Sei sulla pagina 1di 27

Next Generation Maintenance

Supply Chain Optimization Project Business Process Document Business Release 1

September 27, 2001

Purpose of this document The Business Requirements Document captures the As-Is Process, identifies improvement opportunities and defines the To-Be Process. This document has been prepared from the perspective of a Business User. Therefore technicalities and features of the software solution have been explained only to the extent that is required to explain a point. Details of the software solution from a configuration and technical viewpoint have been explained in other documents such as the Solution Architecture Specification, Integration Design Specifications. This document is a supporting tool to decide on the features and functionality needed in the solution. The document should be used at the start of the Design phase, and continually reviewed throughout the implementation as more information becomes available.

Revision History Date Aug 27, 2001 Aug 30, 2001 Sep 06, 2001 Sep 19, 2001 Sep 24, 2001 Sep 27, 2001 Author Wayne Fu R. Zimmerman Dustin Wu R. Zimmerman, Dustin Wu R. Zimmerman R. Zimmerman Updates First Edition First Draft Revision Revision Revision Edit

INDEX

1 Introduction ....................................................................................................................................4 2 Overview of Operations at SWA....................................................................................................4 3 Scope and purpose of Business Release 1 (BR1)............................................................................4 4 The As-Is Process............................................................................................................................6 4.1 FORECASTING FOR EXPENDABLES:...........................................................................................6 4.2 INVENTORY MANAGEMENT FOR ROTABLES:.............................................................................7 5 Improvement Opportunities / Benefit of New System....................................................................8 6 Forecasting Matrix........................................................................................................................10 7 Demand Planning Process.............................................................................................................12 7.1 EXPENDABLES PLANNING PROCESS........................................................................................13 7.2 ROTABLE PLANNING PROCESS...............................................................................................14 7.3 ACO PLANNING PROCESS.......................................................................................................14 7.4 NEW PART INTRODUCTION.....................................................................................................14 7.5 NEW LOCATIONS.....................................................................................................................14 7.6 REALIGNMENT........................................................................................................................15 7.7 PHASE OUT PART / LOCATION................................................................................................15 7.8 RETURNS FORECAST...............................................................................................................16 The returns forecast is a post process that is completed after the consolidated forecast is generated; Service Parts Planner will handle this forecast. The returns forecast is designed to send items back to the repair center with the yield rate applied. These items will come back to the supply chain after the predefined repair lead-time. The process of returning parts back to the supply chain is mainly done in replenishment planning. Therefore, in BR-1 we will output information of part number, quantity and dates for which the repair items are sent to the repair center. See Appendix H for the process flow chart.........................................................................16 8 Exception management.................................................................................................................16 Appendix A: Glossary:.................................................................................................................17

ii

iii

1 Introduction
Southwest Airlines (SWA) has formulated a Next Generation Maintenance Automation Program, which aims to improve the effectiveness and efficiency of key processes in the Maintenance and Engineering organization. One of the first initiatives of the Next Generation Maintenance Automation Program is to improve the spare parts planning process using a sophisticated supply chain planning and optimization tool. The project focused on the spare parts planning process is the Supply Chain Optimization Project. Based on the requirements jointly specified by the Systems and Maintenance & Engineering Departments, i2s TradeMatrix Service Asset Management suite was chosen as the solution. i2s TradeMatrix solution includes the Demand Planner (DP), Service Parts Planner (SPP), and the Service Budget Optimizer (SBO) applications. The TradeMatrix (TM) suite being implemented at SWA will provide decision support capabilities that will allow SWA to have the right quantity of the right part at the right place at the right time. It will achieve this by optimizing the tradeoffs between service levels, inventory levels, and inventory budgets for Maintenance & Engineering spare parts.

2 Overview of Operations at SWA


Southwest Airlines Materials Department manages one of the lowest inventory supplies in the industry, turning its $240 million rotable inventory approximately four times per year, the highest turn rate in the industry. There are approximately 120,000 part location combinations in the SWA supply chain. Inventory is broadly classified into Rotables and Expendables. Rotable inventory is valued at $ 240 million with over 1,800 separate stock item numbers and approximately 350,000 individual serialized units. Southwest Airlines expendable inventory is valued at $50 million with over 41,000 separate stock item numbers and contains over 3 million individual units. Currently the Materials Department is organized into three distinct groups, Repairs, Stores, and Purchasing. Limited material planning is done manually by Buyers or Inventory Analysts assigned to Purchasing. Demand forecasting is accomplished by the Maxi-Merlin system known as Wizard. Wizard tracks about 45,000 unique items that consist of aircraft parts, flight simulator parts, aircraft maintenance tools and test equipment, and non-aircraft parts. It is capable of developing a single period aggregated forecast which it uses to plot basic reorder points for each part location. SWA currently has three primary maintenance/parts stocking facilities, which are located in Dallas, Houston, and Phoenix. There are also 13 line maintenance facilities and over 60 ramp and provisioning locations where parts are allocated. Inventory is also furnished to some third party vendors that perform contract maintenance for Southwest Airlines. As the magnitude and complexity of the maintenance operations increase, SWA requires the ability to optimize its expendable and rotable inventories to ensure efficient use of available resources. The selected solution will work well with the existing inventory transaction system (Wizard) as well as any system selected for use in the future.

