Case Study 1 – Hungary Distribution Center
The case study refers to distribution networks situated in Hungary. They comprise 30 regions with demands to be satisfied. In the initial setting, the region are served from a central distribution center (DC) located in Budapest.
The main operational parameters of the distribution networks are:
- Product: mineral water in 2-liter bottle, its cost and selling price are 0.2 and 0.5$/bottle, respectively.
- Demands: constant over time, proportional to the number of inhabitants in the given regions, 0.4 liter/day/inhabitant.
- Order parameters of the regions: order interval = 5days,
- expected lead time (ELT) = 7days, backorder is not allowed.
- Parameters of the DCs: carrying cost = 0.001$/m3/day, initial cost of an additional DC = 1 million $, the additional DCs use min-max inventory policy with a periodic check of 1 day;
- Transportation: trucks with capacity 50m3 and speed 60 km/h, with cost 0.05$/m3/km calculated with actual routes.
- The time period considered is 1 calendar year, with a 1-month disruption at the central DC in Budapest.
The initial structure (image above) is extremely vulnerable. The structure was augmented step by step: first, one additional DC (in Siófok) was added to the network, which was supplied from the DC in Budapest. Three of the regions were served exclusively from Siófok, three other regions from both Siófok and Budapest, and the remaining 24 regions entirely from Budapest (image on the left, upper part of the image).
The next network was differentiated from the preceding one only in the numbers of how many regions were served solely by Siófok and how many by Budapest and Siófok together. These numbers were 6-6 (image on the left, lower part).
Lastly, two other networks were created both having two additional DCs. Similarly, to the cases with one additional DC, these DCs served some regions exclusively and the same numbers jointly with Budapest (Fig.10).
Comparing the values of the complexity measures, from Tab.1, in rows 2-5, with the values of the basic distribution network controlling only one, central DC (0-0-0, first row), it can be seen that all measures exceed their initial values. Within the blocks of networks with the same number of additional DCs, the increase is monotonous.
The monotonous increase of the entropy values is experienced for all the consecutive networks. Looking at the robustness related measures, maximum of the normalized betweenness centrality and factor R, it can be observed that in case of the networks belonging to the same block, their augmentation with further edges led to growing robustness measures. This outcome is in harmony with the general perception that if a higher proportion of the regions applies multiple sourcing, the network’s robustness increases.
This part shows how the different distribution networks behave in case of disruption. In the case under discussion, the disruption occurs at the central DC.
It causes ripples, its negative effects gradually spread across the whole distribution network, from the central DC, through the additional DCs and finally to the regions. Clearly, the additional DCs play a crucial role in mitigating the ripple effect and consequently, a logical way was to concentrate on their inventory levels.
A large number of KPIs (Key Performance Indicator), such as number of bottles sold, costs related to inventory and transportation, revenue, profit and service level, were determined through simulation.
For this purpose, the AnyLogistix supply chain software was used. The three KPIs analyzed and illustrated were the number of bottles sold (Fig.11), the profit (Fig.11) and the accumulated service level by orders (Fig.12).
Altogether 30 cases were considered, 1 in the basic structure, 4,9,4 and 12 in structures 1-3-3, 1-6-6, 2-3-3 and 2-6-6, respectively. Within the sets with the same structural complexity, the consecutive cases incorporated improved operational complexity, by increasing the parameters of the min-max inventory policy of the additional DCs step by step. In each step, the min and the max parameters were increased by the same amount (500.000 bottles).
As expected, the smallest amount of mineral water was sold in the basic network with no additional DC. The amount monotonically increased each set in parallel with the network’s enhanced operational complexity.
Looking at the profit, it can be seen that the inclusion of the additional DCs having initial cost, in most cases resulted in decreased profitability. This phenomenon is called “the cost of robustness”. However, in the networks with 1-6-6 and 2-6-6 structures, with relatively large inventory policy parameters, the profit could even surpass the value yielded by the reference network.
Interpretation of the results
The disruption considered in the case study lasted for one month at the central DC in Budapest. Throughout that period no region could be served directly from this DC. Therefore, modification had to be initiated in order to relive the consequences of the temporary shutdown.
The question was how to balance the aspects of robustness, complexity and efficiency while mitigating the ripple effect of this disruption on the other parts of the investigated distribution networks. For this purpose, several strategies were implemented and their impacts analyzed. The strategies were as follows:
- The augmentation of the starting distribution network with one additional DC, as structural modifications.
- The use of multiple sourcing in different extents, as structural modifications.
- The step-by-step increase of the min-max inventory policy parameters of the additional DC, as operational modifications;
•Mitigation of the ripple effect in supply chains: Balancing the aspects of robustness, complexity and efficiency, Judit Monostori