روش زمانبندی پروژه برنامه ریزی تولید

We suggest a novel approach to modelling and solving production planning (PP) problems in make-to-order production environments. This approach unifies the capacity and the material flow oriented aspects of PP. Orders are modeled as projects that compete for limited resources. The projects consist of networks of variable-intensity activities which may require several resources at the same time. The goal is to generate production plans that satisfy all the temporal and resource constraints and minimize additional or external resource usage over the planning horizon. The model is applicable at different aggregation levels of PP, as we show with two real-life case studies. Although the capacity and material flow aspects of PP are coupled, a special solver developed for the project model efficiently solves problems of real-life sizes.

Pmductbn pknnhg (PP) matches Mure productlon load and capacttles by generating plans that determlne the flow of materlals and the use of resources over a gken plannlng horbon. Flndlng an approprlate match between load and capacltles Is a chlckewanbegg problem. On the one hand, the actual load - and the correspondlng flow of materlals - should M to the wallable resource wpacltles. On the other hand, the level of capacltles should be determlned so as to meet the demand over the. The problem Is usualy settled by lbtlng elther the load or the capacltles [I]. However, thls approach can hardly be taken nowiadm when the tradttlonal boundarles of companles are gettlng dlssoked. In the era of supply chaln management, vlrtual enterprlses and produdlon netwrks the tasks of PP may cross the organhatlonal boundarles of the fhs. Declslons on the use of resources should concern both Internal and externel capacltles; the Internal flow of materlals should be synchronhed wtth the Inmmlng and outgolng nDws [2]. All thls makes the PP problem extremely hard to solve. Conversely, the complex sltuatlons call for emclent, robust declslon support methods at each node of a produdlon network [2,3]. Hence, there Is a need ol Intultke and flexlble models end fast, rellable solutlon technlques that scakup WBII also to large problem Instances. An Important pndlcal requlrement for any new method Is that lt should be able to wrk by uslng data stored In exlstlng - smcalled legacy - produdlon Informatlon systems. There are long-standlng reclpes to handle complextty. A-bn removes detalls In the representatbn of products and orders, produdlon processes, resource capacltles, and the [I ,4]. Slmllar problems formulated wtth more detalls are llmlted by shorter plannhg horizon. The solutlon Is generated In a process where hlgher level solutlons provlde constrahts to lower level problems. Decomparfibn separates PP problems Into a resource and a load dented subproblem. Decompmltlon Is usually applled on several levels of aggregatlon. E.g., decomposed plannlng fundlons ire tradltlonally termed as master produdlon schedullng and materlal requlrements plannlng on the slde of the load, whereas roughcut capaclty plannlng and capaclty requlrements plannlng on the slde of resources [I]. Below we suggest a novel apprmch to modelllng and soMng long and medlum-term capactty and produdlon plannlng problems. Our method Is based on a generalhad verslon of the mzwrcecons?mhed pmm zcheduhg problem [5] and unmes the resource and the materlal flow orlented aspects of PP. By Introduclng ecthfikz, we handle together resource and temporal constralnts. Hence, plannlng can wrk wtthou uslng lead tlme estlmates that, In a turbulent envlronment, cannot represent IndMdual orders any more [6]. The projed- based approach caplures the strong goal-orlanted nature of memrder and engfnee-rder produdlon. The paper Is structured In the followlng way: In Sedlon 2 the projed model Is presented along wtth the maln prlnclples of the solutlon technlque. Then we descrlbe how to solve PP problems In two dlllerent produdlon envlronments by taklng the suggested project-based approach (Sedlons 3 and 4). Flndly, conclusions of the two case studles are drM In Sectlon 5.

In thls paper w hwe presented a new model for project schedullng and qplled tt to Integrated capadty and productlon plannlng problems. The method Is able to support plannlng even K no datalled Informatlon on productlon technology Is wallable. Cmersely, tt Is appllcable also for detalled plannlng when all product and productlon technology Informatlon are to be explottad. The model can handle deslgn, englneerlng, as well as tradttlonal produdlon processes. Hdng Introduced varlablelntenslty tasks and contlnuous resources, the model can be solved by cuslomhed mathematlcal programmlng methods very emclently. Hence. tt can be applled In a dynamlc setllng when re plannlng Is ldtlded by unexpected changes. The approach Is an attarnative to controCtheoretlc plannlng methods that, remarkably, Irtegrate the load and capaclty orlanted aspects of PP by worklng on contlnuous models [9,10]. The fast response the provldes opportunlty to analyze varlous scenarios and to drive the englne of a declslon support tool that can be used In an exploratory mode of operatlon. Our current goal Is to develop a complete workflow model [Ill whose key components are driven by the project-based planner.