This post was originally written for RPA Journal: https://rpa-journal.org/
Digital twin is a widely-used term which can describe very different things: 3D representation of human body, simulation model of an energy grid, or a CAD model of a car engine. The term is sometimes credited to NASA, having been coined for the purposes of spacecraft development, but there are many alternative definitions and application fields. “Virtual representation that serves as the real-time digital counterpart of a physical object or process” is one simple and versatile definition of a digital twin.
Utility of digital twins in engineering cannot be overestimated. In fact, automotive, industrial and power grid engineering already heavily rely on model-based design practices. Industry 4.0 vision of connected manufacturing goes hand-in-hand with creating digital twins of production processes. Yet, in the wider context of business processes, the idea of digital twin is just starting to take root.
Digital process twin is a virtual representation of a real business process, which in its full form comprises three main components:
- Accurate model of the business process
- Real-time status data received from the process
- Possibility to exert control over the process through adjusting the digital twin
Accurate models of business processes can be created with the help of process mining, which in its modern form applies machine learning to both enterprise application log data and the UI-level human interactions to discern patterns of business processes and create highly accurate process maps.
Process diagram as produced by a process mining suite, image credit Arvato Systems
Translating raw as-is processes into orderly to-be diagrams is probably the one step that requires the most human judgement and expertise to be done right. At this stage organizations have the opportunity to improve their efficiency and customer experience and avoid “garbage in, garbage out” mistakes in automation.
Shipment process of a hardware retailer in BPMN notation, image credits bpmn.org
Real-time data integration is made possible by using workflow engines, which record the current status of every work item (aka process instance) throughout the business process.
Business process in BPMN notation with real-time work item data, image credits Camunda
Real-time control over the process can be achieved through variable parameters of intelligent automation robots which perform certain sub-processes, as well as through diverting work items to alternate process paths depending on the current state of the overall process and the specific work item.
Benefits of digital process twins
Business process management has gone through several waves of innovation, starting with Taylors’s Principles of Scientific Management, all the way through the advent of enterprise resource planning and business process reengineering to the modern process-oriented organizational structures. Yet businesses are only now starting to realize the potential of fully connected digital process twins, and the opportunities are very significant.
First of all, real-time insight into business processes creates true transparency. What can be measured, can be managed. Knowing what exactly is happening in the businesses enables a new level of analysis and continuous improvement beyond the once-in-a-while process optimization studies.
Resulting deep understanding of the business processes provides a data-driven decision basis for implementing automation where it is most effective, eliminating guesswork from automation pipeline decision making. Furthermore, most RPA platforms can integrate seamlessly with workflow engines, thus providing an additional level of transparency and modularity for the robots.
One of the most significant advantages of using digital twins is the possibility of simulation. Any advanced engineering these days is done on a simulated digital-twin model first, before a prototype is built. Ability to simulate business processes would enable organizations to identify and reinforce potential points of failure. As demonstrated by Covid and the ensuing supply chain crises, organizational resilience has never been more important than now.
Using standardized, real-time business process models enables end-to-end process integration, involving both administrative and operational subprocesses. For example, upon receiving customer order for a single-piece flow product, the system could automatically trigger component orders to suppliers, monitor production progress and report on the current status of the individual work order, enabling unparalleled transparency into the process. Ultimately, the customer could track the status of production just as easily as tracking a parcel.
Having real insight into running processes would also provide granular data for benchmarking. Best-practice bodies such as AQPC already provide platforms for process-level efficiency comparisons, which would be brought to a new level by the availability of exact, real-time subprocess data for analysis.
Digital process twin is an approach and a philosophy, much rather than a specific stack of technologies. It can be implemented in various ways. Broader adoption of digital process twins is a specific manifestation of the larger trends toward increased connectivity, transparency and integration, and it will, in our opinion, characterize business process management in the next decade.