Building Blocks of Supply Planning Solutions: Digital Distribution Network Design

Michael Jermann Manager, PwC Switzerland 15/12/22

Are you facing the challenge of having to transform your supply planning processes? The pandemic has prompted many supply chain leaders to improve their planning processes and invest in new software solutions to enhance future resilience. This trend is further accelerated by the move to S/4HANA, the new version of SAP, which many companies are using as an opportunity to re-evaluate existing operating models.

Against this backdrop, this white paper series takes a look at supply planning solutions. Supply planning refers to the process of creating production, capacity utilisation, distribution and material requirement plans based on a forecast of demand. Most software offerings in this space bring a core platform of capabilities that is then configured to a product that fits the customer’s specific needs. Supply organisations therefore play a critical role in shaping these solutions by formulating clear functional requirements. This paper zooms in on the foundational building blocks, and instead of giving a broader view of capabilities discusses these topics in greater detail. The intention is to make these aspects of the supply planning tool more tangible and foster a deeper awareness of the critical design decisions on your transformation journey. This is important, because the design of these components of your planning tool has a fundamental impact on the overall solution. Moreover, as these building blocks touch on almost every function of the tool, it’s critical to get their design right the first time to avoid costly rework during the implementation. Your business’s specific functional requirements in this area might also have an impact on your selection of vendor.

Design your digital distribution network for resilience

In Part 1 of this series, we looked at the most common solver methods that are used in supply planning solutions. In Part 2 we turn to the digital distribution network. If we think of the solver as the engine technology of a railway, then the network represents the tracks, for the solver can only navigate within the given network structure. Unlike a railway network, which is mostly static and changes only slowly, the supply chain network in many industries is evolving. New warehouses are opened or closed, products are launched or discontinued, demands and order patterns change, or constraints require a redesign of distribution strategies. In this section we give our view on the optimal design of this network and how it can impact your supply planning system.

What is a digital distribution network?

A digital distribution network consists of supply and demand nodes as well as the transportation lanes that connect these locations. As such, connections are often not universally applicable to all types of products in your network: they are often product-specific. Therefore, when we talk about the distribution network, we mean all the possible combinations of a product, the shipping location and the destination location.

A standard goods flow is illustrated in Figure 2. It consists of the movement of raw materials from the supplier to the manufacturing site where the finished good is produced. Then the finished product is sent to a distribution centre (DC) and from there to the final sales node of the customer. In very dense networks where supply and demand nodes are very close together, it can frequently happen that these nodes are much more interconnected. For example, it may be that high volume products are shipped to customers directly from manufacturing sites or that there are flows between DCs to rebalance stocks from changing demand patterns. These networks can look very different across companies, and even within companies, as different countries can have substantially different value chains and route-to-market approaches. A good solution provides great flexibility for configuring this fundamental building block of your supply planning.

How to define the locations in a digital supply network

Defining the internal part of the network is normally quite straightforward, and would contain the real warehouse and factory locations. Defining the scope of the external network, on the other hand, is a bit more complicated. This part of the network includes materials suppliers and the final customer (such as a retailer or wholesaler). The reason for the complexity is that suppliers and final customers normally have substantially more nodes than internal network nodes. As pointed out in Part 1 on solvers, high-complexity supply chains can negatively impact solver runtimes. Moreover, adding a substantial number of nodes could increase the number of parameters planners need to maintain. It is therefore important to map the network only to the relevant granularity.

Suppliers need to be included in the network to enable companies to be able to do effective material resource planning and have the option of constraining finished goods products by material availability or supplier capacity. As a guiding principle, we would advise you to think about the location level at which you would set material constraints to determine the constraint granularity. In most cases, aggregating these nodes by supplier without mapping individual supplier manufacturing sites provides enough flexibility and avoids the parameter maintenance necessary when supplier information is received manually ‒ which owing to a lack of system integration is most often the case. During the design phase it’s critical to work closely with the supplier to establish collaboration frameworks ensuring that the supply planning solution integrates well with the envisioned process.

The final customer is normally a ship-to with a concrete address. As there can be potentially thousands of ship-to customers depending on the industry, they are normally clustered together. It’s crucial to understand this clustering properly, as it should reflect the granularity you need to formulate effective supply chain strategies. Supply planning should receive the forecast from demand planning at this level of granularity. This means it’s especially important to reduce the number of these nodes, not only because of potential impacts on the performance of the final solution, but also because forecast accuracy generally decreases if the demand is split across more locations. We recommend assessing this grouping across three dimensions:

Lead time and transport constraints: This is the most straightforward dimension, as it is mostly defined by the actual distance and time needed for the journey to the final customer. Transport teams normally have a good framework in terms of what lanes they frequently negotiate for with carriers, and have commonly used lead time assumptions when planning these flows. The degree of detail can be a city, a region or even a country in the event that there are only negligible differences in terms of lead time or transport costs.
Contractual obligations or other agreements: It can be that certain customers have different contractual conditions that strongly influence planning approaches. One example could be consignment warehouses where stock that is exclusive to the customer is kept. It could also be that some customers organise their transport themselves, while for others the internal transport team needs to organise transport. In such cases it might make sense to maintain separate clusters to allow these factors to be considered in planning. This could be in addition to the geographic component (for example Paris-Pickup and Paris-Delivery).
Demand priorities: If you want to dictate to your supply chain solver demand priorities or if your strategies differ in terms of the exact key account or commercial segment, it can be beneficial to mirror this level of granularity in your network.

