The use of construction equipment productivity rate model for estimating fuel use and carbon dioxide (CO₂) emissionsCase study: bulldozer, excavator and dump truck

Abstract

Carbon dioxide (CO₂) emission has drawn a great attention in recent years all over the world, and it plays a very important role in the process of global warming. The off-road equipment, which includes construction equipment, accounted for 7.5% emission of CO₂. The objective of this article is to propose a groundwork for a methodology that can be used to estimate the total fuel use and CO₂ emissions from construction equipment activities based on its productivity rate. Currently, there is not a methodology or technique for estimating productivity, fuel use and emissions at once. The multiple linear regression analysis has successfully modelled the productivity rate and good to be used as a benchmark for estimating fuel use and CO₂ emissions from various types of construction equipment performing earthwork activities. The proposed methodology will help the contractor to estimate the total expected pollutant emissions for the project, which would be valuable information for a preliminary environmental assessment of the project. By using construction plans and specifications, the methodology and tool presented in this research can be used to estimate cost, fuel use and emissions from commercial, residential, industrial or heavy highway. By the proposed approach, it is possible to develop new fuel use and emissions inventories for construction industry in general.

Keywords::

• construction equipment
• productivity
• fuel use
• emissions

1. Introduction

According to the US Environmental Protection Agency (EPA) report 2007, construction industry has produced approximately 1.7% of total greenhouse gas (GHG) emissions. Equivalent to 6% of total US industrial-related GHG emissions, this quantity places construction as one of the top emitting sectors (EPA 2007b). Although construction practices typically do not produce large quantities of GHG compared with the operations of many other sectors, the sheer number of construction projects results in significant aggregate emissions for the sector. The Sector Strategies Division of EPA recently released the report which presents historical emissions estimates for 14 industrial sectors that produces 84% of industrial GHG emissions in the USA. According to this report, the construction sector produced 6% of total US industrial GHG emissions in 2002. CO₂ emission has drawn a great awareness in recent years all over the world, and it plays a very important role in the global warming. According to EPA, the on-road vehicles released 92.5% of the total CO₂ emission, while the off-road equipment, which includes construction equipment, accounted for the remaining 7.5% emission of CO₂. Among this 7.5%, the construction equipment accounted for 40%, which is very severe (EPA 2007a).

Currently, most construction estimators are able to estimate the costs related to construction equipment and its activities that they perform. However, most construction estimators seldom concerned themselves with the fuel consumption and emissions from the equipment that they used (Lewis and Hajji 2012b). Since some new environmental regulations appear on the practices of other industries, construction professionals can no longer disregard the issues of energy use and emissions aspects of their works. They need a methodology that can be used with their current expertise to help them quantify the total fuel use and emissions of the construction equipment activities. The objective of this article is to propose a groundwork for a methodology that can be used to estimate the total fuel use and CO₂ emissions from construction equipment activities. This methodology can be used along with other common estimating approaches to gain an overall understanding of the fuel and emissions footprint for construction activities and ultimately an entire project. This proposed methodology can also be developed into a stand-alone model, or into a module that can be used collaboratively with other existing estimators.

2. Related studies

Productivity (P) is simply defined as the ratio of the quantity (Q) of work completed to the duration (D) of time that it took to complete the work (Lewis and Hajji 2012a). For construction equipment activity, if a bulldozer hauls 1000 bank cubic yards (bcy) of earth in 10 hours (h), the productivity rate is 100 bcy/h. This ratio also reveals that the duration of a construction equipment activity is inversely proportional to productivity – as productivity decreases, the activity duration increases. When the duration increases, it will result in higher fuel consumption and higher emissions. Thus, it is necessary to predict the productivity rates of the construction equipment in order to estimate its emissions. To predict productivity rates of the construction equipment, this article uses multiple linear regression (MLR) analysis. MLR analysis is a powerful tool that provides a simple method for establishing a functional relationship between predictor and response variables. It has been used frequently to conduct construction-related studies (Akinsola 1997; Akintoye and Skitmore 1994; Edwards, Holt, and Harris 2000). Furthermore, the use of deterministic MLR analysis is well established in construction engineering research, particularly in predicting productivity (David and Gary 1993; Dunlop and Smith 2003; Han and Halpin 2005; Ok and Sinha 2006; Smith 1999). In this article, MLR is used to determine the statistical relationship between productivity rate as response variables, and construction equipment working attributes, such as engine horsepower (hp), distance and soil types as explanatory variables.