3 Scope and purpose of Business Release 1 (BR1)


This Supply Chain Optimization Project is divided into three Business Releases (BR). BR1 focuses on modeling the demand patterns of rotables and expendables. Appropriate statistical models that best represent the demand patterns of the various products will be modeled. Modeling will be done using both the Demand Planner (DP) and Service Parts Planner (SPP) components of i2s TradeMatrix suite. Models

will include trend and seasonality; both are based purely on historical demand. Causal and event driven forecasts will be based on Maintenance Checks and Flight Schedules. As the current demand planning processes in SWA are very basic, the models that are developed initially will be kept as simple as possible. The purpose of BR1 is to generate the consolidated forecast and mean absolute deviation (MAD) for both rotables and expendables. The demand planning process will include a combination of demand forecasting using the i2 products as well as experience-based manual forecast adjustments of the statistical forecasts. The number of item location combinations to be reviewed will be managed by exception reports. The scope for BR1 includes all rotables and expendables at all locations (Store, Line, Provisioning and Ramp). The scope excludes planning for Ground Equipment. The functionality at the end of BR1 will include:

Demand planning carried out in DP and SPP. Demand plans from DP will be sent as input to SPP. SPP will generate a returns forecast to project the quantity of repairable parts returning to the unserviceable pipeline (repair loop). Demand plans from SPP will be sent to Wizard through appropriate interfaces. Wizard will calculate the reorder point and carry out further processing as per current logic. Essentially, Wizards forecasting will be replaced by the demand plan from the i2 solutions. Everything else in Wizard will continue to operate the same way as it does currently Repairs forecast will be used for review and information purposes only. Currently, there are no systems in SWA that can utilize this output. During BR2, the repairs forecast will be an input for carrying out the inventory and replenishment planning processes in SPP Exception reports of item-location combinations, where the forecasts need manual intervention, prior to being released.

The planning horizon for demand forecasting will be 12 months for DP and SPP. The original output from DP and SPP will be a time-phased forecasted demand plan. This aligns very well with the inventory planning and replenishment planning processes envisioned for BR2. In order to interface with Wizard at the end of BR1, the i2 solution will be configured to generate the average demand over lead-time. A high level landscape at the end of BR1 is shown below:

Remedy

OpenView

Maintenance Schedule A B I N I T I O

Scope of BR 1
Demand Planner (forecast for 12 months)

Service Parts Planner (Service Level Planning)

Flight Schedule

Service Parts Planner (Base Replenishment Planning)

Corporate DataWarehouse

Service Parts Planner (Average Demand over Lead-time)

Wizard

Service Budget Optimizer

Scheduling Tool

4 The As-Is Process


The goal of this project is to have the right part at the right location at the right time. This will be achieved by optimizing the tradeoffs between service levels and inventory levels. Reorder point (ROP) is the key to managing inventory at the most optimized level while maintaining the service level at the predefined target. Wizard calculates a reorder point for each part location at the end of every transaction. Additionally, Wizard recalculates all part location reorder points at the end of each month. If the calculated reorder point is above the location balance on hand a stock action request will be generated. Looking into the formula of ROP calculation, we can easily find that forecasted demand is the real driver of the business performance measure, service level and inventory level. The forecasted demand affects the ROP dramatically. Currently there are two main processes for demand planning of parts in SWA one for forecasting expendables using Wizard and the other to plan for rotables. These are explained below.

4.1

Forecasting for expendables:

In Wizard, forecasting is done for expendables only (Tracking Level 0 & 1). The formula used by Wizard to calculate the forecast is:

F1 = F2 + [A * (U1 F2)] Where: F1 = Current Forecast F2 = Old Forecast A = Alpha U1 = Previous Period Use This computation is executed for the previous 12 periods stepping backward from the oldest period. The resulting forecast calculation is used as part of the reorder point calculation. Materials requirements for specific maintenance events (Quarter, Half and Full D Checks, Aircraft Change Orders) are planned by Purchasing Agents or Inventory Analysts. This is a manual process where-by the responsible individual calculates from the maintenance plan, the materials required to support a specific maintenance event. The person doing the planning maintains the forecast for required materials on a spreadsheet. Periodically the requirements for the planned maintenance event will be compared to material on hand to ensure adequate quantities are available to support the maintenance event.

