Project Management Technique Simulation Modeling Analysis By: Jackie Holohan & Penny Leahy December 7, 2000 Simulation and Modeling Analysis Fall 2000 Professor Ernesto Gutierrez-Miravete 1 Table of Contents Topic Executive Summary 1.0 Introduction 1.1 Purpose & Performance Metrics 1.2 Scope 1.3 Target Audience 1.4 Application of Simulation 1.5 Flow Diagram of PMT 1.6 Data Collection 2.0 Model Details 2.1 Locations 2.2 Entities 2.3 Path Networks 2.4 Input Factor & Processing 2.5 Model Assumptions 3.0 How to run the Model 4.0 Model Results 4.1 Simulation Options 4.2 Example Problem to Solve 4.3 Analysis of Results 4.4 Confidence Intervals 4.5 Conclusions from Example 5.0 Model Verification & Validation 5.1 Verification 5.2 Validation 6.0 Conclusion References Page Number 3 4 4 5 5 5 6 6 8 8 15 15 17 20 21 22 22 22 28 30 31 31 31 31 34 35 2 Executive Summary Currently project managers in Information Technology are leading projects using Project Management Technique, a checklist of project tasks to complete a project. Often, projects are delivered late or over budget because hasty estimates are made with respect to the schedule and cost estimates. The goal of this project is to create a model of managing a project. The model can be tailored based on project complexity and experience level of the team members. The number of projects completed per time period to be studied can also be altered. The output of the model is the time to complete the project based on the input specified (complexity and experience level of the team). Cost can be calculated based on the time estimate. The model will enable both the 30-day project manager and 30-year project manager to make an educated guess for the schedule and budget. Project goals will be met and proper expectations will be set for upper management and customers. 3 1.0 Introduction There are several Information Technology (IT) challenges existing in business today varying in levels of complexity, duration and experience level of the resources. In order to increase the productivity and decrease variability in IT project management, businesses are designing methodologies to be used through the entire project. Project Managers need to meet their schedule and budget targets; therefore, they should be given sufficient time ad funds to do so. Generic procedures for project management are defined, including standard tasks and checkpoint reviews. These checkpoints ensure the project is proceeding correctly and will deliver a quality solution that customers want. Assuming that all projects delivered will meet the success criteria while using PMT, two fundamental questions IT professionals need to answer while managing a project are: How much will this project cost? How long will the project take using these resources? 1.1 Purpose The purpose of this simulation is to develop a tool that can be used by IT professionals to assess the cost and time requirements of a project. It is based on a five-step methodology called Project Management Technique (PMT). Each step includes necessary project tasks, as well as a checkpoint reviews. The simulation begins when the user specifies the following: Number of people and size of the project Complexity of the project Experience level of the resources Performance Metrics: The simulation will help to estimate the following: The amount of time to complete the 5 steps of a project. The cost to complete the project can subsequently be calculated based on the time applied. The results will help IT professionals make educated decisions about how to proceed with the project. Unreasonable guesses of project completion time and cost will be minimized. Instead, project leaders can be set up for success, given reasonable goals and targets. This simulation will help businesses create strategies, prioritize projects and plan resource allocation. The number of people, which people and scope of the project, can be optimized at the beginning. In addition, it will reduce idle time of resources (i.e. money, employees, etc.), enable people to complete realistic project plans and set a project team up for a successful delivery of a desired outcome. 4 1.2 Scope The topics/tasks that are in scope are: Information Technology projects using PMT methodology Estimate time to complete all project tasks in the PMT methodology Model should estimate the time to complete each phase in the PMT, as well as time to complete the entire project Cost can be determined based on the time to complete the project The topics that are out of scope are: All projects outside Information Technology All projects in IT not using the PMT methodology 1.3 Requirements The requirements of this project, given by project leaders at GE, are: Gather time data for each project task in PMT. Use the time in the simulation in the processing of project task times. Consider projects of various size, complexity, number of people on the team and experience levels on the team. Develop a model to simulate the amount of time to complete a project, as well as the time to complete each project task and phase. Allow the user to deduce the cost to complete a project. Be sure all bugs are worked out through the verification process. Validate the model with real-world data from the IT department at GE Indusrial Systems. 1.4 Target Audience IT Project Managers. The simulation will provide accurate cost and time estimates at the time of project conception. IT Project Team Members. This simulation will be a visual guide to project management and act as a training tool. Upper Management in the IT Organization. Upper Managers will be able to estimate the number of projects that can be done in a year. Planning for budgets and resources on a large scale will be feasible. 