Engineering Management 1 Lecture 2: Project Scheduling I Ruel Ellis Department of Mechanical and Manufacturing Engineering Network Scheduling Critical Path Method Program Evaluation & Review Technique Issues in Network Scheduling
Net w or k S cheduling Net w or k Scheduling Phases Activity Definition Activity Sequencing Duration Estimating Do what? Activity r elations hip? Duration? I n what order? Adj ust schedule performance? Project Scheduling Schedule Control
Act ivit y Definition: S t ar t ing Point Must define all activities needed to complete pr oj ect What are natur al starting points for activity definition? Outputs of activity definition phase Activity list S uppor ting documentation T he Wor k Breakdown S tructure (WBS ) Act ivit y Sequencing: Def ining R elationships Activities typically have inter dependencies Pr ecedence cons tr aints Nature of precedence r equir ements
Act ivit y Sequencing: Def ining R elationships Appr oaches Experience Hier ar chical Yellow s ticky Defining pr ecedence relationships challenging in complex projects Dur at ion/ Cost E stimation Strategies Top-Down Estimating Bottom-up Estimating Level of Effort Standard Costs and Time Historical Relationships Simpson s Rule
T op-dow n Estimating I dentify large blocks of effort from the WBS (task/subtask level) Estimate cos ts and time S imilar work/expert assessment/cost es timating r elations hips S trength Quick Weakness May be inaccur ate and/or less precise B ot t om-u p Estimating P ro je ct I dentify small blocks of effort from the WBS (work package/activity level) Level 1 Level 1 S um for higher level cost estimates S trength Greater confidence in small increment es timation Level 2 Level 2 Level 2 Weakness More time cons uming May stack-up Level 3 contingency Level 3 planning Level 3 by estimators
L evel of Effort Appr oach Perform at high level I dentify individuals needed Number and j ob clas s ification Estimate period each individual needed How long will each individual be needed L evel of Effort S trength Quick Weaknes s Focus on people NOT tasks needed done May be inaccur ate and/or less precise
S t andar d Costs and T ime Known/Repeatable activities and s tandar d oper ations Little var iance in similar operations Not dependent on particular proj ect S t andar d Costs and T ime S trength Accur acy increased, variance decreased Weakness Assumes standard activities Assumes historical record available and current
H istorical Relationships Future = f(past) or f(parameter) Stable ratios may exist -- for example: documentation, known activities, or prior efforts (C 1 /C 2 )=(Lines of code 1 /Lines of code 2 ) n Cost Estimating Relationships -- duration may vary as a function of some $ variable cost = f(weight of spacecraft) cost=f(square feet of interior space) S imps on s Rule for E stimation T ime or cost may vary from nominal Can lessen anxiety about es timating Uses nominal, minimum, and mos t likely values to es tablis h estimate X exp = X min + 4X nom + X max 6
P r oj ect Scheduling What is the pur pos e of project s cheduling? T he WBS gives no s equencing of the different work packages T he WBS gives no r elations hip between the wor k packages Must develop the logical relationships between work activities By hand --- computers? Project Scheduling Approaches Most common approaches Milestone Chart Gantt CPM (Critical Path Method) PERT (Prog Eval and Rev Technique) PDM (Precedence Diagramming Method) Other approaches - (sim, GERT,TOC)
Milestone and Gantt Char ts: Milestone Char ts Depict principal project milestones (events) Good Over view T ool Great for progress reports Milestone and Gantt Char ts: Gantt Charts (aka B ar Charts) S how activity start and s top times Do not show precedence r elations hips May but becomes cumbersome Great management presentation tools
P r oj ect Net w or ks Net w or k Analysis T wo appr oaches Activity on Arrow (AOA) Activity on Node (AON) More typical definition CPM - deterministic pr oj ect schedule PERT - probabilistic pr oj ect schedule
