Using SimVision for Organization Development. 2. SimVision and Organization Design

Transcription

1 Using SimVision for Organization Development 2. SimVision and Organization Design

2 Contents The Information Processing View of Organizations Fast-Tracking Projects The Hidden Work in Fast-tracked Projects How SimVision Models Hidden Work Managing Information Overload Managing Information Overload with Modularization How SimVision Models Modularization Managing Information Overload by Increasing Organizational Slack Examples of Increasing Organizational Slack Using SimVision Using Matrix Organizations to Manage Information Overload Understanding Task Interdependence Pooled Interdependence Sequential Interdependence Reciprocal Interdependence Coordination Methods for the Three Types of Interdependence Types of Workflow Thompson s Model of Workflow Pooled Workflow Sequential Workflow Reciprocal Workflow Types of Decision-Making Concurrent Engineering and Parallel Workflow Types of Work About Constraints The Principle/Agent Problem A Quick Tour of the SimVision Application Terms and Concepts Understanding Models, Programs, Projects, and Cases Examining the Startup Model Examining the Default Project Using the Tree Pane Reasons for Using SimVision Using SimVision With Hands-On Examples Step 1: Define Model Requirements Hands-On: Defining the Requirements for the Arnold House Model Step 2: Build a Baseline Model Step 2. 1 Define a Project Work Process Plan Hands On: Defining the Milestones and Tasks for the Build House Project Step 2. 2 Create the Model Hands On: Creating the Arnold House Model Step 2.3: Define the Project Organization Hands On: Defining the Build House Project Personnel Step 2.4: Assign Each Task to a Responsible Position Hands On: Assigning Tasks to Positions in the Build House Project Step 2.5: Set Task Work Volumes Hands On: Setting Task Work Volumes in the Build House Project Step 2.6: Model Task Interdependencies Hands On: Modeling Task Interdependencies in the Build House Project Step 2.7: Set Probabilities Hands On: Setting Probabilities in the Arnold House Model Step 2.8: Add Skills to Tasks and Positions Hands On: Adding Task and Position Skills to the Build House Project Step 2.10: Add Meetings

3 Hands On: Adding Meetings to the Build House Project Step 2.11: Add Cost Data Hands On: Adding the Cost Data to the Build House Project Step 3: Simulate the Baseline Model Hands On: Simulating and Analyzing the Baseline Arnold House Model What to Do Next Background Reading

4 The Information Processing View of Organizations The term information processing refers to people, or actors, as processors of information. The fundamental idea behind information processing is that organizations involve a volume of information that needs to be processed by a certain number of actors. Information processing is concerned with the rate at which the actors can process the information, and the factors that affect this rate, such as skill levels, experience, and the complexity of the tasks involved. The basic ideas of information processing were formulated by March and Simon at Carnegie Mellon University, and crystallized by J. Galbraith in the early 1970s. Galbraith s theory of information processing Galbraith further posited the following: That people carry out tasks and it takes time, the measurement of which, with no calculation of exceptions or other distractions, is termed the CPM (Critical Path Method). That tasks have exceptions, or errors that occur during their execution. That organizations are basically exception-handling machines. That people look for help most frequently laterally from their peers, which he called the gangplank method of coordination. That the gangplank coordination method leads to matrix organizations to reduce bottlenecks occurring at the Project Management level. You ll learn more about matrix organizations later in this class and in Class 4. Institutionally appropriate behavior People act instrumentally (i.e. for reward) but also because of who they think they are. There are a number of sociological institutions such as home, school, church whose rules and patterns of behavior affect how people behave. Behavior ruled by such institutions is called institutionally appropriate behavior. 2-4

5 Fast-Tracking Projects Most projects feature a large number of tasks that must be undertaken concurrently. Furthermore, these tasks are often interdependent, meaning that errors in one cause rework in others. Today s competitive markets often require that project schedules be reduced, in many cases unrealistically. Such time-to-market shrinkage results in fast-tracking also known as concurrent engineering or the process of scheduling many interdependent tasks concurrently. Fast-tracking almost inevitably results in higher levels of coordination and rework, which in turn can overwhelm project managers forced to deal with increasing levels of information under tighter and tighter time constraints. The Hidden Work in Fast-tracked Projects The problem with the CPM method of calculating the duration of a fast-tracked project is that CPM assumes that all actors are working without interruption on their tasks, and that tasks overlapped for the purpose of fast-tracking can be carried out concurrently or partially concurrently without adverse effect. The CPM method assumes that the only work that will occur is the direct work involved in each task. In reality, overlapping tasks generate what is known as hidden work, which cannot be modeled by the CPM method. The more tasks are overlapped, the more exceptions tend to occur, and thus the more coordination and rework is required. This hidden work prolongs tasks, delays others, and can have a knock-on, or even an exponentially slowing effect in a project. 2-5

6 How SimVision Models Hidden Work In SimVision, the basic unit of simulated work (a work item ) typically has a volume of one day, depending on the total task work volume (which can be almost any duration or work volume). Work items are distributed to workers according to the assignment and staffing structure in the model. Every worker maintains an inbox queue of events requiring his attention. Each individual work item may give rise to an exception (problem), or a communication event, both of which are examples of hidden work. An exception triggers a cascade of exception handling events through the supervision structure, leading to more hidden work. A communication is passed to appropriate colleagues as defined by the communication links. You can also model the coordination and rework arising from overlapping tasks, and the resulting indirect work is factored in to the overall schedule. The model explicitly shows predicted coordination and rework with green and red links between interdependent tasks, as shown here: 2-6

7 During simulation, these links translate into indirect work that you can tangibly see in the simulations charts. In SimVision, indirect work is categorized as either rework, coordination, or decision wait time. The program and project Breakdown charts shows these three types of work along with direct work for each task in a project, as follows: Managing Information Overload There are three ways described below for dealing with the bottlenecks of information that occur when projects are fast-tracked: modularization, increasing organizational slack, and using lateral coordination in a matrix organization. 2-7

8 Managing Information Overload with Modularization Interdependency is expensive, time-consuming, and requires more coordination and rework. It invariably leads to information overload at the project management level. One way to manage the information overload that results from high levels of interdependency is to modularize, or decentralize management in the organization. By partitioning a project into nearautonomous sub-projects, significant authority can be delegated to the project managers of these sub-projects, reducing information processing bottlenecks. How SimVision Models Modularization SimVision allows you to model the centralization of decision-making within an organization. You do this by setting the Centralization property for a project or a suite of projects. By default, centralization is set to medium, which means that most decisions are made at a middlemanagement (subteam leader) level. You can set centralization to High, modeling an organization where most decisions are referred to and made by high-level managers. To cope with the information bottlenecks that might result from high centralization, you can set centralization to Low, modeling a situation where subteams and low-level team members are empowered to make most decisions and very few are referred up the hierarchy. In terms of how you model project team configurations with SimVision, there is a three-level hierarchy you can use. Organizations: At the program (suite of projects) level, you can model organizations that you link with individual projects. Organizations are represented on the program page, along with the program s milestones and project, as shown here. Departments and subdepartments: Within each organization, you can model a hierarchy of departments and sub-departments. Departments and subdepartments are modeled on their respective organization s page. Organizations on the program page Departments within an organization Persons: Within each department and subdepartment, you can model the individual persons. Persons in SimVision are not represented by icons like most other objects, such as organizations, project, and departments. Instead, each department or 2-8

9 subdepartment has an associated list of persons, which you can create and manipulate in that department s Person List dialog box. You can also use this dialog box to staff positions with actual persons. If you are modeling a project for which the personnel structure needs to be made more or less modular, you can adjust the structure of organizations, departments, and their staffing persons either to move people out of or into functional silos. For example, the following illustration shows departments in a software company organized by function: The next illustration shows the same software company, this time with its departments organization by project: Managing Information Overload by Increasing Organizational Slack The following are some other ways of managing information overload on a project, and should be considered when you are making interventions in a SimVision model. You can: allocate larger budgets schedule longer activities increase the project duration (not usually an option!) over-design components for independence and redundancy use more staff use staff with higher skills increase the skills of existing staff 2-9

