USER'S GUIDE for DSM@MIT
Contents
1. this sequence change Figure 12 illustrates this procedure in the AEAP worksheet al alel el el a iolt iapiapie 1 Select the cell i j in the lower diagonal_y 2 Execute Swap the sequences to change the sequence of tasks i and j FIGURE 12 SWAPPING TASK SEQUENCES FROM AEAP WORKSHEET 14 Soo Haeng Cho soohaeng_cho yahoo com 3 3 6 Move tasks with nonzero level slack A task with nonzero level slack can be placed in a different hierarchy between ts level in the AEAP DSM and that in the ALAP DSM without affecting the structure of the project If the task is planned to start with some lag i e later than the as early as possible schedule the DSM can show the better structure of the project by moving it to the level where tasks in the same level are possibly performed in parallel with it The tool allows a user to move those tasks and create different DSM views in Customized and Cus collapsed worksheets This command must be executed in the Analysis worksheet as illustrated in Figure 13 1 Select the cell i j where task j with nonzero level slack is going to be moved downstream to the level of task i within a dotted box 2 Press Ctrl and select other cells to be moved at the same time 3 Execute Move tasks with nonzero AA 8 CIO EF GI2. ajja alal wn 4 4 o w m n e w Nn b c d 6 f moj og h j k m n 14 15 165 Eid 18 o a9 1R MIAP DSM Input FIGURE 4 INPUTS FOR STRUCTURING MODULE 3 2 Analyses in the Structuring Module By executing Analyze DSM in the menu the tool performs analyses in the structuring module The results are shown in the five worksheets AEAP ALAP AEAP collapsed ALAP collapsed and Analysis Figures 5 through 7 show the analyses results of the sample in Figure 4 8 Soo Haeng Cho soohaeng_cho yahoo com Sample for User s Guide 5 Number of tasks in the DSM FIGURE 5 AEAP AND AEAP COLLAPSED WORKSHEETS E Sample for User s Guide dali Block1 ABA N w prg rg ajo jejo VO FIGURE 6 ALAP AND ALAP COLLAPSED WORKSHEETS 9 Soo Haeng Cho soohaeng_cho yahoo com By partitioning a project iteration loops cycles are identified and tasks are rearranged A coupled block represents a set of tasks constituting a loop within which there is at least one dependency mark in the upper diagonal of the matrix Due to its cyclic relationships among constituent tasks iterations may take place potentially among the tasks within a coupled block A level is determined such that tasks in the same level constitute a coupled block or they are independent of other tasks not belonging to the same coupled
3. 39 5 43 3 48 56 9 Major Critical Sequences with Leveled Resources Major Critical Paths without Resource Constraints 9 11 10 17 18 19 69 2 10 18 19 52 2 10 18 19 18 9 11 17 10 18 19 lt Note gt Task indices are based on the indexing system in Project worksheet FIGURE 21 LEAD TIME AND CRITICAL PATHS The tool also computes conventional and resource constrained slacks and criticalities of tasks In addition the measures computed for determining resource priorities are provided which reveal relative importance between tasks Table 7 shows an example of these measures Task ID Taskname Stack avg s RC_Slack avg sd ao o e T eS T T 2 Bocan 8 57 ss 7373 20 70 oo o0 15 DBioo2 Duration Jo T T T T T T 16 aa PC E CT MARCHI pz ft __ seas l 1849 1 o 90 1 30 68 Ee e o 10 o0 10 ae e ee lt Note gt Be careful not to print all Project worksheet since there exist data not seen in the screen Table 7 Measures for Task Criticalities The tool also draws three simulated Gantt charts out of many scenarios in Gantt Chart P1 worksheet for the results without resource constraints and in Gantt Chart P2 worksheet for the results with resource constraints Figure 22 shows the examples of a simulated Gantt Chart without resource constraints in a and with resource constraints in b The highlighted tasks indicate that they are on critical path s In
4. 32 S s vases ee eee AAA AAA AAA KAWAWA 18 ARI RESOULCES ki tase iest 19 BAB Iteration issia rriste aa se cctavasd osadensa tend si ES PRSES a ESEA kada 20 34 3 1 Sequential Ieraton eere akaa Es AREVA eo iEn OTENE Es hai miba 21 343 2 NA YE ay 0 10 E ko AA AA AA AAA 22 3 5 Analyses in the Modeling Module cccccececececccecececececececececececeeeeeeeeeeeeeeeeeeeeeeeeeeees 23 3 5 1 Analyses Results for Coupled BlockS wwwmmennananananmnnananananwmaananawwa 23 3 5 2 Analyses Results for the Project cccccececccececececececececeeeceeeeeeeceeeeeeeeeeeeeeeeeeeeeeess 24 3 6 Transferring Analyses Results from the Structuring amp Modeling Modules to the Schedulins Modules aa KK Kia Kasai kawai kei kakalia kasa ka kaakaa 26 A OPTIONS EEE OE E E EE UAE 28 4 1 Display Options ai khasa haka ameita mh a a 28 ALA Column SiE 25 sic 5 ce Sasoves chance seshad inci ctasanstdeadin ss EEE T E EEE E Aa 28 Ae V2 1O yaa AAA AAA ata cudaes AAA hand oe obe Sedans odes ded AS 28 4 1 3 Conversion in Analysis Sheets sities er hil cc aoe eR ead E E ed 29 AD Resource Option nininini coding tee beeps Sollee a Tekan deh i ama 30 42 Within BIOCKS EE GA shea are a US esa eas 30 422 OUtSIdE BIOCKS cxsasins cacediendaPasstaeaeasgnasdoass ehaceean n ae ia taa 30 4 3 Simulation Option iii cscccccscsscsccsssasccckassoaccoatacccusabscneccubavscebetacancoadanceceabasnecoubanscesataas 31 4 3 1 Latin Hypercube Samp
5. Basic Editor Choose Modules folder under ProjectGlobal GLOBAL MPT in the Project Explorer Window Right click and choose Import File and select DSM MIT bas from the folder in which you saved this file Create a map DSM to Gantt Chart in Microsoft Project Start Microsoft Project From Menu choose Open Select Microsoft Excel Workbooks from the drop down menu Files of type and then the excel file containing a sample application Open the file Click New Map Type DSM to Gantt Chart in the Import Export map name field and check Tasks Resources and Export header row Import includes headers in the Option tab Click O K 5 Soo Haeng Cho soohaeng_cho yahoo com 5 In the Task Mapping tab select MS Project from the drop down menu of Source worksheet name Match the fields From Worksheet FieH to To Microsoft Project as follows To Microsoft Project From Worksheet Field ID Task ID Name Task Name Duration Likely Duration Predecessors Predecessors Resource Names Resource Names Priority Priority lt Note gt Make sure deselect Complete from the Worksheet Field 6 In the Resource Mapping tab select Resource from the drop down menu of Source worksheet name Match the fields From Worksheet Field to To Microsoft Project as follows To Microsoft Proj