Act ivit y on Node: F undamental Relationships A B C J Y K M X L Z J L K M Act ivit y on Arrow: F undamental Relationships A 2 7 3 B A 2 7 F 14 3 6 A B 7 F R 3 6 6 D
Net w or k S cheduling Cr it ical Path Met hod (CPM) Project Networks: B as ic R u les Flow from left to r ight Activity cannot begin until connected activities have completed Arrows indicate pr ecedence & flow Each activity has unique I D Looping (or cycling) not allowed Conditional statements not allowed Have unique s tar ting point
E xample: T he Engineer ing Center Activity D e scriptio n P redece sso rs D ura tio n (da ys ) A Application approval N one 5 B C onstruction plans A 15 C Traffic study A 10 D Service availability check A 5 E Staff report B,C 15 F C ommission approval B,C,D 10 G C enter Construction F 170 H Occupancy E,G 35 I mplement ing CPM: Act ivit y on Arrow Will illustrate AOA CPM on the Engineer ing Center example T ake s pecial note of: Dummy activity Forward and backwar d pass Earliest/latest Start times Earliest/latest Finish times Concept of slack (float) Concept of critical path Assumption of unlimited r es our ces
Act ivit y on Arrow Not at ion A(ES,EF) 1 2 Dur(LS,LF) AOA vs. AON: A Matter of Preference AON** Advant ages No dummy activities Events not used Easy to dr aw Activity emphas is AOA Advant ages Path tracing easy Easier to dr aw for complex proj ects Highlights key events
AOA vs. AON: A Matter of Preference AON** Dis advant ages Path tracing difficult Network drawing may be complex AOA Dis advant ages May require dummy activities Detracts from activities P r ogr am E valuat ion and R eview T echnique ( P E R T )
P E R T versus CPM CPM is det er minis t ic One dur ation for each project activity Proj ect duration therefore deterministic one value PERT is pr obabilistic T hree dur ation for each project activity to es tablish expected dur ation Proj ect duration therefore pr obabilistic a r ange of possible values for proj ect duration P E R T versus CPM Both develop a cr itical path Both assume unlimited (unconstrained) resources PERT provides for statistical analysis of project durations
P E R T : Not hing is Certain S imilarity to CPM method Appr oach to activity duration es timation Role of the cr itical path Proj ect duration stochastic Use of the CLT to assume nor mality Determining mean and var iance Calculating P(pr oj ect duration > X) Under lying as s umptions Analys is of t he P ER T Net w or k - E XAMP L E
P E R T Analysis Example (b-a)/6 (a Will + 4m demonstrate + b)/6 concepts using example 3 1 2 5 6 4 Activity a m b E(duration) Std Dev (duration) 1-2 17 29 47 30 5 2-3 6 12 24 13 3 2-4 16 19 28 20 2 3-5 13 16 19 16 1 4-5 2 5 14 6 2 5-6 2 5 8 5 1 P E R T C P at h Var iance I F YOU ASSUME INDEPENDENCE the var iance of any path = sum of activity variances for all activities on that path NORMALLY DI S T RI BUT ED Variance of the PROJECT = variance of the CRI T I CAL PATH I f more than one cr itical path, PROJECT VARIANCE=largest of CRI T I CAL PATH VARIANCES
P E R T Variance S ince NORMALLY DI S T RI BUT ED can estimate pr obability of completing pr oj ect on time can estimate pr obability of completing pr oj ect by any target date if critical path expected = 64, STD DEV= 6 T arget= 70 Z = (70-64)/6 = 1 probability = 1 -.84 =.16 P E R T Estimates S o what do you mean by optimistic, pessimistic? value you expect to be exceeded at a probability level and not exceeded at 1-α probability PROBLEM: estimating the MOS T LI KELY duration of most things is hard Asking es timator s to come up with What won t be ex ceeded 95% of the time is a guess at best.
I ssues in Net w or k S cheduling S hor t Exercise Net w or k Scheduling I dentify three cr itical issues, assumptions, or limitations in the use of network diagrams as a pr oj ect management tool?
Project Scheduling I ssues A typical project has less than 10% of activities which are identified as critical Point vs. probabilistic estimates Project Scheduling I ssues CPM or PERT: Research on PERT /CPM pr oj ects T echnical performance -- No significant difference Probability of cost & s chedule overruns -- S ignificant difference Adequacy of estimates & WBS Unconstrained r es our ces