10 Examples of Increasing Organizational Slack Using SimVision SimVision models project staff as positions, which can represent one or multiple persons, and individual persons who staff the positions. Positions are measured in FTEs (full-time equivalents), which you apply as a property of the position, so it s easy to increase the FTE value of a position by just changing the value of its FTE property. Of course, when adding FTEs, you have to take into account the ramp-up time that new resources will require, not to mention the negative impact of adding a new member to an already formed team, and the possibility that the new FTE does not have the right skills at the right levels, or the right experience for their assigned tasks. Skills are also modeled as properties of both positions and persons, so increasing the skill level is just a matter of adjusting the level of the Skill property. This and other methods of reducing information overload and decision-wait time are discussed later in the class, as you do the handson exercise. Using Matrix Organizations to Manage Information Overload Another solution for the information overload problem lies in the development of matrix organizations. These organizations rely heavily on lateral coordination Galbraith s gangplank method referred to earlier which means communication between peers, who solve problems among themselves and relay the solutions to common supervisors. Matrix organizations will be described in greater detail in a later class. Shoe-size management problem One challenge that can occur in a matrix organization is that the people with the loudest voices and/or biggest egos tend to be heard the most. This leads to so-called shoe-size management, i.e. the person banging the biggest shoe on the table wins out. Understanding Task Interdependence Interdependencies occur when tasks are broken into specialized pieces. There are various ways in which tasks or people can be interdependent. Thompson describes three types of interdependence: reciprocal, pooled, and sequential. Pooled Interdependence Pooled interdependency occurs when the branches of an organization have common goals, and when all branches are required to succeed for the organization to succeed. For example, two airplanes in the same company might require maintenance simultaneously, but there is only one maintenance crew. 2-10

11 Sequential Interdependence Sequential interdependency occurs when one branch of an organization is dependent on the output of another branch to begin functioning. For example, one branch might need to receive parts produced by another branch before they can begin assembling the parts. With sequential interdependence, you might still have pooled interdependencies, but you also need to micromanage the information flow between sequential workers who rely on the worker before them in the workflow. This type of interdependence makes scheduling difficult, because if one person in the workflow is late with their output there s a knock-on effect, which can become very costly. You also have to factor in uncertainties, such as the weather, which increase coordination requirements. There s a need with sequential interdependence for lateral coordination, such as that found in matrix organizations, to reduce management bottlenecks. Two ways to deal with increased coordination are to reduce task length and to make tasks more parallel. Reciprocal Interdependence Reciprocal interdependency occurs when people or tasks have conflicting goals that need to be negotiated. For example, in an office building project the architect is likely to care about having wide open spaces with no columns obstructing the view and interaction of the workers; the structural engineer, in contrast, cares about pounds of steel per cubic foot and is likely to want columns to reduce the load-bearing spans. The two sets of requirements represent negatively interacting goals (columns vs no columns) that require tradeoffs. Coordination Methods for the Three Types of Interdependence There are different ways of coordinating interdependent tasks, depending on the type of interdependency. Pooled: With pooled interdependencies such as one that involves a shared resource, rules and standards are more efficient for coordinating the work. In this situation, project managers are likely to coordinate between specialists using fixed rules with little room or need for mutual adjustment of goals. Project groups can be larger because the decisions are governed by the rules and standards and there is less need for communication or goal adjustment. Sequential: With sequential interdependencies, there is more of a need to micromanage the information flow between sequential workers, so scheduling is key. Sequential workflows are more susceptible to the uncertainty of outside factors, such 2-11

12 as the weather, and this uncertainty increases the need for communication so some lateral coordination is also required, and rules and schedules need to be updated frequently to avoid the knock-on effect of changing circumstances. Reciprocal: With reciprocal interdependencies, rules and standards are likely to be too rigid and mutual adjustment of work practices and goals is required for flexibility. In reciprocal workflows, groups are typically smaller and more self-contained and tangential to reduce coordination requirements. Types of Workflow In response to the three types of interdependence, there are three corresponding types of workflow, pooled, sequential, and reciprocal. Thompson s Model of Workflow In Thompson s model of the flow of work in an organization, shown below, managers are directly responsible for support (admin) staff and for individual workers. 2-12

13 In this model, you can standardize on the following: Input/required skills: e.g. journeymen for bricklaying Work processes: e.g. making a concrete cylinder Outputs (specs): e.g. required concrete strength after 28 days Pooled Workflow In response to pooled and sequential interdependence, there are two types of workflow, the first of which is pooled workflow. In pooled workflow, workers have a common workflow, as shown by the green arrows in the following diagram. 2-13

14 As discussed, the Project Manager coordinates between specialists using rules and standards, which are imposed by the support or admin staff, as shown in the following diagram. These rules and standards determine input skills, processes, and outputs. Constraints of resource sharing occur, e.g. a shared computer onsite requires scheduling. Sequential Workflow In sequential workflow, a series of specialists work interdependently but sequentially, as shown by the green workflow arrows in the following diagram. 2-14

15 These specialists are coordinated by direct supervision and hierarchical planning, as shown by the orange arrows in the following diagram. An example of sequential workflow is on a construction site where the site has to first be cleared; the foundation is then laid; then rebar has to be put in; then the walls are built, and so on. In another example, a patient admitted to hospital for surgery first has to undergo pretreatment, then the anesthetist has a turn, followed by the surgeon, followed by nurses for post-surgical care. In both these examples, you can see how, if one team member is delayed, it affects all others down the line. Reciprocal Workflow With reciprocal workflow, tasks have negatively interacting sub-goals, as shown by the green arrows in the following diagram. 2-15

16 These negatively interacting goals are best coordinated by mutual adjustment, which can be achieved by lateral coordination, as shown by the orange arrows in the following diagram. Types of Decision-Making In response to the types of interdependencies, there are at least two ways of making decisions in an organization: Sequential decisions: Simply put, the first person in line wins. For example, in the pooled interdependence example of the airplanes, whichever airplane arrives first in the maintenance hangar gets attended to first. Parallel decisions: Synchronous decision-making that requires tradeoffs, generates more coordination, and is generally more difficult to carry out. The advantage of this method of making decisions is that it can reduce rework, because all interested parties are involved in the decision-making process. Concurrent Engineering and Parallel Workflow Also called over the wall engineering, concurrent engineering is a highly parallel way of doing business that involves input from all disciplines in an organization. An example is a car manufacturing company that employs designers in L.A. who throw the finished design over the wall to the manufacturers in Detroit. Problems arise when the design is flawed, generating costly rework in the form of change orders. A more balanced approach is to have manufacturing involved in the design, which is likely to eliminate some of the rework because design flaws can be eradicated before the design is approved. Types of Work The following table shows three types of work that can occur in an organization: Unskilled Workers do not need any specific skills, training can be carried out on the job e.g. busboys, dishwashers Craft Only outputs are specified, skills are assumed Workers need a license e.g. journeymen, electricians Professional Workers have to have degrees and/or board certification e.g. architect, engineer, doctor 2-16

17 About Constraints There are a number of constraints affecting the successful operation of an organization or a branch thereof. Constraints typically necessitate sequential tasks. An example is a hazard constraint, for example, welding need to occur in an area where flammable paint is being applied. To avoid the fire hazard, weld first, paint after. The Principle/Agent Problem The principal/agent problem occurs where workers play games that hamper coordination. For example, in a situation where an important resource needs to be shared between groups, the resource may be hoarded by one or more groups. The resource may be a piece of equipment such as a crane, a critical person, or even office space that is at a premium, such as lab space in a university. The manager trying to schedule resources in this situation will have problems identifying the real priority of needs because everyone is claiming that their need for the resource is critical. 2-17

