Per Eilif Wahl

Constraint handling in stochastic optimization algorithms for natural gas liquefaction processes

Abstract Liquefaction of natural gas requires energy intensive refrigeration. A fair comparison of different process concepts and energy efficient designs requires some use of optimization. In near optimal designs, the driving forces in heat transfer are small. Thus, rigorous thermodynamics providing accurate and reliable temperature profiles must be applied for the solution to have a practical value. Owing to the characteristics of the process and the thermodynamics, the optimization problem is non-convex. Furthermore, the optimal solution is expected to be located close to the boundary of the feasible region, suggesting the importance of constraint handling. In this paper, a single-mixed refrigerant process (PRICO®) has been optimized using adaptive simulated annealing. A constraint handling method utilizing process characteristics is proposed and compared with static penalty function formulations. The results indicate the importance of constraint handling, and the best solution found exceeds previously published results.

Constrained non-linear optimisation of a process for liquefaction of natural gas including a geometrical and thermo-hydraulic model of a compact heat exchanger

Abstract A great deal of effort has been put into improving natural gas liquefaction processes, and a number of new process configurations have been described. Recent literature has identified a need for more realistic heat exchanger models to obtain optimum design and operating conditions that do not compromise safety, or that are unrealistic. Here we describe a concept for finding the design and operating conditions of a single mixed-refrigerant process which gives minimum power consumption under given space or weight constraints. We use a sophisticated heat exchanger modelling framework that takes into account system geometry and resolves the details of the heat exchanger through conservation equations coupled with accurate models of thermo-physical properties. First, we find the feasible region which does not compromise safety with Ledinegg instabilities. We then identify the optimal operating conditions for a specific design within this region, before identifying the process design that requires least power consumption. We illustrate how this differs from a purely thermodynamic optimisation, and discuss our key results.

Formulating the optimization problem when using sequential quadratic programming applied to a simple LNG process

Abstract Sequential quadratic programming (SQP) may be very efficient compared with other techniques for the optimization of simple processes for the liquefaction of natural gas (LNG), and can be combined with process evaluation using commercial flowsheet simulators. However, the level of success is dependent on the formulation of the problem. In this work, effects of varying different aspects of the optimization problem formulation is investigated, such as variable selection, formulae for the estimation of derivatives, initial values, variable bounds, and formulation of constraints. Especially the formulation of the constraint for the temperature difference between the hot and cold composite curve is essential. The commonly used minimum temperature difference constraint should generally not be employed in gradient based optimization. Recommendations regarding optimization of simple LNG processes using SQP and flowsheet simulators are provided.