4.2

Inventory management for rotables:

Wizard does not forecast or generate a reorder point for any rotable (Tracking Level 2 & 3). The system does track by serial number, the location and status (serviceable/unserviceable) for each rotable component owned. Total ownership, serviceable and unserviceable pipeline, and history of each rotable are also tracked in Wizard in MSRH. This is a stand alone screen in Wizard and is not integrated to any other screen(s) in Wizard. As such any changes to MSRH do not affect any other functions in Wizard. Determination of optimal ownership levels for spares is done manually. Usually a request by someone to the Inventory Management Group is the initial input that starts an analysis of spare ownership. Once the target stock item is identified for review, an Inventory Analyst will be tasked to conduct an analysis of the stock item to determine optimal ownership. The Analysts will review current ownership, usage, and repair vendor lead - time to determine the optimal level of spare ownership. The following procedure is used to help determine this level: The data points required to determine an optimal rotable spare level are average monthly usage, average vendor lead time, probability and standard deviation. In most cases the required information can be obtained through Wizard and the mathematics model for standard deviation in Excel. The components that make up the Optimal Spare Model include unserviceable pipeline, safety stock and demand during leadtime requirements. The following formula is used to determine the desired ownership level: Am + Av + P (Sd) = Os Am = Av = P = Sd = Os = Average Monthly Usage (Greater of Use or Allocations) Required quantity to meet demand through Average VLT Area under normal distribution curve (Probability/Desired Spare Level) Standard Deviation Optimal Spare Level

Step by step methodology: From Wizard (CCFRS) determine the Average Monthly Usage by adding the past 12 months use in the Scheduled and Unscheduled columns, then divide by 12 to determine the average. The greater of this number or required allocations will be used as the Average Monthly Usage.

From Wizard (MSPD) determine the Average Vendor Lead - Time by adding the lead times for repair orders and dividing that number by the number of events. Consideration should be given for discarding excessively long or short lead-times, as they tend to skew the results. The formula for Av is: Av = Lt(Am 30) Lt = 30 = Average Vendor Lead-Time Days per Month

For instance, if there are 5 Repair Orders that have been closed and the total vendor lead-time for all 5 orders was 180 days, you will divide 180 by 5. This will give you an average of 36 days (Lt). It is necessary to determine how many parts are required for a 36-day period. Plug the numbers into the above equation to determine Av. This number will be used as the required quantity to meet demand through lead-time. Probability of a normal distribution can be ascertained from any Standard Normal Probability Distribution Chart. In most instances we will be applying a 95% probability, the factor would be 1.65. The more common probability factors are listed below. Only use the standard deviation for the monthly use this in effect will become the safety stock input. Factor .84 1.28 1.29 1.65 2.33 Probability .79955 .89973 .90147 .95053 .99010

Standard Deviation for a Normal Distribution can be determined by using the Standard Distribution Model in Excel or manually determine the standard deviation. The preferred method is the Excel model.

Once all data points are determined, the numbers are used in the equation and solved for O s. Normal rules for rounding apply, but no rounding is done until the equation is solved for O s.

5 Improvement Opportunities / Benefit of New System


Significant improvement opportunities exist in the current inventory planning system at SWA. Introduction of a formalized material planning process will improve material availability at the location it is required. The new planning system is constructed of three well-defined modules, forecasting, inventory planning and replenishment planning. The first BR will focus on the forecasting process, which will improve the visibility of the demand plan one year into the future. The demand forecast will be applied to parts at all tracking levels. On the contrary, the current system provides only the projection for the upcoming month for expendables only. Furthermore, the new system will provide SWA with more statistical forecasting algorithms to fit different patterns of demand, this is currently done with a single formula in Wizard. The demand planning process using the capabilities of i2s TradeMatrix DP and SPP present significant improvement opportunities with respect to current SWA processes. These are listed below:

Features of As-Is Processes

Features of To-Be Processes

Demand Planning for Rotables is manually triggered and managed Forecast is done with one single algorithm - moving average.

Formal Demand Planning process for Rotables using statistical algorithms Demand Planner has multiple forecasting algorithms build in, including multiple regression, triple plus, moving average, modified crostons, periodic, etc.The software also has a Pick Best feature to automatically select the algorithm that is best suited for every item-location combination The new demand planning process will cover parts at all tracking level, including consumables, expendables, repairable expendables and most important of all, rotables. Demand planner will project the demand plan 12 months into the future Modified Crostons algorithm is designed specifically for sparse demand. The new demand planning process will forecast base on the usage history, maintenance events and flight schedule and MTBF. The system will pick the best forecast based on the MAPE. Planners can also reference different type of forecast for finetuning the demand plan. The new system will be handled by planning group. This will put the purchasing agent and inventory analyst back to their focus and increase the capability of inventory management at SWA. Review this for intent and readability Dual Mode Demand Planning combines sophisticated statistical algorithms with experiential intervention, as appropriate Forecasting will also include event based forecasting such as Maintenance Schedules and causal factors such as MTBF based on Flight Schedules apart from historical consumption / usage. This will be a change in perspective from rear-view to forward looking Returns forecast will be done using the functionality for the same in Service Parts Planner (SPP) Forecasting for new parts, new locations or parts for a new series of aircraft will use a combination of the historical substitution feature of the i2 product and experiential honing of the same Phase out of parts / locations can be handled in the system (Venky to check veracity of this statement)

Forecast is done for expendables.