18 A Quick Tour of the SimVision Application The trick of good model building is what you leave out Ray Levitt Terms and Concepts Before you start using SimVision, you should understand the following terms and concepts: Model-A visual representation of a program and its projects. Program-A set of related projects that together produce the product or products. A program also includes the associated responsible organizations, milestones, and relationships between projects. Project-A set of related product development tasks including positions (groups of one or more individuals) that perform tasks and the dependencies between tasks and positions. Case-A specific instance of a program. Portfolio-A program of multiple projects. Pane-An area of the workspace. SimVision has five panes: the Model, Tree, and Properties panes, Table View, and the Fix Simulator Errors and Warnings pane. Workspace-The area of the application where you work, which contains the Model pane, other optional panes, and the toolbars. Simulator-Software that simulates the work done by positions as they perform individual project tasks, including both planned direct work and coordination and rework. Simulation charts-charts that summarize and provide details of the simulated performance of the program and the individual modeled projects. Shapes-A set of color-coded objects that represent projects, milestones, tasks, positions, organizations, departments, meetings, ghost milestones, and ghost tasks. Links-A set of color-coded arrows that represent the relationships between shapes. Objects-An umbrella term for shapes and links. Understanding Models, Programs, Projects, and Cases Before you begin constructing a SimVision model of your project, it's helpful to understand how SimVision defines models, and the difference between a program, a project, and a case. A model is the graphical representation of a program. Programs contain one or more projects and organizations. A case is a specific instance of a program, and allows you to capture iterations of a model and analyze the effect of the changes in each iteration. The following illustration shows how a SimVision model visually represents a program, how the projects and organizations within the program are depicted, and how the various cases of the model are depicted. 2-18

19 model cases organizations (green tabs) projects (yellow tabs) The hierarchy of programs, cases, projects, and organizations is best illustrated by the Tree pane. For example, in the following illustration the ABC Company Demo program has six cases, each with the same six projects, four program milestones, and two organizations. The case names indicate what factors changed in each case. You can switch between cases, projects, and organizations using the tabs under the Model pane. You can also use the Navigate menu for this purpose. The menu is useful when you have so many projects and organizations in a program that you have to scroll among their tabs under the Model pane. 2-19

20 Examining the Startup Model Launch SimVision. If the default settings are in operation, the Model pane displays a model called the startup model. (Whether the Model pane displays this model, opens empty, or displays the previously opened model depends on a setting in the Options dialog box. See Setting Options for New Models in Help.) The default model consists of a program with Start and Finish milestones and a single default project called Project1. The default program has a title, which you can customize by clicking and retyping, and a default company logo (epm), which you can augment or replace with a custom logo. See Setting Options for New Models in Help. There is a single case, called Baseline. The project has a tab, the Project1 tab beside the Program tab. Click this tab to see the project pane and the default project elements. Examining the Default Project The default project can be empty or it can contain default objects, again depending on a setting in the Options dialog box. If default settings are in operations, the default project looks like the following illustration. 2-20

21 The default project objects are: A logo and title independent of the program logo and title but can be similarly customized. Start and Finish milestones. A position, meeting, and task. Horizontal blue bars used as annotations to visually delineate the areas of the model. For example, in complex models these bars can be very helpful for separating positions, meetings, and tasks. For more information, see Using the Visual Model Delineators in Help. Using the Tree Pane Have a look at the Tree pane at the left. When you expand its contents by clicking the plus signs beside the object names, you can see all the elements in the startup model, as shown here. 2-21

22 The Tree pane's object hierarchy operates similarly to the folder hierarchy in Windows Explorer, with some added features. Click a + (plus) or (minus) sign to expand or contract a group of objects. Click a shape in the Tree pane to select that shape in the Model pane, a really useful feature for complex models. If the program, project, or organization to which that shape belongs is not currently visible in the Model pane, clicking the object in the Tree pane opens the appropriate page in the model pane. For example, if the program page is displayed and you click a position in the Tree pane, the position's project page displays in the Model pane. Right-clicking an element in the Tree pane centers that element in the Model pane. This is another useful way of navigating complex models. For example, if a project has forty tasks spread widely across the Model pane so that scrolling is required to view all the tasks, you can focus on a specific task by right-clicking it in the Tree pane. Reasons for Using SimVision Engineers design products like bridges, airplanes, and semiconductors by building, validating, testing, and refining computations virtual prototypes of these complex systems, rather than by trial and error experiments. The SimVision methodology and software enable project managers to design their work processes and organizations in the same way. Raymond Levitt SimVision is a unique modeling and simulation tool that rigorously integrates organizational and work process views of a strategic, time-critical project or suite of projects. By building, viewing, analyzing, and refining a SimVision model of a large and complex project with aggressive schedules and performance targets, the project management team can: Clarify interdependencies between tasks and roles/responsibilities of positions. Predict backlog of positions due to direct work/coordination/rework, and impact of backlog on schedule and quality. Simulate alternatives for reducing schedule while maintaining quality. There are a number of reasons to build and analyze a SimVision model, for example: To formalize and clarify planned work process and organization for the project. To identify schedule and quality risks in a baseline (most current) plan. To identify alternative management interventions to mitigate risks. To compare duration, cost, and quality tradeoffs for interventions and select the best. The following are six clearly identified reasons for building SimVision models: 1. Pre-project strategic planning 2. Align goals and build consensus 3. Project checkup 4. Capture and refine work process knowledge 5. Evaluate outsourcing 6. Manage portfolios [CLASS BREAK: 15 MINUTES] 2-22

23 Using SimVision With Hands-On Examples Whatever your reason for building a SimVision model, there are five basic steps you will follow to create the model. This section explains the theory behind each of these steps, describes how to carry them out, and takes you through a hands-on example of designing and building a real model using the steps. The five steps are: 1. Define model requirements 2. Build a baseline model 3. Analyze the baseline 4. Evaluate the analysis results, comparing them to the requirements 5. Explore alternative scenarios Step 1: Define Model Requirements Before building a model, it s essential that you spend some time with the client so you can both get a clear understanding of the business objectives and the possible trade-offs among them. Make sure you meet with the person who actually owns the project schedule and risk. Ask them why they are undertaking the project and what benefits they hope to achieve by completing it successfully. There are four practical things you can do with the client in this early planning stage: 1. Define the key business objectives for the project. 2. Define the worst-case values (e.g. time-to-market or cost) that would cause the project to be cancelled. 3. Delineate the trade-offs among project objectives: ask the client questions such as what it s worth to them to have the product reach the market one month or maybe one day earlier? Conversely, what penalties might be incurred if the product is delayed by a month or a day? In this way, you can identify the outcomes that would be unacceptable. This information will be useful when you are suggesting alternate scenarios. 4. Set critical schedule milestones and attach real dates to them. These milestones should be closely related to the business objectives. You can rename SimVision s default Start and Finish milestones to anything you like, and add interim milestones. For example, your program s milestones might end up looking like this: 2-23

24 Hands-On: Defining the Requirements for the Arnold House Model It s October 2005 and the Arnolds have approached you to manage the project of building their house. Their goal is to move into the house by February 10th Their financial advisor has told them that they can afford to spend $200,000 on the house. They want to see if it s possible to build their house on time and within budget. A further time constraint for the Arnolds is that the lease on their current residence expires on February 1st, after which they will be forced to rent on a month-to-month basis, which will increase their rent by 50%. If there is any way the house could be finished by January 31st, the Arnolds would be willing to pay a bonus of $2,000. Conversely, for every week the project slips, you agree to a penalty of $500. The business objectives of the Arnold House program are twofold: Project duration The Arnolds want their house built by February 10th 2008 (the project start date is November 2nd 2007). A $2,000 bonus for finishing by January 31st, and $500 p/week penalty for finishing late are further incentives for meeting this objective. Project cost The objective is to keep the labor cost below $200,000. The Arnolds must balance a potential bonus payment to their contractor against penalty payments received and higher rent incurred. Now that you have identified the business objectives, you can crystallize the information you ve heard from the client into the following form. This form is intended to help you clarify the basic 2-24

