SKU duplication on a unidirectional picking line

Fivaz, Desima (2013-03)

Thesis (MComm)--Stellenbosch University, 2013.

Thesis

ENGLISH ABSTRACT: PEP is a devision of Pepkor Retail Limited and is the biggest single brand store network in Southern Africa and also owns and runs the largest clothing factory in Southern Africa. It was founded in 1965 and has since grown to more than 1 400 stores in 9 African countries (there is a PEP store in almost every town and village in South Africa). Currently the warehouse management system (WMS) implemented by PEP only allows for a stock keeping unit (SKU) to be placed on one picking line in one location when the distribution list (DBN) is released. Because pickers are only allowed to walk clockwise around the conveyor belt, they are forced to pass a location at least the same number of times as the number of branches to which the SKU is to be distributed to. Therefore if the SKUs with the highest pick frequency can be assigned to 2 locations (it is duplicating the SKU), the number of times each of these locations must be passed may be reduced. In this study 4 questions are considered when 15 algorithms are constructed that will determine how an algorithm assign the SKUs to picking lines. Question 1 determines whether the original picking lines are to be treated separately (PS) or to combine them rst (PC). The second question is to decide if the SKUs are rst to be duplicated and then assigned to picking lines (DA) or if they are rst assigned to picking lines and then duplicated (AD). Question 3 determines whether the non-duplicate and duplicate SKUs are treated separately (ND) or simultaneously (S) when they are assigned to the picking lines. The nal question is to specify how the SKUs are assigned to the picking lines. Three assignment methods (cyclical, set length subset sequential assignment, remaining high, low cyclical assignment) and 6 clustering algorithms are introduced. The conclusion is made that the SKUs with the highest pick frequency is duplicated rst to yield the biggest saving in the number of cycles. Between 40{70% of the SKUs should be duplicated, dependant on the algorithm used. The only decision that has a major in uence on the number of cycles is the assignment method used. Algorithm 5 and 8 yielded the greatest saving in the number of cycles (40.7% and 39.8% respectively), both implementing set length subset sequential assignment, followed by the clustering algorithms.

AFRIKAANSE OPSOMMING: PEP is 'n afdeling van Pepkor Retail Limited en is die grootste enkel-handelsmerk winkelnetwerk in Suidelike Afrika. PEP besit en bestuur ook die grootste klerefabriek in Suidelike Afrika. PEP is gestig in 1965 en het sedertien gegroei tot meer as 1 400 winkels in 9 Afrika lande (daar is 'n PEP winkel in amper elke dorp in Suid-Afrika). Op die oomblik laat die pakhuisbestuurstelsel, wat deur PEP in sy distribusie sentrum ge mplementeer word, slegs toe dat voorraadeenhede (VEs) in 'n enkele vakkie langs 'n enkele uitsoeklyn geplaas word. Aangesien werkers slegs toegelaat word om kloksgewys om die vervoerband te beweeg, word hulle gedwing om ten minste soveel keer verby elke vakkie in die uitsoeklyn te loop as wat die aantal winkels is waarna die VEs in daardie vakkie versprei moet word. Dus indien die vakkies wat die VEs bevat wat na die meeste winkels versprei moet word, tussen 2 vakkies verdeel word (die VE word gedupliseer), verminder die aantal kere wat beide vakkies besoek moet word. In hierdie studie word 4 vrae beskou wat geantwoord moet word wanneer die 15 algoritmes opgestel word, wat sal bepaal hoe die algoritme die VEs hanteer. Vraag 1 bepaal of die oorspronklike uitsoeklyne wat deur PEP verskaf is apart hanteer word en of hulle eers gekombineer moet word. Die tweede vraag bepaal of die VEs eers gedupliseer word en dan aan die onderskeie uitsoeklyne toegedeel word en of die VEs eers aan die uitsoeklyne toegedeel word en dan gedupliseer word. Vraag 3 is slegs van toepassing wanneer die VEs eers gedupliseer word en dan toegedeel word aan die uitsoeklyne, en bepaal of die nie-gedupliseerde en gedupliseerde VEs apart of gelyktydig hanteer word. Die laaste vraag spesi seer met behulp van watter metode die VEs toegedeel word aan die onderskeie uitsoeklyne. Drie toedelingsmetodes (sikliese toedeling, vaste lengte subversameling opeenvolgende toedeling, oorblywende hoogste/laagste sikliese toedeling) en 6 bondelalgoritmes word voorgestel. Die gevolgtrekking word gemaak dat die VEs met die hoogste uitsoek frekwensie eerste gedupliseer moet word om die grootste besparing mee te bring in die aantal siklusse om al die VEs uit te soek. Tussen 40{70% van die VEs moet gedupliseer word afhangende van die algoritme wat gebruik word. Die enigste besluit wat 'n noemenswaardige invloed op die aantal siklusse het is die toedelingsmetode. Algoritme 5 en 8 lewer die grootste besparing in die aantal siklusse (40.7% en 39.8% onderskeidelik), beide implementeer die vaste lengte subversameling opeenvolgende toedeling, gevolg deur die bondelalgoritmes.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/79925
This item appears in the following collections: