1

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40

FLEET MANAGEMENT CRITERIA: IDENTIFICATION OF OPTIMAL DISPOSAL POINTS

USING EQUIVALENT UNIFORM ANNUAL COST
Paul Kauffmann1, Ed Howard2, Jason Yao3, Drew Harbinson4, Newell Brooks5, Richard Williams6,
Christine Gurganus7
1

Corresponding Author
Professor
Email: kauffmannp@ecu.edu
2
Associate Professor
Email: howardw@ecu.edu
3
Associate Professor
Email: yaoj@ecu.edu
6
Richard Williams
Associate Professor
Email: williamsric@ecu.edu
7
Christine Gurganuss
Email: Gurganauss10@students.ecu.edu
Department of Engineering
East Carolina University
236 Slay Building
Greenville, NC 27858
Phone: 252-328-9645, Fax: 252-737-1041
4

Drew Harbinson
Email: dharbinson@ncdot.gov
5
Newell Brooks
Email:nbrooks@ncdot.gov
North Carolina Department of Transportation
Equipment & Inventory Control Unit
4809 Beryl Road, 5066 Mail Service Center
Raleigh, NC 27606-5066
Phone: (919) 733-2220

TRB 2012 Annual Meeting

Word count: 5406+ 4 figures @250 +4 tables @250 = 7406

Submission date: July 31, 2011

Paper revised from original submittal.

2
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

Abstract
This paper presents a decision model developed to identify the optimal asset life for six equipment classes
including three on road and three off road. This model was developed as an element of a comprehensive
study to identify a methodology for evaluating aging (or depreciation), disposal points, and overall
utilization. The asset life optimization model discussed in this paper is a core result of that of that study
and presents the opportunity to reduce overall cost, improve the age of the fleet and its readiness to serve
the public, and improve overall utilization. This paper examines salvage values and identifies trend
models for market value decline based on historical records. Using internal cost information, the model
identified trends in cost of operation and use as equipment ages. This information was integrated into an
optimal economic life model based on equivalent uniform annual cost to identify the optimal disposal
point for the six classes. The analytical models developed in the study provide a foundation for long term
analysis of fleet size and the cost effectiveness of disposal points.

TRB 2012 Annual Meeting

Paper revised from original submittal.

3
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49

Introduction
This paper describes a model developed to identify the optimal (minimal) cost life cycle for six equipment
classes, the largest North Carolina Department of Transportation (NCDOT) cost investment groups:
Class 0201 pickup trucks
Class 0205 single axle dump trucks
Class 0206 flat bed and miscellaneous trucks
Class 0314 loader-backhoe
Class 0900 motor grader
Class 2002 front end loader
This model was a key component of a larger study involving approaches for evaluating aging (or
depreciation), disposal points, and overall utilization. For brevity, this paper focuses only on the
optimization model and these other areas of the study will be presented in future papers. The
management goals for the model involved development of a practical decision tool which would help
reduce cost, improve the age of the fleet and its readiness to serve the public, and promote improved
overall utilization. This paper examines the development and application of equivalent uniform annual
cost (EUAC) as a critical element of this effort, applied to identify the optimal disposal point for these
classes.
The following sections of the paper are organized to present the logical flow in building the
information necessary to develop the optimal life model.
Literature review examines the kinds of related literature we found and the current models related
to our work.
Age and usage trends reviews an important model component- how equipment use declines over
time.
Operational cost growth with age analyzes trends in equipment cost based on age.
Integration of the data to develop the optimal life model.
Literature Review
Although a large number of publications address concepts related to fleet management, current literature
contained very few specific methodologies which could be directly applied to the solution of this problem
and guide us specifically in addressing the primary goal of the project. Consequently we used the
literature search to provide general benchmarks for our analytical approaches. In general, references fell
into three categories:
Engineering Economics Texts: Advanced texts in engineering economics typically include
chapters addressing the methods and issues of replacement analysis. [1] and [2] are examples of
comprehensive text books containing a thorough discussion of these engineering economic
methods. They provide excellent information on replacement analysis methodologies but do not
contain specific methods and examples related to the problems addressed in this research.
Economic life models: The literature review did find papers on several economic life models.
[3], [4], and [5] are examples which addressed general theoretical concepts which are broadly
applicable to this study. On the other hand, [6] describes a total cost model applied to passenger
vehicle life cycles. However, it did not integrate time value of money concepts and employed
calculus based optimization methods. Consequently, its approaches are not of significant impact
to this research.
Practitioner references and handbooks: [7] and [8] are examples of this type of reference material.
Once again, these practitioner handbooks provide general information which serves as a
foundation to examine issues related to this research. However, they did not provide specific
guidance or insight into solutions of the issues identified in the project.

