American Journal of Civil Engineering and Architecture, 2018, Vol. 6, No.

2, 80-92
Available online at [Link]
©Science and Education Publishing
DOI:10.12691/ajcea-6-2-5

A Statistical-based Approach to Evaluate the Production

of Crawler-type Dozer in Construction Projects
Hamed Nabizadeh Rafsanjani1,*, Yaghob Gholipour2, Xiaoxiang Xue3
1
Durham School of Architectural Engineering and Construction, University of Nebraska-Lincoln, Lincoln, NE 68588, USA;
formerly, School of Civil Engineering, University of Tehran, Tehran 14155, Iran
2
School of Civil Engineering, University of Tehran, Tehran 14155, Iran
3
Durham School of Architectural Engineering and Construction, University of Nebraska-Lincoln, Omaha, NE 68182, USA; Project
Engineer, Jacobs Engineering Group, Omaha, NE 68134, USA.
*Corresponding author: hnabizadehrafsanj2@[Link]

Abstract Estimating the actual production rate of construction machinery which clearly differs from nominal
production provided by machinery manufacturers always is a critical challenge in construction projects execution.
Studies indicate that true estimation of actual production is a key element in estimating the time and cost required to
terminate construction operations. However, current literature shows that it is still a quite challenging to estimate
actual production. In particular, there are various independent parameters that affect the actual production rate.
Understanding the role and importance of each parameter could lead to an accurate estimation for production rates.
To this end, this paper presents a statistical-based approach to find the discrepancies between the nominal and actual
production rates of crawler-type dozers and to understand how various parameters could affects the actual
production rates. The data for the actual production of machine were records from productivity measurement of 39
dozers. Working condition, type of materials, and ground slope are three main independent parameters considered
and evaluated for each machine. The results obtained from statistical analyses on the data and a comparison between
these results with the data provided by Caterpillar and Komatsu manufacturers show a) the discrepancies between
the actual and nominal hourly production, b) the effect of individual parameters on actual production, and c) the
relationship between machinery working age and discrepancies. The findings of this study could be a unique help for
project managers in planning of machinery and equipment in a project site.
Keywords: heavy construction machinery, crawler-type dozer, actual hourly production estimation, production
parameters, statistical analysis, case study
Cite This Article: Hamed Nabizadeh Rafsanjani, Yaghob Gholipour, and Xiaoxiang Xue, “A Statistical-
based Approach to Evaluate the Production of Crawler-type Dozer in Construction Projects.” American Journal
of Civil Engineering and Architecture, vol. 6, no. 2 (2018): 80-92. doi: 10.12691/ajcea-6-2-5.

management and construction project performance [3].

Machinery manufacturers generally provide an ideal
1. Introduction and Motivation hourly production of their machinery for users. This ideal
hourly production named nominal hourly production
Construction machinery and equipment are one of clearly differs from the actual hourly production of a
the necessary resources to the accomplishment and machine in construction projects. The actual production
success of construction projects. Today contractors depends mainly on the conditions of project sites.
undertake many types of construction activities that Estimating actual production and therefore the
require different types and sizes of machinery and discrepancies between the nominal and actual production
equipment [1], and they typically invest the largest value of rate is a key element in estimating the time and cost
project costs in machinery and equipment specifically for required to terminate the construction operations. In order
the equipment-intensive projects [2,3]. In this context, to estimate this production, it is a critically need to
adequate knowledge of machinery and equipment know how different project site conditions affect the
management critically needs to decrease various risks machinery production [3]. In fact, since the ideal
associated with increasing the overall project cost. Fan et conditions considered by manufacturers could rarely be
al. [4] and Tatari and Skibniewski [5] provided the critical achieved in real construction projects, the actual
factors for machinery management. In particular, the production could differ with the nominal production.
success of a construction project is highly connected to its A growing body of research has studied different
machinery production, and it has been universally subjects of construction machinery and equipment
accepted that the machinery hourly production is the key management [13-21]. Within this area of interest,
factor in success of construction projects [1,3,6-12]. This researchers have extensively provided new algorithms,
factor is the main relationship between the machinery methods, models, and techniques to study about
81 American Journal of Civil Engineering and Architecture

