Inventory management strategy evaluation system and method5963919Abstract A system and method for evaluating an inventory management strategy combines multiple management strategies in a single inventory management system. The system analyzes the inventory portfolio on an item-by-item basis to assign the most suitable management strategy for that item. The inventory management system provides a high level of flexibility for the users to define input parameters to ensure a desired level of customer satisfaction. Additionally, it determines whether the inventory items are forecastable before predicting future demands. Claims What is claimed is: Description BACKGROUND OF THE INVENTION
______________________________________
a. Four Factors
i. Who: Customer Order Entry generates replenishment order
directly, e.g., sales
ii. When: Upon placement of a customer order for the product
iii. How: Triggered by customer order input
iv. How Much: Lot-for-lot
b. Primary Characteristics
Product lead time falls within the customer requirement interval,
such
that it can be supplied directly from manufacturing (or outside
supplier) in time to meet customer demand without requiring stock.
High cost-per-order items
Demand is rare and random, and highly variable
Demand trend is not increasing
Cost is high relative to volume
c. Secondary Data Required for Management
Risk of obsolescence
Setup cost
Convertibility - can be made from, or made into, another finished
good (configurability)
______________________________________
2. Replenish-To-Order (Kanban)--Optimal strategy with rare demand items. Customer order interval is sometimes shorter than the supply lead time. There is a small stock in the warehouse, which is replenished as it is depleted by customer order.
______________________________________
a. Four Factors
i. Who: Warehouse generates replenishment order
ii. When: Upon fulfillment of a customer order from stock
iii. How: Per-use review; triggered by picking activity
iv. How Much: Lot-for-lot
b. Primary Characteristics
Product lead time at least occasionally does not fall within the
customer requirement interval
High cost items
Demand is rare and random, and highly variable
Demand trend is not increasing
Cost is high relative to volume
"Other" demand insufficient to cover usage ("other" demand such as
MRP dependent demand for an item that is used both as a
sub-assembly and sold as a finished good or spare part)
c. Secondary Data Required for Management
Risk of obsolescence
Setup cost
Convertibility - can be made from, or made into, another finished
good (configurability)
Quantity discounts (if purchased)
Holding cost (size, weight, fragility, shelf life)
Demand source
Desired customer service level
Cost to expedite
______________________________________
3. Warehouse Replenishment (EOQ)--Strategy appropriate for items that are replenished based on make-to-stock reorder point (ROP). There is a cost-optimized level of stock held in the warehouse, replenished at ROP.
______________________________________
a. Four Factors
i. Who: Warehouse generates replenishment order
ii. When: Whenever inventory level falls below reorder point
iii. How: Per usage or continuous review
iv. How Much: Economic Order Quantity
b. Primary Characteristics
Product lead time does not fall within the customer requirements
Cost low relative to volume, and below some absolute value
Demand and lead time is relatively stable
Relatively forecastable
c. Secondary Data Required for Management
Seasonality and trend
Risk of obsolescence
Setup cost
Convertibility - can be made from, or made into, another finished
good (configurability)
Quantity discounts (if purchased)
Holding cost (size, weight, fragility, shelf life)
Demand source
Desired customer service level
Engineering change frequency
Stockout cost
Planned sales promotions
Forecast (especially if demand is high or variable, and/or forecast
quality is good, and lead time is not variable)
______________________________________
4. Fixed-Rate Supply--Strategy that works well with high volume, stable demand, commodity item. Continuous production allocated as product comes off manufacturing line.
______________________________________
a. Four Factors
i. Who: Manufacturing sets fixed-rate production based on
near-zero variability future demand
ii. When: Continuous production, lead-time essentially zero
iii. How: Customer orders filled as product comes off the line
iv. How Much: Fixed rate
b. Primary Characteristics
High-volume, low-cost
Very low demand variability
Product lead time falls within the customer requirement interval,
such that it can be supplied directly from manufacturing in time to
meet customer demand without holding stock. Typically, a product
that is produced continuously and allocated to the customer order as
it rolls off the line.
c. Secondary Data Required for Management
Demand trend
Freight cost
Control of capacity (manufactured in-house or under capacity
contract with supplier)
______________________________________
5. Multi-Input Expert Planning--Strategy optimal where cost, trend, or variability of item demand justifies expert planning. While EOQ, and often forecast, implementation details are provided, user is strongly encouraged to look more closely at these items before assigning any inventory strategy due to the high risk factors involved.