1.5 Application of the Simulation Prediction of cycle time and reduction of cost for IT project development work Prioritization of tasks in PMT and improvement of project quality Planning of work-flow for the entire project life cycle Analysis of throughput (i.e. number of projects which can be completed per year) Estimation of the optimal number of resources given project complexity, size and experience level of the team 5 1.6 Basic Flow Diagram of PMT The Project Management Technique is designed to help an individual, or a team, navigate through a project. The following flow chart displays the 5 phases of PMT. Once the required tasks are completed in each phase, the team meets with managers and stakeholders for a project update in a checkpoint review. The individual modules of PMT are discussed in the following sections. Figure 1.6 Project Management Technique Process Flow Project Management Technique (PMT) Process Flow C Control C 6 Checkpoint I Create Checkpoint B Investigate Checkpoint Baseline Checkpoint S Checkpoint Scope 1.6 Data Collection 18 managers, each experienced with PMT, were surveyed to find out the time statistics of each project task in different project scenarios. In order to allow the user of the Simulation model to study projects ranging in complexity and resource experience, an input factor (described in detail in section 2.4) was employed to change the task times depending on the project the user is studying. The basis of this input factor is in Figure 1.7. It was gathered from a group of highly experienced IT professionals who lead a project with low complexity. This type of project was chosen as the basis because it has the least variability, being a project with low complexity. In addition, this project would be run optimally considering the high team experience. 18 project leaders, ranging from various levels of experience, were asked the following for each of the project tasks: Based on your team's experience level, how long did it take to complete each project task most of the time? Based on your team's experience level, what is the maximum amount of time it takes to complete each project task? Based on your team's experience level, what is the minimum amount of time it takes to complete each project task? Figure 2.4d shows the responses to the above questions as given by 18 IT managers. These values were used to determine the adjustment factor (outlined in section 2.4). The type of project teams considered were: Teams with little project management technique experience (0 - 1 years) Teams with moderate project management technique experience (2-5 years) Teams with strong project management technique experience (>5 years) The feedback from each of the 18 project leaders was analyzed, validated and fitted in the simulation model as a triangular distribution. The triangular distribution is used because it provides a rough model in the absence of data. The 18 project managers estimated the minimum, maximum and mode (most likely value) for each PMT phase. Although the triangular distribution will create more variation than the true distribution, and the extreme values will not be captured, there isn't enough data to estimate which type of distribution each task takes. The extra variation is acceptable for the purposes of this simulation. 7 Figure 1.7: Project Task Time Data (In Hours) High Experienced Resources working with a Low Complexity Project Phase Task Max 2 1 3 2 3 4 5 6 7 Evaluate Alternatives/Risk Assessment Cost/Benefit Analysis Work/Resource Management Quality Plan /Customer Needs Architecture/Requirements Training (Project Team) 4 3 8 5 8 16 2 1 4.5 3 6 8 6 4.5 10 7 10 24 Baseline 8 9 6 10 Project Management Process Maps (As Is) Architecture/Requirements Resources/Cost Benefit/Schedule Status 16 8 8 5 14 6 6 1 18 10 10 7 Investigate 8 9 11 12 6 13 14 Project Management Process Maps (To Be) Ranking of Defects/Designing Solution Support Plan Architecture/Requirements Test Scenarios Peer Reviews 16 8 24 4 8 24 2 14 6 20 3 6 20 1 18 10 32 5 10 28 3 Create 8 15 13 12 14 16 Project Management Development of Solution Test Scenarios Support Plan Peer Reviews Prototype 16 24 24 4 2 12 14 20 20 3 1 8 18 36 28 5 3 16 Control 8 17 7 14 18 19 20 Project Management Finalize Development/Testing Training (User Groups) Peer Reviews Contingency Plans Lessons Learned Team Celebration 16 8 16 2 3 1 4 14 6 8 1 2 0 0 18 16 24 3 4 2 16 Scope 1 Mission/Problem Statement Mode Min 8 2.0 Model Details The following sections outline the theory behind the PMT simulation model. The task processing times used in the model are based on the data gathered from 18 projects. Section 2.4 describes how the user can adjust the model based on the complexity level of the project and the experience of the team. The simulation model consists of 20 locations and 1 entity, described below. 2.1 Locations There are 20 locations represented in the simulation by a yellow box and a code number. Locations are the project tasks to be completed at each project task phase. The locations belong to the "common task" phase or specifically to one of five core modules including: Scope Baseline Investigate Create Control The Figure 2.1 displays the tasks associated with each PMT phase. The task code numbers are included since they are used to represent the tasks in the simulation model layout. The "common task" section of the chart contains project tasks to be completed at more than one of the phases. For example, task C.3 (Project Management) needs to be completed at the Baseline, Investigate, Create and Control phases. 9 Figure 2.1 Locations and Codes Module Scope Code S.1 S.2 S.3 S.4 S.5 Baseline B.1 Resources/Cost Benefit/Schedule Status Investigate I.1 Ranking of Defects/Designing Solution Create Cr.1 Cr.2 Control Co.1 Finalize Development/Testing Co.2 Contingency Plans/Lessons Learned Co.3 Team Celebration Common Tasks C.1 C.2 C.3 C.4 C.5 C.6 C.7 Task Name Mission/Problem Statement Evaluate Alternatives/Risk Assessment Work/Resource Management Cost/Benefit Analysis Quality Plan /Customer Needs Development of Solution Prototype Architecture/Requirements Training (Project Team) Project Management Process Maps (To Be) Support Plan Test Scenarios Peer Reviews There is also a location called "Checkpoint Review", in which the team meets with management to describe the progress of the project. This meeting is held when all of the required tasks are completed for each of the five core modules. 