Uncertainty in refinery planning presents a significant challenge in determining the day-to-day operations of an oil refinery. Deterministic modeling techniques often fail to account for this uncertainty, potentially resulting in reduced profit. The stochastic programming framework explicitly incorporates parameter uncertainty in the problem formulation, thus giving preference to robust solutions. In this work, a nonlinear, multiperiod, industrial refinery problem is extended to a two-stage stochastic problem, formulated as a mixed-integer nonlinear program. A crude-oil sequencing case study is developed with binary scheduling decisions in both stages of the stochastic programming problem. Solution via a decomposition strategy based on the generalized Benders decomposition (GBD) algorithm is proposed. The binary decisions are designated as complicating variables that, when fixed, reduce the full-space problem to a series of independent scenario subproblems. Through the application of the GBD algorithm, a feasible mixed-integer solution is obtained that is more robust to uncertainty than its deterministic counterpart.

中文翻译：

---

不确定条件下多期炼油厂规划问题的优化

炼油厂规划的不确定性对确定炼油厂的日常运营提出了重大挑战。确定性建模技术通常无法解释这种不确定性，从而可能导致利润减少。随机规划框架明确地将参数不确定性纳入问题表述中，从而优先考虑稳健的解决方案。在这项工作中，非线性、多周期、工业炼油厂问题被扩展为两阶段随机问题，并表示为混合整数非线性程序。在随机规划问题的两个阶段使用二元调度决策开发原油排序案例研究。提出了一种基于广义Benders分解(GBD)算法的分解策略解决方案。二元决策被指定为复杂变量，当它们被固定时，将全空间问题简化为一系列独立的场景子问题。通过GBD算法的应用，得到了一个可行的混合整数解，它比确定性对应物对不确定性更鲁棒。