1 An Introduction into Modelling and Simulation Prof. Dr.-Ing. Andreas Rinkel af&e andreas.rinkel@hsr.ch Tel.: +41 (0) 55 2224928 Mobil: +41 (0) 79 3320562 Goal After the whole lecture you: will have an fundamental understanding what simulation is and what it is used for knows the Simulation project phases and deliverables Gather requirements Build Model Deliver Results the simulation process some basics in statistics have trained your knowledge in a series of small assignments with SIMIO in different areas will have some rules on hand when simulation is an appropriate approach for the problem solving process can apply the learned on small real projects will have an academic SIMIO simulator for your ongoing self studies 2 Literature Rapid Modelling Solutions: Introduction to Simulation and Simio, SIMIO LLC, C. Dennis Pegden. David T. Sturrock Simio & Simulation, Mc Graw Hill 2011, W. David Kelton, Jeffery S. Smith, David T. Sturrock Simulation Modeling and Analysis with SIMIO: A Workbook, SIMIO LLC, Jeffrey A. Joines, Stephen D. Roberts, 2013, third Edition Process Analysis and Improvement: Tools and Techniques, McGraw-Hill, Irwin, 2005 Applied Simulation Modeling, McGraw-Hill, 2003, Andrew Seila, Vlatko Ceric, Pandu Tadikamalla First (and Second) Steps in Statistics, SAGE 2010, D.W. Wright and K. London 3
2 Structure Theoretical Part Introduction and Problem Description What are Reasons to Start a Simulation Project? What at least is Simulation and what are the Simulation Phases? What is the Scope of (Simulation) Work? Practical Part A series of step by step labs to get familiar with the simulation tool Simio to learn different aspects of modelling and simulation each lab ends up with an assignment of a small case study Second lecture in summer semester 2015 will come up with a bigger Case study in the area of production line, supply chain, resource planning 4 The Problem Mainly described in a Word Model Given Word Model: We plan an office that dispenses automotive license plates. In our first approach we want to divide its customers into categories to level the office workload. Customers arrive and enter one of three lines based on their residence location. Each customer type is assigned a single, separate clerk to process the application forms and accept payment, with a separate queue for each. After completion of this step, all customers are sent to a single, second clerk who checks the forms and issues the plates (this clerk serves all three customer types, who merge into a single first-come, first-served queue for this clerk). So, is this a good architecture? By the way what is probably the expected average and maximum time of a customer in the system for all customers? how many servants do we really need and what is their utilization? How much space do we need in front of the clerks Did we oversee something? Phuuuu, we should ask an consultant! Lets start an Simulation Project to answer all these questions!!!??? 5 What are Reasons to Start a Simulation Project I Study Highly Variable Processes, Variability is inherent in all systems disrupts systems without variability, performance would often be easy to predict accurately capturing system variability will result in better analysis and decision making simulation has the ability to handle variability Understand Process Interdependencies Processes should not be analyzed in a silo All systems or Subsystems have interdependencies Users can change one or more variables in the model and clearly understand how the entire system is impacted simulation allows analysts to study system interactions 6
3 What are Reasons to Start a Simulation Project II Identify Bottlenecks: key output statistics of a simulation study: The average time entities spend waiting for resources the average numbers of entities waiting for a resource These statistics are automatically calculated by the simulation software Queuing theory is at the core of simulation Experiments (simulation is an experimental approach!) with the model to determine which changes will reduce the bottleneck in the real world Analyse the temporal Behaviour Simulation allows the look at a system dynamic over time Relying on average values for planning can be misleading Use simulation to plan staff schedules and resource availability by time of day, week or any planning horizon 7 What are Reasons to Start a Simulation Project III Animation Advantage: builds confidence Seeing the system dynamically change over time A valid model, backed by real data and compelling animation, will help leaders to make decisions Disadvantage: expensive (Time, Costs, intellectual power) Find the right balance between animation and modelling! Process Complexity Simulation projects can handle Complexity Simulation models are dynamic and based on real-world variability over time Making decisions regarding complex systems using a simple spreadsheet and average values can be dangerous With highly complex systems, simulation will provide more detailed information and understanding to allow for better decision making 8 Simulation is: a very broad term methods and applications to imitate or mimic real systems, usually via computer Applies in many fields and industries e.g. production, business processes,.. a very popular and powerful method At least, simulation is the process of designing a Model of a concrete System and conducting Experiments with this model in order to understand the behaviour of a concrete system and/or to evaluate various strategies for the operation of the system (Shannon, 1975) Simulation is an experimental method! 9
