Log Ware 5.0

SELECTED COMPUTER
PROGRAMS
FOR LOGISTICS/SUPPLY CHAIN
PLANNING
Version 5.0
Ronald H. Ballou
Weatherhead School of Management
Case Western Reserve University
(C) Copyright !!"#"00$ Ronald %. &allo' (ll rights reserved
2
S)*)C+), C-M.U+)R .R-/R(MS
0-R *-/1S+1CS2SU..3 C%(14 .*(4414/
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 Pag
e
FORECAST Forecasts time series data by means of exponential
smoothing and time series decomposition methods 5
ROUTE Determines the shortest path through a networ of
routes !
ROUTESEQ Determines the best se"uence to visit stops on a route #$
ROUTER Develops routes and schedules for multiple trucs
serving multiple stops #5
INPOL Finds optimal inventory ordering policies based on
economic order "uantity principles 2%
COG Finds the location of a single facility by the exact center&
of&gravity method $$
MULTICOG 'ocates a selected number of facilities by the exact
center&of&gravity method $(
PMED 'ocates a selected number of facilities by the )&median
method %#
WARELOCA * warehouse location program for specifically analyzing
the +semore ,oap -ompany case study %5
LAYOUT )ositions products in warehouses and other facilities %(
MILES -omputes approximate distance between two points
using latitude&longitude or linear&grid coordinate points %!
TRANLP ,olves the transportation method of linear programming
5#
LNPROG ,olves general linear programming problems by means
of the simplex method 5$
MIPROG ,olves the mixed integer linear programming problem
by means of branch and bound 55
MULREG Finds linear regression e"uations by means of the
stepwise procedure of regression/correlation analysis 5(
SCSIM ,imulates the flow of a product through five echelons of
a supply channel .2
/ach module is selected from the following master screen by clicing on the
appropriate button.
$
HARDWARE REQUIREMENTS
LOGWARE is designed for microcomputers operating under 0I1D20, !34 154 26664
or 7). *t least #.89 of :*8 should be installed. ;ard drive space of at least #689
should be available. * color monitor capable of producing at least .%6x%36 pixels
resolution is needed4 although 366x.66 is better and #62%x(.3 is preferred. :esolutions
greater than #62%x(.3 pixels are not supported. * laser printer is preferred. * mouse is
needed. * $< floppy drive and/or a compact dis reader are needed.
INSTALLING THE SOFTWARE ON A HARD DRIVE
)lace the program compact diss in the appropriate drives. In 0I1D20,4 clic on the
Start button and then select the Run option from pop&up menu. 5ype =X:Setupe!e>
?=7> being the letter designated for your -D drive@ . 5he program may also be installed
with 0indowsA Start" Sett#n$%" C&ntr&' Pane'" A(()Re*&+e Pr&$ra*%" In%ta'' option.
-hange the subdirectory under which the program will be installed if the default
subdirectory is not desired.
RUNNING THE PROGRAMS
*fter the program is installed4 clic on the Start button and select Pr&$ra*%. -hoose the
L&$,are icon to activate the program. -lic on the desired program module. * shortcut
icon on the Destop may also be created.
%
EDITING THE DATA
In those modules where a screen data editor is present4 the first action is to open a data
file by clicing on the moduleAs Start button. If a file is named that is not in the current
list of files4 a data shell will be created into which a new problem may be entered. 5he use
of the editor is simple and transparent with a little practiceB however4 a few comments
about its use may help to get started.
)ress the In% ey to start a new line of data in a matrix. 5he normal action is to
insert a text row at the end of the matrix. 5he A(( button may also be pressed.
5his will allow a row to be added at the end of the matrix as well as within the
matrix. )osition the cursor in the matrix row where the row is to be added.
)ressing the E%- ey clears a matrix cell.
)ressing the De'ete button deletes the row in a matrix highlighted by the current
cursor position.
If C&'u*n ar#t.*et#- is to be used4 highlight the matrix column on which the action
is to apply.
*lternatively4 the data for each module except ,-,I8 may be created and edited with
the use of /xcel. It is expected that the user has a basic nowledge of /xcel use.
COPYING THE INSTRUCTIONS AND THE SOFTWARE
5his software and the associated instructions may be copied as long as they are used for
educational purposes. *ll copied materials must display the following copyright notices.
-opyright #!!2&266% :onald ;. 9allou *ll rights reserved.
:onald ;. 9allou offers this software for educational purposes only and does not
warrant the software to be fit for any particular application. 5he user agrees to release
:onald ;. 9allou from all liabilities4 expenses4 claims4 actions4 and/or damages of any ind
arising directly or indirectly out of the use of these computer programs4 the performance
or nonperformance of such computer programs4 and the breach of any expressed or
implied warranties arising in connection with their use. If these conditions are not
acceptable4 the software should be returned to :onald ;. 9allou.
)rofessor :onald ;. 9allou
0eatherhead ,chool of 8anagement
-ase 0estern :eserve +niversity
-leveland4 2; %%#6. +,*
5elC ?2#.@ $.3&$363
FaxC ?2#.@ $.3&.256
/&mailC :onald.9allouD0eatherhead.-0:+.edu
+p to date information about the software may be found at www.)ren;all.com/9allou.
5
14S+RUC+1-4S 0-R )5.-4)4+1(* SM--+%14/
(4, +1M) S)R1)S ,)C-M.-S1+1-4 0-R)C(S+14/
FORECAST
FORECAST is computer software that forecasts from time series data by means of
exponential smoothing and/or time series decomposition methods. In logistics/,-4 such
time series may be product sales4 lead times4 prices paid for goods4 or shipments. 5he
philosophy of time series forecasting is to proEect an historical pattern of the data over
time4 and4 if present4 account for trend and seasonality. /xponential smoothing is a
moving average approach that proEects the average of the most recent data and adapts the
forecast to changing data as they occur. 2n the other hand4 the time series decomposition
approach recognizes that maEor reasons for variation in data over time are due to trend
and seasonal components. /ach of these is estimated and combined to produce a forecast.
For bacground information on the forecast models used in FORECAST4 see -hapter 3
of the 9usiness 'ogistics/,upply -hain 8anagement 5e textboo.
5o run FORECAST4 select the appropriate module from the LOGWARE master
menu. 2pen an existing file or select a new one. )repare or change the database. ,elect
the appropriate model type4 which may be either some form of an exponential model
?'evel only4 'evel&5rend4 etc.@ or the time series decomposition model. -lic on S&'+e to
generate a forecast.
INPUT
5he input to both forecasting modules consists of the time&ordered series of data4 raned
from the most historic to the most recent observations4 and various parameter values that
guide the execution of the models. 5he dimensions of the models allow observations for
up to 266 periods and a forecast of up to 56 periods. 9oth model types run from the same
database although some of the parameters are not used in the time series decomposition
model.
Para*eter% an( La/e'%
5his portion of the screen sets the parameters for both the exponential smoothing and
time series decomposition models. 5hese guide the overall action of the models. -onsider
each element on this screen.
Problem label. 5his is a label given to the problem you are solving. 0*:1I1FC Do
not use commas ?4@ or double "uotation mars ?G@ in the label since this will cause an error
in reading the data file.
Number of data points. ,pecify the number of data periods in the time series. +p to
266 points are allowed. 9e sure that the number of points specified here matches the
number of data points actually entered in the time series.
Initialization period. 5he initialization period is the number of the oldest data points
used to determine starting values for the exponential smoothing model. * minimum of $
periods of data should be declared for this purpose. If a seasonal model is to be used4 at
least the number of periods in one full seasonal cycle must be specified.
Error statistics. 5he number of data periods needed to compute forecast error
statistics is referred to as the validation period. 5hese error statistics are the mean
absolute deviation ?MAD@4 the bias ?BIAS@4 and the root mean s"uare error ?RMSE@. 5he
6
validation period is the last N periods of data. /nough data points should be used from
this validation period to strie a reasonable average for these statistics.
MAD is defined as the average of the absolute differences between the actual values
and the forecast values for the validation period. BIAS is the average of the differences
between actual and forecast values for the validation period. RMSE is s"uare root of the
average of the s"uared differences between the actual and the forecast values for the
validation period.
Model type. ,electing the model type refers to the exponential smoothing model or
the time series decomposition model. 5here are four variations of the exponential
smoothing model for best representing the character of the time series. 5hese are the
'evel only4 'evel&5rend4 'evel&,easonal4 and 'evel&5rend&,easonal. ,elect the type that
best represents the data. *lternately4 select the time series decomposition model.
Smoothing constant search. 0hen one of the forms of the exponential model is
selected4 indicate whether a search for the smoothing constants is to be performed using
FORECAST. If not4 the smoothing constants for the model type selected must be
specified.
If FORECAST is to search for the smoothing constants4 indicate the smoothing
constant increment for searching. WARNING: ,ignificant running time may result from
too small a search increment4 especially for the more complicated model forms such as the
'evel&5rend&,easonal model. *n increment of 6.# wors reasonably well. 9e prepared to
wait when smaller increments are chosen.
Seasonal length. 0hen using seasonality in either the exponential smoothing or time
series decomposition models4 indicate the number of data periods representing a full
seasonal cycle. WARNING: 5he initialization period should be a seasonal cycle plus 2
data periods. If a model type without seasonality is used4 specify a seasonal length of 6.
;owever4 be sure to use at least three data periods for initialization.
Forecast length. +p to 56 future periods may be specified for forecasting. ;owever4
remember that these are historical proEection methods such that forecasting beyond .
months or a full seasonal cycle can lead to a significant forecasting error.
T#*e Ser#e% Data
5his screen section is for entering time series data by period. * period may be any time
segment such as a day4 wee4 month4 or "uarter.
Period label. * time segment4 or period4 may be given an identification label. Food
practice is to eep the number of letters or numbers to less than or e"ual to #5.
0*:1I1FC *void using commas ?4@ or double "uotation mars ?G@ in the labels.
Observations. *ll time series data should be entered chronologically4 with period #
representing the oldest observation. +p to 266 data points are allowed. *lthough larger
numbers are permitted4 scaling data so that entries are not larger than . digits is good
practice.
EXAMPLE
5he prices for a certain purchased component have been observed over one year and a
half. 5hese prices are as follows.
7
Period, mo. Price, $/unit Period, mo. Price, $/unit
26664 Han. #!.$. 266#4 Han. $2.52
Feb. 25.%5 Feb. $#.$$
8ar. #!.($ 8ar. 25.$2
*pr. 2#.%3 *pr. 2(.5$
8ay 26.(( 8ay 2..$3
Hune 25.%2 Hune 2$.(2
Huly 2$.(! Huly 2!.#%
*ug. 23.$5 *ug I
,ep. 2..36 ,ep. I
2ct. 25.$2 2ct. I
1ov. 25.22 1ov. I
Dec. 2(.#% Dec. I
5he prices are to be forecast to the end of 266#.
* full seasonal cycle is #2 months. 5herefore4 the initialization period is set at #2 J 2
K #% periods. 5hree periods are selected for validation. * 'evel&5rend&,easonal
exponential smoothing model is to be tested with a level smoothing constant ?@ K 6.24 a
trend smoothing constant ?@ K 6.$4 and a seasonal smoothing constant ?@ K 6.#. 5he
data screen is prepared as shown in Figure F2:/-*,5&#.
2nce the data have been prepared4 select the S&'+e button. * validation chec of the
data will be made and the data processed according to the selected model. 5he results are
shown in Figure F2:/-*,5&2. * graphical display of the data and the forecast can be
seen in Figure F2:/-*,5&$.
F#$ure FORECAST01 Data ,creen for the /xample )roblem.
8
F#$ure FORECAST02 ,olution :eport for the /xample )roblem.
F#$ure FORECAST03 5ime ,eries Data and the Forecast.
TIME SERIES FORECASTING
Curve fitting and model validation
TREND EQUATION: !"#$ %"&'#(T
MODE) T*+E: Time Serie, De-om.o,ition
+rd +eriod Sea/
no" la0el////////// A-tual Fore-a,t Trend ,onal
Error
!# MAR 11! #"& &"(# 2"3' "'
!"32
!$ A+R 11! 2"#& #"(' 2"(' "''
!"$3
!2 MA* 11! $"&( 3"'$ ("( "'&
!"3
!( 4UNE 11! &"2 &1"32 ("$( "(&
/$"2#
!' 4U)* 11! '"!3 ("3& '"1( !"11 "
2!
T5E FORECAST
MAD 6 "'$ 7IAS 6 /!"#3 RMSE 6 3"11
T8e fore-a,t, for .eriod, 1 to 3 are:
+eriod Fore-a,t
1 &&"(1
! &!"(2
&1"1&
9
14S+RUC+1-4S 0-R +%)
S%-R+)S+ R-U+) .R-/R(M
ROUTE
ROUTE is a program to find the shortest route through a networ of nodes connected by
arcs4 which usually are roads4 rail lines4 streets4 pipes4 or wires. * path4 or arc4 is a
connection between nodal pairs over which travel may occur. * route is a set of arcs
connected head to tail for traveling from one node on the networ to another. ,ince there
are usually many ways between points on a well&connected networ4 we see the route
minimizing travel distance or time between two selected points. ROUTE finds all of the
minimum&cost routes from a specified origin node to all other nodes in the networ. For
bacground information on the shortest route method used in ROUTE4 see -hapter ( of
the 9usiness 'ogistics/,upply -hain 8anagement 5e textboo.
INPUT
Input is managed by three sections of the data screen. 5he first calls for a title to the
problem and identifies the origin node by number. 5here may be up to 566 nodes in the
networ.
5he second section is used to create a master list of nodes and identify them with a
descriptor. If a graph of the problem solution is desired4 coordinate points for the nodes
may be added. 5o determine these coordinate points4 place a linear grid over a map of the
networ and read the X,Y coordinates for each node. 5he 646 coordinates must begin in
the southwest corner of the grid.
5he third section is used to define the available arcs between nodes and the cost
?distance or time@ to traverse the arcs. 1odes are numbered. -ost may also be expressed
as a weighted index of both time and distance. 5he relative weights depend on the balance
desired between the goals of shortest time and shortest distance. 5here may be up to !66
of these connecting arcs.
*fter completing the data entry4 permanently save the data in a file by selecting the
Sa+e button and choosing a file name to receive the data. +se an input file of the form
:F'xx.D*54 where xx refers to the problem number such 6#4 654 #64 etc.
RUNNING O!"E
*fter creating or retrieving data from a file4 clic the ROUTE button to solve the problem
for the data as displayed on the screen. 5he resulting optimal paths to each destination
point are shown.
Lou may also view a graph of the networ by clicing on the P'&t button. 5he
problem solution may further be displayed for the current run. ,elect the S&'ut#&n pat.
button and indicate the destination point of interest. 5he optimal path will be highlighted.
EXAMPLE
-onsider the problem as shown in Figure :2+5/&#. 5he maEor highways between
*marillo4 5exas and Fort 0orth4 5exas are given. 5he approximate travel times between
nodal pairs are shown. 2ur tas is to find the route that offers the lowest driving times to
go from *marillo4 57 to Fort 0orth4 57.
10
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
0
1
2
3
4
5
6
7
8
9
10
11
12
13
A
90
B
84
C
E 84
D
F
G
H
I
J
Destination
Fort Worth
Origin
Amarillo
Note: All times are in minutes
Oklahoma
City
!oor"inates
#