acquisition, costing, maintenance, purchase, repair, there are so few publications about production estimation
replacement, retirement, selection, and sustainability gives evidence to the fact that less attention has been paid
topics [2,5,22]. In addition, various related computer to this domain than to the other main domains of
software has been developed and proposed to help project machinery; the available research in production estimation
managers make decisions about construction machinery. domain is significantly insufficient and not proportional to
In particular, the success of such research work and the importance of the topic. Therefore, in order to advance
software is highly depends on machinery actual the basis of topic, this paper considers a case study of
production [9,11,23-26]. Rafsanjani et al. [10] discussed various models of crawler-type dozer to evaluate the
that the nominal to actual hourly production could get a machine production in project sites and to discuss about
ratio of less than 0.3 in a project site. Such a high the main parameters affect the production.
efficiency drop might critically affect results and This paper is organized as follows: research objectives
conclusions by machinery research working on other is presented in section 2. Section 3 describes the research
fields (e.g., costing, replacement, and selection). However, methodology. In section 4, the results are provided.
current literature shows that the accurate estimation of Section 5 makes a discussion on the results. Limitations of
hourly production has only intrigued very little research in this research and recommendation for future research
the past decades. In 1994, Edmonds et al. [27] took the work are presented in section 6. Finally, section 7 presents
actual production of various earthmoving operation conclusions.
machinery into account, and viewed the actual production
as a percentage of full capacity. By using several methods
such as short range analysis and analysis of running time- 2. Objectives
speed, they finally conclude that the actual production of
machinery is approximately 52.5 percent of the nominal The main objective of this paper is to develop a step-
production. It is worth mentioning that providing a by-step approach in order to assess actual hourly
specific amount of production drop for a range of production of crawler type dozer. This is achieved through
machinery is not a feasible approach being implemented statistical analyzing the real data of a case study for
in different project sites, and therefore their percentage various models of crawler-type dozer. This objective is
result could not be extended to construction machinery translated into three sub-objectives:
industry. Zou [28] used the 3-D HSV color-space digital (1) Find the discrepancies between the actual and
image processing method to study how different site nominal hourly production of the machines. The
conditions could influence on excavator idle time and results of this objective give an overview of how
truck load cycle time. Through the proposed method, he the efficiency of machinery and equipment in a real
tried to achieve more realistic results of machinery project site can be found. These results could greatly
production and discussed how this method can be help estimate the most realistic time machinery and
extended to understand the effects of site conditions on equipment fleet needed to complete their assigned
production of other machinery and equipment. However, tasks.
finally no data for actual hourly production of the (2) Determine the individual parameters which affect
machines was provided. Rafsanjani et al. [8,10] studied the machine production. This step is essential to
the hourly production of different machinery models and present why the lowest discrepancies might be achieved.
provided the actual hourly production for some machines: (3) Determine how a machine working age affects its
excavators, graders, loaders, and rollers. In their work to actual hourly production. A special emphasis on this
find the actual production a three stage analysis was sub-objective helps to understand the relationship
employed: long-range analysis, ratio analysis, and analysis between the working age and hourly production, and
of variance. In addition, the authors provided the actual how the production can be predicted based on the age.
conditions of construction sites where the data of Addressing these sub-objectives constitutes a significant
machines was collected. However, they did not discussed contribution to the field of machinery management. It is
how different projects site conditions affected the worth mentioning that although this paper only focuses on
machinery production. Based on their previous studies, crawler-type dozer, the approach is inclusive and can be
Rafsanjani et al. [11] then proposed a regression-based extended to other machinery and equipment employed in
model to estimate actual production of wheel-type loader. construction sites.
The model uses the engine power (hp) of machine to The main reason for the selection of dozer for this study
estimate the actual production. However, the authors did is that this machine either crawler or wheel-type is
not discussed how the model could be used for a variety of typically the starter machine in a fleet of machinery
project conditions. In fact, the model could not be working in a project site, and it actually acts as a feeder
extended in-practice to all models of machines working in for the machinery fleet. Therefore, any delay in the dozer
various project sites. Recently, Rashidi et al. [23] used a work could have negatively effects on the other machinery
linear mixed model to estimate productivity of a model of and equipment and thus a negative effect on the project.
dozer. They presented factors affect machine productions. Therefore, there is a need to know how this machine
However, their results could not be generalized to works in a site. In particular, a crawler-type dozer can
different models of dozer typically used in project sites. work on a variety of soils. While sharp-edged pieces of
In particular, the review of literature apparently shows stone are always a problem for wheel dozer, such pieces
that over the past 20 years, construction literatures have are not destructive to a crawler type one [3]. A crawler
written little information to advance practical basis for dozer can also work on mud-slick surface. In addition,
actual machinery production estimation. The fact that compared to other machines and equipment, this machine
 American Journal of Civil Engineering and Architecture 82

provides the highest power, traction, and flotation needed prevent bias in data that could happen by only considering
for doing different variety of working conditions; it can a model of machine. It is worth mentioning that while
easily operate on a slope of 100% [3]. Therefore, the crawler-type dozer could work with a variety of
crawler-type dozer is often used in project site. It is worth blade such as straight or angel blade, all the machines
mentioning that Rashidi et al. [23] and Ok and Sinha [26] considered for this research worked with Semi-Universal
studied dozers to provide their proposed estimation (SU) blades.
models. To this end, the focus of this study is on the
crawler-type dozer machine. Table 1. Number and Age for each Model
Model Number Age (years)
D6N (H) 6 8 10 10 11 13 14
3. Methodology D6T (G) 5 8 9 11 11 14

Caterpillar
D7R (F) 6 6 10 10 10 12 13
A step-to-step methodology was developed to achieve D9T (C) 2 4 11
the sub-objectives. First, in order to get the actual D10T (B) 2 7 9
production data, the author employed a real survey on D155A-2 6 2 4 4 5 5 6
different models of crawler-type dozer in real projects.