TRB 2012 Annual Meeting

Paper revised from original submittal.

4
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49

The next paragraphs highlight the literature most closely related to our work. Many of the
publications found were focused on specific types of fleets, such as those of freight carriers, rental
companies, transit authorities, or the military. For example, [9] benchmarked the best practices in fleet
management of a number of private companies and public entities, including state Departments of
Transportation. One of their findings was that most state DOTs were in the early stages of developing
performance measures to assist in fleet management.
Two of the more applicable reports for this project dealt specifically with state DOTs. The most
recent of these reports [10], prepared for the Oregon DOT, provides a summary of the types of models
that have been used to make replacement decisions, including computation of an economic life, utilization
of repair cost limits, and comprehensive cost models. One finding from their research was that decreasing
utilization as a function of age was not considered in most models. That is, a constant utilization rate was
assumed in the models. The Oregon report concluded that if a simple ranking criterion is to be used to
prioritize replacement decisions, then equipment age, rather than other factors such as maintenances cost,
utilization rate, etc., is the most cost effective model. If equipment is utilized in a manner such that newer
equipment is generally preferred over older equipment (such as would be expected in a DOT fleet), then
the overall fleet operating costs will be higher than if the equipment is equally utilized.
A report prepared for the Virginia DOT [11] found that a relatively good correlation between a
units variable cost as a function of the fuel costs for that unit could be made with a logarithmic model,
but the data contained a great deal of variation. Much of this variation could be attributed to differences
in how data was entered into the states database. The Virginia report recommended that an estimate of
the cost of operation for the next year could be obtained by computing the ratio of labor and parts cost per
fuel dollar year to date to the labor and parts cost life to data. A limitation to utilizing this approach was
the amount of data that was deleted at each year end. The report recommended that more cost data be
archived to achieve more confidence in cost modeling.
A key theme from this literature is that the feasibility of any model is limited by the availability
(effort and cost) and accuracy of the data required for the model. Repair and maintenance costs are by
their nature extremely variable and, as a result, highly accurate correlation of fleet cost data to any model
is challenging. In addition, integrating feasible data base fields into a comprehensive decision model to
yield realistic results is also a challenge.
Age and Usage Trends
As a foundation for the economic life model, we started with study of the 2006-09 data base for trends
related to usage of the six equipment classes. One target for the study involved determining whether the
economic life model should be differentiated based on the three geographic regions of North Carolina.
Due primarily to data variability, we did not find statistically significant differences in the geographic
regions relative to usage patterns for these six classes. However we did find that each of the years of the
study had a particular personality. In general, due to a variety of operational and financial circumstances,
we focused our analysis of usage and age patterns on 2008 as the best example of a typical year in the
study data set. Figure 1 shows a typical pattern or trend in miles per year based on age for single axle
dump trucks (class 0205) in 2008 and indicates an annual mileage decline of 540 miles per year of age
from the base of 12,650 annual miles. A few issues are evident in Figure 2:
Age is not uniformly distributed and it is clear that purchases of new units were inconsistent.
The data shows significant variation. We discussed the issues of filtering high and low data and
other methods to eliminate extreme points. Other than obvious outliers, management wanted to
leave the data as it existed since this represented a benchmark, not only for equipment planning,
but also for management and use of the information system.

