Files
entrix_case_challange/optimizer/optimzer_plan.md
2025-05-03 20:46:14 +02:00

1.4 KiB

Optimizer Definition for a constraint n-forecast Trading-Problem

We want to optimize the performance of an energy trader given the forecast for n steps. The battery:

  • holds 1MWh
  • charges/discharges at max. 1MW per hour (we can add/loose x*1MW, x \in R ) Prices are stable for the given hour (t) and we sell and buy for the same price.

Considerations:

  • Single variable, P (=x), for each hour t from 0 to n-1.

  • If P > 0, it represents discharging (selling power) with a magnitude of P.

  • If P < 0, it represents charging (buying power) with a magnitude of -P.

  • If P = 0, it represents holding (doing nothing).

  • if we have forecasts for t_n, t_n+m we might have to interpolate between n .. m

  • or... we work with the gaps and dt as charge time .... no

Variables:

  • price_t = price per MWH at t (eq)
  • B (t=0..n) = State of Battery in MWH
  • P (t=0..n) = Charge/Discharge factor given the possible base rate of 1MW/h
  • max_p = 1 (charge/discharge limits) & and battery capacity limits (both=1)
  • SoB_initial = 0
  • h = horizon \in N^+

Objective

  • We Maximize: Sum_{t=0}^{n-1} (price_t * P)

Constraints

  • Fixed starting state: SoB_0 = SoB_initial
  • Charge/Discharge Limit: (-max_p <= P <= max_p) for all t = 0, ..., n-1
  • Storage Limit: (0 <= B+(1*P) <= max_p) for all t = 0, ..., n-1
  • Future B State: SoB_{t+1} = (B + P) for t = 0 to n-1