Project management and scheduling under OFBiz

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionProje t management and s heduling under OFBizCh. DHIB1,2, O. Heintz1, A. Soukhal2, E. Néron21Néréide ompany3 bis, Les Isles37270 VERETZ2Laboratoire d'InformatiqueUniversité François Rabelais ToursPolyte h'Tours - Département Informatique64, Avenue Jean Portalis 37200 ToursNovember 8, 2012Apa heCon Europe 2012, Sinsheim - GermanyCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 1/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusion1 Context2 Introdu tion3 Appli ations4 Multi-Skill proje t s heduling models5 Implementation6 Con lusionCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 2/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionPlan1 Context2 Introdu tion3 Appli ations4 Multi-Skill proje t s heduling models5 Implementation6 Con lusionCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 3/ 30

Context ontextA Phd thesis entitled: Resolution methods for multi-skill proje ts heduling problems [2010-2013℄Collaboration between Néréide and S heduling team (OC team) of theComputer s ien e Laboratory of University of Tours (Polyte hni s hoolof Tours)

ContextInvolved partiesNéréide: An ERP integrator, espe ially Apa he Open for Businesssolutions.OC team: spe ialized in s heduling problems and de ision aid problems.It is atta hed to CNRS a (ERL CNRS 6305)Figure: Pra titioner ooperate with a ademi ian around OFBizaNational enter for s ienti� resear h(http://www. nrs.fr/)

ContextObje tiveDe�ne one or more proje t s heduling model, taking into a ountresour e availabilities and skillsPropose good s heduling methods for these modelsIntegrate these models as will as proposed solutions into OFBiz asadd-ons

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionResultsA ademi results3 national onferen es (ROADEF'2010, ROADEF'2011, ROADEF'2012)3 international onferen es (IESM'2011, OR'2011, PMS'2012)a s ienti� paper is going to be submitted in an international journalbefore the end of this yearCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 5/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionResultsPra ti al resultsIntegration of Drools into OFBiz through an add-onUser de�ned rules for tasks prioritiesTwo proje t s heduling models are studiedAdd-ons integrating the �rst model are developed and an be used fordemonstration on OFBizProje t generator for both models is implemented, to generate a proje tdemo data automati allyDeveloped models work with any time periodi ity (hourly, daily, et ,.)All developments are a essible with GPL li enseCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 5/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionIntrodu tionA proje t is a set of tasks that must be realized with respe t to some onstraints and at aim to optimize one or more obje tivesA task is hara terized in general by a workload ( expressed in man/dayor other), and an exe ution time (start date, ompletion date)Tasks are submitted to pre eden e relationships (a task Ti annot startbefore the end of task Tj)To realize a task, one or more resour es (human resour es, ma hines,et ) are neededCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 7/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionMore task hara teristi s ...A task an have a release date, a due date or a deadlineThe resour e requirement an be expressed in terms of skillsA ost an be asso iated to a task dire tly or dedu ed from its durationand/or assigned resour esA task an be allowed or not to be interrupted during its exe ution andresumed later, we alled it a preemptive taskIf a task has a due date then it has a tardiness whi h is equivalent to thedi�eren e between its a tual ompletion time and its due dateAnother notion alled lateness whi h is zero if the a tual ompletiontime is before or equal the due date, and the di�eren e between thesetwo dates else.Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 8/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionResour es main hara teristi sThree main type of resour es an be distinguished:1 Renewable resour es (ma hine available per day)2 Nonrenewable resour es (money)3 Doubly onstrained resour es (Proje t budget (limited per period andglobally))Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 9/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionResour es main hara teristi sThree main type of resour es an be distinguished:1 Renewable resour es (ma hine available per day)2 Nonrenewable resour es (money)3 Doubly onstrained resour es (Proje t budget (limited per period andglobally))In all ases availability and skills of resour es are always onsidered.Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 9/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionProje t s heduling obje tives1 Minimize the proje t ompletion time2 Minimize the ost of proje t3 Minimize the number of delayed tasks4 et .

Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 10/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionProje t s heduling obje tives1 Minimize the proje t ompletion time2 Minimize the ost of proje t3 Minimize the number of delayed tasks4 et .Two or more obje tives an be onsidered simultaneouslyCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 10/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionClassi al approa h1 Identify tasks2 Estimate task durations3 Establish relations between tasks4 Allo ate ne essary resour es to ea h task5 Make a planningCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 11/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionSome questions1 How to allo ate resour es if many possibilities exist ?2 How to verify manually the availability of ea h resour e ?3 How to be sure that the resour e assigned to a task is able to do it ?4 How to be sure that your s hedule is optimal or near to optimal withrespe t to the riteria to be optimized (Proje t ompletion date, proje t ost, et .)Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 12/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionProje t s heduling appli ationsThe proje t s heduling problems are useful for many industrial appli ations1 Constru tion proje ts2 Software development proje ts3 Many proje t s heduling models are a generalization of produ tions heduling (Job-shop s heduling)Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 14/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionExample of resour e onstrained s hedulingThe well known proje t s heduling model alled RCPSP for (Resour e onstrained proje t s heduling problem) is des ribed in the following exampleExampleCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 15/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionWhat hange when skills are onsideredTask has no dedi ated resour esMore than one assignment is possible for a task


Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionWhat hange when skills are onsideredTask has no dedi ated resour esMore than one assignment is possible for a task

⇒ Resour es assignment be omes non trivialCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 16/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionModel I : Preemption and syn hronizationExampleA0 A1 A2 A3 A4 A5[4, 0, 1]

[0, 3, 1] [2, 3, 0][0, 2, 0] Person analysis web DB UnvailabilityP1 1 1 1 [4, 5[P2 1 0 1 [5, 6[P3 1 1 0 -

b Skills syn hronizationb Preemption for ertain tasksb One person per skill duringtask exe ution

Time50 10P1P2P3 A1,Analyse

A1,BD A2 ,BD

A2 ,web A3,webA3,Ana A3 ,Ana A4,webCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 18/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionModel II : One skill per task, release and due dates,preemption only by unavailabilityExample


Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionResolution methods1 Exa t methodsBrute-for e sear hLinear programminget .2 Heuristi methodsglutton algorithmMeta-heuristi s...Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 20/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionPro ess1 Proje t sele tion : �rst estimation of time and ost an be done usings heduling algorithm2 If the proje t is ontra ted and has to be releaseda proje t team is sele tedtask requests per skill is estimated in hours, day, half a day or other timeunit measuretime slots of availability for the proje t is asso iated to ea h workerthe proje t manager or planning responsible person apply the s hedulingalgorithm3 if no feasible solution found, he an try with other parameters (givemore time, et ), or add more resour es availability4 if a solution is found, ea h sta� member has an a ess to his planningCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 22/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionHeuristi algorithmRequire: A: set of a tivities to be s heduledt ← 0while A 6= φ and t < horizon doES t : set of eligible tasks at t, sorted a ording to priority rule R1i = 0while i < |ES t | dofeasible ← TryS hedule(ES t (i), t)if feasible thenA ← A \ ES t(i)end ifi ← i + 1end whilet ← nextEvent()end whileCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 23/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionDemo data generatorbased on PSPLIB instan es(http://129.187.106.231/psplib/) on�gurable interfa e to manage the di�erent proje t parameters(number of skills, number of persons, et .)


Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionDemo data generatorbased on PSPLIB instan es(http://129.187.106.231/psplib/) on�gurable interfa e to manage the di�erent proje t parameters(number of skills, number of persons, et .)Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 24/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionQuality measure


Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionQuality measureCan we know whether the solution was good ?


Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionQuality measureCan we know whether the solution was good ?Answer ompare it, with the best known method if existsdeveloping good lower bounds (in ase of minimization) and then ompare with themCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 25/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionPro ess in ase of disruptionDisruption is unavoidable during proje t exe utionA disruption may be due to task request underestimation, new tasksthat was not onsider at the beginning, worker unexpe ted absen eHow to deal with itFix the s hedule ?Res hedule from s rat h the remaining a tivities ?Other approa h ?Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 26/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionOur approa hRes hedule the remaining tasks but ...Don't hange assignment for tasks in progressIn addition to the original optimized riteria, we minimize the maximumassignment hange with respe t to the a tual planning (so, we try tokeep assigned tasks to ea h person as mu h as possible identi al tothese in a tual planning )Ch. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 27/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionSomme on lusionsCon lusionTwo models of proje t s heduling onsidering skilled-resour es arestudiedResolution methods are proposed and implementedAdd-ons, integrating algorithmi solutions to OFBizCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 29/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionSomme on lusionsCon lusionTwo models of proje t s heduling onsidering skilled-resour es arestudiedResolution methods are proposed and implementedAdd-ons, integrating algorithmi solutions to OFBizOngoing worksIntegrating all methods into OFBiz as add-onsAdd a multi-skill proje t s heduling module in the ofbiz-extraCh. DHIB1,2, O. Heintz1 , A. Soukhal2 , E. Néron21Néréide ompany 3 bis, Les Isles 37270 VERETZ 2Laboratoire d'Informatique Université François Rabelais Tours Polyte h'Tours - Département Informatique 64, Avenue Jean Portalis 37200 ToursProje t management and s heduling under OFBiz 29/ 30

Context Introdu tion Appli ations Multi-Skill proje t s heduling models Implementation Con lusionThank you !Questions ?


Project management and scheduling under OFBiz

Documents

Transcript of Project management and scheduling under OFBiz