Simulation is the most flexible of all operations research modeling techniques. It is almost always implemented on a computer that is used to generate random numbers and keep track of various statistics. The essence of a computer simulation is the representation of the elements of a situation or system on a computer. The system being modeled can either be one in existence or on the drawing board. The analysis of the simulated data enables the manager to make inferences about the real world system by manipulating the computer representation.
Computer simulations play two broad roles.
- Build skills to manage similar systems by offering a broader range of experiences and challenges than the individual could accumulate purely through actual work experience.
- Evaluate different policies for the design and/or operation of the system.
The military is one of the leading users of simulations as a training tool. Flight simulators are used by civilian airlines as well as the military to train and retest pilots. War game simulations enable senior military commanders to learn from handling a wide variety of complex battle situations. Business schools also use simulations to teach their students by providing them with simulated experiences of a wide range of business problems and opportunities. In each case, the strength of simulation as a training tool lies in the ability to simulate a greater variety of potential situations than the individual would ever likely face in actual experience. Computer simulation is also the basis for many popular games such as Sim-City as well as the entire array of sports simulation games.
Simulations are widely used to design a facility or develop operational policies. A computer simulation of a manufacturing plant or fast-food restaurant is designed primarily to evaluate different layouts and management policies. The goal is to reduce waiting time and increase the rate of production. In general, these simulations are discrete event stochastic simulations. That is, the computer simulation includes probability functions that are used to reflect randomness in the customer arrival pattern or the time to complete a task. The simulation model can be used to identify and avoid possible bottlenecks in a proposed system or to develop solutions for bottlenecks in an existing system.
The growth and refinement in the use of simulation has paralleled the growth of the computer. The first computer simulations were written in general purpose programming languages. Later, special simulation languages were developed. More recently, dramatic improvements were made in the dynamic visual representations of the system on the computer screen. The computer can represent a complex manufacturing line and the manager can watch as parts move from work station to work station and see where bottlenecks grow. The restaurant manager can watch customer lines grow and decline in a fast food restaurant. There are also simple simulation programs that are add-ons to spreadsheet software such as @Risk, an add-on to Excel.
The role of computer simulations in society as a planning and educational tool will continue to grow. Computers will become ever more powerful and ubiquitous and their graphics more realistic. We live in an uncertain world in which changes are happening faster and faster. Stochastic simulation is the best tool for carrying out low cost experiments while capturing more of the real-world situation than traditional analytic models are capable of doing. In addition, computer software and graphics will continue to improve, making it easier and easier to build simulations. However, an intuitive understanding of probability constructs is critical for interpreting the output of a stochastic simulation. Developing this understanding in anyone who might one day be a user or a builder of stochastic simulations is the largest remaining challenge.
Areas of Application
Probabilistic simulations have been used to characterize complex systems such as production lines, inventory systems, a police patrol force, a criminal court system, armies in battle, an emergency room, hospital populations, and so forth. Civil engineers who design transportation systems routinely build simulation models to study highway and street traffic flows. One of the earliest simulations was used to study tollbooths and traffic through tunnels leading into New York City. New airport runways with entrances and exits are designed using simulations to also identify and address safety concerns. Communications traffic flowing through a telecommunications network has also been simulated, as has the spread of pollution as it flows through groundwater. Simulation models are also used in public policy studies. Models have been developed to study the impact of prison sentencing guidelines on prison populations or to explore different strategies for controlling the spread of AIDS in Africa. Different policies regarding the collection and distribution of transplanted organs have also been studied using computer simulations.
The simplest class of simulations involves basic single and multiple server queueing systems. "Analytic" queueing models can be used to determine long-term averages, but they require restrictive modeling assumptions in order to obtain closed form solutions for these averages. In contrast, a stochastic simulation can calculate a variety of system performance statistics over a much broader range of assumptions. In addition, a computer simulation can describe the ebbs and flows in the system and not just long-run averages. These queueing simulations have been used to study banks, fast food restaurants, airport ticket counters or ships lined up waiting to enter the Suez Canal.