My background is in process improvement, and I started modelling factory simulations over ten years ago. I first used modelling to help evaluate and justify improvement projects and then used it to help validate scale-up plans. As a factory simulation modeller you want two things:

1. For your model to be perceived as useful and,
2. any recommendations coming from the model to be implemented.

I found starting a modelling project has lots of parallels with process improvement projects. In that both projects should have clear goals and align with the business’s strategy. Without this you will struggle to get your projects supported and any recommendations implemented.

There are two things you need to get right before you start creating a factory simulation (or also known as manufacturing simulation):

1. Understand what operational question you are trying to answer.
2. Accurately define a process map.

Operational problem definition for factory simulation

I often talk about defining the operational question before starting any modelling project. This involves talking to the process stakeholders and the people commissioning the project and finding out what they hope to achieve with the model, and what other operational concerns they have. The question needs to be defined in a way that is deliverable by the model. Having a question defined gives purpose to your model, and helps you focus on work that helps answer the questions. Without this purpose it is easy to waste time on collecting data or modelling processes that are not relevant to answering the question.

A second thing to ask yourself is “does the question align with the business’s strategy”. If you can define the question in a way that it clearly aligns, then you will find it much easier to gain support in constructing the model and you are more likely to have any finding the model produces acted upon.

Defining a question that meets all these criteria, and that all the stakeholders are happy with can be difficult. However, it is worth the effort and makes the later stages of creating the model much easier.

Examples of questions that a simulation can answer

The complexity of the model depends upon the question you need to answer. A well-defined question can help focus and speed-up the data collection and modelling phases and helps ensure you get the answers you are looking for. I like to ask myself the question at the end of each project phase to ensure the project is on track.

Here are some typically questions we have been asked:

• Where are utilisation losses coming from?
• What is the optimum re-order level?
• What is the payback period for this new piece of equipment?
• How quick does a conveyor need to be repaired?
• What is the best product mix for our current resources?
• What is our activity-based costs?

Check some short videos demonstrating these common questions. You can find a YT playlist here: https://www.youtube.com/playlist?list=PLHOPtCr3e1lkCDf_ALM8UoTT1eSsRARby

Managing the relevancy of secondary questions

It can be very tempting to add in a few more secondary questions. The phrase ‘wouldn’t it be nice if …’ is often heard. If you are not careful this can lead to mission creep and significant delay getting the answer to your primary question. Before adding a secondary question to your project brief, consider the following:

• Is the secondary question important and will not significantly delay answering the primary question? If the answer is ‘yes’ then add the question to the model brief, as it will add perceived value to the model with little extra effort.
• Is the secondary question important but will significantly delay answering the first question? If the answer is yes, then leave out the question but build the model in way that it can be developed to answer the secondary question in the future.
• Is the secondary question important? Does it naturally fit with your planned model? Is it worth the extra effort to answer the secondary question? If the answer is ‘no’ to any of these, then leave the secondary question out.

Process mapping for factory simulation modelling

Once you have defined your question you can now take a look at the process you are planning on modelling. To create a useful model, you need a good understanding of the process itself. The best way to do this is to map the process with the process owner. This could be the engineer who designed the process or the manager of the process and operators that run the process. You will likely find discrepancies between how the process is thought to be run and how it is actually run. It is important to be clear which version of the process you are modelling 😊.

Developing a process map that all the stakeholders agree on is important, because a mistake here could mean making structural changes to the model at a later date. I personally struggle with late structural changes, as they can be complicated and time consuming to implement and generally lead to the tricky bugs that are hard to find.

Tools and methods for process map drafting

If you are very lucky an engineer will provide you with a process flow diagram (PFD). This is a great start, but PFD’s can be complex and contain all the back-up lines that are not used in normal operations. And, it could also be out of date. I would still create the process map with the process stakeholders and use the PDF as supporting information.

The best method to create a process map is to go old-school with sticky notes, pencils, and coffee. Lock yourself in a room with the process owner and some operators, and work through the process step by step, going backwards when necessary. Depending on the complexity, the first draft should take 1-3 hours. Take photographs of your hard work and convert it to a process flow chart. Make sure you send a copy to the stakeholders for comment, and if you have the chance walk the line and validate the process map. And as a modeller you have a great opportunity to ask all the daft questions, so take full advantage of this.

Model processes first, then data processing

At the start of a project, it is more important to get an accurate process map than it is to get accurate process data. Chasing accurate data often delays the projects and there is a risk of spending too much time collecting unimportant information. Manufacturing simulation software is designed to use variables and internal spreadsheets to hold data. These are much easier to manage and modify as the project develops. I tabulate relevant process data using the internal spreadsheet in Simul8 and populate it with best guesses. This has the benefit of allowing other people in the project team to gather relevant process data whilst you focus on the modelling.

Finally, draft and implement the factory simulation

Armed with a well-defined question and an accurate process map, you are now ready to draft your model.

Whether it is a new process improvement project or a modelling project, at Production Support 56 we always start by understanding the problem and the process.

Related content

If you are interested in factory simulation and manufacturing simulation you might be interested in some of the following SCDA:

• Link: Inventory simulation for optimized stock
• Link: Crane yard simulation in AnyLogic
• Link: Conveyor routing simulation in AnyLogic
• Link: Manufacturing simulation for plant design
• Link: Conveyor system optimization and simulation
• Link: Visual Components financial KPI simulation
• Link: AGV simulation of part routings in AnyLogic
• Link: Parking lot simulator with simmer in R
• Link: Simulation-based capacity planning
• Link: Machine learning and discrete-event simulation
• Link: Queueing systems solved analytically and simulated in AnyLogic
• Link: simmer in R for discrete-event simulation
• Link: An introduction to simple conveyor line models in AnyLogic
• Link: Simulation methods for SCM analysts
• Link: Backlog simulation in FIFO production system
• Link: simmer R-package applied to simulate simple receival inspection process

Dr Michael Stephenson

Co-Founder of Production Support 56. Improving operational performance with process improvement, process development and simulation.