Importance of Project Management

Projects represent change and allow organizations to effectively introduce new products, new process, new programs Project management offers a means for dealing with dramatically reduced product cycle times Projects are becoming globalized making them more difficult to manage without a formal methodology

Project management helps cross-functional teams to be more effective

Management of IT Projects
More than $250 billion is spent in the US each year on approximately 175,000 information technology projects. Only 26 percent of these projects are completed on time and within budget.

The average cost for a development project for a large company is more than $2 million.
Project management is an $850 million industry and is expected to grow by as much as 20 percent per year.
Bounds, Gene. The Last Word on Project Management IIE Solutions, November, 1998.

What Defines a Project?

How does a project differ from a program?

Project Management versus Process Management

Ultimately, the parallels between process and project management give way to a fundamental difference: process management seeks to eliminate variability whereas project management must accept variability because each project is unique.
Elton, J. & J. Roe. Bringing Discipline to Project Management Harvard Business Review, March-April, 1998.

Measures of Project Success

Was the movie Titanic a success?

Delayed Openings are a Fact of Life in the Foodservice, Hospitality Industry

Disney's shipbuilder was six months late in delivering its new cruise ships, and thousands of customers who had purchased tickets were stranded. Even with that experience, their second ship was also delivered well after the published schedules. Universal Studios in Orlando, Fla. had been building a new restaurant and entertainment complex for more than two years. They advertised a December opening, only to announce in late November that it would be two or three months late. Even when facilities do open close to schedule, they are rarely finished completely and are often missing key components. Why do those things happen? With all of the sophisticated computers and project management software, why aren't projects completed on schedule?
Frable, F. Nation's Restaurant News (April 12, 1999)

IT Project Outcomes
101-200% late
16%

More than 200% late

6%

51-100% late

9%

29%

Cancelled

21-50% late

8% 6%

Less than 20% late

26%

On-Time

Source: Standish Group Survey, 1999 (from a survey of 800 business systems projects)

Why do Projects Fail?

Studies have shown that the following factors contribute significantly to project failure:
Improper focus of the project management system Fixation on first estimates Wrong level of detail

Lack of understanding about project management tools; too much reliance on project management software
Too many people Poor communication

Rewarding the wrong actions

Why do IT Projects Fail?

Ill-defined or changing requirements

Poor project planning/management

Uncontrolled quality problems

Unrealistic expectations/inaccurate estimates

Naive adoption of new technology
Source: S. McConnell, Construx Software Builders, Inc.

Have You Ever Lost Sight of the Project Goals?

QuickTime an d a Photo - JPEG decompressor are need ed to see this picture.

Not all Projects Are Alike

[in IT projects], if you ask people whats done and what remains to be done there is nothing to see. In an IT project, you go from zero to 100 percent in the last second--unlike building a brick wall where you can see when youre halfway done. Weve moved from physical to non-physical deliverables. J. Vowler (March, 2001)

Engineering projects = task-centric IT projects = resource-centric

Shenhars Taxonomy of Project Types

Degree of Uncertainty/Risk
Super HighTech ERP implementation in multi-national firm New shrinkwrapped software New cellphone Construction Assembly Projects Auto repair System Projects Array Projects High Advanced radar system

HighTech

MediumTech LowTech

System Complexity/Scope

Project Life Cycle

Required Resources

Phase 1
Formation & Selection

Phase 2
Planning

Phase 3
Scheduling & Control

Phase 4
Evaluation & Termination

Time

Life Cycle Models: Pure Waterfall

Concept Design Requirements Analysis Architecture Design

Detailed Design
Coding & Debugging System Testing

Source: S. McConnell Rapid Development (Microsoft Press, 1996)

Life Cycle Models: Code & Fix

Design, Cost, Time Trade-offs

DESIGN
Required Performance

Target
Budget Constraint

COST

Due Date

Optimal Time-Cost Trade-off

Optional Scope Contracts

Since it is widely accepted that you can select three of the four dimensions (or perhaps only two), what to do?
Fixed Scope Contract specifies SCHEDULE, COST, SCOPE

Optional Scope Contract

specifies

SCHEDULE, COST, QUALITY

(general design guidelines may be indicated)

Importance of Project Selection

There are two ways for a business to succeed at new products: doing projects right, and doing the right projects.
Cooper, R.G., S. Edgett, & E. Kleinschmidt.
Research Technology Management, March-April, 2000.

Project Initiation & Selection

Critical factors
1) Competitive necessity 2) Market expansion 3) Operating requirement

Numerical Methods
1) 2) 3) 4) Payback period Net present value (NPV) or Discounted Cash Flow (DCF) Internal rate of return (IRR) Expected commercial value (ECV)

Project Portfolio
1) Diversify portfolio to minimize risk 2) Cash flow considerations 3) Resource constraints

Payback Period
Number of years needed for project to repay its initial fixed investment

Example: Project costs $100,000 and is expected to save company $20,000 per year Payback Period = $100,000 / $20,000 = 5 years

Net Present Value (NPV) Discounted Cash Flow (DCF)

Let Ft = net cash flow in period t (t = 0, 1,..., T)
F0 = initial cash investment in time t = 0

r = discount rate of return (hurdle rate)

NPV =

t=0

Ft 1 +rt

Internal Rate of Return (IRR)

Find value of r such that NPV is equal to 0

Example (with T = 2):

Find r such that

F0 + F1 + F2 2 = 0 1+r 1+r

DCF Project Example*

Phase I Research and Product De velopment
$18 million annual research cos t for 2 years 60% probability of succ ess

Phase II

Market Deve lopment

Undertaken only if product development is s ucces sful $10 million annual expenditure for 2 years to develop marketing and distribution c hannels (net of any revenues earned in tes t marketing)

Phase III

Sales
Proceeds only if Phase I and II verify opportunity. Production is subcontracted and all cas h flow s are after-tax and occur at year's end. The results of Phase II (available at the end of year 4) identify the product's market potential as indic ated below :

Product Demand High Medium Low

Product Life 20 years 10 years Abandon Project

Annual Ne t Cash Inflow Probability $24 million $12 million None 0.3 0.5 0.2

*Hodder, J. and H.E. Riggs. Pitfalls in Evaluating Risky Projects, Harvard Business Review, Jan-Feb, 1985, pp. 128-136.

DCF Project Example (contd)

Year 1 2 3 4 5 - 14 15 - 24 Expe cte d Cas h Flow (in $ million) -18 -18 0.6 (-10) = - 6 0.6 (-10) = - 6 .6 (0.3 x 24 + 0.5 x 12) = 7.92 .6 (0.3 x 24) = 4.32

What is the internal rate of return for this project?

DCF Example Continued

What if you can sell the product (assuming that both Research and Product Development AND Market Development are successful) to a third party? What are the risks AT THAT POINT IN TIME?

Assume that discount rate r2 is 5%

Probability What is 20 years of cash inflow at $24M/year? What is 10 years of cash inflow at $12M/year? Expected value of product at Year 4: $299.09 $92.66 $136.06 0.3 0.5

DCF Example Continued

Expected cash flows (with sale of product at end of year 4) are now:
Expe cte d Cash Flow $ $ $ $ (18.00) (18.00) (6.00) 75.63

Year Year Year Year

1 2 3 4

Outflow $ 18.00 $ 18.00 $ 10.00 $ 10.00

Inflow $ $ $ $

136.06

Net (18.00) (18.00) (10.00) 126.06

Probability 1 1 0.6 0.6

What is the internal rate of return for this project?

Criticisms of NPV/DCF
1) Assumes that cash flow forecasts are accurate; ignores the human bias effect 2) Fails to include effects of inflation in long term projects 3) Ignores interaction with other proposed and ongoing projects (minimize risk through diversification) 4) Use of a single discount rate for the entire project (risk is typically reduced as the project evolves)

Expected Commercial Value (ECV)

Probability = pc Commercial Success (with net benefit = NPV) Launch New Product Probability = 1 - pc Commercial Failure (with net benefit = 0)

Probability = pt
Technical Success Develop New Product Probability = 1 - pt Technical Failure

Risk class 1

Risk class 2

DCF Example Revisited

0.3 Probability = pt Development Succeeds Research & Product Development Market Development 0.5 Product Demand Medium Product Demand High

0.2 Probability = 1 - pt Development Fails Drop project

Product Demand Low

Discount rate r1

Discount rate r2

Ranking/Scoring Models
Profit abilit y/value 1) Increase in profitability? 2) Increase in market share? 3) Will add knowledge to organization that can be leveraged by other projects? 4) Estimated NPV, ECV, etc. Organizat ion's Strat egy 1) Consistent with organization's mission statement? 2) Impact on customers? Risk 1) Probability of research being successful? 2) Probability of development being successful? 3) Probability of process success? 4) Probability of commercial success? 5) Overall risk of project 6) Adequ ate market demand? 7) Competitors in market Organizat ion Costs 1) Is new facilit y needed? 2) Can use current personnel? 3) External consultants needed? 4) New hires needed? Miscellaneous Factors 1) Impact on environmental standards? 2) Impact on workforce safety? 3) Impact on qualit y? 4) Social/polit ical implications