Many models and methods have been developed to quantify the CO₂ emission from construction equipment. The state government of California has developed California Air Resources Board (CARB) OFFROAD2007 Model. OFFROAD2007 emission model consists of three main modules: population, activity and emissions factor (CARB 2007). These three modules are combined to yield the annual equipment emissions by county, air basin or state. Emission inventories obtained from this model are produced with respect to fuel type (e.g. gasoline, diesel, compressed natural gas), engine type (e.g. two-stroke vs. four-stroke), equipment category and horsepower group. The emissions (tons/day) inventory is calculated by taking the product of maximum rated average horsepower, load factor, annual activity, emission factor and engine population. The CARB also developed an EMission FACtors (EMFAC) model to calculate emission rates from all motor vehicles, such as passenger cars to heavy-duty trucks, operating on highways, freeways and local roads in California (CARB 2003). An emission inventory can be summarized as the ‘product of an emission rate (e.g. grams per pollutant emitted over a mile) and vehicle activity (e.g. miles driven per day)’. The model calculates emission factors and emission inventories for the following primary pollutants: hydrocarbons (HCs), carbon monoxide, nitrogen oxides (NO_x), carbon dioxide (CO₂), particulate matter (PM), fuel consumption, oxides of sulphur (SO_x) and lead. Li and Lei (2010) proposed the implementation of discrete-event simulation (DES) in estimating and analysing CO₂ emission for earthwork construction activities. This method estimates emissions from individual project-specific construction at micro-level and includes three modules: module 1, to calculate the emission factors of CO₂ from the EPA model; module 2, to use DES to calculate the duration of earthmoving process and module 3, to calculate the total amount of CO₂ emission by using duration and emission factors. Ammouri, Srour, and Hamade (2011) proposed the Carbon Footprint Calculation for Construction Projects. This calculator is capable of estimating the total carbon footprint of a construction project taking into consideration the size, landscape and materials. The calculator can be localized to any market based on the database of materials and equipment. It takes project inputs from nine different categories: general project information, site works, concrete works, metal works, wood works, windows and doors, finishes, mechanical works and electrical works. The total carbon from an observed work is embodied in the raw material, transportation carbon footprint and workforce carbon footprint.

Although there are some existing methodologies and techniques that estimate emissions from construction equipment, these tools typically do not address its important relationship with productivity rate and total fuel use. Conversely, most construction productivity estimating models accurately address the productivity aspects of construction equipment, but do not address the issues of energy use and emissions. There currently is not a methodology or technique for estimating productivity, fuel use and emissions at once.

3. Methodology

The proposed estimating tool is developed by combining the productivity rate model from a reliable construction estimating data sources and the calculation algorithm employed by the EPA's NONROAD model (Figure 1). In order to develop productivity models, simple earthwork activities involving bulldozer, excavator and dump truck were selected. The data for these activities were collected from RSMeans Heavy Construction Data 2011. This is parts of RSMeans estimating references, which is widely accepted and used by most construction professionals. These data provide technical information required for preparing quantity takeoffs and complete construction estimates for major construction projects (RSMeans 2011). It helps design and construction professionals to compare the estimates of design alternatives, perform cost analysis and value engineering and review estimates quotes and change orders prepared by others.

Figure 1 Fuel use and CO₂ emission estimation methodology.

The data also include information in sizing, productivity, equipment requirements, design standards and engineering factors – all organized according to the latest 2004 Construction Specification Institute Master Format classification system, which has 48 divisions ranging from General Requirements to Electrical Power Generation. The selected activities are provided in RSMeans data on Division 31: ‘Earthwork’. This division has sections representing major earthwork activities: clearing and grubbing, tree and shrub removal and trimming, stripping and stockpiling, grading, excavation and fill, erosion and sedimentation controls, soil treatment and stabilization, shoring, underpinning, dams and tunnel construction. Bulldozer activities are provided in section 31.23.16.46, excavator in section 31.23.16.13 and dump truck in section 31.23.23.20. The activity input data for these three types of equipment are shown in Table 1.

Table 1 Activity input of selected construction equipment.

Equipment	Number of data	Activity input	Unit/type/range
Bulldozer	72	Engine size	80–700 horsepower (hp)
		Distance	50–300 feet (ft)
		Type of soil	Sand–gravel; sandy-clay loam; common earth; clay
Excavator	394	Bucket size	0.5–3.5 cubic yard (cy)
		Trench depth	1–24 ft
		Soil type	Sand–gravel; sandy-clay loam; common earth; clay
		Excavator type	Regular; with truck mounted; with trench box
Dump truck	240	Loading capacity	22–60 cy
		Distance	2000 ft to 4 miles
		Speed	5–25 miles per hour (mph)
		Wait–load–unload time	15–25 minutes (min)

Based on the data from RSMeans, the productivity rate models were developed by using MLR. The MLR model is written as follows:

(1)

where y_i is the response that corresponds to the levels of explanatory variables x₁, x₂, …, x_p at the ith observation and β₀, β₁, β₂, …, β_p are the coefficients in the linear relationship. For a single factor (p = 1), β₀ is the intercept and β₁ is the slope of the straight line defined; ϵ₁, ϵ₂, …, ϵ_n are errors that create scatter around the linear relationship at each of the i = 1 to n observations. The regression model assumes that these errors are mutually independent, normally distributed and with a zero mean and variance σ². To make estimates of the coefficients in the regression model, the method of least squares is used.

In this article, MLR is used to determine the statistical relationship between a response (productivity rate) and the explanatory variables in construction equipment activities. The response variable is expressed in terms of loose cubic yard per hour (lcy/h) as productivity rate.

