Manual Logware

SELECTED COMPUTER PROGRAMS

FOR LOGISTICS/SUPPLY CHAIN PLANNING

Version 5.0

Ronald H. Ballou

Weatherhead School of Management

Case Western Reserve University

(C) Copyright 1992-2004 Ronald H. Ballou All rights reserved

SELECTED COMPUTER PROGRAMS

FOR LOGISTICS/SUPPY CHAIN PLANNING

LOGWARE is a collection of selected software programs that is useful for analyzing a variety of logistics/supply chain problems and case studies. It contains the following modules.

Module Page

FORECAST Forecasts time series data by means of exponential smoothing and time series decomposition methods

5

ROUTE Determines the shortest path through a network of routes

9

ROUTESEQ Determines the best sequence to visit stops on a route 13

ROUTER Develops routes and schedules for multiple trucks serving multiple stops

15

INPOL Finds optimal inventory ordering policies based on economic order quantity principles

24

COG Finds the location of a single facility by the exact center-of-gravity method

33

MULTICOG Locates a selected number of facilities by the exact center-of-gravity method

37

PMED Locates a selected number of facilities by the P-median method

41

WARELOCA A warehouse location program for specifically analyzing the Usemore Soap Company case study

45

LAYOUT Positions products in warehouses and other facilities 47

MILES Computes approximate distance between two points using latitude-longitude or linear-grid coordinate points

49

TRANLP Solves the transportation method of linear programming

51

LNPROG Solves general linear programming problems by means of the simplex method

53

MIPROG Solves the mixed integer linear programming problem by means of branch and bound

55

MULREG Finds linear regression equations by means of the stepwise procedure of regression/correlation analysis

57

SCSIM Simulates the flow of a product through five echelons of a supply channel

62

Each module is selected from the following master screen by clicking on the appropriate button.

HARDWARE REQUIREMENTS

LOGWARE is designed for microcomputers operating under WINDOWS 98, NT, 2000, or XP. At least 16MB of RAM should be installed. Hard drive space of at least 10MB should be available. A color monitor capable of producing at least 640x480 pixels resolution is needed, although 800x600 is better and 1024x768 is preferred. Resolutions greater than 1024x768 pixels are not supported. A laser printer is preferred. A mouse is needed. A 3½ floppy drive and/or a compact disk reader are needed.

INSTALLING THE SOFTWARE ON A HARD DRIVE

Place the program compact disks in the appropriate drives. In WINDOWS, click on the Start button and then select the Run option from pop-up menu. Type “X:Setup.exe” (“X” being the letter designated for your CD drive) . The program may also be installed with Windows’ Start, Settings, Control Panel, Add/Remove Programs, Install option. Change the subdirectory under which the program will be installed if the default subdirectory is not desired.

RUNNING THE PROGRAMS

After the program is installed, click on the Start button and select Programs. Choose the Logware icon to activate the program. Click on the desired program module. A shortcut icon on the Desktop may also be created.

EDITING THE DATA

In those modules where a screen data editor is present, the first action is to open a data file by clicking on the module’s Start button. If a file is named that is not in the current list of files, a data shell will be created into which a new problem may be entered. The use of the editor is simple and transparent with a little practice; however, a few comments about its use may help to get started.

• Press the Ins key to start a new line of data in a matrix. The normal action is to insert a text row at the end of the matrix. The Add button may also be pressed. This will allow a row to be added at the end of the matrix as well as within the matrix. Position the cursor in the matrix row where the row is to be added.

• Pressing the Esc key clears a matrix cell.

• Pressing the Delete button deletes the row in a matrix highlighted by the current cursor position.

• If Column arithmetic is to be used, highlight the matrix column on which the action is to apply.

Alternatively, the data for each module except SCSIM may be created and edited with the use of Excel. It is expected that the user has a basic knowledge of Excel use.

COPYING THE INSTRUCTIONS AND THE SOFTWARE

This software and the associated instruc¬tions may be copied as long as they are used for educational purposes. All copied materials must display the following copyright notices.

 Copyright 1992-2004 Ronald H. Ballou All rights reserved.

Ronald H. Ballou offers this software for educational purposes only and does not warrant the software to be fit for any particular application. The user agrees to release Ronald H. Ballou from all liabilities, expenses, claims, actions, and/or damages of any kind arising directly or indirectly out of the use of these computer programs, the performance or nonperformance of such computer programs, and the breach of any expressed or implied warranties arising in connection with their use. If these conditions are not acceptable, the software should be returned to Ronald H. Ballou.

Professor Ronald H. Ballou

Weatherhead School of Management

Case Western Reserve University

Cleveland, OH 44106 USA

Tel: (216) 368-3808

Fax: (216) 368-6250

E-mail: Ronald.Ballou@Weatherhead.CWRU.edu

Up to date information about the software may be found at www.PrenHall.com/Ballou.

INSTRUCTIONS FOR EXPONENTIAL SMOOTHING

AND TIME SERIES DECOMPOSITION FORECASTING

FORECAST

FORECAST is computer software that forecasts from time series data by means of exponential smoothing and/or time series decomposition methods. In logistics/SC, such time series may be product sales, lead times, prices paid for goods, or shipments. The philosophy of time series forecasting is to project an historical pattern of the data over time, and, if present, account for trend and seasonality. Exponential smoothing is a moving average approach that projects the average of the most recent data and adapts the forecast to changing data as they occur. On the other hand, the time series decomposition approach recognizes that major reasons for variation in data over time are due to trend and seasonal components. Each of these is estimated and combined to produce a forecast. For background information on the forecast models used in FORECAST, see Chapter 8 of the Business Logistics/Supply Chain Management 5e textbook.

To run FORECAST, select the appropriate module from the LOGWARE master menu. Open an existing file or select a new one. Prepare or change the database. Select the appropriate model type, which may be either some form of an exponential model (Level only, Level-Trend, etc.) or the time series decomposition model. Click on Solve to generate a forecast.

INPUT

The input to both forecasting modules consists of the time-ordered series of data, ranked from the most historic to the most recent observations, and various parameter values that guide the execution of the models. The dimensions of the models allow observations for up to 200 periods and a forecast of up to 50 periods. Both model types run from the same database although some of the parameters are not used in the time series decomposition model.

Parameters and Labels

This portion of the screen sets the parameters for both the exponential smoothing and time series decomposition models. These guide the overall action of the models. Consider each element on this screen.

Problem label. This is a label given to the problem you are solving. WARNING: Do not use commas (,) or double quotation marks (") in the label since this will cause an error in reading the data file.

Number of data points. Specify the number of data periods in the time series. Up to 200 points are allowed. Be sure that the number of points specified here matches the number of data points actually entered in the time series.

Initialization period. The initialization period is the number of the oldest data points used to determine starting values for the exponential smoothing model. A minimum of 3 periods of data should be declared for this purpose. If a seasonal model is to be used, at least the number of periods in one full seasonal cycle must be specified.

Error statistics. The number of data periods needed to compute forecast error statistics is referred to as the validation period. These error statistics are the mean absolute deviation (MAD), the bias (BIAS), and the root mean square error (RMSE). The validation period is the last N periods of data. Enough data points should be used from this validation period to strike a reasonable average for these statistics.

MAD is defined as the average of the

...