6. Compute lead time with resource constraints With the second option selected the tool creates the worksheet when executing Compute lead time without resource constraints With the third option selected the tool does not create the worksheet 33 Soo Haeng Cho soohaeng_cho yahoo com APPENDIX Definitions of Heuristic Priority Measures Al Rank Positional Weight RPW A rework adjusted rank positional weight of task i is defined as follows rpw d Yd J where d rework adjusted duration of task i which is defined as the expected value of the sum of the duration of its first execution and the total amount of successive rework it creates for its predecessors assuming no resource constraints in the project network La sum of all rework adjusted durations over all successors of task i j Note a set of successors includes all downstream tasks that receive outputs from the task By adding the summation part of the rank positional weight it measures global importance of a task while it measures only local importance among neighbor tasks without it A2 Cumulative Resource Equivalent Duration CUMRED Cooper defined a cumulative resource equivalent duration of task i of each resource as follows RUD L E R where RUD resource usage duration of resource k L last time period of requirement for resource k E total excess requirement for resource k R total available amount of resource k RUDMAX
7. Maximum RUD Where k 1 2 K number of resources K RED r R RUD RUDMAX d k 1 where RED resource equivalent duration of task i r amount of resource k required by task i CUMRED RED RED j where CUMRED cumulative resource equivalent duration of task i L RED Sum of RED over all successors of task i J j 4 Cooper D Heuristics for Scheduling Resource Constrained Projects An Experimental Investigation Management Science Vol 22 No 11 pp 1186 1194 1976 34 Soo Haeng Cho soohaeng_cho yahoo com
8. The copyright of this program belongs to Soo Haeng Cho Steven D Eppinger and Massachusetts Institute of Technology This program is protected by copyright law and international treaties Unauthorized reproduction or distribution of this program or any portion of it may result in severe civil and criminal penalties and will be prosecuted to the maximum extent possible under the law The integrated method streamlines project planning and control using three modules structuring modeling and scheduling as illustrated in Figure 1 In the structuring module the design structure matrix DSM method is used to structure the information flows among tasks and capture the iteration loops By classifying various types of information dependencies a critical dependency path is identified and redundant constraints are removed for the modeling and scheduling analyses In the modeling module a generalized process model predicts complex behaviors of iterative processes using advanced simulation techniques such as the Latin Hypercube Sampling and parallel discrete event simulation The model computes the probability distribution of lead time and identifies critical paths in a resource constrained iterative project Using the results of analyses from the structuring and modeling modules a network based schedule in the form of a PERT Gantt chart is developed in the scheduling module The schedule is used as the basis for monitoring and control of the pro
9. b tasks Block2 m and i compete for the limited resource and the tool determined priorities based on the rules explained earlier 25 Soo Haeng Cho soohaeng_cho yahoo com Task ID TaskName Simulated Duration TJ f T Pi fa TT esb ie ee a KAA kaa ei op OO a E D aa Pig fn T se J T J ee State aA A a at 5 YA al Siste Duaton sofa _47 _as 2i___7at___aat_e TotalLeadTime 4asf T a Without Resource Constraints Saistuston sal asl asfasl rool ar aa Y Total Lead Time sia _ b With Resource ee lt Note gt Tasks a and b are completed and Block and e are in progress FIGURE 22 SIMULATED GANTT CHARTS OF A PROJECT 3 6 Transferring Analyses Results from the Structuring amp Modeling Modules to the Scheduling Module When executing Analyze Project the tool creates MS Project worksheet Using the data map created in INSTALLATION section a user can transfer the data in this worksheet to Microsoft Project as follows 1 Start Microsoft Project 2 From Menu choose Open Select Microsoft Excel Workbooks from the drop down menu Files of type and then the excel file having analyses results from the structuring and modeling modules Open the file 3 Choose DSM To Gantt Chart for Import Export map to use for importing and Open Then a user must have a Gantt chart in Microsoft Pr
10. block This multi level structure presents a simple hierarchy of a project as well as concurrency of tasks independent tasks and or blocks in the same level can work in parallel In AEAP tasks are sequenced based on the as early as possible rule which assumes that a task starts immediately after all the inputs necessary to begin are available In ALAP tasks are sequenced based on the as late as possible rule in which the starting time of a task is delayed to the extent that it does not delay entire project AEAP collapsed and ALAP collapsed worksheets show the collapsed views of the DSMs in AEAP and ALAP respectively where coupled blocks are collapsed to block tasks In the collapsed views information flow from a single task to a block task is marked as 2 when there exists at least one second type flow between the single task and any constituent task within the block in the DSM In contrast any information flow from a block task to any single or block task is regarded as the second type flow as it is very likely that preliminary outputs of tasks within the block are transferred to downstream tasks before converging to their final forms at the end of iterations Ef Sample for User s Guide eee _ CID E FIGIH I JIK LIMIN Ol POS Number of elements in the collapsed DSM i lt note gt index based on AEAP collapsed FIGURE 7 ANALYSIS WORKSHEET 10 Soo Haeng Cho so