10 2.1.1 Scope The first section of Project Management Technique is the scope. During this step the project team defines the mission statement, evaluates risks, alternatives and resource needs, and prepares the team for the project with necessary training. The following chart outline the tasks required for completion of the scope module. Please note, the mode, minimum and maximum values are in hours. Figure 2.1.1 Scope Requirements Code S.1 S.2 S.3 S.4 S.5 C.1 C.2 Task Name Mission/Problem Statement Task Requirements Define Objective, Scope & Approach Plan for the project. Evaluate Alternatives/Risk Assessment Define High Level Risks Risk Abatement Procedure Evaluate Alternative Designs or Software Packages Work/Resource Management Identify Project Approach Progress Report. Determine Detailed Work Tasks, Who is Responsible and Cost/Benefit Analysis Cost Analysis and Benefit classification Quality Plan / Customer Needs Identify Customer Requirements Quality Scorecard Architecture/Requirements Define & Document Project Requirements Define & Document Architecture Requirements Training (Project Team) Project Required Training Min Mode Max 1 2 3 2 4 6 4.5 8 10 1 3 4.5 3 5 7 6 8 10 8 16 24 TOTAL HOURS 25.5 TOTAL DAYS 3.188 46 5.75 65 8.1 Each task is assigned a code, used so that the task can be identified in the simulation model layout. The minimum, maximum and mode service time, listed above in hours, is based on a triangular distribution. This is because of the natural large variation in project task times and the lack of sufficient data. 11 2.1.2 Baseline The second module in the PMT process is called baseline, in which the team performs background studies on the project. The "As Is" business and information technology process is studied. There are 4 major tasks listed below that need to be completed before the checkpoint review. Figure 2.1.2 Baseline Requirements Code Task Name C.3 Project Management C.4 Process Maps (As Is) C.1 Architecture/Requirements B.1 Resources/Cost Benefit/Schedule Status Task Requirements Min Mode Max Team Meeting 14 Project Plan Progress Report Resource Management Quality/Risk Abatement Plan Cost/Benefit Analysis Updated Identify Financial Structure 6 Benchmark Current Processes Define Business Requirements Process Map Summary Review & Document Project 6 Requirements. Obtain sign-off Review & Document Architecture Requirements. Obtain sign-off Modify Schedule/Resources 1 Update Cost Benefit Analysis TOTAL HOURS 27 TOTAL DAYS 3.375 16 18 8 10 8 10 5 7 37 4.63 45 5.6 12 2.1.3 Investigate During the investigation phase of PMT, the team begins designing the solution and developing the test scenarios. Alternatives of the "To Be" solution is designed; options are investigated. There is also a step called Project Management that includes time for team meetings and administrative tasks. The following tasks must be completed before the checkpoint review meeting. Figure 2.1.3 Investigate Requirements Code C.3 C.4 I.1 C.5 C.1 C.6 C.7 Task Name Project Management Task Requirements Min Mode Max Team Meeting 14 Project Plan Progress Report Resource Management Quality/Risk Abatement Plan Process Maps (To Be) Develop Future Process Maps 6 Ranking of Defects/Designing Solution Rank the defects 20 Design a solution without defects Support Plan Specify Document Needs 3 Document Prototypes Architecture/Requirements System Availability 6 Future Application Deployment Test Scenarios Develop Test Scenario Model 20 Develop a Test Strategy Peer Reviews Conduct structured walk1 through meetings to review the code or design documents TOTAL HOURS 70 TOTAL DAYS 8.75 16 18 8 24 10 32 4 5 8 10 24 28 2 3 86 106 10.8 13 13 2.1.4 Create In the forth phase of PMT, the team begins building the solution and finalizing the test strategy. They also take time to do peer reviews, to gather feedback from the team on the project details. The plan to support the application post launch is also created. The following 6 tasks in Figure 2.2.4 must be completed before the checkpoint review can occur. Figure 2.1.4 Create Requirements Code Task Name C.3 Project Management Cr.1 Development of Solution C.6 Test Scenarios C.5 C.7 Support Plan Peer Reviews Cr.2 Prototype Task Requirements Min Mode Max Team Meeting 14 Project Plan Progress Report Resource Management Quality/Risk Abatement Plan Design Database 20 Identify Security Profiles User Process Narrative Conversion Data Mapping Setup Configuration Develop Test Strategy 20 Develop Detailed Test Plans Identify who will Test Design Production Support 3 Conduct structured walk-through 1 meetings to review the code or design documents Create a prototype of the design and 8 pilot it to key users groups. TOTAL HOURS 66 TOTAL DAYS 8.25 16 18 24 36 24 28 4 2 5 3 12 16 82 106 10.3 13 14 2.1.5 Control During the last phase of PMT, the team installs the solution and runs the final test scenarios. They also take time to train users and write a contingency plan. Upon completion of the tasks the team participates in a celebration then gives a final report during the last checkpoint review. Figure 2.1.5 Control Requirements Code Task Name C.3 Project Management Co.1 Finalize Development/Testing C.2 Training (Project Team) C.7 Peer Reviews Co.2 Contingency Plans/Lessons Learned Co.3 Team Celebration Task Requirements Min Mode Max Team Meeting 14 Project Plan Progress Report Resource Management Quality/Risk Abatement Plan Run Test Scenarios 6 Installation Test Systems