This work presents a systematic methodology for driver and power plant selection based on mathematical programming. Optimal designs are generated given a set of mechanical and electric power demands, an economic scenario and other design constraints. The solution consists of the optimal number, type and actual size and model of the main compressor drivers, helper motors or generators and power plants, as well as the compressor stage arrangement. The discrete nature of gas turbines is considered and gas turbine drivers and pre-engineered power plants are selected from a group of candidates. The use of piecewise linearisation in some constraints maintains the model robustness, without leading to significant inaccuracies. Parallel compression sets are allowed in the formulation in order to benefit from the associated step improvement in plant availability. Plant availability has also been included in the objective function under the form of lost profit. Despite the associated non-linearities, this adds a new dimension to the classical approach and allows a more comprehensive exploitation of the trade-offs between capital costs, operating costs and availability. When ignoring process heating and neglecting any potential steam system that is not part of a preengineered power plant, a simplified formulation can be applied to heavily power dominated processes, such as LNG. However, a more comprehensive formulation, allowing waste heat recovery and integration with a multilevel steam system, is able to produce more thermally efficient systems which become competitive under some economic scenarios despite their increased capital cost and complexity. Furthermore, addressing the supply of process heat via the steam system extends the applicability of the methodology' to a wider range of process power-to-heat ratios. The formulation has proved to be flexible and robust enough to produce solutions ranging from nosteam to all-steam systems. Its application to industrial case studies produces optimal and near-optimal systems for a variety of economic scenarios.