The 336th Transportation Group (Forward) conducted
a study to determine the point in a vehicle’s life when
the value of a refurbishment at or above the –20 level
is maximized. [Maintenance at or below the –20 level
is performed by unit mechanics.] Our expectation as we began
the study was that a refurbishment would curb the rapid escalation
in cost and frequency of repairs and extend the usable life
of the vehicle. To explore this hypothesis, I collected data
from the CALIBRE Integrated Logistics Analysis Program (ILAP)
support office at Camp Arifjan, Kuwait, and data from the
336th Transportation Group’s Unit Level Logistics System-Ground
(ULLS–G) computer. [CALIBRE is a management and technology
services company serving the Army and the other services.]
I aggregated the data from both data sets, matching dates,
mileages, repair costs, and days down for the M1114 up-armored
high-mobility multipurpose wheeled vehicle, M915 truck, and
M1070 heavy equipment transporter truck.
To test the hypothesis, I built a simple time/value microeconomic
model, where mileage represented time on the x-axis and both
days down and repair cost represented value on the y-axis.
I created a scatter plot for each vehicle type and added a
trend line to clarify patterns in the data. I also conducted
a regression analysis for the parts cost only to determine
the relationship between percent change in repair cost with
respect to mileage.
I expected repair costs to increase gradually early in the
vehicle’s life. I also hypothesized that a readily identifiable
point would exist in the mileage of a vehicle, after which
the cost and frequency of repairs would increase rapidly.
Performing a refurbishment at that point would flatten the
cost curve and extend the life of the vehicle, maximizing
the value of the refurbishment. What follows is a summary
of my results for each vehicle type.
A weak relationship exists between higher mileage and greater
number of days down per repair for the M1114 vehicle type.
Over a 70,000-mile life of M1114s in the Army’s fleet,
the average number of days down per repair increased only
3 days, from 38 to 41 days. This represented only an 8-percent
increase over 70,000 miles, which was far less than I initially
For parts cost, the trend line appeared almost flat, showing
no increase in parts cost over the lifetime of the M1114.
Analyzing the greatest concentration of vehicles, which had
between 45,000 and 65,000 miles of use, supported this trend.
No significant increase in parts cost over the lifetime of
an M1114 was observed. This conclusion also contradicted
my hypothesis that parts cost would increase over the life-time
of a vehicle.
The regression analysis also contradicted my hypothesis.
The slope of the regression equation was -0.121, meaning that
an increase of 1 unit of mileage is related to a 12-percent
decrease in 1 unit of parts cost. Essentially, increased mileage
is related to decreased parts cost. [Regression analysis measures
the relationship between one or more independent variables
and a dependent variable. In this case, mileage is the independent
variable and parts cost is the dependent variable. A regression
equation is generated from the dataset, and the coefficient
of correlation is derived from the regression equation. The
coefficient of correlation measures the strength of the relationship
between the two variables. A coefficient of correlation of
1 is the perfect linear relationship: an increase of 1 unit
of mileage is correlated to an increase in 1 unit of parts
cost. A coefficient of correlation of 0 indicates no relationship.
To say that a 1-unit increase in mileage “causes” a
1-unit increase in parts cost is a misnomer. We can only
state that a strong relationship exists; we cannot say that
there is causation.]
A weak relationship also exists between higher mileage and greater days down
per repair for the M915 trucks, mirroring the results from the M1114 series.
The trend line increased from 38 to 42 days down per repair over the lifetime
of the M915 series. This represented an 11-percent increase over 100,000 miles,
from 20,000 to 120,000 miles. I initially expected the relationship between
higher mileage and greater days down per repair to be stronger, especially for
vehicles with over 100,000 miles of use.
Counterintuitively, an inverse relationship exists between mileage and parts
cost for the M915. While I expected that higher mileage would positively correlate
with higher parts cost, I found that, as mileage increased, parts cost decreased
over the lifetime of the M915. While the trend line demonstrated this strong
relationship, the result may be caused by problems in the data set, which I will
discuss later in this article.
The regression analysis returned a different result from the trendline analysis.
The slope of the regression equation was 0.012, meaning that mileage accounted
for approximately 1.2 percent of the change in parts cost.
A weak relationship also exists between higher mileage and greater number of
days down per repair for the M1070 vehicle type. Over the lifetime of an M1070,
the number of days down per repair increased from 35 to 38. This represented
only a 9-percent increase over 130,000 miles, which was far less than I originally
expected, especially for such a long lifetime.
A positive relationship exists between higher mileage and greater parts cost
for the total data set of the M1070. A $20,000 increase in cost occurred between
0 and 200,000 miles. Breaking the data set down to observe the majority of the
fleet’s concentration, which had accrued between 25,000 and 50,000 miles,
revealed an increase from $1,900 to $2,200 per repair. The $300 increase represented
a 16-percent increase over the parts cost for repairs at 25,000 miles—a
statistically significant result.
The regression analysis returned an interesting result. The slope of the regression
equation was 0.149, which meant that mileage accounted for approximately 15
percent of the change in repair cost. This represented the strongest relationship
returned by a regression analysis.
For all vehicle types, the fundamental assumption of this analysis was that all
vehicles of a specific type are fungible. In other words, one M1114 with 50,000
miles of use is exactly the same as another M1114 with 50,000 miles of use. On-the-ground
experience undermines this assumption. Company-level leaders know that different
vehicles have different “personalities,” which may result in one
vehicle needing more repairs than another. Perhaps these “personalities” are
due to manufacturing irregularities and will be eliminated as the Army increases
demand for Six Sigma quality during manufacturing. This assumption of vehicle
fungibility was one of the weaknesses of the study.
Another weakness was that data collected from ULLS and ILAP are snapshots in
time. Battle damage and battle loss costs and days down are not reported to
either the unit’s ULLS box or ILAP, so they are not reflected in this study.
Vehicles used by surge units may keep reported mileages and repair costs artificially
low. When surge units complete their short tours, they usually turn their vehicles
over to year-long tour units, and year-long tour units subsequently turn their
vehicles in for refurbishment. This further complicates tracking repair cost
over the lifetime of an individual vehicle.
For all vehicles, the data collected may be inaccurate. The imperatives of conducting
combat logistics patrols trump accurate data collection, and this may cause irregular
reporting of repair costs at specific mileages. For the M915 vehicle type, too
few data points existed at the time the data were collected. This problem may
have caused the counterintuitive result for decreasing parts cost over increasing
mileage. For the M1070 vehicle type, the data indicate a positive relationship
between higher mileage and higher parts cost, especially for vehicles in the
25,000- to 50,000-mile range in which the majority of the fleet is concentrated.
Unfortunately, neither the M1114 nor the M915 vehicle types indicate such a strong
positive relationship for either parts cost or number of days down per repair.
I observed less than a 10-percent increase in parts cost over the lifetime of
the M1114 and M915. These results are insignificant considering that the increases
occur over 100,000 miles of use.
My initial hypothesis was incorrect. The presence of a readily identifiable point
along the trend line representing the mileage of a vehicle, after which costs
would rise rapidly, did not exist for any vehicle type. A gradual increase early
in the lifetime of a vehicle was also not present in the data. Overall, a very
slight increase in a linear relationship over the lifetime of a vehicle was observed.
The regression analysis also returned a weak correlation between increased mileage
and increased parts cost, with the exception of the 16-percent relationship on
The conclusion for Army leaders? This study could find no significant statistical
relationship for higher mileage accounting for higher repair cost, except for
the M1070. Perhaps a follow-up study would find different conclusions, but operators
and mechanics are still our best way of determining which vehicles need to
be refurbished. As a result, refurbishment programs should be conducted on a
vehicle-by-vehicle basis, as determined by the unit’s maintenance officer,
with input from the line mechanics and operators.
First Lieutenant James B. (Brad) Hamlett is the Supply and Services Officer
the 336th Transpor-tation Group at Camp Arifjan, Kuwait. He holds a bachelor’s
degree from Furman University and is a graduate of the Transportation Officer