MANAGING MANUFACTURING PROJECTS USING SIMULATION AND META HEURISTICS

MANAGING MANUFACTURING PROJECTS USING SIMULATION AND META HEURISTICS A THESIS Submitted by K.RAJA in fulfilment for the award of the degree of DOCTOR OF PHILOSOPHY FACULTY OF MECHANICAL ENGINEERING ANNA UNIVERSITY: CHENNAI - 600 025 March 2010

ANNA UNIVERSITY: CHENNAI 600 025 BONAFIDE CERTIFICATE Certified that this thesis titled MANAGING MANUFACTURING PROJECTS USING SIMULATION AND METAHEURISTICS is the bonafide work of Mr. K.RAJA who carried out the research under my supervision. Certified further that to the best of my knowledge the work reported herein does not form part of any other thesis or dissertation on the basis of which a degree or award was conferred on an earlier occasion of this or any other candidate. Place : Date : Dr. S. KUMANAN RESEARCH SUPERVISOR Professor and Dean Department of Production Engineering National Institute of Technology Tiruchirappalli - 620 015. 1

ABSTRACT Companies today find themselves in a highly competitive environment of rapidly changing operational requirements. Many are losing their competitive ability to balance operational processes against difficult business realities, including the need to manage increasing product complexities, shorter time to market, newer technologies and threats of global competition. Given these changes, businesses are trying to improve their workload processes with the use of effective project management techniques. The most popular methods for project planning and management are based on a network diagram such as Program Evaluation and Review Technique (PERT) and the Critical Path Method (CPM). These tools did not consider number of factors which are important for real-life project management. The availability of improved tools such as decision CPM (DCPM), Graphical Evaluation and Review Technique (GERT), and Venture Evaluation and Review Technique (VERT) are limited in application. The major hurdle with these tools is the assumption that there are infinite numbers of resources available for each activity of a project. Hence, the need for powerful graphical and analytical tools arises in project management. Simulation is a powerful technique for solving a wide variety of problems. Modeling and simulation can help in reduce projects cycle time; reduce cost and the amount of testing that must be done during development project. This thesis has discussed the literature in project management functions and their use in manufacturing management. It reports the development of PN based modeling and simulation software PETRI PM with suitable extensions to effectively deal with the Project management applications for planning, scheduling and control functions. Petri nets offer a versatile modeling framework for complex, distributed concurrent systems and have been used in a wide range of modeling applications. They are graphical in nature and are 2

backed up by a sound mathematical theory. The primary difference between Petri nets and modeling tools is the presence of tokens which are used to simulate dynamic, concurrent and asynchronous activities in a system. The Petri net model can be subjected to qualitative analysis to check system properties such as reachability, liveness, bound ness and conservativeness Information like maximum activity usage and probable project completion time are displayed graphically. Spread sheet flexibility and versatility makes Excel an appropriate platform for designing generic models that can be easily customized to the needs of a specific problem. A process model is developed, and using Microsoft Excel it is simulated. The model incorporates interrelated dynamic tasks with iterations, uncertainty of activity duration and complex resource scheduling that can represent realistic behavior of a complex design project. Using the SimQuick software, projects is modeled and the completion time is estimated. The estimation is done for unlimited and constrained resource environment. This thesis includes new methods for allocation of resources for manufacturing projects since the existing methods are limited. Proposed procedures provide optimal alternate solutions in terms of minimizing the project cost. This work confirms an alternative and efficient methodology for solving resource constrained project scheduling problems and opening the application of bacteria foraging algorithm, genetic algorithm to the optimization issues for scheduling of manufacturing projects. The existing resource leveling algorithms various resources are limited. Therefore, Memetic algorithms and particle swarm optimization algorithms are developed for this research and validated.application of resource leveling technique in manufacturing industries is exemplified. The proposed methods demonstrate that it is easy to handle real time changes of the project as well as create competent resource leveled schedules that flexibility to the user. 3

Managing multiple projects is a complex task. The researchers have proposed many tools and techniques for single project scheduling Mathematical programming and heuristic are limited in application.this work proposes the use of a heuristic and genetic, memetic algorithms for scheduling a multi project environment with an objective to minimize the makespan of the projects.the proposed method is validated with numerical examples and is found competent.. 4

