Accelerating Start-up Using Digital Twin Technology for Combined Virtual Commissioning and Operator Training

Commissioning of process plants is a time-consuming and potentially unpredictable stage on the critical path towards a fully-operating plant. Key technologies for assuring start-up are virtual commissioning and operator training, but these are often disjoint activities, leading to unnecessary expense and investment in model development. This paper presents an approach that streamlines the commissioning and start-up phase while minimizing the total cost of ownership and also results in a digital twin that can be used for multiple further activities over the plant’s lifetime.

During the virtual commissioning phase, the automation system design is tested against a relatively simple virtual representation of the process and equipment, which simulates the responses of all devices that communicate with the automation system. Virtual commissioning aims to debug the logic and programming of the control system and Human Machine Interface (HMI) in a virtual environment before commissioning any physical equipment. Correcting even simple bugs can cost orders of magnitude more and add significant delays to start-up.

Similarly, the training of plant operators on the HMI is vital for them to carry out their tasks effectively and safely, particularly on a new process, and to avoid unnecessary shutdowns in the crucial first few months of operation. This training is conducted using an Operator Training Simulator (OTS), which is usually created from scratch as part of the automation system contract.

The process start-up can be significantly streamlined if the virtual commissioning model can be integrated with a full physics-based model of the process. Such an integrated system presents multiple benefits. A consistent look-and-feel of the automation system and OTS means that operators can be trained before the plant is fully commissioned. The use of high-fidelity process models results in an OTS that more accurately represents the dynamic behavior of the plant. Shorter construction time with reduced overall cost can be achieved. Finally, the resulting fully-fledged digital twin can be used for multiple other purposes, such as designing and verifying new control strategies and testing new recipes or revised operating policies.

This work shows how the simulation models developed for testing the automation system can be interfaced with dynamic process models developed and used during process design to provide a high-fidelity OTS. The approach leverages investment in both areas to minimize the incremental cost of creating an OTS while providing a fully-operational digital twin of the process that can create value in other areas.

We describe the workflows of creating such a high-fidelity digital twin, including the challenges and how to address them, using an integrated reaction-and-distillation process as an example. We also describe how the digital twin is a crucial element of a process lifecycle approach that [1] leverages existing investment in R&D and digital design models to reduce start-up time and cost and [2] goes on to utilize such models during operation, in conjunction with real-time plant data, as part of digital operations applications for equipment and process monitoring, soft-sensing, real-time optimization and what-if analysis based on current plant state.