The productivity models in this article were built by using stepwise regression selection method and validated by using data splitting method. In data splitting method, the original data-sets are split into a model-building set and a validation set (Kutner, Nachtsheim, and Neter 2005). If the number of data is within 6–10 times the number of predictor variables, it is enough for making an equal data split. If the entire data are not large enough under these circumstances, the validation set will need to be smaller than the model building data-set. The model-building set is used to develop the model. The validation set is used to evaluate the reasonableness and predictive ability of the selected model. A means of measuring the actual predictive ability is to use the model to predict each case in the validation data-set and then to calculate the mean of the squared prediction errors, denoted by MSPR or mean squared prediction error. The MSPR is calculated as:

(2)

where Y_i is the value of the response variable in the ith validation case; is the predicted value for the ith validation case based on the model from using the model building data-set and n is the number of cases in the validation data-set.

If the MSPR is fairly close to the mean squared errors (MSE) based on the regression fit to the model-building data-set, then MSE for the selected regression model is not seriously biased and gives an appropriate indication of the predictive ability of the model.

The other information such as rated engine horsepower (hp) and engine model year will be used to estimate the fuel use and CO₂ emission rates. It needs the brake-specific fuel consumption (BSFC) factor (Table 2). This factor is an approximation of amount of fuel consumed by a particular type of equipment during a unit of use. The BSFC for this estimation was based on the calculation algorithm used by the EPA's NONROAD model (EPA 2010). BSFC is reported in pounds per horsepower-hour (lbs/hp-h). The EPA assumes a constant BSFC for each engine tier within a specified engine horsepower range, without regard for the cycle being performed. Assuming a constant fuel consumption rate for all engine tiers may make the calculation of emission rates easier, but may not be a good representative of actual working conditions (EPA 2010). This factor used by NONROAD is based on engine dynamometer test data and adjusted accordingly to account for in-use operation that differs from the typical test conditions.

Table 2 Steady-state BSFC for nonroad diesel engines.

Engine horsepower (hp)	Tier level	BSFC (lb/hp-h)
>0–100	0–4	0.408
>100–750	0–4	0.367
>750	0–4	0.367

Nonroad engines are mainly tested during steady-state condition. However, the steady-state condition typically used for emission testing is not necessarily representative of the operation of engines in many nonroad applications. Some of the disparities can happen due to engine load, engine speed or transient demands. Transient adjustment factors (TAFs) are applied to Tier 0, Tier 1, Tier 2 and Tier 3 BSFC factors. Transient control is expected to be an integral part of all Tier 4 engine design considerations (EPA 2010).

The tier level is the emission standard that corresponds to the equipment model years and technology types (represented by engine horsepower) (EPA 2001). EPA has regulated emissions from nonroad engines used in most construction and port cargo handling equipment since 1999. As shown in Table 3, these regulations continue to be phased in under a four-tier system, with emission standards based on engine horsepower (hp) and equipment model year.

Table 3 General guide to EPA tier level for off-road diesel engines (EPA 2010).

Engine power	Tier	Years	Engine power	Tier	Years
hp < 11	1	2000–2004	100 ≤ hp < 175	1	1997–2002
	2	2005–2007		2	2003–2006
	4	2008+		3	2007–2011
				4	2012+
11 ≤ hp < 25	1	2000–2004	175 ≤ hp < 300	1	1996–2002
	2	2005–2007		2	2003–2005
	4	2008+		3	2006–2010
				4	2011+
25 ≤ hp < 50	1	1999–2003	300 ≤ hp < 600	1	1996–2000
	2	2004–2007		2	2001–2005
	4	2008+		3	2006–2010
				4	2011+
50 ≤ hp < 75	1	1998–2003	600 ≤ hp < 750	1	1996–2001
	2	2004–2007		2	2002–2005
	3	2008+		3	2006–2010
				4	2011+
75 ≤ hp < 100	1	1998–2003	hp ≥ 750	1	2000–2005
	2	2004–2007		2	2006–2010
	3	2008		4	2011+
	4	2008+

Tiers 1–3 standards are largely being met by enhanced engine design and manufacturing improvements; they require little or no exhaust after-treatment, and do not address fuels. The Tier 4 standards require dramatic reduction in NO_x and PM emissions. The off-road NO_x and PM standards under Tier 4 are approximately 10 times lower than the Tier 3 standards for most engines. They will be phased in between 2008 and 2015. To comply with this rule, engine manufacturers will need to produce engines with advanced emissions control technologies similar to those that will be used for on-road trucks. This rule also requires fuel producers to reduce the sulphur content of diesel fuel used in nonroad engines. Reducing the level of sulphur in diesel fuel is necessary to prevent damage to emission control systems.

The calculation of fuel use rate using BSFC is as follows:

(3)

The calculation of CO₂ emissions rate is as follows:

(4)

where BSFC_SS is the converted steady-state fuel consumption factor (gal/hp-h), 453.6 is the conversion from pounds to gram, HC is the in-use adjusted HC emissions (g/hp-h), 0.87 is the carbon mass fraction of diesel and 44/12 is the ration of CO₂ mass to carbon mass.