ACKNOWLEDGEMENT I express my wholehearted and sincere gratitude to my research supervisor Dr. S. Kumanan, Professor, Department of Production Engineering, National Institute of Technology, Tiruchirappalli. I am profoundly indebted to him for his constant attention at every stage of the work. His deep commitment towards research, innovative thinking and invaluable comments formed the basis for the success of this endeavor. I thank my doctoral committee members Dr. T. Selvaraj, Professor of the Department of production Enginering and Dr.A.Noorul Haq Professor, Department Production Engineering, National Institute of Technology, Tiruchirappalli, for their valuable suggestions during the course of this research work. I owe special thanks to Dr. M. Chidambaram, Director, National Institute of Technology, Tiruchirappalli, for providing excellent research facilities in the campus. I acknowledge the help and support rendered by all the faculty members and non teaching staff in the Department of Production Engineering, National Institute of Technology, Tiruchirappalli. I am thankful to my parents Kandhasamy and Sushila, sisters Amutha, Manimegala brother in law Gunasekaran and Ramasamy wife Kavitha for their consistent support and to my sister son Sakthi for the patience shown by them during the research. K.RAJA 5

CHAPTER NO PAGE NO TABLE OF CONTENTS TITLE ABSTRACT LIST OF TABLES LIST OF FIGURES LIST OF SYMBOLS AND ABBREVATIONS 1 INTRODUCTION 1.1 BACKGROUNG 1.2 MOTIVATION OF RESEARCH 1.3 SCOPE OF RESEARCH 1.4 ORGANIZATION OF THESIS 2 LITERATURE REVIEW 2.1 INTRODUCTION 2.1.1 PROJECTMANAGEMENT 2.1.2 PROJECT MANAGEMENT TECHNIQUES 2.1.3. TRADITIONAL MODELS x ii xiii xix 2.1.4 INADEQUACIES OF CONVENTIONAL MODELS 2.2 PROJECT SCHEDULING 2.3 MODELING AND SIMULATION OF PROJECT 2.3.1 PETRI NETS 2.3.2APPLICATION OF PETRI NETS FOR SIMULATION 2.3.4.SPREADSHEET SIMULATION 2.4 RESOURCE ALLOCATION 2.5 RESOURCE LEVELING 2.6. MULTI-PROJECT SCHEDULING 2.6.1MULTI-PROJECT SCHEDULING WITH RESOURCE CONSTRAINTS 2.7 NON TRADITIONAL SEARCH TECHNIQUES 2.7.1 GENETIC ALGORITHM 2.7.2 PARTICLE SWARM OPTIMIZATION ALGORITHM 6

2.7.3 MEMETIC ALGORITHM 2.7.4 BACTERIA FORGING ALGORITHM 2.8. SUMMARY 3 PROJECT MANAGEMENT USING SIMULATION 3.1 INTRODUCTION 3.2 PROBLEM STATEMENT 3.3 SIMULATION OF PROJECTS WITH EXCEL 3.3.1. DYNAMIC CRASHING 3.3.2. NET PRESENT VALUE 3.5 SIMULATION OF PROJECTS USING PETRI NETS 3.5.1PROPOSED PETRI NET REPRESENTATION 3.5.2 PROJECT ANALYSIS 3.5.3REACHABILITY ANALYSIS 3.5.4 INVARIANTS ANALYSIS 3.5.5FEATURES OF THE PETRI-PM SOFTWARE 3.5.6 PROJECT ANALYSIS MODULE 3.5. 7 NUMERICAL EXAMPLE 3.5.8 SIMULATION 3.6 SUMMARY 4 RESOURCE ALLOCATION USING META HEURISTICS 4.1 INTRODUCTION 4.2 PROBLEM DESCRIPTION 4.3.GENETIC ALGORITHM APPROACH 4.4.MEMETIC ALGORITHM APPROACH 4.5.BACTERIAFROAGING ALGORITHM 4.6.SUMMARY 5 RESOURCE LEVELING USING METAHEURISTICS 5.1.INTRODUCTION 5.2 RESOURCE LEVELING PROBLEM 5.3 PROPOSED MEMTIC ALGORITHM 5.3.1 MEMETIC PARAMETERS 5.4.PARTICLE SWARM OPTIMIZATION (PSO) 7

5.5 SUMMARY 6 MULTI PROJECT SCHEDULING USING METAHEURISTICS 6.1 INTRODUCTION 6.2 MULTI PROJECT SCHEDULING PROBLEM 6.3 MULTI PROJECT SCHEDULING USING HEURISTIC 6.4 MULTI PROJECT SCHEDULING USING GENETIC ALGORITHM 6.5 MULTI PROJECT SCHEDULING USING MEMETIC C ALGORITHM 6.6SUMMARY 7 CONCLUSIONS 7.1 CONTRIBUTIONS OF THESIS 7.2 SCOPE OF FUTHURE WORK 8