11. com 4 OPTIONS 4 1 Display Option Option FIGURE 25 DISPLAY OPTION 4 1 1 Column Size This option gives a user to adjust cell sizes in worksheets directly DSM entry cell is the width of cells in the DSM Level column is the width of columns showing levels Task name column is the width of columns showing task names Gantt Chart unit bar is the width of a bar for one unit of duration in simulated Gantt charts 4 1 2 Border If the first checkbox is selected the tool draws borders around diagonal cells in the DSM as shown in Figure 26 With the second checkbox selected by default the tool draws borders when running Analyze DSM as shown in Figure 27 28 Soo Haeng Cho soohaeng_cho yahoo com Oe ame yame Rano we tuu web Cee Aa OCR Baw OL oma o NAMERE SASNAL T ET SEA FELE EG HILEL D 1m wawa tanta index of a task Taskname Levei 1 2 f 4 5 ef 7 9 ja a Bo Ser b In the Worksheet AEAP FIGURE 26 BORDERS AROUND DIAGONAL CELLS ___ Task Name Level fal 2l sf al s ef 7 ja E FIGURE 27 BORDERS IN DSM VIEWS 4 1 3 Conversion in Analysis Sheet As explained in Analyze DSM if non binding dependencies between tasks are marked as 2 in AEAP and ALAP collapsed views they are converted from 2 to I blue marks in the Analysis worksheet by default With this checkbox unselected the tool
12. in the DSM Input worksheet S e Pragati tare HOTE 1 Erber Proci aso a 2 4 Proact Marsa WOTE gt 1 Enter Pyopact Mate ct 2 A angie 3 freer attack nares for entries n EN Sarok 2 Erter aitad ranar tor ertrias h DSN Teas Neree 12345 6 6 910111213 1415 Tew rare 123 5 6 6 10 11 12 13 14 m i 8 J b 2 gt 2 lt 3 E 3 vi s a 5 10 1 k 12 1 13 m 14 fi 15 i 10 t ti 17 n 13 n 14 1 Select the row 2 Execute Insert a new task 4 Execute Analyze DSM lt 3 Enter a task name and information dependencies FIGURE 9 INSERTING A NEW TASK IN DSM INPUT WORKSHEET 3 3 2 Delete a task in the current row A user can delete a task from DSM Input or AEAP worksheet After deleting the selected task the tool automatically performs Analyze DSM Figure 10 illustrates this procedure in the AEAP worksheet tarer of foes nine OSV spate new nee uses eames ates nies pent T Fintor of tak n Po DEM E antes nra ndio ued Tast Hi ss mea Tak Nore T 2 3 alale z ol apopupafis e HEHEH aE pa 1 fiaj iF jaj jpe Abit J 1 1 1 12 3 4 5 6 a 8 shomi 1 Select the row gt 2 Execute Delete a task FIGURE 10 DELETING A TASK FROM AEAP WORKSHEET 13 Soo Haeng Cho sooha
13. is a binding dependency which is not critical binding Following critical binding dependencies along the tasks a critical dependency path goes through the entire process hierarchies of a project from the first level to the last level This path is different from a critical path because it is determined without considering time aspects However this path provides guidance for process improvements in the early stage of planning processes when detailed data for durations are not available In Figure 7 the black or blue marks at 6 3 8 4 10 6 7 3 and 10 5 represent non binding dependencies The marks in first three cells are converted from 2 in the AEAP collapsed DSM to l blue marks since there are no benefits from overlapping between downstream and upstream tasks with non binding dependencies The pink marks at 3 1 4 3 7 4 8 7 9 8 and 10 9 represent critical binding dependencies The brown marks at 5 2 6 4 9 5 and 9 6 represent non critical binding dependencies The network diagram in Figure 8 shows equivalent project structure of the AEAP collapsed DSM except that non binding dependencies are omitted It is noticed that block tasks B1 and B as well as singe task m can be placed in different levels since they have nonzero level slack The path along the tasks a b e i n o is a critical dependency path of the project linked by critical 11 Soo Haeng Cho soohaeng_cho y
14. mT aog z o o d lt Note gt 1 The optimistic estimate must be smaller than or equal to the most like ly estimate which must be smaller than or equal to the pessimistic estimate 2 The percentage of completion must be between 0 and 100 but does not need to be accurate 3 A user can also enter duration estimates for coupled blocks directly to Project worksheet without computing them in the modeling module This is further explained in the Options 4 There is no need of entering duration estimates in the rows such as Block1 and in Block2 Table 3 Duration Inputs from Project worksheet 3 4 2 Resources The tool assumes that there exists a fixed resource pool throughout the entire project duration Resource worksheet is given for receiving data for this resource pool Table 4 shows an example where Jo Max Units represents the total amount of each resource available for the project Max Units 100 200 Po BB 200 Oo o oB a B 10096 Table 4 A Resource Pool from Resource worksheet The tool receives a resource requirement of each task which is assumed to be constant over the entire period the task is processed It also takes the resource requirements for block tasks assuming that those resources are required over the entire period that all constituent tasks are processed When two or more tasks are competing for limited resources in a certain period of time the t
15. repetition until it reaches Max Learning which is the minimum percentage of original duration when the task does the same work repeatedly Thus rework amount is calculated as the original duration multiplied by rework impact and learning curve Figure 17 shows an example of rework impact and learning curve inputs The tool takes an approach similar to Browning and Eppinger Modeling the Impact of Process Architecture on Cost and Schedule Risk in Product Development Working paper No 4050 Sloan School of Management MIT 2000 21 Soo Haeng Cho soohaeng_cho yahoo com steaming Curve gt lt Rework Impact gt Task Name Learning Max Learning of 10 11 a ae a 05 0 so olos E o e lt Note gt Learning must be higher than or equal to Max Learning FIGURE 17 REWORK IMPACT AND THE LEARNING CURVE FROM BLOCK WORKSHEET 3 4 3 2 Overlapping The tool receives inputs for overlapping iteration between the tasks of which information flow is marked as 2 in the DSM These inputs need to be provided from Block worksheet for tasks within a coupled block and Analysis worksheet for other tasks The overlap amount i j represents the planned overlap amount between tasks i and j and it is a fraction of the expected duration of task i This carries the assumption that the downstream task cannot be completed before the upstream task finishes The overlap impact i j repre
