1. Introduction
The drive axle housing of the truck, located at the end of the drive train, is a critical chassis bearing and power transmission component of the vehicle [
1]. The drive axle housing mainly bears the reaction force from the ground, as well as the vertical force, longitudinal force, transverse force, and braking torque between the road frame or vehicle body [
2]. It requires that the strength and stiffness of the drive axle housing meet its specified service requirements [
3]. To realize the lightweight design of a drive axle housing, many scholars have carried out lightweight designs based on different methods.
Yucun Zhou took a light truck drive axle housing as the research object, selected the design parameters, limited stress, and displacement constraints with light weight as the optimization objectives, and realized the purpose of light weight [
4]. Bingbing Zhou analyzed the dynamic characteristics and fatigue of the drive axle housing and then carried out the lightweight optimization design to verify the optimized drive axle housing [
5]. Yang Chen put forward a reliability analysis method based on the Monte Carlo method by studying the optimization scheme of the drive axle housing of off-road vehicles. On this basis, they carried out a lightweight design and reduced the weight of the drive axle housing by 12.48% [
6]. Guo Zhongjia optimized the rear axle housing of a light truck via finite element method based on the mechanical calculation to achieve the lightweight purpose. The low-order-mode natural frequency of the drive axle housing is too high [
7]. Zhang Jun optimized it to increase the low-order mode frequency by about 10 Hz under the restriction of mass [
8]. Xu Wenchao carried out the Six Sigma robust multiobjective lightweight design of the drive axle housing, studied the influence of wall thickness of the drive axle housing on its performance through the methods of entropy weight and TOPSIS, and combined RBF and NSGA-II algorithms to design a multiobjective lightweight drive axle housing, which significantly improved the performance [
9]. Based on static analysis, Yu Yunyun carried out DOE experimental design for 12 design parameters, established the restraint and optimization objective functions, and carried out a multiobjective optimization design, which not only reduced the mass but also guaranteed the performance stability of the drive axle housing [
10]. To optimize the space structure of the drive axle housing, Xu Kang carried out the topological optimization design under the limited conditions of different working conditions, which made the stress distribution more uniform while reducing the weight and lowering the maximum stress [
11].
Several scholars have studied the dynamic characteristics of the drive axle housing to varying degrees. Wang Xuemei analyzed and optimized the dynamic characteristics of the drive axle housing by the separation method, which improved the minimum fatigue life while reducing the mass [
12]. Zheng Bin carried out static and dynamic characteristics analysis and multiobjective optimization design under three different conditions, which improved the comprehensive performance of the drive axle housing [
13]. To study the resonance characteristics of the drive axle housing, Li Huilin carried out modal and harmonic response analysis and identified the dynamic characteristics of the drive axle housing [
14]. Liu Guozheng analyzed and verified the vibration and noise of the drive axle housing by combining a simulation with a bench test, and obtained the contribution of vibration and noise at different locations [
15]. To study the NVH performance of the drive axle, Jiao Dongfeng established a set of NVH performance-analysis methods for the drive axle housing and analyzed the vibration and noise of the passenger car at 60–65 km/h, which verified the validity of the methods [
16].
Numerous studies have been carried out to explore the fatigue life of the drive axle housing. As the fatigue failure life of drive axle housing cannot be calculated accurately by simulation analysis or bench test, Shao Yimin put forward a new analysis method based on dynamic strain measurement of actual mine pavement conditions and combined it with finite element analysis to calculate the fatigue failure life, which provides the technique for drive axle housing simulation [
17]. Zhao Wei studied the dynamic response and fatigue life prediction of vehicle systems under a random road spectrum and analyzed the magnitude of dynamic stress and fatigue life in the resonance frequency region [
18]. To study the vertical fatigue of the drive axle housing, Meng Qinghua proposed a seven-degree-of-freedom dynamic model to predict the fatigue failure under dynamic load and put forward an optimization scheme on this basis [
19]. Shang Minzheng used ABAQUS software to carry out dynamic and static characteristics analysis and fatigue analysis of automotive drive axle housing under sinusoidal load spectrum, obtained cloud diagrams of the position and life distribution of failure zone, and proposed improvement plans [
20]. Shou Xusong took the old drive axle housing as the research object and conducted a comparative study on residual state and predicted life to achieve high fatigue prediction accuracy [
21]. To study the fatigue life uncertainty of the drive axle housing design, Xian Zhongyu carried out static strength analysis and fatigue life analysis by using ABAQUS and researched the cumulative damage and limit safety coefficient [
22]. Li Jianmin established a dynamic test system for the drive axle housing of the loader and calculated the fatigue life of the drive axle housing. It was found that the fatigue area is different from the position where the maximum static stress occurs [
23]. Fan Zhimin obtained the failure zone of the drive axle housing through transient dynamic analysis and identified the failure zone of the drive axle housing by fatigue analysis. The comparison results between the bench test and the fatigue analysis are consistent [
24]. Liu Weiwei carried out dynamic and static characteristics analysis and fatigue life analysis under the maximum vertical force condition of the drive axle housing, extracted the fatigue life and safety factor, and further optimized the design [
25]. Yang Zhiqing used HyperMesh software to predict the fatigue life of the drive axle, obtained the distribution of the fatigue life of the drive axle, and verified the rationality of the design [
26]. Gao Jing utilized MSC. Patran carried out the finite element analysis of the drive axle, and the simulation results were compared with the fatigue test of the platform to reach a consistent conclusion [
27]. Guo Dongqing carried out the modal analysis and optimization of the drive axle, which reduced the mass of the drive axle housing by 27.3% [
28].