TRB 2012 Annual Meeting

Paper revised from original submittal.

5
1
2
3
4

This last bullet is an essential point. Managements plan in the long term focused strategically on use of
this information and the optimal life model to improve not only equipment performance but also data base
accuracy and consistency.

Annual Miles by Age, Class 0205, 2008

35000

y = -539.52x + 12650

30000

Annual Miles

25000
20000
15000
10000
5000
0
0

10

15

20

25

Age, Years

5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

FIGURE 1 Annual Miles by Age, Class 0205, 2008

Cost Trends
Annual operating cost was calculated as the sum of the following items: PM labor, repair labor, PM parts,
repair parts, fuel, oil, and tires. This sum was divided by the number of miles (for on road) or operating
hours (for off road) to develop an annual operating cost for each asset. One of the major concerns at the
start of the project was how to calculate the change in operating cost with age over a multiple year period
in a manner which would efficiently consider the cost changes in items like fuel, tires, parts and other
components which demonstrated inconsistent annual cost changes. After examination of a number of
different methods, all of which would have entailed significant staff and administrative costs in the long
term for NCDOT to perform a periodic review of optimal life of eventually 200 model classes, we
identified an approach which normalized cost growth for an individual year. Figure 2 demonstrates this
method and plots the cost per mile of class 0201 (pickup trucks) for 2009.
As a starting point to explain analysis of operating cost, for each year of study data, the annual
operating cost per mile or cost per hour was plotted versus the age of the item and best fit modeling
equations were investigated. In general, exponential equations provided a better fit to model the
operating cost data based on two factors.
First, the coefficients of determination1 were higher for the exponential equations compared to
other equations (such as linear).

R2 is commonly called the coefficient of determination and is the proportion of variability in a data set that is
accounted for by the fitted equation. It provides a general measure of how well future outcomes are likely to be
predicted by the model.

TRB 2012 Annual Meeting

Paper revised from original submittal.

6
1
2
3
4
5
6
7

Second the exponential equations were more accurate in predicting the total annual operating
cost. For example, the exponential equation shown in Figure 2 estimated the total annual cost for
with only a 6.6% error compared to over 27% for the linear equation.
It is important to emphasize that this analysis is a foundational starting point and proof of concept for a
multiple year effort to refine these models. Consequently, we plan to continue to improve our
understanding of the operational system reflected in this data and the fit of modeling equations over time.

Cost per Mile, Class 0201 Pickup, 2009

2.50

y = 0.1737e0.0663x
R = 0.14

Cost per Mile, $

2.00

y = 0.0358x + 0.1793
R = 0.0148

1.50
1.00
0.50
0.00
0

10

15

20

Equipment Age, Years

8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

FIGURE 2 Cost per Mile, Class 0201, 2009

One of the major benefits of this approach is that the rate of cost increase can be obtained by
manipulation of the exponential fit equation. The following description explains the relationship of the
exponential equations and the annual estimated percentage change in cost. An equation that predicts the
cost by a percentage increase per year has the form:

P(1 i) n

Equation 1

Where F = the future cost per mile or hour when the piece of equipment is n years old, based on a present
cost per mile or hour of P when the equipment is new and an annual percentage cost increase of i. Since
the exponential curve fit uses a base e, the equation has the form:
y Ae Bx
Equation 2
Equation 2 can be rearranged as:

A(e B ) x

Equation 3

Equations 2 and 3 are identical if y = F, A = P, x = n, and

eB

(1 i )

Equation 4

Therefore, the annual cost increase rate is:

TRB 2012 Annual Meeting

Paper revised from original submittal.

