Quantitative Multivariant HAZOP Analysis Of A Complex Process System Using Dynamic Simulation

HAZOP is one of the most widely used process hazard analysis (PHA) tool. Currently, HAZOP is undertaken at design stage by a multidisciplinary team of human experts and more recently, by expert systems with a knowledge or model based framework that aids the human team. In the interest of conserving time, economy and more comprehensive results, research for computer aided HAZOP techniques has been undertaken. The knowledge based framework is one such attempt; however it may not be comprehensive enough due to dependence on human expertise. Model based framework can furnish comprehensive plant behavior if the simulation used is accurate and extensive.This study takes Residue Fluid Catalytic Cracking (RFCC) unit as a case study. RFCC is a very important chemical process in petrochemical industry since it cracks the long chain hydrocarbons into smaller and economically viable products. It is also a very complex chemical process with many variables affecting each other along with the control strategy making the deviation propagation hard to predict. The purpose of the study undertaken is to make HAZOP less expert knowledge intensive and overcome the shortcomings inherent in the process. Dynamic simulation coupled with multistep and multivariant analysis yields accurate deviation effects. The dynamic simulation being employed here is an accurate first principle model with real-time and accelerated computing. Methodology includes the use of dynamic simulation to provide high fidelity plant data for a complex chemical process, with the ability to study multistep effects down the process line. Qualitative analysis is undertaken for understanding the process and its dependencies by observing trends in the data responses and constructing signed digraph and adjacency matrix. Quantitative analysis by using the data obtained through simulation, to give design intent data for normal operating condition and effects of the introduced deviations.The quantitative multistep analysis can be further divided into two parts:single deviation and multivariant deviation propagation. In the first part two unique deviations, namely increase in steam and increase in raw feed oil flowrate are investigated in case study 1 & 2 respectively. The deviations effect product quality adversely enough to call for remedial action; however the effect dissipates along the process line. In the second part, three parameter deviations are considered in combination. Both case study 3 & 4 consider the same parameters with one different guideword-parameter combination. The effects minimize for the first combination but exaggerate for the second one causing an unplanned shutdown instead of the expected temporary loss of product quality. The multistep, much less the multivariant analysis is very hard to get through conventional HAZOP.The scenario consequences given by the study may have been easily missed during conventional HAZOP due to the propagating nature and is extremely undesirable due to the downtime, economic losses and possible harm it poses to the equipment and personnel. It can also be used to present successful remedial actions for deviations as well as plant behavior information for operator training. The study presents the multistep effects of a single and multivariant deviation accurately and quantitatively in order to aid HAZOP; make it more comprehensive, improve predictability of possible hazard identification arising from deviations and ascribe actions to mitigate the effect.