16. A EIS i za Steck 2 a ooh ew Souro Meio SWT shit sip Ny vin vi ms gt m s mo p3 1 ay Wa o gt wy A is ier Mas me Mei at wif a iba t HA a V a gg hai ed emmah ee mpat x tisessa Na UA ett gio D Fe Bad Pad aD a ox oe s ai ua n on oo 64 oR Lead Tame 3s u vex 1 ti o 5 b c 2 E 5 D cumulative probability FIGURE 19 ANALYSES RESULT FOR COUPLED BLOCKS FROM THE MODELING MODULE 23 Soo Haeng Cho soohaeng_cho yahoo com The computed durations of a coupled block are also transferred to Project worksheet as shown in Table 6 By default 10th 50th and 90th percentiles of simulated durations are assigned to optimistic most likely and pessimistic durations of a dummy task respectively Task D TaskName Opt Duration Likel Duration Pess Duration Complete Act Duration 11 Bock Goo d La dn o d g d a d e d d Ee a Ee A a S a E E E ae a ee ee ee ee ee eee 15 D Block2 Duration 84 ns 1583 Table 6 Simulated Durations of a Block Task in Project Worksheet The tool also draws three simulated Gantt charts out of many scenarios for ith block in Gantt Chart Bi worksheet Figure 20 shows an example of a simulated Gantt chart Sampled Duration represents the duration of one time execution sampled from the triangular distribution based on the estimated durations and percentiles Si
17. ATION MARK 3 4 Inputs for the Modeling Module Using the analyses results from the structuring module the modeling module performs analyses with additional inputs such as durations resources overlapping and sequential iterations A table for the inputs to the modeling module is created in Project and Resource worksheets after the analyses in the structuring module The tool assumes that sequential iteration takes place only among tasks within coupled blocks 17 Soo Haeng Cho soohaeng_cho yahoo com 3 4 1 Durations Table 2 shows the part of the table created in Project worksheet after analyzing the sample in the previous section A coupled block has its constituent tasks and a dummy task with indentation For instance Block 1 in Table 2 is composed of single tasks c d g and f and a dummy task D Block1 Duration A dummy task is used to represent the duration of a coupled block For each task a user is asked to enter three estimated durations optimistic most likely and pessimistic for its expected duration of one time execution The expected duration is the duration between the start and end of its continuous work even though the task may iterate more than once afterwards If a task is in progress a user is also asked to enter actual duration and percentage of completion In this case the estimated durations must be for a remaining duration of a task If a task within a co
18. H I J K LIM NIO PIA RIS TYVI 3 Sample pj A E OF PIGS rer ot stenenirin tes 9 cnote gt index based on AEAP colspsed FIGURE 13 MOVE TASKS WITH NONZERO LEVEL SLACK 15 Soo Haeng Cho soohaeng_cho yahoo com lt Note gt 1 The tool assumes that the sequence of coupled blocks remains the same after moving tasks with nonzero level slack For instance the index of block 1 should always be smaller than that of block 2 2 If a successor of a selected task also has nonzero level slack and its level is not higher than the target level of the selected task the successor is also moved to the level which 8 higher by one than the target level 3 3 7 Enter block names Without entering the names of coupled blocks manually to all different worksheets a usercan enter those names once to the form provided when executing Enter Block Names as shown in Figure 14 Mock Name Humber of Bineks in the Project 2 Subirame Dosgn avd FEA Kucha Design and FEA FIGURE 14 ENTER THE NAMES OF COUPLED BLOCKS 3 3 8 Insert the unplanned iteration mark The tool assumes that planned iteration only occurs between the overlapped tasks or tasks within coupled blocks Unplanned iteration represents a possible iteration in a system level or between major development phases usually a long feedback loop in the DSM which is not accounted in a project plan The loop usually feeds back from the task such as major
19. The first type represents the case that a downstream task requires final output information from an upstream task to begin its work The second type represents the case that a downstream task uses final output information in the middle of its process and or begins with preliminary information but also receives a final update from an upstream task 7 Soo Haeng Cho soohaeng_cho yahoo com a First Type b Second Type FIGURE 3 TYPES OF INFORMATION FLOW BETWEEN TWO TASKS When there is the first type of information flow from task a to task b enter 1 into i j of the square matrix where i and j are the unique indices representing tasks b and a respectively For the second type of information flow enter 2 Reading across a row reveals the tasks where the inputs of the task corresponding to the row come from Reading down a specific column reveals the tasks receiving outputs from the task corresponding to the column If the sequence of tasks in the matrix is the sequence of execution a nonzero element in the upper diagonal represents a feedback Figure 4 shows a sample of the inputs for the structuring module ey Sample for User s Guide A IBICDIE FIGH I J KI LIM NDONDI GIRIS Ti Ul Vv iw X Project Name lt NOTE gt 1 Enter the name of a project in cell 2A Sample 2 Enter all task names in column A Task Name 2 3 4 5 6 7 8 9 10 1112 13 14 15516 47 18 19 20 21 22 2312 mi BS Eia D