7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

eB 1

Equation 5

For the plot shown above, the annual cost increase rate is:

e 0.0663 1 0.685 or 6.85%

Equation 6

Optimal Life Cycle Cost Models

This section applies the relationships from the previous discussion to develop the optimal life model. It
begins with an overview of life cycle models, examines sensitivity issues, discusses the specific data
which is integrated into the model, and then uses the optimal life model to identify a target retirement
period for an example class.
Engineering economics employs a number of models for equipment replacement studies. A
fundamental concept in these models involves the economic life of an asset. The economic life defines
the operating interval which minimizes the equivalent uniform annual cost (EUAC) of the asset. A key
strength in this approach is that an existing asset can be evaluated based on either an optimum economic
life or a new technology alternative. Most important, these models can be updated on a periodic basis as
new or improved economic data becomes available. This assures thorough and ongoing evaluation of
improvement opportunities. We selected the EUAC framework since it provides a flexible foundation for
the current and future asset decisions which may face NCDOT management.
The primary principle of the model is that the asset should be kept as long as the marginal cost of
one more year of life is less than the EUAC of the previous year (or the challenger) over its economic life.
The analytical approach involves development of a total cost model which reflects cost factors which are
significant in economic impact and important to decision makers. In the case of the six classes in our
study, the challenger is the new unit with equivalent capabilities but with improved operating cost and the
challenger- defender framework can be simplified to an optimal economic life analysis.
Figure 3 demonstrates the basic principle of optimal life models. The total cost curve (at the top)
is comprised of the sum of two components: the capital cost and operations / maintenance cost. As time,
use, or age of the current asset increases, the total asset cost declines initially. In this first part of asset life
(left half of Figure 3), capital cost declines on an annual basis as asset life is spread over more years. On
the other hand, the cost of operation and maintenance (O&M) becomes more expensive but does not
override the decline in capital. As the asset continues to age (right half of Figure 3), O&M cost increases
begin to outpace the decline of the capital cost component The point where the total cost reaches a
minimum is the optimal economic life of the asset and the point at which it should be replaced.

TRB 2012 Annual Meeting

Paper revised from original submittal.

Optimal Economic Life:

Min EUAC
Total Cost
Operations and
Maintenance
Cost

EUAC

Capital Cost

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

Time or Use

FIGURE 3 Graphical Example of Optimal Life Model

In general terms, Equation 7 defines the EUAC2 of asset x (sometimes expressed as the annual
cost) as the sum of annual expense terms, such as maintenance and operating costs, plus the capital
recovery cost incurred by keeping the equipment.
EUACx= (annual O&M and related expenses) + (CR: capital recovery cost)

Equation 7

CR, the capital recovery cost, is calculated in Equation 8 with P = the present market value and S
= the salvage value.
CR = P (A/P, i%, n) S (A/F, i%, n)

Equation 8

Capital Recovery cost (CR) represents the difference between current market values and salvage
and becomes less as the unit ages. Terms such as (A/P, i%, n) represent the standard engineering
economic factors to convert values to present (P), future (F) , or annual worth (A) amounts based on a
given interest rate (i%) and the number of compounding periods (j, k, or n).
The optimal life model we employed builds on the basic EUAC Equation 7 using a year by year,
iterative approach which can be reevaluated annually as new information is identified. To demonstrate
the application of this methodology, we present an example of calculating the EUAC in a specific year of
an asset. On a before tax basis, the present worth of a new asset through year k in the future (PWk) is
described by Equation 9:
k

28

PWk = I-MVk(P/F, i%, k) +

E j (P/F, i%, j)

Equation 9

j 1

29
2

Additional information on EUAC and annual cost can be found in Canada, John, William Sullivan, Dennis
Kulonda, and John White. Capital Investment Analysis for Engineering and Management, 3rd Edition. Prentice Hall,
New York, 2005, p.144 and pp. 274-291.

TRB 2012 Annual Meeting

Paper revised from original submittal.