!
o
o
r
"
i
n
a
t
e
s
F#$ure ROUTE01 /xample 1etwor of the 8aEor ;ighways between *marillo4 57 and
Ft. 0orth4 574 with Driving 5imes.
0e will label the nodes so that AK#4 BK24 etc. 5he database screen segments are as
follows.
Parameters and labels
Node identifcation and location
+ro0lem la0el/ E9AM+)E +RO7)EM
Origin node num0er: !
+oint Node
no" no" Node name 9/-oord */-oord
! ! A/AMARI))O "&1 !!"!1
7 2"&1 !!"!1
& & C $"$1 ("#1
3 3 D ("1 "(1
# # E !&"'1 !!"11
$ $ F !&"!1 $"11
2 2 G !"#1 !"'1
( ( 5 !$"$1 3"'1
' ' I/O: CIT* !'"1 !1"21
!1 !1 4/FT ;ORT5 !'"$1 &"&1
+oint ////////From/////////// //////////To///////////
no" Node no" Node name Node no" Node name Co,t
! ! A/AMARI))O 7 '1"11
! A/AMARI))O & C !&("11
& ! A/AMARI))O 3 D &3("11
3 7 # E (3"11
# 7 & C $$"11
$ & C 3 D !#$"11
2 & C $ F '1"11
( 3 D 2 G 3("11
' # E ' I/O: CIT* (3"11
!1 # E $ F !1"11
!! $ F ( 5 $1"11
! $ F 2 G !&"11
!& 2 G ( 5 3("11
!3 2 G !1 4/FT ;ORT5 !#1"11
!# ( 5 ' I/O: CIT* !&"11
!$ ( 5 !1 4/FT ;ORT5 !$"11
!2 ' I/O: CIT* !1 4/FT ;ORT5 !$"11
11
Node connections
5he solution to this problem has the following output form.
It will tae $3% minutes to drive the shortest route from *marillo to Fort 0orth. 5his
is route ! " # $4 or A B E I % in the notation of Figure
:2+5/&#. * plot of the problem with its solution is shown in Figure :2+5/&2.
C
E
D
F
H
J
B A
I
G
F#$ure ROUTE02 )lotted ,olution to /xample )roblem.
EXAMPLE PROBLEM
Origin node number = 1 Number of nodes = 10
Number of r!s = 17
"#or$es$ %$#s from origin node 1 $o && des$in$ion nodes
'os$ P$#
90.00 1 () *
1+8.00 1 () +
*9,.00 1 () + () ,
17,.00 1 () * () -
**8.00 1 () + () 6
++6.00 1 () + () 6 () 8 () 7
*88.00 1 () + () 6 () 8
*-8.00 1 () * () - () 9
+8,.00 1 () * () - () 9 () 10
1*
14S+RUC+1-4S 0-R +%)
+R(V)*14/ S(*)SM(4 .R-/R(M
ROUTESEQ
ROUTESEQ is a heuristic program for solving the traveling salesman problem. It will
se"uence up to 26 stops on a route plus an origin point. ,tops and the origin point are
identified with linear coordinate points. /uclidean ?straight&line@ distances are computed
in terms of these coordinate points. For bacground information on the coincident origin&
destination ?traveling salesman@ problem4 see -hapter ( of the 9usiness 'ogistics/,upply
-hain 8anagement 5e textboo.
INPUT
*n input file is prepared with the use of the data editor or /xcel spreadsheet. 5here are
four types of recordsC ?#@ the X,Y coordinates of the origin point4 ?2@ a circuity factor4 ?$@
a map scaling factor4 and ?%@ the coordinate points for each stop. * typical data screen
might loo as followsC
F#$ure ROUTESEQ01 5ypical Data ,creen.
Problem label. 5his is a problem descriptor. /nter any convenient label. 0*:1I1FC
Do not use commas ?4@ or double "uotation mars ?G@ in the label since this will cause an
error in reading the data file.
#ircuity factor. 5his is a multiplier greater than # to convert straight&line distance to
road4 rail4 air4 etc. actual travel distance. For example4 if linear coordinates are used4 a
multiplier of #.2# is a good average to convert coordinate&computed distance to road
miles. Do not use a value less than #.
Map scaling factor. * multiplier to convert coordinates to a desirable distance
measure. -oordinate divisions on a map or grid may be arbitrary. 5he map&scaling factor
typically converts these coordinates to miles or ilometers. For example4 a map showing a
scale of #K56 miles and line marers every inch would have a scaling factor of 56.
1+
$epot coordinates. 5hese are the X,Y coordinates for the point where the route starts
and ends. X and Y are linear grid coordinates4 although other coordinate systems may be
used with care.
Stop data. 5he points to be visited on a route are identified by their X,Y coordinates.
X and Y are linear grid coordinates4 although other coordinate systems may be used with
care. +p to 56 stops are allowed on a route.
RUNNING O!"ESE%
*fter the data have been prepared on the data screen4 clic on the S&'+e button to find the
best route. 5he stop se"uence may be specified4 or LOGWARE will design the route.
2nce a solution is available4 it may be graphically presented by clicing on the P'&t button.
EXAMPLE
* truc is to be routed from its depot to five stops. 5he data screen giving location of the
depot and the stops was previously shown. * grid overlay of the depot and stops is shown
in Figure :2+5/,/M&#. 1ote that a circuity factor of #.2# is used. *lso4 note that the
map&scaling factor is # in this case.
5he computed results as they appear on the screen are as follows with the plotted
route shown in Figure :2+5/,/M&2.

.igure RO/0E"E1(* )lot of /xample )roblem ,olution.
STO+ SEQUENCE RESU)TS
Sto. ,e<uen-e i,:
DE+OT ! 3 # & DE+OT
Total route di,tan-e 6 !3"!$2
1,
14S+RUC+1-4S 0-R
V)%1C*) R-U+14/ (4, SC%),U*14/
ROUTER
ROUTER is a software program to determine the best routes and schedules for a fleet of
privately controlled vehicles. 5he typical problem is one where trucs are domiciled at a
central depot4 maing deliveries or picups to a number of stops4 and returning to the
same depot. ,tops need to be assigned to vehicles and then se"uenced on each vehicle
route. 5he obEective is to minimize the total distance traveled on all routes4 and indirectly
to minimize the total number of vehicles needed to serve the stops.
E&ample. * food company maes daily deliveries from its warehouse to retail
stores as shown in Figure :2+5/:&#. 5he dispatcher must plan how many routes
there should be4 which stores should be assigned to the routes4 and in which
se"uence the stores should be served. -ertain of the stores can accept delivery
only at specified times of the day4 the drivers are to wor 3 hours or be paid
overtime4 and the trucs are limited in their carrying capacity. 5hese restrictions
are to be respected in designing the routes.
F#$ure ROUTER01 Nehicle :outing from a -entral Depot.
ROUTER is a commercial&grade model that has been installed in actual dispatching
operations and has many capabilities that are not fully described in this set of abbreviated
instructionsB however4 many of these become obvious by exploring the data screens. In
addition4 there are limits placed on the problem size of this educational version of the
model. It handles up to .6 stops. 5he modelOs features includeC
9oth picup and delivery stops are permitted on the same route. )icups may be
mixed on the vehicle with deliveries4 or they may only be permitted on the vehicle
after all deliveries have been made.
8ixed vehicle types are allowed.
'oads on a vehicle may be controlled by weight4 cube4 or number of stops.
* variety of coordinate systems for stop and depot location is allowed.
1-
Distances between depot and stops4 or between stops4 are computed from
coordinate geometry4 or they may be specified.
5he maximum time or distance on a route may be specified.
9arriers may be defined to represent laes4 pars4 rivers4 etc. through which a route
cannot penetrate.
5he earliest time for a vehicle to leave a depot and the latest time for it to return may
be specified.
,top loading/unloading times may be calculated based on weight and cube4 or they
may be specified for each stop.
9reas such as for lunch and overnight may be specified.
,peed zones are used to define the speed between groups of stops4 or speeds may be
specified between selected stop pairs.
5ime windows in which deliveries or picups are to be made can be specified for
each stop.
:oute design may be computed with the model using any one of three methods4 or
the user may specify the design.
:oute costs are determined based on vehicle fixed and variable rates4 driver fixed
and variable rates4 and overtime rates.
Incremental costs of serving a stop on a route are calculated which can be compared
with an alternative transport method of serving the stop.
For bacground information on the routing and scheduling method ?savings method@
used in ROUTER4 see -hapter ( of the 9usiness 'ogistics/,upply -hain 8anagement 5e
textboo.
INPUT
*ll data are entered by means of a screen editor. First4 select a file from among existing
ones or start a new one. /nter data in the appropriate screens as chosen from the various
folders.
5he data needed for vehicle routing consist primarily of data about stops4 vehicles4
costs4 and constraints on route design. /ach data element is discussed below.
C&&r(#nate P&#nt%
5he depot and all stops are located geographically with coordinate points. * variety of
coordinate systems is permitted such as latitude&longitude or a simple linear grid.
-oordinates are used to compute the approximate straight&line distance between point
pairs and to locate them in relation to each other for mapping purposes.
S-a'#n$ Fa-t&r%
,caling factors are applied to the coordinate&computed distances to convert them to
actual distances. 5hese scaling factors4 both horizontal and vertical4 are made up of two
parts. 5he first is to convert the coordinates to straight&line miles. Do this by multiplying
the coordinate points by the map&scaling factor. ;owever4 if longitudes and latitudes are
used4 the scaling factors will depend on the approximate latitude of the stops and depot.
5hese conversion factors for latitude&longitude are given in 5able :2+5/:&#. 2thers
may be obtained from the particular map used.
16
,econd4 a circuity factor is applied. 5his factor adEusts straight&line miles to
approximate actual road miles. * commonly accepted factor ?used by the Department of
5ransportation@ is #.2#4 if linear grid coordinates are used. 5his means that road miles are
about 2# percent longer than straight&line miles where distance is calculated from linear
grid coordinates. If latitude&longitude coordinates are used4 a factor of #.26 is more
accurate for road distances in the +.,. ,ee 5able #%&% of 9usiness 'ogistics/,upply -hain
8anagement 5e for circuity factors for other parts of the world.
5herefore4 the scaling factor input into the ROUTER database is the product of the
above two factors. For example4 if the map scaling factor is 6.$ and the circuity factor is
#.2#4 the combined factor is 6.$ #.2# K 6.$.$. 5he final value may be altered to reflect
any adEustments that may have been made in the coordinate points. If linear coordinates
are used4 horizontal and vertical scaling factors are liely to be the same4 but they will be
different for latitude&longitude coordinates.
Dep&t T#*e Re%tr#-t#&n%
Depots fre"uently have time restrictions on them reflecting their open hours. It may
be desirable to have vehicles not leave the depot before a certain time ?earliest starting
time@ and to return no later than a certain time ?latest return time@.
R&ute Re%tr#-t#&n%
It is often necessary or desirable to restrict routes by time and or distance. * time
restriction is the maximum number of hours that a vehicle is allowed on a route. 5he
distance restriction is the maximum number of miles that a vehicle may travel before it
must return to the depot. Be &ure t'at t'e&e re&triction& do not con(lict )it' t'e de*ot
time re&triction&.
Spee( 4&ne%
5he depot and the stops may be grouped into boxes called speed zones and the speeds
then defined between the zones. 5his allows speed differences to be recognized between
rural and city areas4 between congested and non&congested regions4 or between any
groups of stops.
* speed zone is a rectangular box into which stops are assigned. 5hese boxes are
defined with the coordinates of their sides4 i.e.4 left4 right4 top4 and bottom. 5he
coordinates of each box must use the same coordinate system as for the depot and the
stops.
2nce the speed zones have been defined4 the speeds between the boxes and within the
boxes should be specified. For example4 speeds between stops within a city zone might be
25 mph4 while 56 mph might be appropriate between stops in a rural zone. ;owever4
speeds between the two city zones crossing a rural zone would be represented by a
combined speed4 possibly reflecting the proportion of the distance spent in each zone. If
the approximations seem crude4 a greater number of smaller boxes may be created to
improve accuracy.
17
For our problem4 there would be . speed combinations4 assuming that the speed
between say zone # and zone 2 is the same as between zone 2 and zone #. 'isting the
speed combinations4 we would haveC

+rigin De&tination S*eed
,one ,one m*'
# # 25
# 2 %5
# $ $5
2 2 25
2 $ %5
$ $ 25
If a speed for a zone combination that is needed has not been specified and there is no
specific speed for the stop combination4 the default or the specific speed between stops is
used.
,peeds may also be specified for selected combinations of stops4 or depot and stops.
Finally4 a default speed is given.
ROUTER prioritizes its choices of speeds. First4 the computer searches for a specific
stop&to&stop speed. If none is available4 an attempt is made to find a speed from the ,peed
Pones list. If none is available4 the default speed is used. * default speed should always
be defined. Be &ure t'at &*eed& are &u((icientl- large &o t'at all &to*& can .e &er/ed
)it'in t'e time re&triction&.
"able O!"E'( 0engt'& o( +ne Degree 0atitude and
+ne Degree 0ongitude

