Optimization-Based Planning (SAP Library - Supply Network Planning Run)

Optimization-Based Planning

Purpose

The SNP optimizer offers cost-based planning. This means that it searches through all feasible plans in an attempt to find the most cost-effective (in terms of total costs). Total costs refers to the following:

· Production, procurement, storage, and transportation costs

· Costs for increasing the production capacity, storage capacity, transportation capacity, and handling capacity

· Costs for violating (falling below) the safety stock level

· Costs for late delivery

· Stockout costs

You use the SNP cost profile to fine-tune the relative importance of different cost types.

In the optimizer view, a plan is feasible when it satisfies all the Supply Chain Model constraints that you set in the SNP optimizer profile. A feasible solution might involve due date or safety stock constraint violations. Due dates and safety stocks are Soft Constraints (constraints to which you assign violation costs). The optimizer only proposes a plan that will violate soft constraints if, according to the costs specified in the system, it is the most cost-effective plan.

The optimizer makes sourcing decisions within optimization-based planning. This means that it uses costs as a basis for deciding the following:

· Which products are to be produced, transported, procured, stored, and delivered and in which quantities (product mix)

· Which resources and which production process models (PPMs) or production data structures (PDSs) to use (technology mix)

· The dates and times for production, transportation, procurement, storage, and delivery

· The locations for production, procurement, storage, delivery, and the source and destination locations for transportation

Since you can enter PPMs or PDSs with fixed resource consumption in master data, you can also include setup operations in Supply Network Planning. Therefore, you can also use the SNP optimizer for lot size planning. The optimizer also supports cross-period lot size planning where orders are grouped into large lots due to high set up costs.

The optimizer uses the linear programming method to take account of all planning-problem-related factors simultaneously within one optimal solution. As more constraints are activated, the optimization problem becomes more complex, which usually increases the time required to solve the problem. As a rule, you should run optimization as a background job.

The optimizer makes a distinction between continuous linear optimization problems and discrete optimization problems.

Linear Optimization

You can choose one of the three following methods in the SNP optimizer profile to solve continuous linear optimization problems:

· Primal simplex method

· Dual simplex method

· Interior point method

All three methods arrive at an optimal solution. Runtime could be the main influencing factor when deciding which of these methods to use. However, there is no general rule for selecting the best method for a given problem (apart from to test each method individually). A good way of assessing the application is to do a benchmarking based on a test scenario. This is because the optimal choice of method depends more on the structure of the supply chain and less on the input data. Therefore, in a productive environment, daily benchmarking is not necessary.

Discrete Optimization

A problem is not continuous (and is therefore discrete) in Supply Network Planning, when the model contains:

· Discrete (integer-value) lot sizes for transportation or PPMs/PDSs

· Discrete means of transport

· Discrete increase of production capacity

· Minimum lot sizes for transportation or PPMs/PDSs

· Piecewise linear cost functions for transportation, production, or procurement

· Fixed PPM/PDS resource consumption

· Fixed PPM/PDS material consumption

· Cross-period lot size planning

If you want the optimizer to consider any of the above constraints, you must use one of the discrete optimization methods from the SNP optimizer profile.

The piecewise linear cost function that you can define in master data makes a distinction between the convex cost function (cost per unit increases for higher volumes; for modeling overtime or night shifts for instance) and the concave cost function (cost per unit decreases for higher volumes; for modeling freight rates for instance).

Convex cost functions do not complicate the planning problem and can be solved efficiently. However, they can also be modeled using alternative modes without using piecewise linear cost functions.

Example

· Mode 1 with $50 per unit and a limited capacity of eight models

· Mode 2 with $100 per unit and a limited capacity of six models

· Convex functions of labor cost per day, assuming eight normal working hours and a maximum of six hours of overtime paid at double time

In contrast, concave piecewise linear cost functions cannot be solved by an LP solver but only by using discretization methods (mixed integer linear programming). If piecewise linear functions are modeled but the optimizer is run without discretization or the discretization horizon is smaller than the planning horizon, the optimizer takes into account the linear cost function defined in addition to the piecewise linear cost function.

Note

The discrete optimization method cannot be used with strict prioritization (see below).

Using the discrete optimization method can significantly increase runtime requirements. Note that Supply Network Planning is a medium-term planning function and its focus should not be on solving integer problems (that is, using the discrete optimization method).

