Skip to main content

Advertisement

Table 3 Performance of one week simulation using different optimization frameworks

From: Scaling: managing a large number of distributed battery energy storage systems

Algorithms Revenue [TEuro] average violations discharging
Global Optimization 34.60 (100%) 0%
Heuristic 1: without function 32.79 (95%) 11.1%
Heuristic 2: bounded SoC 24.53 (71%) 3.7%
Heuristic 3: linear 27.99 (81%) 8.5%
Heuristic 4: concave piece-wise linear 31.21 (90%) 9.3%
Heuristic 5: non-concave piece-wise linear 29.40 (85%) 10.4%
  1. Interestingly, heuristic 1 without any function modeled performs quite well regarding revenue, however by taking into account relatively high violations in the control. Vice versa, bounding the SoC with heuristic 2, provides poor performance while being quite robust in provoking not many violations. Given its effort, heuristic 5 does not perform well. A reason could be the mixed-integer problem, which was not always solved to optimality