According to the research carried out by the above scholars, the main research objectives of the drive axle housing are mechanical characteristics, dynamic characteristics, fatigue damage life, and lightweight design with different working conditions. However, due to the uncertainty of the forced vibration of the drive axle housing, the vibration of the drive axle housing is unpredictable. At the same time, different manufacturers and users have different design requirements for the drive axle housing design. To improve the characteristics and design scheme of the drive axle housing, this paper establishes the three-dimensional model of the drive axle through SolidWorks software. The static analysis with four different working conditions is conducted in the ANSYS workbench to obtain the dangerous position and distribution. The fatigue life analysis with the maximum vertical force working condition is carried out to obtain the fatigue distribution of the drive axle housing with alternating loads at different locations. The vibration characteristics and resonant response frequency of the drive axle housing are also analyzed through dynamic characteristics. The random vibration characteristics of the drive axle housing under rough road surfaces are analyzed through random vibration. At the same time, two optimal design schemes are carried out based on mechanical analysis, which are topology optimization design and multiobjective optimization design. Based on the topology optimization and multiobjective optimization method, two lightweight design schemes are provided and two lightweight ideas of drive axle housing are proposed, which provide a reference for the drive axle housing when facing different requirements.
The paper is arranged as follows. In
Section 2, the finite element model of the drive axle housing is established. The force analysis diagram of the drive axle housing with four different typical working conditions is given.
Section 3 is the finite element analysis of typical working conditions, mainly including the maximum vertical force condition, the maximum traction working condition, the maximum lateral force working condition, the maximum braking working condition, and fatigue life analysis.
Section 4 is the dynamic characteristic analysis of the drive axle housing. The vibration characteristics of the drive axle housing are studied through modal analysis, harmonic response analysis, and random vibration response analysis.
Section 5 is the optimization design of the drive axle housing using topology optimization and multiobjective optimization to meet different requirements.
Section 6 is the conclusions.
5. Optimization Design of the Drive Axle Housing
5.1. Topology Optimization Design
The mass of the drive axle housing accounts for about 10% of the total mass of the truck. To realize the light weight of the truck, the lightweight design of the drive axle housing should be carried out first. In terms of lightweight design, topological optimization design has apparent advantages, such as optimum allocation of material space ratio and more design freedom [
41,
42]. Topology optimization guarantees the maximum utilization efficiency of structural materials without significantly impacting performance. For the topology optimization design of the drive axle housing, the optimum area should be determined first. In selecting the optimization area, the matching properties between the drive axle housing and other parts should be comprehensively considered. The optimized structure should not affect the matching properties of other parts. Therefore, the middle position is chosen as the topological optimization area, with the flange discs at both ends of the half-shaft sleeve as the boundary. The solution with the maximum vertical force condition is input into the topology optimization module, and the variable density optimization method is used as the design method to optimize the topology of the drive axle housing.
The mathematical model of the variable density method is shown in Formula (19).
where
ρ is the unit pseudo-density,
di is the relative density, and
ρi is the actual density of each unit.
Based on Formula (19), when
di is equal to 1, the material should be retained. When
di is equal to 0, the material is removed. Because the expression of relative density is discrete, it must be continuous in practice. Flexibility minimization is the optimization objective of topology optimization of drive axle housing; that is, to maximize the stiffness of the drive axle housing. The relationship between the two is expressed as Formula (20).
where
C is the flexibility of the structure,
F is the load matrix,
U is the structural displacement matrix, and
K is the overall stiffness matrix.
