Microgrid dispatch simulation with solar, diesel and battery storage

Tidewater Island at about 12.5° N has no grid connection to the mainland. Three 400 kW diesel generators supply 2,300 people, a fish-processing plant and a small clinic. Peak demand is about 650 kW at 19:00 to 21:00, the night base is around 250 kW, and the island uses roughly 3.3 GWh a year. Diesel is delivered by barge at USD 1.30 a litre. In bad weather the barge is late, and twice last year the co-operative had to ration power.

A development bank has offered a concessional loan for 800 kWp of PV and a 600 kW / 1,200 kWh battery. Before signing, the co-operative's board wants an independent answer to four questions: - How much diesel will the system really save in a year? - Which dispatch strategy should the controller use? - Are the proposed sizes right, or would a different PV and battery combination save more per dollar? - Will supply be as reliable as today, with enough spinning reserve when a cloud passes over the array?

The operating rules are fixed by the station engineer: no generator may run below 30% load for more than an hour, online capacity plus available battery power must always cover the load plus a reserve, and at least one generator stays online unless the battery can carry the island for a full hour on its own.

You will build hourly load and PV data, write component models for the generators and the battery, implement load-following and cycle-charging dispatch, compare them, sweep PV and battery sizes, and work out the levelised cost of energy. Everything runs in Python on a laptop. You may compare one case against a commercial tool if you have access, but you do not need one.

Done looks like a simulation repository and a board report with fuel saved, renewable fraction, generator hours and starts, battery cycles, unserved energy, reserve violations and levelised cost for each strategy and size. The recommendation should say which size and strategy to buy and what risks remain.