Scheduling and Cntrls f Prject Manufacturing Mhamed El-Mehalawi, Ph.D., PMP Prject Management Cnsultant Abstract Prductin cntrls are usually based n either mass prductin r jb-shp manufacturing. These techniques are tailred t manage prductin f quantities f the same prduct. Outside f that spectrum is the manufacturing f sets f unique prducts. Every prduct is the ultimate result f a prject. Althugh these prducts are created in manufacturing envirnments similar t the jb-shp cnditins, they still need different management and cntrls techniques. In prject management, prduct manufacturing is usually a phase in a larger prject such as EPC (engineering, prcurement, and cnstructin.) Mst EPC prjects invlve manufacturing that is either internal t the perfrming rganizatin, r utsurced. Usually the schedule f manufacturing is lumped with the prcurement phase as ne lng task. This shws a discnnectin in the prject schedule. The prject manager has n cntrl ver that task because manufacturing by itself is managed by jb-shp r make-t-rder prductin management techniques. Prjects require a unique prduct t be manufactured. Prject planning and scheduling fr unique prducts is nt cmmn in mst manufacturing plants. This paper presents and redefines the cncept f prject manufacturing and based n a unique and temprary prduct. The paper examines the scheduling, sequencing, and resurce pling peratins required fr a plant dedicated t prject manufacturing. A methdlgy fr autmating the creatin f a schedule based n the critical path methd is utlined in the paper. An implementatin is discussed t prve the usability f the prpsed prcedure. The gal f this prcedure is twfld. The first is t integrate the manufacturing schedule with the rest f the prject schedule which gives the prject manager mre cntrl ver the whle prject including the manufacturing phase. The secnd is t intrduce a new methd fr managing and resurce lading in prductin plants dedicated t prject manufacturing. That enables the prductin cntrls persnnel t manage each prject individually and in the same time cmbine all prjects wrked cncurrently at the plant. Overview Prject manufacturing r engineer-t-rder (ETO) manufacturing is knwn and practiced in the industry but nt in a frmal way. Prject manufacturing is t prduce r assemble ne unit f each unique prduct. Althugh it is a manufacturing envirnment, it fllws the definitin f the prject f being temprary and unique. In mass prductin, there is a prductin r assembly line that prduces thusands f units frm a certain prduct. The methds used fr prductin planning, scheduling, and cntrls f mass prductin cannt be emplyed t prject manufacturing. Instead, a cmpany that wns r perates manufacturing plants dedicated t manufacturing r assembly f prjects use jb-shp techniques t schedule and cntrls their prductin. Nt nly that, they try t implement state f the art prductin prcess imprvements that are designed mainly fr prductin f high vlume such as mass prductin r jb-shp manufacturing. Blevins (1999) intrduced the tpic f prject manufacturing in a very nice and simple way. He stated that the prject manufacturing business has a set f islands and urged that they need t be integrated fr a better planning and cntrls. Fx et al (2009) intrduced a cmprehensive list f challenges and surces f cmplexity that face prject manufacturing. Interested readers are encuraged t review Fx s article. Sme f these challenges are: Giving an authrity t the custmer. That is typical in prject envirnments but nt very helpful in scheduling prject manufacturing. Change f pririties f individual custmers makes the prject manufacturing schedule t stp and resume several times during the lifecycle f the prject. High number f cmpnents that are needed fr the assembly f a single sub-prduct. Fr example a side wall f a biler may need mre than 100 tubes in different shapes and sizes. Cmpnents fr the same sub-assembly may have a high variatin in delivery times. 1