Integration Test Test Report Training Manual 8 Develop On-line Help Conduct structured walk1 through meetings to review the code or design documents Determine plans for the 2 unexpected. Document lessons learned and share with other teams. Celebrate hard work with the 0 entire team TOTAL HOURS 31 TOTAL DAYS 3.875 16 18 8 16 16 24 2 3 3 4 4 16 49 6.13 81 10 15 2.2 Entities The system has one entity -- the person or group of people working on the project. The entity enters the system at the mission statement task in the Scope module. This is the first task to complete in the project management life cycle. The arrivals are set at a frequency of 4 (number of projects per year). The frequency of the projects can be changed in the model at the discretion of the user by going to the ProModel menu option: Build Arrivals Type in frequency number (This is the number of projects to be completed in the time allotted). 2.3 Path Networks and Processing When the entity enters the system, it starts at the mission statement task in the scope module. It proceeds through the system, going to each task and stopping at the checkpoint review after completion of each module (Scope, Baseline, Investigate, Create and Control). Figure 2.3 shows an entity flow diagram that maps the entity as it works its way through the PMT requirements. 16 Fig. 2.3 Entity Flow Diagram S.3 S.4 S.5 C.1, C.2 S.2 Scope ENTER 1 S.1 1 C.3, C.4, C.1 B.1 Baseline 2 2 C.3, C.4 I.1 Investigate 3 C.5, C.1, C.6, C.7 3 C.3 C.6, C.5, C.7 4 C.3 Cr.1 Create 4 Cr.2 Co.1 C.2, C.7 Co.2 EXIT Control Co.3 Common Tasks: C.1 C.2 C.3 C.4 C.5 C.6 C.7 Checkpoint Reviews: 17 2.4 Input Factor and Model Processing Input Factor There are several things to consider when planning for a project. Simplification is used to model the complexity of project management. This model considers 2 major factor when identifying the type of project to run through the ProModel simulation: Team experience level Number of team members and complexity of project This "input factor" depends on the team experience, number of team members and complexity of the project. The average experience level of the team is displayed in figure 2.4. Figure 2.4a Experience Level Rating Rating 3 2 1 Experience Level Low Medium High Years using PMT 0-1 2-5 >5 For the purpose of this simulation model, the number of team members and complexity are rated in the following chart. Figure 2.4b Duration and Complexity Rating Rating 1 2 3 Complexity Low medium High People 1-2 3-4 5-6 Name Small Project Medium Project Large Project Once the user finds the appropriate rating from the above tables, he or she can choose the input factor from the following chart. Figure 2.4c Input Factor Table Team Experience Level Rating Duration Or Complexity Rating Small Medium Large 1 2 3 Low 3 1 4 7 Medium High 2 1 2 5 8 3 6 9 18 Section 3.0 describes how the user changes the input factor from the default value of 1. To change the task times based on this input factor, the model uses the "adjustment value" (specified in Figure 2.4d) corresponding with the input factor chosen by the user. The task times, (listed in section 2.1) are multiplied by the below adjustment value. See Appendix B for more detailed calculations. Figure 2.4d Adjustment Factor Input Factor 1 2 3 4 5 6 7 8 9 Adjustment Value 1.3 1.2 1.1 3.2 2.8 2.4 5.7 4.8 3.9 Model Processing Based on the input factor, the service times at each of the locations (project tasks) are multiplied by the according adjustment factor. Several "If...then…else" statements are executed in the processing model in ProModel to adjust the service times accordingly. The service times are adjusted to account for the user's complexity and team experience level of the project. To summarize: Small projects that have low complexity and a low experience level (input factor 1; adjustment factor 1.3) will take longer to complete than small projects with low complexity and a high team experience level (input factor 3; adjustment factor 1.1). A large complicated project with a low team experience level will take the longest (input factor 7; adjustment factor of 5.7). A medium sized project with high experience on the team (input factor 6; adjustment factor 2.4) will take less time than a medium sized project with low experience on the team (input factor 4; adjustment factor 3.2). 19 Detail calculations behind the adjustment factor: Figure 2.4d Data Collected from 18 Projects Proj Input # Factor Scope Baseline Investigate Mode Min Max Mode Min Max Mode Min Create Control Adjust Factor Max Mode Min Max Mode Min Max 1 2 1 1 AVG 72 50 61 30 38 34 90 78 84 45 40 53 30 49 35 70 46 58 100 124 112 90 94 92 149 131 140 100 120 110 85 150 87 125 86 138 70 62 66 40 110 42 112 41 111 1.316 3 4 2 2 AVG 60 54 57 30 34 32 84 74 79 45 39 47 27 46 33 52 60 56 100 114 107 88 80 84 120 143 132 105 93 99 87 122 73 132 80 127 68 56 62 35 104 41 97 38 101 1.227 5 6 3 3 AVG 45 47 46 30 21 26 60 70 65 33 22 41 32 37 27 40 50 45 88 84 86 50 90 70 114 98 106 90 74 82 63 107 69 108 66 106 30 70 50 35 26 31 7 8 4 4 AVG 145 152 149 88 76 82 212 210 211 115 86 140 123 88 152 119 87 146 280 272 276 220 230 225 350 337 344 267 210 344 259 212 336 263 211 340 150 90 266 170 110 272 160 100 269 3.225 9 10 5 5 AVG 120 140 130 77 67 72 190 174 182 100 77 125 110 75 127 105 76 126 250 232 241 196 200 198 277 318 298 222 173 400 240 199 200 231 186 300 144 140 142 89 240 85 230 87 235 2.82 6 6 AVG 114 110 112 70 52 61 190 126 158 70 70 130 108 62 90 89 66 110 205 209 207 170 166 168 300 217 259 190 157 255 210 161 257 200 159 256 177 50 155 65 100 242 121 75 199 2.42 13 14 7 7 AVG 200 100 328 192 264 146 334 412 373 200 150 255 226 158 259 213 154 257 499 483 491 400 404 402 650 563 607 460 388 670 480 366 537 470 377 604 286 175 465 288 181 483 287 178 474 5.727 15 16 8 8 AVG 270 120 312 175 126 318 223 123 315 150 130 250 206 132 186 178 131 218 440 400 420 400 380 390 506 518 512 299 291 295 310 510 330 508 320 509 290 200 245 100 405 200 396 150 401 4.808 17 18 9 9 AVG 190 150 259 172 50 249 181 100 254 181 100 254 140 152 146 146 300 380 340 340 250 300 275 275 412 420 416 416 300 340 320 320 256 261 259 259 195 199 197 197 125 116 121 121 11 12 112 100 106 106 177 175 176 176 412 416 414 414 20 77 89 83 1.1 330 317 324 3.925 324 Calculations for Adjustment Factor: For input factor 1 ---(using the averaged times): Each time was divided ( i.e. time for mode of scope) by the corresponding time in the row for input factor 3. Those values were summed 15 numbers in the row for input factor 1. That total was divided by 15 to get the average multiplier or adjustment factor. Example (for input factor of 1): 61/48 + 34/26 + 84/65 +………..