Scoring Attributes
To convert various measurement scales to a (0, 1) range. LINEAR SCALE: value of attribute i is vi xi =

xi - L U- L

EXPONENTIAL SCALE: value of attribute i is vi xi =

1.00 0.90 0.80 0.70

1 - exp L - xi . 1 - exp L - U

Attribute Value

0.60 0.50 0.40 0.30 0.20 0.10 0.00 1 2 3 4 Res ponse 5 6 7 Linear Scale Exponential Scale

Ranking/Scoring Example

Attribute
1) Does project increase market share? 2) Is new facility needed? 3) Are there safety concerns? 4) Likelihood of successful technical development? unlikely 1 1 unlikely 1

Measurement Scale
2 yes likely 2 2 unsure 3 3 4 4 5 5 3 4 no no likely likely 5 likely

Attribute Weight (wi)

30% 15% 10% 20% 25%

5) Likelihood of successful commercial development? unlikely

Ranking/Scoring Example (contd)

Project Score (V j)

Attribute
Project A Project B

#1
4 2 0.75 0.25 0.97 0.64

#2
yes no 0.25 0.75 0.64 0.97

#3
likely unsure 0 0.5 0.00 0.88

#4
4 3 0.75 0.5 0.97 0.88

#5
1 4 0 0.75 0.00 0.97

Linear Scale
Project A Project B
0.413 0.525 0.581 0.845

Exponential Scale
Project A Project B

Analyzing Project Portfolios: Bubble Diagram

Prob of Commercial Success
High

Zero

High

Expected NPV

Low

Analyzing Project Portfolios: Product vs Process

Extent of Process Change

Source: Clark and Wheelwright, 1992

Key Elements of Project Portfolio Selection Problem

1. Multi-period investment problem

2. Top management typically allocates funds to different product lines (e.g., compact cars, high-end sedans)
3. Product lines sell in separate (but not necessarily independent) market segments 4. Product line allocations are changed frequently 5. Conditions in each market segment are uncertain from period to period due to competition and changing customer preferences

Stage-Gate Approach
Initiation Define Design Improve Control

Initiation Project Review Charter

Work Statement Risk Assessment Purchasing Plan Change Mgt

Detail Design Schedule & Budget Contingency Plan Product & Performance Reviews

Installation Plan Facility Prep Training Plan Implementation

Production close-out Lessons learned Post-project audit

Source: PACCAR Information Technology Division Renton, WA

Project Selection Example

Y e a r (t) 2 $10 ($25) $20

1 Project A Project B Budget Limit (B

t)

3 $20 $50 $40

4 $20 $50 $55

($40) ($65) $120

Phases of Project Management

n n

Project formulation and selection Project planning

u u u u u

Summary statement Work breakdown structure Organization plan risk management Subcontracting and bidding process Time and schedule Project budget Resource allocation Equipment and material purchases Cost control metrics Change orders Milestone reports

Project scheduling
u u u u

Monitoring and control

u u u

Project Planning
n

Summary Statement
u

u
u u u
n

Executive summary: mission and goals, constraints Description and specifications of deliverables Quality standards used (e.g., ISO) Role of main contractor and subcontractors Composition and responsibilities of project team Managerial responsibilities assigned; signature authority Cross impact matrix (who works on what) Relationship with functional departments Project administration Role of consultants Communication procedures with organization, client, etc.

Organization Plan
u u u

u
u u

Importance of Project Planning

The 6P Rule of Project Management: Prior Planning Prevents Poor Project Performance
If you fail to plan, you will plan to fail
Anonymous

Work Breakdown Structure (WBS)

1) Specify the end-item deliverables

2) Subdivide the work, reducing the dollars and complexity with each additional subdivision
3) Stop dividing when the tasks are manageable work packages based on the following: Skill group(s) involved Managerial responsibility Length of time Value of task

Work Packages/Task Definition

The work packages (tasks or activities) that are defined by the WBS must be: Manageable Independent

Integratable
Measurable

Design of a WBS
The usual mistake PMs make is to lay out too many tasks; subdividing the major achievements into smaller and smaller subtasks until the work breakdown structure (WBS) is a to do list of one-hour chores. Its easy to get caught up in the idea that a project plan should detail everything everybody is going to do on the project. This springs from the screwy logic that a project managers job is to walk around with a checklist of 17,432 items and tick each item off as people complete them.
The Hampton Group (1996)

Two-Level WBS

WBS level 1

1. Charity Auction

WBS level 2

1.1 Event Planning

1.2 Item Procurement

1.3 Marketing

1.4. Corporate Sponsorships

Three-Level WBS
WBS level 1
1. Charity Auction

WBS level 2

1.1 Event Planning

1.2 Item Procurement

1.3 Marketing

1.4 Corporate Sponsorships

1.1.1 Hire Auctioneer

1.2.1 Silent auction items 1.2.2 Live auction items 1.2.3 Raffle items

1.3.1 Individual ticket sales

1.3.2 Advertising

1.1.2. Rent space

WBS level 3
1.1.3 Arrange for decorations
1.1.4 Print catalog

Estimating Task Durations (contd)

Benchmarking Modular approach Parametric techniques Learning effects

Beta Distribution
Probability density function Completion time of task j

Time
Optimistic Timeto j
Expected duration =

Most Likely Time =tm

Pes simis tic Time tp j

Beta Distribution
For each task j, we must make three estimates:

to j most optimistic time most pessimistic time tp j tm j most likely time

p m to j + tj + 4tj Expected duration j = 6

p o2 t j - tj 2 Variance of task j = j = 36

Estimating Task Durations: Painting a Room

Task: Paint 4 rooms, each is approximately 10 x 20. Use flat paint on walls, semi-gloss paint on trim and woodwork. Each room has two doors and four windows. You must apply masking tape before painting woodwork around the doors and windows. Preparation consists of washing all walls and woodwork (some sanding and other prep work will be needed). Only one coat of paint is necessary to cover existing paint. All supplies will be provided at the start of the task. Previous times on similar painting jobs are indicated in the table below.
hours 27 38 33 17 26 22 14 30 28 21 23 27 23 37 17 17 m in 25 25 12 44 7 1 2 27 30 13 59 44 15 6 54 13 hours 31 19 26 30 25 24 32 32 13 42 22 32 32 27 26 21 m in 52 15 27 27 21 28 58 1 43 45 57 15 31 15 11 52

What is your estimate of the average time you will need? What is your estimate of the variance?

Estimating Task Durations with Incentives Task: Consider the painting job that you have
just estimated. Now, however, there are explicit incentives for meeting your estimated times. If you finish painting the room before your specified time, you will receive a $10 bonus payment. HOWEVER, if you finish the painting job after your specified time, you will be fined $1000. Revised estimated time =

Estimating Task Durations with Incentives Task: Consider the painting job that you have
just estimated. Now, however, there are explicit incentives for meeting your estimated times. If you finish painting the room before your specified time, you will receive a $10 bonus payment. If you finish the painting job after your specified time, there is no penalty.

Revised estimated time =

Role of Project Manager/Team

Client Project Manager Subcontractors Project Team Regulating Organizations

Top Management

Functional Managers

Responsibilities of a Project Manager

To the organization and top management
Meet budget and resource constraints Engage functional managers

To the project team

Provide timely and accurate feedback Keep focus on project goals Manage personnel changes

To the client
Communicate in timely and accurate manner Provide information and control on changes/modifications Maintain quality standards

To the subcontractors
Provide information on overall project status

Project Team
What is a project team?
A group of people committed to achieve a common set of goals for which they hold themselves mutually accountable

Characteristics of a project team

Diverse backgrounds/skills Able to work together effectively/develop synergy Usually small number of people Have sense of accountability as a unit

I design user interfaces to please an audience of one. I write them for me. If Im happy, I know some cool people will like it. Designing user interfaces by committee does not work very well; they need to be coherent. As for schedule, Im not interested in schedules; did anyone care when War and Peace came out?
Developer, Microsoft Corporation
As reported by MacCormack and Herman, HBR Case 9-600-097: Microsoft Office 2000

Intra-team Communication
M = Number of project team members L = Number of links between pairs of team members
If M =2, then L = 1

If M =3, then L = 3

Number of Intra-team Links

L = Number of Intra-team Links =

N 2

N(N-1) 2

Importance of Communication
On the occasion of a migration from the east, men discovered a plain in the land of Shinar, and said to one another, Come, let us build ourselves a city with a tower whose top shall reach the heavens. The Lord said, Come, let us go down, and there make such a babble of their language that they will not understand one anothers speech. Thus, the Lord dispersed them from there all over the earth, so that they had to stop building the city.
Genesis 11: 1-8

Project Performance and Group Harmony

What is the relationship between the design of multidisciplinary project teams and project success? Two schools of thought: 1) Humanistic school -- groups that have positive characteristics will perform well 2) Task oriented school -- positive group characteristics detract from group performance

Project Performance and Group Harmony (contd) Experiment conducted using MBA students at UW and Seattle U using computer based simulation of pre-operational testing phase of nuclear power plant* Total of 14 project teams (2 - 4 person project teams) with a total of 44 team members; compared high performance (low cost) teams vs low performance (high cost) teams

Measured:

Group Harmony Group Decision Making Effectiveness Extent of Individuals Contributions to Group Individual Attributes
*Brown, K., T.D. Klastorin, & J. Valluzzi. Project Management Performance: A Comparison of Team Characteristics, IEEE Transactions on Engineering Management, Vol 37, No. 2 (May, 1990), pp. 117-125.