Abdelmaguid and Nassef (2010) stated that the jb shp scheduling prblem (JSP) is a traditinal decisin making prblem that is encuntered in lw vlume high variety manufacturing systems which are knwn as jb shps. Althugh jb-shp scheduling is dedicated fr lw vlume manufacturing, it is still a gd technique fr repetitive prducts but nt as gd fr a prject manufacturing. Every final prduct and all f its cmpnents are ttally different frm any ther prduct. Therefre, it is unwise t apply an apprach fr repetitive prducts int nnrepetitive parts. Anther simple reasn t examine prject manufacturing prblems is that manufacturing the prduct is ne part f the whle prject. Integrating prductin schedule with the rest f the prject schedule will be a nightmare if prductin is scheduled using jb-shp apprach. Carn and Fire (1995) urged t find an innvative apprach t integrate manufacturing and lgistics with prject management. Their apprach was suitable back then befre the ppularity f the enterprise prject management tls. On their study f cnstructin prject cmplexity, Bertelsenand Kskela (2005) urged that using tw different systems fr prject management and prject prductin adds t bth the cmplexity f the prject and the uncertainty f the bjectives. The fllwing table shws sme differences between the repetitive prductin and prject manufacturing. These differences are given t emphasize the need f a different methd fr scheduling and cntrlling prject manufacturing. Categry Repetitive Manufacturing ETO / Prject Manufacturing Prducts Makes standard prducts Prducts are unique Pricing Uses a price list Estimates, qutes, and Bidding Number f cmpnents per subassembly Lw High Inventry Based n part number N Inventry Engineering N r nly a few engineering changes A significant number f engineering changes Value Lw value Typically higher in value Prductin ruting Standard Custmized fr every prduct Lead time days r weeks mnths r years Shipping Ships frm finished gds Ships frm WIP(wrk-in-prgress) Prgress measures Measures cst variance frm the standard cst. Measures cst variance frm the riginal budget. This paper presents an apprach fr scheduling and cntrl f prject manufacturing using the critical path methd. The apprach prvides slutins t the tw requirements f the prject manufacturing business: i. Integrating the manufacturing schedule with the rest f the prject schedule. ii. Managing the verall lad f the manufacturing plant using enterprise prject cntrls systems. In fact, this apprach has been implemented in a few plants dedicated t prject manufacturing f heavy pwer generatin prducts. The apprach simplified the planning and cntrls that used t be perfrmed using jb-shp methdlgies. It prvided mre flexibility in stpping the prject and resuming it again. The prpsed system can be integrated with the manufacturing executin system f the plant. That will make it fully autmated which can prvide real time status f the prject. If there are nt enugh capabilities t integrate the system with manufacturing executin systems, it can be implemented by itself and the initial schedule and its prgress culd be updated manually. The rest f the paper is dedicated t the descriptin f the system and hw it is integrated with ther systems. The descriptin f hw t use the system and t get the benefit frm it is als utlined belw. 2
Creatin f a Manufacturing Prject Schedule Fr every prject r prduct a unique ruting sheet is always develped by the manufacturing engineering team. The ruting sheet cnsists f a set f ruting sheets fr shippable prducts r subassemblies f the final prduct. Every shippable prduct is cnsidered as a single nde in the wrk breakdwn structure (WBS) f the whle prject. Each ruting sheet at a WBS nde cnsists f a set f wrk-rders. A wrk rder describes the sequence f peratins that are needed t prduce a smaller subassembly. Prducts frm wrk rders may be assembled using a set f peratins described in anther wrk rder. A wrk rder may have ne r mre peratins. Typically each peratin in the ruting sheet cntains the fllwing: peratin duratin labr class number f needed man-hurs wrk center number f machine hurs material being prcessed number f material units In the critical path methd (CPM) wrld, each peratin can be cnsidered an activity. Therefre, the ruting sheet allws the creatin f the activity list cmplete with its duratin, labr resurces, machine resurces, and material resurces. If the ruting sheet is in electrnic frmat, which is mst prbably the case, it can be easily cnverted int a list f activities in the prject scheduling sftware. That activity list will be cmplete with all resurces laded. Mving activity data frm the ruting sheet int the scheduling system can be autmated and can be perfrmed in few mments even if the prject is very large. T have a cmplete schedule, ne imprtant feature is still missing. That is the activity