Demand projection is done for one month Moving average does not handle the sparse demand pattern very well. The usage history is the only source of demand forecasting.

Existing demand planning process is distracting purchasing agent and inventory analyst from their area of concern. Review this for intent and readability Binary mode Demand Planning is either a fully automated or fully manual process Only historical consumption / usage data is used for forecasting ----

Returns (for repairs) are not forecast Forecasting for new parts, new locations or parts for a new series of aircraft was experience based

Phase out of parts / locations handled manually

Forecasting is a black box process

Forecasting will be a process driven, and controlled by the planners and customized for their use

6 Forecasting Matrix
SWAs Demand Planning process will be based on three types of inputs, namely: Historical data based forecasting Maintenance Schedule based forecasting Flight Schedule based forecasting

These are explained below:

Demand Planning Approach Forecast based on history Demand plan based on maintenance schedules Demand Plan based on flight schedules

Description Historical usage available from the Wizard transaction system is used to calculate forecasts. Maintenance schedules and the type / location of the maintenance are available. The parts to be replaced are determined using the Bill of Material (BOM) for the maintenance and the probability of usage of each of those parts. Flight schedules and reliability statistics are available. Both should be used to forecast usage at SWA stations.

Ideally, all forecasts should be a combination of these three forecasting approaches. However, due to constraints imposed by current business processes, the applicability of a particular approach will depend on a combination of material type and location type at SWA. The applicability of a demand planning approach to a combination of material type and location type is shown in the table below and explained subsequently.

Material Type

Location Type History Y

Type of input Flight hours X Maintenance Y

Expendables Store

10

Line Ramp Provisioning Store Rotables Line Ramp Provisioning


Legend: X = Not Applicable, Y=Applicable

X X X Y Y Y Y

X X X X Y Y Y

X X X Y X X X

Historical data based forecasting: For expendables, this will be applicable only at the Stores level. This will not be applicable at Lines and Ramps due to the current inventory management process for expendables at these locations. Currently, expendable inventory is tracked only at the Store level. Visibility of expendable material is lost in the system the moment it is issued from a Store. This business process is not expected to change. Flight hours based forecasting: A forecast for rotables will be generated for the Line, Ramp, and Provisioning levels based on fight schedule and component reliability data (MTBF). This will not be applicable at the Store level. Expendables will be excluded due to the current inventory management process for expendables at these locations. Currently, expendable inventory is tracked only at the Store level. Visibility of expendable material is lost in the system the moment it is issued from a Store. This business process is not expected to change. Planned Maintenance based forecasting: Using planned maintenance events, rotable and expendables will be forecast at the Store level. This will not be applicable at the Line level; scheduled maintenance events are not planned at that level. Implications of the Forecasting Matrix: In those instances (material type + locations type) all three forecasting approaches are applicable, the Demand Planner of i2 will generate three different forecasts. Subsequently using the Pick Best feature, it will select the forecast which is most accurate. Where, only two of the three forecasting approaches are applicable, Demand Planner will generate only two forecasts. Subsequently, the Pick Best feature will select the better of the two forecasts. The reduced applicability of the forecasting approaches for a few material/location type combinations may give the impression of a compromise in the ability to pick the best-forecast possible. However, a forecasting approach needs to meet the needs of the business and not the other way around. Given a certain business need at a specific location for a specific material, the forecasting matrix provides the best possible approach to forecasting. The forecasting matrix will however need to be reviewed and updated when: New systems bring about changes in business processes. For example, new systems may introduce material tracking for expendables at Lines.

11

Business processes change due to changing needs. Certain types of maintenance are introduced at the Line Stations.

7 Demand Planning Process


During the Design Phase, the Team determined that a single process would not be sufficient to take care of various SWA needs. Therefore, sets of processes were designed for SWA. Some of these processes cater to the core Demand Planning requirements and need to be run on a regular basis. The other processes are situational and come into effect only when certain non-routine business needs trigger them. The Demand planning processes designed for SWA are as follows: 1. Demand Planning for Expendables/ACO items 2. Demand Planning for Rotables/ACO items 3. Demand Planning for new parts/locations 4. Demand Planning for parts/locations being phased out 5. Demand Planning for products/geographies during realignment 6. Returns forecast 7.1 Demand Planning for Expendables Scope of the process: The scope of demand planning for expendables includes all TL 0 and 1 expendable materials at the Store level. All TL 0 and 1 expendable materials at other than Store locations are out of scope. Business Requirements/Expectations: A usable forecast based on historical usage and planned maintenance activity is the expected output of demand planning. It is expected that this process will improve material availability at the Store locations. Key Considerations: Required historical usage data is not available for forecasting at Line, Ramp, or Provisioning levels. Without the required usage data it is not possible to forecast at these levels. Aircraft Change Orders and Component Change Orders are not planned, therefore material requirements to support their execution cannot be forecast. Salient Features of the Process: DP will generate a forecast based on historical usage and demand patterns. Materials Planners will review exceptions to the forecast and make necessary adjustments. The forecast will be finalized and used as an input to Wizard for generation of a ROP for each Store location.

