Optimal geographical positioning of a warehouse or distribution centre is the single most important logistics network decision. The positioning of single, or multiple, warehouse facilities and the allocation of customer demand to those facilities will have a long-term impact on transport cost, inventory cost and service levels.
The positioning of warehouse facilities, through warehouse location analysis, is usually a long-term decision that must be fully aligned with the business strategy, specifically against plans for territory expansion, order profile changes and volume changes. The location of the warehouse will drive the business’s long-term supply chain performance and consequently it is a decision that demands deep-dive analysis.
To undertake this deep-dive analysis, and to arrive at a location decision that is right for the business in the long-term, there are a series of steps that must be undertaken. Each of these steps is outlined below.
Step 1: Frame the problem that needs to be solved
The very first thing that needs to be undertaken is an accurate description of the problem that needs to be solved. Should the location of your facilities minimise outbound transport costs, or both inbound and outbound transport costs? Are there service parameters i.e. each customer must be served within x hours that will drive the need to establish multiple warehouse locations? If multiple locations are required should the cost of inventory be factored into the analysis?
To give you a rule of thumb, usually, where a single warehouse is required then the analysis is simply a balance of inbound and outbound transport costs. Where multiple locations are required it is normal practice to also consider warehouse facility costs and inventory costs.
Once you have accurately framed the problem, and decided on the cost and service factors to be optimised (inbound transport, outbound transport, warehouse costs, inventory costs) then you need to establish the data-set to be modelled, starting with the demand profile.
Step 2: Establish the demand profile
To establish the demand profile, you will need the points of demand (customer addresses) along with the volume allocated to each point (sales by customer). Caution is urged here – investing in new warehouse facilities is a long-term commitment and consequently just using historical demand, or a demand snapshot, may not reflect the planned future of your business. It is better to model a long-term profile that captures the likely growth of the business in terms of territory expansion, changes to order profiles and changes to order volumes (facilitate this through the S&OP Process if you have one).
Step 3: Build a cost profile
A cost profile needs to be established for each of the elements you need to optimise. Below is a quick guide to how you should structure each cost profile:
- Inbound and outbound transport costs – it’s usually best to produce a simple linear cost profile i.e. cost per unit (tonne, m3 etc.) per km;
- Warehouse costs – build a profile that captures the variable cost per unit plus the expected step changes in fixed costs per x number of units;
- Inventory costs – you will need a profile covering the expected inventory level, per the demand profile used, along with the working capital cost of that inventory.
Step 4: Select your modelling method
There are many different modelling approaches to warehouse location, but all of them tend to fall within three categories, known as Precise, Heuristic and Constrained Simulation. Each category has its own strengths and weaknesses dependent on the scale and scope of the analysis you wish to undertake and the complexity of the model (multiple locations is significantly more complex than determining a single location). Below we have provided an explanation of each of the three categories:
Precise, or sometimes referred to as ‘Exact’, location modelling is a no short-cuts approach to arrive at the mathematically optimum answer. Usually, where the problem being solved is relatively simple i.e. finding the optimal location for a single facility by balancing only inbound and outbound transport costs, then the Precise method can be deployed. This is because finding the optimal answer in this scenario does not require multiple iterations which are time consuming. In fact, the Precise method can normally be undertaken in spreadsheets using the map coordinates of the demand points in conjunction with a trigonometry formula to calculate distance.
Constrained simulation is where, instead of calculating the optimal, or near optimal answer, you model different scenarios based on pre-existing warehouse locations. This type of modelling will usually be undertaken to determine where a new product should be distributed from, or for periodic optimisation of logistics networks against an existing network of warehouses. Again, Constrained Simulation can usually be undertaken in spreadsheets, modelling and analysing each possible scenario individually.
Heuristic, in this context, means a best guess on the information available. This does not necessarily mean not analysing and modelling the data, but rather finding short-cuts to achieve a sufficient level of analysis that arrives at a near-optimal answer without the need for multiple iterations. This can be driven by selective evaluation based on practical experience, or alternatively there are many systems with heuristic based algorithms to derive a near-optimal answer quickly. Heuristic models are most suited to problems that involve locating multiple warehouse locations against multiple cost and service constraints.
If you need support in modelling the optimal geographic location for your warehouse, or warehouses, please contact our supply chain consultants today. The team at Paul Trudgian are experts in all aspect of supply chain modelling and implementation, and can assist you with warehouse location analysis, warehouse facility search and warehouse design.