It is very common to use a flowchart or a VSM (Value Stream Map) for visualizing a business process, an operation process, or a manufacturing process. Not only both flowchart and VSM are very good visualization tool, but the flowchart can easily explain a loop, a branch and a merge in the process, and the VSM can easily explain the cycle-time and the waiting-time in the process. Both flowchart and VSM can be used for the process improvement.
But both flowchart and VSM are the statistic model. They are pretty hard to use when a problem in a dynamic behavior needs to be solved. It is because the statistic model is hard to simulate the process with the events and the parameters which dynamically change. To solve a problem in the process which behavior dynamically change, Discrete-Event Simulation Model is used instead of the flowchart or the VSM.
I know the following commercial software for the discrete-event simulation.
- SimEvents library – MATLAB/Simulink
- Arena – Rockwell Automation
- Extendsim – Imagine That Inc.
They are similar to Microsoft VISIO, and can do almost similar things. But my favorite discrete-event simulation software is SimEvents because the library is so powerful and flexible with MATLAB/Simulink functions.
Of course, the biggest advantage with the software is the capability of dynamic process simulation by changing the parameter values on the process model. In other words, you can make an experiment of the process at your desk. The experiment (i.e., discrete-event simulation) on the process model is much cheaper and faster than an actual implementation of a new process.
I am call it as Process DOE (Design of Experiment) because you can use very similar techniques of DOE when the process model is small. You can apply the Screening DOE to the process model to identify the main factors in the process. If you apply the Modeling DOE to the process model, you can create an analytical mathematical model of the process. And if you use the analytical mathematical model for optimization, you can tune the parameter values on the process model.
Even when the process model is too big for the Screening/Modeling DOE, you can apply the Monte Carlo Simulation to the process model to identify the capability of the process with probability of distribution.
The buffers (or queues) in the process dynamically change their status. The number of waiting queues depends on the frequency of events and the capability of the process. If you simulate the process model, you can dynamically identify the number of WIP (Work in Process/Progress), the number of queues, and the bottleneck process. You can solve a problem in the process using TOC (Theory of Constraints). Implementation of a new solution to the actual process is risky in terms of time and cost, and the probability of success is unsure. If you simulate the process model and analyze the result statistically before the actual implementation, you could say that the new process will be success with 95% confidence level, and the new process reduce the lead-time 15% +/-2%. This can be a powerful statement in your project presentation.