Group Harmony: High vs Low Performing Groups

Extent of Individual Contribution: High vs Low Performing Groups

Decision Making Effectiveness: High vs Low Performing Groups

Project Organization Types

Functional: Project is divided and assigned to appropriate functional entities with the coordination of the project being carried out by functional and high-level managers Functional matrix: Person is designated to oversee the project across different functional areas Balanced matrix: Person is assigned to oversee the project and interacts on equal basis with functional managers Project matrix: A manager is assigned to oversee the project and is responsible for the completion of the project Project team: A manager is put in charge of a core group of personnel from several functional areas who are assigned to the project on a full-time basis

Project Organization Continuum

Functional Matrix

Project Matrix

Functional Organization

Balanced Matrix

Project Team Organization

Project fully managed by functional managers

Project fully managed by project team manager

A Business School as a Matrix Organization

Dean
Associate Dean for Undergraduate Program Accounting Department Chair Marketing Department Chair Finance Department Chair Associate Dean for MBA Programs Director of Doctoral Program

Larry

Zelda

Diane

Curly

Bob

Barby

Moe

Gloria

Leslie

Matrix Organizations & Project Success

Matrix

organizations emerged in 1960s as an alternative to traditional means of project teams popular in 1970s and early 1980s

Became Still

in use but have evolved into many different forms

Basic question: Does organizational structure impact probability of project success?

Organizational Structure & Project Success

Studies by Larson and Gobeli (1988, 1989) Sent questionnaires to 855 randomly selected PMI members Asked about organizational structure (which one best describes the primary structure used to complete the project) Perceptual measures of project success: successful, marginal, unsuccessful with respect to : 1) Meeting schedule 2) Controlling cost 3) Technical performance 4) Overall performance Respondents were asked to indicate the extent to which they agreed with each of the following statements: 1) Project objectives were clearly defined 2) Project was complex 3) Project required no new technologies 4) Project had high priority within organization

Study Data
Classification of 547 respondents (64% response rate) 30% project managers or directors of project mgt programs 16% top management (president, vice president, etc.) 26% managers in functional areas (e.g., marketing) 18% specialists working on projects Industries included in studies 14% pharmaceutical products 10% aerospace 10% computer and data processing products others: telecommunications, medical instruments, glass products, software development, petrochemical products, houseware goods Organizational structures: 13% (71): Functional organizations 26% (142): Functional matrix 16.5% (90): Balanced matrix 28.5% (156): Project matrix 16% (87): Project team

ANOVA Results by Organizational Structure

Controlling Cost Ave (SD) 1.76 (.83) 1.91 (.77) 2.39 (.73) 2.64 (.76) 2.22 (.82) 2.12 (.79) 10.38* Meeting Sche dule Ave (SD) 1.77 (.83) 2.00 (.85) 2.15 (.82) 2.30 (.79) 2.32 (.80) 2.14 (.83) 6.94* Technical Performance Ave (SD) 2.30 (.77) 2.37 (.73) 2.64 (.61) 2.67 (.57) 2.64 (.61) 2.53 (.66) 7.42* Overall Results Ave (SD) 1.96 (.84) 2.21 (.75) 2.52 (.61) 2.54 (.66) 2.52 (.70) 2.38 (.70) 11.45*

Organizational Structure A B C D E Functional Organ ization Functional Matrix Balance d Matrix Project Matrix Project Team Total Sample F-statistic Scheffe Results

N 71 142 90 156 87 546

A,B < C,D,E E<D A,B < C < D,E A,B < C,D,E A,B < C,D,E

*Statistically significant at a p<0.01 level

Summary of Results
Project structure significantly related to project success New development projects that used traditional functional organization

had lowest level of success in controlling cost, meeting schedule,

achieving technical performance, and overall results Projects using either a functional organization or a functional matrix had a significantly lower success rate than the other three structures Projects using either a project matrix or a project team were more successful in meeting their schedules than the balanced matrix Project matrix was better able to control costs than project team

Overall, the most successful projects used a balanced matrix, project

team, or--especially--project matrix

Subcontracting = Business Alliance

n

When you subcontract part (or all) of a project, you are forming a business alliance....

Intelligent Business Alliances: A business relationship for mutual benefit between two or more parties with compatible or complementary business interests and/or goals Larraine Segil, Lared Presentations

Communication and Subcontractors

What types of communication mechanism(s) will be used between company and subcontractor(s)?

WHAT a company communicates.....

HOW a company communicates.....

How is knowledge transferred?

Personality Compatibility
Subcontractor Personality

Corporate Personality

Project

Individual Personality

Subcontracting Issues
What part of project will be subcontracted? n What type of bidding process will be used? What type of contract? n Should you use a separate RFB (Request for Bids) for each task or use one RFB for all tasks? n What is the impact on expected duration of project? n Use a pre-qualification list? n Incentives? Bonus for finishing early? Penalties for finishing after stated due date? What is impact of risk on expected project cost?
n

Basic Contract Types

n

Fixed Price Contract

u

Client pays a fixed price to the contractor irrespective of actual audited cost of project

Cost Plus Contract

u

Client reimburses contractor for all audited costs of project (labor, plant, & materials) plus additional fee (that may be fixed sum or percent of costs incurred)

Units Contract
u

Client commits to a fixed price for a pre-specified unit of work; final payment is based on number of units produced

Incentive (Risk Sharing) Contracts General Form:

Payment to Subcontractor = Fixed Fee + (1 - B) (Project Cost)
where B = cost sharing rate

Cost Plus Contract B=0 Linear & Signalling Contracts

Fixed Price Contract B=1

Why Use Incentive Contracts?

Expected Cost of Project = $100M Two firms bid on subcontract Firm 1 Firm 2

Fixed Fee (bid)

Project Cost

$5 M
$105 M (inefficient producer)

$7 M
$95 M

What is result if Cost Plus Contract (B = 0) used?

Washington State Bid Code (WAC 236-48-093)

n

n n n

WAC 236-48-093: A contract shall be awarded to the lowest responsible and responsive bidder based upon, but not limited to, the following criteria where applicable and only that which can be reasonably determined: 1) The price and effect of term discounts...price may be determined by life cycle costing if so indicated in the invitation to bid 2) The conformity of the goods and/or services bid with invitation for bid or request for quotation specifications depicting the quality and the purposes for which they are required. 3) The ability, capacity, and skill of the bidder to perform the contract or provide the services required. 4) The character, integrity, reputation, judgement, experience, and efficiency of the bidder. 5) Whether the bidder can perform the contract with the time specified. 6) The quality of performance on previous contracts for purchased goods or services. 7) The previous and existing compliance by the bidder with the laws relating to the contract for goods and services. 8) Servicing resources, capability, and capacity.

Competitive Bidding: Low-Bid System

n

In the low-bid system, the owner wants the most building for the least money, while the contractor wants the least building for the most money. The two sides are in basic conflict.
Steven Goldblatt Department of Building Construction University of Washington The Seattle Times, Nov 1, 1987

Precedence Networks
Networks represent immediate precedence relationships among tasks (also known as work packages or activities) and milestones identified by the WBS Milestones (tasks that take no time and cost $0 but indicate significant events in the life of the project)

Two types of networks: Activity-on-Node (AON)

Activity-on-Arc (AOA) All networks: must have only one (1) starting and one (1) ending point

Precedence Networks: Activity-on-Node (AON)

Start

End

Precedence Diagramming
Standard precedence network (either AOA or AON) assumes that a successor task cannot start until the predecessor(s) task(s) have been completed. Alternative relationships can be specified in many software packages: Finish-to-start (FS = a): Job B cannot start until a days after Job A is finished Start-to-start (SS = a): Job B cannot start until a days after Job A has started Finish-to-finish (FF = a): Job B cannot finish until a days after Job A is finished

Start-to-finish (SF = a): Job B cannot finish until a days after Job A has started

Critical Path Method (CPM): Basic Concepts

Task A 7 months

Task B 3 months

Start End

Task C 11 months

Critical Path Method (CPM): Basic Concepts

ESA = 0 LFA = 8 ESStart = 0 LFStart = 0 Start Task A 7 months ESB = 7 LFB = 11 Task B 3 months ESEnd = 11 LFEnd = 11

