Introduction

In the typical Sales & Operations planning process (S&OP), the Sales, Finance and Operations Departments first agree on a common Forecast for the different products (or products families) and Customers (or regions). This forecast is known as (unconstrained) Consensus Forecast). The subsequent step consists in executing the Supply Planning. In this step, the supply planner creates an high level supply plan and determines – if possible – the feasibility of such a plan.

The tricky thing here is to fulfil business goals which partly overrule each other: match customer demand on time and in full, cover inventory targets and utilize the available capacities efficiently at the same time.

To achieve a constrained and feasible demand and supply plan SAP IBP offers different algorithms:

  1. Performing the time series-based unconstrained heuristic, including manual adjustments in case of capacity or supply shortages. The manual adjusted plan can then be propagated downstream towards the customer
  2. Running the time series-based optimizer includes an objective function (maximizing profitability or revenue plus different soft or hard constraints along the supply chain). The result is a feasible supply plan.

In this article we focus therefore on the three different SAP IBP time-series based solvers within the Sales and Operations Planning (S&OP) module of IBP:

  • Heuristic
  • Supply Propagation
  • Optimizer 

We will go through a system demo to cover the following scenarios:

  • Enough capacity available but production constraints due to machine maintenance
  • Mitigation of capacity bottlenecks by
    • Delivery maximization 
    • Applying fair-share
    • Profit maximization

The idea here is, to provide a basis for a judgement or evaluation on how the solvers will meet the planner’s objectives.

The evaluation of each solver will therefore be based on:

  1. Planning quality
    • How feasible is the initial resulting plan? 
    • Are all the supply chain constraints correctly considered? 
    • Do the results reflect the business objectives and priorities?
  2. Manual effort
    • What amount of effort and time is required by the user to adjust the plan in order to make it feasible?
  3. Complexity
    • Are the results comprehensive and transparent to the end users?
  4. System Performance
    • Can the system handle the supply chain model complexity in a timely way?

Demo Scenario Description

The scenario we have tested in our SAP IBP Test and Demo System comprises:

  • three articles: IBP-300, IBP310 and IBP 320
    • IBP-300 short margin (1%) /High volume
    • IBP-310 medium margin (10%) /Medium volume
    • IBP-320 high margin (50%) /Medium volume
  • two Locations: 1720 (DC = Distribution Centre) and 1710 (Production) and one resource: Assembly @ 1710

Business Objectives (example) (sorted based on priorities): 

  1. Respect hard constraints (such as the capacity of the assembly @1710)
  2. Maximize operational profit (i.e. high margin articles have higher priority)
  3. Fulfilcustomer demands
  4. Respect safety stocks
  5. Stable production (stable resource utilization). (note: the production planner avoids peaks where possible and prefers a stable plan)
  6. Minimize inventories

Note: the objectives may vary from customer to customer. In this demo we will assume the six objectives listed above.

The supply chain network of our demo is displayed below:

Figure 1 Supply Chain Network

Scenario #1: Running Unconstrained Heuristic

Initial Demand Situation

The demand in our simple scenario is constant except a peak in March:

Figure 2 unconstrained (consensus) Forecast

Heuristic results (1)

After running the time series-based heuristic the customer receipts match the demand 1:1.

Figure 3 heuristic results (note: lot size is 1 piece)

There is enough capacity available:

Figure 4 capacity utilization after running the heuristic

Production Constraints

Due to an exceptional production campaign or machine maintenance the article IBP 320 (the high margin product) cannot be planned between calendar weeks 6 and 9 in the month of February; there is however capacity available in January.

Figure 5 prevent production for article IBP-320

To maintain a downtime in IBP you can either adjust the resource capacities or enter as above a value in the key figure “Production Receipts Ads. Interactive” (key figure ADJUSTEDPRODUCTION), which belongs to the category “Adjustment key figures”: any value entered here is considered as fixed by the heuristic. 

In screenshot below you can see how the supply is planned in the distribution centre and in the production plant. Due to the production stop, shortages are generated for IBP-320 requiring user attention. The initial plan is not feasible:

Figure 6 inventory overview 

The parameter CARRY_OVER_NEGATIVE_PROJECTEDINVENTORY is not set, therefore the shortages in the Projected Stock key figure are not considered in future periods by the heuristic.