0atitude 0ongitude
0atitude Statute 1ilo2 Statute 1ilo2
3Degree&4 mile& meter& mile& meter&
6 .3.(6% ##6.5.! .!.#(2 ###.$22
5 .3.(#6 ##6.5(3 .3.!## ##6.!62
#6 .3.(25 ##6..6$ .3.#2! #6!..%$
#5 .3.(5# ##6..%% ...3$6 #6(.55$
26 .3.(3. ##6.(6# .5.62. #6%..56
25 .3.32! ##6.((6 .2.(2! #66.!5$
$6 .3.3(! ##6.356 5!.!5. !..%!6
$5 .3.!$5 ##6.!%# 5..(25 !#.2!6
%6 .3.!!$ ###.6$% 5$.6.$ 35.$!(
%5 .!.65% ###.#$2 %3.!!5 (3.356
56 .!.##5 ###.2$6 %%.552 (#.(66
55 .!.#(5 ###.$2( $!.(.. .$.!!(
.6 .!.2$6 ###.%#5 $%..(% 55.36$
.5 .!.23# ###.%!( 2!.$#5 %(.#(3
(6 .!.$2% ###.5.( 2$.(2! $3.#33
(5 .!.$.6 ###..25 #(.!.6 23.!6%
36 .!.$3. ###.... #2.65# #!.$!%
35 .!.%62 ###..!2 ..6%! !.($5
!6 .!.%6( ###.(66 6.666 6.666
18

D#%tan-e%
,pecified distances may be used in place of the coordinate point computed distances.
,uch specified distances always tae precedence over the approximated distances. 5hey
are used typically on a selected basis where greater accuracy is needed4 such as when the
route is to represent a drive&time standard. In addition4 they may be needed to accurately
represent distances when one&way streets4 laes4 mountains4 or other barriers on the route
mae distance approximations unacceptable. 5hese distances can be specified by depot&
to&stop or by stop&to&stop pairs.
5rea6 T#*e%
,ome route designs may re"uire the driver to tae a brea during the tour4 such as for
lunch or a rest. +p to two brea times are permitted4 and they are expressed in minutes.
5hey are to tae place after a specified number of minutes into the planning horizon.
T#*e W#n(&,%
*ny stop may have certain times within which deliveries or picups can be made.
5hese are referred to as time windows. 5hey are specified as a beginning time and an
ending time in minutes. If the first of two time windows is non&constraining4 then the
times should be set as wide as possible. * good choice would be to use the earliest start
and latest return times for the depot. * non&constraining second time window is to set the
beginning and ending times to a number beyond the planning horizon4 say !!!!.
St&p V&'u*e%
,tops are designated as either a delivery ?D@ or a picup ?P@. * delivery stop is one
where goods originate from the depot and are destined for the stop. * picup stop is one
where goods originate from the stop and are destined for the depot.
,top volumes may be expressed in one or two measures4 typically weight and cube.
0eight may be hundredweight4 cases4 units4 ilograms4 or other similar measure. It uses
the )eig't carrying capacity of the vehicle. -ube4 on the other hand4 uses the &*ace
carrying capacity of the vehicle. It is expressed as cubic feet4 cubic meters4 or other
appropriate space measure. -ube may also be used as a surrogate for stops. 5hat is4 by
declaring cube to be # and then specifying cube capacity to be the number of stops
allowed on a route4 the number of stops on a route may be controlled.
It is not always necessary to control both weight and cube. 0eight is the primary
measure such that cube may be inactive at times. If so4 use zeros for all stops and use a
vehicle cube capacity of any size.
P#-6up P&'#-7
0hen picups are to enter a delivery route4 there will be the "uestion of whether
picups should be made )'ile delivery volume is present on the vehicle or whether they
should only be allowed a(ter all deliveries have been made. 5his policy is expressed in
percentage terms of the available vehicle capacity. * vehicle loaded value of 6 percent
means that picups are allowed only after all of the deliveries have been made4 or the
vehicle is empty. * $5 percent value means that picups are allowed when the vehicle load
has dropped to $5 percent of capacity or less. )icups are not allowed as long as the
picup and delivery volume is more than $5 percent of capacity. * value of #66 percent
19
means that picups are allowed any time on a route. )ercentage values near #66Q allow
the most flexibility in route design.
Ve.#-'e Capa-#t7
Nehicles are categorized first by uni"ue type and then by the number of vehicles and
their characteristics within that type. 5he various vehicle types are numbered
consecutively. 5he number of vehicles within that type is then declared. /ach type is
given a capacity by weight and by cube. 0eight and cube units should match those given
as stop volumes.
It is a good idea to declare more vehicles than actually available in the fleet.
ROUTER will only use enough of them to form the routes4 which may be more than
currently being used due to meeting the restrictions placed on route design. 5he current
fleet may not be meeting these restrictions and therefore can operate with fewer vehicles.
C&%t%
-osts of the routes are determined by summing the costs associated with the vehicle
and the driver. Nehicle costs are based on two rates. 5he first is the fixed rate per vehicle.
5his is the fixed charge associated with owning4 or leasing4 and maintaining the vehicle for
the period represented by route planning4 i.e.4 a day4 a wee4 or a month. 1ext is the
variable rate per mile for operating the vehicle. 5his rate represents such costs
apportioned on a mileage basis as fuel4 tires4 and oil. 5hese rates can vary with the vehicle
type.
Driver costs are based on fixed and variable rates plus an overtime rate. 5he fixed
driver rate is the charge associated with fringe benefits4 salary minimums4 and other
charges that do not vary with the time on the route. 5he variable rate is associated with
the costs that are dependent on the hours spent on the route such as the wage rate. 5he
overtime rate is the per&hour rate that taes effect after a specified number of hours have
been spent on a route.
R&ute 5arr#er%
5here may be certain areas where the vehicles must go around rather than penetrate
them. ,uch areas may be laes4 rivers4 pars4 etc. In ROUTER4 a vehicle comes up to
the barrier and runs around the &'orte&t &ide. 5he distance around the barrier is added to
the route distance.
9arriers are always expressed as a %&sided figure. 5his figure may be an irregular
shape or a rectangle. /ach corner of the figure is expressed as horizontal and vertical
coordinates. 5he coordinates must be from the same coordinate system as for the stops.
5he coordinates of each barrier should be expressed as the northwest4 northeast4
southwest4 and southeast corners of the barrier. 8ore than one barrier may be used in any
one problem.
Note5 5he shape of the barrier must always bulge outward.
RUNNING O!"E
*fter preparing the data screens4 the solver may be activated by clicing on the S&'+e
button. * submenu will appear from which you may select to have ROUTER determine
the route design4 or you may specify the route configuration. 5he latter is useful in
establishing the costs associated with a current route design.
*0
O!"E0De%#$ne( R&ute%
0hen ROUTER designs the routes4 the solution procedure is a heuristic method based on
the savings method of -lare and 0right. ,olution taes place on the database as shown
on the current data screens.
NO"E) If you see the message that there is an I1*D/M+*5/ 1+89/: 2F
N/;I-'/, *N*I'*9'/ 52 -2N/: :2+5/,4 add more vehicle capacity ?large
vehicles or more vehicles@4 even if the current total vehicle capacity exceeds total current
stop volume. 5he additional capacity is needed for intermediate computations4 but will
not necessarily appear in the final solution.
U%er0De%#$ne( R&ute%
5he user has control over the route design through the route designer/editor. 5his
editor serves two purposes. First4 it allows the user to create route designs with the aid of
graphics4 statistics about developing routes4 and whatever principles of good route design
may be available. ,econd4 it allows the user to specify routes4 vehicles4 and stop se&
"uences4 and to edit routes that have been developed. 5his procedure may be used as an
alternative to a ROUTER&designed route.
OUTPUT
5he output is presented in two formsC a report of the routes and associated statistics4 and a
graphical display of the routes. 5he route is displayed graphically upon completion of the
solution process. For the :egals 8etals database4 the resulting route design is displayed
in Figure :2+5/:&2.
*1
F#$ure ROUTER02 Fraphical Display of the :oute Design for :egals 8etals
* report may be obtained by clicing on the Rep&rt button after a solution run is
completed. 5he report provides summary information for all routes as well as detailed
costs and time statistics on individual routes. * sample report is shown in Figure
:2+5/:&$.
ROUTER SO)UTION RE+ORT
)a0el/ E9AM+)E
Date/ #=2='2
Time/ ':1:31 AM

>>> SUMMAR* RE+ORT >>>

TIME=DISTANCE=COST INFORMATION

Route Run Sto. 7r? Stem
Route time@ time@ time@ time@ time@ Start Return No of Route Route
no 8r 8r 8r 8r 8r time time ,to., di,t@Mi -o,t@A
! &"# !#"( 2"2 "1 !&"( !:11AM !!:'+M 2'! "11
'"$ "$ 2"1 "1 #"# !:11AM 1#:&2AM 3 !!&! "11
& (" !"( $"& "1 $" !:11AM 13:1'AM 3 !1' "11
3 #"& 3"3 '"( "1 1"2 !:11AM 13:!#AM # ! "11
Total !&&"$ !1"2 &1"( "1 3$"& !# #!&$ "11
BE5IC)E INFORMATION

Route Be8 ;eig8t DelvrC +i-?u. ;eig8t Cu0e DelvrC +i-?u. Cu0e Be8i-le
no tC. -a.-tC Deig8t Deig8t util -a.-tC -u0e -u0e util de,-ri.tion
! ! &11 &1 1 2$"2E '''' 1 1 "1E TRUC: / !
!1 !1 1 !11"1E '''' 1 1 "1E TRUC: /
& !1 !'1 1 '1"#E '''' 1 1 "1E TRUC: /
3 ! &11 '# 1 '("&E '''' 1 1 "1E TRUC: / !
Total !11 '# 1 '1"2E &'''$ 1 1 "1E

**
DETAI)ED COST INFORMATION
////////Be8i-le//////////// ////////////Driver////////////////
Route Total FiFed Mileage Total FiFed Regular Overtime
no -o,t@A -o,t@A -o,t@A -o,t@A -o,t@A time@A time@A
! "11 "11 "11 "11 "11 "11 "11
"11 "11 "11 "11 "11 "11 "11
& "11 "11 "11 "11 "11 "11 "11
3 "11 "11 "11 "11 "11 "11 "11
Total "11 "11 "11 "11 "11 "11 "11
>>> DETAI) RE+ORT ON ROUTE NUM7ER ! >>>