9
1
2
3
4
5
6
7
8
9
10
11
12
13

where I = initial capital investment, MV = market value at the end of the year of interest, and the
summation is the total of the annual operating and maintenance expenses (Ej) from the current year to
year k in the future.
The total marginal cost of an incremental year of ownership (TCk) is a critical quantity for the
optimal life decision. The difference in present value of the marginal cost of an additional year of
ownership from year (k-1) to year k can be developed using Equation 10.
TCk (i%) = MV k-1 MVk + i*MV k-1 + Ek

Equation 10

where E = annual expenses incurred in the year of interest. Using Equation 10, EUAC can be calculated
using Equation 11.
k

14

EUACk = [

TC j (P/F, i%, j)] (A/P, i%, k) =

Equation 11

j 1

15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44

Table 1 shows the spreadsheet application of these equations for a set of parameters (noted in
Table 2) for class 0201 pickup trucks. For a given year and using the column numbers on the bottom line,
the following bullet points describe the column content:
Column 1 (market value): represents the anticipated market value in a given year. In year 3, this
asset is anticipated to have a market value of $13,956 from its original purchase price of 19,239.
(Note: this example assumes that MV declined at a uniform annual rate defined as the MV
Decline Rate.
Column 2 (Loss in MV): Calculates the incremental loss in MV for keeping the asset one more
year. For example, MV declines from $17,287 to $15,533 from year 1 to 2 for a loss in MV of
$1,952.
Column 3 (Capital cost of MV): The cost of maintaining the investment in the asset and not
liquidating is the interest rate (in this case 3%) times the market value. For example, in year 2,
$15,533*3% = $466.
Column 4 (Miles): This is the anticipated usage of the asset in miles or hours.
Column 5 (Annual Operating Expense): This column multiplies the anticipated cost per mile for
specific year times the annual miles. For year 2, the estimated cost per mile is approximately
$0.20.
Column 6 (Total marginal cost): This column sums columns 2, 3, and 5 to calculate the total
marginal cost of an incremental year of asset use.
Column 7 (PV of marginal cost): This column takes the marginal cost in column 6 to a present
value at the time of asset purchase. For example, the present value of $5,210 at 3% and 2 years is
$4,911.
Column 8 (Total PV): This column adds the PV values for the previous years from column 7. For
example for year 2, $10,069= $5,158 +$4,911.
Column 9 (EUAC): Column 9 converts the total PV in column 8 to a series of uniform annual
costs. For example the equivalent uniform annual cost of a present value of $10,069 at 3% for
two years is two equal payments of $5,262.
The minimum EUAC can be seen in year 8, highlighted in bold.

TRB 2012 Annual Meeting

Paper revised from original submittal.

10
1
2

$5,158
$4,911
$4,701
$4,523
$4,373
$4,245
$4,137
$4,044
$3,965
$3,895

$5,158
$10,069
$14,771
$19,294
$23,667
$27,912
$32,049
$36,093
$40,058
$43,953

EUAC
through
current
year

$5,313
$5,210
$5,137
$5,091
$5,069
$5,069
$5,088
$5,123
$5,173
$5,234

Total PV

$2,842
$2,990
$3,142
$3,299
$3,459
$3,622
$3,788
$3,955
$4,123
$4,291

PV of
marginal
cost

$519
$466
$419
$376
$338
$304
$273
$245
$220
$198

Total
marginal
cost for
year

$1,952
$1,754
$1,576
$1,416
$1,273
$1,143
$1,027
$923
$829
$745

15,710
15,234
14,758
14,282
13,806
13,330
12,855
12,379
11,903
11,427
10,951

Annual
Oper.
Expense

Miles

$19,239
$17,287
$15,533
$13,956
$12,540
$11,268
$10,124
$9,097
$8,174
$7,344
$6,599

