5. Distribution planning: Eldorado Distribution Network Design Problem description Eldorado Company, a large electronics retailer in Russia with stores in 350 cities, needed to determine the optimal number of warehouses, and where they should be located in order to better fulfil customer demand and minimize delivery and inventory costs. Model development The analysis showed that the problem could be solved with the introduction of the decision support system AnyLogic Logistics Network Manager. Solution development Input data provided by the customer described potential locations subject to rent cost, investments for building new or modernizing old warehouses, average level and cost of storage, overall costs for staffing and security, etc. The simulation model included in the solution also considered the warehouse and retail store GIS (geographic information system) coordinates, and distances between cities (Fig. 3). Figure 3: Potential locations Testing the solution Users can carry out several experiments with the model. The parameter variation experiment checks all possible scenarios of warehouse location taking into account fixed warehouses and their maximum number. The result of such an experiment is the best combination of warehouses that cost the least amount of money. Based on this information, an optimization experiment calculates in-store warehouses floor space. In a simple experiment, a user manually chooses warehouses from the list, and launches the model with this combination in order to receive statistics about it. Analysing the results The system as introduced allowed the client to simulate, in detail, several kinds of activities: - Daily basis (model time): goods are sold in stores, and losses from the shortage of demanded goods are counted. - Weekly basis: inventory is supplemented to target level, transportation costs are counted, and deferred payments to suppliers are planned. - Monthly basis: warehouse levels are renewed according to monthly sales levels of stores, transportation routes from warehouses to stores are generated, and franchisee shipments are planned. Monthly sales numbers conform to average sales numbers, while daily sales are generated stochastically. Implementation The solution allowed the customer to choose the warehouse network out of \63,000 combinations. The software implementation costs are paid off during the first two months of work when using the distribution network system recommended by the model. The decision support system is expected to operate for a long time, as it allows the users to find new optimal distribution system setups in case the market situation changes (change of transportation tariffs, warehouse parameters, amount of stores and sales, etc.). Case Transportation: direct shipment vs milk-runs Transportation consolidation may help in reducing transportation unit costs. However, in some cases it is impossible to use a distribution centre or warehouse to make use of consolidation effects. For example, organic farms do not usually use a warehouse for perishable goods and deliver directly to the markets (Fig. 1). But also in this case it may be possible to increase efficiency. For the organic farms two options will be analysed: independent distribution with direct shipment from each farm to the customers vs cooperative distribution using the milk-run system through the formation of partnerships between farms located in the same region. The cost analysis has to answer two questions: (1) the truck size to be used according to the replenishment interval; and (2) whether it is worth using direct shipment from each farm to the customers or to apply a partnership formation utilizing the milk-run method. In an organic supply chain in Brazil, four farms are located in Itaipava, a city located 95km away from the centre of Rio de Janeiro. These farms supply six street markets, ten restaurants and four organic markets per week. The trucks they use for transportation have a capacity of 6t. Total weekly demand (restaurants, street markets and stores) is around 36t. The costs in this analysis are estimated as close to the reality of a supply chain in Rio de Janeiro as possible. The following assumptions exist: 1. All customers are located close to each other so the transportation costs inside the city can be neglected. 2. Maximum replenishment interval is three days. 3. Each supplier delivers exactly one product to all customers (i.e. carrots, lettuces, tomatoes and onions), so there are four different products in this supply chain with proportionally distributed demand. 4. At present, trucks with 6t capacity are used. 5. Three truck sizes can be considered in the analysis: capacity of 4t, 6t and 10t respectively. 6. Fixed costs for one delivery is R$100, R$150, and R$180 respectively. 7. Costs per delivery at the distance of 95 km is R$20 assuming R$0.2105 per km. 8. Ordering cost is R$6, unit price is R$2 and interest rate is 0.3%. 9. Daily cost of small truck = R$120.00 10.Daily cost of medium truck = R$ 170.00 11.Daily cost of large truck = R$ 200.00 12. Replenishment can be chosen for every two days, three, and every day (assuming that 1 week has 6 working days) Questions: A. If the delivery follows direct shipment and the weekly demand for each farm is 9t (1.5t per day), determine the most efficient truck size to be used according to the batch-size and replenishment interval? B. Re-evaluate the decision that you have taken if the delivery follows milkruns?