See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/3928134

Base camp facility layout

Conference Paper · February 2001

DOI: 10.1109/ICSMC.2001.973756 · Source: IEEE Xplore

CITATIONS READS

2 1,591

3 authors, including:

Barry Ezell
Old Dominion University
59 PUBLICATIONS 1,001 CITATIONS

SEE PROFILE

All content following this page was uploaded by Barry Ezell on 26 June 2017.

The user has requested enhancement of the downloaded file.

 Base Camp Facility Layout

Matthew U. Robertson, Barry C . Ezell, Assistunt Professor, and Michael L. IWGinnis, Professor, United
States Military Academy

and capstone projects at the United States Military

Abstruci--The primary purpose of this project was to aid in Academy, have an interest in finding an optimal base camp
the developiiient of an optimal base camp facility layout by layout.
studying and identifying proximity relationships between
essential facility components. These relationships were The purpose of this project was to aid in the development of
determined through extensive research, surveys, and a base camp facility layout optimization system by
interviews with key stakeholders. Three separate understanding the proximity relationships between base
relationship charts were created representing the mean, camp components, developing a facility layout domain, and
median, and mode response scores from research and an comparing generated layouts to existing models and cainps.
online survey of critical base camp stakeholders from The first phase involved collecting input from a variety of
around the world. These charts were then used in sources to gain an understanding of the proximity
conjunction with score and rectilinear distance matrices to relationships between fifteen common base camp facilities.
organize the data for software optimization. The This data was then entered into the Computerized Relative
Computerized Relative Allocation of Facilities Technique Allocation of Facilities Technique (CRAFT) software in
(CRAFT) was used to solve for a near-optimal layout for order to generate a camp layout. The final phase involved
three base camps with 15 facilities. These three layouts comparing the results to mid-1980s Theater Construction
were then compared to each other as well as standard base Management System (TCMS) drawings and Camp
camps found in Theater Construction Management System Bondsteel’s layout.
(TCMS) software and Camp Bondsteel.

11. DATACOLLECTION AND TRANSFORVATION

1. INTRODUCTION
To optimize a base camp the facilities, or components,
Base camps have been in the Army’s repertoire since its which make up the physical infrastructure of a base camp
formation. A base camp is an evolving military facility that must be arranged in the most efficient manner. Some
supports the military operations of a deployed unit and common types of facilities include motorpools, chapels, and
provides necessary support and services for sustained ammunition holding areas. Both passive and active
operations [l]. Because of the recent proliferation of research methods were used to collect data about base camp
deployments, the Army now uses base camps from Haiti to facility proximity relationships. Two passive methods,
South America to Kosovo. These camps usually begin as research of military publications and web-based research,
tactical assembly areas and become defacto base camps were used to obtain general information about the
over time. This process calls upon the Army’s engineers to relationship between various facilities. Overall, one-third
develop and construct camp layouts that are tactically sound of the data came from passive research. An online survey of
and help soldiers to complete their missions in less than critical stakeholders and interviews with a commander of an
ideal situations. Because of the increase in the number of engineer brigade constituted the active research portion of
Operations Other Than War (OOTW) deployments and the the project and Glled in gaps left by the passive methods.
ever-increasing coniplexity of base camps, some type of Two-thirds of the data came from this type ofresearch. The
layout optimization tool would be useful in planning and end result of this phase was a completed relationship chart
constructing base camps. for a base camp with 15 different facilities.