LIST OF TABLES TABLE No. TITLE PAGE No. 2.1 Table 2.1 Benefits of Modeling and Simulation 3.1 Table Details of Case study for Excel simulation 3.2 Details of statistical measures for Excel simulation 3.3 Rework probability with learning 3.4 statistical measures with rework 3.5 statistical measures with increased capacity of resources 3.6 Dependency relationships for the manufacturing project network used for the example 3.7 7 minimum(a), most likely(b) and maximum(c) duration and crashing cost for the each activity of the project 3.8 comparisons between without crashing and optimal crashing 3.9 minimum (a) most likely (b) maximum(c) duration and crashing cost for each activity of the project 3.10 Results of the dynamic method implementation 3.11 Activities are crashed in each trial of the dynamic method 3.12 sales departments provide an estimate of the general and administrative costs. 3.13 Summarizes information of cash flow analysis 3.14 Discount factors over time for the development project 3.15 present an example iteration of the development project cash flows. 3.16 Product development project details under study 9

3.17 Details of model parameters for the manufacturing project under study for simulation using simquick 3.18 Project completion time under unlimited resources. 3.19 PD completion time under constrained resources. 3.20 Interpretations of places and transitions for car body painting Project 3.21 Input data for the project 3.22 PETRI-PM developed software output on Traditional computation 3.23 Simulation Result: Critical Indices after 100 Simulations 3.24 Input Data for the project for Petri net simulation 3.25 PETRI-PM output on traditional computation 3.26 Simulation result: Critical indices after 1000 simulation runs 3.27 Incidence matrix of Petri net for Car body painting project 3.28 Reachabilty Table of Petri Net for Car Painting Project 4.1 Early schedule of the project 4.2 Optimal schedule of the project by the proposed method 4.3 Optimal schedule (Alternate solution) of the project by the proposed method 4.4 Numerical example from literature 4.5 Early schedule of the project from literature 4.6 optimal schedule of the project by proposed method 4.7 optimal schedule (Alternate solution)of the project by proposed method 6.1 A Selection of Common Priority Dispatching Rules 6.2 Details of nodes and activities for all projects P1, P2, P.3 6.3 Details of Activity, Duration and Resources used for Project P1 6.4 Details of Activity, Duration and Resources used for Project P2 10

6.5 Details of Activity, Duration and Resources used for Project P3 6.6 The starting node, ending node and duration of all the projects 6.7 Available quantity of resource and delivery due date for each project 6.8 Details of Project Completion Dates for Different Priority Rules 6.9 Details of Nodes and Activities for Projects A, B 6.10 Details of activity, duration and resources used for project A 6.11 Details of activity, duration and resources used for project B 6.12 The starting node, ending node and duration of all the projects 6.13 Details of projects A, B completion date for different priority rules 6.14 Details of nodes and activities for all projects P1, P2, P3, P4 & P5. 6.15 Details of Activity, Duration and Resources used for Project P1 6.16 Details of Activity, Duration and Resources used for Project P2 6.17 Details of Activity, Duration and Resources used for Project P3 6.18 Details of Activity, Duration and Resources used for Project P4 6.19 Details of Activity, Duration and Resources used for Project P5 6.20 Starting node ending node and duration of the project 6.21 Details of completion time of all projects 6.22 Details of nodes and activities for all projects PI, PII, PIII & PIV 6.23 Details of activity, duration and resources used for project P1 6.24 Details of activity, duration and resources used for project P2 6.25 Details of activity, duration and resources used for project P3 6.26 Details of activity, duration and resources used for project P4 11

6.27 The starting node ending node and duration of the project 6.28 Details of completion time of all projects LIST OF SYMBOLS AND ABBREVIATIONS 12

1. INTRODUCTION Companies face stiff competition and need to manage increasing product complexities, shorter time to market, newer technologies. Given these changes, companies are trying to improve business processes with the use of effective project management techniques. The most popular methods for project planning and management are based on a network diagram such as Program Evaluation and Review Technique (PERT) and the Critical Path Method (CPM). These tools and its extensions did not consider number of factors which are important for real-life project management. Development of newer tools and techniques to manage manufacturing projects needs attention. 2. BACKGROUND OF RESEARCH The shift from the classical bureaucratic structures in the manufacturing industries to "lean, mean, flat" organizations is underway. Project Management (PM) approach is drawing increased attention in manufacturing management in recent years and it forms an essential decision making aides it suits the current trend and characteristics of manufacturing (Badiru 1996). Literature survey has shown increasing rate of embrace of network techniques in manufacturing settings. (Dereyck and Herrroelen (1997) Harhalakis (1989) Bowman (1995), Abdul-Nour et. al, (1998)). The existing methods for managing manufacturing projects have been successful in off-line planning and scheduling, it is difficult to dynamically monitor and control the progress of the project and to model resource constraints because information is loosely coupled. Researchers attempt to explore newer methods for managing projects. Resource leveling procedures are aimed to get the even usage of resources and to avoid high peak or very low resource requirements. Many researchers have focused on resource leveling procedures (Burgess and Kilebrew 1962, Sheng-Li et al 2006). Based on the problem characteristics, the code scheme, genetic operators and algorithm structure needs attention Resource allocation procedures are aimed to get the shortest project schedule by allocating the available