Komatsu
D155A-6 5 2 2 3 7 7
Then, manufacturers’ publications were used to collect D275A 3 2 3 5
data for the nominal production. Finally, required D375A 4 3 3 4 6
statistical analyses were employed to analyze all data in
order to achieve the sub-objectives.
3.3. Data Collection
3.1. Case Study of Real Projects 3.3.1. Nominal Production
Crawler-type dozer generally used for dozing and The data for the nominal hourly production was derived
making compacted soil, materials, and earth loose in from machinery charts and performance handbooks of
project sites. In construction industry, this machine is the two manufacturers: Caterpillar Performance Handbook
mainly used in borrow pits to dig the earth. Therefore, the [32], and Komatsu Specifications & Application Handbook
projects chosen for this case study are five different [33]. Figure 1 presents charts for nominal hourly
borrow pits which are located around Iran. These pits have production of crawler-type dozer. The horizontal axis of
generally been dug for use in different construction each chart shows the value for average of dozing distance
projects, e.g. highways and dams. The borrow pits are while the vertical axis presents the hourly production for
located in areas containing almost similar climate different models. The average dozing distance for all
conditions and annual rainfall. In addition, the local models of Komatsu is considered from 15m to 180m.
weather of these areas is hot and dry. The soil samples However, the upper bound of this distance is limited to
taken from all five borrow pits had very little natural 90m for model F, G, and H of Caterpillar, while that of the
moisture. Therefore, it was assumed the pits soil is dry. other models of caterpillar is more than 180m. In addition,
Generally, although bias can occurs at any step of the charts show that the highest hourly production in the
research, data collected during an experiment or survey ideal case is approximately 2700 Lm3 for both Caterpillar
study is more likely to be biased [29,30]. Bias in data and Komatsu dozers.
could mislead the statistical analysis results and study
conclusions. Accordingly, machinery data collected only 3.3.2. Actual Production
based on a project could be biased due to certain condition The actual production data was collected from
of the project. In this context, five different pits chosen for records of different models of dozers in the five borrow
this study could help avoid the bias in data. pits. In this step, each model was individually considered
at a project site. In data gathering phase, based on
3.2. Machine Selection manufacturers’ production charts (see Figure 1) the
average dozing distance of machine was considered as a
The choice of machine in this study was limited to parameter to achieve the production presented in a way
two manufacturers’ models: Caterpillar, and Komatsu. that the manufacturers present their machinery production.
Caterpillar believes to control more than half of the U.S. In this context, the hourly production for all machines in
construction equipment market and one third of the world some specified dozing distances was recorded. As shown
market [31]. Komatsu dozers are also typically employed in Figure 1, the manufacturers typically consider some
in many earthmoving operation [23]. The machine models dozing distances of interest to present their results of
considered for this study are as follow: hourly production (e.g., 15m, 30m, and 45m). In the case
• Caterpillar: D6N, D6T, D7R, D9T, D10T of this research, before the data collection step, a pre-
• Komatsu: D155A-2, D155A-6, D275A, D375A survey was done to determine the typical dozing distance
Table 1 shows the number and age of each model. in five pits. Based on the results of this survey, 15m, 30m,
There were totally 39 machines with average age of 7.3 and 75m were finally chosen as the three main dozing
working years. In addition, the Caterpillar models worked distance of interest for all models. In other words, these
6 years on average more than the Komatsu models. The three distances were the only distances that the author
variety range for models and their age could provide this collected the real data of all the models. In particular, a
opportunity to understand how the age might affect the comparison between the collected actual data and found
productivity of machines. In addition, they can also help nominal data could address the first sub-objective.
83 American Journal of Civil Engineering and Architecture

Figure 1. Manufacturers charts for crawler-type dozer hourly production (Caterpillar performance handbook 2012, Komatsu specifications &
application handbook 2009)

In addition to the dozing distance, working condition, Furthermore, the skills and working experience of
ground slope (i.e., grade%), and materials (i.e., soil) machine operators also have great influence on machine
conditions have critical effects on dozer production production. Elazouni and Basha [35] and Alwi [36]
[3,10,21,22,23,34]. Indeed, these three independent mentioned that the poor performance of machine operators
parameters are the main parameters critically play can lead to a low production. Figure 2 shows the
important roles to estimate dozer actual hourly production. distribution of working experience of machine operators
Therefore, in this study, several surveys were conducted in the case of this research. The majority of operators were
on different conditions to study the effect of each more than five years of working experience. It is
parameter on machine productions. In order to address this noteworthy that the operators assigned randomly to
step as the second sub-objective of this study, it was different machines that not only leads to achieve more
needed to study various operating conditions of dozer at reliable results extended to machinery industry, but also
the borrow pits. In this context, based on the projects helps avoid bias in data.
conditions, project machinery managers’ experience, and The time of data gathering in the five projects was
the aforementioned literatures, the author categorized each approximately 8 months. During this period, the dozers
of these three independent parameters as follow: run for one 8-hour shift per day and 20 days per month. In
• Working condition: (1) Good; (2) Medium; (3) Weak particular, it was tried hard to collect the actual production
• Ground slope: (1) Zero; (2) +15%; (3) −15% data of each model for various project site conditions
• Types of Materials: (1) Loose soil; (2) Soil during this 8-month period of time that could lead to
containing rubble stones; (3) Blaster rocks achieve more reliable results. In addition, a dozer actual
In a good working condition, it was assumed an hourly production was estimated by dividing the total
experienced operator operates a machine with low-working volume of soil loosed by machine by the total operational
age in an ideal site condition. In this condition, the hours for a working day.
efficiency for time-working of a machine is approximately In addition, based on the three sub-categories
50min per hour [3,9,10,12,32,33]. 40min per hour and considered for each main parameter, 27 (i.e., 3×3×3)
30min per hour are the time-working for medium and different scenarios could be defined for each dozing
weak working conditions, respectively. Depending on distance of each model. For example, a scenario for a
various operating terms and conditions, a project site machine with dozing distance of 15m can be: to work with
condition can be considered as one of the three working blaster rocks in a good working condition on a grade of
conditions. It must be noticed that the experience of +15%. Accordingly, 26 other scenarios could be
machinery manager in construction projects plays the considered for this machine. In order to construct and
main role to make this critical decision. In particular, the study all scenarios for all dozers, during the 8-month
machinery and equipment manager of five borrow pits has period of time, all models were separately tracked to
28 years of experience in earthmoving operations. collect all required data.
 American Journal of Civil Engineering and Architecture 84

Figure 2. Working experience (years) of machine operators

Figure 3. A sample set of real data

3.4. Data Analysis each model were finally collected. Figure 3 presents a
sample set of the data collected for model D7R of
After collecting the required actual data, appropriate Caterpillar dozer with dozing distance of 30m. This
statistical analyses were employed to find best representative samples shows that actual hourly production data ranges
for each scenario of each model. In this context, for each approximately from 150 to 200 m3 that shows a 25%
individual scenario a sample of 40 data was collected for difference in the range of data.
each individual project. These samples were not collected According to the classification of the collected data,
in consecutive days; collecting data in consecutive days One-Way ANOVA was selected and employed to analyze
might provide opportunity for happening the bias in data. the data and find a representative for each scenario. One-
In addition, the minimum number of data more than 30 for Way ANOVA is a numerical statistical method for
each sample could help that the sample set would be comparing three or more independent sample groups in
approximately normal [37-41]. Such data sets would be order to find whether the differences between means of
suitable to conduct statistical analysis with reliable results the groups are statistically significant or not. In other
[42-47]. Since there were 5 borrow pits in the case of this words, this method is typically used to statistically
research, 200 (i.e., 5 × 40) samples for each scenario of understand the differences among independent groups of
85 American Journal of Civil Engineering and Architecture

samples. In this research, the One-Way ANOVA tested a Table 3 and Table 4 provide the results of One-Way
hypothesis on sample means for each scenario of ANOVA for this scenario. Figure 4 and Figure 5 also
individual dozers. In this context, the null and alternative show the production charts for the data provided in
hypotheses are as follow: Table 3 and Table 4, respectively. Based on the found
- Null hypothesis: There are no statistically significant three points for each model, the best curve was fitted to
differences among means of 5 sample sets for each model the points. Compared to Figure 1, Figure 4 and Figure 5
of dozer. present the actual hourly production for Caterpillar and
- Alternative hypothesis: There are statistically Komatsu dozers, respectively. For models of Caterpillar
significant differences among means of 5 sample sets for dozer presented in Figure 4, by increasing the dozing
each model of dozer. distance, the rate of production for model of B and C
In the context of One-Way ANOVA, when the null decreases by an amount significantly higher than that of
hypothesis is not rejected, the grand mean (i.e. the mean for F, G, and H. However, this rate remains approximately
for all samples) is selected as representative for all the the same, specifically for dozing distance more than 30m,
data. In this research, the null hypothesis was tested for all models of Komatsu (see Figure 5).
through five independent groups of 40 samples for each
machine. Table 2, as an example, shows the results for Table 3. Actual Hourly Production for Models of Caterpillar
One-Way ANOVA analysis employed on the data Model
Dozing
Actual Hourly Production (Lm3/hr)
collected for Caterpillar model D7R with dozing distance Distance(m)
of 30m (see Figure 3). In this example, F-value is less than 15 149
the critical value, and therefore the null hypothesis was D6N (H) 30 101
failed to be rejected. In other words, the result indicates 75 43
that the grand mean is the representative for the data of
15 281
this scenario. Accordingly, all data sets for all models
were tested, and for all them the null hypothesis was failed D6T (G) 30 144
to be rejected; for few machines the null hypothesis was 75 62
marginally failed to be rejected. It is worth mentioning 15 305
that all estimations and tests were performed in the open-
source statistical language R. D7R (F) 30 174
75 86
Table 2. Sample of One-Way ANOVA Analysis
15 912
Source df SS MS F Critical Value D9T (C) 30 536
75 289
Model 4 2401.93 600.48 2.54 3.42
15 1353
Error 195 46086.15 236.33
D10T (B) 30 738
Total 199 48488.08 75 304

Table 4. Actual Hourly Production for Models of Komatsu

4. Results Model
Dozing
Distance(m)
Actual Hourly Production (Lm3/hr)

15 527
Based the One-Way ANOVA analysis, the best
representative for each scenario of the three assigned D155A-2 30 290
dozing distances for each model was finally found.
In this research, it was decided to present the results in 75 95
practical implications (e.g., see Figure 1). In this context, 15 746
therefore, the author first chose one of the scenarios as the
base scenario. Then, by finding the correlations among the D155A-6 30 351
base and other scenarios, the correction factors which 75 140
adjust the results of the base scenario to other scenarios
were estimated. It is worth mentioning that manufacturers 15 1041
generally use this method to present production of their D275A 30 571
machinery and equipment. In particular, this method for
presenting the results mainly helps better understanding 75 270
the discrepancies between nominal and actual hourly 15 1759
production.
In this context, the scenario in which machines work D375A 30 915
with loose soil in good condition on ground with zero 75 450
slope was chosen as the base scenario for all models.
 American Journal of Civil Engineering and Architecture 86

Figure 4. Chart for actual hourly production for models of Caterpillar

Figure 5. Chart for actual hourly production for models of Komatsu

87 American Journal of Civil Engineering and Architecture

Furthermore, Table 5, Table 6, and Figure 6 present the Table 6. Correction Factors for Materials
correction factors which should be employed to achieve Type of Materials Caterpillar Komatsu
the results for the other scenarios. These correction factors
were also resulted from One-Way ANOVA analysis. Loose soil 1.00 1.00
Depending on the project site operating conditions (i.e., Soil + Rubble stones 0.62 0.69
working condition, material condition, and ground slope),
the appropriate correction factors might be applied to the Blaster rocks 0.52 0.46
charts of actual hourly production data (see Figure 4 and
Figure 5) to estimate the production of interest. In
particular, the formula for production calculation is:
Estimated Actual Hourly Production
= Actual Hourly Production from Charts (1)
× (Π Correction Factors )
For example, a Caterpillar model D7R (i.e., F) assumed
works in a project site with the following conditions:
(1) Dozing distance: 30m.
(2) Working condition: Medium
(3) Materials: Blaster rocks
(4) Ground slope: +10
The data and correction factors for aforementioned
conditions are as follow, respectively:
(1) 174 (m3/hr) (see Table 3)
(2) 0.60 (see Table 5)
(3) 0.52 (see Table 6)
(4) 0.70 (see Figure 6)
Therefore, the machine production is: 174 (m3/hr) × (0.60
× 0.52 × 0.70) = 38 (m3/hr)
This example shows a big discrepancy between Figure 6. Correction factors for ground slope
nominal and actual hourly production which can be
happen in a real construction project due to different
project site conditions. It is worth mentioning that Table 5 5. Discussion
presents a range for correction factors related to the
working conditions. In this context, choosing a number Caterpillar and Komatsu manufacturers regularly
form this range is mainly depends on the machinery provide a new version of their own catalogues for their
manager’s decisions in a project site. For this example, customers. The catalogues includes different charts, tables,
0.60 was chosen from the range from 0.52 to 0.67 and figures to show how their machines and equipment
provided for medium working conditions of the dozer. works in ideal conditions; such assumed ideal conditions
In addition, for most of the correction factors provided could be hardly achieved on a real project.
through Table 5, Table 6, and Figure 6, the amount of This paper investigated the actual production of
Komatsu factors are higher than that of Caterpillar factors crawler-type dozers that work in borrow pits. The main
which leads to higher production in a same project site parameters which affect a machine production were also
condition. Compared to the Komatsu dozers, the investigated. A comparison between estimated results
Caterpillar models worked 6 years on average more; it from this study and manufacturers’ results shows (see
could be considered as the main reasons for lower Figure 7 and Figure 8) a critical discrepancy between the
production. However, Table 6 shows that for blaster rocks, nominal and actual production in the ideal project site
a Caterpillar dozer can work better than a Komatsu dozer. conditions for all models of dozers; an average drop of
This might lead to concluding that Caterpillar dozers 500 Lm3/hr exist for Caterpillar models at dozing distance
typically work with blaster rocks better than Komatsu of 15m. In addition, Figure 7 shows that by increasing the
Dozers. In addition, Figure 6 shows for downhills, a dozing distance, the rate of change for both nominal and
model of Caterpillar could work better than a Komatsu actual productions remains approximately the same for
model. The effect of gravitational force helps movement model B of Caterpillar, while this rate for actual
of a machine in on downward slopes [3]. This effect production of models C, F, G, and H is lower than that of
named grade assistance is resulted from machine weight, nominal production. In other words, compared to the
and it is consistently independent from machine age. dozing distance of 15m, there is not such a big
discrepancy in production for models of C, F, G, and H at
Table 5. Correction Factors for Working Conditions
dozing distance of 75m. Furthermore, Figure 8 shows that
Working Condition Caterpillar Komatsu for all models of Komatsu studied in this research, by
Good 0.83 ~ 0.61 0.83 ~ 0.77 increasing the dozing distance, the rate of change in
Medium 0.67 ~ 0.52 0.67 ~ 0.59 production is lower for actual production than nominal
Weak 0.50 ~ 0.36 0.50 ~ 0.44 production.
 American Journal of Civil Engineering and Architecture 88

Figure 7. Charts for nominal and actual data for models of Caterpillar

Figure 8. Charts for nominal and actual data for models of Komatsu

Table 7 and Table 8, and Figure 9 present the differences materials. Since the estimated limits for range was too
between the estimated correction factors and the factors short, a point estimation is provided. In addition, Table 8
provided by two manufacturers. Table 7 shows that this shows that for the four ranges provided by manufacturers’
study results present a range for correction factors for data, the estimated correction factors are within the
models of Komatsu, while the manufacturer presents a limitation of ranges. Figure 9 presents that for all models
point estimation for factors; considering a range for of Caterpillar and Komatsu dozers work on a downhill
correction factors could be more logical. In addition, for slope, the estimated correction factors for ground slope is
all working conditions of Caterpillar dozers, the range for approximately as same as the factors provided by the
estimated correction factors is within the limits of ranges manufacturers. As discussed, since the gravitational force
provided by the manufacturer (see Table 7). Table 8 helps movement of a machine in on downward slopes, the
presents the correction factors for different types of correction factors remains the same. On the contrary, the
89 American Journal of Civil Engineering and Architecture