4. Results and discussion

The MLR approach was successful in providing model for predicting the productivity rate of the bulldozer, excavator and dump truck activities. Based on the value of R², the MLR equation for predicting productivity rates can adequately explain the variability of the data. With α = 0.05, all parameters (excluding the intercept) in the models had p-values < 0.0001 and were statistically significant to be included in the models. For bulldozer, in the result of original regression function, it was found that the plot of residuals against the predicted values showed evidence of unequal variance. The unequal error variances and non-normality of the error terms frequently appear together. To remedy the non-normality in the data, a Box–Cox analysis has been conducted and the result recommended that the best lambda (λ) for transforming the response variables (Y) is 0.2. Thus, for bulldozer the form of transformed regression function for productivity can be written as:

(5)

(6)

The overall results of MLR analysis and validation results of productivity models for bulldozer, excavator and dump truck are shown in Table 4.

Table 4 Validation results of productivity models.

Equipment	n	R^2	MSPR	MSE
Bulldozer	72	0.9534	0.0023	0.0061
Excavator	394	0.9195	214.5	225.52
Dump truck	240	0.9432	49.28	46.79

Because all data-sets were large, the model was also validated by using cross-validation procedure or data splitting. As shown in Table 4, the MSPR from the productivity model are fairly close to the MSE based on the regression fit to the model-building data-set. The MSE for the regression models of productivity rate used here are not seriously biased and give an appropriate indication of the predictive ability of the model. The MLR analysis results of productivity models for bulldozer, excavator and dump truck are shown in Table 5.

Table 5 MLR results for productivity models of three equipments.

Variable	Parameter estimates	p Value
Bulldozer
Intercept	1.87659	< 0.0001
Engine size (hp)	0.00350	< 0.0001
Distance (ft)	− 0.00240	< 0.0001
Soil type 1 – sand–gravel	0.23656	< 0.0001
Soil type 2 – sandy clay–loam	0.21667	< 0.0001
Soil type 3 – common earth	0.16644	< 0.0001
Excavator
Intercept	− 3.946	0.3656
Soil type 1 – common earth	8.465	0.0193
Soil type 2 – sandy clay–loam	14.907	< 0.0001
Soil type 3 – sand gravel	16.412	< 0.0001
Depth	− 2.069	< 0.0001
Bucket size	55.131	< 0.0001
Excavator type 1 – regular	3.317	0.1676
Excavator type 2 – with trench box	4.166	0.4743
Dump truck
Intercept	58.799	< 0.0001
Loading capacity	2.079	< 0.0001
Speed	1.625	< 0.0001
Cycle distance	− 12.056	< 0.0001
Cycle time	− 2.789	< 0.0001

Concerning the types of excavator, since the p-values of type of excavator are bigger than α = 0.05, it can be concluded that different types of excavation do not significantly lead to different rates of productivity. However, since the type of excavator is categorical variable, it can be used to represent the categories of a qualitative explanatory variable in the regression model.

The overall calculation formula for obtaining the total fuel use and CO₂ emissions estimates are formed by combining the productivity rate models with calculation algorithm of the NONROAD model. In order to estimate the fuel use and CO₂ emissions from a certain quantity of soil performed by a bulldozer, or an excavator or a dump truck, the total duration of activity is needed. The total duration in hours (h) can be obtained by dividing the total soil quantity with the productivity rate in loose-cubic yard per hour (lcy/h). Once the total duration is obtained and engine horsepower (hp) is known, the total emission in grams (g) and total fuel use (gal) can be calculated by multiplying the CO₂ emission rates (g/hp-h) or the BSFC (gal/hp-h) from NONROAD with horsepower (hp) and duration (h).

(7)

(8)

(9)

By using the basic formula in (7), the overall calculation formulas for obtaining the total fuel use of each type of equipment are shown in Table 6. To demonstrate the total fuel use estimate for bulldozer, a case of 150 hp bulldozer, model year 2003, that hauls 1000 cy common earth in 300 ft is presented. The engine size, distance and type of soil are used to generate productivity rate, while engine size and model year are used to obtain engine tier level, BSFC and TAF. The results showed that the productivity rate is 20.02 cy/h and it needs 49.95 h to complete the work. This bulldozer is in Tier 2 level and has BSFC of 0.367 lbs/hp-h and TAF of 1.01. The total fuel consumed to complete 1000 cy common earth is 333.37 gal. The total fuel use estimate for the excavator is presented in the following case: a regular excavator with 3 cy bucket size has to dig a 100 ft long – 10 ft wide – 12 ft deep trench in a sand–gravel type of soil. The engine has 400 hp and model year 2003. The type of soil, type of excavator, soil quantity and bucket size are utilized to calculate the productivity rate, while the engine size and model year will determine the engine tier, BSFC and TAF.

Table 6 Productivity-based total fuel use models.

Equipment	Total fuel use model
Bulldozer	fs (soil type factor): sand–gravel = 0.236; sandy-clay loam = 0.217; common earth = 0.166; clay = 0	(10)
Excavator	fs (soil type factor): common earth = 8.465; sandy-clay loam = 14.907; sand–gravel = 16.412; hard clay = 0ft (scraper type factor): excavator = 3.317; truck-mounted = 4.165; trench-box = 0	(11)
Dump truck		(12)

Note: Q, quantity of soil dozed/moved/excavated (cy); hp, engine horsepower; BSFC, brake specific fuel consumption (gal/hp-h); TAF, transient adjustment factor (unitless); D, distance (ft) – miles for truck; d, depth (ft); B, bucket capacity (cy); C, loading capacity (cy); S, speed (mph); t, cycle time (min).

