Page 86 - ExtendSim User Guide
P. 86

60 Simulation Concepts
Additional modeling terminology
• An output parameter is a value determined by the input parameters and the operation of the system—output parameters specify some measure of the systems performance or system dynamics.
Constant values and random variables
You enter parameter values in block dialogs to specify settings for a model. Constant values never change; random values are based on distributions and change each time they are used. Models that have no random input parameters are referred to as deterministic models. Models that are based on one or more variables that are random are said to be stochastic, as discussed below:
• Deterministicmodelscontainonlynon-random,fixedcomponents.Nomatterhowmany times a deterministic model is run, unless some parameter is changed there is no uncertainty and the output will be exactly the same. Thus the behavior of the model is “determined” once the inputs have been defined.
The advantage of a deterministic model is that only one run is necessary, since it produces an exact measurement of the model's performance. It is also helpful in when initially building a model since you can be assured that changes in results will be due to changes made to the model and not to randomness. The disadvantage is that these types of models can only accu- rately be used to model a few types of processes, since real-world systems typically contain some element of randomness.
• Addingrandomnesstooneormoreinputstoadeterministicmodelchangesittoastochas- tic or Monte Carlo model. Stochastic models are run repeatedly and then analyzed statisti- cally to determine a likely outcome. Notice that the occurrence of randomness does not mean that the behavior of a process is undefinable or even that it is unpredictable. Random variables vary statistically as defined by a distribution. This means that their range and pos- sibility of values is predictable.
While stochastic models can be applied to very complex systems, a disadvantage is that the output is itself random—the average of the simulation runs provides only an estimate of the model's true behavior.
ExtendSim provides several methods for including randomness in models. For instance, as you saw in the chapter “Building a Model”, the Random Number block (Value library) allows you to select a random distribution or enter a table of values which specifies an empirical distribu- tion of probabilities. For more detailed information about ExtendSim’s random number capa- bilities, see“Random numbers” on page 703.
Tutorial


































































































   84   85   86   87   88