The study of a better use of hydrogen in refineries is essential due to its increasing use in hydrotreating fractions obtained from petroleum. Since several factors affect this process, it is essential not only that the hydrogen network must be able to operate feasibly in various conditions but also accomplish it with minimum costs. In this work, a systematic approach is proposed considering a multi-scenario optimization problem formulation for the network design coupled with the flexibility evaluation of the proposed design to verify the flexibility and identify critical scenarios that are used to update the previous set of scenarios for the design problem. As a result, it is obtained a cost-efficient flexible design. The proposed approach can be used for new designs or for the retrofit case. For the design a superstructure-based MILP and MINLP multi-scenario models were developed and completely described, to optimize hydrogen networks through uncertainties in hydrogen consumption in consumer units. The flexibility index problem formulated for hydrogen networks is presented. All the optimization models were implemented in the modeling system GAMS. The initialization strategy consist of using the network obtained from linear optimization as a starting point for nonlinear optimization. In addition, it was also used the proposed technique of virtual compressors, able to reduce the cost of capital even further. Two case studies were used to validate the proposed approach. A case study from the literature was used and also a second case using real data of a Brazilian refinery. Compared to the initially proposed network, the model through optimization achieved flexible design with a reduction of more than 13.8% (example 1) and 16% (example 2) in the operating cost. For both cases, the procedure could find a cost-efficient flexible design that can be coupled with the refinery production planning for the whole process economy.

研究在炼油厂更好地利用氢气是必不可少的，因为它越来越多地用于从石油中获得的加氢处理馏分。由于有几个因素会影响这个过程，因此氢网络不仅必须能够在各种条件下可行地运行，而且还必须以最低的成本完成它。在这项工作中，提出了一种系统方法，考虑网络设计的多场景优化问题公式以及对所提议设计的灵活性评估，以验证灵活性并识别用于更新先前场景集的关键场景。设计问题。结果，获得了具有成本效益的灵活设计。建议的方法可用于新设计或改造案例。对于设计，开发并完整描述了基于上层结构的 MILP 和 MINLP 多情景模型，以通过消费者单位氢消耗的不确定性来优化氢网络。提出了为氢网络制定的灵活性指标问题。所有的优化模型都在建模系统 GAMS 中实现。初始化策略包括使用从线性优化中获得的网络作为非线性优化的起点。此外，它还使用了提议的虚拟压缩机技术，能够进一步降低资本成本。两个案例研究被用来验证所提出的方法。使用了文献中的案例研究以及使用巴西炼油厂真实数据的第二个案例。与最初提出的网络相比，该模型通过优化实现了灵活的设计，运营成本降低了13.8%以上（例1）和16%以上（例2）。对于这两种情况，该程序都可以找到一种具有成本效益的灵活设计，可以与整个流程经济的炼油厂生产计划相结合。