Key Data/Integration Requirements: It will be necessary to ensure only accurate historical data is used to create the forecast. Where historical records indicate incorrect demand, the record will have to be corrected before it is used. Demand will be credited to where it should have been supported not where it was actually supported. This will ensure we capture the demand stream for how we want the distribution network to function and not just the exceptions.

12

Until usable fractional Bills of Material can be created and administered it will necessary to use only those Bills of Material that will have 100% usage rate associated to them. Improvements: The net improvement expected from demand planning is an increase in material availability at the location it is required. This will be achieved through improvements in forecasting algorithms, Pick Best feature, and projection of the demand plan for 12 months into the future. Gaps and recommended resolution: Forecasting at the Line, Ramp, and Provisioning Levels will not be able to be performed until required data are available. Specifically it will be necessary to change the business process of not issuing expendable material at these levels. Issue, Credit, and Transfer (ICT) transactions will have to be captured at the Line, Ramp, and Provisioning Levels before a forecast at those levels can be generated. Forecasting for Aircraft Change Orders and Component Change Orders will not be able to be performed until required data are available. The data required for generating a forecast for ACOs/CCOs are not currently available. The specific data that would be required are Bills of Material for each ACO/CCO and a plan by date of when the activity is to occur.

7.1

Expendables planning process

Generating an initial forecast: (Systems) The demand planning process for expendables starts with the preparation of data from the inventory transaction system. Once the data is transformed into the required format and loaded into DP, an initial forecast based on historical usage and demand patterns will be computed. When the initial forecast is complete, the DP engine will stop. Users can then access this information with the DP user interface. Reviewing the initial forecast and inputting forecast adjustments: (Planning) Exception reporting is the most useful tool available to determine the item location combinations requiring review. For item location combinations that fall within the exception criteria the Planner will be required to make adjustments as necessary. Prior to making an adjustment to the forecast, the Planner will need to review usage history, analyze consumption patterns, and gather any additional information to be used as input to the demand planning system. Once an adjustment is completed the new forecast will be copied to the final forecast, and the Planner may move onto the next exception. With the exception of parts used to support Aircraft Change Orders (ACO) and Component Change Orders (CCO), item location combinations that do not fall within the exception criteria require manual intervention. Parts used to support Aircraft Change Orders (ACO) and Component Change Orders (CCO) must go through the same process as those items identified as exceptions. Output consolidated forecast: (systems) Once adjustments to the initial forecast are completed, the final consolidated forecast will be exported from DP and to SPP. Once loaded to SPP, the post-process designed for the integration between Trade Matrix and Wizard will be executed. See Appendix A for process flow chart. Note:

For expendables at Line Stations, SWA does not have inventory visibility. It is currently a manually triggered pull mechanism, not a push system. Actual usage is not captured for expendables at Line Stations. Therefore, for expendables, forecasting can only be done at the Store location Issues of expendable items to a Line from a Store other than the primary supporting Store will be corrected in the usage history at the time the data is extracted.

13

Transfers between Stores are not counted as usage. The transaction for usage at a store is the issue. Issues are documented in the transaction system. Forecasting at Store locations will include dependent demand from Lines and independent demand from itself.

7.2

Rotable Planning Process

The planning process for rotables is basically the same as the process for expendables. The major exception is that all category A items will be reviewed by the Planner regardless of their exception status. The major steps in forecasting rotable material include generating an initial forecast, reviewing the initial forecast, inputting forecast adjustments, and finally generating a consolidated final forecast. Since the rotables are generally high value, it is important to closely monitor their usage at the demand planning stage. Planners will review all category A rotable items regardless of their exception status. See Appendix B for process flow chart.

7.3

ACO planning process

The DP planning engine will generate an initial forecast to include parts used for ACOs. The required information includes a schedule for ACOs and the parts list (BOM) needed for completing each ACO. The planning process for ACOs are the same as the processes discussed in previous sections, depend on type of part, rotable or expendable. See Appendix A for process flow chart. Note: Current ACO schedules are not fully planned in advance, it is a daily or weekly planning process. Therefore, the ACO related demand forecasting would not be included in the planning system at the first business release.