______________________________________
a. Four factors
i. Who: Expert master planner (or expert system) with
Marketing and Operations visibility
ii. When: Using heuristics based on forecast, BOH, planned
customer orders, and capacity
iii. How: Periodic or continuous review
iv. How Much: Decision driven by planner, based on best
estimates of forecast, EOQ, marketing input
b. Primary Characteristics
High cost and high volume
Cost high relative to volume
Demand is extremely variable
Lead time is extremely variable
c. Secondary Data Required for Management
Inventory on hand in warehouses
Capacity
Planned customer orders
Setup cost
Quantity discounts
Holding costs (size, weight, fragility, shelf life)
Convertibility - can be made from, or made into, another finished
good (configurability)
Demand source
Engineering change frequency
Stockout cost or customer service level
Sales promotions
Forecast (or demand history)
______________________________________
6. Forecast Optimal--Optimal strategy for items having a demand history with patterns supporting statistical forecasting.
______________________________________
a. Four Factors
i. Who: Expert master planner with forecast input
ii. When: Weekly input based upon MRP/MPS requirements
iii. How: Periodic or continuous review
iv. How Much: Decision based upon weekly forecast
requirements and desired service level
b. Primary Characteristics
Significant increasing or decreasing trend
Significant seasonal or cyclic pattern
Low "noise" other than trend or seasonality
Low intermittence (weekly zero-demand occurrence)
c. Secondary Data Required for Management
Inventory on hand in warehouses
Capacity
Planned customer orders
New product introduction
Planned obsolescence
Holding costs (size, weight, fragility, shelf life)
Convertibility - can be made from, or made into, another finished
good (configurability)
Demand source
Engineering change frequency
Stockout cost or customer service level
Sales promotions
______________________________________
FIG. 3 is a graphical representation of where each of the above six strategy falls on the supply chain in terms of how close to the actual customer order the inventory replenishment order takes place. For example, for strategy #1 (Make-To-Order), replenishment point coincides with the customer order entry point. Replenishment points for strategy #2 (Replenish-To-Order) and #3 (Warehouse Replenishment) are based on the level of stock held in the warehouse. To initiate MISER program 250 in the preferred implementation, a user double-clicks on the program icon and selects an inventory portfolio to be evaluated. Inventory portfolios are generally stored in inventory database 240 as a separate input file using spreadsheet program 260. To convert the spreadsheet file into a "tab-delimited" text file for use in MISER program 250, a user simply selects an option in spreadsheet program 260 to "Save As" a text tab-delimited file. This converts the spreadsheet file into a universally-readable plain text format. Thus, input file is now a "tab-delimited" text file containing essentially a large table holding all the information about a specific portfolio, one item per line. FIG. 4 is a sample table illustrating the input format of an inventory item. The format and content of input table 400 are exemplary only and may be easily modified. For illustration purposes, a column of an input line, representing a field of an inventory item, is shown as a row in an input table 400. For example, column A of the input line contains a line item index number field in an integer format. Other fields include the inventory item's description, number of order, customer requirement interval (CRI), standard deviation of CRI, setup cost, average lead time, unit cost, total demand, and weekly demand. After the input file is properly formatted for MISER program 250, the user may specify input parameters about how the user wants to use the information in the input file. FIGS. 5a-5c are sample input screen displays for specifying input parameters. FIG. 5d is a sample input screen summarizing the input parameters. During this process, the user may override values contained in the input file with user-specified values. For example, as shown in FIG. 5a, MISER program 250 allows the user to specify the output file name, number of weeks of historical demand data, and desired customer assurance level. If the user inputs 95% for desired customer assurance level, MISER program 250 then assures that the level of stock carried, or currently in the pipeline, is sufficient to meet at least 95% of orders coming in based on past trends and future forecasts. FIGS. 5b and 5c show other input fields that the user may define such as global setup cost values, customer required interval, manufacturing lead time, and purchasing lead time. After the user finishes defining input parameters, MISER program 250 presents a summary of the input parameters to the user as shown in FIG. 5d. The user at this point may change any of the displayed values or choose the "continue" icon to run MISER program 250. If the user double-clicks on the "continue" icon, computer 110 runs MISER program 250 to create an output file. FIG. 6 is a sample table illustrating the output format of an inventory item. For illustration purposes, a column of an output line, representing a field of an inventory item, is shown as a row in an output table 600. Various fields include the selected strategy for each item in the portfolio as well as information necessary to implement the strategy. In evaluating the portfolio to select the appropriate strategy, MISER program 250 uses a decision tree 700 illustrated in FIG. 7. Before proceeding through the decision tree, however, MISER program 250 stratifies the portfolio. The goal of stratification is to identify the cutoff values used in various nodes of the decision tree. Thus, after the cutoff values are determined, each inventory item can be individually evaluated using the decision tree by comparing the various input parameters of the item to the cutoff values. Then, an appropriate inventory strategy is assigned based on whether the item meets the cutoff threshold at certain critical nodes of the decision tree. By doing so, MISER program 250 identifies, for example, items which are rare or contribute only slightly to the total portfolio. Stratification techniques are well known in the production/inventory management field and will not be described in detail here. Details on stratification techniques, such as ABC stratification, are provided in Benito E. Fores & D. Clay Whybark, Implementing Multiple Criteria ABC Analysis, APICS Journal of Operations Management 79-85 (1992). Specifically, MISER program 250 stratifies the portfolio based on three criteria: order cost, volume of orders, and number of orders. For each criteria, stratification process produces three threshold values, costlimit, demlimit, and ordlimit, respectively. To do so, MISER program 250 sorts the portfolio items according to each of the three criteria, then for each criteria, MISER program 250 determines a specific cutoff value based on pre-defined or user-specified input parameters. The first stratification criteria, cost per order, identifies items that are "high risk" on an order-by-order basis. For example, MISER program 250 identifies items that represent, for example, the top 5% of the portfolio. Such items expose a business to a high risk of loss if an order is incorrectly placed, thereby either sacrificing a large amount of revenue, or necessitating a large amount of dollar-volume in a warehouse. MISER program 250 assigns the numerical value representing the top 5% of the order cost to variable costlimit. Next, total volume stratification identifies a class of items whose volume falls within, for example, the top 20% of the portfolio. For example, if a portfolio contains ten items and the volume sold for each of the ten items is 1, 2, 3, 4, 5, 6, 7, 8, 9, and 10, respectively, the total volume is fifty five and the last six items account for 82% of the total volume (45 out of 55). Then, according to the volume objective, which is either pre-defined or user-specified, items falling below the objective are isolated for a different strategy. In the above example, if the volume objective is 80%, then the first four items, with volume of 1, 2, 3, and 4, respectively, fall below the cutoff line and are isolated for assignment to a different strategy. Likewise, if the cutoff volume limit is five, then any item with volume of five or greater would pass the test. Those items with fewer than five would fail the test and a different inventory strategy is assigned. MISER program 250 assigns the numerical value representing the volume limit is to variable demlimit. Finally, order frequency stratification determines items having a high order frequency and therefore, a high correlation to customer services. If these items are misordered, it could result in a large number of unfilled orders or excessive volumes to be carried in stockrooms. MISER program 250 assigns the numerical value representing high order frequency to variable ordlimit. As described in connection with specific management strategies, some inventory strategies require forecasting future demand. MISER program 250 implements a two-step process to provide a more accurate assessment of future demand. First, MISER program 250 analyzes each inventory item to determine whether the item is forecastable. Second, if an item is deemed forecastable, MISER program 250 forecasts the future demand of the inventory item. The two-step approach of system 100 forecasts inventory demand only under appropriate and optimal circumstances, and avoids forecast error of any great magnitude. To determine whether the item is forecastable, MISER program 250 divides the items into three categories: forecast-inappropriate, forecast-appropriate, and forecast-optimal. During the initial screening, MISER program 250 analyzes whether an inventory item has an extremely low demand or order frequency with orders occurring in a totally random fashion. In other words, MISER program 250 determines whether there is sufficient amount of historical data and analyzes the