Advanced

Benefits and risks of implementing multiechelon optimisation at Scania Parts Logistics

Nordlund, Sara LU (2014) MTT820 20132
Engineering Logistics
Abstract
The purpose of this project is to find out if it will be beneficial to implement Multi-Echelon Optimisation(MEO) at Scania Parts Logistics (SL). SL is the department at Scania that is responsible for the spare parts distribution. The IT-system used for planning at SL is currently being replaced. Even though the new system, Spare Parts Management (SPM), can perform MEO it was decided not to use it during the initial roll out. SL however wants to know if they should implement MEO in the future.

The project is performed with a deductive approach; the aim is to validate how the theory works in practice. The intention was to prove that MEO will be beneficial for SL to use by comparing the three different scenarios:
1. The old planning... (More)
The purpose of this project is to find out if it will be beneficial to implement Multi-Echelon Optimisation(MEO) at Scania Parts Logistics (SL). SL is the department at Scania that is responsible for the spare parts distribution. The IT-system used for planning at SL is currently being replaced. Even though the new system, Spare Parts Management (SPM), can perform MEO it was decided not to use it during the initial roll out. SL however wants to know if they should implement MEO in the future.

The project is performed with a deductive approach; the aim is to validate how the theory works in practice. The intention was to prove that MEO will be beneficial for SL to use by comparing the three different scenarios:
1. The old planning system
2. The new planning system, SPM, with standard settings
3. SPM with MEO activated
For scenario 1 existing data was gathered from different sources at SL. This scenario was included to be able to validate the outcomes from the new system; since it was the first time anyone at Scania used SPM. For scenario 2 and 3, data regarding the spare parts was gathered and inserted into SPM. SPM was then run, in order to receive results that could be used for comparison.

It was not possible to analyse the entire assortment, which consists of more than 100 000 parts, in the scope of this project. Instead a group of parts that could represent the complete assortment was chosen. SL had divided the assortment into different segments, according to an ABC-classification based on part cost and part demand. These segments were used to find the parts to analyse: a few parts from each segment, about 50 parts in total.

The first analysis compared scenario 1 and 2, in order to see that SPM generated realistic results. It was quickly found that it did not; after performing a data validation, several parameters and settings needed to be adjusted in order to get accurate results. The old planning system has many limitations and uses quite simple methods to calculate forecasts and safety stock. With the new planning system, the expectations were to get more accurate results. Findings showed that the change in forecasts and safety stock levels differed from part to part: the levels both increased and decreased.

Similar findings where made when comparing scenarios 2 and 3, the difference varied highly between parts. With such a high variation of results, it was not possible to state any conclusions based on these numbers alone. However, with MEO the average safety stock levels at Scania Parts Center (SPC), the central warehouse in Belgium, decreased by about 3%. At the seven Regional Warehouses (RW’s) located around Europe, several parts were removed from stock.

Average inventory levels were compared between scenario 1 and 3. It was shown that the average inventory levels at SPC decreased by 25%, while at the RW’s the levels increased. This kind of result is also supported by the theory of MEO.

The RW’s only deliver emergency orders, orders that should be delivered within 24 hours. All orders to refill stock are delivered straight from the central warehouse, SPC. The theory of MEO however, is based on regional warehouses supplying all types of orders. Therefore there is a risk of using MEO at the regional warehouses because the effects of MEO at such installations are not known. SL’s intention is to soon include a larger part of the distribution system in SPM. If this was to happen, the effects of using MEO will likely become greater.

The conclusion was that further analysis is needed before deciding whether to implement MEO. The initial focus should be on making the standard version of SPM work as desired. When that is achieved, further evaluation of MEO can be carried out. Firstly, the entire assortment should be analysed in order to get figures that can be trusted. Alternatively, a better segmentation should be done to find more general parts to represent the assortment. Secondly, SL needs to consider if the RW’s should be included in the scope of MEO. If not, the recommendation will be to wait with implementing MEO until more nodes are included in SPM. Then the benefits of using MEO will likely exceed the cost of resources required to implement it. (Less)
Abstract (Swedish)
Syftet med detta projekt är att ta reda på om det är fördelaktigt för Scania Parts Logistics (SL) att implementera flernivåoptimering, så kallad multi-echelon optimisation (MEO). SL är den avdelning på Scania som är ansvarig för distributionen av reservdelar. För tillfället håller det IT-stöd som används för planering på att bytas ut på SL och i det nya systemet, Service Parts Management (SPM), är det möjligt att aktivera en funktion som utför flernivåoptimering. Till en början har SL valt att inte implementera denna funktion, däremot vill de veta om de borde implementera den i framtiden.