Loss in
MV

Capital
cost of
MV

0
1
2
3
4
5
6
7
8
9
10
Col.
#

Market
Value

Year

TABLE 1 Example of Optimal Life Calculation

$5,313
$5,262
$5,222
$5,191
$5,168
$5,152
$5,144
$5,142
$5,145
$5,153
9

3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29

Sensitivity of Optimal Life Model to Parameter Changes

In exploring the optimal life of the six asset classes, we examined the impact of data uncertainty and
variability on the model results. Examples of model variables and their impact include:
Increasing the O&M costs decreases the optimum life interval. There are several factors
impacting O&M including the rate of increase of annual costs, the rate of decline in use with age,
and the starting usage base level. We explored all of these.
Increasing the initial investment increases the optimal replacement interval.
Increasing the rate of return increases the replacement interval.
Increasing the rate of decrease in salvage tends to increase the optimum replacement interval. We
explored possible changes in the rate of market value decline.
Our approach in sensitivity analysis employed three components.
Base scenario: As a foundation, we searched for the optimal life using the base scenario
information above. In general, this scenario represents current levels of use and operation.
One-at-a-time changes: We examined variable changes one at a time in 10% change steps from
the base values, either increase or decrease as appropriate. This included rate of decline in market
value, the rate of decline of use with age, rate of increase of operating cost and base miles or
hours of use. One at a time sensitivity analysis provides insight as to which parameters have the
largest impact on optimal life and is explored in more detail in the next section.
Most Likely - Several variables changing together: This is the scenario reflecting the anticipated
future. For example, if the fleets are reduced and utilization increases, this will impact the
starting usage level (increase), the rate of decline of usage over time (smaller), and the
acceleration of higher operating cost (increase) compared to the base scenario. Using the model
variables outlined in Table 2 for the most likely scenario, the model predicts the estimated
economic life for the six classes as:
Class 0201-Pickup Truck: Optimal economic life identified in the 8-9 year range.

TRB 2012 Annual Meeting

Paper revised from original submittal.

11
1
2
3
4
5
6
7

Class 0205-Dump Truck: Optimal economic life identified in the 8 year range.
Class 0206-Flat Bed Truck: Optimal economic life identified in the 8-9 year range
Class 0314 Loader Backhoe: Optimal economic life identified in the 8 year range.
Class 0900 Motor Grader: Optimal economic life identified in the 11 year range.
Class 2002 Front End Loader: Optimal economic life identified in the 12 year range.

Class 0201

3%

Class 0205
Class 0206
(truck 2)
Class 0314

3%

TABLE 2 Economic Life Model Input Parameters

Base
Annual
annual
miles (or
Market
Market value
miles (or
operating
value
annual
operating
hours)
base
decline rate
hours)
decline
15,710
$19,239
10%
miles
475 miles
13,915
$51,853
11%
miles
431 miles

3%

$52,130

9%

9,533 miles

181 miles

$0.58

6.9%

3%

$71,574

8%

595 hours

21.6 hours

$12.72

8.9%

Class 0900

3%

$154,030

8%

836 hours

19.6 hours

$19.88

6.3%

Class 2002

3%

$107,364

8%

599 hours

10.6 hours

$15.76

6.0%

Interest
rate

8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

Base cost
per mile (or
hour)

