Understanding Inventory Optimization

Now that we have handled fundamental concepts about inventory, we will dedicate this section to understanding what optimization is.

Important initial definitions:

• Optimization is defined as the process of making something as good as possible reaching a standard (global) optimum for the entire model or a local optimum within a neighboring set of candidate solutions.
• An optimization problem consists of maximizing or minimizing a function by systematically choosing input values from within an allowed set, and computing the value of the function delivering as optimal results to make decisions.

Inventory optimization calculations in a supply chain network are performed to calculate the recommended level of safety stock, level of cycle stock, and the target inventory position finding optimal values for the inputs in the function.

Inventory planning: the objective of inventory planning is to project inventory requirements across the entire supply chain network and planning horizon to meet customer service-level targets. Therefore, inventory planning is the science of making rational and profitable inventory decisions for every inventory stocking in the end-to-end supply chain to accomplish the following:

• Meet desired customer service-level targets.
• Minimize inventory working capital investment.

Inventory Optimization is the application of algorithms and scenario planning that considers demand and supply signals handling large-scale complex supply chains maximizing or minimizing the impact of the inputs over the objective function. To gain a deeper understanding of inventory optimization and the relevant calculations, we need to understand some basic concepts behind supply chain management and inventory planning, as follows:

• Service level
• Normal distribution and statistical probability
• Forecast error calculation
• Demand lag
• Supply variation calculation
• Sequential and cross-over order delivery types

• Service level: A service level represents the percentage of customer orders fulfilled. For inventory service optimization, two analytics calculations are widely in use for service level analytics: Fill rate and Available in full. The fill rate represents the percentage of customer demand quantity met on time, while available in full, also known as on time in full (OTIF), measures the percentage of the orders fully satisfied on time.
• Normal distribution and statistical probability: When you have a large amount of data, its distribution follows the normal curve, according to the central limit theorem used widely in statistics. A detailed understanding of this statistical concept and the central limit theorem are beyond of the scope of this training; however, basic knowledge of these concepts is required to understand inventory optimization calculations.
• Forecast error calculation: For forecast error or demand deviation calculations, forecast values are compared with the actual sales data for every period. Calculated mean, standard deviation, and coefficient of variation of the forecast error are used for safety stock and inventory calculations.
• Demand lag: To control the process of last-minute forecast adjustments and to use a more realistic forecast error value for inventory calculation, demand lag is used. A lag represents the gap between the current period and the forecasted period.
• Supply variation calculation: Supply variation is used for the supply processes involving purchasing, production, and transportation. The variation is calculated for the operation’s lead time.
• Sequential and cross-over order delivery types: Delivery types of the orders can be sequential or cross-over, which is true for production, purchasing, and transport orders. In sequential delivery types, orders are processed on a first-come-first-serve basis, and a delay in one order delays subsequent orders based on a sequential nature. On the other hand, in a cross-over delivery system, a delay in an order may not impact other orders because subsequent orders can be delivered first, before the delayed order.

To put it briefly, inventory optimization is the science of calculating inventory targets to meet desired service goals, doing so at the lowest inventory cost possible, and across the entire supply chain.

What decisions are we making in inventory planning and optimization? How much inventory (Where and from which SKUs?) do I need to minimize cost and maximize end-customer inventory availability (service level)?

The general issue is that setting inventory targets is difficult because things do not always go quite as planned.

Instead of clear requirements for inventory planning and a straightforward, linear planning process, today’s planning has rather more complicated and sophisticated boundary conditions and requirements.

Today’s market is described more accurately by quite complex supply chains, such as those indicated in the figure Supply Chain Complexity. As we see in the figure, inventory at each node can, and should, impact inventory decisions at other nodes. Consequently, it is not evident how much inventory is required at specific points in a supply chain.

Complex Supply Chain Example

The figure represents all points of a supply chain which should be interconnected within a retailer's network in the United States. This network shows an example of supply chain complexity.

Unpredictable Demand

The demand side of inventory planning is not always completely predictable.

Major supply chain demand factors include:

• Simultaneous internal and external demand
• Forecast error
• Seasonal, time-varying demand
• Multiple service levels and inventory thresholds
• Over-forecasting and under-forecasting
• Outliers

Inventory targets must consider variability and uncertainties on the demand side.

After all, suppliers are rarely perfect. Major supply factors influencing the supply side include:

• Batch size requirements
• Late shipments
• Short-shipped supplies
• Frozen forecast windows
• Time-varying bills of material
• Multiple supply sources
• Seasonal supply sources

Consequently, inventory targets must also consider variability and uncertainties on the supply side.

Now, take all of these uncertainties and multiply to get the full picture for the entire supply chain and its complexity.

It becomes evident that variability and uncertainty have interdependencies within the network.

Inventory decisions at every point in the enterprise-wide supply chain are linked. However, traditional systems ignore this complexity.

What makes the inventory challenge even more difficult to achieve?

Inventory planning process practices remain siloed, responding to partial needs of the end-to-end supply chain network resulting in:

• Over-buffers of inventory.
• Bull-whip effect.
• Limitation to determine any postponement strategy.
• Not handling BOM, lot sizes and other supply chain complexities.