+ 111/83 = 19.74 19.74/15 = 1.3 Note: The Adjustment Factor is 1.3 for Input Factor 1.1. 2.5 Model Assumptions To narrow the scope of the project, the following assumptions were made during model creation: Between each location there is a .25 hour (15 minute) wait to account for breaks or meetings. The team members will be working full time on the project. There will be no "sideprojects". The workweek is 40 hours per week. The model is based on a 50 week-year (excluding holidays and vacations), working 5 days per week. Only one type of project (duration, complexity and experience level) can be studied during a simulation run. The inter-arrival frequency of the projects is currently set, but can be adjusted as needed by the user. The model output is in time, but this can be converted to cost in dollars at the user’s discretion using simple calculations. Due to the ProModel student version limitation on the number of locations (causing the use of common tasks) the output could not be divided to reflect the time needed for each module of the project. Checkpoint meetings are most often 1 hour, have a 30-minute minimum and a 1.5hour maximum. This service time for the checkpoint meeting is fit as a triangular distribution model {t(.5,1,1.5)}. 21 3.0 How to run the model 1. Open PMT.mod (Project Management Technique ProModel File). 2. Choose the Simulation Options menu option. 3. The PMT.mod should be set to run 10 replications of 2000 hours (1 years time = 40 hours per week * 50 weeks per year excluding vacation time). Change these options if desired. 4. Press "OK" on the "Simulation Options" screen. 5. Verify arrival rates by choosing the Build Arrivals. The frequency is currently set at 4 (i.e. 4 projects arriving to the team in 1 year, or 2000 hours). Change these options if desired. 6. Choose the Simulation Model Parameters menu option. 7. Press the " Change" button on the "Model Parameters" screen. 8. Change the Input Factor to the desired setting (1-9) in the highlighted cells below. This Input Factor represents the type of project you are simulating, with respect to team experience rating ad the duration/complexity of the project. Duration Or Complexity Rating Small Medium Large 1 2 3 Team Experience Level Rating Low Medium High 3 2 1 1 2 3 4 5 6 7 8 9 9. Press the "Run" button on the "Model Parameters" screen. This will run the Simulation according to the Input Factor specified. Once the simulation has completed, ProModel will prompt the user if he or she would like to see the results. 10. Press "Yes" to load the results. 11. A "General Report Type" screen will appear. You may want to choose to see "All" replications to view the time to complete each of the project tasks in the replication. This will also display the mean time of the 10 replications and the standard deviation. 22 4.0 Model Results The model for completing a project has been built using ProModel Student 4.2, based on the assumptions in section 2.6. See Appendix D for a view of the Model text details. 4.1 Simulation Options The standard model is set to run for 2000 hours, or 1 year's time (40 hours per week * 50 weeks per year -- excluding vacation time and holidays) 10 replications are run, to obtain average values for the time to complete each task. 4.2 Example Problem to Solve using PMT.mod To demonstrate and validate the simulation model, the following 3 scenarios were chosen for testing: Small complexity project with a low experienced team (Input Factor = 1) Medium complexity project with a medium experienced team (Input Factor = 5) Large complexity project with a highly experienced team (Input Factor = 9) These 3 scenarios were chosen because they are the most likely situations to happen in the real world. For example, a manager is more likely to assign a highly experienced person to a more complex project. Teams with low experience are unlikely to be assigned complicated projects. 4.2.1 Background For Example A new Information Technology enhancement has been requested by one of the key customers. The project entails changing the quoting and ordering software package to include one of the new products just released. The IT project manager has several options in planning for this enhancement request. The maximum amount of time allowed to complete this project is 6 months. 1. Scope down the request to be a series of 3 small projects (i.e. with iterative deliverables) with low experience levels on the team. New employees cost $40,000 per person year. The team only would have 2 people on the team. This project will only deliver the customer "satisfiers". The business needs to do this enhancement to stay competitive in the market. 2. Deliver the request as a medium project using moderately experienced people on the team. The team would have 3 people on the team. Moderately experienced people cost $65,000 per person year. 23 This project will only deliver the customer "satisfiers". The business needs to do this enhancement to stay competitive in the market. 