Process Safety

Exothermic reactions carried out in batch reactors need a lot of attention to operate because any insufficient condition can lead to thermal runaway causing an explosion in the worst case. Therefore, a well-designed intervention action is necessary to avoid non-desired events. For this problem, we propose to use resilience-based reinforcement learning, where the artificial agent can decide whether to intervene or not based on the current state of the system. One of our goals is to design resilient systems, which means designing systems that can recover after a disruption. Therefore, we developed the resilience calculation method for reactors, where we suggest the use of dynamic predictive time to failure and recover to better resilience evaluation. Moreover, if the process state is out of the design parameters, then we do not suggest calculating with the adaptation and recovery phase. We suggest using Deep Q-learning to learn when to intervene in the system to avoid catastrophic events, where we propose to use the resilience metric as a reward function for the learning process. The results show that the proposed methodology is applicable to develop resilient-based mitigation systems, and the agent can effectively distinguish between normal and hazardous states 

Nowadays, reactor runaway is still a crucial phenomen from the safety viewpoint. About 120 scientific journal articles are published every year in the last decade in which thermal runaway is a keyword. The possible cause and consequences of reactor runaway are adressed where the worst case is the explosion of the reactor. Prevention steps to avoid the development of thermal runaway include the appropriate design of the reactor, the operation strategy and an early warning detection system. The available assessment methods for thermal risk analysis are addressed in detail. Reactor runaway criteria can indicate early the thermal runaway, which criteria are addressed in this review in detail under three classes: geometry-, sensitivity-, and stability-based runaway criteria. Operation strategy of semi-batch reactors can be designed by calculating Westerterp-diagram whose evolution is cleary presented. Significant works on the field of the reactor design, operation and reactor safety are collected and evaluated. Finally possible further research areas are suggested to improve our knowledge about thermal safety, such as investigating parameter uncertainty in runaway indication or optimize the safety actions to moderate the consequences of runaway. 

Safe operation of thermally sensitive chemical reactors remains a crucial engineering issue. Thermal runaway occurs mainly due to loss of temperature control, and many chemical accidents initiated by thermal runaway can be foreseen by an appropriate analysis of thermal process data. Thermal runaway criteria can be applied to determine the onset of runaway phenomena and safety boundary diagrams can be constructed. However, several runaway criteria exist, which define the runaway-and safe zones differently. In this work nine commonly applied thermal runaway criteria were analyzed and compared based on their critical curves. As a result of this analysis, two new criteria were developed. Reliability of the derived criteria were investigated, where the occurrence of real runaway was determined based on the number of indications applying different criteria. The two new criteria were tested in three general reaction systems and in case a complex problem, where feeding trajectory of fed-batch reactor was optimized applying new criteria as non-linear constraint. One of the new criteria shows the highest reliability in indication of runaway development from all investigated runaway criteria.

With improving industrial chemical technologies the Hazard and Operability Analysis (HAZOP) or other safety analysis becomes more difficult to perform, because the technologies become more integrated and complex. Applying dynamic HAZOP can help the experts to explore all hazard situations in a technology based on dynamic process simulation. In this work a framework for exploring hazard events is suggested to support the safety analysis. The proposed framework works via OPC (Open Platform Communication) connection between MATLAB and Aspen HYSYS dynamic process simulator. Concentration of cumene-hydroperoxide (CHP) in vacuum column is chosen as a case study to show how the framework can be applied to analyze process safety related situations. The hazard events are generated through controller failures with significant changes in setpoints for a specific time. The events can be ranked based on the consequences, which can be significant in different perspectives. The most important is to analyze the temperature and pressure trajectories during the event, although the required time and energy to reach the normal conditions again can be evaluated too. All these parameters are considered in the qualification and ranking process of each investigated process malfunction.