Chapter INTRODUCTION TO Z SIMULATION MODELLING | 4.0 INTRODUCTION Simulation is a very powerful and widely used scientific technique used in the analysis and study of different types of systems. Due to matters associated with complexity, stochastic relations and so on, not all real-world problems can be adequately represented in a model forms. Attempts to solve such problems using analytical models would require so many simplification assumptions that the solutions are likely to be inferior or inadequate for implementation. In most cases, the only credible form of modeling and analysis available to the decision maker is simulation. Computer simulation is the method used to reduce the risks associated with the creation of new systems or the cha in the evolution of existing ones into somethi w. Sometimes iherisks associated with the creation of new sytems rc ono thereby leaving little room for error. Simulation becomes the best bet in subjecting the proposed system to series of tests and analyses Prior to the full implementation of the project that propels the creation of such systems. This challenge is made more imminent by the complex factors that underly most operations that drive systems today. In recent years a lot has gone into the development ofsimulation applications that combine analytical techniques with graphically appealing features. In an ever evolving world where even the simplest of concepts are getting driven by information technology tools, precision is gradually becoming very critical giving te toa seemingly imminent need for a huge decrease in the margin Or error. 1 Scanned with CamScanner2 General Introduction to Modelling and Simulation 4.4 WHAT IS SIMULATION? Simulation is a process (and sometimes, programs) thar (imitate) the behaviour ofa real system. The simulation of g . isthe operation (carried out) on a model, which is a representa tion of that system. The model is open to manipulation which Woulg N Woulg ordinarily have been impossible, too expensive, or too a to perform on the system which it portrays. The operation on th model can be studied, and, from this, properties concerning the pehavior of the actual system can be inferred. Suffice it to say, that simulation uses a model to reach an inference which in tum provides understanding about the behaviour of the real worlg system under perspective. In view of the foregoing computer simulation can therefore be said to be the execution of the processes described above with the help of computer programming. Simulation is used before an existing system is altered or a new one is built. It is used to reduce the chances of failure to meet specifications, to eliminate unforeseen mishaps, to prevent under or over-utilization of resources, and to optimize system performance. A simulation model takes the form of a set of assumptions about the operation of a system, expressed as mathematical or logical relations between the objects of interest in the system. In contrast to the exact mathematical solutions available with most analytical models, the simulation process involves executing or running the model through time, usually on a computer, to generate representative samples of the measures of performance. In view of the foregoing, simulation may be seen as a sampling experiment on the real system, with the results being sample points. Mimic, 4.2 THE NATURE OF SIMULATION e Simulation underlies the operations of a facility or process usually through a computer. It entails the following: o What is being simulated in the system? o Studying the system would necessitate making assumptions/approximations, both logical and mathematical about how such a system works. 0 These assumptions form a model of the system. Scanned with CamScannerIntroducti mulation Mod ing § 3 o Ifthe model structure is ver (analytical) methods could b on the questions of interest, ¥ simple, then mathematical ¢ used to get exact information Most complex systems would also require complex models if they are to be valid. As a result, they must be studied through simulation (evaluating the model numerically and collecting data to estimate the model characteristics.) e Agood example isa manufacturing company seeking to extend its plant facility. o Build it and see if it works out? o Simulate current, expanded operations (many other issues along the way can also be investigated and evaluated cheaply and quickly) 4.3 WHAT IS A MODEL? A model is a simplification of the real world occurrence or phenomenon. It is a representation (usually miniature) of something; an example for imitation or emulation. A model can be analytical (Queuing Theory) or Simulated. A more detailed discourse on modelling will be done in the next chapter. THE REAL WORLD ‘THE SIMULATION STUDY | Simulation Model al) | Simulation Experiment Simulation — | Analyste fl 'GURE 1.1 Schematic istration of simulation modeling process Scanned with CamScanner4 General Introduction to Modelling and Simulation include in your model. if you choose the model 2, / right, you can exgress the real heart of what you want to understand about the real world, without any useless detail. This vail male your: ‘model much easier to analyse. We come up with models because we want guidelines about how a system should nerd ori wilt Je es designedfuit/o _perdee he ceremae third lines] fees] t0Bel and 2 analysis are only life can translate the ling Condusonsback f ay | | int reabwonid | predictions FIGURE 1.2 Graphical illustration of simulation modeling process Scanned with CamScannerIntroduction to Simulation Modellin make predation geablite experiments AZ Some modeh Different models are useful for different