Methods for multi-objective investment and operating optimization of complex energy systems

The design and operations of energy systems are key issues for matching energy supply and consumption. Several optimization methods based on the mixed integer linear programming (MILP) have been developed for this purpose. However, due to uncertainty of some parameters like market conditions and resource availability, analyzing only one optimal solution with mono objective function is not sufficient for sizing the energy system.In this study, a multi-period energy system optimization (ESO) model with a mono objective function is first explained. The model is then developed in a multi-objective optimization perspective to systematically generate a good set of solutions by using integer cut constraints (ICC) algorithm and ε constraint. These two methods are discussed and compared.In the next step, the ESO model is reformulated as a multi-objective optimization model with an evolutionary algorithm (EMOO). In this step the model is decomposed into master and slave optimization.Finally developed models are demonstrated by means of a case study comprising six types of conversion technologies, namely, a heat pump, boiler, photovoltaics, as well as a gas turbine, fuel cell and gas engine. Results show that, EMOO is particularly suited for multi-objective optimizations, working with a population of potential solutions, each presenting a different trade-off between objectives. However, MILP with ICC and ε constraint is more suited for generating a small set of ordered solutions with shorter resolution time.

► Combining MILP with the integer cut constraints (ICC) to generate a set of good solutions for an energy system design.
► Combining the integer cut constraints with the ε method for doing a multi objective optimisation.
► Doing a multi-objective optimization of an energy system design by using the evolutionary algorithm (QMOO).
► Results shown, the multi objective optimisation with the evolutionary algorithm (QMOO) is more effective.
►  MILP with ICC and ε method has a shorter resolution time for limited number of solutions compare to QMOO.