Research Article  Open Access
Prediction Method of the Fuel Consumption of Wheel Loaders in the VType Loading Cycle
Abstract
Wheel loaders in the Vtype loading cycle are characterized by complicated loading conditions, nonlinear powertrain system, and timevariable engine power distribution. Therefore, it is difficult to predict the fuel consumption of wheel loaders in the Vtype loading cycle. The static matching methods cannot provide fuel consumption prediction for the loading cycle. In this paper, the prediction method and model of the fuel consumption for wheel loaders in the Vtype loading cycle were proposed. Firstly, the hydraulic system data were tested when a wheel loader loaded three different materials in a typical Vtype loading cycle. Secondly, the tested data were filtered by the 8thorder Butterworth filter and the dimensionless power deduction equations of hydraulic power system for loading three different materials were obtained by Gaussian and linear fitting based on the filtered data in the loading cycle. Finally, fuel consumption was obtained with the compiling dynamic calculation program as well as input parameters of tested vehicle speed, throttle parameter, and the dimensionless equation. The simulation results agreed well with experiment results. Dynamic calculation program is applicable to calculate loading economy and can provide academic guidance for wheel loader’s design and optimization.
1. Introduction
Vtype loading cycle fuel consumption is the most significant economic index for wheel loaders. The typical Vtype loading cycle, as shown in Figure 1, consists of V1 phase (moving forward to material), V2 phase (loading material), V3 phase (returning to the initial position from material), V4 phase (driving fullload loader to transporter), V5 phase (unloading material), and V6 phase (returning to the initial position from transporter) [1].
In the powertrain system of a wheel loader (Figure 2), the engine power is transported to three different parts: accessories, the hydraulic system, and the driveline system. In the Vtype loading cycle, driving resistance and digging resistance are mainly loaded on the wheel loader. One loading cycle may contain one single load or multiple loads. The engine power distribution is closely related to the external workloads. Wheel loaders in the Vtype loading cycle are characterized by complicated loads, nonlinear powertrain system, and timevariable engine power distribution. Therefore, it is difficult to predict the fuel consumption [2].
Wheel loaders in the Vtype loading cycle had been extensively studied. Filla et al. developed the dynamic simulation platform for wheel loader vehicle performance, economy, and operability by combining the automatic control model with the operator model [1, 3, 4]. Hemami planned an optimal digging path to minimize the digging force and enhance the loading economy and efficiency [5]. Yossawee et al. proposed the method to calculate initial and final positions with the symmetric Clothoid curve and planned a Vtype loading cycle path reasonably [6]. Wu et al. got an optimal loading path by means of test and theory analysis and then controlled the bucket along the path to improve fuel efficiency and operability [7–9].
The Vtype loading cycle was mainly obtained by test and fuel consumption prediction was seldom obtained through simulation. Bohman obtained a simple model by dividing the cycle into different parts to calculate Vtype loading cycle fuel consumption, but the relative error was large and generally ranged from 10% to 20% [10]. This paper aims to develop a method to predict the Vtype loading fuel consumption accurately.
Engine power distribution in the Vtype loading cycle is the key to study fuel consumption of wheel loader. The power from engine is mainly to balance digging resistance and driving resistance and the digging resistance balanced by the hydraulic system is complicated and unpredictable. Takahashi got a curve of force when the loader was digging some materials by tests and the simulation method and the curve showed the complex nonlinear change [11]. However, the power transportation loss in the hydraulic system was unknown. Therefore, it was impossible to calculate the hydraulic system power. Filla got the power distribution map by simulation [12]. Debeleac got the engine power distribution of the wheel loader working in the Vtype loading cycle by means of the theoretical calculation with testing data of the hydraulic system [13]. In the paper, the hydraulic system power distribution characteristic was gotten by the mathematic calculation method [13].
The engine power of a wheel loader is generally diverted to the transmission system, the hydraulic system, and accessories. In this paper, engine power deduction is defined as the power obtained after subtracting the power diverted to the transmission system and can represent engine power distribution. Engine power deduction is widely used in static matching. In general, the static matching method can be expressed as
It means that, at a certain gear, the vehicle velocity can be determined by three different parameters: throttle , power deduction , and running resistance . In other words, velocity reflects the running resistance of vehicle. The traditional static matching cannot be used to predict fuel consumption of a wheel loader in loading cycle. Equation (1) may be rewritten asTaking timevariable throttle , engine power deduction , and velocity as input parameters to compile dynamic calculation program based on the traditional matching program, we obtained fuel consumption ratio curve through the iterative matching. Vtype loading cycle fuel consumption can be acquired by the accumulation method.
