The Production Planning Optimizer (or PP Optimizer, PPO) integrates purchasing, manufacturing, and distribution so that comprehensive tactical planning and sourcing decisions can be supported and implemented based on a single, global consistent model. The PP Optimizer offers cost-based planning which means that it searches through all feasible plans to find the most cost-effective solution (in terms of total costs). The total cost covers the following aspects:
- Production, procurement, storage, and transportation costs
- Costs for increasing the production capacity
- Penalties for violating (falling below) the safety stock level
- Late delivery penalties
With this, the PP Optimizer addresses the weaknesses of the classical MRP logic.
Whereas the classical MRP logic assumes infinite capacity when creating the procurement, manufacturing, and distribution orders, the PPO can consider limited resources and available capacity. Thus, the issues that arise from an infinite planning result are avoided when using the PPO. Furthermore, there is no adequate or optimal decision support to deal with alternative sources of supply in the classical MRP logic other than maintaining simple quotations or priorities. In contrast, the PPO can make a cost-based decision about the selection of various sources of supply and about the distribution of receipts among these alternative options. Finally, any decision in the MRP logic is guided by priorities, whereas the PPO can balance different cost factors to arrive at a cost-optimal result.
The PP Optimizer uses advanced optimization techniques, based on constraints and penalties, to plan the product flow along the supply chain. The results are optimal purchasing, production, and distribution decisions, reduced order fulfillment times and inventory levels, and improved customer service. Starting from a demand plan, the optimizer determines a permissible short- to medium-term plan for fulfilling the confirmed and estimated sales volumes. This plan covers both the quantities that must be transported between two locations (for example, from a producing location to a distribution center), and the quantities to be produced and procured. When making a recommendation, the PP Optimizer compares all logistical activities to the available capacity.
PPO planning decides on tactical planning and source of supply determination. The strengths of PPO planning are: The selection of the source of supply considering the costs, and the determination of the approximate production date considering both the procurement costs and the storage costs. The result of the PPO planning is the answer to What is produced, where and when? The answer to when cannot be more precise than a planning bucket, and it does take (to some extend) the sequence-dependent setup activity into account.
The PPO can be used to support decision-making in a multitude of scenarios. It can decide, which products to manufacture, store, transport, and substitute. Planning can be done either plant centric, or across plants, taking transportation lanes into account. It can optimize the manufacturing process, by choosing between different production versions and by deciding whether production is more suitable at a more expensive alternative comparing to proponing of production on the cheaper alternative and storing the product.
The PPO is therefore able to make the following decisions in particular:
- A product is subject to seasonal fluctuations. Is it cheaper to build up a warehouse stock in low season (storage costs) or to use an expensive source (in-house production or external procurement)? You can either procure a product externally or produce it in-house. In the current requirement situation, is it cheaper to procure a product externally (in this case, the resource is available for the production of another product) or to produce it in-house?
- Considers the impact of bottleneck resources in both directions of the material flow.
- A product is in stock in one plant, but it is required in another plant. Is it more favorable to transfer the product (as a result, a future requirement may have to be covered differently in this plant) or to produce it again in-house?
- A resource can produce a quantity A of products. Different costs are incurred when these products are produced. Another resource can produce a different quantity B of products. The quantities A and B have an intersection. On which resource are the products from the quantities A and B produced at the lowest cost?
The PPO determines the following:
- The production quantities, the procurement quantities, the stock transfer quantities for each product and period.
- The selection of the resources and the production versions for the production.
- The selection of the plants, the warehouses, the suppliers, and the transportation lanes.
The PPO creates and solves a mathematical model. Based on the cost model and demand signals, the PPO creates a mathematically optimized production plan. It can include make or buy decisions, in which sequence to produce, what to produce, where to procure from, whether to prebuild and store or to produce just-in-time, whether extra capacity is required, and which resource to produce on. This model takes the following factors into consideration:
- Valid transportation lanes and production data structures (PDS)
- Lead times
- Transportation costs
- Production capacity (initial capacity, minimum capacity utilization, and extended capacity utilization)
- Production costs (including charges for violating minimum capacity and using extended capacity)
- Storage costs
- Lot size (minimum, maximum, and rounding value)
- Scrap
- Alternative resources
- Penalty costs for not fulfilling demand (supply shortage)
- Delay penalty and maximum delay allowed
- Safety stock violation penalty costs
- Procurement costs
- Cost multipliers
- Location products
- Fixed PDS resource consumption
- Fixed PDS material consumption
The PPO works based on periods/buckets. You can define the granularity of the buckets. That means, you can decide whether to plan with subdaily, daily, weekly, or monthly periods. On the contrary, this means that the sequence of orders within a planning period is not defined. For this granularity of time-continuous planning, DS optimization can be used either as part of production planning optimization or as a separate subsequent step.