sequence. Examining the wrkflw f prject manufacturing, it fund that fr every prject r prduct, there is a unique engineering bill f material (EBOM) nt a standard BOM. Knwing that each wrk-rder represents a cmpnent r part in the prduct, the EBOM specifies the parent assembly f each subassembly. That kind f parent-child relatinships, allws fr generating activity sequencing. Cnsider the fllwing example: Part C is a subassembly f part D. Part C is assembled by welding parts A and B. Part A is a 40 feet tube. It needs t 4 peratins; cleaning, milling, bending, and heat treatment. Part B is a flange and needs nly tw peratins; drilling and grinding. Then the sequence will be A1, A2, A3, A4, C1, C2, C3, D1. In parallel, there will be anther sequence B1, B2, C1, C2, C3, D1 as shwn in Figure 1. Fllwing that prcedure, activity sequencing can be easily generated and autmated. The WBS f the prject shuld be recgnized and fllwed during generating the activity list either manually r autmatically. Impsing activity sequence n tp f that will prduce a rtated shape f the EBOM. A1 A2 A3 A4 S C1 C2 C3 D1 B1 B2 Figure 1: Activity Sequencing Example These tw steps generate the initial resurce laded manufacturing schedule f the prject. It is generated using the CPM and can be easily integrated with the rest f the prject schedule. The benefit f the integratin is knwn t 3
mst peple in the field f prject scheduling. Any changes in the engineering schedule will affect the manufacturing schedule as sn as they are recrded. Althugh such a situatin is nt favrable in a manufacturing envirnment, it is the fact f life. Changes n engineering schedules and their reflectins n manufacturing schedules are much easier t handle using the integrated schedule than having tw different scheduling systems. Hwever, these changes cmplicate the resurce management f the manufacturing plant as shwn belw. EBOM Frm ERP System Ruting Frm ERP System Dcuments Cllected Frm Manufacturing, Accunting, and HR Part Child- Parent Relatinship Operatins and Their Prperties Build Glbal Infrmatin: Data Manipulatin Data Manipulatin Wrk Centers. Labr IDs. Material IDs. Templates. Activity Relatinships Activity List Resurce Lading Prject Cntrls Database Gantt Chart Resurce Lading Earned Value Measures Figure 2: Prcess Data Flw Figure 2 shws the flw f infrmatin t perfrm the prcedure mentined abve. It is assumed that bth the ruting sheets and the EBOM are generated and psted int the ERP system f the perfrming rganizatin. The tw blcks in different clr are assumed t be perfrmed nce in the lifetime f the system and then updated when changes happen. These are meant t build the glbal structures such rganizatin chart and resurce hierarchy f the manufacturing plant in the scheduling system. Updating the Manufacturing Schedule In mst manufacturing plants, there is a manufacturing executin system which cllects actual infrmatin frm the shp flr. The system cllects real-time infrmatin abut each peratin in the ruting sheet. It cllects the fllwing infrmatin: Operatin start time and date Time elapsed n an peratin 4
Labr hurs charged t an peratin Machine hurs charged t an peratin Material cnsumed s far Number f units cmpleted Operatin finish time A baseline based n the initial schedule shuld be generated fr future cmparisn with the prject prgress.on a daily, weekly, r biweekly basis, the prject schedule can be updated autmatically by extracting the actual start date, actual duratin, actual labr hurs, actual machine hurs, and actual material prcessed frm the manufacturing executin system. If the prcess is autmated, the update cycle might be daily. If it is perfrmed manually the update cycle might me weekly r biweekly. The number f unites cmpleted determines the physical r realistic percent cmplete f each activity. The earned value measures (EVM) can be emplyed with a great cnfidence since the percentage cmplete f the activity is based n a ttal bjective measure. Therefre, the system will be able t reprt the prject prgress and mre. Suppse that the baseline states that the activity duratin is 100 hurs and it requires 200 labr-hurs t bend 1000 tubes. Tw wrkers will be wrking n this activity fr 100 hurs. After 5 wrking shifts, 40 hurs f the duratin are gne, 80 labr hurs are charged, and 300 tubes are cmpleted. Using the EVM, this activity is ver budget and behind schedule. Hwever, there is anther cnclusin