A Simulation based Optimization Study of Dynamic Crashing

A Simulation based Optimization Study of Dynamic Crashing in Collaborative Projects Masters Thesis Proposal by Krishna Neelakanta University of Colorado, Colorado Springs Fall 2009 Page 1

Introduction � Time-Cost Tradeoff in Project Management � Crashing a Project Schedule ◦ Deployment of additional Resources to activities or other means to minimize cost � Balance Crashing costs and Penalty Costs � Aim to minimize total cost while delivering the project on time � How can we model collaborative projects as a project network � Introduce dynamic re-evaluation of optimal crashing configuration for the project network � Evaluate the model via simulation Krishna Neelakanta Thesis Proposal Page 2

A Typical Project Network A Project Network with task dependencies Krishna Neelakanta Thesis Proposal Page 3

Project Time-Cost Tradeoff Crashing + Penalty Cost VS Amount of Crashing 60 50 40 30 Crashing Cost + Penalty Cost 20 10 0 0 10/08/2004 5 10 15 Krishna Neelakanta Thesis Proposal Page 4

Static vs Dynamic Crashing � Static Crashing ◦ Evaluate critical path ◦ Run an LP optimizer on the critical path to identify tasks that need to be crashed ◦ LP optimizer tries to minimize Project Penalty + Crashing cost � Dynamic Crashing ◦ The initial step is similar to Static Crashing ◦ Re-evaluate project network at different points in project network, by accounting for task completed or underway (sunken costs at that point in time) ◦ Optimize the reminder of the project via LP optimizer Krishna Neelakanta Thesis Proposal Page 5

Related Work � Critical Path Method (CPM) and Project Evaluation Review Technique (PERT) have been around since 1950 s ◦ CPM – Repeatable activities from previous experience. Low variance ◦ PERT – More focus on research type projects and activity times are uncertain. � � � PERT/CPM is now one area – Hillier and Lieberman [4] Simulation is a tool used commonly for Project Mgmt. Williams [5] Activity times are not certain. CPM falls short here Project managers want to determine the optimal crashing configuration Industrial Strength COMPASS (ISC) Hong and Nelson 2006 [6] – convergent, discrete optimization via simulation Dynamic Crashing approach in simulation based optimization outlined in Kuhl et. al (2008) [7]. Basis for extending the work. Krishna Neelakanta Thesis Proposal Page 6

Goal of the Thesis � � � Design a model/template Project Network for Collaborative efforts. Address Bottlenecks, context based handoff’s, reviews, addition of tasks in mid project in the model Develop a simulation-based dynamic optimization method for the generalized stochastic time-cost tradeoff decision problem ◦ Should include the initial evaluation to determine optimal configuration ◦ Should include dynamic re-evaluation to determine optimal configuration as the project progresses � Implement the simulation-based optimization tool in a Project Management tool like Microsoft Project Krishna Neelakanta Thesis Proposal Page 7

Thesis Approach � � Design a model/template Project Network for Collaborative efforts. The model should address bottlenecks, context based handoff’s, review processes, addition of tasks in mid project Build a simulator to test the model Hook up the simulator to a proven industrial strength Linear Programming Optimizer to evaluate crashing points, optimal crashing configuration. ◦ Use the Industrial Strength COMPASS – Nelson et. al [8] � � Perform dynamic re-evaluation and crashing of the project network Questions that we would like to answer. ◦ Distribution of project completion times, Project Costs and Savings ◦ How does a Collaborative Project compare with a Traditional Project (network on Slide 3) Integrate the proposed model into an Common Project Management Tool like MS Project How does the model perform for larger project networks. Krishna Neelakanta Thesis Proposal Page 8

Assumptions & Considerations � Initial Parameters ◦ Initial Project Network ◦ Optimistic, Pessimistic and mean times for each activity and dependencies ◦ Simulator uses above and a Beta Distribution to generate activity times during each run. ◦ Project Cost is a linear function of completion time, penalty costs and crashing costs Highly dynamic nature of the Collaborative Process and reduction in centralized planning � Autonomy of Collaborators and its reflection in the model � Simulation comparison and conclusions for Collaborative Model vs Traditional Model that is in Literature � Integration of the Collaborative Model Simulation into MS Project � Krishna Neelakanta Thesis Proposal 10/08/2004 P ag e 9

Project Network Simulator �A simulator is to be built and tested. Language : C++. � Simulator to incorporate, initial evaluation of Project Network, dynamic re-evaluation based on the LP optimizer. Idea is to minimize the cost function � Simulator should support task addition and in mid project � Allows user input to select characteristics of desired project Network Krishna Neelakanta Thesis Proposal Page 10

Current Status � � � � Performed a literature Study of Simulation based Optimization in Project management and Collaboration efforts Studied related work in the area of collaborative computing Studied the Industrial Strength COMPASS (ISC), Linear Programming Optimizer Installed a version of ISC and ran some tests on the same Brushed up on Simulation Concepts and ideas Installed Microsoft Project 2007. Exploring ways to integrate simulator and the optimizer system into Microsoft Project, by studying the API’s available Krishna Neelakanta Thesis Proposal Page 11

Deliverables � The simulation tool, and the design and implementation documentation. � Integrating the Simulator to MS Project to help the Project Manager try the system � A thesis report documenting the design and implementation of Simulator System, related algorithms, the analysis of findings, and the lessons learned in thesis, and future work to be done. Krishna Neelakanta Thesis Proposal Page 12

High Level Timelines � Committee Proposal Approval – Dec 2009 � Develop model and design and implement simulator – Dec-Jan 2009 � Document findings, thesis report – Jan-Feb 2010 Krishna Neelakanta Thesis Proposal 10/08/2004 P ag e 13

References [1] Haga, W. A. 1998. Crashing PERT networks. Ph. D. Dissertation, University of Northern Colorado, Colorado. [2] Haga, W. A. , and K. A. Marold. 2004. A simulation approach to the PERT CPM time-cost trade-off problem. Project Management Journal, 35(2): 31 -37. [3] Haga, W. A. , and K. A. Marold. 2005. Monitoring and control of PERT networks. The Business Review, 3(2): 240 -245. [4] Hillier, F. S. , and G. J. Lieberman. 2008. Introduction to Operations Research. 9 th Ed. , Mc. Graw-Hill, New York. [5] Williams, T. (2004). Why Monte Carlo Simulations Of Project Networks Can Mislead. Project Management Journal, 35(3), 5361 [6] Xu, J. , B. L. Nelson, and L. J. Hong. 2007. Industrial Strength COMPASS: A Comprehensive Algorithm and Software for Optimization via Simulation. Website <http: //users. iems. northwestern. edu/~nelsonb/ISC/>. [7] Michael E. Kuhl, Radhamés A. Tolentino-Peña A DYNAMIC CRASHING METHOD FOR PROJECT MANAGEMENT USING SIMULATION-BASED OPTIMIZATION Proceedings of the 2008 Winter Simulation Conference Pg : 2370 -2376 [8] Nelson, L. J. and B. L. Nelson. 2006. Discrete optimization via simulation using COMPASS. Operations Research, 54: 115 -129. Krishna Neelakanta Thesis Proposal Page 14