3. Combine the enhancement with the large project currently in progress with highly experienced personnel. These employees cost $90,000 per person year. This project will have 5 people on the team. This project will deliver customer "satisfiers" and customer "delighters". The business will gain market share and sales if this project is completed. Figure 4.2.1 displays the output for the 3 simulation runs. Figure 4.2.1 Output from Examples Number of Employees Number of Projects Average Cost per Employee / year Average Cost per day Input Factor Phase Scope Task 1 3 1 $ 65,000.00 $ 250.00 5 5 1 $ 90,000.00 $ 346.15 9 1 2 3 $ 40,000.00 $ 153.85 Mean Stdev 26.828 2.465 25.654 1.702 Work/Resource Management 30.580 2.668 Cost/Benefit Analysis 12.002 1.818 Quality Plan /Customer Needs 34.493 5.470 39.412 1.675 Architecture/Requirements 66.946 6.447 Training (Project Team) Subtotal (hours) 235.916 9.679 Days 29.489 1.210 Total Cost for Scope $ 4,536.84 $ 186.13 1 Mission/Problem Statement 2 Evaluate Alternatives/Risk Assessment 4 3 5 6 7 Baseline 2 3 8 Project Management 9 Process Maps (As Is) 6 Architecture/Requirements 10 Resources/Cost Benefit/Schedule 67.544 46.486 39.412 16.746 1.984 5.527 1.675 3.358 Mean 5.048 11.091 21.507 7.302 14.065 22.064 43.165 124.241 15.530 $ 3,882.53 44.905 22.290 22.064 13.539 Stdev Mean 1.275 7.0314 2.359 15.4478 3.483 29.9559 1.021 10.1706 2.194 19.5903 1.758 30.7318 6.704 60.1219 8.557 173.050 1.070 21.631 $ 267.40 $ 7,487.73 1.429 1.160 1.758 3.418 Stdev 1.77625668 3.28631313 4.85162492 1.42194392 3.05562677 2.44832728 9.33779075 11.918 1.490 $ 515.69 62.5457 31.0462 30.7318 18.8581 1.99038453 1.61596985 2.44832728 4.76042647 4.262 143.182 .533 17.898 133.17 $ 6,195.37 5.935 .741 $ 256.82 Status Subtotal 170.188 Days 21.274 Total Cost for Baseline $ 3,272.85 $ Investigate 8 Project Management 9 Process Maps (To Be) 11 Ranking of Defects/Designing Solution 12 Support Plan 6 Architecture/Requirements 13 Test Scenarios 14 Peer Reviews 67.544 46.486 100.100 15.632 39.412 93.835 7.580 6.968 .871 134.01 1.984 5.527 4.953 0.369 1.675 2.769 0.311 102.797 12.850 $ 3,212.41 $ 44.905 22.290 71.246 11.601 22.064 68.783 5.732 1.429 1.160 6.540 0.512 1.758 3.649 0.535 62.546 31.046 99.236 16.159 30.732 95.805 7.984 24 1.990 1.616 9.110 0.713 2.448 5.082 0.745 Create Subtotal 370.590 Days 46.324 Total Cost for Investigate $ 7,126.72 $ 8.350 1.044 160.57 246.620 30.828 $ 7,706.88 67.544 114.246 93.835 15.632 7.580 49.352 Subtotal 348.189 Days 43.524 Total Cost for Create $ 6,695.94 $ 1.984 4.494 2.769 0.369 0.311 3.520 6.665 .833 128.18 44.905 76.567 68.783 11.601 5.732 34.214 241.802 30.225 $ 7,556.30 1.429 62.546 9.616 106.647 3.649 95.805 0.512 16.159 0.535 7.984 4.041 47.655 11.167 336.796 1.396 42.099 $ 348.97 $ 14,572.89 1.990 13.394 5.082 0.713 0.745 5.628 15.554 1.944 $ 673.01 67.544 38.345 66.946 14 Peer Reviews 7.580 18 Contingency Plans/Lessons Learned 11.832 20 Team Celebration 24.391 Subtotal 216.638 Days 27.080 Total Cost for Control $ 4,166.11 $ 1.984 4.806 6.447 0.311 0.902 9.874 12.923 1.615 248.52 44.905 30.374 43.165 5.732 8.287 14.980 147.442 18.430 $ 4,607.57 1.429 62.546 7.804 42.306 6.704 60.122 0.535 7.984 0.842 11.543 9.139 20.865 13.871 205.366 1.734 25.671 $ 433.46 $ 8,886.04 1.990 10.870 9.338 0.745 1.173 12.729 19.320 2.415 $ 835.96 14.656 1.832 281.84 $ 1.360 0.170 26.15 8 Project Management 15 Development of Solution 13 Test Scenarios 12 Support Plan 14 Peer Reviews 16 Prototype Control 8 Project Management 17 Finalize Development/Testing 7 Training (User Groups) Checkpoint Review Meetings (hours) Days Cost for Checkpoint Meetings $ Grand Total (Hours) 1356.176 91.506 Days 169.522 11.438 Total Cost $ 26,080.31 $ 1,759.72 Total Weeks Total Months Total Months / 1 project * 33.90 8.48 2.83 $ 1.027 0.128 32.09 $ 7.945 343.507 .993 42.938 248.27 $ 14,863.29 $ 0.081 0.010 2.53 $ 1.027 0.128 44.43 $ 11.066 1.383 478.80 0.081 0.010 3.50 863.929 88.458 1202.927 123.176 107.991 11.057 150.366 15.397 $26,997.79 $ 2,764.31 $ 52,049.74 $ 5,329.71 21.60 5.40 5.40 30.07 7.52 7.52 * 3 project examples do compare with the real world. Small projects usually take under 3 months, medium projects take from 3-6 months and large projects are > 6 months. 25 4.2.2 Run Simulation #1 1. Change the arrivals to have only 3 occurrences (Build Arrivals. Change occurrences to be 3 projects) 2. Change the simulation options to run for 6 months (8 hours per day * 5 days per week * 4 weeks per month * 6 months = 960 hours). This can be done by going to the Simulation menu, then choose Options. The run hours can be changed to 960. The replications can stay at 10 to get an average time to complete the 3 projects. 3. Set the input factor to be 1 (Small project with low experience levels on the team). Go to the Simulation Model Parameters menu. Change the Input Factor to = 1. 4. Press "Run" on the Model Parameters screen. Output Phase Scope Baseline Investigate Create Control Total Cost Total (Days) Total (Weeks) Total (Months) 95% Confidence Interval for Cost Cost Mean $ 4536.84 $ 3272.85 $ 7126.72 $ 6695.94 $ 4166.11 $ 26,080 Stdev $ 427.78 $ 241.21 $ 338.22 $ 258.60 $ 467.77 $ 1,760 Time (Days) Mean 29.49 21.27 46.32 43.52 27.08 Stdev 2.78 1.57 2.20 1.68 3.04 169.52 33.90 8.48 11.44 2.29 .57 95% Confidence Interval for Time 26 4.2.3 Run Simulation #2 1. Change the arrivals to have only 1 occurrence (Build Arrivals. Change occurrences to be 1.) 2. Be sure the simulation options will run for 6 months (8 hours per week * 5 days per week * 4 weeks per month * 6 months = 960 hours). This can be done by going to the Simulation menu, then choose Options. The run hours can be changed to 960. The replications can stay at 10 to get an average time to complete 1 project. 3. Set the input factor to be 5 (Medium project with medium experience levels on the team). Go to the Simulation Model Parameters menu. Change the Input Factor to = 5. 