A TRUC: / ! leave, at !:11AM on daC ! from t8e de.ot at Toledo O5

Sto. Drive Di,tan-e Time
Sto. Arrive De.art time to ,to. to ,to. Dind
No de,-ri.tion time DaC time DaC Min Min Mile, metG
C8i-ago 1#:1AM ! !:1+M ! 31 &1"( $2 *ES
! MilDau?ee 1:1+M ! 1&:11+M ! 31 !!'"2 !11 *ES
De.ot !!:'+M ! /////// // /// #1'" 33
Sto. Sto. volume In- -o,t to ,erve ,to. Ca.a-itC in u,e
No de,-ri.tion ;eig8t Cu0e In A In A=unit ;eig8t Cu0e
2$"2E "1E
C8i-ago !1 1 "11 "1 $"2 "1
! MilDau?ee 1 1 "11 "1 "1 "1
Total, ;eig8t: Del 6 &1 +i-?u., 6 1 Cu0e: Del 6 1 +i-?u., 6 1
Route time: Di,tan-e:
Driving !#"( 8r To !,t ,to. $2 mi
)oad=unload 2"2 From la,t ,to. 33
7rea? "1 On route !11
Total &"# 8r Total 2'! mi
MaF alloDed !$("1 8r MaF alloDed '''' mi
Route -o,t,:
Driver Hreg timeI A"11
Driver Hover timeI "11
Be8i-le HmileageI "11
FiFed "11
Total A"11
F#$ure ROUTER03 /xcerpts from a ,olution :eport for the :egals 8etals -ompany
*+
14S+RUC+1-4S 0-R
14V)4+-R3 C-4+R-* S-0+W(R)
INPOL
INPOL is a software program to compute inventory policies under the reorder *oint
?fixed order "uantity4 variable order interval@ and *eriodic re/ie) ?variable order "uantity4
fixed order interval@ methods of economic order inventory control. /conomic order
"uantity principles are used to find the optimal policies. 5hese policies are a result of
answering two "uestionsC
;ow much to order of a productI
0hen should the product be orderedI
5he policy variables can include replenishment "uantity4 reorder&point "uantity4 time
between inventory level review4 and the target "uantity for orders.
* single echelon4 single inventory is assumed4 which schematically can be represented
as follows.
INPOL will compute4 and optimize if called upon4 the following total cost expression.
5otal cost K )urchase cost J 5ransport cost J -arrying cost
J 2rder processing cost J 2ut&of&stoc cost J ,afety stoc cost
5here are a number of options available to determine inventory policy. 5hese areC
:eorder point or periodic review methods of inventory control may be selected.
5he customer service level may be specified4 or it may be computed if out&of&stoc
costs are nown.
5he order "uantity may be specified or computed.
5he products may be ordered singly or Eointly.
*,
*n average inventory level may be specified to represent existing conditions and the
costs determined.
5hus4 a variety of practical situations may be represented. 5he inventory control methods
used in the INPOL module are discussed in 9usiness 'ogistics/,upply -hain 8anagement
5e.
INPUT
5he input data are entered by means of a data screen editor or /xcel spreadsheet. Four
data sections need to be prepared. /ach section is illustrated using the data from the test
problem shown as product #4 and the data elements on the screens are defined.
E&ample. 5he test problem involves two products that are to be Eointly ordered.
5he inventory level review time is set at #.5 wees for both products. 5he service
indices are specified to be 6.!6 and 6.36 respectively. 2ther data are shown on the
illustrations below.
T.e PARAMETERS AND LA5ELS Se-t#&n
5his section of the data screen has the following elements.
+ro0lem la0el/TEST PROBLEM
Num0er of .rodu-t,: 2
Time frame H!6Calendar daC,@ 6;or?ing daC,@ &6;ee?,@
36Mont8,@ #6*earI: 3
Are t8e item, JointlC ordered H*=NIG Y
Are t8e ,ervi-e indi-e, ,.e-ified H*=NIG Y
Do Cou Dant a ,en,itivitC analC,i, H*=NIG Y
Joint-Order Parameters
4oint/order .ro-urement -o,t: 100
Order -C-le time Hin t8e unit of time frameI: 1.5
Individual-Order Parameter
Are t8e order <uantitie, to 0e ,.e-ified H*=NIG N
* description of each item in this section is as follows.
Nu*/er &8 Pr&(u-t%
,pecify the number of products that are to be analyzed. * maximum of 56 product
items may be processed at one time. If the items are to be Eointly ordered4 data for at least
two items must be entered in the database.
T#*e Fra*e 91:Ca'en(ar (a7%" 2:W&r6#n$ (a7%" 3:Wee6%" ;:M&nt.%" <:Year=:
*-
,elect the time dimension for the demand4 lead&time4 and carrying cost data. 9e sure
that the data are expressed in the same time units throughout. -alendar days represent
$.5 days per year. 0oring days represent 256 days per year4 or about 5 woring days
per wee. 0ees4 months4 and year have their usual definition.
Are t.e Ite*% >&#nt'7 Or(ere( 9Y)N=?
If items are to be ordered separately4 select 1o ?N@. If two items or more are to be
Eointly ordered4 select Les ?Y@. Do not respond with Y and select that order "uantities are
to be specified at the same time.
Are Ser+#-e In(#-e% Spe-#8#e( 9Y)N=?
If you are specifying the customer service levels for the product?s@4 select Les ?Y@. If
the optimum service levels are to be computed4 select 1o ?N@. -omputing the service
levels re"uires that you now the out&of&stoc costs.
D& Y&u Want a Sen%#t#+#t7 Ana'7%#% 9Y)N=?
* sensitivity analysis computes various inventory related costs and parameters for
incremental changes in service levels. Do not use this option if the average inventory is
given values greater than zero ?6@.
>OINT0ORDER PARAMETERS
>&#nt0Or(er Pr&-ure*ent C&%t
5his is a common cost as incurred for processing all Eointly ordered items
simultaneously. It is expressed as R/order.
Or(er C7-'e T#*e
Lou may specify a common order cycle time4 or lead&time4 for all Eointly ordered items.
It must be the dimension as specified in the 5ime Frame. INPOL will compute the
remaining parameters. Lou must first have declared that items are to be Eointly ordered.
5hen4 if order cycle time is specified as zero ?6@4 the optimum review time will be
computed. ,pecify an order cycle time greater than zero ?6@ if you want to fix the review
time at a particular value. +se this option only if items are to be Eointly ordered.
INDIVIDUAL0ORDER PARAMETER
Are t.e Or(er Quant#t#e% t& /e Spe-#8#e( 9Y)N=?
5his selection applies only to separately ordered items. Lou may assign the order
"uantity for an item and INPOL will complete the remaining calculations. :espond with
Les ?Y@. If 1o ?N@ is selected4 INPOL will compute the optimum order "uantity. Lou
must select 1o ?N@ if you have declared 1o ?N@ to items being Eointly ordered.
T.e DEMAND)LEAD TIME DATA Se-t#&n
5his section has the following layoutC
*6
+rd Average Std" dev" Average Std" dev"
no" +rodu-t la0el/// demand of demand lead time of lead time
! +RODUCT ! 111 !11 !"#1 1"11
! +RODUCT #11 2 !"#1 1"11
* description of each item in the section is as follows.
Pr&(u-t La/e'
Five an identifying label to the data. Food practice is to use no more than #5
characters4 however4 do not use a comma ?4@ or a double "uotation mar ?G@ in the label.
:etain the same product se"uence throughout all data screens.
A+era$e De*an(
5his is the itemOs average demand for the period that you selected in PARAMETERS
AND LA5ELS section. It may be the item demand forecast or the average demand over
the period. For example4 average demand is 2666 lb. per year. 9e sure the same time
dimension is used as for lead&time and carrying cost.
Stan(ar( De+#at#&n &8 De*an(
5his is the standard deviation of the demand. It may be the standard error of the
forecast or the standard deviation as calculated from the distribution of demand. Demand
is assumed normally distributed. For example4 the standard deviation of demand is #66 lb.
per year.
A+era$e Lea(0T#*e
5his is the average lead&time as calculated from the distribution of lead&times. 5he
period must be the same as for demand and carrying cost. For example4 the average lead&
time is 6.66!. years ?5 wees@4 but it is expressed in years because demand is in yearly
time units.
Stan(ar( De+#at#&n &8 Lea(0T#*e
5his is the standard deviation of the lead&time distribution. It is assumed that demand
and lead&time distributions are independent of each other4 which may not be the case. *
large standard deviation of lead&time can produce very high safety stoc levels and4
therefore4 high in&stoc probabilities. If the service level for the item is very high4 reduce
this value. For example4 the standard deviation of lead&time is 6.66665 lb. per year. 'ead
times are assumed normally distributed.
T.e PRICE)COST DATA Se-t#&n
5his screen has the following layout.
*7
Tran,/ CarrC/ Order Out/of/
+rd Unit .ort ing .ro-" ,to-?
no" +rodu-t la0el/// .ri-e rate -o,t -o,t -o,t
! +RODUCT ! "# 1"1111 1"11#( 1"1111 !"11
+RODUCT !"'1 1"1111 1"11#( 1"1111 1"2#
* description of each item on the screen isC
Pr&(u-t La/e'
characters and do not use a comma ?4@ or a double "uotation mar ?G@ in the label. :etain
the same product se"uence throughout all data screens.
Un#t Pr#-e
5his is the price paid for the item in inventory. It may be a delivered price or an f.o.b.
factory price. -are must be taen in how the transportation rate is specified. )rice should
be expressed in R/unit. For example4 the unit price is R2.25 per lb.
Tran%p&rt Rate
)rice plus transport rate represent the landed price of the item in inventory. If the
transportation charge is already included in the price4 as it would be for a delivered price4
no transport rate needs to be included. 0ith an f.o.b. factory price4 the transport rate
should be included. It should be expressed in R/unit. For example4 the transport rate is
R6.25 per lb.
Carr7#n$ C&%t
-arrying cost represents such components as capital tied up in inventory4 insurance on
the inventory4 personal property taxes4 obsolescence4 and any other costs that are incurred
due to the level of inventory held. It is expressed as a fraction of the item value per unit of
time. For example4 a carrying cost of $6Q per year would be inputted as a fraction4 i.e.4
6.$6. Be &ure carr-ing co&t i& ex*re&&ed in t'e &ame time dimen&ion a& demand and lead2
time, i.e., i( demand i& in )ee6&, carr-ing co&t &'ould al&o .e in *er2)ee6 unit&.
Or(er Pr&-e%%#n$ C&%t
5his refers to the cost to process a specific item on an order. For Eointly ordered
items4 there may be a common cost in addition to this item cost or instead of it. 5he
common cost is assigned in the PARAMETERS AND LA5ELS section. *n example of
the item order processing cost would be R#.$5 per order.
Out0&80St&-6 C&%t
5his is the cost associated with being out of stoc. It refers to lost profit4 lost future
sales4 or additional processing costs due to bacordering. If a customer service level is
specified4 you do not necessarily have to provide this cost. If no customer service level is
specified4 this cost is re"uired to find the service level and the optimal policy. 5he value
may be set at zero ?6@ if the service level is given. * value greater than zero ?6@ must be
*8
used if no service level is selected in the PARAMETERS AND LA5ELS section. It is a
cost expressed as R/unit4 for example4 R6.55 per lb.
T.e MISCELLANEOUS DATA Se-t#&n
5his screen has the following layoutC
* description of each item in this section is as follows.
Pr&(u-t La/e'
characters and do not use a comma ?4@ or a double "uotation mar ?G@ in the label. :etain
the same product se"uence throughout all data screens.
A+era$e In#t#a' In+ent&r7
If you wish to find the cost of a particular inventory level4 you may specify the average
initial inventory. 5he level is expressed in the same units as demand. For example4 the
average initial inventory is #.3! lb. 5he value is usually set at zero ?6@. ,et the sensitivity
option in the PARAMETERS AND LA5ELS section to 1o ?N@.
Or(er Quant#t7
5his is used to specify a particular order "uantity rather than have INPOL calculate it
for you. 2therwise4 it is left as zero ?6@. 5he units are the same as those used for demand.
For example4 order "uantity is #25 lb. Do not use for Eointly ordered items.
Ser+#-e In(e!
5his is the probability of being in stoc during an order cycle. 5his index is given a
value when there is no out&of&stoc cost provided in the database. It presets the service
level and INPOL minimizes cost based on it. It is expressed as a fraction of #4 e.g.4 6.!6.
+se when service levels are to be specified as a selected option in the PARAMETERS
AND LA5ELS section.
RUNNING INPO*
*fter the data have been prepared on the data screen4 clic on S&'+e to compute the
inventory policy. 5he results are shown as an output report and graphical displays for
various combinations of output variables.
+rd Avg initial Order Servi-e
no" +rodu-t la0el/ inventorC <uantitC indeF
! +RODUCT ! 1 1 1"'1
+RODUCT 1 1 1"(1
*9
Output Rep&rt
+sing the example problem shown under INPUT above4 the following will be
presented in report form. 5he first portion of the report shows the values of the policy
variables. 5here are no values shown under the reorder point policy since this is a Eoint&
order problem and only a periodic review policy is appropriate. ;ence4 we should see.
and annual costs for various cost categories by productC
and a summary of the cost and investment in all products in the databaseC
COMPUTED STOCKING POLICIES
+eriodi- RevieD +oli-C
+rodu-t ! +rodu-t
Average inventorC !@2 &(#
Order <uantitC &@111 2#1
MaF )evel $@ !@#!1
Order revieD time !"#1D? !"#1D?
Turnover ratio $1 $(
Inve,tment A&@(23 A2&
Demand in/,to-? ''"2&E ''"(E
ESTIMATED ANNUAL COSTS
+rodu-t ! +rodu-t
+ur-8a,e -o,t A &3@111 3'@311
Tran,.ort -o,t 1 1
CarrCing -o,t !@1!( !#
Order .ro-" -o,t !@2&& !@2&&
Out/of/,t? -o,t (# &#
SafetC ,t? -o,t !#1 $
SUMMARY DATA
+ur-8a,e -o,t A (&@311
Tran,.ort -o,t 1
CarrCing -o,t !@(
Order .ro-" -o,t &@3$$
Out/of/,t? -o,t &1
SafetC ,t? -o,t !#$
Total -o,t ((@#21
Total inve,tment A 3@$1$
+0
If a sensitivity analysis is selected4 a report of the following type is displayed.
Grap.#-a' D#%p'a7%
,ensitivity plots can be displayed. 5he four choices areC
5otal costs against service level
5otal costs against average inventory
*verage inventory against max level
,ervice level against average inventory
5hese are plots whose data are taen from the sensitivity results in a solution run. *n
example is shown in Figure I1)2'&#.
Sensitivit An!"sis Res#"ts $%& Pe&i%'i( Revie) Sste*
P&%'#(t 1
Servi-e Servi-e MaF RevieD Average Total
indeF level@ E level time@ D? inventorC -o,t@ A
1"#1 '2"21 $@111 !"#1 !@#11 &'@!3!
1"#! '2"2( $@11# !"#1 !@#1# &'"1#$
1"# '2"(3 $@11' !"#1 !@#1' &'@111
" " " " " "
" " " " " "
" " " " " "
1"'( ''"'$ $@&## !"#1 !@(## &2@1&1
1"'' ''"'( $@313 !"#1 !@'13 &2@1&(
!"11> !11"11 !@2&(@1#! !"#1 !@2&&@##! !@31$@1&$
>5ig8 value, for a ,ervi-e indeF of !"11 re.re,ent infinitC"
+1
F#$ure INPOL01 )lot of 5otal -ost *gainst ,ervice 'evel
+*
14S+RUC+1-4S 0-R
C)4+)R#-0#/R(V1+3 0(C1*1+3 *-C(+-R
COG
COG is computer software to locate a single facility by means of the exact center&of&
gravity method. 5he problem is one where a single facility4 such as a warehouse4 is to
serve ?or be served by@ a number of demand ?or supply@ points with nown locations and
volumes. 5he obEective is to find a location such that total transportation cost4 as
represented in the following expression4 is minimizedC
( ) ( )
[ ]
78 9 R 1 X X Y Y
i i i i
i
N
7
= +
=
2 2
#
where
78 K total transportation cost
N K the number of origin/destination points in the problem. +p to 566 points may be
used.
Xi,Yi K the geographical locations of the origin/destination points represented using linear
X,Y coordinate points.
7 K a power factor in the distance computation formula. Distances are computed
from coordinate points using the following formula.

( ) ( )
[ ]
Distance K 1 X X Y Y
i i
7
+
2 2
where
X Y
i i
4
represent the origin/destination points and X Y 4 represent the facility.
5he power factor 7 controls the linearity of the distance between points. 5he
value of 7 is usually 6.54 which is a straight line between points.
1 K a scaling factor to convert coordinate distances to miles.
9 K the volume of an origin/destination point in any appropriate demand units.
R K the transportation rate between the facility to be located and the
origin/destination points4 expressed in a monetary unit per unit of volume per
unit of distance4 such as R/unit/mile.
For bacground on the center&of&gravity method used in the COG module4 see
-hapter #$ of 9usiness 'ogistics/,upply -hain 8anagement 5e.
INPUT
5he data inputs consist of coordinates for locating origin/destination points4 volumes of
origin/destination points4 transportation rates between the facility and the
origin/destination points4 and miscellaneous factors.
++
EXAMPLE
,uppose that we have a small problem as shown in Figure -2F&#. 5he product is
-hemicals. 5here are #6 marets to be served from a single warehouse location. 5he
warehouse is supplied from a single plant. 5he total amount of product shipped by the
plant is the sum of the volume demanded by the marets. 5he product is shipped over
road networs. 5he annual volumes of the marets and the transportation rates are given
as followsC
Point,
i
9olume,
l..
Rate,
$/l../mile
8
#
$46664666 6.6626
8
2
546664666 6.66#5
8
$

#(46664666 6.6626
8
%
#246664666 6.66#$
8
5
!46664666 6.66#5
8
.