The model showed the results that the productivity rate is 156.34 cy/h or equals to 2.84 h of total duration to complete the trench. The excavator is categorized in Tier 2 engine and has the BSFC of 0.367 lbs/hp-h and TAF of 1.01. To complete digging the trench, the excavator needs 50.53 gal of fuel. The sample case for truck is presented as a 30 cy off-road truck has to haul 1000 cy of soil in 1 mile. The truck has 535 hp – model year 2003 with 10 mph of average hauling speed. The truck needs 15 min on average to load and dump the soil. The truck capacity, average speed, hauling distance and load–dump time are utilized to determine the productivity rate, while engine size and model year will define the engine tier level, BSFC and TAF. The fuel use model showed the results that the truck has productivity of 83.54 cy/h or needs 11.97 h to haul all soil quantity. Since the engine is categorized as Tier 2, the truck has BSFC of 0.367 lbs/hp-h and TAF of 1.01. To complete the task, the truck needs 284.85 gal of fuel in total.

By using the basic formula in (8) and (9), the overall calculation formulas for obtaining the total CO₂ emissions of each type of equipment are shown in Table 7. To demonstrate the CO₂ emission model for bulldozer, a case is presented where a 150 hp of bulldozer model year 2003 has to haul 1000 cy of common earth in the distance of 300 ft. The result showed that the estimated productivity of this bulldozer is 20.02 cy/h and can approximately complete the job in 49.94 h. Based on the engine size and model year, this bulldozer is categorized in Tier 2 engine level. This engine also has 0.367 lbs/hp-h of BSFC. When completing the job, this bulldozer was estimated to emit 3.4 tons of CO₂. To demonstrate the model for estimating total CO₂ emissions from excavator, a case of a 400 hp – model year 2003 excavator is presented. The excavator has to dig a 100 ft long – 10 ft wide – 12 ft deep trench in sand–gravel soil with its 3 cy bucket.

Table 7 Productivity-based total CO₂ emission models.

Equipment	Total CO2 emission model
Bulldozer	fs (soil type factor): sand–gravel = 0.236; sandy-clay loam = 0.217; common earth = 0.166; clay = 0	(13)
Excavator	fs (soil type factor): common earth = 8.465; sandy-clay loam = 14.907; sand–gravel = 16.412; hard clay = 0ft (scraper type factor): excavator = 3.317; truck-mounted = 4.165; trench-box = 0	(14)
Dump truck		(15)

Note: E, total emissions (g); Q, quantity of soil dozed/moved/excavated (cy); hp, engine horsepower; EFss, steady-state emission factor (g/hp-h); BSFC, brake specific fuel consumption (gal/hp-h); TAF, transient adjustment factor (unitless); DF, deterioration factor (unitless); S_PM, adjustment to PM emission factor for fuel sulphur content (g/hp-h); HC, in-use adjusted hydrocarbon emissions (g/hp-h); 453.6, conversion factor from pounds to grams; 0.87, carbon mass fraction of diesel; 44/12, ratio of CO₂ mass to carbon mass; D, distance (ft or miles); d, trench depth (ft); B, bucket capacity (cy); C, loading capacity (cy); S, average hauling speed (mph); t, load–dump time (min).

The results showed that the excavator has the productivity rate of 156.34 cy/h or needs 2.84 h to complete digging the trench. Based on the horsepower and model year, the engine is categorized as Tier 2 engine and has 0.367 lbs/hp-h of BSFC. When finishing the trench, this excavator released 0.5 tons of CO₂. To illustrate CO₂ emission models for truck, a simple case is presented as follows: a 535 hp – model year 2003 truck is used to haul 1000 cy of soil in 1 mile of distance. The truck has 30 cy of loading capacity with average hauling speed of 10 mph. For loading and dumping the soil, the truck needs 15 min in average. The estimated productivity rate of this truck is 83.54 cy/h, and for hauling 1000 cy of soil, the truck needs 11.97 h. Based on the horsepower and model year, the truck is categorized as Tier 2 engine with 0.367 lbs/hp-h of BSFC. When the job is completed, the truck released 2.9 tons of CO₂.

5. Sensitivity analysis

Sensitivity analysis is conducted to analyse the effect of changes in explanatory variables against the output: total fuel use and CO₂ emissions. The analyses are useful to understand the environmental impact of a certain earthwork activity performed by construction equipment in different set of conditions.

The sensitivity analysis for bulldozer are constructed by two different work conditions: first, as shown in Table 8, bulldozer has to haul 1000 cy of soil in 300 ft of distance, using various size of engine and all types of soil; second, as shown in Table 9, 564 hp bulldozer – model year 2003, has to haul 1000 cy of all type of soil in various distances. Based on the information in Tables 8 and 9, there is an inverse relationship between productivity rate and the fuel use and CO₂ emissions; that is, as the productivity rate decreases, the fuel use and emissions increase. The productivity rate also decreases with the dozing resistance based on soil type; sand–gravel has the highest productivity rate whereas clay has the lowest. Likewise, all the emissions increase as the soil resistance increases. Furthermore, for a specific soil type, the productivity rate increases as engine size increases, and decreases as the dozing distance increases. In general, Table 8 shows the inverse relationship between engine size and total fuel use and CO₂ emissions; that is, for all types of soil, as the bulldozer uses bigger size of engine or bigger rated horsepower, the emissions become lower. There is little difference in fuel use and CO₂ emissions for sand–gravel, sandy clay–loam and common earth. The soil type with the highest estimated total emissions based on engine size is clay.

