Supply Planning is an important function in every industry which tries to achieve a balance between customer service, inventory targets and resource utilization. A supply planner’s primary role is to deliver a tactical plan which meets the demand on time, covers safety stock targets at the distribution centres in the supply chain network while also ensuring that resources in the factories are utilized efficiently. A supply planner basically works on integrating the manufacturing and sales side of business and aims to optimize both ends.

The Manufacturing or Production function always prefers to produce according to cost effective lot sizes, minimal changeovers, production cycles and sometimes other constraints like labour or semi-finished material. On the other hand, Sales or Customer Fulfilment functions have different priorities which mainly include meeting customer demand in full and on time, and meeting inventory targets without going below or above targets significantly. These two sides of business often contradict each other and achieving a plan which has been optimized from both angles is challenging.

Usually, a medium term tactical supply plan is generated periodically (weekly, fortnightly or monthly). This plan is normally capacity constrained and contains decisions regarding production, sourcing and distribution of products through the supply chain.

A lot of companies who purchased  SAP’s Supply Chain Management (SCM) application use the Advanced Planning and Optimisation (APO) Supply Network Planning (SNP) Optimizer as a tool to generate the capacity constrained mid-term plan. It is a powerful tool which delivers a plan that respects resource contraints and optimizes the supply chain within a given cost framework. If implemented correctly, it can result in big cost savings for any organization. In this blog, I will try to cover several aspects related to the SNP Optimizer.

SNP Optimizer - The Basics

The SNP Optimizer is a cost based planning engine/algorithm which converts the supply chain planning problem into a linear programming equation, or cost function with an objective of minimizing the cost of the overall solution. The SNP Optimizer iterates through several ‘feasible’ plans until it reaches an ‘optimal’ plan.

It can work in 2 different modes – Linear/Continuous (without lot size constraints) or Discrete (with lot size constraints). Linear mode is normally used to generate a rough-cut capacity plan, or as part of an S&OP process as the purpose is to just ascertain whether the forecasted demand volume can fit within the production capacity available in the factory. Normally, discrete mode is used to generate a mid-term tactical plan which is more feasible from practical point of view as it follows the minimum lot sizes and rounding values for production, transportation and a few other constraints.

There are 2 types of constraints which are dealt with differently by the SNP Optimizer;

• Hard Constraints: Constraints which cannot be violated. An example is the capacity of a production or transportation resource
• Soft Constraints: Constraints which can be violated but a penalty cost if incurred for each violation. Examples are safety stock violation, stock out or non-delivery of demand

The costs are applied to various soft constraints in the order of their importance. In a typical setup, the biggest focus is on meeting the demand on time. If demand cannot be met on time, the next focus is on meeting the demand with an acceptable delay if possible. Once demand has been met, the next priority is to meet the safety stock targets. Once safety stock targets are met, any additional supplies need to be stored and hence incur inventory carrying cost or storage cost in APO terminology. The 4 constraints just mentioned are the most commonly used constraints when setting up SNP Optimizer functionality.

Constraints for Planning

Commonly applied constraints within the SNP Optimizer like production capacity, stock out or non-delivery, storage, procurement and transportation costs are some of the important parameters for generating a constrained plan. However, these constraints may not be able to deliver a plan which is ‘feasible’ from business point of view especially from the manufacturing side. When the mid-term plan flows into the short-term horizon, it falls under the responsibility of the scheduler. It may not be possible for the scheduler to follow the plan delivered by the supply planner as some of the crucial manufacturing constraints may not have been considered. This could lead to several manual updates to the plan in the short term which means not realizing the benefits of investing in the tool and a disconnect between the mid-term and short-term plans. Ideally, the mid-term plan should flow seamlessly into the short term and the scheduler should work making the mid-term plan more granular without having to reshuffle the plan considerably.

Some of the common drivers for the Production function across industries are minimizing changeovers, cost effective minimum lot size for a product family or products of a similar nature/characteristics and line constraints. In some other cases, criteria such as production cycles, semi-finished product availability, shelf life or labour could be important.

The SNP Optimizer is capable of handling these constraints effectively through various techniques. These constraints cannot be implemented directly but need to be modelled. Basically, the constraint has to be mapped in terms of setting up the master data and costs in a particular way that the Optimizer understands the given set of data and dependencies to drive the desired behaviour. However, it does not imply that the SNP Optimizer is a replacement for a short-term scheduling solution. If the mid-term plan has proposed to produce products A,B,C within a week, the scheduler still needs to work out the correct sequence between the 3 products according to short term constraints like changeovers, material availability and shift patterns.

