By Walt ~ March 31st, 2009. Filed under: Best Practices, Supply Chain / Inventory Systems / Logistics, Systems Engr..
Retail stockage policy, for merchandising or military applications, all have in common three decision rules which can be simulated (with or without constraints such as budget, quantity discounts, or warehouse space) using empirical or stochastic demand distributions to determine optimum customer satisfaction levels. Simply said, these rules answer the three questions of what to stock on the shelf, how much to stock, and when to reorder. The decision of what to stock, often referred to as stockage breadth, determines the number of Stock Keeping Units (SKUs) to put on the shelf. (SKU is a term used in retail merchandising such as Home Depot or Sears. The US Army uses the term Authorized Stockage List (ASL).)
Stockage breadth determinations are often made by a simple rule of thumb using the frequency of demands for an item as a model. For example, to add an item to the ASL at the US Army retail level, 9 recurring demands within 360 days are required. Given an estimated order-ship-time (OST) of 20 days this policy guarantees that 180 days of stock outs before an item can be added to the ASL. The loss of critical combat equipment for this length of time significantly reduces combat readiness. In a merchandising operation, 180 days of waiting for an item can result in significant loss of profit and loss of customers. Moreover, the demand quantity is not considered in current models using frequency of demand stockage rules. A better approach would be to use the demand frequency and the demand quantity in an algorithm that considers the cost associated with not having an item in inventory. This algorithm is explained below.
Simulation models with algorithms that determine stockage breadth based on a decision rule of not stocking (lost profit) versus the cost of stocking (holding an item in inventory in anticipation of a customer demand) will optimize overall profit (military readiness), customer satisfaction and cost savings. Quite simply, the rule is: If the cost of not stocking is greater than the cost of stocking, add the item to the inventory, otherwise do not stock the item. In this manner, the benefits derived from stocking an item will be in proportion to the related inventory carrying costs. Aside from using empirical customer demand for determining what to stock, merchandisers, for example, can benefit from simulation based on changing fashion trends, changing technology, advertising campaigns, and marketing forecasts.
Stockage depth determination of how much to stock (order quantity, Q) and when to re-order (reorder point, RP) is also subject to simulation and sensitivity analysis which can identify significant cost saving in inventory investment.
Q determinations can be simulated using Economic Order Quantity (EOQ) using various stochastic demand distributions such as Poisson and constraints (e.g. budgetary or spatial), in order to determine optimum customer satisfaction.
The RP is the sum of the Order-Ship-Time (OST) and the Safety Level (SL). OST is the forecasted demand quantity for an item between the interval of re-ordering and receipt of an item in inventory, often referred to customer demand during lead time. The SL is an added quantity to allow for fluctuations in demand (standard deviation) during lead time and fluctuations in the delivery time (also a standard deviation): a condition of dual uncertainty.
The components Q, OST and RP are called the Requisitioning Objective (RO) in the Army model. In an asset triggered re-order system or point-of-sale system such as at Home Depot, an order is generated whenever the Net Assets are less than or equal to the RO. Net assets are comprised of the Balance on Hand (BOH) plus Due In (DI) – stocked items on order plus Due Out (DO) – back orders awaiting stock replenishment. The only difference for a merchandising operation in this model is the absence of customer requests awaiting stock replenishment (DO). Either the item is on hand or else there is a loss sale. No back order is generated.
Sensitivity analysis simulations of each of the RO components can be accomplished to optimize customer satisfaction and inventory investment. Moreover, with the capacity of computing power currently available, virtual sensitivity adjustments can be made in inventory investment in real time.
Foresight Systems M & S has the software know-how to make virtual inventory optimization a reality. Some examples include:
- Transportation costs skyrocket as the fuel prices increase, triggering an increase in ordering costs which, in real time, will adjust upward the economic order quantity (EOQ). A virtual optimization of “cost to order” and “cost to hold” will result in overall lower inventory costs.
- Day-to-day changes in the cost of not stocking an item can be implemented without delay.
- Limited time quantity-discount offers from vendors and resulting changes in Q, OST and SL can be evaluated in near real time.
- Changes in the stochastic demand model used for forecasting can be accommodated in real time.
- “The inventory investment interest rate decreases thus lower holding costs which recalculates a higher Q that minimizes total costs of ordering and holding.
Foresight sensitivity simulation and virtual inventory optimization is the way of the future for most retail, asset triggered reordering systems. We are actively pursuing the application of Foresight’s proven systems modeling tools and methodologies to inventory and supply-chain management and will soon be bringing tailored, innovative solutions to market. If you have interest in such applications, please contact us. If you have questions or comments relevant to this post, we invite you to register and comment.
(Note, We did not address the Just-In-Time (JIT) inventory model . JIT is not applicable to conditions of uncertain demand and fluctuations in demand during lead time and fluctuations in the length of the delivery time. JIT is best used for manufacturing operations where demand can be more accurately forecasted for planned production. Also, with increasing fuel prices and transportation costs, JIT may not be the most cost effective solution.)