Supply Chain Simulation

An Analysis of Traditional Versus Collaborative Supply Chains In The Softgoods Industry

By Tim Curran and Jim Lovejoy

Textile Clothing Technology Corporation

September, 2001

1 Supply Chain Traditional and Collaborative Simulations
A major focus of the DAMA Project was developing models to prove that a collaborative supply chain would provide superior performance in responsiveness and inventory control. In an effort to provide this simulation capability, multiple models were developed to simulate the supply chain alternatives. These models include all of the planning, production, and distribution functions that exist in the current business process. Production is driven by forecasts that are developed through Demand Planning, Corporate Resource Planning and Product Resource Planning. Incoming orders are processed through Order Fulfillment and into the Warehouse for shipping to customers.

The first models developed focused on improving lead-time through the supply chain. These simulation models, the Traditional Industry Supply-Chain Simulation - Lead Time (TISS-LT) and Collaborative Industry Supply-Chain Simulation - Lead Time (CISS-LT), were developed to show the direct impact that collaborative business processes have on lead-time through the pipeline. As expected, the effects on lead-time were dramatic and showed clearly that CISS-LT was indeed more responsive to fluctuations in consumer than demand than TISS-LT (see results on page 12).

Once these initial models were proven, more detailed models were developed to show that collaboration had positive effects on both lead-time and inventory costs. Using the original models as a foundation, the Traditional Industry Supply-Chain Simulation (TISS) and Collaborative Industry Supply Chain Simulation (CISS) models were developed. These models include additional logic to model both lead-time and inventory cost. The CISS and TISS models are used to demonstrate the broader impact of building a supply chain based on the DAMA Architecture.

2 The DAMA N-Tier Collaboration Models
The DAMA N-Tier Collaboration Model is a product of the DAMA project, and was developed over a period of time. The contributions of the many companies involved in the pipeline analyses, documentation of the business process models, prototype tools and pilots has been a key factor in developing both the DAMA N-Tier Collaboration Business and Detail Models.

Before the DAMA N-Tier Collaboration Models were developed, the DAMA Model for Supply Chain Collaboration was published. And in order for DAMA to understand the complete supply chain, it was necessary to understand the "As-Is" information model of the textile industry today. Typically, a textile supply chain consists of several manufacturers, each representing a sector of the industry; i.e. fiber, textile, apparel (sewn products) and retail. A model of the industry was documented that shows the flow of information between these sectors, and is represented in Figure 1.

Figure 1. AS-IS Model of U.S. ITC

Information is passed between sectors in the form of Electronic Data Interchange (EDI) transactions, and typically each sector is customer focused (fiber focuses on the textile customer), rather than consumer focused (all sectors focus on consumer demand). Internal to each company there are a number of business processes that occur (forecasting, planning, scheduling, purchasing, etc.). And typically, the customer and supplier in the supply chain have little knowledge of those transactions that are occurring within the other Trading Partner Company(s).

In order for all members of the supply chain to respond to consumer demand, a new collaborative paradigm was required. This new paradigm will provide supply chain visibility to critical information for all members of the supply chain. The DAMA Model for Supply Chain Collaboration has been developed to show how all sectors of the supply chain would participate collaboratively in the major business processes that traditionally have occurred only within the four walls of a particular company. This model suggests that retail, apparel, textile and fiber companies within a particular supply chain share information and collaboratively make decisions about forecasting, planning, scheduling, product delivery and expediting orders. The DAMA Model for Supply Chain Collaboration is a high-level model for collaboration to achieve Demand Activated Manufacturing as shown in Figure 2. There are six collaborative activities that may be employed in this model:

1. Develop Business Planning Agreements,
2. Define Products,
3. Forecast and Plan Capacity Commitments,
4. Schedule Product and Product Delivery,
5. Expedite Production and Delivery Exceptions, and
6. Populate Supply Chain Utility.

Figure 2. DAMA Model for Supply Chain Collaboration

For each of the first five collaborative activities, the trading partners must populate the Supply Chain Utility, the sixth activity in this model. The Supply Chain Utility is a set of applications implemented to support collaborative product definition, forecast visibility, planning, scheduling, and execution.

This model assumes a collaborative supply chain, with multiple trading partners, working collaboratively to meet consumer demand. Trust must exist between all trading partners, and technical data security must be implemented. Working together, the trading partners share information about their products, manufacturing capabilities, allocations of capacity to the partnership, and day-to-day operational status. Each of these collaborative activities will be discussed in detail in the following sections.

