Safety Stock Planning (SAP Library - Supply Network Planning)

Safety Stock Planning

Purpose

Safety stock planning in Supply Network Planning (SNP) allows you to attain a specific service level by creating safety stock for all intermediate and finished products at various locations throughout the entire supply chain.

Safety stock is used to safeguard stock in the supply chain against uncertain influencing factors. Supply chain planning is susceptible to a number of influencing factors that cannot be predicted with certainty in advance. For example, when forecasting customer demand, the quantity required is usually uncertain. In addition to this, disruptions in production or transportation time fluctuations lead to discrepancies in planned replenishment lead times. To safeguard yourself against such uncertainties when planning, you can take one or more of the following measures:

· Overestimate customer demand

· Underestimate production output quantity

· Overestimate procurement lead times

· Use safety stock

While overestimating customer demand is a Demand Planning tool for safeguarding against forecasting errors, you can model an underestimation of production output quantity and an overestimation of procurement lead times using production process models (PPMs) and transportation lanes.

You have to address the following questions as part of safety stock planning:

...

1. At which locations within the supply chain do you want to have safety stock?

2. How much safety stock do you want to hold at a particular location?

Example

This graphic is explained in the accompanying text

With the aid of this simple supply chain, it immediately becomes clear that the question about where to hold safety stock is a highly complex problem due to the variety of possible combinations (calculated as 2 to the power n – in this example, there are already 64 possibilities). Due to this wide range of possibilities, it is advisable to make use of the planner’s experience and allow the planner to simulate selected planning scenarios.

Safety Stock Methods

Since safety stock is usually necessary for products at different locations, you can select a safety stock method in the product master for every location product. The different methods are split into standard and extended methods here.

Standard Methods

The standard methods differ from one another in their observance of time. For these methods, the planner enters the safety stock information directly into the system:

	Not time-based (static)	Time-based (dynamic)
Safety Stock	SB	MB
Safety days' supply	SZ	MZ
Max {Safety stock, safety days’ supply}	SM	MM

For example, you might want to define safety stock directly as a safety days’ supply or as the maximum of the safety stock and safety days’ supply. With the maximum option, safety stock can be adjusted dynamically to meet the demand flow, and not fall below the defined safety stock. You specify safety stock that is not time-based on the Lot Size tab page in the product master, but define safety stock that is time-based in interactive Supply Network Planning.

Note that for safety stock method MM, the SNP optimizer only considers independent requirements as well as dependent and distributed demands caused by fixed orders, since these demands and the demand locations are already known before the optimization run.

Extended Methods

While the standard methods are based exclusively on the planner’s experience, the proposed safety stock levels of the extended methods are determined by the system based on scientific safety stock planning algorithms. The starting point is a service level that you want to attain through observance of the calculated safety stock. You can define this service level in the Lot Size tab page in the product master. It can be interpreted as follows (based on the business process):

· Shortfall-event-oriented (alpha service level): The service level in percentage means that no shortfall is expected in x percent of the buckets within the planning period.

· Shortfall-quantity-oriented (beta service level): The service level in percentage means that x percent of the expected total customer demand can be fulfilled within the planning period.

Example

Bucket	1	2	3	4	5	6	7	8	9	10
Expected demand	100	100	100	100	100	100	100	100	100	100
Shortfall quantity	0	0	0	0	0	10	0	0	0	10
Shortfall event	-	-	-	-	-	x	-	-	-	x

Total of shortfall quantities: 20 -> beta service level: 1 – (20 / 1000) = 98%

Total of shortfall events: 2 -> alpha service level: 1 – (2 / 10) = 80%

To decide which service level to use, answer this question: Are the costs for subsequently delivering a shortfall quantity dependent on the shortfall quantity or not? If these costs are not dependent on the shortfall quantity (fixed costs), we recommend that you use an alpha service level; if they are dependent on the shortfall quantity (variable costs), a beta service level would be more appropriate.

The stockholding method used by SNP when planning demands has a major influence on the algorithm for calculating safety stock. The following two different stockholding methods exist:

· Reorder cycle method: With this method, the system makes a purchase order decision on a time basis, which means that procurement can only be triggered for all t buckets.

· Reorder point method: With this method, the system makes a purchase order decision on the basis of stock, which means that procurement can be triggered if stocks fall below a certain level s (the reorder point).

Stockholding Method

This graphic is explained in the accompanying text

The two service levels result in these four model-supported safety stock methods:

	Reorder cycle method	Reorder point method
Alpha service level	AT	AS
Beta service level	BT	BS

The prerequisite for using this method is that shortfall quantities are delivered subsequently (“back order case” as opposed to “lost sales case”). If this prerequisite is met, the system can calculate the safety stock on any step of the supply chain and for each bucket in the planning period.

Forecast Error Determination

When calculating the safety stock, the system can take into account a forecast error in both demand and procurement. The following key figures form the starting point for the forecast error calculations:

Demand	Procurement
Planned demand quantity key figure	Planned replenishment lead time key figure
Realized demand quantity key figure	Realized replenishment lead time key figure

The system calculates the forecast error by determining the planned actual deviation of the relevant key figures. The standard deviation of the planned actual deviations is interpreted as the forecast error. A forecast error is thus determined from the historical data and the future forecast is based on this forecast error. To more accurately support the future forecast, it is a good idea to interpret this forecast error as a relative forecast error, so that instead of keeping the forecast error itself, you keep the relationship between forecast error and forecast (variation coefficient). This is clarified in the following example:

Example

Mean value of the planned demand quantity: 100

Standard deviation of the planned actual deviations: 10

Bucket	1	2	3	4	5
Demand Forecast	100	1000	1000	100	100
Forecast error if the standard deviation is constant	10	10	10	10	10
Forecast error if the variation coefficient is constant	10	100	100	10	10

If the forecast error is not dependent on the forecast, an incremental forecast unexpectedly causes the safety stock to fall because the forecast error decreases in relation to the forecast. For this reason, it is more advisable to use relative forecast errors in a dynamic environment than constant forecast errors.

If there is a forecast error in procurement (a replenishment lead time forecast error), the demand forecast error is adjusted based on the assumption that the two forecast errors are independent of each other.

You can also enter the demand forecast error and replenishment lead time forecast error directly in the location product master. We recommend that you do this in the following circumstances:

· If there is no historical data (because the product is new for instance)

· If the amount of historical data is so small that it is impossible to calculate a statistically relevant forecast error

· If the forecast error can be considered constant

In the safety stock planning profile, you can specify if the system is to calculate the forecast error during extended safety stock planning from the historical data or from the location product master data.

Modifying the Parameters

Multilevel safety stock planning is a very complex issue for any supply chain structure. It is therefore a good idea to implement high-performance heuristics. Algorithms are thus the focus of attention for single-level, non-time-based safety stock planning that is then incorporated within a multilevel, time-based supply chain planning by adjusting the input parameters.

This makes it necessary for the system to adjust the forecast and forecast error for demand and procurement.

For demand, the system first determines all the location products supplied by the safety stock location product. It then projects all the forecasts and forecast errors onto the safety stock location product (considering all the quantity and time relationships) to calculate the safety stock.

For procurement, the system first determines all the location products that are to supply the safety stock location product. It will do this until it finds a safety stock location product, or until it finds external supply for the supply chain. It then determines the critical supply path by calculating the maximum replenishment lead time. All forecasts and forecast errors along this critical path are then projected onto the safety stock location product for safety stock planning.