In this paper, we deal with the optimization of a cluster comprising two key processes in petroleum refinery: catalytic reforming, and light naphtha isomerization. We propose a novel hybrid methodology which combines direct and indirect approaches. In the direct approach we consider the complete kinetic model and we adopt derivative-free approaches since the analytic expression of the model is not available. In the indirect approach we propose a surrogate modelling based optimization and we apply state-of-the-art solver to retrieve the optimal operating condition and flowsheet configuration of the process. The hybrid method consists of two steps: (i) we compute different good feasible solutions by means of the indirect approach, and (ii) we apply the direct approach using the solutions of the previous step as starting points. Computational experiments for different scenarios are finally discussed.

在本文中，我们将对包括两个炼油厂关键过程的集群进行优化：催化重整和轻石脑油异构化。我们提出了一种新颖的混合方法，结合了直接方法和间接方法。在直接方法中，我们考虑了完整的动力学模型，并且由于模型的解析表达式不可用，因此我们采用了无导数方法。在间接方法中，我们提出了一种基于代理建模的优化方法，并应用了最新的求解器来检索过程的最佳操作条件和流程图配置。混合方法包括两个步骤：（i）我们通过间接方法计算出不同的可行方案，并且（ii）我们以上一步的解决方案为起点应用直接方法。