coefficient of variation among the historical data to determine the level of fluctuation among the available data. If there is insufficient amount of historical data or if the historical data fluctuates beyond a desired level, MISER program 250 renders the item as forecast-inappropriate. If not, the item is forecast-appropriate. In the next layer of the screening process, MISER program 250 ascertains which of the forecast-appropriate items are forecast-optimal. This screening process, which will be described in detail below, takes place within a decision tree 700 shown in FIG. 7. As will be explained below, if the item is classified as forecast-optimal, MISER program 250 assigns strategy #6 (Forecast Optimal) and predicts future demand based on historical data. After obtaining the three cutoff values by stratifying, MISER program 250 proceeds with the evaluation process according to decision tree 700. MISER program 250 first checks whether "Average Order Cost>costlimit" (step 710). Average Order Cost of the inventory item is the total dollar sales for the total period surveyed divided by the total number of orders for the item during that period. As previously mentioned, costlimit is obtained by stratifying the portfolio by average order cost per line and represents a 95th percentile limit of average order cost. The decision criteria of step 710 identifies items that are high-risk in terms of total expenditure for a single order relative to the total product portfolio. If Average Order Cost exceeds costlimit, then MISER program 250 determines whether "Average Demand>demlimit" (step 715). Average demand is obtained by dividing total demand by the number of periods for which the demand data is provided. As mentioned above, delimit is obtained by stratifying the portfolio by total volume per line item and represents a certain percentile of an entire product portfolio. The decision criteria of step 715 identifies highest volume line items in the portfolio. If Average Demand exceeds demlimit, MISER program 250 assigns inventory strategy #5 (Multi-Input Expert Planning) to the inventory item. If Average Demand falls below demlimit, MISER program 250 checks whether the item is "forecastable," i.e., whether future demand of the item can be properly predicted based on historical demand data (step 720). Specifically, if the inventory item has fewer than 32 periods of zero-demand, MISER program 250 computes a coefficient of variation among the past demand data. If the coefficient of variation for the non-zero demand period is less than 1.0, the item is considered forecastable and MISER program 250 assigns strategy #6 (Forecast Optimal) to the inventory item. Thereafter, MISER program 250 forecasts future demand of the item using a generally known forecasting technique such as the Croston methodology. Croston method is a modified version of a forecasting technique known as exponential smoothing used to accommodate intermittent demand. Exponential smoothing is defined in detail in D. D. Bedworth & J. E. Bailey, Integrated Production Control Systems (2d ed. 1987). If the item contains no more than one period of zero-demand, an autoregressive error variability is small relative to trend magnitude, and forecasting may be performed using the autoregression method. If the item is not forecastable, MISER program 250 first checks whether "Supply Lead Time (SLT)<CRI for custsat % of the time" (step 725). SLT is the time required to manufacture or otherwise procure a specific item for delivery to a customer. On the other hand, CRI is the time period between placement of an order and the customer-specified delivery date. Custsat% is the user-specified level of customer satisfaction. For example, if the user specified the customer assurance level as 95%, then custsat% is 95%. If the time necessary to supply a given item is less than the interval of time the customer requires for delivery, there is no need to maintain an inventory of item on stock. In other words, if SLT<CRI, the item should be made to order. Because SLT and CRI are random rather that absolute values, there is always some chance that the item will not be deliverable within the customer required interval. MISER program 250 weighs the probability of on-time delivery of each inventory item given its demand and supply distributions according to the graph shown in FIG. 8. MISER program 250 takes the weighed probability and ascertains whether it falls within a user-defined customer assurance level to ensure an on-time delivery of the item. For example, if a user-defined customer assurance level is 95%, the probability that the item will be late must be 5% or less. To determine whether the probability is below the threshold, for example, below 5%, a solution Z is obtained by assuming the following: 1. Lead Time, LT, is a normal random variable with mean .mu..sub.LT and standard deviation .sigma..sub.LT. 2. Customer Requirement Interval, CRI, is a normal random variable with mean .mu..sub.CRI and standard deviation .sigma..sub.CRI. 3. CRI and LT are independent random variables. Given these assumptions, the solution