purposes, and different purpares require different models, 4.4 SOME TERMINOLOGIES © System: A group of objects that are joined together in some regular interaction or interdependence toward the accomplishment of some purpose. In practice, however, this definition generally tends to be more flexible. The exact description of the system usually depends on the objective of the simulation study. For example, what may be a system for a particular study may be a subset of the overall system of another. © Entity: this is an object of interest in the system e.g., customers at a bank © Attribute: this is a property of an entity e.g., account balance © Activity: this represents a time period of specified length. It can also imply a collection of operations that transform the state of an entity, e.g., making bank deposits © Event: an action that brings about a change in the system State.E.g., arrival; beginning of a new execution; departure ® State Variables: they define the state of the system. Simulation can be restarted from state variables. E.g., length of the job queue. Scanned with CamScanner6 — General Introduction to Modelling and Simulation this is the sequence of events ordered based on the function of time. NO’ to three alternative ways © ont, process, and activity) give rise the three concepts (eve : ! t f building discrete simulation models, TABLE 1.1 Example of a system and its components EVENTS | STATE VARIABLES gystem | ENTITIES arrriuTes | ACTIVITIES Banking | Customers ‘Account Making Arrival; | Number of busy is balance | deposits Departure tellers; Number of customers waiting 2| Airline | Passengers: Ticket Booking ‘Arrival; | Number of pilot; air number; ID flights:flying} departure | passengers hostesses | Number the plane wailing to board 3 4] ontinuously (continuot NOTE: State Variables may change ¢ only at a discrete set system) over time or they may change points (discrete sys.) in time. 4.5 QUEUING SYSTEMS tities request service from a shar! These are systems where ¢ resource(s) and have to wait for service until resource is availab Some examples include: o Customers waiting in a bank Packets waiting in a network router/switch Travelers waiting in an airport for flight registration oO ° o Tasks requesting execution in a processor (semaphores) ° database Bank transactions requesting access to a Scanned with CamScanner_Introduction to Simulation Modelling 7 Customer FIGURE 1.4 Simple queuing model e Queuing models are used to: » Describe the behavior of queuing systems > Evaluate system performance e . Queuing System has: Queue (Buffer): with a finite or infinite size < Server: with a given processing speed » Customers: Entities requesting the shared resource (i.e. server). > Events: Arrival/Departure (with given rates) Customers ofthe sistem © @) Servers YY DEPARTING ARRIVING pr aueue customers ce FIGURE 1.5 Airline ticketing queue and model representations Scanned with CamScannerral Introduction to Modelling ad Simulation current Sen output eterenceciode / FIGURE 1.6 Electric bulb and model representations 4.6 COMPUTER SIMULATION mulation is a computer program that attempts to stract model of a particular system as well as describe hysical) system and its evolution in time. ed by state variables. The em is describ d, modifies the state variables Computer Si simulate an ab: the behavior of a real (P The behavior of the sys simulation program on the other han to emulate the evolution. Fundamentally, computer simulation can branch of applied mathematics. The emergence ol high computing power and advanced programming techniques as well as the seeming difficulties associated with accurately describing complex real world systems using analytical or mathematical models heralded the birth of computer simulation. Computer simulation can therefore be defined as the use of computers to mimic the operations associated with real world processes based on adequately developed assumptions which may take the form of logical, statistical or mathematical relationships developed and structured into a model. Simulation can be a complex and time consumi ivi the modeling stage through to the output analysis Levene! necessitates the involvement of some measure of expertise and high level decision making throughout the entire process. In order be classified as a f computers with Scanned with CamScannerTatroduction to Simulation Modelling — 9 toget it right the following possible drawbacks must be prevented Unclear objectives Using simulation when an analytic solution is appropriate ii, invalid model. simulation model too complex or too simple. ssumptions. Erroneous = Undocumented assumption: ‘This is extremely important and it is strongly suggested that assumptions made at & stage of the simulation modeling and analysis exercise be documented thoroughly. \ii. Using the wrong input probability distribution. vii. Replacing a distribution (stochastic) by its mean (deterministic). ix, Using the wrong performance measure. x. Bugs in the simulation program. xi. Initial bias in output data. xi. Making one simulation run for a configuration. 4.7 WHY WE NEED SIMULATION Minimize production risk ¢ It may be too difficult, risky, or expensive to observe a real, operational system ¢ Parts of the system may not be observable (e.g., internals of a silicon chip or biological system) Make design more efficient * Analyze systems (performance, behavior) before they are built ¢ Reduce the number of design errors ¢ Optimize design to improve the behavior ¢ Analyze operational systems ¢ Create virtual environments for W_______training, entertainment __| Scanned with CamScanner4.8 WHY MODELLING AND SIMULATION? ] WHEN TO USE SIMULATIONS; WHY USE MODELS? 