by [TC]²
Sourcing Simulator
Version 2.1 (SS)
The Sourcing Simulator is an interactive software package that provides the capability to:
1. Simulate retailing and retailing/manufacturing scenarios
for a line of product (Sourcing Analysis Model).
2. Model the relationship between selected operational parameters and/or
environmental factors and retail performance measures (Decision Surface
Model).
The package is designed to:
1. Allow the user to evaluate the impact of various retail and manufacturing
operating practices under different operating conditions
2. Provide a training tool for retail buyers, manufacturers sales reps
and managers.
It is designed to run on Windows-based systems without requiring extraneous
proprietary software.
Overview of the Sourcing Analysis Model
The Sourcing Analysis Model simulates the retailing process for a line of product, which is offered over one or several consecutive Selling Cycles of fixed duration, with any leftovers being liquidated at the end of the final cycle. The simulated activity includes execution of an ordering (sourcing) strategy, interaction of a stream of customers with the shelf stock resulting in sales (or lost sales) and, depending on the scenario, price promotions, and/or markdowns. Many performance measures are evaluated.
The model tracks inventory by SKU throughout the Selling Cycle. SKUs are differentiated by style, color, and size.
The vendor of product can be modeled in any one of three modes: Perfect Supplier, Modeled by Fill Rate, or Detailed Model. In the "Perfect Supplier" mode, any product ordered from the vendor is assumed to arrive on-time (in the lead-time specified) and complete (no shortages). In the "Modeled by Fill Rate" mode, a user-specified percentage of orders arrive "short," i.e., in-complete. The amount of shortage is user-specified. In the "Detailed Model" mode, the vendor model includes things like: manufacturing lead time, capacity, quality, shipping time, raw material inventory and supply, collaboration with the retailer, make-to-order or make-to-stock inventory control, and production plans are specified by the user. This mode allows a more thorough examination of the vendor's ability to respond to the orders placed on them by the retailer.
The program generates a random stream of customers, week
by week throughout the Selling Cycle. The arrival rate can vary from
week to week to reflect seasonality.
When a customer arrives, the program randomly assigns an associated
style, color and size from discrete probability distributions that describe
consumer preferences. If the desired SKU is in stock, one unit is purchased.
If the desired SKU is out of stock, the customer may choose another
specific SKU or leave.
In order to use the simulation model, the user must first define the scenario to be executed. This is done by interactively entering (or accepting default values for) a number of data items.
One of the key inputs is the buyer's plan. This consists of the length of the Selling Cycle (up to 260 weeks), the anticipated total demand (volume), the number of styles, colors, and sizes in the line, the anticipated percentage of demand (mix) by style, color, and size and the anticipated demand seasonality (i.e., percent of demand in each week of the season).
An important concept embodied in the model is that of buyer forecast error. This permits the execution of scenarios in which the buyer's plan is not an accurate estimate of the actual demand for the season. Error comes in three forms: volume error, mix error, and seasonality error. Volume error represents a difference between the actual demand volume and the plan volume. Mix error represents differences in the anticipated and actual percentages of demand by style, by color, and by size. Seasonality error represents a difference between the planned (i.e., anticipated) and actual percentages of demand in each week. Mix error can change depending on the week within the Selling Cycle.
The stocking (and restocking) of the product on the shelves is governed by a sourcing strategy. The model allows the user to select from six strategies. Two of the strategies rely totally on the buyer's plan figures; one permits limited use of point-of-sale (POS) data; while the other three permit extensive use of POS data to re-estimate demand and place within season reorders. Depending on vendor reliability, the content of deliveries may or may not match the content of the orders and there may be minimum order size requirements. The model also provides break-even analysis comparisons between alternative strategies.
Promotions, if any, reduce the selling price for a specified string of weeks while markdowns reduce the selling price until the end of the Selling Cycle (or a subsequent markdown). Promotions are initiated at specified starting weeks; markdowns may be initiated either at specified starting weeks or by lower than expected sales. Both promotions and markdowns can impact customer behavior in two ways: by increasing the customer arrival rate and by increasing the proportion of customers who look for another item when encountering a stockout of their current choice.
While simulating the season, the model calculates a large number of performance measures related to revenues, costs, inventory, and customer service on either a weekly or complete scenario basis. These can be viewed in tabular and, for some measures, in graphical format.
Manufacturing Output Statistics
Raw Material Inventory Measures
Residual Units is the number of units of raw material not consumed, i.e., the number of units leftover after the last shipment to retail.
Units Ordered is the number of units of raw material ordered from the supplier.