#646664666 6.66#2
8
(
2%46664666 6.6626
8
3
#%46664666 6.66#%
8
!
2$46664666 6.662%
8
#6
$646664666 6.66##
#%(46664666
)
#
#%(46664666 6.6665
'ocate the single warehouse so that transportation costs are minimized.
5he inputs to COG would loo lieC
+ro0lem la0el: EFam.le
+oDer fa-tor HTI: "#
Ma. ,-aling fa-tor H:I: #1
+oint +oint 9 -oor/ * -oor/ Tran,.ort
no" la0el//// dinate dinate Bolume rate
! M! "11 !"11 &111111 1"111
M #"11 "11 #111111 1"11!#
& M& '"11 !"11 !2111111 1"111
3 M3 2"11 3"11 !111111 1"11!&
# M# "11 #"11 '111111 1"11!#
$ M$ !1"11 #"11 !1111111 1"11!
2 M2 "11 2"11 3111111 1"111
( M( 3"11 2"11 !3111111 1"11!3
' M' #"11 ("11 &111111 1"113
!1 M!1 ("11 '"11 &1111111 1"11!!
!! +!! '"11 $"11 !32111111 1"111#

+,
F#$ure COG01 'ocations of 8aret )oints ?M i@ and )lant ?P@ on a 'inear Frid.
RUNNING #O+
:unning COG re"uires you to first create the database for a particular location problem.
5hen clic on the S&'+e button to compute the center&of&gravity coordinates. Lou may
choose to have COG compute the facility location coordinates4 or you may specify these
coordinates. If you choose have the coordinates computed4 the simple center&of&gravity is
found. 5o see if this initial location can be improved upon4 as for additional
computational cycles. 0hen there is little or no change in costs between successive
cycles4 as for no further computations. :ead the results from the screen or print them.
For the example problem4 the results will appear as shown in Figure -2F&2. *fter 56
computational cycles4 the best location for the facility is
X Y = = . 2!3 .%3% . 4 .
for a total annual transportation cost of R5546#5465(. *t this point4 you may as for the
points and the facility location to be plotted on a linear grid by selecting this option from
the master menu. ,uch output is shown in Figure -2F&$.

0 1 * + , - 6 7 8 9 10
1
*
+
,
-
6
7
8
9
10
0
o
o
o
o o
o
o
o
M
M
M M
M
M
M
M
M
M
1
6
7 8
9
10
-
,
*
+
"!&e2 1 = -0 mi&es
X
3
o
o
P
1
o
+-
F#$ure COG03 )lot of 2ptimized Facility 'ocation for /xample )roblem.
COG
LOCATES A $ACILITY BY T+E E,ACT CENTER O$ GRA-ITY
MET+OD
Iteration K K
num0er 9 -oord * -oord Total -o,t
1 $"# #"'$' ##@3$'@#'&
L/ COG
! $"&$1 $"&1$ ##@1$!@!($
$"&21 $"31' ##@13@!1#
& $"&#( $"333 ##@1!(@23'
3 $"&33 $"3#' ##@1!$@'#&
# $"&& $"3$2 ##@1!$@1$(
$ $"&& $"32 ##@1!#@#''
2 $"&!2 $"32# ##@1!#@&3(
( $"&! $"322 ##@1!#@!3
' $"&1( $"32' ##@1!#@!3!
!1 $"&1$ $"3(1 ##@1!#@!1
!! $"&13 $"3(! ##@1!#@1(!
! $"&1 $"3( ##@1!#@121
!& $"&1! $"3(& ##@1!#@1$3
!3 $"&11 $"3(& ##@1!#@1$!
!# $"&11 $"3(& ##@1!#@1#'
!$ $"&11 $"3(& ##@1!#@1#(
!2 $"'' $"3(3 ##@1!#@1#(
!( $"'' $"3(3 ##@1!#@1#(
#1 $"'( $"3(3 ##@1!#@1#2
F#$ure COG02 -omputational :esults for the /xample )roblem.
+6
14S+RUC+1-4S 0-R
C)4+)R#-0#/R(V1+3 MU*+1.*) 0(C1*1+3 *-C(+-R
MULTICOG
MULTICOG is computer software to locate multiple facilities by means of the exact
center&of&gravity method. 5he problem is to locate one or more facilities ?sources@4 such
as warehouses4 to serve a number of demand points ?sins@ of nown locations4 volumes4
and transportation rates. 5he number of facility locations is specified. 5he obEective is the
find the coordinates of the facilities such that the following expression is minimized.
( ) ( )
78 9 R 1 X X Y Y
i: i: i : i :
i
N
:
M
= +
= =

2 2
# #
where
78 K total transportation cost
i K demand ?sin@ point number up to a total of N
: K facility ?source@ number up to a total of M
9
i:
K volume associated with a demand point
R
i:
K transport rate to a demand ?from a supply@ point

X Y
i i
4
K coordinates of a demand ?or supply@ point
X Y
: :
4 K coordinates for the facility location :.
1 K scaling factor to convert coordinates to distance units. 8ultiply 1 by #.2#
to approximate road distance and #.2% to approximate rail distance.
For bacground on the center&of&gravity method used in the MULTICOG module4
see -hapter #$ of 9usiness 'ogistics/,upply -hain 8anagement 5e.
INPUT
5he inputs to the program consist of ?#@ a problem descriptor4 ?2@ coordinates for demand
points4 ?$@ volumes of demand points4 ?%@ transportation rates between the facility and the
demand points4 and ?5@ a scaling factor. ,pecificallyC
* problem descriptor using any combination of letters and numbers
5he number of demand points in the problem. +p to 566 may be used.
5he geographical locations of the demand points represented by linear X,Y
coordinate points. *ny linear grid system may be used.
5he number of facility locations to be analyzed. +p to 26 locations may be used.
* scaling factor to convert coordinate distances to miles
5he volume of a demand point in any appropriate units
5he transportation rate between the facility to be located and the demand points
expressed in R/unit/mile or other distance measure
*ll input can be prepared from the screen editor or using an /xcel spreadsheet.
+7
EXAMPLE
,uppose we have a small problem as shown in Figure 8+'5I-2F&#. 5he product is
-hemicals. 5here are #6 marets that being served from two warehouse locations. 5he
product is shipped over road networs. 5he annual volumes of the marets and the
transportation rates are given as followsC
Point,
i l..
Rate,
$/l../mile
8
#
$46664666 6.6626
8
2
546664666 6.66#5
8
$

#(46664666 6.6626
8
%
#246664666 6.66#$
8
5
!46664666 6.66#5
8
.

#646664666 6.66#2
8
(
2%46664666 6.6626
8
3
#%46664666 6.66#%
8
!
2$46664666 6.662%
8
#6
$646664666 6.66##
#%(46664666
'ocate the two warehouses so that annual transportation costs are minimized.
5he input to MULTICOG would loo lieC
L&-at#&n Data
//+ARAMETERS AND )A7E)S//
Ma. ,-aling fa-tor H:I: #1"11
//)OCATION DATA//
+oint +oint 9 -oor/ * -oor/ Tran,.ort
no" la0el///////// dinate dinate Bolume rate
! M! "11 !"11 &111111 1"111
M #"11 "11 #111111 1"11!#
& M& '"11 !"11 !2111111 1"111
3 M3 2"11 3"11 !111111 1"11!&
# M# !1"11 #"11 !1111111 1"11!
$ M$ "11 #"11 '111111 1"11!#
2 M2 "11 2"11 3111111 1"111
( M( 3"11 2"11 !3111111 1"11!3
' M' #"11 ("11 &111111 1"113
!1 M!1 ("11 '"11 &1111111 1"11!!
+8
0 1 * + , - 6 7 8 9 10
1
*
+
,
-
6
7
8
9
10
0
o
o
o
o o
o
o
o
X
X
M
M
M M
M
M
M
M
M
M
1
6
7 8
9
10
-
,
*
+
"!&e2 1 = -0 mi&es
X
3
F#$ure MULTICOG01 'ocations of 8aret )oints ?
M
i
@ on a 'inear Frid.
RUNNING M!*"I#O+
*fter preparing a database4 clic on S&'+e to find the facility coordinate points. 5hen4
select how many facilities are to be located. ,elect whether you wish the program to find
the locations or whether you want to specify the location coordinates. If the latter is
chosen4 you will need to assign each demand point to a facility location. :ead the results
from the screen or direct them to a printer. 5he results may be displayed graphically by
clicing on the P'&t button.
* computed solution for two warehouses in the example problem is shown in Figure
8+'5I-2F&2. 5he first location is to be located at XK5.64 YK3.64 should serve marets
.4 (4 34 !4 and #6. 0arehouse 24 which is to be located at X!K!.64 Y!K#.64 is to serve
marets #4 24 $4 %4 and 5. 5he suggested locations are plotted in Figure 8+'5I-2F&$.
+9
F#$ure MULTICOG03 * )lot of the ,olution for 5wo 0arehouses for the /xample
)roblem.
.. PROBLEM
SOLUTION ..
Title: E9AM+)E +RO7)EM
Sour-e 9/Coordinate */
Coordinate Bolume
Co,t
! 3"'''(& 2"'''(2&
!11@111@111 !2@1#2@1&
("''212 !"113!&3
32@111@111 (@'&@231
Sour-e Allo-ated demand
.oint, to ,our-e .oint,
! $ 2 ( ' !1
! & 3 #
Total -o,t 6 #@'('@2$3"11
F#$ure MULTICOG02 :eport of ,olution :esults.
,0
14S+RUC+1-4S 0-R
.#M),1(4 MU*+1.*) 0(C1*1+3 *-C(+-R
PMED

PMED is computer software to locate multiple facilities by means of the )&median
approach. 5he problem is to locate one or more facilities ?sources@4 such as warehouses4
to serve a number of demand points ?sins@ of nown locations4 volumes4 and
transportation rates. Fixed costs of a candidate set of facilities may also be nown. 5he
candidate set of facilities is selected from the demand points. 5he obEective is to find the
best P locations from the M candidate sites where P is less than or e"ual to M. 5he )&
median problem is expressed asC
Minimi,e 78 9 R d X ; Y
X Y i
X i :
X *
X i :
i i i: i: : :
: : i
i: :
:
i:
i:
:
i:
= +
=
=
=
=

subEect toC
?for all @
7 ?for candidate 4 pairs@
?for all 4 pairs@
L
# if facility is open
6 if closed
iE
E
? 4 @ 01
whereC
X
i :
X
:
Y
:
78
i N
: M
9 i
R i
d i :
*
i :
: :
:
i
i
i:
4
4
4
=