2. Dynamic Power Deduction in the VType Loading Cycle
2.1. Test
When the wheel loader works in the Vtype loading cycle, the hydraulic system power varies with time. The pressure and flow of working pump and steering pump (red circle shown in Figure 3) on the output side of the gearbox are tested while the loader is digging sand, stone, and freestone. The sensor placement is shown in Figure 3. The pressure and flow signals were tested with pressure flow sensors. The inlets of working pump and steering pump were connected to the tank. Therefore, only one of the inlet pressure and flow should be tested. The testing controller could be used to collect the data. The collected data were stored in the internal storage unit of controller and the sampling frequency was set to be 100 HZ.
2.2. Data Processing of Hydraulic System
The signal of a wheel loader is a typical lowfrequency signal and only stable flat pass band performance could ensure the integrity of the useful signal at the most. Therefore, IIR digital filter was chosen to process the data. Butterworth filter attenuation rate was slower than other IIR digital filters, but its attenuation rate was very flat without amplitude variation. The higher order means that its performance is closer to the ideal lowpass filter [14]. Squared magnitude response of Butterworth lowpass filter can be expressed by the following [15]:where is the order of the filter and is the cutoff frequency of the lowpass filter.
The 8thorder butterwort lowpass filter with a cutoff frequency of 2 HZ was selected in order to filter the inlet and outlet pressure data of hydraulic pump when the wheel loader was used to load freestone material in the Vtype loading cycle. The result is shown in Figure 3.
As shown in Figure 4, represents the outlet pressure of working pump; represents the outlet pressure of steering pump; represents the pressure of intake pipe. The mathematical statistics analysis indicated that the pressure signals of pump became smoother after filtering and that the completeness and accuracy of the filtered data were maintained (Table 1). The filtered data are more conducive to the model loading.

(a)
(b)
After processing the flow data of hydraulic pump in the same way, the curves are obtained and shown in Figure 5.
(a)
(b)
In Figure 5, is the flow of working pump; is the flow of steering pump.
2.3. Power Deduction Calculation
The power of hydraulic systems was calculated by pressure difference between the inlet and outlet flows of hydraulic pumps. The hydraulic system power shows the complicated change in Vtype cycle and can be calculated aswhere is the differential pressure between the inlet and outlet pressures, is the volume flow rate, and is the overall efficiency.
The ratio of hydraulic pump power to rated engine power plus the accessory power ratio equals the engine power deduction in the Vtype loading cycle. Accessory power consumption is nonlinear. But limited by the condition of technical test, we have simplified the auxiliary power ratio whose number is 10% according to engineering experience. The engine power deduction can be calculated by (5). The engine power deduction curve is shown in Figure 6. Considerwhere is the engine power deduction, is the accessory deduction ratio which is 10% in this paper, is working pump power, is steering pump power, and is the rated engine power.
In Figure 6, the power deduction in the whole Vtype loading cycle is complicated and the data have no linear relationship with time in each phase. Therefore, fitting engine power deduction data in each phase with nonlinear regression function can provide good fitting effects.
In the test, we tested the loader’s related data when loading three different materials (freestone, sand, and stone) in the Vtype cycle. It can be considered that working conditions of a wheel loader remain unchanged when the same driver operates the same wheel loader except the V2 phase. That is to say, the engine power deduction remains unchanged except in the V2 phase. And when loading different materials, bucket movement and vehicle velocity show little differences except temporal variations.
Through fitting each phase in loading cycle with Matlab Cftool as well as a lot of mathematic tests, we found that Gaussian function could accurately fit the engine power deduction curve and that linear fitting could ensure the function continuity between two working phases. Gaussian function and linear function are, respectively, expressed as follows:
Table 2 shows the coefficients of dimensionless power deduction equation when loading freestone material. , , and are the Gaussian fitting factors and is the linear fitting factor.

In the power deduction curve based on test data (Figure 7), the fitting equation curve and the relative error directly show that the equation has accurately described the power deduction.
In the same way, we fit the engine power deduction data when loading sand material and stone material in V2 phase via Gaussian and linear fitting (Table 3 and Figures 8 and 9).

3. Fuel Consumption Calculation in the VType Loading Cycle
The calculation method of fuel consumption in the Vtype loading cycle is described as follows.
3.1. Mathematic Calculation Method
In the Vtype loading cycle, wheel loader’s fuel consumption ratio (mL/s) in different time can be obtained via static matching calculation. Each additional period s is set as a small interval and the fuel consumption ratio (mL/s) multiplied by the is equal to the fuel consumption within the interval, . This is the approximate calculation. The total fuel consumption (mL) from to can be obtained aswhere is the fuel consumption at time when vehicle velocity is and is the fuel consumption at time when vehicle velocity is .