End

Task C 11 months ESC = 0 LFC = 11

ESj = Earliest starting time for task (milestone) j LFj = Latest finish time for task (milestone) j

AON Precedence Network: Microsoft Project

Ta sk A 2 Wed 1 2/20/00 7d Th u 1 2/28/00

Ta sk B 3 Fri 1 2/29/00 3d Tu e 1 /2 /0 1

Start 1 Wed 1 2/20/00 0d Wed 1 2/20/00 5 Wed 1 /3 /0 1 0d Wed 1 /3 /0 1 En d

Ta sk C 4 Wed 1 2/20/00 11d Wed 1 /3 /0 1

Critical Path Method (CPM): Example 2

ES A = LFA = ES F = LFF = ES D = LFD =

Task A 14 wks
ES ST ART = 0 LF ST ART = 0

Task F 9 wks

ES B = LFB =
START

Task D 12 wks
ES E = LFE =

ES END = LFEND=

Task B 9 wks
ES C = LFC =

END

Task E 6 wks

Task C 20 wks

Example 2: Network Paths

Expected Duration (wks)
35 32 30 27 26

Path
1 2 3 4 5

Tasks
START-A-D-F-END START-A-D-E-END START-B-D-F-END START-B-D-E-END START-C-E-END

Example 2: CPM Calculations

EA R L I ES T L A T ES T

Tas k or Milestone
ST ART

Duration ( ti ) 0 14 9 20 12 6 9 0

Start Time (ESi) 0 0 0 0 14 26 26 35

Finis h Time 0 14 9 20 26 32 35 35

Start Time 0 0 5 9 14 29 26 35

Finis h Time (LFi) 0 14 14 29 26 35 35 35

A B C D E F
END

Example 2: Calculating Total Slack (TSi)

Total Slack for task i = TSi = LFi - ESi - ti

Tas k or Milestone
ST ART

Duration ( ti ) 0 14 9 20 12 6 9 0

Earliest Start Time (ESi) 0 0 0 0 14 26 26 35

Lastest Finis h Time (LFi) 0 14 14 29 26 35 35 35

Total Slack (TSi)

0 0 5 9 0 3 0 0

Critical Tas k ?
Yes Yes No No Yes No Yes Yes

A B C D E F
END

Slack (Float) Definitions (for task i)

Total Slack (TSi) Free Slack (FSi)
= LFi - ESi - ti = ESi,min - ESi - ti

where ESi,min = minimum early start time of all tasks that immediately follow task i = min (ESj for all task j Si)

Safety Slack (SSi)

= LFi - LFi,max - ti

where LFi,max = maximum late finish time of all tasks that immediately precede task i

= min (LFj for all task j Pi)

Independent Slack (ISi)

= max (0, ESi,min - LFi,max - ti)

Example #2: LP Model

Decision variables: STARTj = start time for task j END = ending time of project (END milestone)

Minimize END
subject to

STARTj FINISHi
STARTj 0

for all tasks i that immediately precede task j

for all tasks j in project

where FINISHi = STARTi + ti = STARTi + duration of task i

Example #2: Excel Solver Model

Gantt Chart

Microsoft Project 4.0

Project Budgeting
The budget is the link between the functional units and the project Should be presented in terms of measurable outputs

Budgeted tasks should relate to work packages in WBS and organizational units responsible for their execution
Should clearly indicate project milestones Establishes goals, schedules, and assigns resources (workers, organizational units, etc.) Should be viewed as a communication device Serves as a baseline for progress monitoring & control

Update on rolling horizon basis

May be prepared for different levels of aggregation (strategic, tactical, short-range)

Project Budgeting (contd)

Top-down Budgeting: Aggregate measures (cost, time) given by top management based on strategic goals and constraints Bottom-up Budgeting: Specific measures aggregated up from WBS tasks/costs and subcontractors

Issues in Project Budgets

How to include risk and uncertainty factors? How to measure the quality of a project budget? How often to update budget? Other issues?

Critical Path Method (CPM): Example 2

ES A = 0 LFA = 14 ES F = 26 LFF = 35 ES D = 14 LFD = 26

Task A 14 wks
ES ST ART = 0 LF ST ART = 0

Task F 9 wks

ES B = 0 LFB = 14
START

Task D 12 wks
ES E = 26 LFE = 35

ES END = 35 LFEND= 35

Task B 9 wks
ES C = 0 LFC = 29

END

Task E 6 wks

Task C 20 wks

Project Budget Example

Task or Milestone
ST ART

Duration (tj) 0 14 9 20 12 6 9 0

Early Start Time (ESj)

0 0 0 0 14 26 26 35

No. of No. of Latest Start Resource A Resource B Time (LSj) worke rs worke rs
0 0 5 9 14 29 26 35 2 4 3 0 1 4 0 12 14 8 0 10 $ $ $ $ $

Material Costs
340 125 200 560 90 -

Direct Labor Labor + Materials Cost/wk

$ $ $ $ $ $ 800 8,800 9,600 4,800 400 7,600 $ $ $ $ $ $ 1,140 8,925 9,600 5,000 960 7,690 -

A B C D E F
END

Cost for Resource A worker = $400/week Cost for Resource B worker = $600/week

Project Budget Example (contd)

Early Start Time s Tas k 1 A 1140 B 8925 C 9600 D E F
Wee kl y Sub to ta ls Cu mu la ti ve

Week
2 800 8800 9600 3 800 8800 9600 4 800 8800 9600 5 800 8800 9600 6 800 8800 9600 7 800 8800 9600 8 800 8800 9600 9 800 8800 9600 9600 9600 9600 10 800 11 800 12 800

19665 19665

19200 38865

19200 58065

19200 77265

19200 96465

19200 115665

19200 134865

19200 154065

19200 173265

10400 183665

10400 194065

10400 204465

Late Start Time s Tas k A B C D E F

Wee kl y Sub to ta ls Cu mu la ti ve

Week
2 800 3 800 4 800 5 800 8925 6 800 8800 7 800 8800 8 800 8800 9 800 8800 9600 10 800 8800 9600 11 800 8800 9600 12 800 8800 9600

1 1140

1140 1140

800 1940

800 2740

800 3540

9725 13265

9600 22865

9600 32465

9600 42065

19200 61265

19200 80465

19200 99665

19200 118865

Cumulative Costs

Range of feasible budgets

Weekly Costs (Cash Flows)

Managing Cash Flows

Want to manage payments and receipts Must deal with budget constraints on project and organization requirements (e.g., payback period)

Organization profitability

Cash Flow Example

Make payment of $5000

M1
Task A 2 mos

Task D 8 mos Task C 4 mos

Receive payment of $3000

START

END

Task B 8 mos

Task E 3 mos

M2

Receive payment of $3000

Cash Flow Example: Solver Model

Material Management Issues

When to order materials? How much to order?
Example: Single material needed for Task B (2 units) and Task E (30 units) Fixed cost to place order = S Cost of holding raw materials proportional to number of unit-weeks in stock Cost of holding finished product greater than the cost of holding raw materials Project can be delayed (beyond 17 weeks) at cost of $P per week

Material Management Example

LS A = 0
Task A 4 wks

LS B = 4
Task B 8 wks

LS C = 12
Task C 5 wks

Start

2 units

LS D = 6
Task D 6 wks

LS E = 12
Task E 2 wks

LS F = 14
Task F 3 wks

End

30 units

Lot-Sizing Decisions in Projects

To minimize holding costs, only place orders at Late Starting Times Can never reduce holding costs by delaying project Time 1 2 3 4 5 6 7 8 9 10 11 12

Demand:

30

Order option #1: 32 Order option #2: 2 30

Choose the option that minimizes inventory cost = order cost + holding cost of raw materials

Time-Cost Tradeoffs

Time-Cost Tradeoff Example

A
Start

C
End

Tas k
A B C D

Normal Duration
7 6 15 10

Marginal Cost to Crash One Normal Cost Week

$60 $85 $55 $120 $8 $5 $10 $4

Time-Cost Tradeoff Example (contd)

Project Duration (weeks) Critical Path(s)
Start-A-C-End Start-A-C-End Start-A-B-End Start-A-C-End Start-A-B-End

Task(s) Reduced

Total Direct Cost

$320 $328 $338

22 21 20

A C

19
18

Start-A-C-End
Start-A-B-End Start-A-C-End Start-A-B-End

C
A, B

$348
$361

Linear Time-Cost Tradeoff

In theory, the normal or expected duration of a task can be reduced by assigning additional resources to the task

Cost
Crash Point
Crash cost = Cc j Slope (bj) = Increase in cost by reducing task by one time unit

Normal Point

Normal cost = CN j

Crash time =

tc j

Normal time

N = tj

Time

Balancing Overhead & Direct Costs

Cost

Total Cost
Indirect (overhead) Costs

Direct Costs

Crash Time

Minimum Cost Solution

Normal Time

Project Duration

Time-Cost Tradeoff (Direct Costs Only)

N Given Normal point with cost CN j and time tj

and Crash point with cost Cj

c and time tj

N Cc j - Cj Assume constant marginal cost of crashing task j = bj = c N tj - tj

Decision Variables: Sj = Starting time of task j END = End time of project tj = Duration of task j Minimize Total Direct Cost = Sj Si + ti
N tc j tj t j

bj tj
j

for all tasks i Pj

for all tasks in project

END = Tmax t j, Sj 0

General Time-Cost Tradeoffs

Minimize Total Costs =

bj tj
j

+ I (END) + P L

where I = indirect (overhead) cost/time period P = penalty cost/time period if END is delayed beyond deadline Tmax L = number of time periods project is delayed beyond deadline Tmax QUESTION: HOW TO DEFINE L?