25 layout of your model, for example, what it will be called, how many projects the program will contain (in this case, just one), and the planned start and end dates: Program Information Program Name Arnold House Business Objectives Finished by February 10th 2006 Cost of under $200k Program Milestones: 1. Start Planned Date: 11/02/07 2. Finish Planned Date: 2/10/08 Project Names: Build House Step 2: Build a Baseline Model This section provides a roadmap for each step in the process of building a baseline case of your model. The steps have been broken down as follows: 1. Define a project work process plan 2. Create the model and its projects 3. Define the project organization 4. Assign each task to a responsible position 5. Set task work volumes 6. Calibrate a CPM zero-probabilities simulation 7. Refine the CPM zero-probabilities baseline model Step 2. 1 Define a Project Work Process Plan One of the big advantages SimVision has to offer is its ability to graphically represent project work processes and organization. Working with the client to create the graphical project model eliminates a lot of ambiguity and vagueness about responsibilities and relationships among tasks and those responsible for them. Being forced to crystallize this information into a graphical model is an excellent way to clarify what may remain hazy or inconsistently understood among project team members otherwise. Getting down to the nuts and bolts of using SimVision, defining a project work process typically involves 3-5 important milestones, tasks, and positions responsible for those tasks. The suggested strategy for modeling is first to get the model topology that is, the set of milestones, tasks, positions, and relationships among them right. Then, you can add the details to describe the attributes (in SimVision terms, properties) of each task and position. Hands On: Defining the Milestones and Tasks for the Build House Project Having talked with the Arnolds contractor about the phases of building the house and who is responsible for each task, you can now create a form for the project that lists the positions, milestones, and tasks involved in the project, as shown next. These milestones, positions, and tasks will translate directly to tangible objects when you create the model. 2-25

26 Project Information Project Name: Build House Positions: 1. Foreman 6. Electrician Mason 7. Plumber Framing Carpenter Finish Carpenter Painter Milestone 1: Name: Start Date: 11/02/07 Task 1: Grade Lot Task 2: Lay Foundation Milestone 2: Responsible Position: Responsible Position: Name: Foundation Complete Mason Mason Task 1: Frame House Task 2: Frame Roof Task 3: Install Electric Task 4: Install Heating Task 5: Install Plumbing Milestone 3: Responsible Position: Responsible Position: Responsible Position: Responsible Position: Responsible Position: Name: Ready for Drywall Framing Carpenter Framing Carpenter Electrician Electrician Plumber Task 1: Install Drywall Task 2: Paint Responsible Position: Responsible Position: Finish Carpenter Painter 2-26

27 Task 3: Install Appliances Responsible Position: Electrician Milestone 4: Name: Finish Date: 02/10/08 Step 2. 2 Create the Model Creating a model in SimVision is as easy as dragging objects from a toolbar to the model pane. Now that you have identified your program name, its single project, and the milestones and tasks, creating the model will take less than 10 minutes. Tip: At any time during this exercise, you can check your model against a finished version of the model in SimVision Course Models\Arnold House Finished.vpm. Hands On: Creating the Arnold House Model The steps in creating the model are: 1. Launch SimVision. 2. Set the program start date. 3. Set the program end date. 4. Save and name the program. 5. Give the model a title. 6. Create the project. 7. Define project milestones. 8. Define project tasks. 9. Set task precedence. Launch SimVision Launch SimVision by double-clicking the SimVision application icon on your desktop, or clicking Start menu > All Programs > epm > SimVision. The SimVision workspace opens with the program s default project and milestones displayed in the Model pane. 2-27

28 Tree pane Properties pane Model pane Default model title Set the program start date Next, you need to set the start date of the program so that all the dates are calculated from the same date as this tutorial. Otherwise, your dates will all be calculated from the actual date on which you begin this tutorial. You want the house building project to start on 11/02/2007. To set the program start date 1 At the bottom of the Model pane, select the Program tab. The program s properties appear in the Properties pane. 2 In the Properties pane, select the month portion of the Start Date property. 3 Change the month to Change the day to 2. 5 Change the year to Set the program end date It is good practice to specify absolute Planned Dates for all milestones so that when you run a simulation, you can tell at a glance how closely the model meets the planned milestones, because the planned dates are graphically displayed alongside the simulated dates. If you don t specify an absolute planned date, the assumed (Relative) planned date for all milestones is the program start date. You now assign a planned program end date by assigning a planned date to 2-28

29 the program Finish milestones. You assign 2/10/2008 as the planned finish date, since this is the date by which the Arnolds want their house built. To set the program end date 1 In the Model pane, select the program s Finish milestone. The milestone s properties appear in the Properties pane. 2 Under Units for the Planned Date property, make sure Absolute is selected. 3 Under Value, change the date to 2/10/2008. Save and name the program Next, save and name the program. There are three places in the workspace where the program name appears: the Tree pane, the application title bar, and the Properties pane. When you save the program, its new name appears in the Tree pane and the application title bar. You can also name the program by changing the Name property in the Properties pane. Renaming this property means that the correct program name appears in all the simulation charts instead of just Program. You will now save and name the program, and change its Name property. To save and name the program 1 On the File menu, click Save. The Save As dialog box appears. 2 Navigate to the SimVision Course Models folder. 3 Save the program file as Arnold House.vpm. The name Untitled changes to Arnold House in the Tree pane, and the filename Arnold House.vpm displays in the SimVision window title bar. If you expand all nodes of the tree pane, you can see the default elements of the program and its project. named program in Tree pane 2-29

30 4 In the Properties pane, change the value of the Name property from Program to Arnold House. (If the property doesn t say Program, it s because you have the project selected instead of the program. Select the Program tab at the bottom of the Model pane to see the program s properties.) Give the model a title You now give the model a title, replace the existing default title. Tip: You can format the title text by right-clicking it and clicking Properties. In the Properties dialog box, you can change the text s font, line, fill, size, and position. To give the model a title 1 In the Model pane, double-click the default title. The title is highlighted. 2 Type Arnold House and click elsewhere in the model. The title is changed to Arnold House. Create the project Because the Arnold House program involves only a single project, you can simply rename the default project. You would create subsequent projects by adding a project shape to the Program tab in the Model pane. To rename the default project and set its dates 1 On the Program tab of the Model pane, double-click the default project, Project1. The name is highlighted. 2 Type Build House and click elsewhere in the Model pane. The project is renamed Build House. Notice how the project tab below the Model pane reflects the new name too. 2-30

31 renamed project in Model pane renamed project tab Add project milestones The Build House project has Start and Finish milestones. Next, you add the two intervening milestones to the project, Foundation Complete and Ready for Drywall. First, you need to select the tab for the Build House project, because this is where you add the project milestones. Select the Build House tab at the bottom of the Model pane. default position default meeting default milestones and task Build House project tab You see the default title, milestones, task, meeting, and position of the Build House project. The blue horizontal lines are just annotations for visually delineating the three main areas of the model. Because the Arnold House model is so simple, you won t need these lines. You also set the project s start and planned end dates by setting dates for the Start and Finish milestones. This is so that you can see accurate start and planned end dates in the simulation charts for both the program and the project. To add project milestones 1 In the Model pane, select and delete the Title and the horizontal blue model delineators. 2-31

32 2 Select the Finish milestone and drag it as far to the right as you can. When you reach the edge of the Model pane, it keeps scrolling until you reach the edge of the canvas. Position the Finish milestone at the canvas edge as shown. 3 Drag the default task up and out of the way. Don t worry about the black links connecting the task to the milestones, just leave them where they are for now. 4 On the Project Shapes toolbar, click Milestone. 5 Add a milestone one-third of the way between Start and Finish. 6 Double-click the milestone so its name is highlighted. 7 Rename the milestone Foundation Complete and click elsewhere to apply the name. 8 On the Project Shapes toolbar, click Milestone again. 9 Add a second milestone two-thirds of the way between Start and Finish and rename it Ready for Drywall. 10 Click Save. Your model should now look like the following illustration. 2-32

33 To set the project s start and planned end dates 1 Select the Start milestone. 2 In the Properties pane, set the Planned Date property to 11/02/ Select the Finish milestone. 4 Set the Planned Date property to 02/10/2008. Add project tasks The Build House project has eleven tasks. By default, each new project contains one task, Task1. You can move and rename the default task to use it as one of the project tasks, or you can delete it and add the tasks you want in the required positions. In this step, you rename and resize the default task for the Foreman s project-long site supervision task (Unless you assign a task to the Foreman, the cost of this position cannot be included in the project cost). You then add the other ten tasks. Tip: When you need to add multiple copies of the same shape, you can use the Rubber Stamp tool to save time. This tool places multiple copies of the shape you select without your having to reselect the shape tool each time. To add the project tasks 1 Double-click the default task, Task1. 2 Type Site Supervision as the new task name. 3 Select the task and drag one end of it so it stretches from the Start to the Finish milestone. Reposition the meeting and position as shown. 4 On the Model toolbar, click Rubber Stamp. 5 On the Project Shapes toolbar, click Task. 2-33

