This paper deals with the optimal-reinforcement-design problem on a meshed gas transportation network. The question is to determine the correct role of optimization in the resolution of such problems. Thanks to the analysis of simple network cases, structural difficulties in the implementation of optimization tools are pointed out. Furthermore, highlights of the errors made in cost evaluations lead to questioning the notion of "optimality" itself. Giving up the search of unique optimal solution, we suggest that a decision-helping tool proposing a set of relevant solutions and relying on optimization could be of great benefit for the project design engineers.
GRTgaz operates, maintains and develops its gas transportation system in order to guarantee gas shipment while ensuring the highest safety and environmental standards. To meet the increasing capacity demand, GRTgaz invests massively on its transportation system and therefore contributes to the european gas market growth. These additional capacities are created through new pipes as well as compressor stations power increase. In both cases, the cost of these investments is very high, therefore detailed design studies are commissioned. They are crucial and must take into account many parameters of completely different kinds (compression station age, input/output flows, political agreements, consumptions forecasts, etc.). What makes them even more complex is that, at this stage, some of these parameters are uncertain or difficult to define. For instance the cost of a compression power increase is linked with the number of installed compressors which can vary significantly. Moreover, even if all those parameters are assumed to be known, the problem consisting in reinforcing the French main gas transmission network remains difficult. Indeed, the meshed structure of this network leads to a non-convex optimization problem, i.e. a problem that cannot be solved by direct use of usual non-linear solvers. This paper highlights and analyzes the complexity of the main gas transmission network reinforcement optimization. Technical and mathematical aspects are pointed out leading to the proposition of an original optimization strategy. The idea is to compute not one reinforcing solution, the so-called "global optimum", but several relevant ones. In section 2, we analyze the mathematical complexity of the reinforcement problem. In particular we present the difficulties induced by the meshed structure of the network, pointing out the limitation of the use of non linear optimization solvers. In section 3 we detail the structure of investment costs to give an order of magnitude of the errors induced by assumptions or unknown parameters. This invite us to discuss the "optimality" of reinforcement solutions.
A gas network is conceived in order to transport gas while satisfying flow and pressure constraints both at the inlets and outlets of the network. Increasing flows induce saturations: they lead to higher pressure drops that are not suitable with these constraints.