4 System A System is a facility or process with defined boarders/interfaces to its environment, actual or planned, or at least a combination of all Examples abound Manufacturing facility, Bank operation, Airport operations (passengers, security, planes, crews, baggage), Transportation/logistics/distribution operation. Business process (insurance office) A real life problem is usually a so called complex or cybernetic system The science of Cybernetics is founded of the work from Nobert Wiener in the 1940 s Cybernetics defines the science of the communication and control of independent (complex) systems (human, technical or abstract processes) In this case, the different areas like machines, human beings, organisations and their processes could be seen as one big organism consist of an arbitrary number of nested and depended subsystems 10 Model A Model is a: set of assumptions/approximations about how the system works Study the model instead of the real system usually much easier, faster, cheaper, safer Can try wide-ranging ideas with the model Make your mistakes on the computer where they don t count, rather than for real where they do count Often, just building the model is instructive regardless of results (understand the system and its behavior) Model validity (any kind of model not just simulation) Care in building to mimic reality faithfully Level of detail (as precise as necessary not as precise as possible!!! What are your questions on the system) Get same conclusions from the model as you would from system 11 Experiment An Experiment is: a test or investigation or procedure carried out under controlled conditions to determine the validity of a hypothesis or make a discover or as basis for a decision the act of conducting such an investigation or test or procedure an attempt at something new or different Due to Variability results of an experiment will differ a little bit to each other a series of experiments must be done to use some analytical statistic methods to make statements about the output data Mean Variance Coincidence Intervals the selection of appropriate input data is a critical task and needs a carefully analysis of the system parameters and as well some knowledge in descriptive statistics 12
5 Types of Models Physical (iconic) models Tabletop material-handling models Mock-ups of fast-food restaurants Flight simulators Logical (mathematical) models Approximations and assumptions about a system s operation Often represented via computer program in appropriate software Exercise the program to try things, get results, learn about model behavior 13 Simulation: Classification I Static vs. Dynamic Does time have a role in the model? Continuous-change vs. Discrete-change Can the state change continuously or only at discrete points in time? Deterministic vs. Stochastic (to describe variability) Is everything for sure or is there uncertainty? Most real operational models: Dynamic, Discrete-change, Stochastic 14 Simulation: Classification II Dynamic Simulation continuous Time controlled discrete simulation can State changes occur continuously be used to mimic over the time line. Model continuous behaviour description by differential equations or real physical models like the wing of a plane to study streaming behaviour. time controlled There are fixed periods of time Δt defined. All events which occur in Δt will be evaluated simultaneously at the end of each time slot. So the simulation or numerical evaluation of differential equations is also possibly, e.g. Euler Algorithm. State changes occur at discrete points of time. State changes are instant event driven Each event has its own point of time. An event may produce a state change as well as it may lead to following events Process/Agent oriented A process in this case is just an element to structure the description of the behaviour in a more natural way. So it just simplifies the process of modelling. 15
6 Some Principles to Improve Processes I Variation degrades performance Increasing utilization increases WIP/Waiting Times A CONWIP strategy has less WIP for the same throughput A single queue decreases WIP Shortest Processing Time first decreases WIP Moving variability downstream decreases WIP Moving fast servers downstream decreases WIP Buffer space increases throughput and decreases WIP Buffering the Bottleneck increases throughput and decreases WIP Feeding the bottleneck increases throughput and decreases WIP Minimizing changeovers increases throughput and reduces WIP Task splitting may improve performance Worker flexibility improves performance Buffer flexibility improves performance 16 Some Principles to Improve Processes II Make a model of the actual or proposed System under study Use a modern Simulation-Tool like Simio to capture the influence of randomness on the dynamic behaviour of your System Use the model to vary buffer sizes, introduce new entity types, modify server characteristics, etc., and see the impact of your proposed changes. The previous principles provide ideas for changes to consider in your system Come up with a simulation result as a basis for better management decisions 17