A new computer strategy cut long-running or unfinished grid-switching calculations across 100 demand scenarios.
The study addresses a planning problem in which every transmission line can be switched on or off to reduce electricity-generation costs. The researchers combined a graph-based method for setting tighter limits in the mathematical model with an iterative procedure that finds good feasible grid configurations and uses them to guide the search for the best solution.
In tests using the standard IEEE 118-bus network and 100 demand scenarios, the combined strategy performed better than conventional approaches described in the literature. The authors say the iterative procedure still converges to the global optimum, even though the procedure used to find intermediate solutions is heuristic.
What the grid tests showed
The researchers used graph simplifications and a relaxed longest-path calculation to derive tighter values for the model's large numerical limits. They also developed an iterative search that generates low-cost feasible configurations, helping the integer optimization solver narrow its search.
Across 100 demand scenarios on the IEEE 118-bus network, the approach reduced the number of instances that remained unsolved after one hour by 57%. It also reduced average computational time by 44% compared with conventional approaches in the technical literature. The authors report that the best performance came from combining the tighter limits with the iterative procedure.
Why faster calculations matter
Allowing transmission lines to be switched can give a power system more options for arranging its network as operating conditions change. But calculating the best arrangement becomes especially difficult when every line is switchable and the network must remain connected.
By reducing the time and number of unfinished calculations in the reported tests, the strategy could make this type of grid optimization more practical, including for systems facing highly variable operating conditions. The study does not show how the method performs in a real operating grid.
Evidence and caveats
The evidence comes from numerical optimization experiments on the standard IEEE 118-bus network under 100 demand scenarios. The results are comparisons with conventional methods reported in the technical literature, rather than measurements from a power grid operating in the field.
The excerpts report results for this network and these scenarios only; they do not establish that the same reductions will occur on other grid designs, with other operating conditions, or in real-time operations. The method for generating intermediate feasible solutions is heuristic, although the paper states that the overall iterative procedure converges to the global optimum.
// Source
Top · 2026 · DOI: 10.1007/s11750-026-00733-6
Authors: Marina Aguilar-Moreno, Salvador Pineda, Juan Miguel Morales