As shown in Figure 10, the fuel consumption of each interval can be expressed aswhere , and are, respectively, fuel consumption at times , and .
In the slowing down phase, a driver loosens engine throttle valve and brakes slightly and the torque converter may work in the reverse transfer condition that makes engine in idling state. Under this condition, the fuel consumption ratio is idling fuel consumption (mL/s). The total fuel consumption (mL) in reversed transmission condition equals the product of idling time (s) and :
Therefore, the total fuel consumption in the Vtype loading cycle can be calculated bywhere is the number of sections in which the torque converter works under the traction condition.
3.2. Dynamic Calculation
Based on the abovementioned fuel consumption calculation method and the static matching program, we compiled a dynamic calculation program to acquire the vehicle traction characteristics and fuel consumption characteristics at different time and obtained the dynamic traction force and dynamic fuel consumption ratio in the whole Vtype loading cycle phase. The block diagram of dynamic calculation is shown in Figure 11. The fuel consumption prediction model for wheel loaders in the Vtype loading cycle is shown in Figure 12.
The simulation settings are as follows: fixedstep BogackiShampine solver; step length of 0.1 s; the total iterative calculation time which is the time length of a Vtype loading cycle. Taking the working conditions of loading freestone material as an example, the input parameters are shown in Figure 13. Velocity, throttle, and power deduction can be rebuilt based on the input parameters for different wheel loaders.
3.3. Static Matching
The static matching mentioned above is described as follows. If the engine power and throttle at the given time are known, the static matching performances between engine and torque converter, including common input characteristics, common output characteristics, traction characteristics, and fuel consumption characteristics at the time, can be acquired. The block diagram of static matching is shown in Figure 14. The vehicle traction characteristic shows the relationship between traction force and velocity at different gears. The fuel consumption characteristic shows the relationship between fuel consumption ratio and velocity at different gears.
After linear fitting with the testing data of engine and torque converter, the obtained fitting curves are, respectively, shown in Figures 15 and 16. Equations (11) and (12) express the engine throttle characteristics and original torque converter characteristics, respectively. Considering the environment conditions for engine experiment such as the temperature and humidity are similar with the conditions for Vtype loading cycle, we neglect the influence of environmental factors on the performance of the engine. Considerwhere and are engine torque, is engine power, is specific fuel consumption, and , , and are fitting coefficients. One haswhere is speed ratio, is capacity factor, and is efficiency.
Vehicle force balance is calculated bywhere is traction force, , is rolling resistance, , is wind resistance, , is slop resistance, , and is acceleration resistance, .
Fuel consumption ratio can be calculated bywhere is unit weight of fuel.
The fuel consumption mathematical model is the key to predict the fuel consumption of the Vtype loading cycle. Comparing the calculated running fuel consumption based on (14) with testing data, as shown in Figure 17, we can see that the relative errors are 0.62% and 0.51%, respectively. The results show that the model is accurate for predicting the fuel consumption.
After extracting engine throttle and power deduction ratio at s, the static matching performances are obtained (Figure 18).
(a)
(b)
(c)
(d)
After extracting the velocity and gear data at s, the transient traction force and fuel consumption ratio can be obtained by the interpolation method based on the traction characteristics and fuel consumption characteristics.
4. Simulation and Results
Figure 19 shows the engine output characteristics including dynamic engine gross power and specific fuel consumption. Figure 20 shows fuel consumption characteristics including fuel consumption ratio and total fuel consumption. Comparing Figure 19 with Figure 20, we can find that the tendency of vehicle fuel consumption characteristics is consistent with that of engine dynamic output characteristics. According to (14), fuel consumption ratio is proportional to the product of specific fuel consumption and engine output gross power. In the Vtype loading cycle, the change of the specific fuel consumption is not large. Therefore, the engine output power is a crucial factor of fuel consumption ratio.
Dynamic traction force reflects the vehicle dynamic performance in the Vtype loading cycle to some degree. As shown in Figure 21, we can see that average output power of turbine shaft is higher in noload or fullload running phases excluding V5 phase than that in V2 phase. The reason is that the lower velocity leads to the lower output power from turbine shaft in V2 phase, but the higher sticking force to insert material and the higher velocity in the running phase lead to the higher average output power in noload and fullload running phases.
Fuel consumption comparison between test results and simulation results when loading different materials (freestone, sand, and stone) in the Vtype loading cycle is shown in Figure 22. In general, the simulation curve fluctuates near the test curve which truly reflects the fuel consumption change in each phase of Vtype loading cycle. The relative errors are calculated by (15) and the fuel consumption comparison is shown in Figure 23. Considerwhere is the tested fuel consumption and is the simulated fuel consumption.