7.4

New Part Introduction

The determination to plan for a new part is the responsibility of the Purchasing Department. The Purchasing Agent will make the determination if a new part should be planned for at the time it is set-up in Wizard. If the part is to be planned, this new item will be included in the New Parts Report for review. Planners will decide if a manual forecast must be input or not. See Appendix C for process flow chart. Note: Initial Data load - A new Item list will be generated for last six months to enable the first planning run. After the initial data load, a New Parts Report will only have items introduced since the last planning cycle. The Planner will be supplied with a report of new items. The Planner will decide whether a given new item is to be manually forecast or to use the initial forecast from DP. New items will have a separate data load so DP can capture the new item in the system. Six months of history, after the first non-zero usage data, will be the minimum requirements to begin using historical usage to forecast.

7.5

New locations

New Store:

14

Chicago Midway Airport (MDW) is scheduled to become a Store in the near future. This type of event happens approximately once in 5 or more years. See Appendix D for flow chart. Preprocess for generating initial forecast The process of forecasting for a new Store location starts with a decision from Maintenance Management to create a new Store location. Once the decision to create a new Store is finalized, Maintenance Management will communicate with the Material Planning Group the details of the decision. Specific input from Maintenance Management will include if there are any shops to be supported by the new Store, and if so, is it a new shop or an existing shop being moved from another location? The forecasting process will be slightly different depending on the answer to these questions. 1. No Shop included in the new Store: The Maintenance Schedule will be used to forecast parts requirements at the new Store location. Use of the Maintenance Schedule to forecast requirements will mitigate the need to have usage history available until such time as an adequate level of history is established. 2. Shop included in the new store: If the new Store location is to support a shop, historical usage will also be used to generate the demand forecast. The process for generating usage history will be predicated on whether the Shop is new or being moved from another location. If a Shop is to be moved from an existing location to a new location the usage history from the old location shall be used to generate the forecast for the new location. Additionally, the historical usage at the old location must be adjusted to remove the usage attributed to the Shop that is being moved. If a new shop is being created at the new Store location the Planner must either use historical substitution or copy the required information from a similar type of shop. Reviewing and output forecast: After the above process is complete, all the data required will be in place to run the forecast. DP will generate the initial forecast the same as usual. After the initial forecast is generated, all remaining steps are same as the process for expendable and rotable forecasting. New Line: When a new Line location is introduced into the system, the Planner will generate a list of line items and quantities to be allocated to the new Line location. This list will be the basis for manually adjusting the forecast. See Appendix E for process flow chart.

7.6

Realignment

Once the Materials Management Department has determined the need for geography or product realignment, the Systems Department will create a Wizard transfer file for usage history. Any realignment will be reflected in the respective hierarchy tables, which will reflect the new parent-child relationship. After the realignment file is created and put in place, all database files will be realigned in Demand Planner during the monthly planning process. See Appendix F for the process flow chart.

7.7

Phase Out Part / Location

This process is used when it is determined that a part or location is to be phased out. The Inventory Management Group is responsible for communicating all information regarding the phase out of parts and

15

locations to the Systems Department and Materials Planning Group. Normally information regarding location(s) to be phased out will originate the Schedule Planning Department or Maintenance Management. When the Systems Department receives the information, they will create a file with deletion records using the data extraction program. Once the deleted records file is loaded to DP the planning process is the same as for expendable and rotable materials. Note:

If deletion is not indicated - zero usage assumed and forecast will be populated Item will be deleted from the Trade Matrix system once the file is loaded If the item is not specifically deleted, the forecast would still be made with zeros Similar to Item phase-out. The item-location combination will be used to delete locations Once a location or item is phased out, the record in DP will be deleted.

7.8

Returns forecast
The returns forecast is a post process that is completed after the consolidated forecast is generated; Service Parts Planner will handle this forecast. The returns forecast is designed to send items back to the repair center with the yield rate applied. These items will come back to the supply chain after the predefined repair lead-time. The process of returning parts back to the supply chain is mainly done in replenishment planning. Therefore, in BR-1 we will output information of part number, quantity and dates for which the repair items are sent to the repair center. See Appendix H for the process flow chart.

8 Exception management
Exception reporting is commonly used to manage the review process in the service industry. It is also an efficient approach to handle high volume planning results. Use of exception reporting will assist in setting the level of work to be done as well as help manage the screening process workload. With the introduction of exceptions, we can filter out the item location combinations that require manual intervention. These item location exceptions will become the work queue for each of the planner. The three exceptions that will be used in DP are: 2. Forecast too high/low 3. Forecast error too high/low 4. Historical Outlier ( 3 sigma) Exception for Expendables:

Forecast too high/low (compared with the usage of the latest month). +/- 20% comparing to last month average. (Item A and B) +/- 40% comparing to last month average. (Item C and D) Historical outlier report (3 sigma difference compared to 12 month forecast average)