is: ##EQU1## where Z is the standard normal distribution coefficient of the user-defined customer service target level. If Z is less than the tolerable probability of a late delivery, in the above example 5%, then MISER program 250 assigns strategy #1 (Make-To-Order) to the inventory item. If not, MISER program 250 proceeds to the next node in decision tree 700. Next, MISER program 250 checks whether the item has a history of lumpy, or highly variable, demands (step 730). MISER program 250 computes a standard coefficient of variation (COV), which is a ratio of standard deviation of demand per period to average demand per period. A COV greater than 1.0 indicates a non-forecastable demand. If the item has demands, then MISER program 250 assigns strategy #5 to the item. If not, the item is assigned strategy #3 (Warehouse Replenishment). Returning to step 710, if Average Order Cost does not exceed costlimit, then MISER program 250 determines whether "Cost/Volume>maglimit" (step 735). Cost/Volume is the unit cost of an item divided by total demand. Maglimit is the limit on cost/volume ratio, whose numerical value is dependent on product portfolio characteristics. The numerical value assigned to maglimit is dependent on the product portfolio. In this step, the ratio of cost and volume per line is calculated and compared to the upper limit, maglimit. Testing has shown that setting maglimit=1.0 has proven effective at identifying products that have exceptionally high cost to volume ratios. This node distinguishes line items that are expensive and have relatively low demand. Such items have significant potential for inflating inventory costs if mistakes are made in the procurement process. If the ratio of cost and volume per line exceeds maglimit, MISER program 250 determines whether the SLT is highly variable (step 740). If so, MISER program 250 assigns strategy #5 to the item. If SLT is not highly variable, then MISER program 250 ascertains whether the item is forecastable (step 745). Similar analysis is performed as described in connection with step 720. If the item is forecastable, MISER program 250 assigns strategy #6 to the item. If the item is not forecastable, MISER program determines whether "SLT<CRI" (step 750). Similar analysis is performed as described in connection with step 725. If SLT<CRI, then MISER program 250 assigns strategy #1 to the item. If not, strategy #2 (Replenish-To-Order) is assigned. Returning to step 735, if the ratio of cost and volume per line does not exceed maglimit, MISER program 250 determines whether "SLT<CRI" (step 755). Similar analysis is performed as described in connection with step 725. If SLT<CRI, then MISER program 250 determines whether the item has a history of highly variable demands (step 760). Similar analysis is performed as described in connection with step 730. If the item does not have highly variable demands, the item is assigned strategy #4 (Fixed-Rate Supply). If the item has highly variable demands, MISER program 250 ascertains whether the item is forecastable (step 765). Similar analysis is performed as described in connection with step 720. If the item is forecastable, MISER program 250 assigns strategy #6. Otherwise, it assigns strategy #1 to the item. Returning to step 755, if SLT is not less than CRI, MISER program 250 determines whether the item has highly variable demands (step 770). Similar analysis is performed as described in connection with step 730. If the item does not have highly variable demands, the item is assigned strategy #3. If the item has highly variable demands, MISER program 250 ascertains whether the item is forecastable (step 775). Similar analysis is performed as described in connection with step 720. If the item is forecastable, it is assigned strategy #6. Otherwise, MISER program 250 determines whether "Number of Orders>ordlimit" (step 780). As explained above, ordlimit is an annual limit on the number of orders placed based on a percentile of an order-based stratification of the entire portfolio. If the number of orders exceeds this ordlimit, MISER program assigns strategy #5. Otherwise, it assigns strategy #2 to the item. Once MISER program 250 determines the recommended optimal inventory management strategy for each portfolio item, it calculates additional values necessary to implement the recommended strategy. The following details the calculations performed to provide this information:
______________________________________
1. Make-To-Order
Reorder Point = 0
Safety Stock = 0
Kanban = 0
Economic Order Quantity = 0
Initial and Average Inventory Levels = 0
2. Replenish-To-Order (Kanban)
Reorder Point = 0
Safety Stock = 0
Economic Order Quantity = 0
Kanban = K = .mu..sub.demand + k.sigma..sub.demand
.mu..sub.demand = average demand per period
.sigma..sub.demand = standard deviation of demand per period
L = lead-time
k = multiplier corresponding to normal distribution
function for designated percentile identified as
desired service level
Initial Inventory Level = Kanban Size
Average Inventory Size = Kanban - D*L
D = average demand per period,
L = average lead-time.