limited resources to project activites. Managing multiple projects is a complex task. It involves the integration of varieties of resources and schedules. The researchers have proposed many tools and techniques for single project scheduling. Mathematical programming and heuristics are limited in application. Most of the techniques developed in the past favored scheduling a single project or multiproject represented single project (Bowers et al (1996), Badiru (1996), Hsing-Pei kao et al (2006)) Recent literature shows multi-project management has become prevalent and its solution methodology needs attention. Simulation is a powerful technique for solving a wide variety of problems (Bank et al, 1998). The application of simulation is so general that it would be hard to point out disciplines or systems to which it has not been applied ( Deo, 2000; Slawomir and Peter 2008). The spreadsheet simulation was suggested for solving management science and operations research problems by Bodily (1986). Spread sheet provide a natural interface for model building are easy to use in terms of inputs. Petri nets are effective tool for modeling discrete event system. Their essential advantage is the possibility of mapping concurrency, synchronism and hierarchism of modeled system., solutions and report generations and allow users to perform what if analysis. Petri nets have been applied successfully in the areas of Performance evaluation, communication protocols, legal systems, and decision making models (Murata, 1989) Petri nets offer many advantages to project managers (Kumanan et al 2000, 2001). The Power of Petri nets in managing projects are to be explored. Literature review reveals there is increased use of stochastic optimization algorithms called metaheuristics intended to be the last resort before giving up and using random or brute-force search. Such algorithms are used for problems where you don't know how to find a good solution, but if shown a candidate solution, you can give it a grade. The algorithmic family includes genetic algorithms, hillclimbing, simulated annealing, ant colony optimization, particle swarm optimization, and so on. (Drezet and Tecquerd 2003). Valls et al. (2003) and Zhang et al. (2005) attempted the use of some Meta heuristics for solving resource allocation problem. The application of metaheuristics in managing projects needs attention.

3. SCOPE OF THIS RESEARCH WORK The scheme of this research work is shown in Figure 1. The work is focused as following o o o o o Identification of issues and challenges in managing manufacturing projects. Application and investigation on use of modeling, simulation and analysis of project Development of methodologies for resource leveling of projects using Meta heuristics Development of methodologies for resource allocation of projects using Meta heuristic optimization techniques algorithms Development of methodologies based on Meta heuristics for optimization of multi project planning.

LIST OF PUBLICATIONS International Journals 1. Kumanan S., Jagan jose, and Raja.K Multi project scheduling using heuristic and genetic algorithm(2005)international Journal of Advanced Manufacturing technology, Vol. 87, pp. 11-15.2005 2. Kumanan S.,and Raja.K Resource Levelling using Petrinet and Memetic approach.american Journal of Applied sciences 4(5)312-322,2007 3. Kumanan S.and Raja.K Modeling Simulation of Manufacturing Projects, American Journal of Applied sciences Vol5(12)1742-1749,2008 4. Kumanan S. and Raja.K (2008) Petri net and Particle Swarm optimization technique based Resource leveling International Journal of Manufacturing science and Production Vol 9 No 3-4 193-202,2008 5. Kumanan S.and Raja.K Multi Project scheduling using a Heuristic and Memetic algorithm. International Journal of Manufacturing science and Production ( accepted for publication) 6. Kumanan S.and Raja.K Resource allocation of manufacturing projects using Genetic, Memetic algorithm and Bacteria foraging algorithm.(working paper) International/National Conference 1 Kumanan.S., Raja.K, Beyond Traditional Manufacturing management techniques for planning modern manufacturing system, published International conference on Emerging Technologies(ICET 2003) 2 Kumanan.S., Raja.K Petri net- A modeling and simulation tool for Modern Manufacturing management(2004) National conference on Business Research,Coimbatore India pp 4.35-4.41. 3. Kumanan.S., Raja.K Project Network Techniques for Modern Manufacturing, Proceeding of National conference on Modeling analysis of production systems(maps 2004) 1

VITAE K.RAJA was born on 29 th December, 1976 at Tiruchengode, Tamil Nadu, India. He received his Bachelor s degree in Mechanical Engineering from Government college of Engineering, Salem affiliated to Madras University in 1998. He obtained his Master s degree in Industrial Engineering with a first class from Bharathiyar University in 2002. He has Three years engineering construction experience in Alpha Beeta engineering construction Pvt Ltd Mumbai. Seven years teaching and research experience in Government College of Engineering Salem His research area of interest includes Manufacturing Management Knowledge Based Modern Manufacturing system and new product development. 2