University of Twente Student Theses


Optimization of the inbound supply chain for small sized materials with low demand

Vriezekolk, B. (2020) Optimization of the inbound supply chain for small sized materials with low demand.

[img] PDF
Abstract:When operating in a saturated market, innovation is key to persuade new customers and to keep current customers loyal. One of the many innovative products that Heineken has brought to market are the Blade and Brewlock products. Not only the product itself is unique, but also the inbound supply chain is unique for Heineken Netherlands Supply (HNS). All Blade and Brewlock materials are single-sourced from suppliers all over Europe to several independently organised breweries across Europe. This does not only bring economies of scale for the production of these materials for the supplier, but it also brings a risk in terms of security of supply to the different Heineken breweries. This unique situation including the small size of materials and the production speed of packaging lines result in an unusual and probably inefficient supply chain, as HNS is used to big volumes. We identify the transfer price being too high as an action problem. Eventually, the price of the Blade and Brewlock products that HNS calculates to Heineken Netherlands Commerce and other Operating Company (OpCo)s will decide if these relatively new products will be a success and profitable to HNS. The quantities shipped (call-off order quantities) to HNS are sourced with lot-size based quantity discounts. The impact of these lot-size based quantity discounts are unknown for HNS as they are used to big volumes. Secondly, this research is focused on the future situation, where a new packaging line is taken into use at different brewery. Inter-brewery transport could be of use to move materials cheaply between the two breweries. This way more lot-size based quantity discounts and lower total cost can be achieved We have investigated the contracts and mentioned discounts. The demand forecast mentioned in contracts are often very optimistic and the minimum order quantities are not always easily achieved by an individual OpCo, so contracts can be misleading for predicting lot sizes. To get some insights in the potential savings, we have compared the current situation against the situation where all lot-size based quantity discounts are used. The maximum potential savings of using larger order quantities are e 66,372 in two years. If we would never use the smallest lot size, already 83% of these savings are achieved. As annual demand increases for a material, the importance of lot-size-based quantity discounts increase as well, as larger demands without increasing order quantities result in a larger number of orders, where discounts are missed more often. Inter-brewery transport could be of use to move materials cheaply between the two breweries, this gives the possibility to merge orders from both breweries to the supplier. The single-source situation of Blade and Brewlock materials makes Vendor Managed Inventory (VMI), where the supplier is in control of distribution material to the breweries, interesting. However, VMI may be hard to implement, because Heineken is a very decentralised organisation, where every brewery can have different working methods. Our research is focused on two breweries in the Netherlands, both part of the HNS. The problem we are optimizing can be described as a multi-item, multi-location lot-sizing problem. Our problem instances seem quite small, so computation time should not be of the biggest importance. So, we can use AIMMS as our modelling software, HNS is also familiar with this software. A MILP formulation is used for solving our lot-sizing problem, because this is a common formulation and because the model is formulated linearly, this makes it relatively fast to solve. We have extended a standard multi-item lot-sizing problem with quantity discounts, a rolling horizon, multiple inventory locations, with transportation between the locations and the possibility of certain material types to free ride on the quantity discounts of other materials. A rolling horizon is used to simulate the current ordering process. We have built several versions of the model to test different scenario’s. A simple one-location model is used to validate the model against the current situation. We have verified the model and we assume the model to be correct because the model gave logical output given the historical shipping data we have used as input. All of the average order quantities were larger than the average weekly demand quantities, therefore the model bundles demand of several weeks. And the average order quantities increased as the planning horizon is increased, except when all of the quantity discounts are achieved. Also did the total cost decrease as the model was solved with more information, i.e. a longer planning horizon. The effect of incorporating an extra location with more demand has a big effect on the average order quantities and promotes the use of lot size based quantity discounts. When the demand increases (by adding an extra location), the importance of using a v longer planning horizon becomes less important as the maximal quantity discounts are more easily reached. The result of the scenarios we have tested resulted in significant savings, of maximally e 373,714 which can be achieved by increasing the planning horizon and using a two-stop-route with direct delivery for the two materials sourced from the same supplier. Average order quantities for four out of seven have increased by 5% to 33% by the outcomes of the base scenario. So, it would be cost-efficient to make more use of these discounts, in general as these four materials are more important. To bundle the demand in one order quantity decision we can make use of inter-brewery transport or a direct delivery with two stops. Direct delivery is always cheaper when the average order quantity is above twelve pallets. This can be achieved by a using a two-stop-route. And solve the model with shared ordering cost for the materials that are sourced from the same supplier result in even larger order quantities. Two material types should not be ordered in larger quantities, holding costs are high and lot-size based quantity discounts are small. So, inter-brewery transportation is also not useful for these material types because inter-brewery transportation is only useful for full pallets and the quantities on a pallet for these materials are too large. The model proposed in this thesis can be used in future for HNS as well. The model is made as generic as possible and has an understandable user interface that can be used by the procurement department. The model is transferred to the Tactical Supply Chain Planning (TSCP) department and can be adjusted if needed by employees with knowledge of AIMMS. By modifying some constraints and using different input, several interesting scenario’s can be tested. The model can be used to investigate the usefulness of inter-brewery transport, maybe for other materials, this can be a really interesting way of dividing inventory across breweries. There does not exist a tool that easily takes into account the free-riding of materials. The model can be used to investigate other materials where free-riding is already happening or possible. Also, the model can be used to make better decisions for the trade-off between groupage transportation and direct delivery. Where you can easily calculate when direct delivery becomes cheaper, the model calculates the expected average order quantities given the prices for groupage transportation or direct delivery. Furthermore, the ordering process itself can be investigated, especially changes in the planning horizon and planning interval. And lastly, the use of warehouse space during the year can already be predicted by the model. However there are other ways to calculate the warehouse space required, the model gives an outcome for each period and can therefore predict tight periods. In general, the model can be of use for new materials or situations where an alternative way of ordering could be better than the current. A change in demand, costs or space available can trigger situations where the inbound supply chain should be alternatively organised. The model can explore settings and help in decisions to be made to organise this supply chain. We hope the proposed model and the insights found in this thesis do not only help for improving the Blade and Brewlock inbound supply chain but also helps in future situations.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:31 mathematics, 58 process technology, 85 business administration, organizational science
Programme:Industrial Engineering and Management MSc (60029)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page