engine power of a machine plays the critical role when a In summary, these results shows that for different
machine works on upward slopes. Whereas the dozers project sites conditions, there is a significant discrepancy
studied in this research were not new models, there is a between actual and nominal hourly production (see
significant discrepancy between correction factors provided Figure 7 and Figure 8), while the correction factors
by manufacturers with those provided by this study. In mostly remain unchanged (see Table 7 and Table 8, and
particular, since the average working age of Caterpillar Figure 9).
models is more than that of Komatsu models, this The working condition considered in this study as a
discrepancy is higher for Caterpillar models (see Figure 9). parameter which includes the effect of machinery age,
operators’ skills, and weather condition. The manufactures’
Table 7. Correction Factors for Working Conditions charts are on the basis of ideal condition that for such
Working Obtained Results Manufacturers Data conditions, a machine is typically considered as a new
Condition* Caterpillar Komatsu Caterpillar Komatsu machine, the operator is considered as a fully-skilled
Good 0.83 ~ 0.61 0.83 ~ 0.77 0.83 ~ 0.49 0.83 one, and the weather condition is ideal. However, in
Medium 0.67 ~ 0.52 0.67 ~ 0.59 0.67 ~ 0.40 0.67
real project sites, achieving to these ideal conditions
Weak 0.50 ~ 0.36 0.50 ~ 0.44 0.50 ~ 0.30 0.50
rarely occurs. In the presented case study for this research,
*The definition of working condition has some changes with the there were no new machines (see Table 1), and the
manufacturer definition.
machine operators always were not the best fully skilled
Table 8. Correction Factors for Materials operators (see Figure 2). These two sub parameters
of working condition, therefore, could lead to significant
Type of Obtained Results Manufacturers Data
Materials* Caterpillar Komatsu Caterpillar Komatsu
discrepancies between nominal and actual hourly production
Loose soil 1.00 1.00 1.00 1.00 ~ 0.81 (see Figure 7 and Figure 8).
Soil + Rubble
0.62 0.69 0.67 0.81 ~ 0.67
stones
Blaster rocks 0.52 0.46 0.67 ~ 0.50 0.50 ~ 0.36

Figure 10. Linear regression model for models of Caterpillar

Figure 9. Correction factors for ground slope Figure 11. Linear regression model for models of Komatsu
 American Journal of Civil Engineering and Architecture 90

Table 9. Results of Linear Regression Models

Coefficient Estimated Coefficient Standard Error T-Value P-Value R

-16
(Intercept) 9.74345 0.41005 22.8 < 2 × 10
Caterpillar 0.8413
Age 4.61067 0.03772 123.0 < 2 × 10-16
(Intercept) 12.21153 0.34291 37.05 < 2 × 10-16
Komatsu 0.7791
Age 5.90495 0.07833 73.37 < 2 × 10-16

In addition, a statistical comparison between the land clearing in a project site. It could be considered
machinery age and discrepancies between nominal and as the main limitation for this work.
actual production can illustrate how the age of a machine Future research is recommended to study crawler
might affect its production. This comparison could type dozer in different working sites besides dozing
address the third sub-objective of this research study. In in order to understand how the machine acts for
particular, therefore, simple linear regression models are such working conditions.
developed to estimate a mathematical relationship 2. There are uncertainties in collected data. Such
between machinery age and discrepancy for different uncertainties specifically can happen during data
models of Caterpillar and Komatsu. In fact, these collection steps and therefore could affect the
regression models help understand how a machine age results.
affect is production discrepancy. In this context, for the It is recommended that future research employs
regression models, the machine age and percentage of various machine learning methods such as neural
discrepancy are considered as explanatory and response network, and statistical methods such as
variables, respectively. Figure 10 and Figure 11 presents multivariate analysis in order to find and void such
the developed regression models for dozers of Caterpillar uncertainties.
and Komatsu, respectively. It is worth mentioning that all 3. Operators were assigned randomly to different
data for dozers points out on Figure 10 and Figure 11 to models of dozer.
show discrepancy range found for all model with the same In such work, future research is recommended to
working years. For example, Figure 10 shows that this consider a plan in order to assign an operator to a
range is approximately between 38% and 67% for models machine. Such plans could provide an opportunity
of Caterpillar with 10 working years. to understand how operators’ skills might affect
Table 9 shows the results of linear regression models. machine production.
The results show that two variables are positively 4. Climate conditions and annual rainfall were
associated for each regression model, and the amount of assumed to be equal for the five borrow pits.
p-values and Pearson correlation coefficients (i.e., R) Therefore, the effects of these parameters were
indicates the strength of this associated relationship, neglected for this study.
specifically for Caterpillar dozers. Therefore, it can be Future research might study the effects of climates
concluded that the age of a machine could help predict its conditions and annual rainfall as part of the working
production’s discrepancy and therefore, to predict its condition in order to achieve more realistic results
actual hourly production. and conclusions.
In addition to the aforementioned discussion, compared It is worth mentioning that in such research work, the
to the Rashidi et al. [23] and Ok and Sinha [26] research data collection step is the main step which leads to
which studied actual productions of dozer, this research achieving reliable results and conclusions. This step
provides real charts for productions and indicates how the significantly depends on the aim of a study. In fact, the
correction factors are affected in various project sites. In aim of study should to be clearly defined before starting to
addition, to best knowledge of the author, it is the first collect required data. In particular, the aim determines
research work that presents all the results in the format of not only the required time for data collection step but
manufacturers’ publication. Such results provide a critical also the methodologies needed to be employed to
unique opportunity to visually understand the differences collect required data. In most cases, the data collection
between nominal and actual hourly production when a step is extremely time consuming. Future work is
machine works in a project site. therefore recommended to employ remote sensing
data methodologies to collect required data. These
methodologies are expected to save required time and cost
6. Limitations and Recommendation for needed for the data collection step.
Future Research
The results and discussion on the results could address 7. Conclusion
the objectives and sub-objectives of this study. However,
this research is also subjects to some limitations, as follow: This paper employed the One-Way ANOVA analysis
1. This study’s findings and results are based on a case and linear regression model to understand and estimate the
study for dozers working at five borrow pits. Such discrepancies between actual and nominal production for
results might not be extended to other types of crawler-type dozer. In order to collect the required data for
projects in where crawler-type dozer are utilized for this research, 21 different models of Caterpillar with
a different job; the machine also can be used for average working year of 10.1 and 18 different models of
91 American Journal of Civil Engineering and Architecture

Komatsu with average working year of 4.1 were [8] H.N. Rafsanjani, An assessment of nominal and actual hourly
individually considered in five different borrow pits for an production of crawler-type front shovel in construction project,
JCE-IEB. 40 (2012) 115-124.
eight-month period of time. For each machine, three [9] H.N. Rafsanjani, Comparison between Nominal and Actual
different dozing distances were considered, and actual Hourly Production of Crawler-Type Dozer: A Case Study, in:
hourly production was estimated based on the ANOVA Comput. Civ. Eng. 2015, American Society of Civil Engineers,
analysis at each dozing distance. The results show n.d.: pp. 17-24.
significant reductions (i.e., discrepancies) in machine [10] H.N. Rafsanjani, Y. Gholipour, H.H. Ranjbar, An Assessment of
Nominal and Actual Hourly Production of the Construction
hourly production rates; these discrepancies should to be Equipment Based on Several Earth-Fill Dam Projects in Iran,
considered in planning of machinery and equipment. In Open Civ. Eng. J. 3 (2009) 74-82.
particular, the results from linear regression models [11] H.N. Rafsanjani, S. Shahrokhabadi, A. Hadjahmadi, The Use of
indicate that a linear relationship could exist between Linear Regression to Estimate the Actual Hourly Production of a
Wheel-Type Loader in Construction Projects, in: ICSDEC 2012,
working age of a machine and its production discrepancies. American Society of Civil Engineers, n.d.: pp. 727-731.
Such relationship could help estimate the actual [12] S.W. Nunnally, Managing construction equipment, 2nd ed.,
production of a machine based on its working age. In Prentice-Hal, 2000.
addition, this paper also studied three parameters that [13] Modeling the Dynamics of Heavy Equipment Management
affect a machine production in a project site. The Practices and Downtime in Large Highway Contractors, J. Constr.
Eng. Manag. 135 (2009) 939-947.
correction factors for each individual parameter were [14] G. Lucko, A Statistical Analysis and Model of the Residual Value
found. These correction factors approximately remained in of Different Types of Heavy Construction Equipment, 2003.
the range provided by manufacturers. [15] A. Tsimberdonis, E. Murphree, Equipment Management through
The author believe that the results of this study could Operational Failure Costs, J. Constr. Eng. Manag. 120 (1994)
provide appropriate prospect for machinery managers in 522-535.
[16] C. Schexnayder, HEAVY CONSTRUCTION EQUIPMENT
construction sites; the results present promising options REPLACEMENT ECONOMICS, 1980.
for future research into machinery and equipment [Link]
production. [17] A. Tavakoli, J. Masehi, C. Collyard, FLEET: Equipment
Management System, J. Manag. Eng. 6 (1990) 211-220.
[18] H. Tajeen, Z. Zhu, Image dataset development for measuring
Disclaimer construction equipment recognition performance, Autom. Constr.
48 (2014) 1-10.
[19] N. Pradhananga, J. Teizer, Automatic spatio-temporal analysis of
The opinions, findings, and conclusions expressed in construction site equipment operations using GPS data, Autom.
this paper are those of the authors and do not necessarily Constr. 29 (2013) 107-122.
prove or disprove the view of Caterpillar and Komatsu. [20] H. Yip, H. Fan, Y. Chiang, Predicting the maintenance cost of
construction equipment: Comparison between general regression
neural network and Box-Jenkins time series models, Autom.
Constr. 38 (2014) 30-38.
List of Abbreviations and Symbols [21] A. Hajji, The use of construction equipment productivity rate
model for estimating fuel use and carbon dioxide (CO2)
Lm3 Loose cubic meters emissionsCase study: bulldozer, excavator and dump truck, Int. J.
Lm3/h Loose cubic meters per hour Sustain. Eng. 8 (2015) 111-121.
[22] Comparison of Two Models for Estimating Equipment
Lm3/hr Loose cubic meters per hour Productivity for a Sustainability Quantification Tool, in: ICSDEC
Lyd3/h Loose cubic yards per hour 2012, American Society of Civil Engineers, n.d.: pp. 626-633.
LCY/hr Loose cubic yards per hour [23] A. Rashidi, H.R. Nejad, M. Maghiar, Productivity estimation of
ft Feet bulldozers using generalized linear mixed models, KSCE J. Civ.
m Meter Eng. 18 (2014) 1580-1589.
[24] S. Han, T. Hong, S. Lee, Production prediction of conventional
hp Engine horse power and global positioning system-based earthmoving systems using
simulation and multiple regression analysis, Can. J. Civ. Eng. 35
(2008) 574-587.
References [25] K. Schabowicz, B. Hola, Mathematical-neural model for assessing
productivity of earthmoving machiner, J. Civ. Eng. Manag. 13
[1] D. Gransberg, C. Popescu, R. Ryan, Construction Equipment (n.d.) 47-54.
Management for Engineers, Estimators, and Owners, Taylor & [26] S.C. Ok, S.K. Sinha, Construction equipment productivity
Francis, 2006. estimation using artificial neural network model, Constr. Manag.
[2] K. Samee, J. Pongpeng, Structural equation model for construction Econ. 24 (2006) 1029-1044.
equipment management affecting project and corporate [27] C.D. Edmonds, B. Tsay, W. Lin, Analyzing machine efficiency,
performance, KSCE J. Civ. Eng. (2015) 1-15. Natl. Public Account. 39 (1994) 28-44.
[3] R. Peurifoy, C. Schexnayder, A. Shapira, R. Schmitt, Construction [28] J. Zou, HSV Color-space Digital Image Processing for the
Planning, Equipment, and Methods, 8th ed., McGraw-Hill, 2010. Analysis of Construction Equipment Utilization and for the
[4] Building Intelligent Applications for Construction Equipment Maintenance of Digital Cities Image Inventory, University of
Management, in: Comput. Civ. Eng. 2007, American Society of Alberta (Canada), 2006.
Civil Engineers, n.d.: pp. 192-199. [29] G.P. Hammer, J.-B. du Prel, M. Blettner, Avoiding Bias in
[5] O. Tatari, M. Skibniewski, Integrated agent‐based construction Observational Studies, Dtsch. Ärztebl. Int. 106 (2009) 664-668.
equipment management: Conceptual design, J. Civ. Eng. Manag. [30] C.J. Pannucci, E.G. Wilkins, Identifying and Avoiding Bias in
12 (2006) 231-236. Research, Plast. Reconstr. Surg. 126 (2010) 619-625.
[6] G. Heravi, H.N. Rafsanjani, Critical Safety Factors in [31] Innovation in Construction Equipment and Its Flow into the
Construction Projects, Adv. Mater. Res. 255-260 (2011) Construction Industry: Journal of Construction Engineering and
3921-3927. Management: (ASCE), (n.d.).
[7] H.N. Rafsanjani, A Computer Model for Selecting Equipment in [32] Caterpillar Inc., Caterpillar performance handbook, 42nd edition,
Earth-Fill Dam Projects, J. Civ. Eng. Archit. 5 (2011) 465-470. (2012).
 American Journal of Civil Engineering and Architecture 92

