Consultants use a variety of tools for their six sigma process improvement projects, but many get confused as to when to bring in simulation, and when to make another choice.
For the past forty years, there is a litmus test that has been true for all our consulting projects that helped us decide the correct tools to use for each project. Here is where we will share that experience.
Process simulation has proven valuable in many projects, yet it can be costly and time consuming. Cost and time is especially relevant when viewed in light of risk, return and the ability to solve the problem by another method. If a project could be worth $40 million dollars and the system is difficult to solve by traditional methods then simulation is very attractive. On the other hand, advanced analysis with any method that will save only minutes out of a workers day is a waste of analysis time. So, with that broad range of possibilities how do you make a decision of which tool to select?
In our process improvement projects, we found that there is a “sweet spot” providing greatest opportunity. These areas that provide the highest consulting revenue are projects exhibiting: complexity, which usually means many steps or many options; variability in processing times and/or processing flows; resources that are shared or interdependencies; and potential for return that is ten times greater than doing the project. This “sweet spot” is also a prime area for process simulation. It is the combination of complexity, variability and interdependencies that make process improvement estimation almost impossible with other methods.
Process simulation provides a framework for defining and solving problems with complexity, variability and interdependencies by using a unified model that ties together all the information into an interactive system. The main parts of a model include:
- Data: Describing the time to process at each step
- Resources: People and equipment used in the process
- Flow Diagram: Defining flow options
- Shifts: Times of flow, people and equipment availability
- Arrivals: Patterns of items being introduced into the process
The picture on the right shows the interaction of the data defining the model. Since the unified model contains very descriptive information about the system and all of the data is linked, a change in the model will reflect what would happen during a change in the real system. It’s like watching changes ripple through a spreadsheet but with two important differences. With a simulation model, complex and variable systems can be represented, and you can watch the processing in the animated simulation model from a “birds eye” view. This means process improvement changes can be tested by observation of the model before implementing in the real system.
With other analysis types, many of data elements used in a unified simulation model are not considered or considered in less detail. For example, in a value-stream map, high level data is collected about queue sizes and average processing time. The value-stream map is a “picture” of the current state of the system. This “picture” is effective for making changes to systems that are not complex (see earlier definition), don’t exhibit variability and have few interdependencies. The reason why a value-stream map is less effective for making changes to a complex system is that the results of a change can’t be predicted. Nothing is connected to make it behave like the real system. It I just a picture of the system.
Furthermore, when a change is made to the system real system, the value-stream map or “picture” becomes invalid. All the information that defines the picture would need to be recollected after the real process “settled.” Why not just simulate a value-stream map? Well…you would have to add of the information of a unified simulation model. Then, it wouldn’t be a value-stream map.
If the problem is simple enough to be solved with value stream mapping, then don’t waste your time developing a simulation model. Value-stream mapping was developed to work on production lines exhibiting low complexity, low variability and no shared resources, which is not the “sweet spot” for simulation. Simulation modeling requires more time, but solves problems that can’t resolved confidently with other methods.
So…does a process improvement project really need to be worth $40 million dollars as a litmus test before using simulation modeling? Of course not! Some projects only have a potential of a $50,000 in savings, but the simulation model development time is in hours! Remember the “sweet spot.”