=
=
=
=
=
=
=
1 if demand or supply node is assigned to facility
64 if otherwise
#4 if a facility is sited at node
64 if otherwise
#4 if is open
64 if not open or a candidate site
total transportation and facility fixed costs
demand or supply point ?sin@ number up to a total of
candidate facility ?source@ number up to a total of
volume of demand or supply point
transportation rate associated with demand or supply point
distance between demand or supply point and facility
number of facilities to locate
For more discussion on the )&median method used in the PMED module4 see -hapter
#$ of 9usiness 'ogistics/,upply -hain 8anagement 5e.
,1
INPUT
5he inputs to the program consist of ?#@ a problem descriptor4 ?2@ coordinates for demand
and supply points4 ?$@ volumes of demand and supply points4 ?%@ transportation rates
associated with a demand or supply point4 ?5@ a candidate list of facilities from which the
number of locations are selected for analysis4 ?.@ fixed costs for candidate facilities4 ?(@
type of coordinate system used4 and ?3@ map scaling factors. ,pecificallyC
* problem descriptor using any combination of letters and numbers
5he number of demand and supply points in the problem. +p to a maximum of .5
are allowed.
5he geographical locations of demand and supply points are represented as linear
grid coordinates or as latitude&longitude coordinates. If latitude&longitude
coordinates are used4 they should be expressed in degrees. ,pecify the type of
coordinate system to be used.
5he particular demand or supply points representing a candidate set of potential
facility locations. +p to #5 points may be in the candidate list. 5hese sites are
identified with an 5 in the database. 1on&candidate sites are left blan.
,caling factors for the linear coordinates to convert them to a distance measure such
as miles. ,caling factors may include a correction for actual road distance as a
multiplier on the computed straight&line distance. 5ry a multiplier of #.2# for linear
coordinates for intercity travel. * circuity factor can be applied to latitude&longitude
coordinate distances to account for actual traveled distance. 5ry a factor of #.2 for
intercity travel and #.%% for travel within cities. ,ee 5able #%&% of 9usiness
'ogistics/,upply -hain 8anagement 5e for additional circuity factors.
5he volume of a demand or supply point in any appropriate dimension
5he transportation rate for a demand or supply point expressed in R/unit/mile or
other distance measure.
Fixed cost for a candidate facility
*ll input data are prepared on the screen data editor or using an /xcel spreadsheet.
EXAMPLE
,uppose that we have a small problem as shown in Figure )8/D&#. 5here are #2
marets being served from up to 5 candidate warehouse locations. 5he product is shipped
over a road networ. 5he annual volumes of the marets4 the transportation rates4 and the
candidate sites with their fixed costs are shown in the figure. 9e sure to select the type of
coordinates being used to locate the points in the problem.
RUNNING PME$
*fter preparing a problem database4 specify the num.er of facilities to locate. 5his
number must be less than or e"ual to the number of candidate sites indicated by an 5 in
the problem database. 1ext4 clic on the S&'+e button to find a problem solution. 5he run
results are shown in Figure )8/D&2. * graphical display of the solution is shown in
Figure )8/D&$.
,*
F#$ure PMED01 * ,mall 'ocation )roblem with Five -andidate ,ites
F#$ure PMED02
:eport on :un
:esults.
S-*U+1-4 R)SU*+S 0-R .#M),1(4 .R-&*)M
N%. $!(i"it n!*e/ -%"#*e Assi0ne' n%'e n#*1e&s
1 M1 32000 1
2 M3 342000 2 3 3 5 11
3 M5 332000 6 5
3 M7 672000 4 7 10 12
T%t!" 1532000
T%t!" (%st8 932337.20
,+
F#$ure PMED03 Fraphical Display of ,olution :esults
If latitude&longitude coordinates are specified4 a map of the +.,. will be displayed
instead of the linear grid. *n example is shown in Figure )8/D&%. 2ther geographic
maps are not available.
F#$ure PMED0; Fraphical Display of :esults when +sing 'atitude&'ongitude
-oordinates.
14S+RUC+1-4S 0-R S-0+W(R) +- (CC-M.(43
,,
+%) US)M-R) S-(. C-M.(43 C(S) S+U,3
WARELOCA
WARELOCA is a computer program specifically designed to assist in the analysis of the
+semore ,oap -ompany case study. +sers should refer to the case study in -hapter #% of
9usiness 'ogistics/,upply -hain 8anagement 5e for bacground and data.
5he program solution procedure is based on linear programming. 5he user selects the
particular scenario of plants4 warehouses4 and customer service level to be considered4 and
the program optimally finds the best allocation of demand to the warehouses and plants.
Nariable costs are minimized4 subEect to customer service and plant capacity constraints.
Fixed costs are not handled within the solution process and must be added to the solution
results.
INPUT
5he database for the case study has been prepared and stored in a file called +ma6#.Dat.
-all it by using the Open 8#'e button. 5he data should only be temporarily changed during
a computer run. If it is desired to save data changes4 it is good practice to save them to a
file of a different name from +ma6#.Dat to preserve the original data.
5he number of customers4 warehouses4 and plants are limited to those in the databaseB
however4 the data associated with each may be altered as desired. :emember that demand
and capacities are expressed in hundredweight ?cwt.@.
RUNNING ,-E*O#-
5o mae a solution run4 it is first necessary to select the plants and warehouses to be
evaluated. Do this by placing an 5 beside the appropriate facility on the data screen. 9e
sure that there is ade"uate capacity among the selected facilities to satisfy all demand.
1ext4 indicate how far a customer is allowed to be from a warehouse. For example4 if
the distance is $66 miles4 which approximates #&day delivery service4 WARELOCA will
loo for all the warehouses in your list mared with an 5. From among those that are
within $66 miles of the customer4 it will select the warehouse that can lea&t ex*en&i/el-
serve the customer. If no warehouse is within $66 miles of the customer4 WARELOCA
will find the warehouse that is clo&e&t to the customer and assign the customer to it4
irrespective of cost. 5hat is4 service considerations will override cost when the desired
service distance cannot be met.
Lou may change the demand by region of the country. 5he demand data in the
+ma6#.Dat database is for the current year. 5he region in which each customer resides is
shown in the customer data. For example4 region # refers to the 1ortheast section of the
country. Demand growth factors may be applied to customer demand by region.
)revious run scenarios are retained4 and you may alter them incrementally. 2nce the
scenario to be evaluated is prepared on the screen4 the selected list of plants4 warehouses4
and service level is saved.
5o find a solution4 clic on the S&'+e button.
OUTPUT
5he results of a solution run can be obtained in two forms. 5he first is a report showing
costs4 facility throughput4 and product flow paths through the logistics networ. 5he
second is a graphical display of the customer and facility locations4 and the run results.
,-
14S+RUC+1-4S 0-R +%)
.R-,UC+ *(3-U+ .R-/R(M
LAYOUT
LAYOUT is a program for laying out products in warehouses and other facilities. 5he
methods used are ?#@ by popularity4 ?2@ by cube4 and ?$@ by cube&per&order index.
)roducts are allocated to limited space so that total travel cost to retrieve items is
minimized. * picing trip originates at the outbound doc and returns to the same point.
2nly one product type is piced per trip4 but more than one item may be retrieved on a
trip. LAYOUT ran orders products in the way they should be allocated to the available
space beginning at the outbound doc. For bacground on the layout methods used in the
LAYOUT module4 see -hapter #2 of 9usiness 'ogistics/,upply -hain 8anagement 5e.
INPUT
*n input file is prepared with the use of the data editor. 2pen a previous file or start a
new one. First4 label the problem4 avoiding the use of commas ?4@ and double "uotation
mars ?G@. 5hen4 select the method for layout. Finally4 enter ?#@ the annual sales of each
product in units per year4 ?2@ the size of a product unit in cubic feet4 ?$@ the inventory
turnover for each product4 and ?%@ the number of orders per year on which the product
appears. * typical input file might loo lie that in Figure '*L2+5&#.
F#$ure LAYOUT01 5ypical Data /ditor ,creen
RUNNING *-.O!"
5o execute a run4 clic on the S&'+e button. 5he results report will show the products
ran ordered with those products being allocated to space closest to the outbound doc
listed first. 5he re"uired cubic space for each product is also computed to aid in allocating
the products to the warehouse space.
,6
EXAMPLE
* warehouse contains six storage bays as shown in Figure '*L2+5&2. /ach bay has
#4%66 s". ft. of storage area4 and product is staced #. feet high. Data on five products
was given previously as an example of input data ?Figure '*L2+5&#@. 5he five products
are to be allocated to the available space by means of the cu.e2*er2order index method.
)roducts with small values of the index are to be located nearest the outbound doc. 5he
computed results are shown in 5able '*L2+5&#. 5he results are used to develop the
layout as shown in Figure '*L2+5&2.
Ta/'e LAYOUT01 'ayout of the /xample )roblem by -ube&)er&2rder Index.
)A*OUT 7* CU7E/+ER/ORDER INDE9
+rodu-t No" of Item ,ale,@ Item ,iMe@ Cu0e/.er/ Re<" ,.a-e@
Ran? name///// order,=Cr" unit, -u" ft" order indeF -u" ft"
! +RODUCT $ #@111 #@111 !!"#1 (31 !!@#11
+RODUCT # 3&1 &@#11 !1"$1 @2(2 &@(&
& +RODUCT !11 1@111 &"1 $@32 2@!'!
3 +RODUCT ! '1 !1@111 3"#1 &@#(' (@1&$
# +RODUCT 3 &11 #1@111 &"#1 &'@3' &@312
$ +RODUCT & 11 31@111 #"#1 !!(@1(( $3@21$
6 5 1 4 6 2 4
4 4 3 3
3 3
Inbound dock
Outbound dock
Bay
1,400 sq. ft.
16 ft. high
F#$ure LAYOUT02 'ayout of the Five )roducts in the /xample )roblem.
,7
14S+RUC+1-4S 0-R
C-M.U+14/ ,1S+(4C)S 0R-M C--R,14(+) .-14+S
MILES
MILES is a computer program to compute distances from latitude&longitude coordinates
using the great circle formula and from linear coordinate points using the )ythagorean
5heorem. For bacground on distance computation4 see -hapter #% of 9usiness
'ogistics/,upply -hain 8anagement 5e.
INPUT
Inputs are in the form of ?#@ a map scaling factor4 ?2@ circuity factor4 ?$@ distance
dimension4 ?%@ coordinate type4 and ?5@ origin and destination coordinate. *n example of
a problem setup is shown in Figure 8I'/,&#.
F#$ure MILES01 *n /xample of the 5ypical Data for a :un ,etup.
5he map&scaling factor is used to convert coordinate points to miles or ilometers.
+se a scaling factor of # for latitude&longitude coordinates. 5he circuity factor is a
multiplier that converts coordinate&calculated distance to estimated road4 rail4 or other
distance. 5ypical circuity factors areC #.2# for road and #.2% for rail4 if linear coordinates
are used. * factor of #.26 is appropriate for road distances if latitude&longitude
coordinates used. * circuity factor for any situation can be found by averaging a sample
of the ratio of actual distance between point pairs to the calculated distance from
coordinate points. *dditional circuity factors may be found in 5able #%&% of 9usiness
RUNNING MI*ES
,8
-lic on Ca'-u'ate to see the computed distance displayed on the screen. Lou may then
elect to change the origin point4 the destination point4 or both4 and compute again. 5he
distances for subse"uent calculations are accumulated. ,ee Figure 8I'/,&# for an
example of calculation results.
,9
14S+RUC+1-4S 0-R RU4414/ +%)
+R(4S.-R+(+1-4 M)+%-, -0 *14)(R .R-/R(MM14/
TRANLP
TRANLP is a software module that solves the standard =transportation problem> of linear
programming. It will handle a problem of up to $6 rows and $6 columns.
)repare a data matrix by opening an existing file or choosing a new file name for
entering data. File names are of the form 5ran6#.dat. 5he data editor of the type shown
in Figure 5:*1')&# will be displayed. *dEust the matrix size by entering the number of
rows or columns to create the desired problem size. -lic on the appropriate row4 or
column4 for the label to be changed. /nter the desired descriptor in the label box.
F#$ure TRANLP01 Data ,hell on /ditor ,creen with /xample )roblem
5he problem structure follows the standard form of the transportation method of linear
programming. 1o dummy row or column is re"uired. 5he program will add one or the
other when supply and demand are not e"ual. 5o prepare a database4 you will need to
define the following elements.
1umber of sources ?rows@
1umber of sins ?columns@
*vailable supply associated with each source point
:e"uirements for each sin
-ell costs by source and sin
:efer to any good boo on management science for a discussion of the transportation
method.
EXAMPLE
* problem with two sources of supply and three sins might have the matrix setup as
shown in Figure 5:*1')&#. ,upply and re"uirements values should be entered as whole
?integer@ numbers. -ell costs may be entered as decimal values4 if desired. *void
extremely large ?or small@ values in the matrix4 since this may exceed the computer
capabilities. :escale the data4 if necessary.
-0
RUNNING "-N*P
*fter the data have been prepared4 clic on the S&'+e button to execute. If errors are
detected in the input data4 error messages will appear.
OUTPUT
2nce the program has executed4 the results will appear on the screen as shown in Figure
5:*1')&2. *llocation "uantities are shown in the body of the matrix. * line report may
also be obtained by clicing on the Rep&rt button. 5his report for the example problem is
shown in Figure 5:*1')&$.
F#$ure TRANLP02 ,olution ,creen
F#$ure TRANLP03 -omputational :esults of the /xample )roblem
O+TIMUM SU++)* SC5EDU)E
/////////// Cell //////////// Unit Cell Unit,
Sour-e name Sin? name -o,t -o,t allo-ated
A ! #"11 &@#11"11 211
A 3"11 "11 1
A & 2"11 "11 1
Total, &@#11"11 211
Sour-e -a.a-itC 6 !@111
Sla-? -a.a-itC 6 &11
7 ! "11 311"11 11
7 !"11 211"11 211
7 & &"11 &@&11"11 !@!11
Total, 3@311"11 @111
Sour-e -a.a-itC 6 @111
Sla-? -a.a-itC 6 1
Total allo-ated 6 @211
Sla-? re<uired 6 @211
Total -o,t 6 2@'11"11
-1
14S+RUC+1-4S 0-R
*14)(R .R-/R(MM14/
LNPROG
LNPROG is computer software to solve linear programming problems by means of the
two&phase simplex method. )roblems of up to 56 constraints and #66 variables can be
solved. :efer to any good boo on management science for a discussion of the linear
programming method.
INPUT
2pen a previously prepared file or start a new one. 5he inputs to the program consist of
?#@ the number of constraints and variablesB ?2@ the cost coefficientsB and ?$@ the :ight
;and ,ide ?:;,@ values. 5he problem setup for an example problem is shown in Figure
'1):2F&#. Note5 5o reconfigure the matrix for a problem size different than appears on
the screen4 first clic on the Re%#@e button and then change the number of constraints and
number of variables.
F#$ure LNPROG01 Data ,etup ,creen for an /xample )roblem.
9efore preparing the linear programming problem data4 it is necessary to convert any
problem into a standard form. 5hat is4
?#@ ,tate the obEective function as a minimi,ing problem. If the obEective is to
maximize4 multiply the obEective function coefficients by a minus one ?#@
?2@ 8ae all right&hand&side values ?:;, values or =re"uirements>@ of constraints
positive or zero
?$@ 1ote whether each constraint is of the 4 K4 or type
EXAMPLE
,uppose we have the following problem.
MaFimiMe '1X
1

% 21X
2
% (3X
3

subEect toC
-*
!1X
1
% !(X
2
% (9
&
#111
"#X
1
% &X
2
% &X
3
!11
!X
1
% !X
2
% !X
3
$11
/ !X
2
/!11
All X
i
1
0e first standardize the problem. 5he fourth constraint is transformed to X! #66 to
mae a positive right&hand side value. 5he obEective function is multiplied through by a
# to convert it to a minimi,ing problem. 5hus4 the problem statement would now appear
asC
Min /'1X
1
/21X
2
/(3X
3
subEect toC
!1X
1
% !(X
2
% (9
&
#111
"#X
1
% &X
2
% &X
3
!11
!X
1
% !X
2
% !X
3
$11
!X
2
!11
In matrix form4 the problem now appears as setup in Figure '1):2F&#.
RUNNING *NPO+
*fter the data input screen has been prepared4 clic on the S&'+e button to solve the
problem. * data validation chec will be performed4 so there may be some error
messages. 5he solution is for the data as they appear on the screen and not necessarily the
data as may be saved in a file. 5he results for the example problem are given in Figure
'1):2F&2.
+ro0lem la0el: E9AM+)E +RO7)EM
SUMMAR* OF RESU)TS
7a,i, A-tivitC Non0a,i, O..ortunitC
varia0le, level varia0le, -o,t
9H!I 31 // //
9HI !11 // //
9H&I !11 // //
// // 9H3I $
// // 9H#I !
9H$I !$1 // //
// // 9H2I 23
// // 9H(I /23
O0Je-tive fun-tion value HNI 6 &2111
F#$ure LNPROG02 ,olution :esults for the /xample )roblem.
-+
14S+RUC+1-4S 0-R
M15), 14+)/)R .R-/R(MM14/
MIPROG
MIPROG is computer software to solve mixed integer linear programming problems by
means of the branch and bound method. )roblems of up to 56 constraints and #66
variables can be solved. Integer programs are sensitive to problem configurations and
even slight changes to a problem can cause a long running time or an infeasible solution to
occur. :efer to any good boo on management science for further discussion of the
integer linear programming method. *n example of an integer programming application is
shown in Figure #$&5 and the 5echnical ,upplement to -hapter #$ of 9usiness
INPUT
2pen a previously prepared file or start a new one. 5he inputs to the program consist of
?#@ constraints and variablesB ?2@ the cost coefficientsB and ?$@ the :ight ;and ,ide ?:;,@
values4 ?%@ constraint type4 and ?5@ variable type. 5he problem setup for an example
problem is shown in Figure 8I):2F&#. Note5 5o reconfigure the matrix for a problem
size different than appears on the screen4 first clic on the Re%#@e button and then change
the number of constraints and number of variables.
F#$ure MIPROG01 Data ,etup ,creen for an /xample )roblem.
9efore preparing the integer programming problem data4 it is necessary to convert any
problem into a standard form. 5hat is4
?#@ Declare the obEective function to be a minimi,ing or maximi,ing problem
?2@ 8ae all right&hand&side values ?:;, values or =re"uirements>@ of constraints
*o&iti/e or ,ero
?$@ 1ote whether each constraint is of the 4 K4 or type
?%@ Declare whether each variable is of the type $%A&'4 (&)4 or 0*1. * $%A&'
refers to the variable taing on values greater than or e"ual to zero4 but being
unrestricted in size. 5he variable may tae on fractional values. * (&) refers to
the variable having a value greater than or e"ual to zero but being unrestricted in
-,
size4 however4 the variable must be an integer ?no fraction allowed.@ * 0*1 refers
to the variable being allowed to tae on only two possible values6 or #.
EXAMPLE
,uppose we have the following problem.
Maximize !X
1

% X
2
% !X
3

subEect toC
2X
1
% 3X
2
% &9
&
(
3X
1
% 2X
2
% X
3
(
All X
i
1@ and X
1
and X
2
are integer
5he problem now is entered into the editor screen as shown in Figure 8I):2F&#. 5he
/xcel spreadsheet editor also may be used.
RUNNING MIPO+
*fter the data input screen has been prepared4 clic on the S&'+e button to solve the
problem. 5he solution is for the data as they appear on the current screen and not
necessarily the data as may be saved in a file. 5he results for the example problem are
given in Figure 8I):2F&2.