that can be captured. It was assumed that each tube will be bent using 0.2 labr hurs. The prgress shws that each tube cnsumed 0.27 f labr hurs. That cncludes that there is a prblem with the prcess efficiency. This is different frm accelerating a cnstructin r engineering activity t finish n time. The situatin here is abut the efficiency f the manufacturing prcess which shuld be crrected. The measures d nt nly give prgress f the activity but als they reprt prblems with the prductin prcess. Manufacturing management shuld pay attentin t these prblems. Resurce Lading fr Prject Manufacturing Plant One f the biggest advantages f the enterprise cntrl systems is the cmbinatin f resurce requirements fr different prjects. Cmbining the demand n resurces helps the manufacturing managers plan their resurces and priritize the wrklad. Figure 3 shws a requirement fr milling mechanist rle fr all prjects currently active in the enterprise database (that are the active prjects in the plant). This helps the manager f the milling grup prepare fr the peak resurce allcatin n milling mechanist by training ther wrkers. Thse with a secndary rle as milling mechanist will serve as primary wrker when they are needed. Anther planning scenari fr the manager f the milling department is when he is negtiating with the prject managers t mve sme f the activities back and frth in rder t get rid f the demand peak. Of curse the prject manager needs t cnsult with the manufacturing scheduler n the pssible effects mving an activity will cause n the verall schedule due t the dependency f ther activities. Is it pssible t find ne crew s heavily ver-allcated? Yes, it is. Prject managers usually knw their emplyees. They knw wh gets the jb right frm the first time, r can give the initial results fast, r wrks with minimal supervisin, r has many years f experience. During the initial planning f the prject, they ask t put that crew n their prject team. The result an ver-allcatin f a certain crew while under-allcatin f ther crews. It is the respnsibility f the functinal manager t reallcate the extra lad t ther crews wh can als perfrm the jb. The functinal manager can cach the less experienced resurce t achieve the jb efficiently. These resurces wrk fr the functinal manager wh is assumed t knw mst f the details f their wrk. By keeping a clse eye n the lad f each resurce in the grup, the resurce manager can avid the situatin f ver-allcatin frm happening. Changes n the start date f the prject r the start and finish f certain activities are impsed by the client r the engineering grup. These changes will frce the prject and its resurce lad t mve back and frth. That will make the requirements n resurces dynamic and uncertain as nted by Wullink et al (2004). Althugh it cmplicates the resurce management prcess, it reflects the real life. The prpsed system f managing resurces using the CPM may seem t versimplify the situatin but implementing it in a few plants prved mre efficient than cmputatinal systems. The CPM gives the true picture t the resurce managers and leaves the decisins t them. Cmputatinal methds make the decisins withut human invlvements. They are als data hungry. 5
Figure 3: Rle-Requirements Histgram Managing all resurces needed fr all active prjects in the manufacturing plant helps in identifying bttlenecks. Manufacturing prjects use high value machines like an verhead crane, a mbile 100 tn crane, a large sized milling machine, etc. If such a machine is verladed, it represents a bttleneck and everything else will be scheduled arund it. T upgrade frm ne unit f that equipment, anther unit shuld be added. Because adding ne unit is s expensive and requires a big investment, it is imprtant t make sure that bth the added unit and the existing units are almst fully utilized. In a prject manufacturing plant, mst resurces are always under laded with the exceptin f ne wrk center, which is always ver laded. A typical prject uses nly sme f the wrk centers. Hwever, ne wrk center is being used by all prjects that cme t the plant. This is the ver allcated machine. When a new prject is assigned t the plant, it is scheduled based n the availability f the ver allcated wrk center. The authr was helping the plant by implementing the cncepts and tls f enterprise prject manufacturing systems. The prject/prductin cntrls team was used t allcating nly human resurces, but was