(a)
(b)
(c)
As shown in Figure 23, the relative errors for various working conditions of loading freestone, sand, and stone are, respectively, 1.84%, 3.44%, and 2.03%, which are all lower than the required error in the engineering calculation, 5%. The result indicates that the loose degree of material is closely related to fuel consumption: the looser material means the lower fuel consumption.
5. Conclusions
(1)The proposed method in this paper for predicting the fuel consumption of wheel loaders in the Vtype loading cycle can accurately predict the fuel consumption in the Vtype loading cycle. The relative errors for various working conditions of loading freestone, sand, and stone are, respectively, 1.84%, 3.44%, and 2.03%, which all meet engineering calculation requirements. The method can provide theoretical reference for design and optimization of wheel loaders.(2)The model can not only predict the Vtype loading fuel consumption but also provide the turbine shaft output power characteristics and vehicle dynamic output traction characteristics for judging whether the matching between engine and torque converter is reasonable.(3)The dimensionless equations of engine power deduction based on the testing data provided the reference data for the design of wheel loaders.
Conflict of Interests
The authors declare that there is no conflict of interests regarding the publication of this paper.
Acknowledgment
This work was supported by the National High Technology Research and Development Program of China (863Program) (Grant no. 2014AA041502).
References
 R. Filla, “An eventdriven operator model for dynamic simulation of construction machinery,” http://arxiv.org/abs/cs/0506033. View at: Google Scholar
 C. Debeleac, “Vibratory diagnosis of the earthmoving machines for the additional necessary power level evaluation,” in Proceedings of the 9th WSEAS International Conference on Acoustics & Music: Theory & Applications (AMTA '08), pp. 24–26, Bucharest, Romania, June 2008. View at: Google Scholar
 R. Filla, A. Ericsson, and J. O. Palmberg, “Dynamic simulation of construction machinery: towards an operator model,” http://arxiv.org/abs/cs/0503087. View at: Google Scholar
 R. Filla and J. O. Palmberg, “Using dynamic simulation in the development of construction machinery,” http://arxiv.org/abs/cs/0305036. View at: Google Scholar
 A. Hemami, “Motion trajectory study in the scooping operation of an LHDloader,” IEEE Transactions on Industry Applications, vol. 30, no. 5, pp. 1333–1338, 1994. View at: Publisher Site  Google Scholar
 W. Yossawee, T. Tsubouchi, S. Sarata, and S. Yuta, “Path generation for articulated steering type vehicle using symmetrical clothoid,” in Proceedings of the IEEE International Conference on Industrial Technology, vol. 1, pp. 187–192, IEEE, 2002. View at: Publisher Site  Google Scholar
 L. Wu, A study on automatic control of wheel loaders in rock/soil loading [Ph.D. thesis], University of Arizona, Tucson, Ariz, USA, 2003.
 S. Singh, State of the Art in Automation of Earthmoving, 2002, Carnegie Mellon University, 2002.
 J. A. Marshall, Towards autonomous excavation of fragmented rock: experiments, modelling, identification and control [M.S. thesis], Queen's University, Kingston, Canada, 2001.
 M. Bohman, On predicting fuel consumption and productivity of wheel loaders [M.S. thesis], Lulea University of Technology, Lulea, Sweden, 2006.
 H. Takahashi, “Simulation of the resistive forces acting on the bucket of wheel loader by use of DEM,” in Proceedings of the 5th International ISTVS Conference, pp. 37–44, Seoul, Republic of Korea, 1998. View at: Google Scholar
 R. Filla, Operator and machine models for dynamic simulation of construction machinery [Ph.D. thesis], Linköping University, Linköping, Sweden, 2005.
 C. Debeleac, “Performances assessment from an earthmoving machine through power monitoring,” Annals of Dunarea de Jos University of Galati Fascicle XIV: Mechanical Engineering, In press. View at: Google Scholar
 K. S. Erer, “Adaptive usage of the Butterworth digital filter,” Journal of Biomechanics, vol. 40, no. 13, pp. 2934–2943, 2007. View at: Publisher Site  Google Scholar
 R. G. T. Mello, L. F. Oliveira, and J. Nadal, “Digital Butterworth filter for subtracting noise from low magnitude surface electromyogram,” Computer Methods and Programs in Biomedicine, vol. 87, no. 1, pp. 28–35, 2007. View at: Publisher Site  Google Scholar
Copyright
Copyright © 2015 Wenxing Ma et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.