Table 8 Bulldozer – model year 2003; 300 ft distance; 1000 cy soil.

Horsepower	Type of material	Productivity (cy/h)	Fuel use (gal)	CO2 (tons)
80	Sand and gravel	18.33	215.84	1.77
80	Sandy clay and loam	17.42	227.11	1.86
80	Common earth	14.88	265.84	2.18
80	Clay	7.85	503.95	4.14
150	Sand and gravel	24.37	273.75	2.50
150	Sandy clay and loam	23.24	287.10	2.62
150	Common earth	20.02	333.27	3.04
150	Clay	10.99	607.09	5.54
275	Sand and gravel	39.07	313.14	2.86
275	Sandy clay and loam	37.41	326.96	2.98
275	Common earth	32.69	374.18	3.42
275	Clay	19.03	642.80	5.87
400	Sand and gravel	60.14	295.88	2.70
400	Sandy clay and loam	57.79	307.80	2.81
400	Common earth	51.07	348.37	3.18
400	Clay	31.21	570.06	5.20
525	Sand and gravel	89.45	261.08	2.38
525	Sandy clay and loam	86.22	270.89	2.47
525	Common earth	76.93	303.58	2.77
525	Clay	48.95	477.09	4.35
600	Sand and gravel	111.90	238.59	2.18
600	Sandy clay and loam	108.03	247.13	2.25
600	Common earth	96.88	275.42	2.51
600	Clay	62.98	423.81	3.87
700	Sand and gravel	148.61	209.55	1.91
700	Sandy clay and loam	143.75	216.71	1.98
700	Common earth	129.70	240.06	2.19
700	Clay	86.49	359.94	3.29

Table 9 Bulldozer – 564 hp; model year 2003; 1000 cy soil.

Distance (ft)	Type of material	Productivity (cy/h)	Fuel use (gal)	CO2 (tons)
50	Sand and gravel	305.24	82.29	0.75
50	Sandy clay and loam	296.60	84.54	0.77
50	Common earth	271.37	92.57	0.84
50	Clay	191.69	130.95	1.19
100	Sand and gravel	249.14	100.60	0.92
100	Sandy clay and loam	241.79	103.86	0.95
100	Common earth	220.40	113.89	1.04
100	Clay	153.31	163.57	1.49
150	Sand and gravel	201.60	124.43	1.14
150	Sandy clay and loam	195.40	128.45	1.17
150	Common earth	177.39	141.49	1.29
150	Clay	121.35	206.72	1.89
200	Sand and gravel	161.61	155.29	1.42
200	Sandy clay and loam	156.42	160.31	1.46
200	Common earth	141.38	177.36	1.62
200	Clay	94.95	264.17	2.41
250	Sand and gravel	128.24	195.68	1.79
250	Sandy clay and loam	123.93	202.45	1.85
250	Common earth	111.46	225.03	2.05
250	Clay	73.36	341.94	3.12
300	Sand and gravel	100.62	249.36	2.28
300	Sandy clay and loam	97.07	258.40	2.36
300	Common earth	86.84	289.00	2.64
300	Clay	55.89	448.81	4.10

Table 9 shows the emissions impact of the bulldozer activity based on the dozing distance. For each soil type, the estimated productivity decreases as the dozing distance increases. The estimated productivity has relatively the same shape for all types of soil, with little difference of productivity for sand–gravel, sandy clay–loam and common earth. Again, clay gives the lowest estimated productivity for the bulldozer activity based on dozing distance. Table 9 also shows the emissions impact of the bulldozer activity based on dozing distance. There are positive relationship between dozing distance and the fuel use and CO₂ emission; that is, for all types of soil, as the bulldozer has to haul longer distance, the emissions become higher. For all types of soil, Table 9 displays that clay gives the bulldozer the highest fuel use and CO₂ emissions based on dozing distance.

The sensitivity analysis for excavator is applied in two different work scenarios: first, a 400 hp – model year 2003 excavator has to dig 100 ft long – 10 ft wide trench in various depth, using 3 cy of bucket size on all types of soil; second, the same excavator with various size of bucket has to dig 100 ft long – 10 ft wide – 12 ft deep trench on all types of soil as well. Based on the information in Table 10 and Table 11, there is an inverse relationship between productivity rate and the emissions; that is, as the productivity rate decreases, the emissions increase. The productivity rate also decreases with the digging resistance based on soil type; sand–gravel has the highest productivity rate whereas clay has the lowest. Likewise, although with very little difference, the fuel use and CO₂ emissions increase as the soil resistance increases. Furthermore, for a specific soil type, the productivity rate increases as bucket size increases, and decreases as the trench depth increases. For all types of soil, the fuel use and CO₂ emissions increase as the trench depth increases. As the excavator digs a trench not deeper than 12 ft, the estimated emissions for all types of soil are about the same (Table 10). Although the fuel use and CO₂ emissions from all types of soil began to vary when the trench went deeper than 12 ft, particularly for hard clay and common earth, the overall estimated CO₂ emissions are considered the same.