Exceptions for Rotables:


Forecast too high/low (5% for all categories) Historical outlier report (3 sigma difference compared to 12 month forecast average) Forecast error too high (MAPE > 10 %)

16

Appendix A: Glossary:
SWA Southwest Airline DP Demand Planner, i2 demand management application. SPP Service Part Planner, i2 application developed for service industry. It is composed of forecasting, inventory planning and replenishment planning module. BR Business Release MAPE Mean Absolute Percentage Error Net Change A processing approach in which the material requirements are continually retained in the computer. Whenever a change is needed in requirements a partial explosion and netting is made for only those parts affected by the change. APICS Dictionary, Eighth Edition, 1995 ACO Aircraft Change Order CCO Component Change Order Pipeline Stock (Spares Pipeline) Inventory to fill the transportation network and the distribution system including the flow through intermediate stocking points. The flow time through the pipeline has a major effect on the amount of inventory required in the pipeline. Time factors involve order transmission, order processing, shipping, transportation, receiving, stocking, etc. APICS Dictionary, Eighth Edition, 1995 Regeneration A processing approach where the forecasted plan is totally re-exploded to maintain valid priorities. New requirements and planned orders are completely recalculated or regenerated at that time. APICS Dictionary, Eighth Edition, 1995 Repairable Expendable Parts Expendable parts that can be repaired and returned to service. These parts are usually a sub-classification of expendable parts and are tracked separately. Rotable Parts Component parts designed to be repaired and reused. Components such as aircraft wheels, avionics assemblies, and built-up engines are considered rotable parts. Scrap Rate A percentage factor used to increase gross requirements to account for anticipated loss during the repair process of a particular product. APICS Dictionary, Eighth Edition, 1995 Seasonality A repetitive pattern of demand from year to year (or other repeating time interval) with some periods considerably higher than others. APICS Dictionary, Eighth Edition, 1995 Service Level A desired measure of satisfying demand through inventory in time to satisfy the customers request. APICS Dictionary, Eighth Edition, 1995 Stock Keeping Unit (SKU) An item at a particular geographic location. APICS Dictionary, Eighth Edition, 1995. A specific item in an inventory. Quantitative Approaches to Mgt. 1986 Stocking Location A Stores facility that is established to maintain a quantity of material on hand to fill the need of allocated locations. Yield The ratio of usable output from a process to its input. APICS Dictionary, Eighth Edition, 1995

17

Signoff Sheet

____________________________________ Jeffrey Hillis Senior Manager, Southwest Airlines ____________________________________ Rusty Puckett Project Manager, Southwest Airlines ____________________________________ David Boyer Manager, Southwest Airlines ____________________________________ Deepak Purohit Project Manager, i2 Technologies ____________________________________ Chung-Hsien Wu Application Engineer, i2 Technologies

Date: ___________________________

Date: ____________________________

Date: ____________________________

Date: ____________________________

Date: ____________________________

18

E x p e n d a b le / A C O P la n n in g P r o c e s s - D P 2
E x tra c t U s a g e H is to ry fr o m W iz a r d T ra n s fo rm D a ta w it h A b i N it io E x tra c t M X S c h e d u le f r o m LAN C o r r e c t H is to r y fo r N e x t P la n n in g R u n No L o a d D a ta w ith D P T M P A I A d a p te r

S y s te m G e n e ra te d In it ia l F o r e c a s t & C o p y P ic K b e s t in t o F in a l F c s t R o w a s D e fa u lt

S y s te m s

A u to m a te d D ay 0

P la n n in g

In it ia l F o r e c a s t

E x c e p t io n ?

R e v ie w F o r e c a s t H is to ry

H is to ry C o rre c t? Yes

Day 1 - 3

E x c e p tio n R e p o r t -F o re c a s t to o H ig h -O u tlie r R e p o r t 5 M in p e r No

No

A C O Ite m

Yes

No

A d ju s t F o re c a s t

C opy U ser Fcst in to F in a l F c s t

A n a ly z e a n d C o rre c t F o re c a s t

No

F o re c a s t A c c u ra te

F in a l, M o v e to N e x t E x c e p t io n P la n n in g Yes

S y s te m s

O u tp u F c s t to T M P A I F o rm a t

Load To SPP

L o a d T o W iz a r d a n d D a ta W a re h o u s e

E v e n in g o f Day 3

19

R o ta b le P la n n in g P r o c e s s - D P 1 1
E x tra c t U s a g e H is t o r y f r o m W iz a r d T ra n s fo rm D a ta w it h A b i N itio E x tra c t M X S c h e d u le / F lig h t S c h e d u le / M T B F fro m L A N L o a d D a ta w ith D P T M P A I A d a p te r

S y s te m G e n e ra te d In it ia l F o r e c a s t & C o p y P ic K b e s t in t o F in a l F c s t R o w a s D e fa u lt

S y s te m s

C o r r e c t H is t o r y fo r N e x t P la n n in g R u n

N o

A u to m a te d D ay 0

P la n n in g

In itia l F o r e c a s t

E x c e p t io n ?