Prioritization

The optimizer can differentiate between the priority of sales orders and forecast demand. With strict prioritization, sales orders always have priority 1, the corrected demand forecast priority 5, and the demand forecast priority 6. Within every priority class, the system uses all available cost information to determine the final solution. When cost-based prioritization is used, the optimizer uses penalty cost information from the product master data (the SNP1 tab page) to determine the optimal solution.

Decomposition

You can use the decomposition methods, defined in the SNP optimizer profile, to reduce runtime and memory requirements for optimization. Decomposition may also represent the only way for the optimizer to find a feasible solution in the event of large discrete problems. For more information, see Decomposition.

Aggregated Planning - Vertical

To reduce the size of the model to be optimized, the optimizer can restrict planning to location product group level (assuming you have defined the demands at the lower level). Plans are distributed to lower level products based on demand for the lower level products. To plan at product group level, you must define hierarchies for products and locations in the hierarchy master. This data is used to generate the location product hierarchy. You also must define the PPMs or PDSs for the product groups and create the PPM or PDS hierarchy in the hierarchy master. In the SNP optimizer profile, if you set the Aggregated Planning - Vertical indicator, the products are automatically aggregated to the relevant groups for planning and after planning is complete, they are disaggregated again.

Aggregated Planning - Horizontal

This function allows you to plan a subset of your supply chain. You can limit the products or locations to be taken into account during the optimization run. For example, if optimization is only run to plant level but forecasts are at customer level, the optimizer can sum (aggregate) the demands to plant level and use this value during the optimization run. The transportation times, for example, from the plant to the distribution center and to the customer, as well as the duration of the PPMs or PDSs, are also taken into account.

Incremental Optimization

Incremental optimization is the name given to optimization-based planning that is run for only part of the model or on the basis of an already existing plan. The plan might be infeasible with this type of optimization since the optimizer cannot plan receipts for shortages that are the result of fixed orders from previous planning runs. It is also possible that the optimizer will ignore input products (and associated stock) that are defined in PPMs/PDSs or products that are available for procurement at a source location using a transportation lane (source location products).

To prevent this, you can specify in the SNP optimizer profile that the optimizer is to take into account the stocks of non-selected input products or source locations products. You can also specify that the dependent demand and distribution demand of fixed orders is to be treated as an independent requirement, meaning that the optimizer will permit shortages subject to the calculation of penalty costs for not delivering. You define penalty costs for not delivering for customer demand, the demand forecast, and corrected demand forecast in the product master data. You can also set that the optimizer is to consider the dependent demand and distribution demand of fixed orders, and the stocks of non-selected input products or source location products as a pseudo-hard constraint. This means that shortages are possible but are subject to the calculation of infinitely high penalty costs that are internally defined in the optimizer. Setting this ensures that the optimizer will only permit shortages if it cannot find any other feasible solution.

Process Flow

...

1. You run the optimizer.

2. You run deployment.

3. You run the TLB.

The Optimization Run

This graphic is explained in the accompanying text

Note: The optimizer plans all distribution demands for all locations in the distribution network before exploding the BOM and processing dependent demand at the production locations.

Factors Considered During the Run

· Valid transportation lanes

· Lead times

· Transportation capacity

· Transportation costs

· Handling capacity

· Handling costs

· Production capacity

· Production costs

· Storage capacity

· Storage costs

· Time stream (location master data)

· Lot size (minimum, maximum, and rounding value)

· Scrap

· Alternative resources

· Penalty costs for not fulfilling demand (supply shortage)

· Safety stock violation penalty costs

· Procurement costs

· Shelf life

· Cost multipliers

· Location Products

· Fixed PPM/PDS resource consumption

· Fixed PPM/PDS material consumption

Other Considerations

· The optimization run results do not include pegging orders back to the original individual requirements because requirements are bucketed.

· Since orders are not pegged back to the individual requirements, Supply Network Planning does not support order-based planning. After the optimization (or heuristic) run, it is not possible to determine information about links between specific planned orders and original sales orders (however, CTM can provide exactly this information by tracking orders).

· The optimizer considers the entire capacity and the entire alternative capacity that is globally available (at all locations).

· In the event of a capacity overload, the optimizer, depending on the system settings, either does not provide a solution or increases the capacity based on a penalty cost calculation.

· The optimizer considers all active types of capacity constraints, including transportation, production, handling, and storage constraints. The settings in the SNP optimizer profile govern whether or not a constraint is active.

· The optimizer takes into account the shelf life of products in a restricted fashion (for more information about this, see Stock Planning).