3. Warehouse Replenishment (EOQ)
Kanban = 0
Reorder Point = ROP = DL
D = average demand per period,
L = average lead-time.
1 #STR1##
k = multiplier corresponding to normal distribution
function for designated percentile identified as
desired service level
.sigma..sub.demand = standard deviation of demand per period
L = procurement lead time
Formula assumes approximate normal distribution for period
demand. For a desired 95% service level, k = 1.64.
Economic Order Quantity
2 #STR2##
D = Total Demand for survey periods
S = Setup cost per order
H = Holding cost per unit. (unit cost * holding cost
percentage)
Initial Inventory
InitInv = EOQ + Safety Stock, if ROP < EOQ
InitInv = EOQ + Safety Stock + ROPi, if ROP > EOQ
Average Inventory = 0.5*EOQ + Safety Stock
4. Fixed-Rate Supply
Reorder Point = 0
Safety Stock = 0
Kanban = 0
Economic Order Quantity = 0
Initial and Average Inventory Levels = 0
5. Multi-Input Expert Planning
Kanban = 0
Reorder Point = ROP = DL
D = average demand per period,
L = average lead-time.
1 #STR3##
k = multiplier corresponding to normal distribution
function for designated percentile identifled as
desired service level
.sigma..sub.demand = standard deviation of demand per period
L = procurement lead time
Formula assumes approximate normal distribution for period
demand. For a desired 95% service level, k = 1.64.
Economic Order Quantity
2 #STR4##
D = Total Demand for survey periods
S = Setup cost per order
H = Holding cost per unit. (unit cost * holding cost percentage)
Initial Inventory
InitInv = EOQ + Safety Stock, if ROP < EOQ
InitInv = EOQ + Safety Stock + ROPi, if ROP > EOQ
Average Inventory = 0.5*EOQ + Safety Stock
6. Forecast Optimal
Reorder Point = 0
Kanban = 0
Economic Order Quantity = 0
1 #STR5##
k = multiplier corresponding to normal distribution function for
designated percentile identified as desired service level
(Hard-coded to 3.0 (99.8% satisfaction) under forecast
optimal condition in order to insure adequate levels)
.sigma..sub.demand = standard deviation of demand per period
L = procurement lead time
Formula assumes approximate normal distribution for period
demand. For a desired 95% service leve1, k = 1.64.
Initial Inventory = Forecast + Safety Stock
Average Inventory = .5 * Forecast + Safety Stock
______________________________________
After the evaluation process, MISER program 250 also displays a summary of the output results as shown in FIG. 9. The summary information includes, for example, the output file name, number of inventory items, number of weeks of historical demand data, average inventory cost, total inventory turns, and a strategy count for each of the six inventory strategies. CONCLUSION The system and method consistent with the present invention accommodates special characteristics and requirements of various types of inventory portfolios. It also combines multiple management strategies into a single inventory management system to provide an accurate and optimal inventory management tool. Additionally, it predicts future demands for inventories by first determining whether inventory items are forecastable. It will be apparent to those skilled in the art that various modifications and variations can be made in the computer network of the present invention and in construction of this computer system without departing from the scope or spirit of the invention. Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. The specification and examples should be considered as exemplary only, with the true scope and spirit of the invention indicated by the following claims.
|
Same subclass Same class Consider this |
||||||||||