Table 10 Excavator – 400 hp; model year 2003; trench: 100 ft long – 10 ft wide; 3 cy bucket size.

Type of soil	Depth (ft)	Productivity (cy/h)	Fuel use (gal)	CO2 (tons)
Common earth	2.5	168.05	9.79	0.09
Loam and sandy clay	2.5	174.50	9.43	0.09
Sand and gravel	2.5	176.00	9.43	0.09
Clay	2.5	159.59	10.32	0.09
Common earth	5	162.88	20.28	0.19
Loam and sandy clay	5	169.32	19.39	0.18
Sand and gravel	5	170.83	19.22	0.18
Clay	5	154.42	21.35	0.19
Common earth	8	156.67	33.63	0.31
Loam and sandy clay	8	163.11	32.20	0.29
Sand and gravel	8	164.62	32.03	0.29
Clay	8	148.21	35.58	0.32
Common earth	12	148.40	53.20	0.49
Loam and sandy clay	12	154.84	51.06	0.47
Sand and gravel	12	156.34	50.53	0.46
Clay	12	139.93	56.40	0.51
Common earth	17	138.05	81.13	0.74
Loam and sandy clay	17	144.49	77.40	0.71
Sand and gravel	17	146.00	76.68	0.70
Clay	17	129.58	86.29	0.79
Common earth	22	127.70	113.34	1.03
Loam and sandy clay	22	134.14	108.00	0.99
Sand and gravel	22	135.65	106.75	0.97
Clay	22	119.24	121.52	1.11

Table 11 Excavator – 400 hp; model year 2003; trench: 100 ft long – 10 ft wide – 12 ft deep.

Type of soil	Bucket vol. (cy)	Productivity (cy/h)	Fuel use (gal)	CO2 (tons)
Common earth	0.63	17.46	445.34	4.13
Loam and sandy clay	0.63	23.90	326.66	3.02
Sand and gravel	0.63	25.41	307.63	2.84
Clay	0.63	9.00	852.24	8.01
Common earth	0.75	24.35	324.35	2.96
Loam and sandy clay	0.75	30.79	256.56	2.34
Sand and gravel	0.75	32.30	244.64	2.23
Clay	0.75	15.89	497.11	4.54
Common earth	1.00	38.13	207.1	1.89
Loam and sandy clay	1.00	44.58	177.21	1.62
Sand and gravel	1.00	46.08	171.52	1.57
Clay	1.00	29.67	266.17	2.43
Common earth	1.50	65.70	120.28	1.10
Loam and sandy clay	1.50	72.14	109.42	1.00
Sand and gravel	1.50	73.65	107.29	0.98
Clay	1.50	57.23	138.07	1.26
Common earth	2.50	120.83	65.3	0.60
Loam and sandy clay	2.50	127.27	62.09	0.57
Sand and gravel	2.50	128.78	61.38	0.56
Clay	2.50	112.37	70.28	0.64
Common earth	3.00	148.40	53.2	0.49
Loam and sandy clay	3.00	154.84	51.06	0.47
Sand and gravel	3.00	156.34	50.53	0.46
Clay	3.00	139.93	56.4	0.51
Common earth	3.50	175.96	44.84	0.41
Loam and sandy clay	3.50	182.40	43.23	0.39
Sand and gravel	3.50	183.91	42.88	0.39
Clay	3.50	167.50	47.15	0.43

The fuel use and CO₂ emissions from excavator activity based on its bucket size are presented in Table 11. The estimated emissions decrease dramatically while excavator digs the trench with bucket size less than 1.50 cy. In this range of bucket size, clay has the highest emissions, while sand–gravel and sandy clay–loam has the lowest. However, when the bucket size used is bigger than 1.50 cy, the estimated emissions start decreasing slowly and shows that for all types of soil, the fuel uses are the same. It indicates that the productivity rate of excavator based on the resistance of soil or soil type varies only when the excavator uses small size of bucket. For bigger bucket size, the resistance of soil does not have impact on productivity rate.

The sensitivity analysis for truck is applied in two different work situations: first, a 535 hp – model year 2003 truck with 30 cy loading capacity has to haul 1000 cy of soil in various distance, using various hauling speed, and has to wait for dumping and loading for 15–25 min; second, the same truck with various capacity of loading has to haul 1000 cy of soil in 1 mile, using average hauling speed range of 5–25 mph, and has to wait for loading and dumping for 15 min. Based on the information in Tables 12 and 13, there is an inverse relationship between productivity rate and the fuel use and CO₂ emissions; that is, as the productivity rate decreases, the fuel use and emissions increase. For any speed the truck uses to haul the soil, the productivity rate increases as loading capacity increases, and decreases as the haul distance increases.

Table 12 Truck – 535 hp; model year 2003; 30 cy capacity; 1000 cy soil; 10 mph of speed.