Annual increase
in cost per mile
(or hour) rate

$0.17

7.9%

$0.61

8.1%

As noted, the data columns in Table 2 indicate the values employed in the economic life model.
It is important that this model and its parameters are consistently and periodically examined and updated
based on updated NCDOT cost and market data.
One-at-a-Time Sensitivity Analysis
An analysis was performed to determine which parameters employed in the optimal life model showed
the greatest amount of sensitivity to change model results. The parameters analyzed included the interest
rate, initial market value, the market value decline rate, mileage decline, cost per mile to operate and
maintain the vehicle, and the annual cost increase. The goals of this analysis were to examine each
parameter and determine the sensitivity of the model to the value change of the parameter and the relative
sensitivity of the parameters compared to each other.
The first step to performing this analysis was to use the baseline values input into the optimal life
model and determine the disposal year for that vehicle. In the case of class 0201, the disposal year was
determined to be 8 years as shown in Table 1. The six variables used in this calculation are shown in
Table 3 in the column labeled 100%. Each of the six variables were then varied 10% above the baseline
value and 10% below the baseline value and the disposal year was calculated using the optimal life model
changing one variable at a time. For example, to find the optimal disposal age when the interest rate was
10% lower than its original value of 0.03, interest rate was changed to 0.027 while all other variables
were kept at their baseline values. The disposal year calculated for this set of variables was 7 years.
Table 3 shows the disposal year as each variable is independently changed. Within the range studied, the
disposal year varied from a low of 6 years to a high of 11 years.

TRB 2012 Annual Meeting

Paper revised from original submittal.

12
1
2

TABLE 3 Varied Parameters Input in EUAC Model to Determine Optimal Life

Parameter
Interest Rate
Initial MV ($)
MV Decline Rate
Mileage Decline
(miles/year)
Cost/Mile ($/mile)
Annual Cost Increase
Rate

3
4
5
6
7
8
9
10
11
12

90%
Value
0.027
17315
0.0913
427.5

Life
7
6
6
7

0.163
0.0765

9
11

Parameter Value and Optimal Life

100%
Value
Life
0.03
8
19239
8
0.1015
8
475
8
0.181
0.0850

8
8

110%
Value
0.033
21163
0.1116
522.5

Life
8
9
9
9

0.199
0.0935

6
6

The graphical representation of the data in Table 3 is shown in Figure 4. Note that a slight offset
was applied to some of the parameters to make the graph more readable. The slope of each parameters
response line indicates the sensitivity of the disposal year to the given parameter, and graphically, the
slope is the partial derivative of the disposal year with respect to the parameter. A comparison of the
slopes of each parameter indicates the relative sensitivity of the disposal year to each parameter. For
example, it is apparent from the slope of the interest rate data that the disposal year was hardly affected by
the 10% variation. Alternatively, the disposal year is very sensitive to the variation in annual cost
increase as indicated by the steep slope of the response line.
12
Interest Rate

11

Initial MV
MV Decline Rate

10

Mileage Decline

Age, years

Cost/Mile
Annual Cost
Increase Rate

80

13
14
15
16
17
18
19

90

100

110

120

Percent Change in Parameter

FIGURE 4 Baseline Sensitivity Analysis, Class 0201

The value of the sensitivity (slope) of each response line in Figure 4 provides a meaningful
comparison of the relative impact of each parameter in the model. Table 4 shows the average value of the
sensitivity for each parameter as well as the standard deviation and coefficient of variation (standard
deviation/mean) of the sensitivity. When each parameter is ranked from lowest to highest sensitivity

TRB 2012 Annual Meeting

Paper revised from original submittal.

13
1
2
3
4
5
6
7
8
9

according to the magnitude of the slopes of the response lines, the sensitivities are ordered as follows:
interest rate, cost per mile, initial market value, mileage decline, market value decline rate, annual cost
increase rate. The coefficient of variation (COV) is listed in Table 4 because it serves as an indication of
how the sensitivity of a given parameter varies over the analysis space. A high COV reveals that the
parameter has some dependency upon the other parameters within the study whereas a low COV indicates
that the parameter is relatively independent of the other parameters. A good example of a highly
dependent parameter as indicated by COV is the interest rate.
TABLE 4 Sensitivities Disposal Year to Each Parameter

Mean
St. Deviation
COV, %

10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37

Sensitivity of Disposal Year to Parameter (year/percent change)

Initial
Market
Annual Cost
Mileage
Interest Rate
Market
Value
Cost/Mile
Increase
Decline
Value
Decline Rate
Rate
0.035
0.138
0.173
0.154
-0.131
-0.285
0.049
0.050
0.045
0.121
0.047
0.047
140
36
26
78
-36
-39

