Abstract ADNOC manages a complex, multi-entity value chain with hundreds of process units and over 200 planned shutdowns each year. Coordinating these shutdowns to maintain asset integrity while maximizing value requires navigating an enormous, combinatorial search space. This paper presents the design, development, and deployment of the VCO ADNOC Shutdown Tool (VAST), an AI-enabled optimization application that aligns shutdowns across ADNOC's integrated network. VAST combines genetic search for shutdown-date selection with linear and non-linear optimal flow solvers in a hybrid architecture. We describe the modeling journey along with five algorithm versions that culminated in a parallelized, deterministic-assisted solution. User Acceptance Testing (UAT) covered 93 tests with 87 passes (94%), validating network models for upstream, gas, refining, and petrochemicals, and confirming reporting, flowsheet visualization, and API integration. Solver benchmarking identified HiGHS as the preferred LP engine for the full-network problem. Runtime scales with the number of enabled shutdowns (illustratively ~2 hours for ~72 and ~4 hours for ~175), motivating hardware and algorithmic speed-ups. We share lessons learned on model validation, repeatability, usability, and IT architecture, and outline a practical operating model for annual shutdown planning and sensitivity analysis. The resulting system has been deployed in production with SSO, role-based access, and documentation handed over, enabling ADNOC to run integrated shutdown-alignment exercises and scenario comparisons at enterprise scale.
Azoz et al. (Mon,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: