Strengths and Weaknesses of the Generic Methodology

Get Complete Project Material File(s) Now! »

CHAPTER 2 METHODOLOGY CONCEPTUALISATION

INTRODUCTION

Simply stated, the purpose of this chapter is to conceptualise the generic simulation modelling methodology. It is imperative to have a clear understanding of precisely what has to be achieved and how it should be attained, before any attempt is made to begin with the physical process by which the desired goal has to be achieved.
A simulation modelling method or methodology is usually developed with a specific class or type of system in mind. Therefore the first section identifies the characteristics of the class or type of system that is considered in this document. The key characteristics of these systems are the following: continuous processes, two types of discrete events (i.e. the services and failures) and complex interrelationships.
In the second section the implications of these characteristics on a simulation model are explored. Different techniques are considered and two possible candidates emerge, namely: a technique that
uses variables to represent the process flow in a simulation model and a technique that uses a fixed time interval to advance the simulation model in time. Equations are developed to determine the maximum possible throughput of the Synthetic Fuel plant, as a function of time, and also the number of modules that is switched on or off in each of the smaller plants to achieve that throughput, as a function of time. The determination of the maximum possible throughput is no arbitrary task because of the presence of feedback-loops, the division of the output of the Steam and Oxygen plants and the fact that the number of available modules in each of the smaller plants is a function of time.
The Entity-represent-module (ERM) method is described in the third section. The ERM method was originally developed as part of the Magister research and is used by both the original simulation modelling method and the generic simulation modelling methodology. It is an innovative method that determines the number of available modules in each of the smaller plants at any given moment in time. The concept of the ERM method is counter-intuitive because it uses entities to represent the modules rather than the cumbersome Servers or Work Centers that are usually used in simulation software packages. It leads to a compact simulation model size, total control over all the aspects of the services and accuracy. Each of the smaller plants is represented by three separate parts (i.e. the Availability, Service and Failure parts) that are ombined to form a high-level building block. Four types of smaller plants are represented in the ERM method by high-level building blocks (i.e. a smaller plant with a multiple service cycle and failures of the modules, a smaller plant with a service cycle and failures of the modules, a smaller plant with a service cycle of the modules and a smaller plant with failures of the modules). The advanced version of the ERM method (i.e. the one used by the generic methodology) is more compact and accurate than the original version (i.e. the one used by the original method).
The Fraction-comparison (FC) method is detailed in the fourth section. The FC method is the most important innovation of the generic simulation modelling methodology and can be considered as the “jewel in the crown” of the generic methodology. It is an elegant method that identifies the momentary “bottleneck” in a complex system at any given moment in time. The FC method is based on the fact that the actual output throughput values of the possible “bottleneck” points at any given moment in time are in fixed relations in terms of one another for all possible throughput options of the system that is under scrutiny. The fixed relations are expressed as the steady state actual output throughput values of the possible “bottleneck” points and are referred to as the FC method parameter set. The parameter set is unique for every specific system description of the system that is under scrutiny. The FC method provides a solution to one of the major problem areas of the generic methodology.
The determination of the governing parameters is detailed in the fifth section. The governing parameters are the gas-feedback-loop-fraction, steam-division-ratio, oxygen-division-ratio and the FC method parameter set. An iterative-loop technique is detailed that uses a FORTRAN software programme called PSCALC.FOR to determine the governing parameters of the Synthetic Fuel plant for the system description that is provided in Section 1.2. The sixth section considers techniques to identify the “bottleneck” smaller plants in the system that is under scrutiny. The original simulation modelling method uses the throughput utilisation values of the smaller plants to identify the “bottleneck” smaller plants. A distinction is made between primary and secondary “bottlenecks”. Two techniques are introduced to identify the primary “bottlenecks”. The first technique identifies the primary “bottlenecks” based on the time that the smaller plant is the “bottleneck” and the second technique identifies the primary “bottlenecks” based on the production that is lost due to the smaller plant. Flared throughput indicates the existence of a secondary “bottleneck”.
The last section conceptualises the structure of the generic simulation modelling methodology. The seven methods and techniques that are developed in the previous sections are integrated to form the generic methodology. The generic methodology is divided into two separate parts. The iterative-loop technique part determines the governing parameters before the start of a simulation run and the simulation model part uses the six other methods and techniques continuously during the simulation run. The simulation model itself is divided into a “virtual” part that deals with the continuous processes and the functioning of the simulation model and a “real” part that deals with the behaviour of the modules. The “virtual” part is represented in the simulation model by the logic engine high-level building block and the “real” part is represented by the four different highlevel building blocks of the ERM method. The five high-level building blocks can be used to construct simulation models of stochastic continuous systems. Simulation models that are developed with the generic methodology do not need a warm-up period and the advantages of this feature are also highlighted. The last section conceptualises the structure of the generic simulation modelling methodology.
The seven methods and techniques that are developed in the previous sections are integrated to form the generic methodology. The generic methodology is divided into two separate parts. The iterative-loop technique part determines the governing parameters before the start of a simulation run and the simulation model part uses the six other methods and techniques continuously during the simulation run. The simulation model itself is divided into a “virtual” part that deals with the continuous processes and the functioning of the simulation model and a “real” part that deals with the behaviour of the modules. The “virtual” part is represented in the simulation model by the logic engine high-level building block and the “real” part is represented by the four different highlevel building blocks of the ERM method. The five high-level building blocks can be used to construct simulation models of stochastic continuous systems. Simulation models that are developed with the generic methodology do not need a warm-up period and the advantages of this feature are also highlighted.
The complex interrelationships of the Synthetic Fuel plant are manifested in both the process flow and the process logic of the plant. The system description of the process flow indicates that there are several feedback-loops and that the output of both the Steam and Oxygen plants is divided (see Section 1.2 and Table A1). The process logic (rules of operation) of the plant indicates the complexity of the interrelationships between the smaller plants (see Section 1.2 and Appendix B).
The continuous nature of the process of the plant implies that all 147 modules are, in a way, intrinsically interlinked as far as the effect of the service or failure of a module is concerned. Any breakdown in the processing capacity at one point because of the service or failure of a module, does have an immediate effect on upstream and downstream operations.
The fact that these characteristics have to be accommodated in a simulation model that conforms to the design criteria that are stated in Section 1.5 poses the main problem of the generic simulation modelling methodology.
The complexity of the main problem, when viewed in its entirety, seems overwhelming. This challenge, however, can be approached in a meaningful way by segregating the main problem into appropriate smaller manageable units or subproblems and then solving each of them individually. The rest of this chapter identifies the subproblems through the process of logical deduction and then identifies and develops methods and techniques that solve the various problems that are posed by the subproblems.
Leedy (1993:71) postulates that the main research problem usually consists of two to six subproblems and advocates that subproblems should not be confused with pseudo-subproblems. He defines pseudo-subproblems as procedural indecisions and indicates, for example, that the problem to determine the correct sample size is a pseudo-subproblem, because there are various techniques available to determine sample sizes and it is only necessary to identify the correct one to use for each specific application.
In this chapter the terms “method” and “technique” are also used in accordance with the convention that is explained in Section 1.1 concerning the hierarchy of terminologies that are proposed by van Dyk (2001:2-4). According to the convention the term “method” is perceived to be indicative of a higher order terminology, while the term “technique” is perceived to be indicative of a lower order terminology. Hence, the term “method” is used to indicate a “tool” that is used to solve a more complex subproblem and the term “technique” is used to indicate a “tool” that is used to solve a less complex subproblem. The characteristics of the class or type of system that is considered in this document are identified in this section. The key characteristics of these systems are continuous processes, two types of discrete events (chronological and stochastic) and complex interrelationships.

READ  THE FULL RANGE MODEL OF LEADERSHIP IN PRACTICE

IMPLICATIONS OF THE CHARACTERISTICS

Section 1.6 indicates that some authors propose that continuous phenomena can be accommodated by using discrete-event modelling techniques. Harrell and Tumay (1999:35)propose two possible techniques that both use discrete-event modelling techniques to deal with continuous phenomena. The first technique suggests that continuously flowing “commodities” can be converted into discrete entities or “packages” for the purpose of a simulation model. For example, the maximum possible raw gas output throughput of the Gas Production plant is 1596000 nm3/h (40 modules with an output capacity of 39900 nm3/h each). This can be converted into 100 discrete raw gas “packages” of 15960 nm3 each for the purpose of a simulation model, if it is assumed that each raw gas “package” represents 1% of the maximum possible raw gas output throughput. If each raw gas “package” is delayed in a simulation model for 36 seconds (one hour divided by 100) as it leaves the Gas Production plant, then the simulation model simulates a raw gas output throughput of 1596000 nm3/h (100 “packages” of 15960 nm3 each leaves the Gas Production plant in one hour).
The following two major concerns immediately become apparent if the example that is mentioned in the previous paragraph is implemented in a simulation model:
a) The first concern is that the maximum possible accuracy with which the raw gas output throughput of the Gas Production plant can be determined, has been reduced to the size of a raw gas “package” per hour (i.e. 15960 nm3/h) or alternatively 1% of the maximum possible raw gas output throughput. The resolution of an answer that indicates the raw gas output throughput therefore cannot be any better than the size of a raw gas “package”.
b) The second concern is that 100 entities (raw gas “packages”) leave the Gas Production plant during one hour of simulated time. This implies that 100 events (delays of raw gas “packages”) occur at that point in the simulation model during one hour of simulated time. It also implies that over a simulated time period of one year a staggering 864000
The accuracy can obviously be improved by converting the maximum possible raw gas output throughput into more discrete “packages”. For instance, a conversion into 200 discrete “packages” will result in an accuracy resolution of ½% of the maximum possible raw gas output throughput. Paradoxically, this implies that the number of events at that point in the simulation model now doubles. This clearly represents a Scylla and Charybdis situation where the choice lies between “two dangers such that avoidance of one increases the risk from the other.” (The Oxford Compact English Dictionary, 1996:917; Macrone, 1999:20-21).
Kelton et al. (1998:353) also propose a variation on this technique and they indicate that it is usually preferred because it results in fewer entities in the simulation model. The variation on the technique uses a single entity that is looped through a time delay and increases a variable that represents the raw gas output throughput with a fixed amount (i.e. the discrete “package” size) each time a loop is completed. The problem is that this variation on the technique does not address the accuracy and huge number of events in the simulation model concerns that are detailed in the previous paragraphs.
The diminished accuracy and huge number of events that characterise this technique clearly violate some of the design criteria of the generic simulation modelling methodology that is stated in Section 1.5. The concession on accuracy obviously impacts negatively on the accurate modelling ability design criterion. The huge number of events in a simulation model that uses this technique affects the short simulation runtime criterion directly and the short development and maintenance times criteria indirectly, because longer simulation runtimes impact negatively on simulation model development, maintenance and use. The violation of the design criteria leads to an untenable situation. It emphatically disqualifies this technique as a contender to feature in the generic methodology.

INTRODUCTION 
CHAPTER 1: PROBLEM EXPOSITION 
Introduction
1.1 Background Information
1.2 System Description
1.3 Simulation Modelling as a Decision Support Tool
1.4 Shortcomings of the Original Method
1.5 Objective Statement
1.6 Importance of the Research
1.7 Limitations of the Generic Methodology
CHAPTER 2: METHODOLOGY CONCEPTUALISATION 
Introduction
2.1 System Characteristics
2.2 Implications of the Characteristics
2.3 The ERM Method
2.4 The FC Method
2.5 Determination of the Governing Parameters
2.6 Identification of the “Bottlenecks”
2.7 Structure of the Generic Methodology
CHAPTER 3: MODEL DEVELOPMENT 
Introduction
3.1 Investigation of the Simulation Software Packages
3.2 Simulation Model Breakdown
3.3 Simulation Model Construction
3.4 Determination of the Iteration Time Interval
3.5 Determination of the Sample Size
3.6 Simulation Model Verification and Validation
3.7 Simulation Model Enhancement
3.8 Comparison of the Simulation Models and the Simulation Software Packages
CHAPTER 4: MODEL APPLICATION 
Introduction
4.1 Background Information
4.2 Scenario I Results
4.3 Scenario II Results
4.4 Comparison of the Scenario I and II Results and the Conclusions
CHAPTER 5: SYNOPSIS 
Introduction
5.1 Motivation for the Research
5.2 Summary of the Research Process
5.3 Summary of the Generic Methodology
5.4 Comparison of the Original Method and the Generic Methodology
5.5 Strengths and Weaknesses of the Generic Methodology
5.6 Contribution to Knowledge
5.7 The Future Vision
5.8 Lessons Learnt and Reinforced
REFERENCES 
APPENDICES 
GET THE COMPLETE PROJECT

Related Posts