Unfortunately, no known recent work in the field of base

camp layout optimization has been completed. This was the A. Passive Data Collection
main motivator in studying base camp facility design. The
Coiistruction Engineering Research Laboratory (CERL) is The first data collection stage involved intensive study of
currently attempting to optimize base camp layouts given the U.S. Army Europe Engineer’s Red Book and Blue Book.
terrain limitations using GEOBest, a U.S. k r Force site These sources outlined individual facility standards.
location planning tool, but has yet to obtain suitable results. Specifically, the Red Book contained a detailed listing of
Other initiatives, such as the Contingency Facilities in common base camp facilities and their minimum acceptable
Future Base Camps by the Office of the Chief of Engineers standards [2]. The Blue Book was intended to provide
commanders and soldiers deployed to the Balkans a

0-7803-7087-2/01/$10.00 02001 IEEE 2064

definitive guide to the extent and quality of base operations, was the median. The final measure of central tendency, the
products, and services they could expect while in garrison mode, represented what would have occurred if the
[3]. Both sources helped in the determination of the stakeholders voted on the relationship between the facility
primary components of a major base camp. From over 30 proximity. Since the mean, median, and mode each had
types of facilities listed in these sources, 15 were chosen for advantages in measuring facility proximity relationships,
the study. This number represented a compromise between they were calculated for each question and later used to
a realistic base camp, a reasonable optimization calculation generate three different base camp layouts.
time, and available information.

Research on the World Wide Web constituted the second C. Data Transformation
type of passive research. Many base camps and units
deployed to base camps have their own web pages. All of the data collected was eventually placed into
Additionally, online articles documenting successes and relationship charts, a convenient way to organize data. A
failures of various base camps were plentiful. Although standard technique, which can be found in major Operations
these sources were useful in gaining a general Management texts such as Heizer and Render's Operations
understanding about base camp layouts, they provided little Management, was utilized [ 5 ] .If close proximity of base
information directly related to facility layout. camp components was absolutely necessary, the cell
corresponding to the components relationshp received an
B. Active Data Collection A. If the proximity relationship was especially important,
the cell received an E. Important relationships received an I ,
In order to address some of the layout shortcomings of web Okay relationshps received an 0, Unimportant
resources and the Red and Blue Books, correspondence was relationships received a U , and Undesirable relationships
initiated with Colonel Robert McClure, Commander, 1'' received anX. Through the passive research methods
Infantry Division Engineer Brigade (mechanized). His Colonel McClure's interviews it was fairly easy to discern
responses answered numerous questions about the relative and code facility proximity relationshps.
positioning and number of facilities per base camp.
Additionallv, he cited certain components such as The survey data required a more concerted transformation
motorpools, ammunition holding areas, and fuel storage effort, though. The absolute highest and lowest values for
areas that should be separated from populated areas of the any of the three measures of central tendency were 10 and
camp [41. 1, respectively. Score brackets were created to transform
these numerical values into a relationship chart values. For
Despite obtaining information from Army publications, web instance, a survey score between 9 and 10 received an A ,
research, and an interview with a base camp expert, there while a score between 1 and 3 received ank: Because of
were still some unresolved facility proximity issues. In the difference between the mean, median, and mode
some instances, proximity relationships between responses, certain questions had three different letters, one
components could not be found. In other instances, for each of the t h e e measures. Of the nineteen questions
contlicting information about component relationships asked, eleven received at least two different letters
existed. To resolve these issues, an online survey was depending on the measure of central tendency used in the
created. Nineteen questions such as: calculation.

How important is it for Seahut Clusters (housing Because of the differences amongst the survey responses
are) arid Aviation Facilities to be close in proximily? and subsequent central tendency measures, there was a need
to create three separate relationship charts. These charts
were asked. The users were asked rate the importance of were titled Mean, Median, and h 4 d e based on where their
proximity between various base camp facilities on a scale of survey data came from (see iligure 4). 37% of the survey
1 (avoid at all costs) to 10 (extremely vital). A response of data within the relationship charts was different in at least
5 signified that adjacency was unimportant. Twelve one of the charts. One-third of the data within the charts
respondents, consisting of a good blend of civilian and canie from non-survey data and was thus similar. An
military experts from Kansas to Kosovo, responded. The additional 30% of the data came from the survey and was
variety of respondents naturally led to wide variety of identical for each relationship chart.
responses. Coupled with a small sample size, high response
variances were prevalent.

