Abstract
In a petrochemical plant, training new operators to operate plant safely and consistent is a need. Conventional training for operators relies on senior operators’ experience. When experience operator leaves or resigns, it will take longer duration to train new personnel to gain the equivalent experience of previous or senior operators. The operators control the plant by using Distributed Control System (DCS) as an interface and ensure the reliability and productivity of plant sustains. The more challenges they face, the more experience they gain. Most of the operators are trained by real time events they faced on the job but errors or plant upsets should be avoided. It is estimated that in the United States alone, 10 billion USD is lost annually in the process industry caused by errors when faced with plant upsets. Operator training plan is therefore the crucial step to reduce the losses as well as the risk of operating the plant. Operator Training Simulator (OTS) is an alternative solution to safely provide the operators with various realistic operational conditions of the plant, including start-up, shutdown, and emergency conditions. This article discusses on the methodology of building an OTS for Ethylene plant, that are beneficial for educating operator response when faced with critical process upsets and succeeding quicker start-up. A dynamic process model was developed to give accurate dynamic behaviour of the plant process mainly during start-up, shutdown and upset conditions. Testing and verification processes of the models such as Model Acceptance Test (MAT), Factory Acceptance Test (FAT) and Site Acceptance Test (SAT) were deliberated. Few issues or problems faced during the testing were also discussed. Based on the experience of developing the simulator, process model was developed using mass and energy balance method does not reflect the actual plant conditions. It might be suitable for commissioning a new plant. However, once the plant is started all the parameters need to be realigned to meet the actual plant data. Process model for existing plant should always apply the actual plant data as a primary data. Especially during MAT, FAT and SAT processes, experience of the panel man is the best input data to get the best output of the simulations.