Projektet är genomfört med en deduktiv ansats, vilket innebär att teorin kring flernivåoptimering kommer att testas genom att simulera verkligheten.... (More)
Syftet med detta projekt är att ta reda på om det är fördelaktigt för Scania Parts Logistics (SL) att implementera flernivåoptimering, så kallad multi-echelon optimisation (MEO). SL är den avdelning på Scania som är ansvarig för distributionen av reservdelar. För tillfället håller det IT-stöd som används för planering på att bytas ut på SL och i det nya systemet, Service Parts Management (SPM), är det möjligt att aktivera en funktion som utför flernivåoptimering. Till en början har SL valt att inte implementera denna funktion, däremot vill de veta om de borde implementera den i framtiden.

Projektet är genomfört med en deduktiv ansats, vilket innebär att teorin kring flernivåoptimering kommer att testas genom att simulera verkligheten. Målet är att visa att det är mest fördelaktigt för SL att använda flernivåoptimerings-funktionen, genom att testa tre scenarion:
1. Det gamla planeringssystemet
2. Det nya planeringssystemet, SPM, med standardinställningar
3. Det nya planeringssystemet, SPM, med flernivåoptimerings-funktionen aktiverad

Scenario 2 jämfördes med scenario 1 för att verifiera att resultaten från det nya planeringssystemet var godtagbara. Det konstaterades snabbt att så inte var fallet: flera parametrar och inställningar behövde justeras innan resultaten blev korrekta.

Då det inte var möjligt att analysera alla de över 100 000 reservdelar som finns i lager i distributionssystemet, valdes en grupp av artiklar ut som kunde representera det kompletta sortimentet. SL hade en segmentering av sortimentet, baserad på en ABC-klassificering där artiklarna var klassade efter deras kostnad och efterfrågan. Denna segmentering användes som bas för att välja ut de cirka 50 artiklar som analyserades.

Resultaten visade att hur mycket säkerhetslagernivåerna och de beräknade prognoserna ändrades, skiljde sig från artikel till artikel. Då det nya systemet har mer avancerade räknemetoder, var förväntningen att prognoser och säkerhetslagernivåer skulle stämma bättre överens med verkligheten. Det skulle kunna förklara varför skillnaderna både ökade och minskade. Liknande upptäcker gjordes när scenario 2 och 3 jämfördes: skillnaderna i hur lagernivåerna ändrades skiljde sig mycket åt mellan olika reservdelar. Det var därför väldigt svårt att dra slutsatser om hur skillnaderna hade blivit för det totala sortimentet och vilket scenario som faktiskt gav bäst resultat.

Då flernivåoptimerings-funktionen var aktiverad i SPM, sjönk antalet artiklar som skulle lagerhållas på de sju regionala lager som finns i Europa. Samtidigt visades att medellagernivån på dessa platser istället ökade. Slutsatsen var att genom att använda flernivåoptimering så skulle ett mindre antal artiklar hållas på regionala lager, men istället skulle fler enheter av resterande artiklar finnas tillgängliga på lagren.

På centrallagret i Belgien, Scania Parts Center (SPC), skulle medellagernivån sänkas med 25 %, jämfört med nuvarande nivåer. Säkerhetslagernivåerna sänktes ytterligare med 3 % när flernivåoptimering var aktiverat, jämfört med endast standardversionen av SPM. Dessa siffror är inte tillräckliga för att dra en klar slutsats om ifall flernivåoptimerings-funktionen bör aktiveras, men de visar åt vilket håll det verkar luta.

De regionala lager som finns i SL’s distributionssystem hanterar endast akuta order, då reservdelar måste levereras inom 24 timmar. De vanligaste orderna, de som görs för att fylla på i lagren, skickas alltid direkt från centrallagret SPC. Teorin kring flernivåoptimering är baserad på lager som hanterar alla typer av order, det finns därför en risk med att optimera regionallager, då det är oklart vilken effekt flernivåoptimeringen skulle få. I framtiden planerar SL att ta in en större del av distributionssystemet i SPM, då skulle effekterna av flernivåoptimering kunna få en mycket större omfattning.