advised t allcate all resurces needed fr all activities. They did s fr a few mnths until they fund that the ver allcated wrk center was the bttleneck driving the schedule f everything else. Actually, this bttleneck causes mre prblems t the plant than the scheduling prblems d. In a fllw up visit t the plant, it was fund that they nly allcated and planned fr the bttleneck wrk center, while nt paying any attentin t the rest f the resurces. Althugh it sunds reasnable, it is nt recmmended t frget abut the wasted hurs f the under allcated resurces. T maximize the utilizatin f the under allcated resurces, the plant is advised t acquire a secnd unit f wrk center similar t the ne that is ver allcated. Hwever, with that investment, will the secnd machine be reasnably utilized? There is a need t ptimize between the tw factrs t make the decisin that is best fr the rganizatin. If there are many prjects assigned t the plant, in rder fr the tw units t be reasnably laded, then the plant wuld g with the ptin f acquiring anther unit. If this assumptin is nt true, they may study anther slutin t minimize the waste in the under allcated areas. The prpsed system prvides a methdlgy t answering all f these questins and t generating different what-if scenaris. It prvides a the big picture f the resurce lading fr the plant in a dynamic envirnment that can be apply the ptimizatin methds t level resurces r leave the task t the resurce managers as described abve. It als prvides a tl t respnd t custmers abut estimates and qutes fr new prjects. Knwing the current and 6
prjected resurce lad, plant management can determine when a new jb will fit in the plant schedule. Althugh the estimate is nt a clear answer as nted by Wullink et al (2004), every custmer requires a time and cst qutatin fr their prject. Cnclusins The technique described in this paper is the autmatic generatin f the initial schedule and the autmatic update and prgress recrding. T generate a detailed schedule fr a large prject, it takes at least tw weeks frm the scheduler t build it manually withut lading resurces. If resurces are t be laded int the schedule, it will cnsume anther tw weeks and will nt be accurate due t usual human errrs. The new system generates the initial schedule laded with resurces in less than five minutes and gets the prgress infrmatin in n mre than tw minutes per prject. Further mdificatin may enhance the system t reduce the prcessing time and allw a grup f prjects t run by batches. The prpsed system makes use f the enterprise prject scheduling systems t manage the resurce lading f the whle plant. The prpsed system facilitates integrating the manufacturing schedule with the rest f the prject schedule while enabling the resurce management in the manufacturing plant. The system was implemented in tw prject manufacturing plants and prved effective in spite f its simplicity. The system des nt use sphisticated cmputatinal methds but relies heavily n the use f CPM and the cmmn sense. The simplicity f the system makes the manufacturing managers cmfrtable in using it which adds t its success in the implementatins. References Abdelmaguid, T.F. &Nassef, A.O., A cnstructive heuristic fr the integrated scheduling f machines and multiplelad material handling equipment in jb shps, Internatinal Jurnal f Advanced Manufacturing Technlgy, Vl. 46, 2010, pp. 1239-1251 Bertelsen, S. and Kskela, L., Appraches t managing cmplexity in cnstructin prject prductin, The 1st Internatinal Cnference n Cmplexity and the Built Envirnment, 2005 Blevins, P., Prject-Oriented Manufacturing: Hw t Reslve the Critical Business Issues That Impact Organizatinal Cmpetitiveness, The Educatinal Sciety fr Resurce Management, Alexandria, VA, 1999, APICS Internatinal Cnference Prceedings. Carn, F. and Fire, A., Engineer t rder cmpanies: Hw t integrate manufacturing and innvative prcesses, Internatinal Jurnal f Prject Management, Vl. 13, N. 5, 1995, pp. 313-319 Fx S., Jkinen T., Lindfrs, N., and Yle n, J, Frmulatin f rbust strategies fr prject manufacturing business, Internatinal Jurnal f ManagingPrjects in Business, Vl. 2 N. 2, 2009, pp. 217-237 Wullink, G., Hans, E.W., and van Harten, A., Rbust Resurce Lading fr Engineer-T-Order manufacturing, Internal Reprt, Beta Research Schl fr Operatins Management and Lgistics, University f Twente Publicatins, 2004, http://purl.utwente.nl/publicatins/70237 7