34 6 Place ten tasks as shown next. 7 Press esc to deactivate the Rubber Stamp tool. 8 Rename the tasks as shown in the illustration: Grade Lot, Lay Foundation, Frame House, Frame Roof, Install Electric, Install Heating, Install Plumbing, Install Drywall, Paint, and Install Appliances. Setting task precedence Next you define the task precedence by linking the tasks to each other and to milestones with successor links. In each of these shape-to-shape relationships, one shape is the predecessor and one is the successor. Therefore, except for the Start and Finish milestones, every task and milestone must have both a predecessor and a successor. The Start milestone only has a successor; the Finish milestone only has a predecessor. Successor links can be one of three types: Finish-Start The successor task start when the predecessor task finishes, or a defined time (or lag) later. This is the default type of link. Start-Start The successor task starts at the same time as, or a defined time after, the predecessor task starts. You don t use these links in this tutorial. Finish-Finish Used for supervisory tasks. You will use this link type for the Site Supervision task. When you have added all the links, your model should look like the following illustration. 2-34

35 To add the Finish-Start Successor links 1 Right-click the Start milestone. 2 On the right-click menu, click New Successor To. The New Successor To dialog box lists the project s tasks and milestones. Notice that the default link type is Finish-Start. 3 Select Grade Lot, and click OK. A successor link joins the milestone to the task. 4 Add the remaining successor links between the tasks and milestones as described in the following table. 5 Click Save. Link this shape... To this shape... Grade Lot Lay Foundation 2-35

36 Lay Foundation Foundation Complete (milestone) Frame House Frame House Frame House Frame House Frame Roof Install Electric Install Heating Install Plumbing Ready for Drywall (milestone) Install Drywall Install Drywall Paint Install Applications Foundation Complete (milestone) Frame House Frame Roof Install Electric Install Heating Install Plumbing Ready for Drywall (milestone) Ready for Drywall (milestone) Ready for Drywall (milestone) Ready for Drywall (milestone) Install Drywall Paint Install Appliances Install Appliances Finish milestone Making a Task into a Supervisory Task The Foreman is assigned the Site Supervision tasks that is designed to last the duration of the project and cover all his supervisory duties. Rather than trying to give this task a work volume that matches project duration, another way to model the situation is to make the task a Supervisory task that finishes at the same time as the last task in the project. Supervisory tasks are measured in FTE resources rather than in units of time. You make a task into a Supervisory task by adding a Finish-Finish Successor link between the last task (Install Appliances) and the end of the Site Supervision task. You then set the task s Work Type to Supervisory and make the FTE value 1. To make the Site Supervision task into a Supervisory task 1 Add a Successor link from the right side of the Install Appliances task to the right side of the Site Supervision task. 2-36

37 2 With the link still selected, change the value of its Type property to Finish-Finish. 3 Select the Site Supervision task. 4 In the Properties pane, click in the Work Volume property so that a drop-down list of work types appears. 5 Select Supervisory from the list and click elsewhere in the Properties pane to apply this value. The units change from Days to FTEs. 6 Enter 1 for the FTEs value. 7 Click Save. Step 2.3: Define the Project Organization SimVision models a project s participants, their skill and experience levels, their reporting relationships, and the organization s exception handling and communication policies in rich detail. This unique and powerful capability has been made extremely easy to use. You model a project s participants using positions. A single position can represent multiple FTEs (full-time equivalents). You add positions to the model as shown here: positions Guidelines for defining the organization: Define about 10 positions. A position can represent either a single individual or a group of people with similar or different skills who are collectively responsible for one or more tasks. Arrange the positions in an exception-handling hierarchy of supervisors and their subordinates. This hierarchy is intended to represent where each position goes when they have a problem, rather than the formal hierarchy of the organization; so the question to ask when creating the hierarchy is Who asks whom for help when 2-37

38 problems arise? By answering this question, you can link positions with Supervision links to show the exception-handling hierarchy. Assign a role to each position. Roles can be PM (Project Manager), SL (Subteam Leader), or ST (Subteam member). There is usually only one PM. Use the ST role for all positions at the bottom of the exception-handling hierarchy who actually do the work. Use SL for all positions in between. Set each position s capacity. Position capacity is measured in FTEs (full-time equivalents). Thus, a position could represent 2 FTEs, or maybe 1.5 FTEs (one fulltime person and one half-time). If a single person is dedicated full-time to the project, set the number of FTEs for the position to 1.0. If the person is sharing time with other projects or administrative duties, set the FTE value to some fraction of 1.0. If a group of individuals is assigned full-time or part-time to a subteam for this project, add their contributions and set the FTE value for the position to the sum of their available time for this project. For a large subteam (>5), you might want to break out a subteam leader as a separate person to whom the others report. This better represents exception-handling in the group and more accurately assigns just the team leader to attend meetings, where assigning the whole subteam is inappropriate. The persons representing a position are assigned when you staff the position. Persons are not graphically represented in a model, but are added as members of departments and subdepartments. You ll learn about this later. Hands On: Defining the Build House Project Personnel Next, you add the positions, set position properties, then create the exception-handling hierarchy. 2-38

39 To place the positions 1 Drag the default position, Position1, until it is directly above the center of the Frame House task. 2 On the Model toolbar, click Rubber Stamp. 3 On the Project Shapes toolbar, click Position. 4 Place six new positions in the Model pane in roughly the following locations. 5 Press esc to deactivate the Rubber Stamp tool. 6 Rename the positions Foreman, Mason, Framing Carpenter, Finish Carpenter, Electrician, Plumber, and Painter as shown. You double-click a position to rename it. To force a name to go on two lines instead of waiting for it to wrap, press ENTER. 7 Click Save. Setting Position Properties Like all other objects, positions have a set of properties. You work with five of these properties: Name A short label for the position. Names can contain any combination of text, numbers, and punctuation. Role The position s function in the organizational hierarchy. Each position fulfils one of three functions: Subteam (ST), Subteam Leader (SL), or Project Manager (PM). Project Manager is the position that tends to make many of the decisions, and to whom others usually go with questions. Typically only one person per project has this role. Subteams generally do most of the work. Positions between PM and ST are usually subteam leaders. In this tutorial, you use only PM and ST roles. Application Experience How experienced the position is with this type of project. FTE The number of full-time equivalent persons represented by the position. Salary The position s hourly wage, measured in currency units. You assign all positions a salary of $50 per hour apart from the Foreman, whose cost you will model later. To set position properties 1 Select the Foreman position. 2-39

40 The position s properties appear in the Properties pane. 2 Change the Foreman s role to Project Manager by clicking under Value for the Role property and selecting PM from the list. 3 Change the Foreman s Application Experience value to High by clicking under Value and selecting High from the list. 4 Select the Framing Carpenter position and change the FTE value to 2 by clicking under Value and entering 2. 5 Select the Painter position and change the FTE value to 2. 6 Select each position in turn, apart from the Foreman, and change the position s Salary property to 50. Change the Foreman s salary to Leave all the other positions with default properties and click Save. Creating the exception-handling hierarchy Supervision links connect supervisor positions with the positions they supervise. The supervision relationship in SimVision refers solely to the direction in which information is passed. It has nothing to do with the rank of the positions within the organization. Next, you add Supervision links so that the Foreman position is connected to all the other positions as follows: To connect positions with Supervision links 1 Right-click the Foreman position. The right-click menu lists the types of links you can add to a position. 2 Click New Supervisor Of. A list of the project s positions appears. 3 Select Mason and click OK. 2-40