The mass retention of the topology optimization of the drive axle housing decreases from 80% to 30% with a gradient of 10%, and a total of six groups of optimization design schemes are obtained. The convergence accuracy of the control solution is 0.1%, the penalty factor is 3, and the maximum number of iterations is 500. The solution is carried out with the condition of maximum vertical force.
The mathematical model of topology optimization is shown in Formula (21).
where
f is the drive axle housing material retention percentage,
m is the initial design mass of the drive axle housing, and
m1 is the removal mass of the drive axle housing.
In
Figure 13, different colors represent different density distributions. The deletion of drive axle housing materials can be seen from the color distribution characteristics. In the figure, red indicates the low-density area, that is, the area with a relative density close to 0, which represents the deletion of the structural material. Gray represents the high-density area, that is, the area with a relative density close to 1, which represents the retention treatment of the structural material. Yellow is the transition area, which can be selected according to the manufacturing process and processing level of the drive axle housing. Based on the above optimization results, in the actual production, the red removal area needs to be appropriately filled in combination with the work needs, the reserved area should ensure its sufficient stiffness and static strength, and the other regions should be thinned. Most material removal positions are concentrated in the drive axle housing and the main reducer cavity housing. The smaller the mass retention percentage is, the more material is removed from the main reducer cavity shell. Most material removal positions are located at the position with less equivalent stress, which is opposite to the distribution of the stress cloud chart. Select the optimization results with 50% quality retention for the verification and analysis of the optimization results.
Based on the material removal strategy given by topology optimization, the model is modified by following methods.
The upper and lower end faces of the main reducer chamber
Step 1: The thickness of the mounting end of the leaf spring seats is reduced by 20 mm, and the fillet is 5 mm.
Step 2: A cylinder with a diameter of 176 mm and a height of 11.25 mm is cut from the upper and lower surfaces of the main reducer chamber, and the fillet is 60 mm.
Step 3: The fillet of the transition area of the main reducer chamber is 60 mm.
The statics analysis of the optimized drive axle housing is carried out again, and the results are shown in
Figure 15.
After topology optimization, the maximum deformation increases slightly compared to before optimization; the value is 1.2283 mm. The maximum equivalent stress increased by 24.17 MPa to 229.54 MPa compared with that before optimization. Compared with the initial model, the stress and deformation are increased. This is because there are certain errors in the manual processing of the model, and the material cannot be deleted completely according to the topology optimization results, resulting in a slight increase in the deformation stress of the model. The distribution is the same as that before optimization.
The fatigue analysis of the optimized drive axle housing with maximum vertical force is carried out again using the fatigue tool in ANSYS, and the results are shown in
Figure 16.
After optimization, the minimum life and minimum safety factor decreased to varying degrees. The minimum life decreased from 10,723 to 7551.5, a reduction of 29.5%. The safety factor decreased from 0.33578 to 0.30043, a reduction of 10.5%.
Then, the modal analysis of the optimized drive axle housing is carried out, and the results are shown in
Table 4 and
Figure 17.
After optimization, the natural frequency changes little, the vibration types are similar, the maximum deformation increases, the sixth-order frequency drops close to 80 Hz, and the equivalent stress and fatigue life have certain changes. However, they are all within the allowable range. The mass of the drive axle housing decreased significantly, from 504.88 Kg to 416.85 Kg, a decrease of 17.4%. The purpose of light weight is achieved.
5.2. Multiobjective Optimization Design
The drive axle housing is a critical component in the automobile chassis. When optimizing the drive axle housing, its static and dynamic characteristics will be affected, and the optimization results of favoring one over the other will further jeopardize the stability and safety of the vehicle. Based on this background, a multiobjective optimization design method based on the response surface method is adopted. Due to the coupling relationship between the structural parameters of the drive axle housing, the size to be optimized is taken as the design variable, the variable range is given, and the optimization objective is taken as the objective function. Within the given variable range, the best position between the given objective functions is searched through the changes in the design variables [
43,
44]. The multiobjective size-optimization design of the drive axle housing is to seek the optimal solution of the objective function on the premise of ensuring its functional performance. While meeting the lightweight requirement, it can also meet other performance requirements.
Due to the assembly relationship between the half-axle sleeve and the truck, the change in its size may lead to the change of mating parts. At the same time, the mass of the axle shaft sleeve is small in the whole drive axle housing, so it should be eliminated when optimizing the drive axle housing. Therefore, the multiobjective size-optimization design is only carried out for the middle chamber of the half-shaft sleeve, and the optimization design parameters are shown in
Figure 18.