The DAMA Model for Supply Chain Collaboration was published, and presented to the VICS CPFR® Committee as a recommendation for N-Tier Collaboration. The CPFR N-Tier committee has identified the DAMA Model for Supply Chain Collaboration as a basis for documenting alternative business models that N-Tier CPFR will support.

This DAMA Model for Supply Chain Collaboration led to the development of the DAMA N-Tier Collaboration Business Model (see Figure 3). The DAMA N-Tier Collaboration Business Model describes a collaborative paradigm that will provide simultaneous visibility to critical information for all members of the supply chain. The model illustrates how retail, sewn products, textile and fiber companies within a particular supply chain share a single demand plan and collaboratively make decisions about forecasting, production planning and scheduling, and product delivery whenever an exception from the original collaborative forecast occurs.

The value of having a single demand plan visible to all members of the chain is to better coordinate process activities across the supply chain. Not only will trading partners better manage the development of a single shared forecast of consumer demand, but they will also leverage that shared view of the forecast by using it to plan production schedules, inventory levels, manufacturing capacity, and ship dates throughout the chain.

The principles described above are embedded in the DAMA N-Tier Collaboration Business Model (Note: the activities below are numbered in Figure ). The primary activities in the model include:

• Collaborative Planning (1),
• Collaborative Forecasting (2),
• Generate Forecasts (3),
• Generate Production Orders (4),
• Generate Ship Orders (5),
• Collaborate To Resolve Exceptions (3b, 4b, 5b), and
• Identify Front-End Agreement Changes (6).

Figure 3. DAMA N-Tier Collaboration Business Model

Companies Provide Business Planning Data (1a) to establish the guidelines and rules for the collaborative relationship. The Voluntary Inter-industry Commerce Standards (VICS) Collaborative Planning, Forecasting and Replenishment (CPFR®) process has provided guidelines for this step. As described by CPFR®, the front-end agreement addresses each party's business goals and the actions and resources necessary for success.

Inputs to the process for planning multiple manufacturing partners (N-Tier Collaboration) would include not only strategy and goals for the partnership, but also specifications for the product being delivered, and inventory and capacity allocations to the partnership.

Collaborative Forecasting (2) requires several key inputs from the trading partners which, taken together, comprise the framework within which the forecast will be managed. Companies Provide Forecast Information (2a) for this step such as market projections, internal forecasts, and historical data. The output of the collaborative forecasting step is a collaborative forecast, and a commitment by each of the trading partners to meet that forecast according to the plan established by the partnership.

Once the initial collaborative forecast has been decided upon by the partnership, Corporate Resource Planning (3a) in each company provides periodic forecast updates, which should be evaluated and processed to determine if the original forecast stands, or if an exception has occurred. The updates are used to Generate Forecasts (3). When an exception occurs, Companies Collaborate to Resolve Exceptions, and a Forecast Resolution (3b) is created. The Forecast Resolution is then processed to provide new Forecast updates to each of the Corporate Resource Planning (3a) organizations in the trading partner companies.

Within a specified time period (established during the Collaborative Planning (1) phase, and defined in the FEA (1b)), the Supply Chain Utility interprets the forecast, and produces production orders. These production orders incorporate lead times required throughout the product life cycle in the supply chain, and are distributed to Product Resource Planning and Production (4a) in each company unless an exception has occurred. Exceptions might include changes in lead-time, or product specifications (e.g. change in color, or size, as a result of previous forecast updates (3a), or inventory status updates (5a). Again, exceptions require that Companies Collaborate to Resolve Exceptions and create a Production Order Resolution (4b). The Production order resolution is then processed to provide new Production Orders to each of the Product Resource Planning and Production (4a) company organizations.

A second time period (established during the collaborative planning phase, and defined in the front-end agreement) is defined to establish when the updated forecasts are used to Generate Ship Orders (5). If the Ship Orders are within the variance defined by the front-end agreement, they are translated by the bill of materials into individual Company Ship Orders, and sent to each company for processing by the Warehouse, Order Fulfillment and Transportation organization (5a). If a Ship Order is outside the variance, the Ship Order is flagged, then Partners Collaborate to Resolve Exceptions, and a Ship Order Resolution (5b) is used to generate a new Ship Order for each company. Exceptions might occur to a significant change in one of the company's inventory status, thus requiring an increase/decrease in upstream or downstream shipments, or as a result of earlier resolutions for forecast updates, or production orders.

At any time when Partners Collaborate To Resolve Exceptions (3a, 4a, or 5a), and the Resolution Requires Changes to the Front End Agreement (FEA) (6), then those Collaboration Agreement Revisions will require the companies to re-enter Collaborative Planning (1), and revise the Front End Agreement (1b).