Should we include external customer locations in our network?

Having defined the granularity of the final customer, you need to consider whether these clusters are mapped as locations in the network or whether they are mapped to the DC where the goods would be delivered from. These two options are common practice, and what option you pick depends on how flexible your sourcing decisions within your network are. If your clustering of demands results in a clear mapping where one cluster is mapped to one location, it might make sense to map these demands directly to the relevant DC. If, however, demand can be sourced from two or more DCs, having final demand nodes mapped as locations would allow the solver to dynamically plan the sourcing of stock based on the current operational situation. If the demand is mapped statically to a location while there are multiple options to source from, an imbalance in the demand mapping with the real stock situation might result in suboptimal stock transfers, as illustrated in Figure 3. In this example, as more stock is mapped to DC1 than available, this results in an unproductive fictional move from DC2 to DC1. This demand would better be sourced directly from DC2. This negative impact of the static mapping can be partially mitigated by having a regular extensive solve run where the customer locations are exposed. The outcome of this run is then used to update the static mapping table with the cost-optimal allocation, which will remain valid until a substantial demand shift again drives unproductive movements.

The decision on whether to map customer locations also affects your flexibility to dynamically prioritise customer demand. If customer clusters are mapped as a location, we could reflect different costs for a certain cluster in an optimiser. This would allow the solver to dynamically plan and prioritise this demand in the event of constraints. A similar outcome could be achieved in heuristics by changing the order of the solve and putting important clusters earlier in the order and less important clusters later. Having customers mapped as a location in your network can also enable advanced demand sensing capabilities by imbedding external customer data along with other internal data to automatically fine-tune your forecast using machine learning. Depending on the structure of the solution, it might allow additional customer specific parametrisation. We can see from this that network design decisions can have a fundamental impact on what capabilities can be enabled and how having a more customer-centric supply chain can be dependent on such design details.

How do I decide on the valid transportation lanes between locations?

We have now reviewed what a digital distribution network is and what locations it should include, specifically in terms of how to structure our final customer demand nodes. What we now need to define is what connections within a network are valid. Normally not every location can ship to every other location, and some connections, while theoretically feasible, will be too costly to be practical. One way to define this is to manually create these connections in the system by defining the product, shipping and delivery location. However, for large, complex networks this is not feasible and prone to errors because of manual maintenance. Based on our experience in the consumer goods industry, a network can number tens if not hundreds of thousands of connections that can change on a weekly basis. This begs the question of how to structure such a network and maintain it regularly. Failing to do so could mean that our solver “train” cannot operate efficiently, as the necessary “rail” network connections are missing.

For high-complexity networks, we advocate rule-based automatic network creation. The goal of this rule-based approach is to enable complex logistics strategies to be mirrored in the tool, thereby ensuring strategic alignment across planning and execution. In this approach the planner formulates rules based on specific products or demands as well as more general attributes. Each rule might trigger thousands of individual connections, thereby making the network maintainable. As rules are regularly executed, they apply to changes in the product portfolio, changes in the demand patterns as well as other dynamic factors to create a self-updating network. By leveraging existing data from the system of record and using advanced analytics, more sophisticated parameters that can be used in the network generation rules can be leveraged.

How do I validate whether my network is operating efficiently?

Once we have an automatically generated and self-updating distribution network in place, we need to have an effective way of evaluating its performance. A common approach is to measure the percentage of demand that passes through your most cost-efficient network option (cost-optimal network allocation efficiency). This can be measured for multiple tiers of cost optimality, thereby giving you a better understanding of your overall network performance (e.g., by also tracking how much passes through the second most cost-optimal route, the third-most, etc.). For this process to work, it is crucial to cooperate with your finance teams to get accurate cost information and to ensure this cost is accurately captured by the system.

For more granular reviews of network distribution efficiency, it can be helpful to categorise transportation lanes to track a higher level of granularity of unproductive movements. To give one example of this, an expensive transportation lane that connects two manufacturing sites together that can both produce the same product could be classified as a dual-sourcing lane, meaning that in an ideal situation movement between these two locations should be avoidable, as products can be produced locally instead. Tracking these distributions on top of the network productivity KPIs can give management a better view of the optimisation potential of the plan.

Where are you likely to need support?

In this, Part 2 of our white paper series on the building blocks of supply planning solutions, we’ve looked at how you can design your digital distribution network for resilience. A digital distribution network consists of supply and demand nodes as well as the transportation lanes that connect these locations. Getting this part of the equation right involves clustering final customers together and understanding this clustering properly across the dimensions of lead time and transport constraints, contractual obligations and other agreements, and demand priorities. It involves asking whether to include external customer locations in the network, deciding on the valid transportation lanes between locations, and working out how to validate whether the network is operating efficiently. External support at this point can help bring the relevant functional and technical know-how to the different phases of the project. It could include structured guidance in areas such as:

Blueprinting: Support in defining the required scale and granularity of your network (such as customer clusters and suppliers)
Blueprinting: Bringing in performance management expertise to define the right KPIs to measure the quality of your network
Implementation: Translate the sourcing and distribution practices into a detailed functional design that is in line with your business’ complexity (such as an automatic rules-based network creation for complex networks)

Part 1

Supply Chain Solver Methods.
In this, Part 1 of our white paper series on the building blocks of supply planning solutions, we examine the role of the solvers. Solvers, the underlying method that translates business objectives into calculated outcomes, fall into two categories: mathematical optimisation and heuristics.

Read article

#social#