[33] Komatsu Inc., Komatsu Specifications & Application Handbook, [41] M Moosavi, H Nabizadeh, MH Mehranpour, A Nazem, A.N.
30th edition, (2009). Shirazi. An investigation of surface roughness measurement in
[34] P. Lewis, A. Hajji, Evaluation of Construction Equipment Fleets rock joints with a 3D scanning device, (2013).
through Fuel Use and Emissions Inventories, in: Sustain. Transp. [42] H.N. Rafsanjani, Y Gholipoor, H. Nabizadeh, ESCMODEL: A
Syst., American Society of Civil Engineers, n.d.: pp. 138-145. Model for Optimum Machinery Selection in Earth-Fill Dam
[35] Evaluating the Performance of Construction Equipment Operators Projects, World Congress on Engineering and Computer Science
in Egypt, J. Constr. Eng. Manag. 122 (1996) 109-114. 2010 (WCECS 2010).
[36] S. Alwi, Factors influencing construction productivity in the [43] S. Amelian, M. Manian, S.M. Abtahi, A. Goli, Moisture
Indonsesian context, in: Fac. Built Environ. Eng., Fukuoka, Japan, sensitivity and mechanical performance assessment of warm mix
2003. [Link] (accessed February 26, 2016). asphalt containing by-product steel slag, Journal of Cleaner
[37] S. Dowdy, S. Wearden, D. Chilko, Statistics for research, 3rd ed., Production, 176 (2018) 329-337.
Wiley, 2004. [44] S. Amelian, S.M. Abtahi, S.M. Hejazi, Moisture susceptibility
[38] H. Nabizadeh, H.F. Haghshenas, Y. Kim, F.T.S. Aragao, Effects evaluation of asphalt mixes based on image analysis, Construction
of rejuvenators on high-RAP mixtures based on laboratory tests of and Building Materials, 63 (2014) 294-302.
asphalt concrete (AC) mixtures and fine aggregate matrix (FAM) [45] S. Amelian, Y. Kim, Evaluation of Tack Coating Practices for
mixtures, Construction and Building Materials 152 (2017) 65-73. Asphalt Overlays in Nebraska. Nebraska Department of
[39] H. Nabizadeh, Viscoelastic, Fatigue Damage, and Permanent Transportation. SPR-P1(16) M039 (2017).
Deformation Characterization of High Rap Bituminous Mixtures [46] H.N. Rafsanjani, M. Chehreghani, M. Nourbakhsh, Recycling of
Using Fine Aggregate Matrix (FAM), (2015). crushed clay brick as aggregate in concrete, Applied Mechanics
[40] H. Haghshenas, H. Nabizadeh, Y. Kim, K. Santosh, Research on and Materials 94 (2011) 1087-1090.
high-rap asphalt mixtures with rejuvenators and WMA additives [47] M. Khezripour, H.N. Rafsanjani, Bucket with Flexible Width for
(2016). Hoe, 2nd International Conference on Applied Research in Civil
Engineering, Architecture, and Urban Planning (2015).