41 4 Add Supervision links between the Foreman and the other five positions, as shown in the previous illustration. 5 Click Save. Step 2.4: Assign Each Task to a Responsible Position The next step is to connect your organizational model to your work process model. You do this by: Assigning each task to a single position. Setting the number of position FTEs to be assigned to the task. Adjusting the level of detail of positions and tasks to achieve a fit between them. Assigning tasks to positions Each task must have just one responsible position. This rule helps to squeeze out ambiguity in a project team that contains multiple positions engaged in concurrent tasks. It forces the modeling team to define holistic tasks that are manageable, and to assign a single actor who can be held strictly accountable for the holistic task. This is especially valuable when building a model collaboratively with several members of a project team. Note: If you accidentally assign more than one responsible position to a task, the simulator will detect the error on simulation and you will be required to fix it before continuing. To assign a task to a position, you drag a Primary Assignment link into the model and attach its source point to the position and its destination point (arrow) to the task. Setting the number of FTEs to be assigned to the task When you create an assignment relationship, the default number of FTEs assigned from the position to the task is 1.0. Decide how many of the position s FTEs (less than or equal to the total FTE capacity of the position) to assign to each task. You assign the FTEs to the Primary Assignment link that links the position to the task. For positions with significant numbers of subordinates reporting to them, assigning all of the available FTEs to any task will cause the position to become backlogged. Answering subordinates questions takes time, so managerial positions need to be given some slack capacity to deal with their subordinates. As a rough rule of thumb, allow FTE of slack capacity for each subordinate when allocating subteam leaders or project managers to direct work activities. Adjusting the level of detail of positions and tasks Take the time needed to get task and position granularity right. If a task feels like it should have two responsible positions, it s too large in scope. Break it up into two or more tasks for a single position at the level positions are modeled in the organization. Alternatively, consider collapsing the positions into a more aggregate position, if the aggregate position can be assigned to other activities in the model. Coming up with the right level of detail generally takes a few iterations. It is a core SimVision modeling skill that improves with practice. Some guidelines are: Define the organization before you define the work process. Create the intermediate milestones before defining any tasks. Then list 5-10 tasks that accomplish the goal of each milestone. This helps the client s technical team to focus on tasks that are relevant to achieving project milestones at about the right level of detail. Err on the side of too little detail at first: it s much easier to add detail than take it out. 2-41

42 Hands On: Assigning Tasks to Positions in the Build House Project Now you assign tasks to the positions. Each task must have a single primary responsible position. More than one position can work on a task, but only one position is primarily responsible for it. Tasks can only have a single primary responsible position, but positions can have multiple primary tasks. Tasks can have any number of secondary responsible positions. Positions are available to work on secondary tasks only when they are not working on primary tasks. If the position is assigned to two concurrent primary tasks, the position is scheduled to work on both tasks each day. The first task in the queue gets done, then the other, day-by-day. Backlog mounts if the position does not have sufficient FTEs to keep up with the work. In this exercise, you assign only primary tasks to the positions. When you have assigned the tasks to positions, your model will look like this (note that the Foreman was already connected to the Site Supervision task with the default Primary Assignment link because that was the default task): To add the primary assignment links 1 Right-click the Mason position. 2 On the right-click menu, click New Assignment To. 3 In the New Assignment From dialog, select the Grade Lot task, and click OK. 4 Add the remaining links described in the table that follows this procedure. Notice how, as you assign tasks, they are removed from the New Assignment From dialog box because each task can have only a single primary responsible position. 5 Click Save. 2-42

43 You add the following primary assignment links: Link from position To tasks Mason Framing Carpenter Finish Carpenter Electrician Plumber Painter Grade Lot, Lay Foundation Frame House Frame Roof, Install Drywall Install Electric, Install Heating, Install Appliances Install Plumbing Paint Step 2.5: Set Task Work Volumes The most important property of a task is an estimate of its work volume. Tasks fall into two types: Work volume tasks: For these tasks, the total work quantity is fixed so that the task s duration varies in inverse proportion to the number of FTEs assigned. Work duration tasks: For these tasks, duration is intrinsically or externally determined by factors other than the number of FTEs assigned to them. For example, a diagnosis of concrete strength may require that the test concrete cylinder cure for 14 days before testing; or a drug clinical trial protocol may require that patients in the study be administered a specified dose of the drug daily for six months. Adding extra civil engineers to the former task or study nurses to the latter will not speed up the task. Fixed duration tasks are impacted by coordination, rework, and resource constraints. Thus, like any other task, their simulated duration is not fixed. To force a fixed time period between one task and another, you can define a task that should take a fixed duration as a very short task (e.g. 1 day), then give an appropriate Start-to-Finish lag to each of the successor tasks. The simulator will start the successor tasks after the fixed time lag. Estimating amount of work for work volume tasks The question you should ask when trying to assess the work volume for work volume tasks is: How many FTE days of cumulative effort will it take for an individual or subteam with a typical level of experience on this type of task, and with the appropriate skill set, to complete the task, assuming they get it right the first time? Note: Most tasks cannot be broken down beyond a certain point, that is, there is a practical upper and lower limit to the number of responsible positions that can meaningfully contribute to the task. The domain expert must decide whether the number of FTEs allocated to the task remains in the range for which the work volume remains constant that is, for which the assumption of the inverse linear relationship between allocated FTEs and task duration applies. There are two possible ways to estimate work volume for these tasks: Directly from historical data: For example, a design task requires the production of 20 layout drawings, each of which takes one person working full-time about two days. Total work volume = 20*2 = 40 FTE days. Historical unit-productivity rate data for lowlevel work products like drawings or machined parts support this productivity factor method of estimating work volume. 2-43

44 From historical or forecasted task duration and crew size: For example, on comparable projects in the past, a subteam of three engineers took 20 working days to complete this activity. Total work volume = 3*20 = 60 FTE days. Historical unit productivity rate data for higher-level tasks like designing or testing components support this method of estimating work volume. Estimating amount of work for work duration tasks For work duration tasks, you should apply the task s fixed duration to the task. The work volume is then computed by multiplying this duration by the number of assigned FTEs. To define work volume for work duration tasks, you click the arrow beside FTE Volume in the Task Properties pane and select Work Duration. Then enter the task duration in working days (not counting weekends) under Value. When you assign the appropriate number of FTEs from the appropriate position to the task by setting the value of the Primary Assignment link s Allocation property, SimVision assigns that number of FTEs from the responsible position to work on the task from the time it starts until it is completed. Note: SimVision assumes that the fixed duration you provide is in working days and calculates the task s calendar duration by scaling from work days to work weeks using the setting of the project s Work Week field. The default project work week is 5 work days of 8 hours each. Thus, if the Test Concrete Strength task requires 28 calendar days for the concrete to cure and the work week is 5 days, you should set the fixed duration of this task at 20 FTE days, since this value is in working days. Hands On: Setting Task Work Volumes in the Build House Project You now set the amount of work you predict each task will require. All the tasks in the Build House project are work volume tasks. Task costs are calculated by multiplying the work volume by the salary of the assigned position. There are various ways you can express work volume (person hours, weeks, months), but for this exercise you express it in terms of days of work. To set task work volumes 1 In the Model pane, select the Grade Lot task. 2-44

45 The task s properties appear in the Properties pane. 2 Change the Work Volume value to 4 days. 3 Select each task in turn and change its Work Volume value as indicated in the table that follows this procedure. 4 Click Save. Task Work Volume Grade Lot 10 days Lay Foundation 20 days Frame House 20 days Frame Roof 12 days Install Electric 8 days Install Heating 7 days Install Plumbing 8 days Install Drywall 20 days Paint 12 days Install Appliances 4 days 2-45

46 Step 2.6: Model Task Interdependencies In fast-track projects, many highly interdependent tasks tend to be executed in parallel. SimVision models task interdependencies in two ways: Information dependency When there is significant interdependence between the decisions that are made in two tasks, their responsible positions need to communicate to make sure that the choices made for each task are compatible with one another. Linking the two tasks with an information exchange relationship, modeled in SimVision by a green communications link, captures this type of dependency. All tasks have some need for information exchange, but we only add communications links between those that have significant functional interdependence. Typically 25-30% of task pairs need this kind of information exchange. Failure dependency For any complex product or system, there is a chain of functional cause and effect such that changes in one task will trigger rework in other tasks that have already been fully or partially completed. For example, suppose a process plant refines gasoline from crude oil. The quantity and quality of crude oil to be processed results in the section and sizing of the catalytic cracker. Any change to the size or type of catalytic cracker might trigger corresponding changes in the design of pipelines, pumps, motors, electricity supply, and compressed air supply. In SimVision, you can model this kind of task interdependence by inserting a red rework link between the driver task that determines the high-level functional requirements and the dependent task. Unlike communications links, rework links are directional. Rework flows from the driver to the dependent task, whereas information is exchanged between task pairs linked by a communications link. Hands On: Modeling Task Interdependencies in the Build House Project Adding a Rework link In the Build House project, a problem installing appliances in the house could necessitate some rework in the wiring system. To model this, you now add a Rework link from the Install Appliances task back to the Install Electric task. The Rework link has an arrow at the Install Electric (dependent) task, showing that the link is directional and rework only flows one way. To add a Rework link 1 On the Project Shapes toolbar, click Rework. 2 Click the bottom target of the Install Appliances task. 3 Click the bottom target of the Install Electric task. A Rework link is added with the arrow at the Install Electric task. 2-46