Software Project Schedules

Observe that for the programmer, as for the chef, the urgency of the patron may govern the scheduled completion of the task, but it cannot govern the actual completion. An omelet, promised in ten minutes, may appear to be progressing nicely. But when it has not set in ten minutes, the customer has two choices--wait or eat it raw. Software customers have the same choices. The cook has another choice; he can turn up the heat. The result is often an omelet nothing can save--burned in one part, raw in another.
F.P. Brooks, The Mythical Man-Month, Datamation, Vol 20, No 12 (Dec, 1974), pp. 44-52.

Coordination Costs (Software Development Project)

n

n n

Assume you want to develop program that will require (approximately) 50,000 lines of PERL code A typical programmer can write approximately 1500 lines of code per week Coordination time is M (M-1)/2 weeks

No. of Programmers 1 2 3 4 5 6 7 8 9 10 11

No. of Weeks Coding 33.33 16.67 11.11 8.33 6.67 5.56 4.76 4.17 3.70 3.33 3.03

No. of Total Coordination Number of Weeks Weeks 0 33.33 1 17.67 3 14.11 6 14.33 10 16.67 15 20.56 21 25.76 28 32.17 36 39.70 45 48.33 55 58.03

Brooks Law

Adding manpower to a late software project makes it later.

n

F.P. Brooks, The Mythical Man-Month, Datamation, Vol 20, No 12 (Dec, 1974), pp. 44-52.

Compressing New Product Development Projects

Traditional Method
Design follows a sequential pattern where information about the new product is slowly accumulated in consecutive stages
Stage 0 Stage 1 Stage N

New Product Development Process

Overlapped Product Design

Allows downstream design stages to start before preceding upstream stages have finalized their specifications.
Stage 0 Stage 1 Stage N

Issues and Tradeoffs

What are the tradeoffs when moving from a traditional sequential product design process to an overlapped product design process?

Increased uncertainty (that leads to additional work)

Can add additional resources to tasks to reduce duration--but costs are increased

Classic PERT Model Defined

Since task durations are now random variables, time of any milestone (e.g., end of project) is now RV

Assume all tasks are statistically independent

Use values of j to identify expected critical path Since time of event (e.g., ESk) is now sum of independent RVs, central limit theorem specifies that ESk is approximately normally distributed with mean E[ESk] and variance Var[ESk]
Expected early start time of task k = E ES k = max s

tasks j on path s

where there exists s paths to task k

Classic PERT Model (contd)

Thus, expected project duration is defined as:
Expect Project Duration = E[E S END ] =
tasks j on CP

Variance of Project Duration = Var[E S END ] =

tasks j on CP

2 j

Using central limit theorem and standard normal distribution:

P ESEND T max = P z

Tmax - E ESEND Var ESEND

PERT Example #1
Task B Programming Task E Implementation

Start

Task A Requirements Analysis

Task C Hardware Acquisit ion

Task F Test ing

End
Task D User Training

Dur ation Estim ates

Ta sk A B C D E F END De scription Re qui re ments An al ysis Prog rammi ng Ha rd ware acq ui si tio n User trai ni ng Impl emen ta ti on Te stin g En d of proj ect Prede ces sors no ne A A A B, C E D, F Optimis tic 2 4 2 12 3 3 0 Pe ssimis tic 14 12 13 18 7 7 0 Like ly 6 7 8 14 5 4 0

Ex pected Duration 6.67 7.33 7.83 14 .3 3 5.00 4.33 0.00

Va rianc e 4.00 1.78 3.36 1.00 0.44 0.44 0.00

PERT Example #1 (contd)

Task B Programming Task E Implementation

Start

Task A Requirements Analysis

Task C Hardware Acquisit ion

Task F Test ing

End
Task D User Training

Ta sk B,C,D E F En d

Pa th Start-A Start-A-C Start-A-C-E Start-A-C-E-F-End

Ex pected Ea rly Start 6.67 14 .5 0 19 .5 0 23 .8 3 23 .8 3 8.25 0

Va rianc e 4.00 7.36 7.81 8.25 Ex pected CP

Due Date 6 15 20 25

Zi -0.33 0.18 0.18 0.41

Pr(zi) 0.37 0.57 0.57 0.66

PERT Expe cted Duration = PERT Varia nce =

= {Start, A, C, E, F, End}

PERT Example #2
Task A A = 4 2 A =2

Task C
C = 10 2 C=5

START
Task B B = 12 2 B=4 Task D D = 3 2 D=1

END

Example #3: Discrete Probabilities

Tas k A (8.0) Tas k D (9.3) START Tas k B (10.0) END

Tas k C (19.0)

Tas k A
Value 7 8 9 Pr ob 0.333 0.333 0.333

Task B
Value 2 12 Pr ob 0.2 0.8

Task C
Value 5 15 25 Pr ob 0.2 0.2 0.6

Task D
Value 3 12 Pr ob 0.3 0.7

Example #3 (contd)
Tas k A
Com bination 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Value 7 7 7 7 7 7 7 7 7 7 7 7 8 8 8 8 8 8 8 8 8 8 8 8 9 9 9 9 9 9 9 9 9 9 9 9 Pr ob 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333 0.333

Task B
Value 2 2 2 2 2 2 12 12 12 12 12 12 2 2 2 2 2 2 12 12 12 12 12 12 2 2 2 2 2 2 12 12 12 12 12 12 Pr ob 0.2 0.2 0.2 0.2 0.2 0.2 0.8 0.8 0.8 0.8 0.8 0.8 0.2 0.2 0.2 0.2 0.2 0.2 0.8 0.8 0.8 0.8 0.8 0.8 0.2 0.2 0.2 0.2 0.2 0.2 0.8 0.8 0.8 0.8 0.8 0.8

Tas k C
Value 5 5 15 15 25 25 5 5 15 15 25 25 5 5 15 15 25 25 5 5 15 15 25 25 5 5 15 15 25 25 5 5 15 15 25 25 Pr ob 0.2 0.2 0.2 0.2 0.6 0.6 0.2 0.2 0.2 0.2 0.6 0.6 0.2 0.2 0.2 0.2 0.6 0.6 0.2 0.2 0.2 0.2 0.6 0.6 0.2 0.2 0.2 0.2 0.6 0.6 0.2 0.2 0.2 0.2 0.6 0.6

Task D
Value 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 3 12 Pr ob 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7 0.3 0.7

Cr itical Path A, D A, D C A, D C C B, D B, D B, D B, D C C A, D A, D C A, D C C B, D B, D B, D B, D C C A, D A, D C A, D C C B, D B, D B, D B, D C C

Pr ob of CP 0.004 0.009 0.004 0.009 0.012 0.028 0.016 0.037 0.016 0.037 0.048 0.112 0.004 0.009 0.004 0.009 0.012 0.028 0.016 0.037 0.016 0.037 0.048 0.112 0.004 0.009 0.004 0.009 0.012 0.028 0.016 0.037 0.016 0.037 0.048 0.112

Length of CP 10 19 15 19 25 25 15 24 15 24 25 25 11 20 15 20 25 25 15 24 15 24 25 25 12 21 15 21 25 25 15 24 15 24 25 25 A,D
0.00 4 0.00 9 0.00 0 0.00 9 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 4 0.00 9 0.00 0 0.00 9 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 4 0.00 9 0.00 0 0.00 9 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0

PATHS
B, D
0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.01 6 0.03 7 0.01 6 0.03 7 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.01 6 0.03 7 0.01 6 0.03 7 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.00 0 0.01 6 0.03 7 0.01 6 0.03 7 0.00 0 0.00 0

C
0.00 0 0.00 0 0.00 4 0.00 0 0.01 2 0.02 8 0.00 0 0.00 0 0.00 0 0.00 0 0.04 8 0.11 2 0.00 0 0.00 0 0.00 4 0.00 0 0.01 2 0.02 8 0.00 0 0.00 0 0.00 0 0.00 0 0.04 8 0.11 2 0.00 0 0.00 0 0.00 4 0.00 0 0.01 2 0.02 8 0.00 0 0.00 0 0.00 0 0.00 0 0.04 8 0.11 2

