(633h) Using AI to Assess the Integrity Management System of the Rhds Unit

In the past decade significant milestones have been achieved in controlling major accident hazards. Management systems are effectively used to demonstrate legal compliance, assess various risks, and establish operating procedures that address health, safety, environmental, and asset integrity risks. Management systems also tailor workflows to supervise the implementation of the organization's procedures, and invoke continual improvement through audits, reviews, and investigations. Management systems deductively use a multi-tier feedback loop framework to improve performance, and a variety of risk assessment techniques (inductive and deductive) to model likelihood as well as consequences of failure. However, management systems are intrinsically reactive, and are challenged by the lack of integration, conformance, omissions, updates (management of change), technology integration, and no input to improve the organizational structure (to simplify workflow and improve conformance). Therefore, in a refinery an integrated control system is required to interactively assess changes to the process units, and prioritize the response from pertinent refinery teams. As such, the integrated control system requires an architecture, and interface that supports the incorporation of the refinery integrity management system into the refinery Digital Control Systems (DCS). The refinery integrity management system includes Laboratory Information System (LIMS, shared with manufacturing product QC/QA), Risk Based Inspection (RBI), online corrosion monitoring (OCM), online equipment monitoring (integrity operating windows (IOW)). Significantly, quantitative risk assessments are no longer a one-off study, rather a continuous cradle to grave assessment of the refinery performance, and commitment towards investors, licensors, stakeholders and the jurisdiction. Aladasani, et al. (2016). Featured Article (NACE INTERNATIONAL). Materials Performance. 55. 28-32. 10.5006/MP2016_55_9-28.

In this work, artificial intelligence (AI) is used to provide a comprehensive assessment of the integrated control system used for the Residue Hydrodesulfurization Unit (RHDS). The RHDS unit was chosen because in par with the SRU it is prone to the highest number of failure mechanisms based on its process stream properties, and operating conditions, inferred from the Aladasani, et al. (2017). Guidance on enhancing process safety management. Topical Conference at the 2017 AIChE Spring Meeting and 13th Global Congress on Process Safety, Vol (1), Pg. 536-551. ISBN: 978-151084135-2. The AI scope includes the review of the failure mechanisms triggers/preconditions mentioned in API 571 considering published, operating experience, lessons learnt from industry, and any omissions or lapses in API-571. Additionally, the AI scope focuses on enhancing the RHDS integrity by reviewing the RBI, OCM and IOW based on incident reports, and published best practices of leading refinery operators. Finally, the AI is used to analyze the integrated control system architecture in terms of strengths, limitations and required improvements to manage the integrity of the RHDS unit.

In the first step, the AI is used to generate benchmarks of the integrated system components, mainly (a) the Programmable Logic Controller (PLC)/DCS code and interface, and benchmark (b) the leading technologies in corrosion monitoring, risk-based inspection, and condition monitoring including their hybrid capabilities. In the second step, the AI is used to recommend PLC/DCS standardization to improve IOW enforcement, upgrades to the RBI software to improve detection, further to to improvements to condition monitoring system, and the IOW logic such as multivariable decision matrices.

The work in this paper, provides valuable insights to refinery operators, technology/solution providers, engineers, researchers, and AI developers. The use of AI as an auditing/assessment tool is demonstrated in an array of technical specialties in a complex refinery unit. Furthermore, the AI adopts a holistic approach to review the entire integrity management system, cross check libraries of information, move across the refinery organization, and without any barriers. As such, the AI has a powerful feature for management, stakeholder, and oversight bodies. However, the AI has a bias because of its deterministic approach. To improve the AI performance constant validation/verification is required along with a subject matter expert to overcome its bias. The AI concludes the following performance metrics, over a short/medium-time horizon if its proposed recommendations are implemented, 30% reduction in corrosion-related failures 95% IOW compliance, 40% decrease in unplanned shutdowns, 20% longer equipment lifespan, and 50% faster response to IOW breaches. The authors believe that the AI recommendation would improve the integrity management of the RHDS unit but at a moderate pace due to human factors.