Limitations of the PPO approach are:
- The optimization run result does not include the pegging of orders back to the original individual requirements because requirements are bucketed.
- Since orders are not pegged back to the individual requirements, PP Optimizer does not support order-based planning. After the optimization run, it is not possible to determine information about links between specific planned orders and original sales orders.
- In the event of a capacity overload, the optimizer, depending on the system settings, either does not cover the requirements on time or increases the capacity based on a penalty cost calculation. Or does not cover the requirements at all.
The PPO converts a production planning problem into a linear or mixed-integer programming problem. The objective of this problem is the minimization of a cost function. Input data to this model are external demands classified into different categories, resource capacities, master data relating to activities like procurement, production, transport, substitution and storage, and a defined cost, bound, and penalty model. The result is orders of different types depending on the activities chosen in the PPO result as well as an extensive log allowing to analyze the result in detail.
PPO planning problems can consume a considerable amount of runtime and memory. Therefore, you must keep the runtime and memory requirements in mind when you design the model size and the model complexity. In the case of PPO, a distinction must always be made between purely continuous models and discrete (mixed-integer) models. If fixed lot sizes, minimum lot sizes, or piecewise linear cost functions are to be considered, you will require a discrete model. In this case, the system must perform a mixed integer optimization, which requires significantly more memory and runtime compared to a purely linear model.
The continuous models are primarily limited by the memory of the optimization server. Moreover, the required runtime of purely continuous models is linearly dependent on the model size and is generally noncritical as a result. On the other hand, the solution of discrete models using a mixed integral optimization can be considerably more complicated. It is generally not possible to guarantee the global optimum under realistic CPU time targets. Instead, feasible and high-quality solutions must be found with acceptable runtimes. The runtime and memory requirement of PPO optimization depends on the number of the variables. The number of variables is therefore restricted. You require a variable for each production quantity or procurement quantity that you want to determine, using PPO optimization. In other words, you require a variable for each location product for each planning period.
Advantages of PPO are:
- Optimizing procurement costs while considering alternative sources of supply.
- Considering capacities.
- Distributing the production over several periods (during seasonal fluctuations, for example) while taking storage costs into account.
Disadvantages of PPO are:
- Planning takes places in planning periods (buckets). The order sequence is not defined within a planning period.
- It can only provide a rough model for the material flow. It basically occurs only at the period boundaries. Unless the integrated DS Optimizer is involved.
Let's try to imagine how does production planning optimization work with a simple example.
Assume that there is a varying demand for a product in different periods (weeks 27-34). There exist several potential sources of supply for this product. In-house production is represented by two production versions. Production version 0001 is more expensive compared to production version 0002 and is limited at 250 units per week. Production version 0002 is the cheapest source of supply, but limited at 150 units per week. Furthermore, production version 0001 is not available in weeks 31 and 32. The third (and most expensive) source of supply is the transfer from another plant. The task of PPO is to assign supply quantities to different sources of supply in different time buckets (weeks). All demands must be met in time and quantity, while considering available sourcing and production capacities. In this example, the assumption is made that storage is not allowed.
So, the PPO can decide first to cover the maximum possible quantities from the cheapest source of supply that is production version 0002. Thus, a planned production receipt is created for each period in which the production version is available with quantity 150. The demand that is not met by these receipts is met by creating planned production receipts for the second cheapest alternative that is production version 0001. Thus, orders are created to meet the shortfall between the demand and the receipt from production version 0002. It is possible for all weeks, except for those in which production version 0002 is not available. For these periods, the most expensive source of supply, a stock transfer from another plant is selected to fulfill the demand.
This is a very simplified description of the reasoning of the algorithm. In real life, different products compete for the scarce resources and temporal decisions like proponing or postponing production of products complicate matters.
A similar logic must be applied across the multilevel structure of the bill of materials, starting with the input of demand at the highest level (finished goods).
The PPO does a bucket-based planning, because it considers the available capacity and demands on bucket level and therefore suggests a feasible plan on a bucket basis. In the result, the orders are scheduled at the start of the buckets. A bucket-based planning result can be sufficient for many planning scenarios. However, in certain scenarios, a subsequent detailed scheduling step must be carried out.
Detailed scheduling can be done manually, using heuristics or using the DS Optimizer. The DS Optimizer can even be integrated into a PPO planning run. In this case, as part of a PPO planning run, the DS Optimizer is executed to obtain a feasible production plan on a time continuous basis. In this case, a feasible plan is achieved by keeping critical operations within the bucket in which the corresponding order was originally planned by the PPO, whereas noncritical operations could be optionally allowed to start before the bucket in which the order was planned by the PPO. This will in turn solve the overload on resources (at the beginning of each period) introduced because of bucket-based planning of PPO.
Other than PPO, there are more heuristics available in PP/DS that support bucket-based planning. The table compares these heuristics with the PPO.