6.8%

32.0%

61.1%

Example #3 (contd)
Length of CP's 10 11 12 15 19 20 21 24 25 Pr ob 0.004 0.004 0.004 0.108 0.019 0.019 0.019 0.224 0.599 Cum ulative Pr ob 0.00 0.01 0.01 0.12 0.14 0.16 0.18 0.40 1.00

Criticality Indices
Task A
6.8% Task B 32.0% Task C 61.1% Task D 38.8%

Expected Project Duration = 23.22

Monte-Carlo Simulation (PERT Example 1)

Ta sk A B C D E F END Ta sk Duration (Uniform Dist) 4.99 4.75 3.38 12 .2 0 5.94 5.34 0.00 Ea rly Start 0 4.99 4.99 4.99 9.74 15 .6 8 21 .0 2 La te st Finish 4.99 9.74 9.74 21 .0 2 15 .6 8 21 .0 2 21 .0 2 Tota l Slack 0.00 0.00 1.36 3.83 0.00 0.00 0.00 Ex pected Duration 6.67 7.33 7.83 14 .3 3 5.00 4.33 0.00 Va rianc e 4.00 1.78 3.36 1.00 0.44 0.44 0.00

Run 1 2 3 4 5 6 7 19 7 19 8 19 9 20 0 Av e Va r

Proj ect Duration 31 .0 7 27 .4 1 23 .9 7 28 .9 3 26 .8 5 28 .8 2 28 .7 7 30 .3 7 29 .7 8 25 .3 3 29 .7 0 27 .1 3 16 .7 77

t(B) 1 0 1 0 1 0 0 0 1 1 0 48 .5 %

t(C) 0 1 0 1 0 0 1 1 0 0 1 42 .0 %

t(D) 0 0 0 0 0 1 0 0 0 0 0 9.5%

t(E ) 1 1 1 1 1 0 1 1 1 1 1 90 .5 %

t(F) 1 1 1 1 1 0 1 1 1 1 1 90 .5 %

Project Makes pan

95% Confidence interval 99% Confidence interval

Lower Limit 26.56 26.37

Upper Limit 27.72 27.90

Calculating Confidence Intervals

For a confidence interval, we can use the sample mean X and the estimated standard error of the mean sX = s n where s is the sample standard deviation and n is the number of trials

Using a normal approximation, a (1- a) twosided confidence interval is given by

X - za/2 sX

New Product Development Projects

Beta test fails (with probability of 0.25) and rework is needed

Lease Mfg/Office Space

Design of physical unit Assemble prototype Beta test prototype

START END Identify/hire staff Electronics design Software

Beta test fails (with probability of 0.25) and rework is needed

New Product Development Projects (contd)

Lease Mfg/Office Space

Design of physical unit Assemble prototype Beta test prototype Prob = .25

Prob = .75

END

START Identify/hire staff

Electronics design

Beta test fails and rework is needed Software

Critical Chain and the Theory of Constraints (TOC)

Project Goal (according to Goldratt): Meet Project Due Date

Use deterministic CPM model with buffers to deal with any uncertainties, Place project buffer after last task to protect the customers completion schedule, Exploit constraining resources (make certain that resources are fully utilized), Avoid wasting time slack time by encouraging early task completions, Carefully monitor the status of the buffer(s) and communicate this status to other project team members on a regular basis, and Make certain that the project team is 100 percent focused on critical chain tasks

Project Buffer Defined

Project Buffer is placed at the end of the project to protect the customers promised due date

Task B Programming Task E Implementation Task F Testing

Start

Task A requirements analysis

Task C Hardware acquisition

Project Buffer
Task D User User training

End

PERT Example #1 Revisited with Project Buffer

Calculating Project Buffer Size

For those who want a scientific approach to sizing buffers....
For tasks k on critical chain, we can calculate project buffer using following formula that project will be completed within worst-case duration estimates around 90 percent of the time:

Buffer =

tasks k on critic al chain

tp k - k

Implications of Project Uncertainty

Task A

START Task B

END

Assume that the duration of both tasks A and B are described by a normal distribution with a mean of 30 days What is the probability that the project will be completed within 30 days?

Uncertainty and Worker Behavior

Consider a project with two tasks that must be completed serially

The duration of each task is described by a RV with values Ti (i = 1, 2)

Start

Task 1

Task 2

End

Values of T 1
7 8 9

Pr ob 0.3 0.4 0.3 8.0

Values of T 2
14 18

Pr ob 0.5 0.5 16

Parkinsons Law (Expanding Work)

Work expands so as to fill the time available for its completion
Professor C.N. Parkinson (1957)
Set a deadline D = 24 days So T(D) = project makespan (function of D) where E[T(D)] = E(T1) + E(T2) + E[max(0, D - T1 - T2)]
Values of T 1 7 7 8 8 9 9 Values of T 2 14 18 14 18 14 18 Project Mak espan 24 25 24 26 24 27

Prob 0.3 0.3 0.4 0.4 0.3 0.3

Prob 0.5 0.5 0.5 0.5 0.5 0.5

Prob 0.15 0.15 0.2 0.2 0.15 0.15

E[T(D)] = 25 days

Procrastinating Worker
Set a deadline D = 24 days E[T(D)] = E(T1) + E(T2) + E{max[0, D - T1 - E(T2)]}

Values of T 1 7 8 9

Prob 0.3 0.4 0.3 8

E[De lay] = max[0, D - T 1 - E(T2)] 1 0 0 0.3

E[Mak e span] 24 24 25 24.30

Can show that E[T(D)] E[T(D)] D

What are the implications for project managers?

Schoenbergers Hypothesis
An increase in the variability of task durations will increase the expected project duration.

Schoenbergers Hypothesis Illustrated

Task A

START

END

Task B

Duration of Task A 12 14 16 14.0

Probability 0.1 0.8 0.1

Duration of Task B 10 15 12.5

Probability 0.5 0.5

Schoenbergers Hypothesis Illustrated

Realization 1 2 3 4 5 6 Task A Duration 12 14 16 12 14 16 Task B Duration 10 10 10 15 15 15 Probability 0.05 0.4 0.05 0.05 0.4 0.05 Max (A, B) 12 14 16 15 15 16

Expected duration equals 14.55 days Increasing the variance of Task A:

Duration of Task A 12 14 16 14.0 Probability 0.3 0.4 0.3 Duration of Task B 10 15 12.5 Probability 0.5 0.5

Results in an increased expected duration = 14.65 days

Risk Management
All projects involve some degree of risk Need to identify all possible risks and outcomes Need to identify person(s) responsible for managing project risks

Identify actions to reduce likelihood that adverse events will occur

Risk Analysis
Risk Exposure (RE) or Risk Impact =
(Probability of unexpected loss) x (size of loss) Example: Additional features required by client Loss: 3 weeks Probability: 20 percent Risk Exposure = (.20) (3 weeks) = .6 week

How to Manage Project Risks?

Preventive Actions
Actions taken in anticipation of adverse events May require action before project actually begins Examples?

Contingency Planning
What will you do if an adverse event does occur? Trigger point invokes contingency plan Frequently requires additional costs

Risk and Contracts

High

Low

Low

High

Degree of Risk Contractor Client

Fixed Price Contract
Elements can be Firm price renegotiated Incentives

Cost Plus Contract

Cost Plus T&M with with limits Incentives Time & materials

Tornado Diagram
Wage Rate Direct Labor Hours Material Units Needed Early Completion Bonus Material Unit Cos t Interes t rates Energy cos ts Overhead $1260 $1290 $1265 $1310 1310 $1350 $1350 $1380 $1400 $1760 1760 $1700 $1720 $1680 $1690 $1640 $1620 $1625

$1200

$1300

$1400

$1500

$1600

$1700 $1800

Project Cost ($000's )

Sensitivity Chart
Wage Rate Direct Labor Hours Material Units Needed Early Completion Bonus Material Unit Cos t Interes t rates Energy cos ts Overhead 0.85 0.73 0.62 -0.45 0.42 0.28 0.19 0.10 -0.5 0 0.5 1.0

Rank Order Correlation with Total Project Cost

Van Allen Company

Strike (wk s) Pr ob 3 0.45 4 0.3 5 0.25 E[Strike Dur ation]

Expected Dur ation 1.35 1.20 1.25 3.80

Resource Allocation & Leveling

Resource Leveling: Reschedule the noncritical
tasks to smooth resource requirements

Resource Allocation: Minimize project

duration to meet resource availability constraints

Resource Allocation & Leveling

Three types of resources:
1) Renewable resources: renew themselves at the beginning of each time period (e.g., workers) 2) Non-Renewable resources: can be used at any rate but constraint on total number available 3) Doubly constrained resources: both renewable and non-renewable

Resource Leveling
Tas k C 9 wk s Tas k A 3 wk s Tas k D 5 wk s Tas k G 5 wk s END