47 Adding a Communications link Communications links between tasks model the fact that the positions responsible for the tasks must communicate with each other throughout the tasks. For example, the plumber and the electrician need to communicate about which rooms they are installing plumbing and heating in so as to keep out of each other s way, since their installation tasks are concurrent. You now add a Communications link between the Install Heating and Install Plumbing tasks. Since communication needs to flow both ways, you will notice that the Communications link has an arrow at both ends, unlike the Rework link, which is directional. To add a Communications link 1 On the Project Shapes toolbar, click Communications. 2 Click the bottom target on the Install Heating task. 3 Click the top target on the Install Plumbing task. A Communications link is drawn between the two tasks. Note that there are arrows on both ends of the link, indicating that communication is two-way. Step 2.7: Set Probabilities Probabilities refer to the probability that, over the course of the project, indirect work will be generated over and above the direct work modeled by the project s tasks. Indirect work is characterized by things like the need for information exchange, rework arising from errors, and common noise such as everyday conversations, phone calls that interrupt the flow of work, and unforeseen circumstances such as a fire drill. SimVision models four kinds of probability: Information Exchange Probability Measures the level of communication in the project between positions who are responsible for tasks linked by communications links. This probability is typically set in the range 0.2 to 0.5 low if the project involves a high level of routine jobs performed by skilled workers, and high if the project involves many highly interdependent tasks that are being performed by less skilled or very busy workers. You will set the probability to 0.3. Noise Probability Measures the effect of interruptions in the ordinary working day that take time away from doing the project tasks. In any real organization, noise can include distractions like a salesperson calling to sell insurance or a request for help from a peer. The probability of noise is generally in the range 0.01 (low) to 0.10 (significant, but common). You will set the noise probability to 0.1. Functional Error Probability The probability that a task will fail and require rework. Functional errors are errors that are localized to a task and cause rework only in that task by the responsible position. Such errors might be discovered by a self-check procedure, a peer review, or a supervisor s review. When a functional error is detected, an exception is sent to the responsible position. The responsible position can choose to rework, do a quick fix or, or ignore the exception. This probability is typically set in the range 0.05 to 0.10 low if the project involves relatively well-understood technology and standard work processes, and high if the project involves unproven technology or innovative work processes. You will set it to

48 Project Error Probability The probability that a task will fail and generate rework for itself and all failure-dependent tasks, which are tasks connected to it by rework links. This probability is typically set in the range 0.05 to 0.10 (significant but common) low if the project involves relatively standard tasks and routine work processes, and high for nonstandard tasks and innovative work processes. You will set a project error probability of 0.1. When a project error is detected, an exception is sent to the position responsible for the failed task. As for functional exceptions, the responsible position can rework, quick-fix, or ignore. By default, new models in SimVision have their probabilities set to zero. You can see this if you click in any blank space in the program page and look at the program s properties in the Properties pane. However, each time a position completes a subtask (equal to 5% of the task s work volume), the simulator rolls the dice to determine whether a functional error, a project error, an information exchange, or a noise item should be generated. You need to set the probability rates so that the generated items actually have an effect in the simulation data. Furthermore, the rework and communications links that you added between interdependent tasks have no effect unless the probability values are set. Hands On: Setting Probabilities in the Arnold House Model To set the probabilities 1 In the Model pane, make sure the Program tab is selected so that you set the probability rates for the program and not the project. The program s properties appear in the Properties pane, with the probabilities still set to zero. 2 Change the value of the Info Exchange Prob. property to

49 3 Change the value of the Noise Prob. property to Change the value of the Functional Error Prob. property to Change the value of the Project Error Prob. property to Click Save. Step 2.8: Add Skills to Tasks and Positions SimVision s model editor assigns each position a medium level of the default Generic skill, and each task a Generic skill requirement. With these default settings, all tasks are executed at a nominal rate (that is, it takes a position with a capacity of one FTE exactly one day to complete one FTE-day of work volume). In practice, positions are often assigned to perform tasks for which they have less than ideal skills and experience. To model this fact of project life, SimVision provides a sophisticated human-resource modeling capability in which a precise skill set and level of Application Experience can be specified for each position. Here are some guidelines to think about when creating skills for a project. A skill can be: A broad professional skill such as Accounting or Structural Engineering. A narrow professional skill such as Pediatric Cardiology. A trade skill such as Finish Carpentry or Semiconductor Manufacturing Equipment Repair. A managerial or social skill such as Project Coordination or Conflict Resolution. Feel free to use skills that lie outside of normal educational or professional specialties as needed. For example, a SimVision model once featured Local Politics as a skill required by the responsible position for a task named Obtain Zoning Approval. When assigning skills to tasks, you must choose the principal or most important skill required to execute that activity. However, a versatile, broadly skilled position can have multiple skills, each at different levels, which allow them to successfully perform a wide range of tasks with different skill requirements. You can set skills at three levels: Low, Medium, and High. Thus, a position in a logic design team for a custom semiconductor might have a High level of the Logic Design skill and a Low level of the Floor Planning skill. Assigning this position to a task for which Logic Design is the required skill (bearing in mind that each task can have only one required skill) will result in an above average performance. If the position has to work on a task that requires Floor Planning as its principal skill, they will progress more slowly and make more errors than a position with a Medium level of that skill. Note: You can assign a position to a task when the position doesn t have the appropriate skill at all either deliberately or in error. If this occurs, the position will make some progress with the task, but at a significantly slower rate than a position with the required skill. If you notice that a task is taking much longer after you assign skills, check to see whether there s a skill assignment error, also called a skill mismatch between task and assigned position. Setting Application Experience You can further model skills by assigning each position an Application Experience value. This property of positions models the fact that a person can have a high level of the skill required by their assigned task but lack experience working on the type of application represented by the project. For example, an electrical engineer may have a Ph.D. in electrical engineering and many years of experience designing generating systems for power plants, but may never have worked on designing power supplies for consumer electronic products. Such an engineer might have a high Power Engineering skill but low application experience for her task on the current project, which is to design the power supply for a portable television. You can adjust the Application Experience property of positions to either Low, Medium, or High. Application experience works together with the match between required skill and skill set to 2-49

50 determine the processing speed and error rate of tasks. For example, a position with a medium level of the task s required skill but high Application Experience will make slightly faster progress and with slightly fewer errors than a nominally skilled position. Note: You don t set Application Experience values in the hands-on exercise in this class. To assign skills, start by defining the skill requirements for the project s tasks. You do this by building a list of skills that are needed throughout the project. Once you have built the project s list of skills, you can apply them one by one to tasks by selecting each task in turn and selecting a skill from the list in the task s Skills property. To assign skills to positions, you select the position and open the Skill Set dialog box, where you can apply skills and skill levels to the position, as shown: Hands On: Adding Task and Position Skills to the Build House Project First, build the skill set for the Build House project. 2-50

51 To build the project s skill set: 1 On the Model toolbar, click Skills. 2 In the Skill Set dialog box, click Add. 3 Click Add. 4 On the new line that appears, enter Masonry for the skill name. 5 Create the skills Framing, Finish carpentry, Electrical, Plumbing, and Painting. 6 Click OK. Next, apply required skills to the project s tasks and assign skills to its position. 1 On the project page of the model, select the first task, Grade Lot. 2 In the Properties pane, click in the Value field for the Skills property. The list of skills appears. 4 Select Masonry from the list. 5 Apply the appropriate skills to the tasks, according to the assigned position. 6 Select the Mason position. 7 In the Properties pane, click the Edit button for the Skill Set property. 8 In the Skill Set dialog box, select the Masonry skill and click OK. 2-51