Based upon this one-at-a-time sensitivity analysis, the disposal year is most dependent upon the annual
cost increase rate, which has about an order of magnitude the sensitivity as the interest rate and about
twice the sensitivity as the other parameters. In fact, the annual cost increase rate only needs to change
off the baseline value by 3.5% (1/0.285) in order to change the disposal year by one year, whereas the
interest rate needs to change off the baseline value by 28.6% (1/0.035) to affect the same one year change
in the disposal year. It is important to note that this analysis is only valid within the space that has been
analyzed and while a 28.6% increase in the baseline 3.0% interest rate results in an interest rate of only
3.9%, this value is already outside of the studied space. Additionally, this example illustrates the need to
expand the sensitivity analysis as market conditions change.
Conclusions
This paper has presented the application of EUAC methods to identify the optimal life cycle for six
common equipment classes found in departments of transportation. This model requires accurate and
thorough data but once it is organized and implemented, it can serve as a strong management tool to
justify equipment expenditures which make sound business sense. In this time of tight budgets, this is an
essential foundation for equipment decisions which reduce costs and increase utilization.
We have included substantial coverage on sensitivity issues in the model since applications in other
geographic regions, in other equipment classes, or in other operational situations may yield different
results based on input changes. It is important the engineers and analysts include examination of the
impact of variation in model parameters on the decision results.
In the future, we plan to further explain and examine other modeling issues such as the decline in market
or salvage value and the impact of varying levels of utilization. We invite interested collaborators to
contact us so that larger data sets can be analyzed and broader applications can be developed.

TRB 2012 Annual Meeting

Paper revised from original submittal.

14
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31

References
1. Canada, John R., William Sullivan, Dennis J. Kulonda, and John White. Capital Investment
Analysis for Engineering and Management, 3rd Edition. Pearson- Prentice Hall, Upper Saddle
River, NJ, 2005.
2. Park, Chan S. and Gunter P. Sharpe- Bette. Advanced Engineering Economics. John Wiley, New
York, 1990.
3. P.C. Jones, J.L. Zydiak, W.J. Hopp. 1989. "Stable Economic Depreciation and Neutral
Replacement Decisions," The Engineering Economist, 34, 115-128.
4. P.C. Jones, J.L. Zydiak, W.J. Hopp. 1992. "Capital Asset Valuation and Depreciation for
Stochastically Deteriorating Equipment," The Engineering Economist, 38, 19-30.
5. P.C. Jones, J.L. Zydiak. 1999. " A Further Note on the Fleet Design Problem," Engineering
Economist 45.
6. JLARC - Joint Legislative Audit and Review Commission. Estimating Optimum Mileage for
Vehicle Retirement. Retrieved from jlarc.state.va.us/reports/techappn.pdf, July 15, 2008.
7. Gransberg, Douglas D., Calin Popescu, and Richard C. Ryan. Construction Equipment
Management for Engineers, Estimators, and Owners. CRC Press, 2006.
8. Tatum, C. B., Michael Vorster, Mac G. Klingler; and Boyd C. Paulson. Systems Analysis of
Technical Advancement in Earthmoving Equipment. Journal of Construction Engineering and
Management, Vol. 132, No. 9, September 2006.
9. Wyrick, D. A., and B. J. Storhaug. Benchmarking fleet management. Duluth MN: Northland
Advanced Transportation Systems Research Laboratories, University of Minnesota, 2003.
10. Kim, D. S., Porter, J. D., Kriett, P., Mbugua, W., & Wagner, T. (2009). Fleet replacement
modeling final report. School of Mechanical, Industrial and Manufacturing Engineering, Oregon
State University, 2009.
11. Gillespie, J. S. and A. S. Hyde. The replace/repair decision for heavy equipment. Virginia
Transportation Research Council, Charlottesville, VA, 2004.

TRB 2012 Annual Meeting

Paper revised from original submittal.