Because of response variety and the small sample size of 111. MODELING
the survey, determirug a single good measure of central
tendency was difficult, Although the mean is usually the After the data had been collected and organized into three
preferred statistic in measuring the central tendency of relationship charts, some type of optimization was needed
samples, the sample size of this survey was small and thus to determine the best layouts. The desired software
susceptible to outliers. An answer to outlier+pceptibility package was one that would fmd the optimal layout given

2065
fifteen different components and their proximity over 40 components. The end result of CRAFT'S
relationstups. Unfortunately, no readily available software optimization efforts was a colored display of the determined
with this capability was found. Three different alternatives layout. However, optimality was not guaranteed and the
were tried, but only one provided acceptable results. The layout was somewhat dependent on the initial layout
following section describes the issues encountered using entered into the software. Additionally, if a facility was
Production and Operations Management for Windows, created with an area greater than 2x1, its shape might
Excel OM, and CRAFT software to find the optimal layout change during the optimization process [SI. This package
strategy for a 15 facility base camps. was a compromise between the ability to handle a large
amount of data and guaranteed optimality. For our purposes
of determining general proximity relationships between
A. A4odelingAlteniatives facilities and developing an effective methodology for
determining the optimal base camp, CRAFT was sufficient
The first attempt at layout optimization used Production and and preferred over the other available alternatives.
Operations Management (POM) for Windows [SI. POM is
an operations management tool focused on a variety of
operation management problems, not just facility layout. B. Model Inputs and Assumptions
The facility layout module gives the user the option of
using pairwise comparisons or explicit enumeration to find The inputs required for CRAFT were a score (or from-to)
the optimal layout. Using the pairwise comparison method, matrix, the size of each facility and the overall base camp,
POM minimizes the product of the proximity scores from and an initial layout. To create the score matrix, the data in
the relationship chart and rectilinear distances between the relationship chart was quantified. Although some flow
coniponents. T h s method is quicker than using explicit between various departments usually gets entered into this
enumeration, which checks all n! layouts. However, it does matrix, the proximity relationship data represented the
not always find the optimal solution. POM was desire for adjacency in a similar manner. The convention
advantageous because it had the option of finding the used for this process was to code A to 6, E to 5, I to 4, 0 to
optimal layout or using pairwise comparison and was easy 3 , U to 2, and X to 1. The area entered for each facility was
to use. Unfortunately, POM was only capable of handling a one unit squared, or one cell. The motivation for this was
layout with ten facilities, while software capable of preventing the shapes of the facilities from changing
handling fifteen was desired. Cutting the number of drastically. Also, the size of most components depends on
components to ten was considered but eventually discarded the number of troops stationed at the base camp and terrain.
because the large number of assumptions needed to do so The primary goal of this project was not to develop an
greatly diminished the reasonableness of the results. optimal layout for a specific scenario but to determine
proximity relationships between base camp facilities. Since
Excel's Operations Management (OM) was the result of a each of the fifteen components was modeled as a single
search to find software capable of resolving a facility layout cell, the layout of the camp had an area of 15 units2( 5 x 3
problem with 15 components. T h s software checked all units). The final input for CRAFT was an initial layout. To
pairwise comparisons to find the minimum score/flow- prevent bias based on knowledge of existing base camps,
distance value [6]. It did not require any installation from the components were not entered in any particular order.
disk, as it was downloadable from the Internet. However, the Mean, Median, and Mode models all had the
Furthermore, it worked in Microsoft Excel as in add-in. same initial layout.
Most importantly, it allowed the user to enter data for 15+
components. Yet for all of the apparent advantages of Like any other modeling tool, using CRAFT required the
Excel OM, its disadvantages were tremendous. OM used formation of modeling assumptions. The two largest
all of the processing capability of a Pentiuin 111, 233 MHz assumptions were that terrain did not affect the layout and
computer. Attempting another task at the same time as that size and shape of all facilities was equal. In base camp
resolving a layout prevented the user from getting back to design, terrain cannot be ignored. Instead of leveling hills,
the program before it finished. Since any layout with blasting mountains, or rerouting rivers, to create a periectly
greater than 9 facilities caused Excel OM to crash, changing flat area, engineers usually change their layout to fit h e
the active screen with even 10 facilities caused all results to terrain, not vice-versa. The similar size and shape
be lost. Even if Excel OM had performed properly, assumption impacted the optimized layout results. For
resolving an optimal layout for 15 components would have instance, the area of a base camp devoted to seahut clusters
taken nearly 35 years (see figure I ) . Thus, Excel OM was is much greater than the area devoted to an education
not the answer to modeling an optimal base camp layout. center. Making this assumption meant that the results did
not represent the ideal layout of a base camp so much as it
The final attempt at modeling an optimal camp layout used represented the relative importance of facility adjacency.
the Cotnputerized Relative Allocation of Facilities Other assumptions were that not including mail rooms,
Technique (CRAFT) [ 7 ] . T h s software, also downloadable concession areas, unit facilities, and others did not affect the
from the Internet, operated as an Excel add-in. The major generated layouts, similar coniponents such as individual
advantage of CRAFT was that it could handle a layout with seahuts and MWR facilities could be lumped together, and

