Simulation versus Scheduling – right tool for the right job

Quite often Simulation and Scheduling are confused as they share many of the same outputs. A simulation or schedule can predict the output of a given scenario and can be used for decision making but there are some important differences. Through this article I will discuss some of the common methodologies for scheduling and simulation including the benefits and pitfalls.

Simulation

There are many different methods of simulation including Excel, Monte Carlo and Discrete Event to name a few. The purpose of a simulation model is to create a digital twin of a process to make decisions based on a mathematical model. Simulation models can be used to evaluate multiple scenarios within a risk-free environment. There are pros and cons to each type of simulation model that I have described below.

Excel Simulation

Excel is a great tool that is very flexible and easy to use. A simple spreadsheet can be an effective simulation tool used to evaluate multiple attributes to calculate a deterministic result. Attributes can be changed quickly and easily to evaluate the change in the result. Although Excel is easy to use, spreadsheets can quickly become overly complex, time consuming, error prone and rely on an expert. Excel models often omit real life complexity, dependencies and variability.

Monte-Carlo Simulation

Monte Carlo simulations are similar to Excel in that a mathematical model is used to determine a calculated result. The main difference is that where in an Excel simulation, fixed attributes are used, in a monte-carlo simulation random variables and distributions can be used. For example, a machine may have a Mean Time To Repair (MTTR) and Mean Time Between Failures (MTTF) that is evaluated within a distribution. The random time between failure or to repair is calculated and will affect the availability of the machine. Monte Carlow simulations are usually created within a series of equations in Excel, Python or similar application. Models can evaluate complex systems and all kinds of probability distributions can be used. However, validation of the model can be difficult and scenarios unidirectional. To be able to create a monte carlo simulation, experience using a programming language and statistical analysis would be necessary.

Discrete Event Simulation

Discrete event simulation software applications have been widely used for 30+ years within a variety of industries. Most software applications have a graphical interface that allows the user to recreate a process within a 2D or 3D layout. Variability can be added and the model run for a simulated period of time and Key Performance Indicators recorded. Discrete Event Simulation is particularly powerful for visualising a process and making decisions to increase specific KPI’s. Simulation models often require application specific programming and experience to create a realistic model. A simulation model may be run for 1+ simulated year or multiple times to create a Confidence Level of +95% within the results. Therefore, validation of the process and model construction is essential to avoid watching the model run for 1+ year.

Scheduling

Scheduling can have many objectives including maximising throughput, minimising wait time, reducing work in progress and improving on time delivery. Through scheduling, a user can determine what tasks to deliver in what sequence to improve predefined defined KPI’s. Although there are many applications available, their purpose normally falls within the following categories.

Production Planning

Production planning within a manufacturing environment normally considers Sales Orders, Forecast, Works Orders, Purchase Orders, Items and Stock. Material Resource Planning (MRP) can be used to calculate what Works Orders and Purchase Orders are required to satisfy demand from Sales Orders, Forecast and minimum stock levels. The production plan does not include variability and will schedule tasks based on known or normal attributes, for example; earliest start date, deadline date, lead time. Production planning can be performed within Excel but is normally included within an Enterprise Resource Planning (ERP) software application. The ERP application will hold the master data for Customer and Suppliers and is the single source of data and record for the company.

Production planning using an ERP solution is deterministic based on a common set of data at a point of time. This is essential for production planning as you would not want to include variability as this would result in an unrealistic plan. For example, you would not want to base your plan on a machine breaking down next Tuesday or an Operator phoning in sick tomorrow.

Typical productions plans have a time horizon from 6 day to 3 years depending on the industry. They are generally at a high level and planned within buckets ranging from weeks to months.

Production Scheduling

Production scheduling involves the finite scheduling of equipment, labour and materials. Where production planning has a greater planning horizon and lower level of granularity, production scheduling has a greater level of detail and a planning horizon from days to months.

Most software can utilise data from a production plan, ERP or work as a standalone system with all data entered directly. Data used for finite scheduling can include the earliest start date and deadline date, resources, labour, materials, cycletimes and setups. The planned start date and planned end date can be accurately scheduled and transferred back to the ERP solution that may affect the MRP calculation for material due dates.

Production scheduling software often includes a Gantt Chart that allows the Planner to visualise the firm and proposed Works Orders. The production schedule can then be shared with the operators and accurate work lists followed. Operators can update the work list with actual data that can be used for schedule adherence reporting and updating the Gannt chart from live data.

An important feature of production scheduling software is the ability to perform ‘what-if’ analysis used for decision making. The current plan can be copied and multiple scenarios created based on data within the system. Validation of the manufacturing process can be easily performed as the Gantt Chart clearly displays the operations, times, materials, and equipment. Variability is not normally included or required as Works Orders are scheduled using real data and known constraints.

Conclusion & Summary

All of the above solutions have varied benefits and are suitable for different applications. The summary below details a high level assessment of their usages, advantages and disadvantages.

Excel Simulation

·        Easy to use

·        Deterministic results

·        Can omit real life complexity, dependencies and variability.

Monte-Carlo Simulation

·        Variability included

·        Good for financial models

·        Difficult to create and validate

 Discrete Event Simulation

·        Variability included

·        Good for ‘what-if’ scenarios

·        Good to visualise process

·        Not used for day-to-day scheduling

Production Planning

·        Typically included within ERP applications

·        Provides Earliest Start Date and Due Dates for Works Orders and Purchase Orders

·        Can be difficult to create ‘what-if’ analysis

·        Often lacks granularity to provide accurate delivery dates or work to lists

Production Scheduling

·        Used for day-to-day scheduling, capacity planning and what-if analysis

·        Visualisation & validation of the plan is simple within Gantt Chart

·        Granular level of detail can provide accurate delivery dates and work to lists

·        Does not include variability