In the next section, I will elaborate on some of these constraints and how the true potential of the SNP Optimizer can be unleashed by implementing the appropriate constraints to generate plans which are more realistic.

Group Planning

Group planning is a methodology where products of similar characteristics are grouped together for production. This helps in generating a tactical plan which has fewer and smaller changeovers. A typical example of this is to produce SKUs of the same flavour in the same week rather than producing random SKUs every week. Such groups can also be called product families. The product groups/families themselves can be produced within a minimum lot size which is economical.

Figure 1 below contains a diagram depicting the concept of group planning. In this example, Products A,C,E and B,D,F belong to different groups. The Optimizer groups the production according to the groups and plans both groups in different weeks. The diagram on the left depicts a production pattern without using group planning which would normally follow the demand pattern and fluctuations.

Figure 1: Concept of Group Planning

There are many different ways this methodology can be used to build various constraints related to manufacturing. Some of the scenarios where group planning can be implemented are given below:

• Production with a minimum order quantity at group level. e.g. Produce at least 10000 cases of plain salted crisps for it to be economical from a changeover perspective
• Capacity constrained at group level such that group X should not consume more than 30% of overall line capacity
• Not producing 2 groups together in the same week. e.g. tomato and cheese flavour crisps must be produced in different weeks
• Production cycles where certain flavours must be produced every n^th week
• Capacity constraint at semi-finished level can be modelled with the help of group planning
• Certain groups must always be produced together in the same week
• When group X runs on Line A, then group Y must run on Line B

By applying such constraints using group planning, the Optimizer can help in achieving a high-level plan which is realistic and provides a better visibility of the supply chain situation in the medium to long term. The inventory projections are more accurate because the supply plan is better aligned with the manufacturing capacity and constraints. This plan integrates much more easily with the short-term resulting in the improvement of Plan vs Actual.

How is Group Planning Modelled?

Let’s look at some of the under the bonnet stuff. The basic group planning solution is modelled by creating a dummy product called group which is assigned as a component to individual SKUs which are required to be grouped. A dummy supply source called PDS is defined for the group which contains a dummy capacity consumption unless a capacity constraint at group level is desired. The group product’s minimum lot size is defined which is the minimum economical quantity of the group. A high storage cost is applied to the group product.

And here is how it works: When Optimizer needs to produce an individual SKU, it must always produce the group (as it’s an input component). When it generates a production proposal for group (this is a dummy production), it always takes the minimum lot size of the group into account. Once the minimum lot size of the group is produced, it cannot be stored as the storage cost used for a group product is very high. Therefore, the Optimizer pulls the production of other individual SKUs belonging to the same group forward into the same week as the group so that the entire quantity of the group is consumed. Figure 2 below depicts the behaviour in this situation.

Figure 2: Effect of Group Planning on Orders

The Trade-Off

There is always a trade-off between complexity of constraints and the solution quality. In general, the more complex the constraints, the more time the Optimizer needs to arrive at an ‘optimal’ solution. Since, only a limited amount of time can be spent for an Optimizer run during the overnight schedule, it is important that the problem sizing and constraints modelling is done carefully. It is often easy to get carried away while modelling such constraints as planners like to see every possible constraint being considered. Sometimes the constraints could even be conflicting in nature which can result in poor Optimizer results.

Conclusion

Once the right balance is struck in terms of constraints which should be modelled, the SNP Optimizer can provide real benefits in terms of providing efficiencies at factories and distribution centres and a good return on your investment in SAP solutions. The Group Planning solution which is discussed here as a case in point, is one of the ways in which constraints can be modelled. There are several other methods and solutions available which can be built to create a plan suiting different business scenarios and needs. I will cover some of these in future blogs.

At Olivehorse, we have several years of expertise in the SNP Optimizer and have implemented Optimizer solutions at clients across various industries. If you would like to learn more about this topic, have a general query, arrange a Proof of Concept or wish to do a cost vs benefit analysis for such implementation, then please feel free to contact us.

Chandra Shaktawat

Senior SCM Consultant, Olivehorse Consulting