Average Inventory (Raw Material) is the average number of units of raw material held by the manufacturer from the time of the first receipt of raw material until the last shipment of finished goods to retail.
Inventory Turns (Raw Material) is the number of times the raw material inventory turns over from the time of the first receipt of raw material until the last shipment of finished goods to retail. It is computed as the total number of units received divided by the Average Inventory (Raw Material) divided by the number of selling cycles.
Finished Goods Inventory Measures
Units Demanded is the total number of units of product ordered by the retailer.
Units Produced is the total number of units of product actually produced by the manufacturer.
Units Shipped is the total number of units of product actually shipped from the manufacturer to the retailer.
Units Backordered is the total number of units of product ordered by the retailer but not shipped by the manufacturer on time.
Residual Units is the number of units of product produced and left over after the last shipment to retail.
Average Inventory (Finished goods) is the average number of units of product carried in finished goods inventory at the manufacturer.
Inventory Turns (Finished goods) is the number of times the finished goods (product) inventory "turned over" from the time of the first receipt of raw material until the last shipment of finished goods to retail. It is calculated as the Units Produced divided by the Average Inventory (Finished Goods) divided by the number of selling cycles.
Service Measures
% Shipped On-Time is the percentage of product that was shipped to the retailer on-time, i.e., within the lead-time specified.
% Backordered is the percentage of all product ordered by the retailer that was not shipped on-time.
% of Backorders Filled is the percentage of all backordered units that was eventually shipped to the retailer.
Costs
Raw Material Costs is the total cost of raw material ordered by the manufacturer. It is calculated as the Units Ordered times the Raw Material Cost.
Inventory Carrying Costs is the sum of the raw material and finished goods inventory carrying costs. Each is calculated as the average inventory (raw material or finished good) times the annual Inventory Carrying Cost Rate divided by the number of weeks from the first receipt of raw material until the last shipment of finished goods to retail.
Production Cost is the cost of producing product. It is calculated as the Units Produced times the Production Cost.
Revenues
Revenue is the revenue received from the sale of product to the retailer. It is calculated as the number of units shipped for initial inventory times the Initial Wholesale Cost plus the number of units of replenishment shipped times the Replenishment Wholesale Cost.
Liquidation Revenue is the revenue generated by liquidating both residual raw material and residual product. It is calculated as the Residual Units of raw material times the Raw Material Liquidation Price plus the Residual Units of Finished Goods times the Finished Goods Liquidation Price.
Total Revenue is the sum of the Revenue and Liquidation Revenue.
Revenue/Unit Produced is Total Revenue divided by the number of Units Produced.
Margins
Gross Margin is Total Revenue minus Raw Material Costs and Production Cost.
Adjusted Gross Margin is the Gross Margin minus Inventory Carrying Costs.
Overview of Decision Surface Model
Decision surface modeling is a powerful tool that allows you to capture and graph the relationship between selected inputs to the Sourcing Analysis model (e.g., lead time, SKU mix error. wholesale cost) and selected performance measures (e.g., service level, gross margin, GMROI). Once a decision surface model is built you can graph any of the selected performance measures versus any one or two of the selected inputs to see and better understand how the performance changes with changes in the value(s) of the input(s) without having to actually run the simulation.
In the Sourcing Simulator decision surfaces are modeled using neural networks. Neural networks permit the modeling of a wide range of complex surfaces without having to guess beforehand what the analytic shape of the surface is.
In order to create a decision surface model you first have to make a "batch run" with the Sourcing Analysis Model in order to specify the inputs and performance measures you are interested in and generate some data points. In a batch run simulations are executed for all combinations of values in ranges that you specify for one or more inputs. Once the batch run is made, you can create decision surface models relating any of a set of performance measures to any subset of the inputs that were specified in the batch run. Once a decision surface model is created, you may specify the range of interest for any one or two of the inputs in the model and graph the performance measure against the input(s) over the specified range. Up to six graphs, from one or several decision surface models, may be displayed side-by-side for easy comparison.
Output files from batch runs and decision surface models may be saved for future use. Graphs can be printed or saved to a file that can be imported into a word processor.
The Sourcing Simulator Version 2.1 runs under Windows
95, 98, 2000 and Windows NT. For more information call Jim Lovejoy at
Textile Clothing Technology Corp. at 919-380-2184, or e-mail at jlovejo@tc2.com
or Russ King at North Carolina State University at 919-515-5186, or
e-mail at king@eos.ncsu.edu.
For more information, please contact:
Jim Lovejoy
Textile/Clothing Technology Corp.
211 Gregson Drive
Cary, NC 27511
(919) 380-2184
FAX: (919) 380-2181
e-mail: jlovejo@tc2.com