The production processes on industrial sites normally require large amounts of heating, cooling and power for their operation. Therefore, optimisation and synthesis of utility systems is of central interest to the industrial managers. The problem of the global climate change has brought forward the question of reducing significantly the emissions of greenhouse gases into the atmosphere. In this thesis, new models and procedures for tackling the above problems are presented. Firstly, improved models of the utility equipment have been developed, taking into account the various performance factors, and economic and environmental implications of their operation. Secondly, a robust procedure for optimisation of the operation of existing utility systems has been developed. It follows a Successive Mixed Integer Linear Programming (SMILP) procedure, which results in rapid convergence. A top-level analysis for utility systems has been formulated, which allows the evaluation of potential energy saving projects in the processes on a site. The analysis follows a stepwise optimisation procedure, independent of the underlying models. Also, an improved efficiency analysis using the power-toheat ratio has also been developed in order to aid the understanding of the efficiency limits of the utility systems. Finally, a new model and a superstructure-based procedure for the synthesis of process utility systems have been developed. The model consists of a rigorous non-linear and a simplified linear versions. The synthesis procedure includes superstructure construction, initialisation and optimisation. The superstructure optimisation extends the operational SMILP procedure to the synthesis context. Using the synthesis procedure, the problems of optimal synthesis and cost effective de-carbonisation of utility systems are tackled and illustrated by the means of a case study.