4. Press "Run" on the Model Parameters screen. Output: Phase Scope Baseline Investigate Create Control Total Cost Total (Days) Total (Weeks) Total (Months) 95% Confidence Interval for Cost Cost Mean $ 3882.53 $ 3212.41 $ 7706.88 $ 7556.30 $ 4607.57 $ 26,998 Stdev $ 587.33 $ 242.66 $ 486.97 $ 618.17 $ 826.65 $ 2,764 Time (Days) Mean 15.53 12.85 30.83 30.23 18.43 Stdev 2.35 .97 1.95 2.47 3.31 107.99 21.60 5.4 11.06 2.21 .55 95% Confidence Interval for Time 27 4.2.4 Run Simulation #3 1. Change the arrivals to have only 1 occurrence (Build Arrivals. Change occurrences to be 1). 2. Be sure the Simulation Options will run for 6 months (8 hours per day * 5 days per week * 4 weeks per month * 6 months = 960 hours). This can be done by going to the Simulation menu, then choose Options. The run hours can be changed to 960. The replications can stay at 10 to get an average time to complete 1 project. 3. Set the input factor to be 9 (Large project with great experience levels on the team). Go to the Simulation Model Parameters menu. Change the Input Factor to = 9. 4. Press "Run" on the Model Parameters screen. Output: This large project takes 7.52 +/ .77 months. This breaks the limit of 6 months. For an extra 1.5 months and extra $26,000, this project can be completed. Cutting the project off at 6 months will leave the in the "Final Development and Testing" phase. The project will not complete development, testing, identifying lessons learned and best practices, or the team celebration. Phase Scope Baseline Investigate Create Control Total Cost Total (Days) Total (Weeks) Total (Months) 95% Confidence Interval for Cost Cost Mean $ 7487.73 $ 6195.37 $14863.29 $14572.89 $ 8886.04 $ 52,050 Stdev $ 1132.70 $ 467.96 $ 939.14 $1192.18 $ 1594.24 $ 5330 Time (Days) Mean 21.63 17.90 42.94 42.10 25.67 Stdev 3.27 1.35 2.71 3.44 4.61 150.37 30.07 7.52 15.40 3.08 .77 95% Confidence Interval for Time 28 4.3 Analysis of Results Comparing Scenario 1 versus Scenario 2 The total cost of the 2 scenarios is quite similar. Scenario Scenario 1: 3 small projects in 6 months with a low experienced team. Input Factor = 1. Scenario 2: 1 medium project in 6 months with a medium experienced team. Input Factor = 5. Total Cost $ 26,080.31 = Xbar $ 1,759.72 = s $ 26,997.79 = Xbar $ 2,764.31 = s Xbar is the sample mean. "s" is the sample standard deviation. Using Minitab, a 2 sample t test was run to see if the cost of Scenario 1 is significantly different that Scenario 2. H0: The 2 scenarios have equal costs (c1 = c2) H1: The 2 scenarios have unequal costs (c1 <> c2) In Minitab, the menu item: Stat Basic Statistics 2 sample t was run to test the hypothesis. The following results were obtained: Two Sample T-Test and Confidence Interval Two sample T for Design1cost vs Design2cost Design1 Design2 N 5 5 Mean 5160 5433 StDev 1671 2155 SE Mean 747 964 95% CI for mu Design1c - mu Design2c: ( -3159, 2612) T-Test mu Design1c = mu Design2c (vs not =): T = -0.22 P = 0.83 DF = 7 Since the p value is .83, we can not reject the null hypothesis at a 95% confidence level. In other words, there is not enough evidence to conclude that the 2 scenario's costs are significantly different. Either scenario can be completed to get the job done. There is no significant difference between the 2. Scenario 3 Scenario 3 costs $ 52,050 with a standard deviation of $ 5,330. This is nearly double the cost for either Scenario 1 and/or 2. The benefits of spending more are: Scenario 1 or 2 cost less; however, the business will only deliver benefit to its customers. These projects will only deliver the customer "satisfiers". The business needs to do this enhancement to stay competitive in the market. 29 For Scenario 3, approximately 26,000 more will need to be spent; however, more benefit will be delivered to the customers. This project will deliver customer "satisfiers" and customer "delighters". The business will gain market share and sales if this project is completed. 4.4 Confidence Intervals In order to verify that the model is producing data that is representative to the population the following confidence intervals were calculated. The mean of each simulation example falls within the confidence interval therefore verifying that the outcome represents the population. Figure 4.4.6 Confidence Intervals Simulation Example Number 1 Scope Baseline Investigate Control Create Average Hours Standard Deviation Hours Confidence Interval 235.916 22.244 272.51 Average Hours Standard Deviation Hours Confidence Interval 170.188 12.543 190.82 Average Hours Standard Deviation Hours Confidence Interval 370.590 17.588 399.52 Average Hours Standard Deviation Hours Confidence Interval 348.189 13.447 370.31 Average Hours Standard Deviation Hours Confidence Interval 216.638 24.324 256.65 2 199.32 124.241 18.795 155.16 149.56 102.797 7.765 115.57 341.66 246.620 15.583 272.25 326.07 241.802 19.781 274.34 176.63 147.442 26.453 190.96 3 93.32 173.050 26.178 216.11 129.99 90.02 143.182 10.815 160.97 125.39 220.99 343.507 21.704 379.21 307.80 209.26 336.796 27.553 382.12 291.47 103.93 205.366 36.845 265.98 144.76 30 4.5 Conclusion of example above There is no significant difference between Scenario 1 and Scenario 2. Scenario 3 costs approximately twice as much ($26,000); however, the business will gain market share and sales. The question is "How much market share and sales will be gained?" If the market share and sales gained exceed $26K, Scenario 3 is recommended if the customer is willing to wait an extra 2 months for the project to be delivered. If the customer is not willing to wait an extra 2 months for the extra benefits, either Scenario 1 or Scenario 2 are recommended. 5.0 Model Verification and Validation Verification is used to debug the model. Validation is used to relate the conceptual model to the output of the simulation. 