With over 15 years of experience in the field of simulation modelling and analysis, the conversation starter for a potential project that I hear the most is, “We need a simulation. Can you help us with that?” The first response that comes to mind is, “Why do you need a simulation and what benefit will it bring?” Since this is a broad and general question, most people struggle to answer it. Their simulation request is often defined by more vague triggers as a video seen on the internet or because “the neighbors have one and they seem happy with it,” instead having a good business understanding of what problem they want to solve.
Don’t get me wrong, this is not a blog to talk you out of simulation, instead this is a blog to help you understand the power of simulation. To give you a quick insight into where and when you should use it. To present a framework of what different types of simulation can be defined and what use cases could be applied to these different types.
Simulation is an extremely powerful tool that, if done properly, will provide you great benefit and in most cases a fast ROI. Its use however is also often misunderstood, and it comes typically with great costs (labor for modelling and analysis, software and computing power) and a relative long project duration as compared to other less advanced techniques. Not every question needs a simulation to provide the answer, but for some problem’s simulation is the only way forward.
Types of Simulation
The first questions that the simulation professional wants to get answered are ‘what problem do you need to solve?’ (the problem), ‘why do you need a simulation?’ (the purpose) and ‘for whom do you need it?’ (the stakeholders). Answering these questions properly will typically guide the project towards four different possible use cases:
A visualization is not a simulation. It is a graphical (most often 3D) representation of the operational system combined with detailed animated features. The end user will see their (future) operation in great detail. It helps stakeholders to gain an understanding of the design, how the system will operate and interact with its environment. It is a great communication tool to get all stakeholders aligned or to use for training purposes. These benefits are enhanced when a visualization is used in combination with virtual reality solutions. Stakeholders can walk or fly through their operation long before it is actually built. Visualizations however do not generate analytical results that can be used for further processing to understand the system dynamics, capacity, throughput, etc. When we support our clients with visualization, we often use different tools that provide a better experience with less cost and a faster turnaround than typical simulation software.
While a proof of concept gets into the dynamics of the operation, it does not, however, represent a detailed design. A proof of concept will often be done before the detailed design is developed. The client wants to get a basic understanding of what effect the dynamics of the operation will have on its overall behavior. There are a variety of techniques that can be used to do a proof of concept, a simulation is just one of them. The benefit of simulation, as compared to other techniques, is that it gives stakeholders an understanding of how system dynamics will affect the operation. That gives a clearer vision of what capacities are required in order to model the dynamics and prevent operational blockages later on.
Within a detailed analysis, a highly detailed dynamic (3D) model is created based on a (future) operational system design including its internal and external dynamics. It is used to predict and forecast in great detail what the responses the operation will have to itself and to its environment (predecessor steps, successor steps, planning, etc.). A powerful set of what-if scenarios are executed, which gives the stakeholders a strong understanding how to best use the (future) system to gain maximum effectiveness at minimum costs. Simulation models for detailed analysis contains the layout, capacities, material flow and control logic (ERP, MES, WMS, WCS I/O, etc.), to model the future system. Dynamic input parameters such as SKU master characteristics, process time, sequence, customer orders, supplier orders, failure rates (MTBF, MTTR), etc. will feed the model and define the different (what-if) scenarios. While running a scenario with the model, it will collect all sorts of operational data, similar as in the real system, but then in a really short time frame. This data will be used for further analysis and help to make management decisions. Typical data that is collected by the simulation model include, utilization rates, throughput (time), queueing behavior, WIP, etc.
A digital twin is a rapid growing technique in product design. However, in process design, and especially in logistics engineering, it is used today as a bit of a buzz word that lacks a clear definition. Nonetheless, there are slowly appearing more real use cases showing great benefits when using this technique in complex operations. In common with a detailed simulation, there is a (3D) visualization and dynamic modeled environment of an operational system. The main difference as comparted to regular simulation or animation is that the model does not exist independent by itself. Instead it is under control of an external (IT) system such as ERP, WMS, WCS or I/O. There is an interface to a real (IT) system that is controlling the operation (physical twin) and which is also controlling the simulation model (digital twin). Digital twinning can be used to (among other use cases):
Statistics Collection
The simulation model collects all kinds of detailed statistical data elements, similar to or better than what is collected in the real system. The simulation will do that in a really fast time frame, a month of logistics operations can be executed in a couple of minutes up to several hours. An analysis dashboard helps the user to read and understand the results.
A proper simulation collects two types of statistics:
Predictive statistics are the overall key performance indicators (KPIs), such as orders handled per hour, lines picked per hour, etc. These are the statistics you want to improve, predict or get a better understanding of. Explanatory statistics on the other hand are the detailed statistics that help to understand the predictive statistics. Examples are orders per mission, distance travelled, etc. These statistics help to understand why the predictive statistics improve or get worse. This way the simulation users can solve bottlenecks, improve throughput, understand system capacity, etc. for new or existing systems.
Coming back to our original question, “Why do you need a simulation and what benefit will it bring?” When executed properly, simulation is a tool that can be used to improve operations, reduce risks, and help prevent high or unnecessary investment costs. Finding the right type and level of simulation to fit your project requirements is the first step in an exciting analytical journey.
—Auke Nieuwenhuis, St. Onge Company