START

Tas k B 2 wk s

Tas k E 3 wk s

Tas k F 2 wk s

Tas k A B C D E F G

Work ers 7 3 2 10 4 5 6

Duration (tj) 3 2 9 5 3 2 5

Early Start 0 0 3 3 2 2 8

Late Start 0 3 4 3 5 11 8

Resource Leveling: Early Start Schedule

Resource Leveling: Late Start Schedule

Resource Leveling: Microsoft Project

De c 1 7, '0 0 T 25 W T F S S M T De c 2 4, '0 0 W T F S S M T De c 3 1, '0 0 W T F S S M T Ja n 7, '01 W T F S

20

15

10

10

10

10 Overa ll ocated:

10

10

10

10

10

10

10

16

16

16

16

16

21

21

21

Work ers

Al l oca te d:

Renewable Resource Allocation Example (Single Resource Type)

3 workers 6 workers

Task A 4 wks

Task C 1 wk

START

Task E 4 wks Task B 3 wks

5 workers

END

Task D 5 wks
8 workers

7 workers

Maximum number of workers available = R = 9 workers

Resource Allocation Example: Early Start Schedule

Tas k A: 3 work ers Start End Tas k B: 5 work ers Tas k D: 8 work ers Tas k E: 7 work ers Tas k C: 6 work ers

Week
No. of Work e rs/wk Cumulative Work ers "Waste d" work er-wk s

1
8 8 1

2
8 16 1

3
8 24 1

4
11 35 -

5
14 49 -

6
8 57 -

7
8 65 -

8
8 73 -

9
7 80 -

10
7 87 -

11
7 94 -

12
7 101 -

Maximum number of workers available = R = 9 workers

Resource Allocation Example: Late Start Schedule

Tas k A: 3 work ers Start Tas k C: 6 work ers Tas k B: 5 work ers Tas k D: 8 work ers Tas k E: 7 work ers End

Week
No. of Work e rs/wk Cumulative Work ers "Waste d" work er-wk s

1
5 5 -

2
5 10 -

3
5 15 -

4
11 26 -

5
11 37 -

6
11 48 -

7
11 59 -

8
14 73 -

9
7 80 2

10
7 87 2

11
7 94 2

12
7 101 2

Maximum number of workers available = R = 9 workers

Resource Allocation Heuristics

n

Some heuristics for assigning priorities to available tasks j, where number of units of resource k used by task j 1) FCFS: Choose first available task

Rk j denotes the

n n n n n

2) GRU: (Greatest) resource utilization =

Rk j
k

3) GRD: (Greatest) resource utilization x task duration = 4) ROT: (Greatest) resource utilization/task duration = 5) MTS: (Greatest) number of total successors

Rk j k Rk j / tj
k

tj

n
n n n n n

6) SPT: Shortest processing time = min {tj}

7) MINSLK: Minimum (total) slack 8) LFS: Minimum (total) slack per successor 9) ACTIMj: (Greatest) time from start of task j to end of project = CP - LSj 10) ACTRESj: (max) (ACTIMj) 11) GENRESj: w ACTIMj + (1-w) ACTRESj where 0 w 1

Resource Allocation Problem #2

Tas k A1 6 days

Tas k A2 4 days

Start

Tas k B1 3 days

Tas k B2 5 days

End

Tas k C1 2 days

Tas k C2 5 days

Gold Crew

Purple Crew

How to schedule tasks to minimize project makespan?

Priority scheme: schedule tasks using total slack (i.e., tasks with smaller total slack have higher priority)

Gold Crew

Task A1 1 2 3 4 5 6 7

Task B1 8 9

Task C1 10 11 12 13 14 15 16 17 18 19 20

Purple Crew

Task A2 1 2 3 4 5 6 7 8 9 10 11 12

Task B2 13 14 15 16 17

Task C2 18 19 20

Resource Allocation Example (contd)

But, can we do better? Is there a better priority scheme?

Gold Crew

10

11

12

13

14

15

16

17

18

19

20

Purple Crew

10

11

12

13

14

15

16

17

18

19

20

Microsoft Project Solution (Resource Leveling Option)

Solution by: Microsoft Project 2000

Critical Chain Project Management

Identify the critical chain: set of tasks that determine the overall duration of the project Use deterministic CPM model with buffers to deal with uncertainty Remove padding from activity estimates (otherwise, slack will be wasted). Estimate task durations at median. Place project buffer after last task to protect customers completion schedule Exploit constraining resource(s) Avoid wasting slack times by encouraging early task completions Have project team focus 100% effort on critical tasks Work to your plan and avoid tampering Carefully monitor and communicate buffer status

Critical Chain Buffers Project Buffer

: placed after last task in project to protect schedule

Feeding Buffers

: placed between a noncritical task and a critical task

when the noncritical task is an immediate predecessor of the critical task

Resource Buffers
resource type

: placed just before a critical task that uses a new

Critical Chain Illustrated

Feeding Buffers
Tas k C1 2 days Tas k B1 3 days Tas k A1 6 days

Start

End

Tas k C2 5 days

Tas k B2 5 days

Tas k A2 4 days

Resource Buffers

Non-Renewable Resources
12 units

Task B 5 wks
6 units 8 units

ST ART

Task A 6 wks Task C 3 wks

10 units

Task D 2 wks

END

Tas k A B C D

Duration 6 5 3 2

No. of Nonrene wable Resource s Units Needed 6 12 10 8

Early Start 0 6 6 11

Late Start 0 6 8 11

Non-Renewable Resources: Graphical Solution

Cumulative Res ources Supplied

40 36

Cumulative Resources

32 28 24 20 16 12 8 4

Cumulative Res ources Required

10

11

12

13

14

15

16

17

18

19

20

Weeks

Resource Allocation Problem #3

Issue: When is it better to team two or more workers versus letting them work separately?
Have 2 workers, Bob and Barb, and 4 tasks: A, B, C, D Bob and Barb can work as a team, or they can work separately When should workers be assigned to tasks? Which configuration do you prefer?

How to Assign Project Teams?

A Start B D C End

Configuration #1
Bob and Barb work jointly on all four tasks; assume that they can complete each task in one-half the time needed if either did the tasks individually

Configuration #2
Bob and Barb work independently. Bob is assigned to tasks A and C; Barb is assigned to tasks B and D

Bob and Barb: Configuration #1

TASK A Dur ation 6 5 4 Expected dur ation Pr ob 0.33 0.33 0.33 5.0 TASK B Dur ation 9 6 Pr ob 0.667 0.333 TASK C Dur ation 12 7 Pr ob 0.6 0.4 TASK D Dur ation 10 6 Pr ob 0.25 0.75

8.0

10.0

7.0

Configuration #1

Bob and Barb work jointly on all four tasks.

What is the expected project makespan?

Bob and Barb: Configuration #2

Bob and Barb work independently. Bob is assigned to tasks A and C; Barb is assigned to tasks B and D
Bob A+C 18 18 13 13 18 18 13 13 17 17 12 12 17 17 12 12 16 16 11 11 16 16 11 11 Barb B+D 19 15 19 15 16 12 16 12 19 15 19 15 16 12 16 12 19 15 19 15 16 12 16 12 m ax (A+C, B+D) 19 18 19 15 18 18 16 13 19 17 19 15 17 17 16 12 19 16 19 15 16 16 16 12

Re alization #

A 6 6 6 6 6 6 6 6 5 5 5 5 5 5 5 5 4 4 4 4 4 4 4 4

B 9 9 9 9 6 6 6 6 9 9 9 9 6 6 6 6 9 9 9 9 6 6 6 6

C 12 12 7 7 12 12 7 7 12 12 7 7 12 12 7 7 12 12 7 7 12 12 7 7

D 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10 6 10 6

Pr ob 0.03 0.10 0.02 0.07 0.02 0.05 0.01 0.03 0.03 0.10 0.02 0.07 0.02 0.05 0.01 0.03 0.03 0.10 0.02 0.07 0.02 0.05 0.01 0.03

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

Bob and Barb: Configuration #2

Bob and Barb work independently. Bob is assigned to tasks A and C; Barb is assigned to tasks B and D

m ax (A+C, B+D) 12 13 15 16 17 18 19

Pr ob 0.07 0.03 0.20 0.20 0.17 0.17 0.17

Cum ulative Pr ob 0.07 0.10 0.30 0.50 0.67 0.83 1.00

Expected Project Makespan: 16.42

Parallel Tasks with Random Durations

Task A
START END

Task B

Assume that both Tasks A and B have possible durations: 8 days with probability = 0.5 10 days with probability = 0.5

What is expected duration of project? (Is it 9 days?)

Project Monitoring and Control

n

It is of the highest importance in the art of detection to be able to recognize, out of a number of acts, which are incidental and which are vital. Otherwise your energy and attention must be dissipated instead of being concentrated.
Sherlock Holmes

Status Reporting?
One day my Boss asked me to submit a status report to him concerning a project I was working on. I asked him if tomorrow would be soon enough. He said, "If I wanted it tomorrow, I would have waited until tomorrow to ask for it!"
New business manager, Hallmark Greeting Cards

Control System Issues

n n

What are appropriate performance metrics? What data should be used to estimate the value of each performance metric? How should data be collected? From which sources? At what frequency? How should data be analyzed to detect current and future deviations?