52 9 Apply the appropriate skills to the remaining positions, all at Medium level. Step 2.10: Add Meetings Meetings are very much a part of project life. You can model them using SimVision, and the simulation takes account of both their positive and negative sides the coordination and decision making that lower error rates, and the time-sink for busy team members. If some participants miss a meeting, the value of the meeting is reduced both for those who missed it and for others who cannot properly coordinate with them. Thus, project and functional error rates increase on subsequent work for participants who miss a meeting. Setting up meetings in SimVision is as simple as dragging a meeting object onto the model and linking it to the appropriate positions with Meeting Participant links. Normally, only a single representative from a position with multiple FTEs attends a meeting. SimVision models this by setting the default number of FTEs on the meeting participant link to 1.0. If the entire subteam is supposed to attend a meeting, then you should adjust the Meeting Participant link s Allocation property accordingly. Conversely, if a single representative will attend for only a portion of the meeting, you can reduce the allocation percentage of the link. Hands On: Adding Meetings to the Build House Project The Build House project has two meetings, so you use the model s default meeting and add a second: Morning Meeting Attended by the Foreman, the Finish Carpenter, the Painter, the Electrician, and the Plumber every week. Site Planning Meeting Attended by the Foreman, the Mason, and the Framing Carpenter every week. To add the meetings 1 Rename the default meeting by double-clicking it and changing the name to Morning Meeting. If necessary, resize the meeting shape to accommodate its name by dragging any of the shape s handles. 2 The Foreman is already a participant in this meeting, as shown by the grey dashed link. To add the Finish Carpenter as a participant, right-click the Finish Carpenter position. 3 On the right-click menu, click New GoTo Meeting To. 4 In the New GoTo Meeting To dialog box, select Site Planning Meeting and click OK. The Finish Carpenter is joined to the meeting with a Meeting Participant link. 2-52

53 5 Join the Painter, Electrician, and Plumber to the meeting in the same way. 6 On the Project Shapes toolbar, click Meeting. 7 Add a second meeting above the Start milestone, renaming it Site Planning Meeting. 8 Assign the Foreman, Mason, and Framing Carpenter as participants, using the positions right-click menus. 9 Select the meeting. The meeting s properties appear in the Properties pane. 10 To ensure this meeting occurs at a different time from the Morning meeting, change the Start Time to 9a.m. by clicking in the Value field for that property, selecting the hour, and clicking the up-arrow. Your model should look like the following illustration. 2-53

54 Step 2.11: Add Cost Data SimVision tracks and provides detailed breakdowns of the costs and revenues associated with a project. Tracking Project Costs Costs are broken down into labor costs that is, salaries and nonlabor costs incurred by milestones or tasks, such as fees, penalties, materials, or payments. Both kinds of costs are expressed in generic currency units. Labor costs are calculated as the salaries of positions and persons multiplied by the work volume of the tasks they are assigned. Work, rework, coordination, and decision costs are also considered part of labor costs. You identify nonlabor costs in three ways: Fixed cost A fixed costs is a one-time cost associated with the project. For example, a hardware installation project might have as a fixed cost the price of the hardware. Cost rate The cost rate represents cumulative costs per specified unit of time. For example, a building construction project might accumulate equipment rental fees over a period of months. Thus, the cost rate of the project might be 10,000 per month for these fees. Actual cost A placeholder value for comparing actual to projected costs for the project. This value is ignored by the simulator but appears in the Project Statistics chart, where you can compare it to the Total Cost value for the project. Tracking Project Revenues You identify project revenues in three ways: Fixed revenue Fixed revenue is the revenue that projects, tasks, and milestones generate. For example, a housing development project might generate 10 million in revenue when the tract is complete. 2-54

55 Revenue rate Revenue rates are associated with projects and tasks and represent cumulative revenues per specified unit of time. For example, an oil drilling project might generate revenue for every day that oil is produced. Thus, the revenue rate might be 50,000 currency units per day. Actual Revenue A placeholder value for comparing actual to projected revenue for the project. This value is ignored by the simulator but appears in the Project Statistics chart, where you can compare it to the Total Revenue value for the project. Hands On: Adding the Cost Data to the Build House Project In the Build House project, you have already modeled labor costs by assigning salaries to the positions (via their Salary properties) and work volumes to their tasks. Labor costs are calculated as follows: Task cost = (assigned position s hourly salary) x (task s work volume). Nonlabor costs cover such things as materials, machine rentals, and one-time payments such as permits. For this exercise, you simply model nonlabor costs as the $100,000 of materials needed to build the house. To do this, you set the project s Fixed Cost property value to 100,000. To model the nonlabor cost of materials 1 Click the Program tab below the Model pane. 2 In the program model, select the Build House project. 3 In the Properties pane, enter for the Fixed Cost property s value. Do not use a comma in the amount. Step 3: Simulate the Baseline Model You are now ready to simulate the baseline case of your model. When you run the first simulation, look at the project s Gantt chart output. Since you have nonzero probabilities, you should expect the SimVision simulated durations for tasks to be somewhat longer than the CPM durations. Start by looking at the final project milestone and notice how much later it occurs than the CPM planned time for the milestone. Is this date acceptable, given the project s objectives, or is it beyond the project s required completion time? Frequently, the simulated end date will be unacceptably late, so your remaining efforts need to focus on shortening the project schedule without unduly compromising quality and cost objectives. If the baseline Gantt chart simulation output indicates that completion occurs on time, this is good news, and you should focus further evaluation on reducing costs and mitigating any identified risks. Next, you should look at the durations of individual tasks on the project s Gantt chart and compare them to their CPM-planned durations. Because of the time taken up by coordination, rework, wait time, and meetings, you should expect there to be some tasks at risk of taking longer than planned. These tasks will show up red, rather than blue. Your next step will be to do further diagnostic analysis on the simulation results. Hands On: Simulating and Analyzing the Baseline Arnold House Model Simulating the program is as easy as pressing a button. The simulator displays data for the program first by default. After you view the program data, you can view the project data. Note: Your results may differ slightly from those shown in this exercise. This is normal variance. 2-55

56 To simulate the baseline case 1 On the Results menu, click Run Simulation. The simulator runs its default 25 trials. If there are warnings, these are displayed in a dialog box and you can click Proceed to continue with the simulation. If there are errors, the simulation is canceled and you must fix the errors before proceeding. There should be neither errors nor warnings for the initial simulation. If there are, see Troubleshooting Simulation in Appendix B of the User Guide. When the simulation is complete, the Chart Window appears, showing the Program Gantt chart. The Gantt chart s data side is displayed. Viewing the End Dates Because the Arnold House program has only one project, the program and project end dates are the same. You can examine either the Program Gantt or Project Gantt chart. For this exercise, you examine the Program Gantt chart and compare the planned end date to the simulated and CPM end dates. CPM data is off by default, so you start by turning it on. To view the program end dates 1 On the Controls menu, click Show CPM. The CPM data appears. 2-56

57 click here 2 The minus sign (-) beside the Build House project means the project is expanded so you can see its tasks. This is a lot of data. You can just view the program duration and end dates by contracting the project. To do this, double-click the minus sign, then click in the header of the Task column to deselect the Build House row. The project contracts. Scroll to the right so you can see the program Start and Finish milestones on the graphical side of the chart. planned end date CPM end date predicted end date The planned, CPM, and simulated end dates are quite close together, as they should be for the Baseline case. 3 Double-click the Build House project again to expand it. 4 Drag the horizontal scrollbar to the left and the vertical scrollbar down until you can see the Sim. Date, Planned Date, and CPM Date columns with the Finish dates as shown. 2-57

58 You can see that the simulated end date is 03/17/08, about five weeks later than the planned date, and the CPM date is 02/29/08. 5 Move the horizontal scrollbar to the right to view the expanded graphical data. Here you can see cyan dashed lines for meetings, dark red and blue bars for simulated durations, and light red and blue bars for CPM durations. The red bars show critical path tasks and the blue bars noncritical path tasks. What to Do Next In the next class, you will evaluate the baseline simulation results and then explore alternative scenarios for meeting the Arnold s objectives. Background Reading For the next class, read the following papers: Design your Project Organization as Engineers Design Bridges, Levitt & Kunz The Virtual Design Team: A Computational Model of Project Organizations, Jin & Levitt 2-58