Distance (miles)	Load–dump time (min)	Productivity (cy/h)	Fuel use (gal)	CO2 (tons)
0.38	15	91.02	285.97	2.39
0.38	20	77.07	337.75	2.82
0.38	25	63.13	412.17	3.44
0.57	15	88.72	293.26	2.45
0.57	20	74.78	347.90	2.90
0.57	25	60.84	427.79	3.57
0.76	15	86.43	301.06	2.51
0.76	20	72.49	359.09	3.00
0.76	25	58.55	444.44	3.71
1	15	83.54	311.47	2.60
1	20	69.60	373.92	3.12
1	25	55.65	467.60	3.90
2	15	71.48	364.03	3.04
2	20	57.54	452.25	3.77
2	25	43.60	596.92	4.98
4	15	47.37	549.30	4.59
4	20	33.43	778.29	6.50
4	25	19.48	1335.66	11.15

Table 13 Truck – 535 hp; model year 2003; 1 mile hauling distance; 15 min wait–dump time; 1000 cy soil.

Capacity (lcy)	Speed (mph)	Productivity (cy/h)	Fuel use (gal)	CO2 (tons)
22	5	58.78	404.79	3.70
22	10	66.91	355.77	3.25
22	15	75.03	317.21	2.89
22	20	83.16	286.28	2.61
22	25	91.28	260.82	2.38
30	5	75.42	315.55	2.88
30	10	83.54	284.85	2.60
30	15	91.67	259.63	2.37
30	20	99.79	238.45	2.18
30	25	107.91	220.6	2.01
42	5	100.37	237.02	2.16
42	10	108.49	219.41	2.00
42	15	116.62	203.94	1.86
42	20	124.74	190.85	1.74
42	25	132.87	179.19	1.63
60	5	137.80	172.77	1.58
60	10	145.92	163.01	1.49
60	15	154.04	154.44	1.41
60	20	162.17	146.83	1.34
60	25	170.29	139.69	1.28

The NO_x and PM emissions from truck activity based on cycle distance are shown in Table 12, analysed using three different load–dump times: 15, 20 and 25 min. For all load–dump times, estimated emissions increase as the cycle distance increases. The longer the truck has to wait for loading and dumping, the higher the fuel use and emissions. More productive supporting equipment (such as excavator or backhoes) for loading and unloading soil to truck is needed to shorten the load–dump time and improve the truck's productivity rate. The estimated fuel use and CO₂ emissions for three load–dump times have little difference when the truck has to haul within less than 2 miles. For hauling distance more than 2 miles, the estimated emissions vary for three different load–dump times. Truck with 25 min load–dump time increases its fuel use and emissions of CO₂, very rapidly compared with 15 and 20 min load–dump times, as the hauling distance increases.

The productivity rates and emissions impact of truck activity based on loading capacity, and analysed using five different hauling speeds (5, 10, 15, 20 and 25 mph), are shown in Table 13. For all hauling speeds, the estimated fuel use and CO₂ emission decrease as the loading capacity increases. It is understood that the productivity rate of truck improves when using bigger loading capacity, and therefore shorten the hauling duration. Truck with highest hauling speed (25 mph) has the lowest fuel use and emissions. The difference of emissions among all hauling speeds becomes smaller following the loading capacity; for instance, the difference of fuel use and emissions at 25 cy loading capacity is bigger than those at 50 cy or more loading capacity. It indicates that using various speed of hauling does not have bigger impact on the estimated fuel use and CO₂ emissions if the truck uses bigger loading capacity.

6. Conclusion and future works

This article presented a methodology for estimating fuel use and CO₂ emissions for some common earthwork activities performed by bulldozer, excavator and dump truck. The MLR approach proved to be a useful alternative for estimating the productivity rate of these equipments. The MLR models for the productivity rate can explain high percentage of the variability in the data. The models are good to be used as a benchmark for estimating fuel use and CO₂ emissions from certain types of construction equipment performing earthwork activities. The productivity rate from this model (lcy/h) is used with BSFC (g/hp-h) from EPA's NONROAD model to estimate the total fuel use and CO₂ emissions.

Based on the methodology presented in this article, the results revealed several trends related to total emissions of the equipment. For example, the total emissions increase as the trench depth or hauling distance increase, because digging deeper trench or hauling further distance lead to lower productivity. Meanwhile, as the excavator's bucket size or bulldozer's blade or truck's capacity increases, the total emissions decrease, because the productivity gets higher when the equipment uses bigger attachment size.

After the estimating tool has been developed, it is recommended for future research to validate and calibrate the model by real-world fuel use and CO₂ emissions data collected from construction equipment. This will be done by using a portable emissions measurement system that is able to record second-by-second fuel use, emissions and engine data from the equipment performing earthwork activities. The field data collection and analysis process will permit evaluation of the variability in fuel use and emissions rates among equipments based on type, engine size, engine load and usage.

The estimating tool proposed in this article will be an effective means for assessing the environmental impacts of construction activities and will allow equipment owners or fleet managers, policy-makers and project stakeholders to evaluate more sustainable alternatives. The tool will help the contractor to estimate the total expected pollutant emissions for the project, which would be valuable information for a preliminary environmental assessment of the project.