4.9 GENERAL BENEFITS OF COMPUTER SIMULATION 1, Allows experimentation without disruption to existing sy carrying out tests on some already existing systems can pro to be impracticable. However, simulation allows for development of model that depicts such a system and releve variables are altered via simulation to check for the effics associated with the proposed change. When the effects of the alterations and manipulations on the system are examined 2 the result is favourable, a decision can be reached to impleme the changes in the real world system. 2. Concept can be tested before implementation: comput simulation allows ideas and concepts to be tested before th are fully implemented in the real system. The testing has“ tendency to bring to the fore design errors that would h# gone unnoticed and cause danger. Furthermore, where s* errors go undetected, it may prove difficult and expensiv® fix after discovery upon implementation. 3. Detection of unforeseen problems or potential threats/faili P when a system is simulated before installation and found! ss Scanned with CamScannerviable in principle, best practice demands adjusted to capture finer details. Further adjusted model may reveal unfore: future problems or bugs that the syst such potential threats before installa rework costs can be proactivel: - ; 1 vantage associated with the simulation process is that kno ledge about the system under review is enhanced in a variety of ways. Just designed model of a real world system may be vague due toa variety of reasons. However, by the time the mod: brought under review, insights are gained as to the workings of the system through the manipulations that simulation offers. By the time all that needs to be done gets done, the little details come into perspective and alignment with the bigger picture leading scientists into gaining knowledge that was hitherto unknown. The gained knowledge consequently affords the designers the capacity to speed up the process significantly and allow data to flow from individual experts to the simulation activity. . Speed in carrying out analyses: upon the full development of the system model, it can be subjected to simulations that can be run at great speeds that are greater than would be possible in the real world. As a matter of fact, multiple experiments can be run on that one model that is altered in a variety of way through the manipulation of relevant variables. A finished model subjected to simulation, churns out results in less than no time that would normally take a much longer time to discover in the real world system. . Demands system definition: it is important for all the aspects of @ system to be known so as to evolve a valid and workable model of a designated system. If elements of the system are Not properly defined, the model will be far from accurate, Scanned with CamScanneral Introduction (o Modelling and Simulation thereby defeating the purpose for modelling in the first pla Consequently, the goal of a reliable simulation forces analy to properly define all the elements crucial to the working. the system, If some facts cannot be defined in definite tenn, ‘I is not out of place to make a safe guess however, ifthe gues) found to be a liability, it is suggested that a more thorou: investigation be carried out to prevent disaster. gh Enhances creativity: this is a very important factor in that affords scientists and engineers the ability to try out innovations without much to risk other than the time investej in carrying out the investigation and experimentation. This i, because simulation affords them the privilege of proposing ney, ways of solving problems using unconventional and unknow, methods. Where such methods are able to scale through th: testing subjected through simulation and if it offers a cheape and reliable solution, then such new methods can be adopteg as against traditional but expensive methods. Build Consensus. Using simulation to present design change; creates an objective opinion. You avoid having inferences mad: when you approve or disapprove of designs because you simply ations that provided the mos select the designs and modific desirable results, whether it be increasing production o reducing the waiting time for service. In addition, it is much easier to accept reliable simulation results, which have beer modeled, tested, validated, and visually represented, instead of one person's opinion of the results that will occur from é proposed design. : 9. Team Training. Simulation training when designed for that purpose. the team provides decision inputs to the simulati progresses. The team, and individual members can learn by their mistakes, and learn to operate much less expensive and less disruptive tha: learning. 10. Specify Requirements. Simulation can be used to speci requirements for a system design. For example, ; 7. models can provide excellen! Used in this mannet ion model asi! of the teat better. This# n on-the-j2! Scanned with CamScannerspecification lor a particular ly] c HAW type of machine in system to achieve a desired 20: simulating different capabilitie: requirements can be established, a complex ail may be unknown, By 8 for the machine, the 4.40 GENERAL LIMITATIONS OF COMPUTER SIMULATION j. Expensive: often times the creation of be an expensive method of analysis. ‘Thi scale modelling projects require hardware, software, analysis an 4 Computer model can : is because most large Major investment in training, id development time. . Time consuming: because modelling does not always produce quick answers couple with the fact that data collection, model development, analysis and