3 Validation of DAMA Architecture
Although the DAMA Architecture was developed based on sound concepts, it required validation to support the perceived benefits. In this effort, simulation models were written to compare the performance of proposed supply chain solutions and the traditional supply chain operations. The measure of success is shown as improvements in lead-time and inventory levels while maintaining a level of service.

Since most complex, real-world systems with stochastic elements cannot be accurately modeled mathematically, a simulation is often the only type of investigation possible. Stochastic simulation uses computer techniques to imitate or evaluate a model numerically in order to estimate the desired true characteristics of a system having random input components. With a model that mimics a system's behavior, one avoids the risk and cost of trial-and-error experimentation on the real world system and can perform "what-if" analysis to evaluate performance and identify existing or potential problem areas.

There are many simulation languages that can be used to develop such process simulations. However, the specific modeling requirements for the DAMA Supply Chain models fit into the capabilities of Rockwell Software's Arena modeling language. The Arena graphics simulation system (4) is designed for building computer models that accurately represent an existing or proposed application, such as schedule optimization, increased throughput, inventory reduction, delivery performance, and resource utilization. Arena integrates all simulation-related functions--animation, input data analysis, model verification, and output analysis--into a single simulation-modeling environment.

Figure 4. Arena Simulation Programming Environment

Arena is a complete and flexible modeling environment and was combined with an easy-to-use graphical user interface using Microsoft Access. This flexibility allowed for rapid development of the models, and a graphic user interface for data entry (Figure 5). The user interface is completely external to the modeling environment and provides a tailored data entry environment.

Figure 5. Sample of Graphic User of Interface

4 Description of Arena Models
In order to validate the DAMA Architecture, multiple models were required to handle the various supply networks and business processes. The first of the models developed was called the Traditional Industry Supply-Chain Simulation - Lead Time (TISS-LT) and focused on accurately modeling the current supply chain and calculating lead times for the various partners in the supply chain. Subsequent models were built on this framework and added functionality to support the DAMA Architecture. The Collaborative Industry Supply-Chain Simulation - Lead Time (CISS-LT) added the logic for collaboration between the supply chain partners. It resulted in a dramatic decrease in lead-time because it eliminated duplication of effort in the planning and other administrative departments. Once these models were developed, logic was added to determine inventory levels for each of the partners. These two models were simply named TISS and CISS and include the lead-time calculations, as well as, the inventory information.

The first model, TISS-LT, was developed to simulate traditional business processes and included all applicable business functions in the current supply chain. The flow of information in the supply chain begins with retail merchandising generating sales forecasts for the season. This information is sent to the Demand Planning department of the Apparel partner. Demand Planning performs two primary functions. The first is to process the incoming forecasts and generate new forecasts that are sent to the Textile Partner.

The second function is processing the forecasts into manageable monthly "buckets" for Corporate Resource Planning to distribute between manufacturing facilities. Corporate Resource Planning completes this task and transmits the information to Product Resource Planning to generate weekly production orders for the plants.

The other function of Product Resource Planning is to send the raw material requirements that correspond to the production order to Purchasing. Purchasing will send orders for raw materials from the Apparel partner to the Textile partner. These raw materials are eventually received in the Raw Material Warehouse and held for consumption by production orders.

The production orders are sent to the Production area that manufactures the defined quantities of product in weekly batches. Once the manufacturing is complete the product is transported to the finished goods warehouse and is entered into stock to be relieved by incoming orders.

The orders arrive to Apparel from Retail and are processed through Order Fulfillment. If the orders are in line with the forecasted quantities they are simply passed through to the warehouse for shipment. If the order does not fall within the expected tolerances of the forecast, then a notification is sent to Product Resource Planning to produce additional quantities and then the order is sent to the Finished Goods Warehouse to be processed.

In the Finished Goods Warehouse, an order is pulled from stock and then sent through Transportation to the Retail Partner. These same processes are executed for all of the partners. This causes long planning times and slow reaction to fluctuations in consumer demand.

The CISS-LT model incorporates most of the information and product flow that was developed in the TISS-LT model. However, the CISS-LT model includes collaborative planning for the supply chain partners. This planning begins with defining a business agreement and a collaborative forecast before a season starts. It also includes collaborative planning throughout the execution of the selling period. This means that there is no duplication of effort or additional processing of demand information in the supply chain. Once an order is sent from Retail, it is visible to all partners. This allows Textile and Fiber Partners to see fluctuations in demand immediately. Also, the primary functions of Demand Planning, Corporate Resource Planning, and Order Fulfillment are consolidated into the Supply Chain Utility and receive input from all sectors but are not duplicated within each sector.

The TISS and CISS models use the logic from the original models with added functionality to account for inventory levels. The logic required was added to several functional areas within the models. In Order Fulfillment, incoming order quantities are compared to the forecast and then either added or subtracted from current production orders. In the finished goods warehouse, inventory is added from Production and relieved to fill orders. This running inventory is recorded and could be used for cost calculations. The same process of managing inventory is performed in the raw material warehouse as well with inventory increased from deliveries from the preceding partner and then relieved to fill production orders.

5 Lead Time Use Case

In order to fully validate the architecture, a realistic use case was developed that provided industry data to the model. This data was collected from members of the industry team and can be easily and quickly modified through the user interface. For the TISS-LT and CISS-LT models, the supply chain consists of one retailer with 200 stores, one apparel supplier for private label poly-cotton men's pants, a single textile manufacturing source for the fabric and a manufacturing source for the polyester fiber. The cotton supply arrives as needed. There was no outsourcing or contracting of the manufacturing.

The retailer's stores are open 7 days a week and sell about 10 pairs of these pants a day per store. Total forecasted volume was 2000 per day, or 14,000 per week on average. The pants are sold in 3 colors and 20 sizes for a total of 60 SKU's.

The orders to the apparel manufacturer are split by store. A pair of pants requires 2 linear yards of the eight-ounce poly-cotton (50/50-blend) fabric and 28,000 yards are ordered each a week. The other materials required are assumed to be available when needed. There is a 3-week period that freezes the manufacturing plan, plus a week to manufacture the pants. The plant runs 2 shifts per day, 5 days a week. This results in a 4-week lead-time for Apparel. (Note: Not included in the model is the time of 30 minutes to sew a pair of these pants - so 7,000 hours or 200 FTE's @ 35 hours are required in the plant for sewing, plus people for spreading, cutting, inspection and packaging).

The textile plant runs 24 hours 6 days a week. Included in textile production are yarn formation, weaving, and fabric finishing. In this case, textile manufacturing takes 8 weeks.

One and a half pounds of the fiber can normally produce 2 yards of the fabric required for the pair of pants. The lead-time for Fiber is 4 weeks.

The retailer places orders for replenishment once a week and updates a rolling quarterly forecast once a month. The purchase orders detail the SKU's required for each store. (Note: A month has 4 weeks).

The apparel manufacturer receives the rolling quarterly forecast that is updated monthly and orders by store every week.

The textile manufacturer receives a weekly order for the fabric for 28,000 yards with quantities distributed among the three colors and a quarterly forecast that is updated monthly.

The fiber supplier receives weekly orders for 21,000 pounds of polyester fiber and cotton along with a quarterly forecast that is updated monthly.

6 Lead Time Results

This poly-cotton pants use case was run in the models for a period simulating 365 days after a warm-up of 365 days. Running the simulation for this period of time shows the time values for each activity in each sector and minimizes the affect of variability of the weekly activities.

The results from the lead-time simulation models proved that the DAMA Architecture did, in fact, have a positive impact on order lead-time between the partners. The two lead times that were calculated for each Partner were the planning / production lead-time and the order lead-time. The planning / production lead-time is defined as the time when a forecast enters Demand Planning until a completed production order is sent to the Finished Goods Warehouse. The order lead-time is defined as the time from order receipt until it is delivered to the customer. Model results are shown in Figure 6 and described as:

• The TISS-LT model had an order lead-time for the Apparel Partner of 14 days.
• The CISS-LT model showed a 52% decrease in this time to 6.8 days.
• There was an even greater impact on the planning / production lead-time that dropped from 81 to 37 days in the CISS-LT model. This resulted in a 55% decrease in planning / production lead-time within the Apparel Partner's operations without making any changes to the manufacturing system.
• Across all Partners in the supply chain, 16 days were removed from order lead-time, which resulted in a 49% decrease. Also, 137 days were removed from the planning / production cycle which represents a 49% decrease.

Figure 6. Lead Time Summary Results

7 Inventory Use Case
Additional information was added to the Lead-Time Use Case to calculate the quantitative effect of holding inventory. This was important to validate the model and assure accurate comparison between models.

The supply chain participants were assumed to have initiated a collaborative business planning agreement that is similar to a VICS CPFR® agreement. The business goals were to reduce lead-time, reduce inventory, and improve in-stock performance across the supply chain. The participants agreed to: define products and participants, share data, adhere to security procedures, define and measure goals, define volumes, and quality. In order to create a more accurate inventory use case, variation in actual orders was introduced in to the model. The actual orders begin at 14,000 per week and increase to 20,000 in the middle of the season. At the end of the selling cycle (one-year) the orders decrease to 6,000 units per week. This fluctuation in demand is included to add realistic variation to the supply chain. If these orders cannot be filled, the partial order will be sent and the remaining quantity is backordered and added to the next shipment. The same logic applies to production orders consuming raw materials.

The retailer updated the collaborative monthly sales forecast as needed with Point-of-Sale (POS) and changes to the promotional plan and had access to the order forecast to make changes as agreed to in the collaborative business plan contained in the Supply Chain Utility. In this use case there is a three-week frozen period where changes cannot be made without prior agreement. These changes beyond the agreed parameters were handled as exceptions.

The apparel manufacturer, textile manufacturer, and the fiber manufacturer had access to the collaborative monthly sales forecast, the weekly ship forecast and received weekly shipping instructions from the Supply Chain Utility. Each supply chain company updated the forecasts as required.

8 Inventory Results

As with the Lead-Time simulation models, the use case was run for a period of 365 days after a warm-up of 365 days. Running the simulation for this period of time shows the time values for each activity in each sector and minimizes the affect of variability of the weekly activities.

The results from the TISS and CISS models proved that the DAMA Architecture not only had a positive impact on lead-time between the partners, but also reduced the inventory levels that were required to maintain customer service levels. Since the TISS-LT and CISS-LT models were used as the building blocks for the comprehensive supply chain models, the lead-time results remained the same in the TISS and CISS models. The inventory levels that were tracked in the models are the inventory in the Retail Distribution Center and the finished goods and raw material levels for the Apparel, Textile, and Fiber Partners. Model results are shown below and illustrated in Figure 7:

• The TISS model had an average inventory level in the Retail DC of 98,690 units. This was decreased by 80% to 20,086 units in the CISS model.
• For the Apparel Partner, Finished Goods inventory was reduced by 65% from 111,615 units to 38,867 units using the DAMA Model and the Raw Material inventory was also reduced by 42%.
• For each partner there was a reduction in finished goods inventory from just over 6 weeks worth of units to approximately 2.5 weeks in order to provide the same level of customer service.

Figure 7. Inventory Summary Results

9 Architecture Validation
The results from the simulation models, which are summarized in Table 1, show conclusively that the DAMA Architecture is a valid supply chain alternative that results in improved responsiveness and reduced inventory in both finished goods and raw materials. The collaboration between partners allows for faster response to unforecasted fluctuations in consumer demand. For example, in the traditional supply chain it will require weeks or even months for the fiber partner to realize the demand has changed. In the DAMA model the supply chain utility will enable this information to be shared immediately.

Based on the results obtained from the TISS and CISS models, it can be concluded that the improvements associated with employing the DAMA supply chain architecture would be even greater if the variations in demand were increased. Due to the improved communication between partners, the impact of unpredictable consumer behavior would be controlled more efficiently. Currently, this problem is managed through increased finished goods inventories. Because this problem will never completely go away, the collaborative model manages this through the upfront business agreements and the supply chain utility.

Furthermore, the TISS model was based on the assumption that the forecast was not altered between partners so that the Textile and Fiber partners were only producing enough to satisfy the retail partner. In traditional supply chains, this forecast may be manipulated between partners and creates what has been classified as the "bull whip" effect that creates even greater inventories in the supply network. The "bull whip" can have the opposite effect and exaggerate downward trends in demand that result in raw material shortages and backorders. Eliminating the "bull whip" effect is just one advantage to collaborative planning. The VICS CPFR Committee has documented the fact that collaboration results in improved forecasting. This benefit was not modeled in the CISS model because it is difficult to quantify. These benefits of the DAMA Architecture are not included in the TISS and CISS models and would amplify the improvements in lead-time and responsiveness.

Table 1. CISS and TISS Model Outputs

10 Availability of CISS and TISS
The models reside at [TC]² and are available for supply chain analysis. For more information, contact Jim Lovejoy at 919-380-2184 or jlovejo@tc2.com or Tim Curran at 919-653-3513, or tcurran@tc2.com.

References
1. Textile Industry Supply Chain Business Model, DAMA Project, www.dama.tc2.com

2. DAMA Model for Collaboration, DAMA Project

Library Index | Home