We are big promoters for the use of simulation modeling to test Mixed Model line designs from a work-flow perspective. Simulation is really a requirement because most mixed model lines experience a high level of variability, from a number of different sources. Since you don’t want to be making major changes after the fact, the best time to improve a design is before you implement it, and simulation modeling will help you to do that.
Other than work-flow, however, what about the use of simulation to test your material delivery system? When the question of whether simulation is needed or not, we turn to the assessment model shown below.
On the X-axis we are showing the level of Interdependencies: how many resources (workstations, machines) are connected sequentially? A long assembly line, for example, would be an example of high interdependency. On the Y-axis we have the level of variability. In a Mixed Model value stream this comes from a variety of sources: normal human variability, machine capability, different work content for different models, different product routings, and a varying mix and volume of products from day-to-day. The matrix above states that if you have both high variability and high interdependency, your need to simulate the flow is also high. If you are high in either dimension, it is also recommended but not as critical. If you have low interdependency and low variability, simulation can be considered optional.
So where do material delivery systems fall, like testing delivery routes for example? Generally speaking they fall in the low interdependency and low variability quadrant. They are low interdependency because there are very few hand-offs. One tugger driver makes all of the deliveries. It is low variability because the work should be very standard, and not subject to great variation. I have seen several delivery route simulation models, but very little (in my opinion) was learned by viewing them that could not be learned in a static Excel spreadsheet.
Modeling material consumption is a potential simulation target that has greater potential. Testing the robustness of your Kanban system by testing usage against a variety of production schedules may allow you to identify potential weaknesses in either delivery or inventory quantities. A model like that would require access to Bills of Material, and would probably be an extension of a work-flow model as well. In reality I have not seen any models of this type, probably because the cost of developing them would be high, and the potential benefit would be low.
In summary, while we are big advocates for the use of simulation modeling to test the work-flow of Value Streams or production lines, we see less of a need for modeling material delivery systems.