When carrying out a multiobjective optimization design of drive axle housing based on the response surface method, DOE experimental design is required first. The range of design parameters is shown in
Table 5.
During DOE experimental design, the experimental method is selected as the optimal space-filling design method. This method can minimize the number of sample points, accurately reflect the changes of design points, and significantly avoid the inaccurate experimental results caused by too concentrated sample points [
45,
46,
47]. The sample type is CCD, and 27 samples are obtained. The mass, deformation, equivalent stress, service life, and the first, second, and third natural frequency of the drive axle housing with the maximum vertical force are taken as the optimization objective functions, expressed in
y1~
y7, respectively.
Through the response surface method, the corresponding tests are carried out on the test points, and the test results are mathematically analyzed to establish the response surface model. The relationship between design parameters and the objective function is observed through the response surface [
48,
49,
50]. The mathematical expression of the response surface method is shown in Formula (22).
where
y is the dependent variable;
xi and
xj are design variables;
i = 1, 2, 3, …
k;
k is the total number of design variables;
β0,
βi,
βii, and
βij are the response surface regression coefficients;
ζ is the response surface-fitting error.
According to the data obtained from DOE experimental design, multiple quadratic regression equation fitting is carried out to obtain the response value between design parameters and design points. The quadratic response surface mathematical model is established, as shown in Formulae (23)–(29).
The obtained response surface model is shown in
Figure 19.
Through the orthogonal experimental design and analysis of DOE data, the relationship curve between the design parameters of the drive axle housing and the objective function is obtained by the response surface method. It can be seen from
Figure 19a that
x1 has a weak impact on
y1, and the parameter that mainly affects
y1 response value is
x4, which shows a positive correlation with it.
Figure 19b shows that the design parameter that has a significant impact on
y2 is
x4, showing a positive correlation. As shown in
Figure 19c,
y3 is affected by
x2 and
x3 with little difference. It can be seen from
Figure 19d that
x1 and
x2 have a positive correlation with
y4, and the influence of
x1 is greater than that of
x2.
Figure 19e shows that the effect of
x1 and
x5 on
y5 is in the opposite relationship, and the effect alone is a negative correlation.
Figure 19f shows that
x2 has little impact on
y6, and
x1 has a positively correlates with y6.
Figure 19g shows that
x3 has little impact on
y7, and
x4 increases first and then decreases with
y7.
Multiobjective optimization sets the optimization expectation for the design points under the constraints of design parameters and given boundary conditions. The analysis system designs according to the optimization expectation. It makes the design of the drive axle housing meet the expected requirements of each design point to the greatest extent through mathematical or statistical methods.
The multiobjective optimization mathematical model of the drive axle housing is as follows:
where
X is the five design parameter points shown in
Table 5;
is the lower limit of the design parameters;
is the upper limit of the design parameters;
min is used to find the minimum value;
max is used to find the maximum value;
yi is the objective function;
i = 1, 2, 3,…, 7.
The optimization boundary conditions expressed in the above model are used for the solution, and the multiobjective genetic algorithm based on the NSGA-II algorithm is used for the solution. Three groups of optimal candidate points are obtained, as shown in
Table 6.
The above three groups of results are the optimal design solution, which can be selected according to the actual needs in the manufacturing process. After analysis, the first group is selected as the optimization result and compared with the initial data, as shown in
Table 7.
By comparing the data in the table, the mass of the optimized drive axle housing is reduced from 504.88 Kg to 482.92 Kg, a decrease of 4.35%. The deformation increases slightly from 1.06 mm to 1.09 mm, but it is negligible for the actual work of the drive axle housing. The maximum equivalent stress decreased from 205.37 MPa to 162.14 MPa, a decrease of 21.05%. The service life increased significantly from 10,723 to 18,474.10, an increase of 72.28%. The first, second, and third natural frequencies are increased in varying degrees, of which the maximum increase percentage is 4.25%, from 77.68 Hz to 80.98 Hz, and the increase in other orders is slight, which are 1.21% and 0.67%, respectively. According to the data analysis and comparison, the drive axle housing not only realizes the light weight but also ensures the performance of the drive axle housing. Compared with the topology optimization results, although the degree of lightweight is slightly poor, it has no impact on the performance of the drive axle housing. The selection of the two optimization schemes should be made according to the needs of designers.