An approach for a simulation-based analysis of business processes using the systems modeling language (SysML)

For the design and production of extensive systems with many configuration options, business processes (BPs) for creating and customizing engineering data, e.g. technical drawings or manufacturing orders, are established. Such BPs can include multiple process stations from different departments and companies resulting in a network of highly interrelated process stations that is considered in this paper. A suitable allocation of resources to process stations and a reasonable sequence of inputs to the BP avoids bottlenecks, reduces idle process stations as well as allocated resources, and increases process efficiency. Estimating effects of changes to the process is a challenging task due to the number of involved process stations and their dependencies so that a process simulation is a promising approach. This paper presents a toolbox approach for a simulation-based analysis of such BPs using the Systems Modeling Language (SysML). Motivated by an industrial case study that initiated this toolbox development, SysML was selected as it allows for an interdisciplinary approach, bringing together model-based product-line development often based on SysML and, for example, manufacturing and supply chain process modeling. The toolbox supports the simulation of process scenarios with variable resource allocations and different types and sequences of objects used as process inputs. Objects are allowed to be updated, transformed, split, and merged throughout the BP. Simulation results are average process times, idle times of process stations, and number of queued orders at a process station. Results enable the assessment of scenarios and the selection of a suitable resource allocation and order sequence. In the industrial case study, the toolbox has been practically applied using a specific SysML tool. General and tool-specific observations and experiences are provided and discussed, including possible mitigations for tool limitations.