Material Handling Information - How to Get Started - How do I Simulate?
How to Get Started
How do I Simulate?
A sound simulation study follows the protocol outlined below:
Problem Statement: The problem statement is particularly geared towards setting the objectives of the study. This also includes denoting the performance measures of interest, and initially designating the system variables that affect these response measures.
Model Formulation: Here, the analyst expands the idea of the system (independent) variables by drafting a representative logical model reflecting the internal interactions of these variables. In other words, this is where the simulation model is created. Simulation software is geared for this task.
Data Collection: This step is usually concurrent with model formulation. Translating the real system into a simulation model requires an adequate knowledge of the internal workings of the system at hand. Data is usually collected in a time-study fashion to characterize non-deterministic processes. Key aspects, such as arrival rates, processing delays, and order sizes, may vary over time. Data collection attempts to translate such variations into a stochastic or probability distribution function to be used in the simulation model.
Model Verification: In plain terms, this means model debugging. Verification ensures the computer program representing the simulation model runs correctly.
Model Validation: Validation answers the question whether the simulation model is an accurate representation of the real system. This is usually achieved through comparing key performance statistics in both the simulation model’s output and the actual system’s known performance.
Experimental Design: The simulation model represents the interactions between several key system (independent) variables. Each one of these variables has different values (levels), and the combination of which makes up different design options (with different effects on the response measures). Experimental design uses statistical theory to: (1) structure this experimentation with system variables, and (2) draw inferences from their effects on the performance measures. Other related items include simulation run length and number of independent runs.
Results Analysis: The final step involves detecting system weaknesses, such as bottlenecks, and recommending better alternatives.
