Photo: Unsplash
Computers are well-suited for solving complex mathematical relationships involving large sets of variables. This capability sets them apart from humans; computers can be programmed to analyze relationships by solving the same equations hundreds or thousands of times.
One of the best applications of this capability is system simulation. Here, a “system” can be anything that has one or more processes, with a set of inputs and a set of outputs. Simulation enables analysts to model a system and analyze what happens next i.e. what outputs are realized under different initial conditions (inputs).
A good modeling and simulation methodology can result in accurate models, even when some parameters have a high degree of variability or uncertainty.
Simulation programming can employ simple simulation modeling techniques or, on the other hand, be extraordinarily complex. They are used to facilitate research in all academic disciplines, including meteorology, sociology, biology, physics, and engineering. Simulations are also extensively used by organizations to analyze various business processes.
As an example, consider a fast food restaurant scenario. By modeling different aspects of the restaurant (such as how many employees are working at any one time, how fast each one can serve customers their order, and how frequently customers come in for food), the restaurant “system” can be analyzed under various initial conditions to reveal its performance , such as:
- How long customers wait
- The maximum number of customers waiting for food
- How many customers can be serviced over the course of a business day
These performance metrics can be used to determine areas that can be improved to increase customer throughput and shop profitability.
Another well-known application of computers’ number-crunching abilities is optimization. If the mathematical relationships between variables are known, a computer can optimize a parameter—that is, find the values of the input variables that maximizes or minimizes an output value. The mathematical expression that relates the inputs to the output is known as an “objective function.”
So, for example, if the relationships among revenue, warehousing costs, transportation costs, and inventory costs are known for a given enterprise (or can be estimated), then the set of inputs that result in minimized total logistics cost can be determined.
What’s the Difference Between Simulation and Optimization?
Although simulation and optimization are similar and leverage many of the same computational techniques and algorithms, they are different activities. Each has its advantages and disadvantages, and each is better suited for certain types of problems. Here are some key differences between them:
- “What-if” analysis: Simulation is better suited to observing the performance of the simulated system by tweaking the initial conditions (that is, the values of the input variables). Optimization is used more often to determine an optimal system design.
- Constraints: Successful optimization depends on properly identifying the constraints placed on various parameters—for example, a business might have a maximum number of employees it can hire to work on production lines. With simulation, the analyst starts with realistic values for inputs and modifies them within reasonable ranges to determine what happens with the outputs.
- Influence of randomness: Simulations can account for random variation in the parameters—in the barbershop example, each barber’s hair cutting speed can be expressed as a normal distribution around an average. This variability can make a large difference in the accuracy of the results. Optimization works better clearly defined mathematical relationships that don’t have variability.
- Planning and decision support: Optimization methods can be used to support both tactical and strategic planning decisions, because they provide a single “best” answer to a given problem. This is one of the advantages of optimization. Simulation, by contrast, is considered more exploratory.
- Modeling difficulty: Simulations are generally easier to model, because fewer assumptions need to be made. An optimization solution requires either more assumptions about the inputs or more computing power to deal with all the different variables to calculate the optimized result.
Read More About Simulation Modeling:
4 Types of Simulation Models to Leverage in Your Business
7 Key Advantages of Simulation for Business
Value Stream Mapping Examples for Different Industries
How Simulation and Optimization Can Work Together
Although simulation and optimization help solve different problems, they can work together to drive business results. How?
By application of simulation techniques for a system—say, the receiving dock at a warehouse—on the basis of observed factors that influence the efficiency, throughput, or other metrics of the receiving team, a business can get a feel for the factors that have the most pronounced effect on the outputs.
Armed with these insights and data simulation techniques, the business can then make better assumptions about the mathematical relationships between the parameters, which drives better optimization—that is, better decisions about what to change and by how much to improve the team’s performance.
Examples of Application of Simulation in Business
Data simulation tools are used in businesses of all sizes and in all industries to analyze current processes and determine where to focus on improvements. Here are some simulation system examples:
- Royal Dutch Shell: Used a simulation based model to support vessel servicing of offshore oil platforms, including factors such as vessel capacity, storage at port facilities, and more. The simulation showed Shell where best to invest in improvements.
- Cancer center: A major medical center in the Midwest modeled internal patient care processes. This helped them determine the best arrangement of different types of patient populations, thereby minimizing patient, doctor, and caregiver travel times and maximizing operating room utilization.
- Agricultural logistics: A sugarcane producer in Brazil used simulation to improve operational capacity and reduced capital costs, while increasing the efficiency of the vehicle fleet carrying sugarcane from plantation to mill.
- Walmart: Before investing millions into a robotic based system that picks groceries for its online grocery pickup system, Walmart ran a simulation to test viability before making the change – you can read the case study here.
In all simulation exercises, it’s important to account for all the factors that influence the system’s performance and to accurately characterize each one. Simulation modeling requires keen observation and data analysis—in some cases, large amounts of data—to get an accurate picture of the system. It’s also important to validate the model by comparing model data with real system data.
The better the model, the better the simulation’s response to different inputs; good data modeling and simulation can result in better optimization.
Simulation and optimization can therefore be seen as two complementary approaches to solving business problems. With advances in big data simulation software and computing power, simulation and optimization will become increasingly important tools in every company’s decision-making toolkit, enabling better insights and better business decisions.
Wish You Could Prevent Unscheduled Machine Downtime?
You Actually Can With Predictive Modeling