20. USER S GUIDE for DSM MIT TABLE OF CONTENTS 1 OVERVIEW asses al ca Ua 3 Zee INSTALLATION z aea i ki aaa 5 Des MUING TIONS ai eeo ie ai E SEET E R E TEE Ia kaisha 7 3 1 Inputs for the Structuring Module isss essnsies teese snie tesiaren reniri renia eh egosi J 3 2 Analyses in the Structuring Module cccscccccceseesessenececececessesencaeceeeeeseesenneaeeeeeeeeees 8 3 3 Editing the DSM sis sites ee ia aa von suite etiee cued E a ite tebteotecdeesnueeheee 13 3 3 1 Insert a new task in the current TOW cece cececececececececececececececececeeeeeeeeeeeeeeeeeeeeeeeeess 13 3 3 2 Delete a task in the current TOW cecececececececececececececececeeeceeeeeeeeeeeeeeeeeeeeeeeeeeeeeeess 13 3 3 3 Delete a dependency mark from the DSM wwwemenaenanananmannananamanw mwanamama 14 3 3 4 Undelete the mark in the DSM osre rde siene e aKa ERa EEEE aa tat 14 3 3 5 Swap the sequence of two tasks in the same level lt wwwwe 14 3 3 6 Move tasks with nonzero level slack ccecececececececececececececeeececececeeeeeeeeeeeeeeeeeeees 15 3 3 1 Enter block NAMES 27 5 acct vs cavevnsetace chino evesdonabUiaesececedonsgane shed cogs iae akiba 16 3 3 8 Insert the unplanned iteration mark wwmewnenamnnwnwnnananananwnmwananamwa 16 3 4 Inputs for the Modeling Module 2000 00 eee ee eesenccecececeesensnnececececeseessanaaeeeeeeseessesenneaeeees 17 DAL Da ya Ta La 3
21. ahoo com binding dependencies The dotted arrows indicate non critical binding dependencies between tasks The dependency between tasks b and f in the collapsed DSM for instance is omitted since it is non binding Note that the coupled blocks collapsed in these analyses are expanded in the modeling and scheduling analyses 9 0 No oe ee LO LQ LG LA LO LO LO FIGURE 8 NETWORK DIAGRAM REPRESENTING AEAP COLLAPSED DSM lt Note gt The tool uses four different indexing systems for tasks as follows 1 The first one is the one used in DSM Input worksheet when receiving inputs from a user 2 The second one is used in AEAP worksheet which shows new indices after partitioning Indices in ALAP worksheet are based on this indexing system 3 The third one is used in AEAP collapsed worksheet after collapsing coupled blocks Indices in ALAP collapsed worksheet are based on this indexing system 4 The fourth one is used in Project worksheet which shows identical indices with IDs in Microsoft Project 12 Soo Haeng Cho soohaeng_cho yahoo com 3 3 Editing the DSM 3 3 1 Insert a new task in the current row A user can insert a new task from DSM Input or AEAP worksheet After adding a task name and its input output information flows a user must execute Analyze DSM in the worksheet where he has inserted the new task Figure 9 illustrates this procedure
22. does not make these automatic conversions 29 Soo Haeng Cho soohaeng_cho yahoo com 4 2 Resource Option FIGURE 28 RESOURCE OPTION 4 2 1 Within Blocks By default the tool resolves resource conflicts among tasks within coupled blocks if resource inputs are provided If No is selected in Resolving Resource Confliction analyses are performed without resolving resource constraints among tasks within coupled blocks With the checkbox unselected in the Priority Option the tool allows splitting of tasks within coupled blocks in process due to resource constraints and determines resource priorities without giving first priority to the tasks in process 4 2 2 Outside Blocks With the first checkbox unselected the tool allows splitting of tasks not belonging to a coupled block in process due to resource constraints and determines resource priorities without giving first priority to the tasks in process With this option the tool does not compute slack and criticalities of tasks and major critical sequences With the second checkbox unselected the tool considers slack per successor before CUMRED when determining resource priorities 30 Soo Haeng Cho soohaeng_cho yahoo com 4 3 Simulation Option FIGURE 29 SIMULATION OPTION 4 3 1 Latin Hypercube Sampling The tool uses the Latin Hypercube Sampling LHS method and assumes the triangular probability distributio
23. e inputs for iterations within coupled blocks from Block worksheet a user needs to execute Display input layout for coupled blocks under Analyze Block in the menu In order to receive inputs for overlapping iterations between block and or single tasks not belonging to a coupled block from Analysis worksheet he also needs to execute Display overlapping input layout for the project under Analyze Project in the menu The tool assumes that planned rework of a task is generated due to the following causes e receiving new information from overlapped tasks after starting to work with preliminary inputs e probabilistic change of inputs when other tasks are reworked e probabilistic failure to meet the established criteria The first cause gives rise to overlapping iteration and the second and the third causes give rise to sequential iteration Parallel iteration of a limited number of tasks is simulated by combining overlapping and sequential iteration 20 Soo Haeng Cho soohaeng_cho yahoo com 3 4 3 1 Sequential Iteration This type of iteration is modeled using three parameters rework probability rework impact and 2 learning curve Rework probability i j r represents the probability that task i does rework affected by task j in rth iteration In the case of i lt j it represents the feedback rework caused by the change of information from downstream task j or by the failure of downstream
24. ect From Worksheet Field Name Resource Name Group Group Max Units Max Units 7 Click O K and Open 6 Soo Haeng Cho soohaeng_cho yahoo com 3 FUNCTIONS When Microsoft Excel starts the add in also starts automatically after proper set up The menu DSM is located among standard menus in Excel as shown in Figure 2 A user can perform analyses for different projects using different workbooks at the same time He can also save a workbook and perform further analyses based on the data and analyses results stored earlier The tool receives inputs from a user through formatted worksheets The inputs and analyses results of each of three modules are explained in the following sections ERJ Microsoft Excel FIGURE 2 ADD IN MENU IN EXCEL 3 1 Inputs for the Structuring Module By executing New DSM in the menu a user is given the DSM Input worksheet which is formatted for receiving inputs of maximum 240 tasks for the structuring module A project name in cell 2A must be unique to analyze different projects in other workbooks at the same time For analyses in the structuring module a list of tasks and information flows among tasks are required First list the tasks down in the first column in their rough sequence of execution with unique indices The same indices are listed in the row above the square matrix The tool receives two types of information flows between tasks as shown in Figure 3
25. eng_cho yahoo com 3 3 3 Delete a dependency mark from the DSM A user can delete a dependency mark from the DSM Input or AEAP worksheet After deleting the selected mark the tool automatically performs Analyze DSM Figure 11 illustrates this procedure in the AEAP worksheet Tost Here Level IAMINI Taek Harve Levi Beets izi i EIA l HE jep pja gadam Jogdg gt gt 11 a 12 i 2 n 6 u AE ae SE ES TRE Fl i F Lal at a at af af af ananas i 2 Ji 4 s ol 7 ol alio TERE 1 Select the dependency mark ss 2 Execute Delete a dependency gt FIGURE 11 DELETING A DEPENDENCY MARK FROM AEAP WORKSHEET 3 3 4 Undelete the mark in the DSM A user can undelete the marks which have been deleted up to ten before executing another command This function is useful when a user wants to compare different partitioning results with a certain dependency being assumed to be minimal and removed 3 3 5 Swap the sequence of two tasks in the same level From the AEAP or ALAP worksheet a user can swap the sequence of the two tasks both of which are located in the same level but do not belong to a coupled block or both of which are within the same coupled block This command does not affect the partitioning results except for
26. itional weights by default With this option selected it also computes them when analyses are performed without resource constraints 4 3 5 Rework Risk Tolerance When tasks iterate sequentially a choice of rework policy may allow rework concurrency to shorten the lead time For illustration consider the simple example given in Figure 30 When both tasks a and b require new information from task c the work policy of performing the rework of both tasks a and b concurrently may be preferred in order to shorten the lead time In this example he tool assumes that they are reworked concurrently if the total rework probability between the two tasks is less than a rework risk tolerance If this is zero default a downstream task always waits until the rework of anupstream dependent task is completed For more detail refer to the thesis written by Soo Haeng Cho Taname oy a 3 aE a O a Information Flow Diagram b DSM FIGURE 30 AN EXAMPLE FOR TASK CONCURRENCY 32 Soo Haeng Cho soohaeng_cho yahoo com 4 4 Gantt Chart amp Table Option FIGURE 31 GANTT CHART amp TABLE OPTION 4 4 1 Simulated Gantt Chart If NO is selected the tool does not draw simulated Gantt charts for coupled blocks and or a project This option saves computation time significantly 4 4 2 MS Project Table By default with the first option selected the tool creates MS Project worksheet when executing
27. ject Modeling n ww Structuring FIGURE 1 AN INTEGRATE PROJECT MANAGEMENT FRAMEWORK 3 Soo Haeng Cho soohaeng cho yahoo com The primary goal of this work is to develop an integrated method that guides project management efforts by improving the effectiveness and predictability of complex processes The method can also be used for identifying leverage points for process improvements and evaluating alternative planning and execution strategies Better project management will ultimately result in a better quality product with timely delivery to customers The structuring and modeling modules are implemented as a Microsoft Excel Add In The analyses results from these modules can be transferred to the structuring module in Microsoft Project Table 1 summarizes the inputs and analyses results of each module Design Structure Matrix Identification of loops and process hierarchies among tasks Structuring Modeling Scheduling e A list of tasks Information flows among tasks e Information flow patterns Duration estimates of tasks Resource requirements of tasks and capacity of the project Dynamic characteristics of overlapping and sequential iterations including the learning curve Rework risk tolerance Scheduled durations of tasks and coupled blocks chosen from the probability distribution Due date and or project buffer size Differentiation of planned amp unplanned iteratio
28. ling ccc cceecesseceseeeneneceeeeeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeas 31 4 3 2 Task Duration Estimate Percentiles c c csssscesececseeeneeeeeeeeeaeaeaeaeaeaeaeaeaeaeaeaeaeas 31 4 3 3 Direct Duration Inputs for Coupled BlocksS wwwnenananwnmnananannzmmmmwanana 32 4 3 4 Rework Adjusted Durations and RPWS sccccccececeesensnneceeeeeceesenenneaeeeeeeeeees 32 4 3 5 Rework Risk Tolerance oi sec vesase wakadai maka kana Ba kaamka makes saa 32 4 4 Gantt Chart amp Table Option cccccccccccccccccecececeeeeeeeeeee cece eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeees 33 4 4 1 Simulated Gantt Chart 0 ccc cccccecssssseecenecceaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaeaes 33 AAD MS Project Pal eked seek E ee EERE E E EE E a doused 33 APPENDIX DEFINITIONS OF HEURISTIC PRIORITY MEASURES w wwwmememowa 34 Al Rank Positional Weight RPW eesse eeeeeeeeeeeeeeeeeeeeeeeeeeeees 34 A2 Cumulative Resource Equivalent Duration CUMRED n 34 1 Soo Haeng Cho soohaeng_cho yahoo com Soo Haeng Cho soohaeng_cho yahoo com 1 OVERVIEW This computer program is an application tool of the integrated project management framework proposed in the M S thesis An Integrated Method for Managing Complex Engineering Projects Using the Design Structure Matrix and Advanced Simulation by Soo Haeng Cho at the Massachusetts Institute of Technology in June 2001
29. mulated Duration represents a total duration including rework The number inside a bar indicates the overlap amount of a task executed in prior state s lt Scenario 2 of Block 2 gt Task 1D Task Name Sampled Duration Simulated Duration JJ as 2 6 13 dk a ee CS a Wa a Po State Duration 26 6 0f sshd 48 Po tea time fief A o lt Note gt Be careful not to print all Block worksheet since there exist data not seen in the screen FIGURE 20 A SIMULATED GANTT CHART OF A COUPLED BLOCK 3 5 2 Analyses Results for the Project Figure 21 22 and Table 7 show an example of analyses results in the Project worksheet obtained by executing Compute lead time with resource constraints under Analyze Project Figure 21 shows the summary of lead time computations and major critical paths and sequences resource constrained critical paths with percentages that they are critical out of many simulation runs Note that lead time as well as critical paths are different when the tool accounts for resource constraints from those without resource constraints The tool also computes probability distributions of the lead time with and without resource constraints 24 Soo Haeng Cho soohaeng_cho yahoo com Lead Time with Leveled Resources Lead Time without Resource Constraints Avg 51 3 Std 8 3 Avg 44 8 Std 9 6 10 30 50 70 90 41 8 46 5 50 3 54 2 61 9 10 30 50 70 90 35 2
30. n for the expected duration d a task The LHS method divides the input distribution into N Number of strata in probability distribution strata of equal marginal probability and randomly samples once from each stratum Then it sequences the sampled values randomly The tool repeats this procedure N Number of Latin Hypercube Sampling times Thus it generates total N N X N random numbers for each task N is the total number of simulation runs 4 3 2 Task Duration Estimate Percentiles Default of optimistic value is a default percentile of optimistic duration estimates of tasks Default of pessimistic value is that of pessimistic duration estimates 3 McKay M Beckman R and Canover W A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code Technometrics Vol 21 No 2 pp 239 245 1979 31 Soo Haeng Cho soohaeng cho yahoo com 4 3 3 Direct Duration Inputs for Coupled Blocks If YES is selected the tool uses user specified duration estimates for coupled blocks instead of computed durations from the modeling module This option can be effectively used when a user can estimate expected durations of coupled blocks more accurately than using the modeling technique 4 3 4_Rework Adjusted Durations and RPWs When analyses are performed with resource constraints the tool computes rework adjusted durations and rank pos
31. ns Identification of non binding dependencies A critical dependency sequence and level slack of tasks Probability distributions of the durations of the project and coupled blocks Resource constrained or conventional slack and criticalities of tasks Critical sequences or paths with percentages of criticalities Simulated Gantt charts PERT or Gantt chart Schedule risk that the project fails to meet the due date or target Table 1 Summary of Inputs amp Analyses Results of Modules Soo Haeng Cho soohaeng_cho yahoo com 2 INSTALLATION For proper installation a user must have two program files one jpg file and one excel file having a sample application The tool consists of three parts a DSM MIT xla file an add in to Microsoft Excel b DSM MIT bas file a module in Microsoft Project c DSM to Gantt Chart a map for a file transfer from Microsoft Excel to Microsoft Project Installation will be completed following the below four steps STEP1 STEP2 1 2 3 STEP3 1 2 3 4 5 STEP4 1 2 3 4 Save warning sign jpg file under C DSM Install DSM MIT xla file Start Microsoft Excel Go to Tools Add ins Browse Select DSM MIT xla from the folder in which you saved this file Install DSM MIT bas file Start Microsoft Project Go to Tools Macro Visual Basic Editor Go to View Project Explorer in the Visual
32. ohaeng_cho yahoo com Analysis worksheet shows the results of structural analyses based on the AEAP collapsed DSM Level slack of a task is defined as the difference between its level in the ALAP DSM and that in the AEAP DSM Dotted boxes are wrapped around the columns of those tasks between the levels they can be located in For instance Block1 has level slack of two and it can be located between levels two and four The task m and Block have nonzero level slack as well Various types of information dependencies are classified using the results of hierarchical decompositions A binding dependency represents the dependency which is regarded as a constraint between two dependent tasks The delay of information transfer directly causes the downstream task to slip This information dependency is translated to a precedence task relationship of finish to start in the modeling and scheduling modules All dependencies within coupled blocks are regarded as binding A non binding dependency represents the dependency which not regarded as a constraint between two dependent tasks The delay of information transfer does not directly impact the schedule even though there is information flow between two tasks Thus it is regarded as a redundant constraint and omitted in the modeling and scheduling modules A critical binding dependency is a binding dependency between the tasks with zero level slack A non critical binding dependency
33. oject 26 Soo Haeng Cho soohaeng_cho yahoo com 4 Go to DSM in the menu of Microsoft Project and execute Adjust Tasks in Coupled Blocks Figure 23 shows an example of the Gantt chart created in Microsoft Project A coupled block is represented as a rolled up task within which its constituent tasks are arranged without feedback information flows A dummy task is added at the end representing the duration of a coupled block that is computed in the modeling module Po ume ty gst OO re ee Predecessors Eak 5 days 27 1 days 1 4 days 10 days 3 4 days 3 5 days 4 5 27 1 days 6 days 1 9 days 8 7 days 2 18 5 days 9 4 days 5 days 12 5 days 13 18 5 days 5 days 9 3 days 16 8 days 10 11 17 5 days 18 FIGURE 23 A GANTT CHART IN MICROSOFT PROJECT lt Note gt 1 Make sure that Calculation is set Automatic as default in Calculation tab of Options under Tools in Microsoft Project 2 If a project has a small number of tasks as shown below a user needs to delete additional rows other than the rows with task names before executing Adjust Tasks in Coupled Blocks Needs to be xo deleted FIGURE 24 ADJUSTMENT IN MICROSOFT PROJECT 3 When a project is in progress actual durations of tasks are not transferred to MS Project For completed tasks a user needs to change durations from 1 to 0 in MS Project 27 Soo Haeng Cho soohaeng_cho yahoo