How should results of the analysis be reported? To whom? How often?

Controlling Project Risks

Key issues to control risk during projecct: (1) what is optimal review frequency, and (2) what are appropriate review acceptance levels at each stage? Both over-managed and under-managed development processes result in lengthy design lead time and high development costs.
Ahmadi & Wang. Managing Development Risk in Product Design Processes, 1999

Project Control & System Variation

Common cause variation: in-control or normal variation Special cause variation: variation caused by forces that are outside of the system
According to Deming: Treating common cause variation as if it were special cause variation is called tampering Tampering always degrades the performance of a system

Control System Example #1

n

Project plan: We estimate that a task will take 4 weeks and require
n

1600 worker-hours

At the end of Week 1, 420 worker-hours have been used

Is the task out of control?

Control System Example (contd)

Week 2: Task expenses = 460 worker-hours
Planned Cost Week (BCWS) 1 400 2 400
470 460

Actual Cost 420 460

Cumulative Actual Cost (ACWP) 420 880

Cost (in worker-hours)

450 440 430 420 410 400 390 380 370 1 2 We ek 3 4

Is the task out of control?

Control System Example (contd)

Week 3: Task expenses = 500 worker-hrs
We ek 1 2 3 Planned cost (worker-hours) 400 400 400
600 500

Actual cost (worker-hours) 420 460 500

Cumulative cost (worker-hours) 420 880 1380

Worker-hours

400 300 200 100 0 1 2 We ek 3 4

Is the task out of control?

Earned Value Analysis

Integrates cost, schedule, and work performed Based on three metrics that are used as the basic building blocks:

BCWS: Budgeted cost of work scheduled

ACWP: Actual cost of work performed

BCWP: Budgeted cost of work performed

Schedule Variance (SV)

Schedule Variance (SV) = difference between value of work completed and value of scheduled work

Schedule Variance (SV) = Earned Value - Planned Value = BCWP - BCWS

Cost Variance (CV)

Cost Variance (CV) = difference between value of work completed and actual expenditures

Cost Variance (CV) = Earned Value - Actual Cost

= BCWP - ACWP

Earned Values Metrics Illustrated

Present time Planned Value (BCWS) BAC

Worker-Hours

Actual Cost (ACWP) Earned Value (BCWP)

Cost Variance (CV)

Schedule Variance (SV)

Week 1

Week 2

Week 3

Week 4

Week 5

Week 6

Relative Measure: Schedule Index

Schedule Index (SI ) = BCWP BCWS

If SI = 1,

then task is on schedule

If SI > 1,
If SI < 1,

then task is ahead of schedule

then task is behind schedule

Relative Measure: Cost Index

Cost Index (CI) = BCWP ACWP

If CI = 1, If CI > 1,

then work completed equals payments (actual expenditures) then work completed is ahead of payments

If CI < 1,

then work completed is behind payments (cost overrun)

Example #2
W E E K
1 2 3 4 5 6 7 8 9 10
Tas k A (36 work e r-hrs) 6 6 6 8 10

Tas k B (36 work er-h rs) 12 12 12 Tas k C (56 worke r-h rs) 10 Weekly Scheduled Worker-Hrs Cumulative Scheduled Worker-Hrs (BCWS) 10 12 12 12

20

22

22

10

12

12

12

12

18

38

60

82

92

104

116

128

Example #2 (contd)
Progress report at the end of week #5: Cumulative Percent of Work Completed:
Week

2
30%

3
40%

4
60% 25%

5
80% 65%

Task A 15% Task B Task C

Not started yet

Worker-Hours Charged to Project:

Week

1
5

2
6

3
8 Not started yet

4
10 15

5
10 10

Task A Task B Task C

Example #2 (contd)
Progress report at the end of week #5:

W E E K
1
Cumulative Sche duled Work er-Hrs (BCWS) Actual Work erHrs Use d (ACWP) Earned Value (BCWP) Sche dule Variance (SV) Cost Varian ce (CV)

10

6 5 5.4 -0.6 0.4

12 11 10.8 -1.2 -0.2

18 19 14.4 -3.6 -4.6

38 44 30.6 -7.4 -13.4

60 64 52.2 -7.8 -11.8

82

92

104

116

128

Example #2 (contd)
140 BAC 120 BCWS

100

Performance Metric

80 Cost Variance 60 ACWP 40 BCWP 20 Sch edule Variance

0 1 2 3 4 5 We ek 6 7 8 9 10

Using a Fixed 20/80 Rule

Cumulative Percent of Work Completed:
Week

2
20%

3
20%

5
20% 20%

Task A 20% Task B Task C

20% 20% Not started yet

W E E K
1
Cumulative Sche duled Work er-Hrs (BCWS) Actual Work erHrs Use d (ACWP) Earned Value (BCWP) Sche dule Variance (SV) Cost Varian ce (CV)

10

6 5 7.2 1.2 2.2

12 11 7.2 -4.8 -3.8

18 19 7.2 -10.8 -11.8

38 44 14.4 -23.6 -29.6

60 64 14.4 -45.6 -49.6

82

92

104

116

128

Using a Fixed 20/80 Rule

140 120

100
Cost (in Worke r-hour s)

80

BCWS ACWP
60

40

BCWP
20

0 1 2 3 4 5 Week 6 7 8 9 10

Updating Forecasts: Pessimistic Viewpoint

Assumes that rate of cost overrun will continue for life of project.

Estimate at Completion (EAC) = ACWP BAC = 1 BAC . BCWP CI

= (64/52.2) 128 = 1.23 x 128 = 156.94 worker-hrs

Updating Forecasts: Optimistic Viewpoint

Assumes that cost overrun experienced to date will cease and no further cost overruns will be

experienced for remainder of project life

Estimate at Completion (EAC) = BAC - CV = 128 + 11.8 = 139.8 worker-hrs .

Multi-tasking with Multiple Projects

How to prioritize your work when you have multiple projects and goals? Consider two projects with and without multi-tasking
Project A Project B

A-1

B-1

A-2

B-2

A-3

B-3

A-4

B-4

Due-Date Assignment with Dynamic Multiple Projects

Projects arrive dynamically (common situation for both

manufacturing and service organizations)

How to set completion (promise) date for new projects? Firms may have complete control over due-dates or only partial

control (i.e., some due dates are set by external sources)

How to allocate resources among competing projects and tasks (so that due dates can be realized)? What are appropriate metrics for evaluating various rules?

What Does the Research Tell Us?

Study by Dumond and Mabert* investigated four due date assignment rules and five scheduling heuristics

Simulated 250 projects that randomly arrive over 2000 days

average interarrival time = 8 days 6 - 49 tasks per project (average = 24); 1 - 3 resource types average critical path = 31.4 days (range from 8 to 78 days) Performance criteria: 1) mean completion time 2) mean project lateness 3) standard deviation of lateness 4) total tardiness of all projects Partial and complete control on setting due dates
* Dumond, J. and V. Mabert. Evaluating Project Scheduling and Due Date Assignment Procedures: An Experimental Analysis Management Science, Vol 34, No 1 (1988), pp 101-118.

Experimental Results
No one scheduling heuristic performs best across all due date

setting combinations
Mean completion times for all scheduling and due date rules not significantly different FCFS scheduling rules increase total tardiness SPT-related rules do not work well in PM (SASP) Best to use more detailed information to establish due dates

Project Management Maturity Models

Methodologies to assess your organizations current level of

PM capabilities
Based on extensive empirical research that defines best practice database as well as plan for improving PM process Process of improvement describes the PM process from ineffective to optimized Also known as Capability Maturity Models

PM Maturity Model Example*

1) Ad-Hoc
The project management process is described as disorganized, and occasionally even chaotic. Systems and processes are not defined. Project success depends on individual effort. Chronic cost and schedule problems.

2)

Abbreviated: Some project management processes are established to track cost, schedule,
and performance. Underlying disciplines, however, are not well understood or consistently followed. Project success is largely unpredictable and cost and schedule problems are the norm.

3)

Organized: Project management processes and systems are documented, standardized, and
integrated into an end-to-end process for the company. Project success is more predictable. Cost
and schedule performance is improved.

4) Managed: Detailed measures of the effectiveness of project management are collected and used
by management. The process is understood and controlled. Project success is more uniform. Cost and schedule performance conforms to plan.

5) Adaptive:

Continuous improvement of the project management process is enabled by feedback

from the process and from piloting innovative ideas and technologies. Project success is the norm. Cost and schedule performance is continuously improving. * source: The Project Management Institute PM Network (July, 1997), Micro Frame Technologies, Inc. and Project Management Technologies, Inc. (http://pm32.hypermart.net/)