Abstract

The goal of this paper is to analyze the effect of simultaneously solving the supply network flow, internal network flow and distribution network flow of Bega Cheese company. By routing in such a way to minimize the cost and meet the customer demand. We design and develop the network in excel involving supply, internal and distribution flows, build the model with Objective functions, decision variables and constraints to optimize the cost using Linear Programming (LP) tool.

1. INTRODUCTION

The network study of a F&B manufacturer called Bega Cheese, head quartered in Melbourne, Australia wants to optimize the supply and distribution network. The current network considered for the study has 151 suppliers, 6 manufacturing locations, 17 warehouses, 783 customers and 3305 SKUs which includes raw material, Semi-finished goods and finished goods.

This paper study the supply, manufacturing and distribution of cheese products and by-products. The focus is on routing the vehicles which is closely linked to the vehicle cost, the number of chill and frozen locations and the associated handling and inventory costs. It is therefore important to see the effect of simultaneously considering the location and routing of the network.

The contributions of the paper are as follows
1. Prepare the Base Line for the existing Bega Cheese network
2. Develop the existing model in Excel and design and execute different models with respect to current and future requirements to optimize the complete network
3. Integrate open solver with Excel to optimize footprint to reduce operating costs while retaining 100% service level

2. PROBLEM DESCRIPTION AND FORMULATION

Bega Cheese has wide range of cheese products and by-products which are supplied across Australia. No.1 F&B manufacturer in Australia has complex network operation from supplier to customer. The details of the network and associated complexities are briefed below

2.1 Network Nodes
Supply Network: The supply network involves suppliers located in one region, which supplies raw materials directly to manufacturing facilities and through warehouses owing to capacity constraints at manufacturing facilities. Suppliers also supplies semi-finished goods and finished goods which are supplied to customers through warehouses and semi-finished goods to manufacturing facilities for further processing

Internal Network: the internal network involves production and supply of semi-finished goods and finished goods from manufacturing facilities to Customers through warehouses. Also, includes internal flows from one manufacturing facility to another manufacturing facility directly and through warehouses which involves Raw material, semi-finished goods and finished goods

Distribution Network: The distribution network flow involves flow of finished and semi-finished goods to customers through different warehouses which receives from suppliers and manufacturing facilities

2.2 Assumptions
Manufacturing Facilities:
• No change in current manufacturing Locations
• No Planned acquisitions for next 3 years
• No production capacity constraints to meet the customer demand
• No warehouse capacity constraint

Products (SKUs):
• The products are classified as Raw Material (RM), Semi-Finished good (SFG) and Finished good (FG)
• These products are further classified as Ambient, Chill and Frozen
• Active list obtained from the transaction data
• Product classification to identify product movements based on item category
• Conversion factors derived based on product weight & max. pallet weight
• Packaging conversion factors obtained from suppliers based on loading patterns

Demand:
• Demand volume is estimated to be stable for next three years
• Product mix requirement from customers may change based on new product introductions
• Ex-works are picked from warehouses by customers

Inventory policies:
• Current storage volumes & costs used for baseline
• 4 weeks inventory considered for all products for scenario modelling
• Ancillary & case pick charges are considered as lump-sum
• Handling charge calculations are based on existing rates

Sourcing policies:
• All supply locations are from Melbourne to facility/warehouse
• All FIS (Free in Store) movements either to warehouse or facility are with no transportation charge

Transportation policies:
• Customer demand was clustered to 22 domestic routes, one international route for exports (up to Port
Melbourne) and ex-works for customer pickups
• FTL (Full Truck Load) rates considered for internal movements
• Blended rates are derived based on truckload utilization for customer routes

Our objective is to simultaneously find the number of warehouses required and vehicle routes that would minimize overall transportation cost, handling cost and storage cost and also find out the outcome of the user defined models with respect to different costs involved in the multi echelon network.

2.3 Constraints
The constraints are:
1. 3 different types of vehicles to collect ambient, chill and frozen material
2. Transportation, handling and storage costs considered are per pallet
3. Demand fulfillment is 100% and to be met at item level

Further to the above constraints, there will be flow constraints from source to destination at different nodes to match the demand of customer at item level, which will be explained in detail ahead.

2.4 Formulation
Let L be the transportation cost per pallet from a source location to destination, H be the handling cost per pallet with respect each warehouse and S be the storage cost with respect to each warehouse

Input parameters:
• S=Suppliers which supplies RM, SFG and FG to manufacturing facilities and warehouses
• F=Manufacturing facilities which receives RM, FG and SFG from suppliers, warehouses and other facilities
and transfers RM, SFG and FG to warehouses and other manufacturing facilities
• W=Warehouses which receives RM, SFG and FG from suppliers and manufacturing facilities and transfers RM,
SFG and FG to manufacturing facilities and Customers
• Lij=Logistics cost from node i to node j. It is the cost involved to transfer a pallet from a source to
destination by truck
• Hw=Warehouse handling cost of a specific warehouse. It is the cost involved to unload/load a pallet from
a warehouse
• Sw=Warehouse storage cost of a specific warehouse. It is the cost involved to store a pallet per week in
a warehouse
• Rs=Requirement from each supplier. It is the requirement of RM, SFG and FG planned from a supplier
• Rf=Requirement from each manufacturing facility. It is the requirement of RM, SFG and FG planned from a
manufacturing facility
• Dc=Demand from each customer. It is the demand from customer of SFG and FG from warehouse