34. ool determines priorities by the heuristic rules Among tasks within a coupled block a task has a higher priority if 1 it has been in process 2 it has a higher user specified priority See the Appendix for the definitions of the measures used in the heuristic rules These measures are computed when analyzing coupled blocks or a project with resource constraints 19 Soo Haeng Cho soohaeng_cho yahoo com 3 it has a higher rework adjusted rank positional weight and 4 it is sequenced more upstream from 1 to 4 in order of significance which can also be changed in the Options Among block and or single tasks not belonging to a coupled block rule 3 is replaced by higher cumulative RED and smaller slack per successor These measures also imply relative importance of tasks from a schedule perspective A user specified priority of a task can be entered in Priority column of the table out of Highest High Medium default and Low Table 5 shows that task c requires 100 default units of resource A2 and task d requires 100 of resource Al and 150 of A with a Highest user specified priority Task ID Task Name Resource Names Priority Eo eee E AAA AA A ai aed ee pa Ja gt anaso Highest Table 5 Resource Reguirements and User Specified Priorities from Project worksheet 3 4 3 Iteration In order to receiv
35. sents the expected overlap impact when task 1 is overlapped with task j by the given maximum overlap amount It is a fraction of that amount If it is equal to 1 it implies no benefit from overlapping To implement overlapping strategy it should be reasonably less than 1 considering additional risk due to the evolution of volatile preliminary information Figure 18 shows an example of overlapping inputs between tasks within a coupled block lt Overlap Amount gt lt Overlap Impact gt Task Name e ot 101 11 g to 141 FIGURE 18 INPUTS FOR OVERLAPPING FROM BLOCK WORKSHEET 22 Soo Haeng Cho soohaeng_cho yahoo com 3 5 Analyses in the Modeling Module Analyses are performed in two steps First it computes probability distributions for expected durations of coupled blocks Then it analyzes an entire project with coupled blocks collapsed which have computed distributions and separate resource requirements 3 5 1 Analyses Results for Coupled Blocks Figure 19 shows an example of analyses results in Block worksheet obtained by executing Compute durations of coupled blocks under Analyze Block A l Cc BEF SW ise MA OF Ge S T U Y V RJY Z ASABACADAEAP AG AH AJ AL ARAL AX AN ADAP AG Manta of Compe Dech t 2 ee SE erpies Ang 215 ted 47 Se LLA Da ee ed kuwa aburi thpt rorem Weckt i T Totis Oura Nra tes waas rus ja Barsten Ang TET IAT t n 1 see eae Wee hs W
36. task j to meet the established criteria In the case of i gt j it represents the feedforward rework that downstream task i needs to do since upstream task j has generated new information after it has done its own rework Figure 16 shows an example of rework probability inputs A user can specify rework probabilities up to Sth Iteration directly If rework probabilities are provided only in 1st Iteration the tool uses the same probabilities in each iteration Alternatively a user can specify a percentage decrease of probabilities in each iteration by entering a number between 0 and 100 in Decrease optional For instance if task h has 60 in Decrease without inputs in the 2nd iteration a rework probability from task I to task h in the second iteration is computed as 0 8 60 0 48 Max Iteration limits the number of iterations of tasks optional lt Rework Probability gt METER ry 0 3 2nd Iteration h ee as Fe es ie nee KI TAA KAA S da odl FIGURE 16 REWORK PROBABILITY FROM BLOCK WORKSHEET Rework impact i j represents the percentage of task i to be reworked when rework is caused by task j Rework impact is assumed to be constant in each iteration The learning curve in Learning represents the percentage of original duration when a task does the same work for a second time The tool assumes that the learning curve decreases by Learning in each
37. testing or reviewing at the end of each phase This can also be caused by unexpected market changes technology innovation etc in the middle of processes The unplanned iteration is often regarded as a failure mode and requires managerial decision about whether to continue or abandon the project When it is 16 Soo Haeng Cho soohaeng_cho yahoo com decided to pursue another iteration it is common to re plan the project The tool represents this type of iteration flow with warning sign in the DSM after partitioning If this loop is initially marked in the DSM and determined as unplanned after partitioning smaller coupled blocks can be obtained by replacing the dependency mark with the warning sign Even though unplanned iteration is not accounted in the schedule it is very important to be aware of the existence of those iterations so that appropriate contingency plans can be made as part of project planning Figure 15 illustrates this function Tart Kara Laval HEI J ej 7 el af sop as sais vay Fecha a7 Ya al a a al a aaa a ig FTSTSlO e Pr ETOM ER i 1 Ji 2 a gi ol 7 o al iol yl iz 13 14 1 Select the cell i j in the upper diagonal 2 Execute Insert the unplanned where unplanned iteration may occur to task i when task j is finished FIGURE 15 INSERT THE UNPLANNED ITER
38. upled block has been completed before and iterates the actual duration of first execution needs to be provided in Act Duration and 100 in Complete The actual duration of a coupled block including all constituent tasks can also be trackedin Act Duration of a dummy task The percentiles of optimistic and pessimistic estimates can be set from the Options default values are 10 and 90 respectively The tool also allows a user to specify different percentiles for duration estimates of each task Ot task Name _ oe Duration ey Yo aa ae Duration ae es Ja a 6 Salt SS Sie Ht Sr 25 ou NS ST T S T S o s E Lethi T T T T T To o as preot eaa e ft AA Ooo T T T S WA zi a WA KAA SS SS T SS Tonn Pig fo ig Table 2 Project Table from Project worksheet Table 3 shows an example of duration inputs It shows that tasks a b and c are completed and tasks d and e are currently worked on Task d is expected to be completed in 5 days most likely from now Task m is expected to be completed in 4 days with 20 confidence Note that without specified percentiles the default percentiles are used for optimistic and pessimistic estimates of each task 18 Soo Haeng Cho soohaeng_cho yahoo com Task ID Opt Duration Likely Duration Yo Complete Act Duration 100 7 D Blockt Duration AAA AAA Pes fo t00 fT ede t i t e e e io
Download Pdf Manuals
Related Search