Supply Propagation

With Supply Propagation, the supply shortage resulting from the production constraint is propagated downstream (customer receipts missing):

Figure 7 unfullfilled Customer demand

The negative projected stock on the production site is levelled to “0”, as the production issue is “resolved” by propagating it to the DC.

Figure 8 Inventory overview for article IBP-320 at location 1710

The supply plan is feasible, but the result might not satisfy the Supply Planner. The reason is that there is enough capacity in January to pre-produce product IBP-320 in order to mitigate the production issue and fulfil the customer demand.

Scenario #2: Time Series-based (TS) Optimization

In our Optimizer run, we use standard time-independent costs.

The demand is now on time

Figure 9 demand fullfilment after running the TS based Optimizer

Since the TS Optimizer allows early production, the capacities in January are used in order to fulfil the later demand of product IBP-320.

Figure 10 production plan for article IBP-320

By applying the TS Optimizer, we get a more stable production plan. As we can see below, the resource utilization after running the heuristic ranges from 46% to 99%. In case of TS Optimizer, it ranges from 53% to 88%.

Heuristic / Sp[CS1] . Propagation:

Figure 11 resource utilization after running the heuristic

TS Optimizer[CS2] :

Figure 12 resource utilization after running TS optimizer

The production plan stability could be further enhanced by setting, for example, penalty costs for violating the “minimum capacity usage “.

Scenario #3: Planning with Capacity Bottlenecks

Now, we reduce the capacity in Assembly to simulate a bottleneck:

Figure 13 Capacity reduction in March and April

After running the heuristic, we see capacity overloads in calendar weeks 10, 11 and 12:

Figure 14 resource utilization after running the heuristic in a capacity bottleneck scenario

Let us assume Business desires to freeze the production plan for January and February, which means that early production (i.e. before CW10) is not possible (the production constraint for article IBP-320 from earlier scenario has been removed).

We will test now the TS based Optimizer with 3 different approaches:

a) maximize deliveries
b) fair-share
c) maximize profit[CS3] [PA4] 

Figure 15 the 3 different TS Optimizer Operators were configured based on distinct business rules

Maximizing Deliveries

Due to the capacity constraints, the Optimizer generates late deliveries[CS5] [PA6] .

Figure 16 resource utilization after running the TS Optimizer in a capacity bottleneck scenario

The distribution of lateness and shortage considers neither fair-sharing (IBP-300 is more impacted than the others) nor profit maximization(IBP-320 has more late demand than IBP-310)[CS7] [PA8] . 

Figure 17 customer demand fullfilment after running the TS Optimizer in a capacity bottleneck scenario

Fair-Share

In the screenshot below, we can see the settings for activating fair-share for the Optimizer. The number of segments (maximum 10) defines how the costs increase in relation with the severity of the violation. For more information please consult the IBP model configuration guide.

Figure 18 Configuration of fair-share logic in IBP

Since this is a global setting, you may want to deactivate the fair-share policy for some Location or Customer Products. To do so assign value ‘1’ to the attribute “Non-Delivery Fair-Share Cost Policy”.

In our scenario, the lateness and shortage quantities are distributed in a more balanced way. All products are similarly and proportionally impacted irrespective of their margins:

Figure 19 customer demand fulfilment after running the TS Optimizer with fair-share

Maximizing Profit

We adjust now the non-delivery costs accordingly (higher for high margin articles and lower for low margin ones) to invoke a maximization of the profit[CS9] [PA10] .

The system therefore assigns the shortened capacity to the high margin articles.

This was also our initial objective,

Figure 20 customer demand fulfilment after running the TS Optimizer with different non delivery cost rates

Preliminary Conclusions

Time series-based Heuristic

  • simple and comprehensive
  • Not adequate in situations of insufficient capacity (bottleneck)
  • Not adequate for other constraints (such as limited supply of components or fixed receipts) because the plan becomes unfeasible
  • Requires significant user efforts / manual intervention in order to fix the plan.

Supply Propagation

  • Simple although not as much as the heuristics
  • Supply constraints are handled correctly
  • Not adequate in situations of insufficient capacity (bottleneck)

Optimizer

  • Very powerfull. It addresses business priorities (such as fairshare and profit maximization)
  • Even if it is more complex the explanation log pinpoints the constraints resulting in lateness, shortage or safety stock violation
  • If discretization is used for several periods (with lot sizes) the performance can be impacted

Author: Pedro Averwald