Dynamic Speed Control (DSC) on motorways can be enhanced by using models with higher granularity for capturing the evolution of traffic flow. Nonetheless, the required number of micro-simulation runs for reaching an optimal DSC solution and their computational complexity hinder the use of micro-level models for DSC on large-scale motorways. The present study introduces a set of techniques that reduce the number of required micro-simulations, thus improving the computational cost. The proposed techniques: 1) split the motorway into stretches; 2) introduce an approach based on genetic algorithms to reduce the number of micro-simulations; 3) exclude, when possible, VMS combinations by applying solution approximation methods instead of micro-simulation. The method is evaluated through a number of test scenarios in a stretch of motorway A6 in the Netherlands.