Decision Variables:
• Rij>=0 if there the requirement is fulfilled from node i to node j. The decision variable is formulated
in such a way that the requirement fulfilled from a source to destination should be greater or equal to
zero so that there is no negative integer exists while minimizing the cost
• Rij=0 if there is no supply between node i and node j. Zero supply from a source and destination
• Rij=integer, the requirement is fulfilled in Full pallets. The decision variable is formulated in such a
way that the requirement fulfilled from a source to destination should be equal to zero so that the
requirement fulfilled is a whole number which means there are supply of fraction pallets
• Dj=0 the unmet demand at destination point should be zero. The demand requirement from a destination
point should be met 100% and the source point supply should be 100%

Objective Function:
The Objective function is to minimize the cost which is product of Logistics cost and number of pallets transferred from a source to destination, product of handling cost and number of pallets transferred from a source to destination and product of storage cost and number of pallets transferred from a source to destination with 4 weeks of inventory

Min Z = ∑LijRij + ∑HwRij + ∑SwRij

3. Design of Models

3.1 Scenarios
Scenario-1: optimization of existing network
• Optimization of existing network with all the warehouses used in baseline
• Arrive at optimal number of locations/warehouses
• Arrive at the costs of optimal network suggested by the model
• Arrive at new product flows

Scenario-2: centralized network
• 2 warehouses with only 1 having cold storage facility
• Arrive at the cost of the selected network and product flows

Scenario-3A & 3B:
• Entire network will be operated with 3 selected warehouses
• Scenario 3A – all existing costs of the selected warehouses are considered
• Scenario 3B – handling cost of a specific warehouse is changes and arrived at the total cost
• Arrive at the cost of the selected network and product flows

Scenario-4:
• 4 different warehouses are selected for the network design
• Arrive at the total cost of the selected network and product flows

3.2 Material movement
Material Movements Considered for the model
• Supplier to WH/Facility
• Facility to Warehouse
• Warehouse to Facility
• Facility to Facility
• Warehouse to customer

3.3 Modeling considerations

4. Key findings and Recommendations

The network modelling is done for 5 scenarios as explained earlier and 4 runs for each scenario which gives us 20 different options optimized networks which leads to take further decisions on implementation

The 20 different options are compared with the baseline model and costs of different parameters are compared like Logistics, handling, storage, internal, last mile etc. to arrive at the inference. Out of 20 options of 5 different scenarios top 3 options which gives maximum benefits with 100% service level are considered for further comparison and considering other practical implementation difficulties one option is considered for implementation and proposed the recommendations with respect to selected option

The key findings are mentioned below
• Total product volumes moving within the network: Ambient – 42%, Chilled – 57%, Frozen – 1%
• 80% of customer demand originates from 4 routes out of 23 routes
• Fulfilling Customer demand is often taking priority over the cost of supplying the product leading to sub
optimized network
• Inter warehouse movements in the current supply chain network are costing approximately 1.6 million AUD
per annum which is between warehouses which is unnecessary movement in the network
• Scenario-1(Run 3) model produced the best possible outcome with over-all cost savings of approximately
8.8 M AUD
• Scenario-4(Run 3) model is the second-best option with over-all cost savings of 6.85 M AUD

With Scenario-1(Run 3) options being selected for implementation and the following recommendations were
proposed to have maximum benefit

• Restructure the current logistics network of 17 warehouses to 8 warehouses (5- Ambient & 4- Chilled
warehouses) to achieve the 8.8 M AUD annual cost saving
• Shifting Ambient pallets to only 4 warehouses from 10 different warehouses and reducing inventory from
6.01M pallets per annum to 2.41M pallets per annum results in 3.6 M AUD
• Rerouting Chilled routes yielding more benefits of 1.96 M AUD with an additional cost of 0.64 M AUD
incurring in Facility to Warehouse transportation cost. The highest benefits yielding routes are
mentioned below with 100% service level
o Route1 – Changing the supply from present 8 WHS to 2 WHs reduces the transportation cost by 0.96 M AUD
o Route2 – Changing the supply from present 6 WHS to 1 WHs reduces the transportation cost by 0.42 M AUD
o Route3 – Changing the supply from present 7 WHS to 1 WHs reduces the transportation cost by 0.22 M AUD
o Route4 – Changing the supply from present 4 WHS to 1 WHs reduces the transportation cost by 0.20 M AUD
• Reducing the inventory annually from 3.75 M pallets to 2.42 M pallets gives an annual saving of 4.5 M AUD

5. Conclusion

To implement the selected option there are some prerequisites to be met
• Implementation of changes must be done in waves
• Each wave can be derived based on prioritization of modification requirements
• Prioritization strategy for modifications can be done in three ways
o Prioritization of routes giving maximum benefits for quick realization
o Prioritization of customer-route combination that has least impact on customer service/ business
o Prioritization of changes by warehouse
• Prerequisites for transformation based on the following
o Warehouse capacities – Align with 3PLs for required warehouse capacities
o Transportation routes- Align with Carrier companies with the fulfillment routes suggested by the model
o Warehouse/transportation contracts- commence review & adjust the contracts with the suggested warehouses
o Facilities – Align facilities to reroute the items to customer fulfilment warehouses suggested by model
and receive raw materials from the suggested warehouses
o Suppliers – Align suppliers to modify delivery routes to receive items to the respective warehouses

By Kumaraswamy DS – Senior Process Consultant