O+TIMA) SO)UTION

Baria0le Balue Rate Co,t Baria0le la0el
9H!I 6 "1111 !"1111 "1111 Bar O!
9HI 6 "1111 "1111 3"1111 Bar O
9H&I 6 $"$$$2 !"1111 $"$$$2 Bar O&
O0Je-tive fun-tion value 6 !1"$2
F#$ure MIPROG02 ,olution :esults for the /xample )roblem.
--
14S+RUC+1-4S 0-R +%)
MU*+1.*) R)/R)SS1-4 .R-/R(M
MULREG
MULREG is a stepwise multiple regression program that is designed to handle up to 5
variables and !!! observations. 8ultiple regression analysis as a techni"ue is not
described in these instructions. 5he reader is referred to any good boo on statistics
covering multiple regression analysis.
INPUT
2pen a previously prepared file or start a new one. * data screen of the type shown in
Figure 8+':/F&# is to be prepared. /ach data item is discussed.
F#$ure MULREG01 /xample of Data Input ,creen.
Problem label. 5he problem label is an identifier for the problem. Do not use
commas ?4@ or double "uotation mars ?G@ in the label. Seep the descriptor to #5
characters or less.
Number of observations. /ntering the number of observations in a problem changes
the size of the data input matrix.
/ariable label. /ach variable may be identified with a label. Seep the descriptor to
#6 characters or less. Do not use commas ?4@ or double "uotation mars ?G@ in the labels.
/ariable type. +p to five variables are allowed. /ach variable is identified as ?#@ a
dependent variable ?D@ or ?2@ an independent variable ?I@. 2nly one variable can be
-6
declared an dependent variable. 5he remaining four variables may be independent or left
blan.
Step0ise regression. Invoing the stepwise procedure allows the independent
variables to be selected based on their contribution to reducing the error in the dependent
variable. If the stepwise procedure is not used4 the regression analysis is conducted
simultaneously on all independent variables.
0hen the stepwise option is selected4 entering and leaving variables may be controlled
according to their statistical contribution. If a zero ?6@ F&level value is given for both
entering and leaving variables4 all variables will be brought into the regression e"uation.
Incremental F&level values greater than zero may cause some variables to be omitted from
the regression e"uation.
"ransformed input data. Invoing the transform data option allows specified
variables to be transformed according to several predefined functions. 5he particular
variable to be transformed is selected by clicing on the appropriate column
29,/:N*5I21, matrix. 5hen4 clic on the Tran%8&r* button and select the desired
transform function.
Print data. * listing of the problem setup and data can be obtained by clicing on the
Pr#nt (ata button.
Plot data. -licing on the P'&t (ata button invoes plotting capability. * plot of the
dependent variable against the selected independent variable can be seen. 5he plot is made
after any transformation of a variable has been carried out.
Error analysis. )rovides additional information about how well the regression line fits
the data. 5his invoes a plot of residuals in the output report.
O/%er+at#&n Data
5he observations for each variable are entered on the screen. 5here may be a
maximum of !!! entries. Data (or at lea&t t)o /aria.le& mu&t .e *ro/ided. ,caling of the
data can be important to eep it within the range of the programOs plotting and
presentation capabilities. * good rule of thumb is to scale the data so that it has a range of
approximately xxx.x. If the output presentation is not satisfactory4 try re&scaling the data
with the editor or the transformation procedure.
RUNNING M!*E+
*fter the data are prepared with data editor4 execute the analysis by clicing on the S&'+e
button. * report of the solution results will appear.
EXAMPLE
,uppose that we wish to develop a curve to estimate class transportation rates as a
function of distance. From rate data collected for 5666 lb. shipments over varying
distances4 we have 2! observations. * listing of the input data is shown in 5able
8+':/F&# and its graphical presentation is given in Figure 8+':/F&2. 5he results of
analysis are illustrated in Figures 8+':/F&$ to 8+':/F&5.
-7
Ta/'e MULREG01 2bservation Data for /xample problem
From running MULREG" the rate estimating e"uation that we see isC
Rate K #2.$. J .6#2 Di&tance
with an :&s"uare of 6.!!##.
RATE DISTANCE
"&31 (
!!"'#1 '
1"11 ##2
!#"3(1 &2
!2"$21 !3
&1"'21 !#2$
&!"&!1 !$&'
&"#31 !11(
!2"'11 &$
$"#1 !!$'
!3"!$1 !'2
!("&31 3$#
3$"1#1 (1&
'"&1 !2(
3&"!1 32!
3$"'$1 2(#
&$"$!1 !((1
32"(21 '33
"231 (#
1"11 $#$
$"#1 !$'
$"#21 !&#!
3$"'$1 (!(
'"&1 !&2
!#"&'1 &&$
&3"&!1 !('
3$"1#1 (&
32"(21 '$#
!"&31 ($
-8
F#$ure MULREG02 * plot of the input data for the example problem.
ANA)*SIS OF BARIANCE
Sour-e SS df MS F/value
Regre,,ion &2'3"21#1( ! &2'3"21#1( &11"3!
Re,idual &&"'&#( 2 !"#$3&
Total &(("$($$ (
SUMMAR*
T8e e,timating e<uation i,:
* HRATEI 6 !"&#21 % 1"1!! F DISTANCE
Dit8 a ,tandard error of e,timate 6 !"!1'
and an R/,<uare of 1"''!!
F#$ure MULREG03 -omputational results for the example problem.
-9
ERROR ANA)*SIS
O0, A-tual E,timated Dif/ EF.e-ted range Out of
no" * * feren-e FSEE range
! "&31 "&!2 1"1& 1"12# / 3"##'
!!"'#1 !&"32 /!"# !!"&1 / !#"2!3
& 1"11 !'"!1$ 1"'!3 !$"($3 / !"&3(
3 !#"3(1 !#"' 1"#! !"'(2 / !2"32!
# !2"$21 !3"'#1 "21 !"21( / !2"!' *
$ &1"'21 &!"3#& /1"3(& '"!! / &&"$'#
2 &!"&!1 &"!$ /1"'1$ '"'2# / &3"3#(
( &"#31 3"#2! /!"1&! "&' / $"(!&
' !2"'11 !$"23& !"!#2 !3"#1! / !("'(#
!1 $"#1 $"# /1"11 3"(1 / ("2$&
!! !3"!$1 !3"233 /1"#(3 !"#1 / !$"'($
! !("&31 !2"''! 1"&3' !#"2#1 / 1"&&
!& 3$"1#1 3$"&1 /1"21 33"12' / 3("#$
!3 '"&1 2"(3 !"&(( #"$11 / &1"1(3
!# 3&"!1 3"'( 1"( 31"1#$ / 33"#&'
!$ 3$"'$1 3$"!1 1"(#( 3&"($! / 3("&33
!2 &$"$!1 &#"!&2 !"32& &"('# / &2"&2(
!( 32"(21 3("1' /1"!#' 3#"2(2 / #1"2!
!' "231 "$(! 1"1#' 1"3&' / 3"'
1 1"11 1"&1$ /1"($ !("1$3 / "#32
! $"#1 2"2&& /!"!& #"3'! / '"'2#
$"#21 ("22 /"!#2 $"3(# / &1"'$'
& 3$"'$1 3$"#1 1"3#( 33"$1 / 3("233
3 '"&1 2"&3$ !"((3 #"!13 / '"#(2
# !#"&'1 !$"3( /!"1&( !3"!($ / !("$21
$ &3"&!1 &3"#!' /1"1' &"22 / &$"2$1
2 3$"1#1 3$"$2 /1"$ 33"3&1 / 3("'!3
( 32"(21 3("(& /1"3!& 3$"13 / #1"##
' !"&31 "(1 /!"3$ 1"#$1 / #"133
F#$ure MULREG0< /rror analysis for the example problem.
Re,idual, RESIDUA)S +)OTTED AGAINST ESTIMATED *
"2% >
"
"
" >
!"2%
" > >
" >
" > > >
1"(%
" > >
" > >
"/////////////////>////////>//////////////////////////////////////////>
/1"% > > > >
" > > >
" >
" > >
/!"% >
> >
"
"
/"%"""""""""%"""""""""%"""""""""%>""""""""%"""""""""%"""""""""%"""""""""%
!&"# !("3 &"3 ("3 &&"3 &("& 3&"& 3("&
E,timated *
F#$ure MULREG0A )lot of residuals for the estimating e"uation.
60
14S+RUC+1-4S 0-R +%)
SU..*3 C%(14 S1MU*(+1-4 .R-/R(M
SCSIM
SCSIM is computer software to mimic the actual flows of product through a supply
channel having four echelons ranging from factory to customers4 as illustrated in Figure
,-,I8&#. 5he product flows are replicated with a 8onte -arlo&type simulation. *
single4 or aggregate4 product is used and a single facility4 or aggregation of facilities4 is
assumed at each echelon. 5he obEective is to simulate supply channel performance and
costs when various forecasting methods4 inventory policies4 transportation services4
production lot sizes4 order&processing costs4 and the lie are used throughout the supply
channel. :eports and graphs are used to portray supply chain performance for different
simulation runs.
F#$ure SCSIM01 T.e Genera' Supp'7 C.anne' 8&r SCSIM
A5OUT SIMULATIONS
* computer simulation is an experimental methodology where events and occurrences4
such as product demand and transit times4 are replicated by obtaining their values from a
probability distribution. :andom numbers distributed according to a probability
distribution of the event being represented are used to mimic the variability in the event4
such as shipping times. * seed number is used to initiate the random processes. ,tarting
with the same seed number will produce the same results if the data inputs are always
identical. ;owever4 different seed numbers produce separate sample observations without
changing the input data. 5o obtain a correct picture of channel performance4 a number of
runs should be made using different seed numbers to create a reasonable sample size. 5he
results in the sample can then be averaged. -omparisons of the samples having different
data sets are best made by performing hypothesis tests on the averaged results to see if
61
there are statistically significant differences. * comparison of single runs is risy since the
observed differences may simply be due to random variations that are not statistically
significant.
,imulation results can be sensitive to start up conditions. 5hat is4 certain assumptions
are made when starting the simulation4 such as inventories are zero4 that are not true after
the simulation output has stabilized. 5he simulation needs to run for a number of periods
before stabilization can occur. It cannot be nown in advance how long this initialization
will tae4 so watch the graphs to see when steady state ?repeatable patterns@ in the output
has occurred. SCSIM uses a one&year period for initialization. If steady state has not
occurred4 run the simulation for additional years4 and then use the results only from the
later years where steady state has been observed.
For further information about the simulation process4 see a good boo on management
science covering 8onte -arlo simulation or a boo that is specifically devoted to
simulation.
A5OUT THIS SIMULATION
* number of assumptions have been made in the operation of this simulation. 5hese are
not under the control of the user4 but they are noted here for better understanding of the
action of the simulator.
*ll distributions are normal with a minimum average value of #
*ny draw on a normal distribution producing a number less than # is truncated to #
5he time period is one day with a wee being ( days4 a month being $6 days4 and a
year being $.6 days
Muantities arriving at an echelon in the channel for a particular day are added to
inventory before deductions from inventory due to sales are made
5he minimum value for demand4 sales4 and production lot size is #
)roduct volumes are in whole numbers
1umbers as input data or as computed values greater than # billion should be
avoided as they may cause overflow errors
*ll unfilled orders become bac orders4 not lost sales4 until ade"uate inventory is
available to eliminate the bac orders
*lthough the results can be seen in the graphs and reports of the simulation4 the day&
to&day "uantity flows throughout the supply channel can be observed in the file named
:/,+'5,. It is updated each time the simulation is run and is a comma&delimited file that
can easily be read in a spreadsheet program such as /xcel. 5he simulation results are
shown for all periods after the initialization year. NoteC 5he :/,+'5, file can become
"uite large if long simulation runs are re"uested. 9e sure to have ade"uate file storage
space available in the output folder.
INPUT
5he data inputs consist of simulation control information4 demand proEection data4 sales
and cost data4 forecasting and inventory control methods4 and transportation choices with
statistical performance data. *ny units of product flow may be used4 but control the data
scaling so that the numbers are neither too large nor too small. 5ime is always expressed
in days. -osts4 revenues4 and profits are expressed in R. *lthough any monetary unit may
6*
be used4 only R will show on the input data screens and the output reports. /ach data
element is briefly discussed as follows.
In#t#a'#@at#&n
Seed number. * seed number needs to be specified to provide an entry point to the
random number generator. Different seed numbers will provide different simulation results
and4 therefore4 different sample observations. NoteC -hanging a seed number after a run
has already been made will produce a different result than when the same number is used
for the first run.
*ength of simulation. 5he length of a simulation run is specified. +sually Eust a few
simulated years are needed to produce steady&state results. * maximum of 26 simulated
years is allowed4 but the :/,+'5, file becomes "uite large and may not fit on a #.%% 89
floppy dis4 if a floppy dis has been selected for the location of LOGWARE output.
Selling price. 5he selling price is the price charged to the end customer in the supply
channel4 in R/unit.
Cu%t&*er De*an( Pattern
/nd customer demand can be generated randomly or it can be specified.
+enerate demand pattern. ,electing to generate demand will produce a daily demand
pattern from a normal distribution. Inputs re"uire a daily demand average and a standard
deviation4 in units. *ny demand generated that is less than one is set to one unit.
*n annual growth rate is a percentage of the average demand. ,easonal indices may
be specified on a monthly basis. /ach monthly seasonal index is nominally a value of #.
,ince it is a multiplier to demand4 an index of #.25 means that demand for that month is
#.25 times the monthly average demand. *n index of 6.5 means actual demand is one&half
of the average monthly demand.
Specify demand pattern. ,pecific demand values may be specified for each day of a
$.6&day year. 5his pattern is repeated for each simulated year. *n annual growth
percentage may be used as a multiplier to these values.
Input% 8&r Le+e'% 1" 2" an( 3
5he inputs are the same for all three echelons4 so they are discussed Eointly.
Product item data
Ite* +a'ue. Data relating to inventory include item value4 carrying cost4 in&stoc
probability4 and bac&order cost. 5he item value refers to the worth of the product at the
particular level in the channel. 5he item value is the accumulation of purchase4
production4 transportation4 and order&processing costs to the point in the channel where