5.1 Model Verification Model verification was used to remove program bugs from the model. In order to verify that the conceptual system is representative of our model we asked an outsider to try to run the model. As a result of this action we found the following problems with our model: The last task was not connected in the path networks and therefore was not producing output. The model’s output should have been based on hours, but it was giving us minutes. After performing this exercise and running numerous replications of the model we have concluded that our model is representative of the conceptual system of PMT. 5.2 Model Validation In order to validate that our model is producing output that is representative of the actual system data was collected from managers who have worked with projects that match the profile for input factors 1, 5, and 9. The managers were asked the questions listed in section 1.5. Figure 5.2a shows the data that was gathered. Figure 5.2a Example Output Proj Input # Factor 1 2 3 1 5 9 Scope Baseline Investigate Create Control Mode Min Max Mode Min Max Mode Min Max Mode Min Max Mode Min Max 70 33 85 50 30 60 115 95 135 110 85 140 70 40 110 125 70 185 103 75 130 246 200 300 241 185 300 148 90 235 174 100 260 145 100 180 345 280 400 335 260 415 206 130 350 31 To validate that the data collected in Figure 5.2a is the same as data that would have been produced by the simulation model the following hypothesis test was employed: Ho: 1 = 2 Ha: 1 <> 2 In this test 1 is the value as predicted by managers that is displayed in Figure 4.4. The 2 represents the mode value given by the output of the simulation runs from section 4.2 displayed below in the Figure 5.2b. Figure 5.2b Output from Example in Section 4.2 1 Scope Input Factor 2 3 Average Hours 235.916 Standard Deviation 22.244 124.241 18.795 173.050 26.178 Average Hours 170.188 12.543 102.797 7.765 143.182 10.815 Investigate Average Hours 370.590 Standard Deviation 17.588 246.620 15.583 343.507 21.704 Control Average Hours 348.189 Standard Deviation 13.447 241.802 19.781 336.796 27.553 Create Average Hours 216.638 Standard Deviation 24.324 147.442 26.453 205.366 36.845 Baseline Standard Deviation The test statistic is based on an level of 10% and n equal to 10 replications: t(1-/2, n-1) = t(.95,9) =1.833 The tested value (t*) is based on the formula: t* = (1 - 2)/ (s/√n) 32 The t* values for the 3 scenarios in Figure 5.2a are: Figure 5.2c T* Test Values t* values Scope Baseline Input Factor 1 5 9 22.62491 -0.1265 -0.10962 30.047 Investigate 45.10412 -0.087 -0.5454 0.124 -0.21329 Control 54.96669 0.121632 0.199556 Create 18.32975 -0.06438 -0.05283 The hypothesis test criteria is if: |t*| <= t(1-/2, n-1) conclude that the null hypothesis (Ho) is true, else conclude that the alternate hypothesis (Ha) is true. For Input Factor equal to 5 and 9, projects with: Medium experienced people and medium complexity (input factor = 5) Highly experienced people and high complexity (input factor = 9) The absolute values for t* given (in Figure 4.4b) by the model are less than the test statistic t. Therefore we would accept the null hypothesis that the 2 means are the same. For Input Factor equal to 1, projects with: Low experienced people and low complexity (input factor = 1) The t values given by the model are greater than the test statistic t. Therefore we would reject the null hypothesis that the 2 means are the same. This indicates that the outcome of our simulation is not producing data that is from the same population as the data given by this manager. This could be due to inaccurate data given by the team or manager, or an incorrect input factor rating. For example, maybe the project should have been rated more complex or maybe the complexity increased 33 during the project. It is important to point out that because of the large range of IT projects that can be performed, a study of this kind would have to be very specialized to a specific project type and team. Because we were able to accept the null hypothesis for the 2 other tests under all of the modules, we feel that our model is validated. We feel that the data given by the manager should have been given a different input value rating. The outcome of the hypothesis tests verifies that the applications of this simulation are very specified and projects data must be clearly defined to produce adequate output. 6.0 Conclusion The purpose of this simulation model is to provide managers in IT businesses a tool to estimate project duration and costs. In order to provide adequate output, the model must be tailored to reflect actual data collected in certain office situations. Based on the data gathered from a General Electric IT department, the output from the simulation model has been validated. However, in some instances, as shown in Section 4.4, the output may vary from actual data gathered from the population because of Assignment of an incorrect input factor Incorrect task times Increased variation in the project To avoid inaccurate project time output the simulation must be modified to reflect data collected from the office that is being studied. For instance, this model would need to be modified if Rensselaer wanted to use it to predict the time and cost of IT projects in their school. In conclusion, this project has shown that simulation models can be used to predict the time and cost of projects as long as the data used to create the model is based on data collected from the specific population being studied. 34 References Harrell, Charles, Ghosh, Diman and Bowden, Royce. Simulation using ProModel. Boston: The McGraw Hill Companies, Inc., 2000. Law, Averill and Kelton, W. David. Simulation Modeling and Analysis - Third Edition. Boston: The McGraw Hill Companies, Inc., 2000. McKenna, Christine. Project Management Methodology © 1999. General Electric Company. Neter, John, Kutner, Michael, Nachtsheim, Christopher, and Wasserman, William. Applied Linear Regression Models. Chicago: The McGraw Hill Companies, Inc., 1996. 35