Simulation - An Introduction Simulation:- The technique of imitating the behaviour of some situation or system (economic, military, mechanical, etc.) by means of an analogous situation, model or apparatus, either to gain information more conveniently or to train personnel. Oxford English Dictionary
Applications of Simulation design stage - design of new facilities, assess alternatives –assess capacity of system –assess capacity of materials handling system –identify bottlenecks operation stage –evaluate alternative scheduling or sequencing rules
Use of Simulation increased use due to increasing computer power user-friendly packages (e.g. WITNESS, Simul8) –animated graphics –totally interactive and interpretative
Event Based Simulation
Discrete Event Simulation - Example
Some Definitions ENTITIES - elements undergoing simulation –machines, parts, conveyors, vehicles, tracks, labour, pipes ATTRIBUTES - information on the entity –route of part, colour, priority, due date, batch size EVENTS - points in time when changes take place –end of machining on lathe_1
Some Definitions QUEUES - when entities are places when not engaged in activities –logical or physical - jobs waiting for machines or machines waiting for jobs ACTIVITIES - things entities do, or have done on them –generally involves more than one entity, e.g. job_6 & machine_9
WITNESS - Available Entities Parts - variable or fixed attributes Buffers - normal or minimum time Machines - single, assembly, production, batch Conveyors - fixed or queuing Vehicles Tracks Labour - operate, repair, setup
The Simulation Process Data collection Model formulation Analysis
Data Collection Use workstudy, measurement etc. Fixed length activities Random variables –calculate probability distribution –sample sizes
Model Formulation Level of detail –Excessive detail can make simulation times too long –Excessive detail can mean accurate data collection is impossible –Not enough detail can mean some factors which significantly affect results are not modelled Random number generation –Monte-Carlo simulation –Psuedo-random numbers Starting conditions –Estimate ‘normal’ starting conditions –‘Warm-up’system to a normal starting condition before measurement
Analysis Compute measurements required from model –Utilisation –Average time in system –Capacity –etc Statistical design of simulation experiments –need to run model for sufficient time to ensure results from ‘stable’ condition are observed –if a stochastic model is used need to ensure sufficient runs are done
Why Simulate ? No mathematical formulation of the problem exists Simulation the only possible or economic means of experimentation and observation Allows use of random elements and rerunning of model Control over time-scale. –Lengthy periods can be simulated quickly and vice-versa Allows comparison with analytic solutions
Problems in Simulation Validity of results –Level of detail –Initial conditions and steady states –Statistical validity of Monte Carlo simulation Use for comparison of solutions rather than absolute measures
Reasons not to Simulate Can be time-consuming to collect data and to build and run the model It is imprecise, with no measure of how imprecise Each model is unique. Difficult to reuse. Simulations tend to grow large to include many details