for night-time operation. The 2-day and 3-day autonomy add resilience for multi-day cloudy weather.
The cost of the BESS at these sizes (in 2026) is:
- 24 MWh: $6 to $12 million
- 48 MWh: $12 to $24 million
- 72 MWh: $18 to $36 million
The cost is substantial. Most off-grid solar foundries opt for 1 to 1.5 days of autonomy, with a backup diesel generator for the extended cloudy periods. The diesel generator is sized to supply the full load, and it runs only when the BESS is depleted and the PV is not generating. The diesel cost is much lower than the BESS cost, but the CO2 emissions are higher.
The BESS technology choice.
The dominant BESS technology in 2026 is lithium iron phosphate (LFP). LFP is favored for stationary storage because of its long cycle life (5,000 to 10,000 cycles), its thermal stability (lower fire risk than NMC), and its cost (the lowest among lithium chemistries for stationary applications).
LFP cells are assembled into modules (typically 50 to 100 kWh per module), modules into racks (typically 250 to 500 kWh per rack), and racks into containers (typically 1 to 2 MWh per 20-foot container). The container includes the BMS (battery management system), the thermal management, the fire suppression, and the AC disconnects.
The BESS has a round-trip efficiency of 90 to 95 percent (the energy lost in charging and discharging). The calendar life is 15 to 20 years. The cycle life depends on the depth of discharge - 80 percent DoD gives 5,000 to 8,000 cycles, 50 percent DoD gives 10,000 to 15,000 cycles.
The alternative BESS technologies include:
- Sodium-ion: emerging technology with lower cost per kWh but lower energy density and shorter cycle life
- Vanadium redox flow: long cycle life but lower energy density and higher cost
- Iron-air: very low cost but very low efficiency and slow response
- Compressed air (CAES): low cost but requires specific geology
- Pumped hydro: very low cost but requires specific topography
For foundry applications in 2026, LFP is the clear winner for most situations.
The microgrid controller ties it together.
The microgrid controller manages the energy flows between the PV array, the BESS, the load, and the grid (or the backup generator). The controller has to balance multiple objectives: maximize the solar fraction, minimize the operating cost, maintain the BESS state of charge within safe limits, and respond to the load transients from the induction furnace.
A good microgrid controller uses model predictive control (MPC) to look ahead at the solar forecast, the load forecast, and the electricity price (if grid-tied). The controller plans the energy flows for the next 15 to 60 minutes and executes the plan in real time. The result is a higher solar fraction, lower operating cost, and longer BESS life.
Author: MONTE INTELLIGENCE solar microgrid engineering team. For BESS sizing and microgrid design, contact helenxu@cnlymonte.com.