inventory is being held.
In+ent&r7 -arr7#n$ -&%t. Inventory carrying cost is the percent per year of the
item value that it costs to hold a product as inventory. It is the total of the cost of money4
insurance on inventory4 personal property tax4 value loss due to obsolescence4 and storage
costs. 5wenty to forty percent per year is a common range.
In0%t&-6 pr&/a/#'#t7. In&stoc probability is the desired percentage of sales that is
being filled on re"uest. * probability less than #66Q permits some bac ordering to occur4
but setting in&stoc probability at #66Q may force exceedingly high inventory levels to
occur4 depending on the inventory control method selected.
6+
5a-60&r(er -&%t. 5he bac order cost refers to the cost for handling an order
outside of the normal processes when the order cannot be filled on re"uest due to an
inventory shortage. 5he cost can be a result of extra paperwor4 premium transportation4
extra communication efforts4 and the lie.
Cu%t&*er)reta#'er)(#%tr#/ut&r),are.&u%e &r(er 8#''#n$. 5he cost to fill an order
received at a facility refers to the cost associated with credit checing4 inventory
availability checing4 stoc picing4 and shipping preparation. 5he cost is expressed in
R/unit. 5he time to accomplish the order filling is expressed as the average time and the
associated standard deviation4 both in days.
Pur-.a%e &r(er pr&-e%%#n$. 5his is the cost incurred by the buyer to prepare a
purchase order being &ent to an upstream supplier. It should not be confused with the
previous cost for *roce&&ing a purchase order being received by an upstream seller. 5he
purchase order cost generally will be a clerical cost associated with supplier
communication4 item checing4 order transmission4 and the lie. It is expressed in terms
of R/order.
Forecasting method
5hree forecasting methods used to proEect future sales at a facility are allowed.
5wo are forecasting from a time series and the third is a userAs specified forecast. ?,ee
-hapter 3 of 9usiness 'ogistics/,upply -hain 8anagement for more information on
forecasting methods.@
E!p&nent#a' %*&&t.#n$. /xponential smoothing forecasting is the level&only
type. 5hat is4 the forecasting model has the general formC
;t< = At J ?#& @;t
5he smoothing constant ?@ is a value between 6 and #. Nalues between 6.65 and 6.$ are
common. ;igh values give more rapid response to a changing time series4 whereas lower
values cause the forecast to be less affected by random variations in the time series. At are
the sales in period t4 and ;t is the previous forecast in period t.
M&+#n$ a+era$e. 8oving average forecasting is a simple type of the form
N
A
F
N t
t
t
t
+
=
1
where At is the sales in period t4 and N is the number of periods ?days@ over which the
most recent N periods of sales are to be averaged. ,maller N values give a more
responsive forecast to shifting trends and seasonality but can be overly sensitive to random
variations. ;igher N values give the opposite effect.
Spe-#8#e( 8&re-a%t. 5he specified forecast allows the user to provide the average
daily sales value. 5his value remains in effect during the simulation until changed by the
user. 5he interval is specified as an input value. 5he user is prompted for a sales value at
the specified interval.
6,
eorder policy
Inventory replenishment can be controlled by three methodsC reorder point4
periodic review4 and stoc&to&demand.
Re&r(er p&#nt. 5he reorder point method ?also nown as the trigger point
method4 S & policy4 >&system4 and fixed "uantity&variable order interval method@ controls
inventory levels through two valuesC the reorder size ?>@ and the "uantity ?R+P@ that
triggers when to place a replenishment order. 5hat is4 when the inventory level of an item
drops to R+P4 replenish inventory with an amount >. 5he user may select the option to
have these values calculated or may choose to specify them.
Per#&(#- re+#e,. 5he periodic review method ?also nown as the P&system or
fixed order interval&variable order "uantity method@ controls inventory levels through two
valuesC the review interval ?7@ and a target level ?MAX@. 5he inventory level is observed
every 7 days4 and a replenishment order is placed for the difference in the MAX level and
the "uantity on hand. 5he user may select the option to have these values calculated or
may choose to specify them.
St&-60t&0(e*an(. ,toc&to&demand control is similar to the periodic review
method except that the MAX level is determined as a multiple of the num.er o( da-& in the
forecast. For example4 the 8*7 level might be set at #< times the monthly forecast4 or
%5 days. * higher multiple forces more safety stoc into inventory. 5he review interval 7
might be a convenient review interval such as $6 days. 5he user sets these values.
Fa-t&r7)S&ur-e
Production cost and lot size. 5he cost to produce a unit of product at the factory is
the standard cost associated with a particular product. It is expressed in R/unit. 5he lot
size in units of sales is the minimum "uantity to be produced at one time. 'ot size is
related to the standard cost of production.
Production time. 5he time re"uired to produce a lot "uantity is expressed as the
average time and the standard deviation4 both in days.
Purchase cost. 5his is the cost in R/unit to purchase the materials to build the
product.
Order'filling cost. 5he cost to fill an order received at the factory refers to the cost
associated with credit checing4 production scheduling4 and shipping preparation. 5he
cost is expressed in R/unit. 5he time to accomplish the order filling is included in the
production time.
Tran%p&rtat#&n
,hipping between supply channel echelons can be accomplished by a number of
transportation modes and their service offerings. /ach transportation service can be
represented by its cost per unit and its delivery time. ,peed and reliability are expressed as
the average transit time and the standard deviation of transit time4 both in days.
6-
RUNNING S#SIM
E!a*p'e Pr&/'e*
5o illustrate the running of SCSIM4 consider the database shown in 5able ,-,I8&#.
5his problem shows a variety of options throughout the supply channel. 2nce the
database is prepared4 clic on the S&'+e button to execute the simulation. * 2&year
simulation is run4 and some of the results are shown below without Eudging their ="uality.>
Ta/'e SCSIM01 Data/a%e 8&r e!a*p'e pr&/'e*
SIMULATION DATA5ASE
T#t'e: ,ample problem
In#t#a'#@at#&n
#2$%5. ,eed value
2 'ength of simulation4 years
(5 ,elling price4 R/unit
Cu%t&*er (e*an( pattern
?enerate dail- demand
#66 *verage daily demand4 units
26 ,tandard deviation of daily demand4 units
6 *nnual demand growth increment4 Q
Mont'l- &ea&onal indice&
8onth Index 8onth Index 8onth Index 8onth Index
# # % # ( # #6 #
2 # 5 # 3 # ## #
$ # . # ! # #2 #
Reta#'er)Le+e' 1
Product item data
%6 Item value in inventory4 R/unit
5 -ustomer order filling cost4 R/unit
$ )urchase order processing cost4 R/order
2# Inventory carrying cost4 Q/year
# *verage customer order fill time4 days
6.# -ustomer order fill time standard deviation4 days
!5 In&stoc probability4 Q
5 9ac order cost4 R/unit
;oreca&ting met'od
/xponential smoothing
6.$ ,moothing constant
Reorder *olic-
:eorder point control method
-alculate inventory control parameters ?M4 :2)@
D#%tr#/ut&r)Le+e' 2
Product item data
$5 Item value in inventory4 R/unit
6.( :etailer order filling cost4 R/unit
#6 )urchase order processing cost4 R/order
25 Inventory carrying cost4 Q/year
2 *verage retailer order fill time4 days
6.2 :etailer order fill time standard deviation4 days
26 9ac order cost4 R/unit
;oreca&ting met'od
66
8oving average
#26 1umber of periods
Reorder *olic-
,toc&to&demand control method
%5 5arget days of inventory
6 :eview time in days
Ware.&u%e)Le+e' 3
Product item data
$6 Item value in inventory4 R/unit
6.5 Distributor order filling cost4 R/unit
56 )urchase order processing cost4 R/order
26 Inventory carrying cost4 Q/year
2 *verage distributor order fill time4 days
6.# Distributor order fill time standard deviation4 days
#6 9ac order cost4 R/unit
;oreca&ting met'od
8oving average
$.6 1umber of periods
Reorder *olic-
)eriodic review control method
$6 2rder review interval ?5@4 days
.566 5arget level ?8ax@4 units
Fa-t&r7)S&ur-e
Product item data
#2 )roduction cost4 R/unit
#666 8inimum production lot size4 units
6.$ 0arehouse order filling cost4 R/unit
5 *verage production time4 days
6.5 )roduction time standard deviation4 days
%6 )urchase cost4 R/unit
Tran%p&rtat#&n
7ran&*ort .et)een Di&tri.utor and Retailer
#.25 5ransport cost4 R/unit
2 *verage time in&transit4 days
6.2 5ransit time standard deviation4 days
7ran&*ort .et)een @are'ou&e and Di&tri.utor
2.6! 5ransport cost4 R/unit
$ *verage time in&transit4 days
6.$ 5ransit time standard deviation4 days
7ran&*ort .et)een ;actor- and @are'ou&e
2.%5 5ransport cost4 R/unit
5 *verage time in&transit4 days
6.5 5ransit time standard deviation4 days
67
S#*u'at#&n Re%u't%
*fter running a simulation4 clic the Re%u't% button4 which will be enabled as a result
of maing a run. 5he results are shown in both graphical and report forms. Narious
combinations of graphs and reports can be selected. Lou have the option of displaying the
graphs in daily4 weely4 or monthly periods. 0hereas the daily period plots display the
minimal time unit of the simulation4 weely and monthly periods average the results for the
selected period. 0hen entering a graph or report4 you have the option of presenting all of
the simulated years or a portion thereof. 5his is useful for presenting the simulated
period?s@ that are most stable and representative of the steady state condition and for
controlling the display resolution for the graphs.
*n example of graphical output is shown in Figure ,-,I8&2 for the retailer/level # in
the channel. * number of such graphs can be selected. ,elect the period for display.
5able ,-,I8&2 shows a sample performance/financial report for the warehouse/level $
echelon of the channel. * variety of reports may be selected. 0ithin the report4 results are
shown as a yearly average and for the simulated period. 5he yearly average is the a/erage
of a result for the selected period of the report4 whereas the result for the simulated period
is the &um of the result values over the simulated period.
5he reported financial items have the following meaning.
evenue. 5he selling price per unit multiplied by the retailer sales.
#ost of purchased goods. 5he unit purchase cost times retailer unit sales.
+ross margin. 5he difference between retailer sales revenue and the associated
purchased goods cost.
Production cost. :etailer sales multiplied by production cost per unit.
"ransportation costs. 5ransportation rate times the shipping volume moving between
echelons within the channel.
Sales order handling cost. 5he cost for a seller to process the sales orders received.
It is the unit order filling cost times the unit sales volume.
Order'processing cost. 5he cost for a facility to prepare a purchase order placed on
an upstream seller. It is the purchase order&processing cost times the number of purchase
orders placed.
Inventory cost. 5he cost of an item in inventory times the inventory carrying cost
times the average inventory level at a particular echelon.
1ac2'order cost. 5he volume of units bac ordered at an echelon times the unit bac&
order cost.
"otal cost. 5he sum of the costs for a particular echelon.
Net profit contribution. 5he difference between revenue and the sum of all supply
channel costs.
68
F#$ure SCSIM02 An E!a*p'e &8 a Grap.#-a' D#%p'a7
5he reported performance items have the following meaning.
Sales forecast. 5he expected sales for a particular echelon as proEected by the
selected forecasting method.
Sales to channel member. 5he actual sales received by a channel member. 5his will
usually be less than customer demand due to the inability to fill all of demand on re"uest.
-verage inventory on hand. 5he average inventory level for the selected reporting
period.
Inventory turnover ratio. 5he ratio of a channel memberAs sales to the average
inventory on hand.
$aily bac2 orders. * channel memberAs average daily bac&order volume.
1ac2'order occurrences. 5he number of bac&order occurrences that too place
within the reporting period.
-verage demand filled on re3uest. * service measure expressed as a percent of
echelon member demand that could be filled at the time the demand occurred. Demand is
either customer demand4 in the case of the retailer4 or the purchase orders placed on a
distributor4 warehouse4 or factory.
$aily 3uantity on order. 5he average daily units placed as orders on an upstream
seller that have not yet been received.
Number of orders placed. 1umber of purchase orders placed on the upstream
echelon for the selected period.
69
Ta/'e SCSIM02 An E!a*p'e &8 an Output Rep&rt
0*:/;2+,/ :/)2:5 F2: ,I8+'*5/D L/*:, # 52 2
Forecasting methodC 8oving average
Inventory control methodC )eriodic review
Learly ,imulated
average period )/:F2:8*1-/ ,5*5I,5I-,
$.4%## (24322 ,ales Forecast4 units
$54(%3 (#4%!. :etailer sales to retailer4 units
24$!(.$2 *verage inventory on hand4 units
#%.!# Inventory turnover ratio
2...2 Daily bac orders4 units
($.26Q *verage demand filled on re"uest
#42$..66 24%(2 9ac order occurrences
#42!# Daily "uantity on order4 units
#2 2% 1umber of orders placed
FI1*1-I*' )/:F2:8*1-/
.66.66 #4266 -ost to process factory orders
#(43(%.66 $54(%3 -ost for handling distributor orders
#%4$3$.!6 234(.3 -ost for carrying inventory
#24$.6.66 2%4(26 -ost for bac orders
(%4(#$.$2 #%!4%2( -ost for transport to distributor
R##!4!$#.22 R2$!43.2 5otal cost
70

Log Ware 5.0

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Log Ware 5.0

Încărcat de

Drepturi de autor:

Formate disponibile

SELECTED COMPUTER

S-ar putea să vă placă și