report Seneration most times require considerable amount of time. However, the simulation process can be sped up using one of two methods namely: reduction of details (scaling) and the use of generic code libraries. Note that whereas scaling can lead to faster Tesults, it may lead to compromising the accuracy of the expected results. In a situation where similar simulations will be Tun, code library or generic simulation can be created. The reusability of this resource eliminates time wasting asa result of having to recreate what already exists in a perfectly usable and acceptable form. This is a motivation for simulation as a method. np 3. Model Building Requires Special Training: It is an art that is learned over time and through experience. Furthermore, if two models of the same system are constructed by two competent individuals, they may have similarities, but it is highly unlikely that they will be the same, 4. Provides only approximate answers: because discrete event simulation depends on the use of random number generators to provide model input, uncertainty is already associated with the output. To make the results more reliable certain statistical tools become relevant for the interpretation of the output. Hence all outputs are tough estimates of the system's true behaviour, Scanned with CamScannerto Modelling ana simulation 14 General Introdu dation i cess of 5. Difficult to validate: model validation is det Pe se fay ing that a computer model accurately rer ereumbersome in studied. Carrying out this process can e 7 ne if thy system doesn't exist, giving rise to reliance on Mi ¢ opinion 0 vuiti of insuring that the mode intuition asa way the rar Se eneward system will. When this és the case, | runs in , : is safer to adopt 2 conservative point of a ie a oe that the system's predict it ed operation is sah expectations for the actual operation when carried out on th real system. VALUATION : é quantifying the servic, (ofa system)means e rried out through experimental 4.11 PERFORMANCE Performance evaluation (of a delivered by a system. This is ca i imulation methocs- analytical or tri isa measurable quantity that precisel, captures what we want to measure (response time, throughput delay, etc.). The throughput metric accounts for how muci customer the system is able to serve per unit ume. The delay metri: accounts for how much time is spent in wait (in the queue) for: 1, the response time is the summatior particular service. Howeve! ; : of the time spent in the queue and the time spent getting server (service time). © For example, in computer system: > The response time of a processor to ex > The execution time of two programs in a multi-proce: machine. © Innetwork systems, we might evaluate: > The (maximum/average) delay experienced by a voici packet to reach the destination > The throughput of the network > The required bandwidth to avoid congestion s, we might evaluate: ecute a given task. SSO. Scanned with CamScannerIntroduction to Simulation Modelling 15 a coat Recoxtor + [Broaks & Exorcises Organisation HR Database Rocordor + Breaks & Exercises Recorder + Breaks & Exercises FIGURE 1.7 Graphical illustration of performance metric Queuing System SoH Queue Server Queuing Time Service Time FIGURE 1.8 The queuing times Scanned with CamScanner, and Simulation | Iatroduction to Modelli 1.12 SYSTEM PERFORMANCE The performance of a system is heavily affected by the Workloag ‘The workload is characterized by the quantity (number of eNtitie Waiting to be served on the queue ¢.g. number of cars on a road and also nature (the features of the entity waiting for service eg type of cars). TOP LEFT: illustration of heavy traffic situation TOP RIGHT: llustration of an average trattic BOTTOM LEFF: illustration of a free flowing traffic FIGURE 1.9 Road traffic © In the context of Web Servers, system inputs are http request (GET or POST Tequests). The workload characterizes: ~ The intensity of the re received by the web se: the performance. quests: how many requests alt ‘Tver. High intensities deteriorat’ > The nature of the Tequests: the rei Tequest or a request that requir database. The performance will request types. (see fig. 1.7) ‘quest can be simple GE! the access of a remott be different for differet 1.13 BENCHMARK 7 @ Benchmarks: used to generate load that is intended to mimic‘ typical user behaviour. Scanned with CamScanneree Introduction to Simulation Modelling 17 wikipedia definition: computing, a benchmark is the Incomputing, ab nehaark is the act of running a computer program, aset of programs, oF other operations the relative performance of running a number of standard tes , In order to an object, normally by ind trials against it ° penchmarking is usually associated with assessing performance istics of computer hardware: xample: the floating point operation performance ofa CPU, > Software benchmark: management systems. run against compi or databa 4.44 EXERCISES TABLE 1.2 Class Activity SYSTEM | ENTITIES ATTRIBUTES | ACTIVITIES | EVENTS | STATE VARIABLES Banking | Customers | Account Making Arrival; | Number of busy balance deposits | Departure} tellers; Number of customers waiting i, Break into groups of 4 students. Using the table above as a template, think of other systems that exist in our everyday world and identify the various components that describe their entities, attributes activities, events and state variables. ii, Explain the concept of performance metrics and point out its relationship with benchmarking. iii, Delay or response time is a combination of queuing time and service time. Show this relationship with the aid of a queuing system diagram. Scanned with CamScanner