Slutsatserna är att fler analyser är nödvändiga för att kunna bestämma om flernivåoptimering ska implementeras eller inte. För att få tydliga resultat bör hela sortimentet tas med i analyserna, alternativt krävs en bättre segmentering och ett noggrannare urval av artiklar. SL bör också fundera över om de anser att regionala lager bör vara med i optimeringen. Om de lämnas utanför, är rekommendationen att vänta med att använda flernivåoptimering tills fler noder installeras i SPM. Då är det mer troligt att fördelarna med att implementera flernivåoptimering skulle överstiga riskerna. (Less)
Please use this url to cite or link to this publication:
author
Nordlund, Sara LU
supervisor
organization
course
MTT820 20132
year
type
M1 - University Diploma
subject
report number
5756
language
English
id
4463093
date added to LUP
2014-08-25 13:10:44
date last changed
2014-08-26 13:05:43
@misc{4463093,
  abstract     = {The purpose of this project is to find out if it will be beneficial to implement Multi-Echelon Optimisation(MEO) at Scania Parts Logistics (SL). SL is the department at Scania that is responsible for the spare parts distribution. The IT-system used for planning at SL is currently being replaced. Even though the new system, Spare Parts Management (SPM), can perform MEO it was decided not to use it during the initial roll out. SL however wants to know if they should implement MEO in the future.

The project is performed with a deductive approach; the aim is to validate how the theory works in practice. The intention was to prove that MEO will be beneficial for SL to use by comparing the three different scenarios:
1. The old planning system
2. The new planning system, SPM, with standard settings
3. SPM with MEO activated
For scenario 1 existing data was gathered from different sources at SL. This scenario was included to be able to validate the outcomes from the new system; since it was the first time anyone at Scania used SPM. For scenario 2 and 3, data regarding the spare parts was gathered and inserted into SPM. SPM was then run, in order to receive results that could be used for comparison.

It was not possible to analyse the entire assortment, which consists of more than 100 000 parts, in the scope of this project. Instead a group of parts that could represent the complete assortment was chosen. SL had divided the assortment into different segments, according to an ABC-classification based on part cost and part demand. These segments were used to find the parts to analyse: a few parts from each segment, about 50 parts in total.

The first analysis compared scenario 1 and 2, in order to see that SPM generated realistic results. It was quickly found that it did not; after performing a data validation, several parameters and settings needed to be adjusted in order to get accurate results. The old planning system has many limitations and uses quite simple methods to calculate forecasts and safety stock. With the new planning system, the expectations were to get more accurate results. Findings showed that the change in forecasts and safety stock levels differed from part to part: the levels both increased and decreased.

Similar findings where made when comparing scenarios 2 and 3, the difference varied highly between parts. With such a high variation of results, it was not possible to state any conclusions based on these numbers alone. However, with MEO the average safety stock levels at Scania Parts Center (SPC), the central warehouse in Belgium, decreased by about 3%. At the seven Regional Warehouses (RW’s) located around Europe, several parts were removed from stock.

Average inventory levels were compared between scenario 1 and 3. It was shown that the average inventory levels at SPC decreased by 25%, while at the RW’s the levels increased. This kind of result is also supported by the theory of MEO.

The RW’s only deliver emergency orders, orders that should be delivered within 24 hours. All orders to refill stock are delivered straight from the central warehouse, SPC. The theory of MEO however, is based on regional warehouses supplying all types of orders. Therefore there is a risk of using MEO at the regional warehouses because the effects of MEO at such installations are not known. SL’s intention is to soon include a larger part of the distribution system in SPM. If this was to happen, the effects of using MEO will likely become greater.

The conclusion was that further analysis is needed before deciding whether to implement MEO. The initial focus should be on making the standard version of SPM work as desired. When that is achieved, further evaluation of MEO can be carried out. Firstly, the entire assortment should be analysed in order to get figures that can be trusted. Alternatively, a better segmentation should be done to find more general parts to represent the assortment. Secondly, SL needs to consider if the RW’s should be included in the scope of MEO. If not, the recommendation will be to wait with implementing MEO until more nodes are included in SPM. Then the benefits of using MEO will likely exceed the cost of resources required to implement it.},
  author       = {Nordlund, Sara},
  language     = {eng},
  note         = {Student Paper},
  title        = {Benefits and risks of implementing multiechelon optimisation at Scania Parts Logistics},
  year         = {2014},
}