R e v ie w F o r e c a s t H is t o r y

N o

H is to r y C o rre c t?

Day 1 - 3

N o

N o

" A " Ite m o r A C O Ite m

Yes

N o

A d ju s t F o r e c a s t

C opy U ser Fcst in to F in a l F c s t

A n a ly z e a n d C o rre c t F o re c a s t

N o

F o re c a s t A c c u ra te

F in a l, M o v e t o N e x t E x c e p tio n P la n n in g Yes

S y s te m s

O u tp u F c s t to T M P A I F o rm a t

Load To SPP

L o a d T o W iz a r d a n d D a ta W a re h o u s e

E v e n in g o f Day 3

20

N e w P a r t In tr o d u c tio n - D P 5

Day 0

In tro d u c e N e w P a rt

Ite m to b e P la n n e d

N o

N o In p u t to D e m a n d P la n n e r (D P )

P u r c h a s in g

Yes R e p o rt L is t o f N e w It e m s t o b e P la n n e d

Yes

M anual F o re c a s t

N o

P la n B a s e d o n M a n u a l F o re c a s t D a y 1 -3

O u tp u t to T M A P I F o rm a t

S P P d ll A c t iv a t e d w h e n th e r e is P o s it iv e C o n s u m p tio n H is to r y

P la n n in g

L o a d to S P P

21

N e w S to r e L o c a tio n - D P 6

N e w S to r e
D ay O

Shop In v o lv e d

Yes

M o v e /N e w

M ove M a in te n a n c e

N o

New

D ay 1

M a in te n a n c e F o re c a s t

C o p y L is t in t o N e w S to re * E x tra c t In fo . fro m W iz a r d to f e e d fo re c a s t fo r S h o p U sage

H is to r ic a l F o re c a s t

S u b tr a c t H is to r y fr o m o ld S to r e a n d In s e r t in to N e w S to r e

P la n n in g

D a y 1 -3

F in a liz e F o re c a s t a n d M o v e to N e x t Ite m

A d ju s t F o re c a s t

R e v ie w F o re c a s t o f N e w S to re

In itia l F o r e c a s t

S y s te m s O u tp u t to T M A P I F o rm a t D ay 3 L o a d to S P P

22

N e w L in e L o c a tio n - D P 4

D ay 0

D e t e r m in e N e w L in e L o c a t io n

M a in te n a n c e M anagem ent

G e n e r a te L is t fr o m D a ta W a re h o u s e o f L in e Ite m s

R u n F o re c a s t

M a n u a lly A d ju s t

P la n n in g Day 1 - 3

Day 3 S y s te m s L o a d to S P P O u tp u t to T M A P I F o rm a t

23

G e o g r a p h y / P r o d u c t R e a lig n m e n t - D P 3

D e te r m in e G e o g ra p h y / P ro d u c t R e a lig n m e n t

Day 0

M a te r ia ls M a n a g e m e n t

Day 0

C r e a te W iz a r d T r a n s f e r F ile f o r U s a g e H is t o r y

R e a lig n m e n t P ro c e s s : R e a lig n m e n t F ile G e n e ra te d

A ll D a t a b a s e F ile s a r e R e a lig n e d in D e m a n d P la n n e r d u r in g M o n th ly P la n n in g R u n

O u tp u t to T M A P I F o rm a t

S y s te m s

L o a d to S P P

24

P h a s e O u t L o c a tio n / P a r t P r o c e s s - D P 7

Day 0

D e te r m in e P a r t t o be P hased O ut

In v e n to r y M anagem ent

Day 0

D e te r m in e L o c a t io n t o b e P hased O ut

S c h e d u le P la n n in g o r M a in te n a n c e M anagem ent

Day 0

G e n e r a te F ile w ith D e le t io n R e c o r d s

L o a d F ile t o D e m a n d P la n n e r

D e m a n d P la n n e r D e le te s R e c o r d N o F c s t fo r n o n e x is t a n t it e m lo c a tio n

S y s te m s

L o a d to S P P

O u tp u t to T M A P I F o rm a t

25

R e tu rn F o re c a s t P ro c e s s - D P 7

Day 3

O u p u t fro m D em and P la n n e r In itia l L o a d o f S e r v ic e P a r ts P la n n e r ( S P P ) E x tra c t fro m W iz a r d : R e v e rs e


B O D , Ite m M a s te r, B O D (L T ) L o a d to W iz a r d and D a ta w a re h o u s e

R un P a rts R Fo

S e r v ic e P la n n e r e tu rn re c a s t

O u tp u t D e m a n d o v e r L e a d T im e p e r Ite m Q u a n t it y f o r R e v ie w

26