2066
that here were relative and rectilinear distances between static. While the user could modify thc components of the
components. base camp and retrieve an updated project plan, the
drawings themselves were not updated. Thus, limited
comparisons could be made. The drawings were basic and
Iv. ANALYSIS
OF RESULTS did not include a majority of the facilities such as education
centers, fuel storage araas, and MWR sites used in our
study. One generality noted from TCMS and OUT layouts
A . A ltemative Anabsis Methods was that the dining facilitics were adjacent to the seahuts
clusters. However, the seahut clusters were not in the
’center of the camp. TCMS had helipads in the middle. The
Three methods were used to analyze the results from CRAFT generated layouts had aviation facilities on the
CRAFT. The first was a comparison of the three generated camp’s perimeter.
layouts to each other. From this typk of analysis, general
trends about proximity relationships were discovered as The final analysis method was a comparison ofthe CRAFT
well as a few potential abnomialities. The second method layouts to Camp Bondsteel, which was constructed in 1999.
was a comparison of the layouts to base camp dramings in Bondsteel is considered by many to be a model base camp
the Theater Construction Management System (TCMS) [9]. since it is relatively new and the beneficiary of reviews of
Military engineers use this software to aid in constructing older camps in the Balkans. Adjacent facilities in
base camps throughout the world. The final analyzation Bondsteel were recorded and compared to the adjacent
method compared the CRAFT layouts to Camp Bondsteel, facilities from the generated layouts. A summary of the
a two-year-old base camp in the Balkans. Another method findings appears in figure 3. CRAFT adequately modeled
that was considered but ultimately not used because of many facility adjacencies found at Camp Bondsteel. Of the
installation problems was the Consequence Assessment items adjacent to the seahut clusters at Bondsteel, CRAFT
Tool Set (CATS) [IO]. This software could have been used modeled 71% correctly and positioned them in the same
to conduct risk analysis on the base camp layouts using relative position. The two layouts were also similar in that
both natural and man-made disasters. they each had the fuel storage area, aviation facility, and
ammo holding area removed from center of the camp.
Additionally, seahut clusters q d soldier support facilities
B. Discussion of Results and hialyzation EjJorts were adjacent in both layouts.

