The decommissioning of utility services has resulted in increasing needs for optimisation technology that will enable the efficient and economic operation of utility networks, Typical networks consist of rich combinations of operational units and feature choices for turbines of different types and operational characteristics. The appropriate selection, loading and operation of the turbines requires integer and continuous variables that give rise to Mixed Integer Non Linear Programming formulations. Even in simplified versions of the problem where efficiency is treated as a constant, the optimisation effort is impeded by model complexities (nonconvexities, size) that diminish the importance and impact of the results. A conceptual tool introduced in the form of Hardware Composites is extended here to cover more realistic and complicated turbine networks including multiple levels, friction losses, allocated turbines and letdown valves.