In comparing the layouts that C M F T generated, some The Bondsteel-CRAFT comparison also highlighted some
general conclusions were made. In each of the layouts, shortcomings of our models. CRAFT identified only 20%
certain coiiiponents were always adjacent. These of those facilities adjacent to niotoyools. l‘his occurred
components were the dining facility and seahut clusters, because there were over 20 separate niotorpools throughout
motorpool and aviation facility, motorpool and aid station, Camp Bondsteel while in the CRAFT layouts, there was
and the Morale, Welfare, and Recreation (MWR) facilities only one component representing a motorpool. Thus, it
and the education center. Insights into the positioning of appears that our method had a difficult time dealing with
components within a base camp were also gained. Seahut multiple component adjacencies. Also, the CRAFT layouts
clusters were centrally located in each of the three layouts, did not express the actual distance between facilities well.
the fuel storage area and Post Exchange (PX) were located For instance, the ammunition holding and fuel storage areas
fixthest fiom the motorpool and aviation facility, and the at Camp Bondsteel were separated from any other facility
dining facility, education center, SSA facility, postal by hundreds of meters. Although the CRAFT layouts show
facility, MWR facilities, lauindrv collection point, & aid that these components should be on the periphery of the
station were always located near the base’s center. Overall, camp, they did not show the need to separate them from
the layouts were not drastically different. Of the adjacent components significantly. The final shortcoming
components that were adjacent, more than half were the Bondsteel comparison showed was that the generated
adjacent in more than one layout. Figure 2 shows the layouts did not express the relative size of the facilities. In
number of times that facilities were adjacent in the three the CRAFT layouts, each component’s area was a single
C R N T layouts. There were unexpected outcomes such as cell. Thus, a maximum of four components could be
the PX being located on the corner of the camp in every adjacent to a cell in a given layout. At Bondsteel, the
layout and the dining facility and fuel storage area being seahut clusters had 7 adjacent components because it had a
adjacent in one layout. For the most part though, the larger perimeter than other facilities such as chapels and
layouts appeared reasonable. education centers. Thus, a single CRAFT layout could
account for no more than four of the seven adjacencies
The second tool used to analyze the layouts was a found at Bondsteel. Although CRAFT didn’t identify all of
comparison to base camp drawings generated for the the adjacencies of larger facilities, it did identify which of
Theater Construction Management System. The goal was the facilities had the strongest proximity relationships.
to contrast and compare what TCMS considered optimal
layouts in the mid-1980s with our layouts. The difficulty Overall, the CRAFT layouts modeled 49% of the facility
with using TCMS inthis way was that the drawings were adjacencies found at Camp Bondsteel. Of the adjacencies
2067
not identified by CRAFT, over 41% were related to the PX (mechanized), Fred Steinman, DvnCorps, and Edward
and motorpools. Excluding the seemingly inadequate Scott, Oflce of the Chief of Engineers.
modeling of these facilities, nearly 60% of the facility
adjacencies in Bondsteel were determined by CRAFT. B. Ezell, M. Davis, and hL McClnnis, Designing a Decision
S~pportSystem forhiiliturg Buss Curiip Site Selection and Fucilig
Layout.
Deputy Chief of Staff of Engineers. Headquatiers, U.S. h y
Europe, Base Camp Facility Stundards (ILdBook),Operation
v. O N SFUTURE
~ C O h ~ ~ ~ E N D . 4 T lFOR WORK JOINT GUARD, March 1997.
Deputy Chief of Staff of Engineers, Hadquarters, U.S. .%?my
Europe, Long Term Base-Ops (Blue Book), October 1999.
After finishng the modeling methodology described in this R. McClure, Commander, 1* Engineer Brigade. Email to the
paper, there appear to be numerous areas for expansion and authors,19 February 2001 and 23 February 2001.
J. Heizer and B. Render, Operationsh.lanag~nient,6medition,
future work. First, contact with more stakeholders is 2001, pp. 319.
needed. Developing any degree of statistical stability H. J. Weiss, Production and Operutions A4unugemenifor
requires more than the 12 respondents who filled out the Windows Version 2.01 [CD-ROMJ,1996.
online survey; Additionally, the survey itself should be H. J. Weiss, Excel Operutronsl~iunugetiient,[sotlware].
Computerized Relative Allocution of Facilities Technique.
expanded to include questions about the proximity [Software] Available :
relationships of more facilities such as finance centers, unit http://www.me.utrnas.eduiO/b7Ejensenladdinsiinde~.html
areas from battalion to task force level, power plcuts, University of Illinois, Plunt Layout ( C I W T ) )[On
, Line1
wastewater facilities, and concession areas. However, each Available:
http://www.cen.uiuc.edu/courses/ie261 /notes/layout/layout-
additional component greatly increases layout computation Com~uter-2.html
time. Given the inputs ofcamp size and population, an [lo] United States Army Corps of Engineers, Theater Construction
algorithm that could recommend the size and number of Management Systemversion 1.2d, [CD-ROM], 2001.
components would also be usefd in developing a basic 1111 ConsequenceAssessment Tool Set, [CD-ROM]
optimal camp layout. Before putting the facilities in an
optimal position however, terrain must be considered. It is
extremely unlikely that a base camp will be built without
any geographical or geological limitations. Possible
techniques for doing this include the software packages
GEOBest and LOGSPOT. The user of a future system
should also have the capability to enter hard constraints
such as "seahut clusters and fuel storage areas must be non-
adjacent" before optimizing the base camp layout.

After the base camp layout had been determined, some type
of sensitivity analysis on the initial layout in CRAFT
should be conducted to check ifthc generated layout is
strongly dependent on the initial layout. Afterwards, risk
analysis should be conducted to aid in determining if the
layout is truly optimal. The Probabilistic Risk Assessment
and Management (PRAM) and CATS are two possible risk
management software packages.

A decision,support system for Army engineers with these

changes implemented would be a powerl'ul tool. An
engineer could enter values for the camp size and
population, terrain limitations, and hard constraints he
wants to obtain a layout that is an 80-90% solution in a
matter of minutes. This would help greatly to reduce
planning time and give engineers a base case from which to
construct a base camp.

VI. ACKNOWLEDGMENTS

We would like to acknowledge the work of the following

people in support of th~sresearch: Colonel Robert McClure,
Commander 1"' InfantqJ LXvision, I"' Engineer Brigade
2068
 Figure 1

Time to Determine Optimal Layout

with Excel OM

40 -

.-5 3 5 -
-
2 30-
$ 25-
.--
E 20-
15-
5
v)
10-
5-
* 0-
5 7 9 11 13 15
# Components

Figure 2: Number of Time that Components were Adjacent in CRAFT Layouts

l # l Comoonent I 1 I 2 I 3 I 4 I 5 I 6 I 7 I 8 I 9 I101 1 1 I121 131 1 4 1 1 5 ~

2069
 Figure 3 : Adjacent Facilities in Camp Bondsteel

#I Component Name Camp Bondsteel

1 [Seahut Clusters -5 6 7 10 11 12 75
2 Motorpools -3 9 10 I1 12
3 Aviation Facilities -2 15
4 Ammo Holding Area
5 Mess Halls
6 Chapels
-I1 7
7 Education Center 1 6
8 Postal Facility - - - - - - _
9 SSA Facility 2
I O Personal Services Center I 2 11 - 1 5
11 PX 1 2 10 15
12 MWR Facilities 7 2
13 Fuel Storage Area
14 Laundry Collection Point
15 Aid Station

Underlined Numbers - 3 CRAFT layouts where adjacent coniponents were identical to Bondsteel layout
Italicized Numbers - 2 CRAFT layouts where adjacent components were identical to Bondsteel layout
Bold Numbers - 1 C W T layouts where adjacent components were identical to Bondsteel layout
Regular Numbers - N o CRAFT layouts where adjacent components were identical to Bondsteel layout

Figure 4: Relationship Chart for the A4eaw Alternative

-
Component # Component Name Mean Alternative

1 Seahut Clusters
2 Motorpools
3 Aviation Facilities
4 Ammo Holding Area
5 Mess Halls
6 Chapels
7 Education Center
8 Postal Facility
9 SSA Facility
10 Personal Services Centers
11 PX
12 MWR Facilities
13 Fuel Storage Area
14 Laundry Collection Point
15 Aid Station

2070

View publication stats