# Optimizing wheel profiles and suspensions for railway vehicles operating on specific lines to reduce wheel wear: a case study

- 154 Downloads

## Abstract

The selection of a wheel profile is a topic of great interest as it can affect running performances and wheel wear, which needs to be determined based on the actual operational line. Most existing studies, however, aim to improve running performances or reduce contact forces/wear/rolling contact fatigue (RCF) on curves with ideal radii, with little attention to the track layout parameters, including curves, superelevation, gauge, and cant, etc. In contrast, with the expansion of urbanization, as well as some unique geographic or economic reasons, more and more railway vehicles shuttle on fixed lines. For these vehicles, the traditional wheel profile designing method may not be the optimal choice. In this sense, this paper presents a novel wheel profile designing method, which combines FaSrtip, wheel material loss function developed by University of Sheffield (USFD function), and Kriging surrogate model (KSM), to reduce wheel wear for these vehicles that primarily operate on fixed lines, for which an Sgnss wagon running on the German Blankenburg–Rübeland railway line is introduced as a case. Besides, regarding the influence of vehicle suspension characteristics on wheel wear, most of the studies have studied the lateral stiffness, longitudinal stiffness, and yaw damper characteristics of suspension systems, since these parameters have an obvious influence on wheel wear. However, there is currently little research on the relationship between the vertical suspension characteristics and wheel wear. Therefore, it is also investigated in this paper, and a suggestion for the arrangement of the vertical primary spring stiffness of the Y25 bogie is given.

## Keywords

Wheel profile Suspension Optimization Wear Specific line## 1 Introduction

The increase in railway vehicle speed, axle load, and traffic volume exacerbates wheel wear, resulting in shorter wheel re-profiling mileage. At present, wheel wear has become one of the most critical issues affecting the operating cost and vehicle-track performance [1, 2]. Reducing wheel wear, therefore, is a topic of big interest.

*Region A*: wheel tread–rail head. The WR contact is typically located in this region and usually occurs when the vehicle is running on straight tracks or curves with large radii. Lowest contact pressures and lateral forces occur in this region, which results in lower wear between the wheel tread and the rail head; (2)

*Region B*: wheel flange–rail gauge corner. The WR contact tends to occur in this region when the vehicle is running on curves with small radii. The contact patch is much smaller than that in region A. This region yields higher contact pressures and sliding velocities, resulting in severe wear between the wheel flange and the rail gauge corner; (3)

*Region C*: contact between field sides of wheel and rail. Contact is least likely to appear in this region. The occurrence of the contact in this region will result in incorrect steering of the wheelset and severe wheel wear. Therefore, letting the WR contact occur primarily in Region A is the optimal solution to reduce wheel wear, in which a reasonable wheel profile plays a key role.

Besides, some studies have shown that the characteristics of vehicle suspensions could also affect wheel wear to some extent [4, 5, 6, 7]. We believe that unreasonable suspension parameters may lead to two mechanisms that increase WR wear: (1) It may affect the WR creepages and sliding velocities, thereby affecting wheel wear; (2) It may also cause WR contact to occur frequently in Region B described in Fig. 1, resulting in severe wear between the wheel flange and the rail gauge corner.

Based on the above considerations, this work aims to reduce wheel wear by optimizing the wheel profile and the vehicle suspension system.

### 1.1 Existing methods

#### 1.1.1 Wheel profile optimization methods

A suitable wheel profile can reduce wheel wear and improve running performance. The optimization of wheel profiles, therefore, has been a meaningful topic since the dawn of railway vehicles. The approaches with different strategies for the development of a new, theoretical wheel profile over the past two decades can be mainly classified into three categories [8, 9]: (1) bio-inspired optimization algorithm; (2) target-based technique; and (3) direct evaluation (wear model).

### Bio-inspired optimization algorithm

The most common bio-inspired algorithm used to optimize wheel wear is the genetic algorithm (GA) [10]. In terms of the single-objective optimization of wheel profiles, Santamaria et al. [11] designed an optimal wheel profile using GA, in which the rolling radius difference function (RRD) was considered as the optimization object. Dynamic simulation results showed that the optimized wheel profile calculated by this method could obtain a lower wear rate and higher running performance. In terms of the multi-objective optimization of wheel profiles, Persson and Iwnicki [12] applied GA to reflect the influence of different profiles on various factors including wear, contact stress, track shift force, derailment quotient, and passenger comfort. Choi et al. [13] applied GA to minimize the flange wear and surface fatigue of a wheel, in which the boundary conditions were given, such as derailment coefficient, lateral WR force, the possibility of overturning and vertical load. Novales et al. [14] used GA to select the optimal wheel profile to balance the derailment coefficient, wear and WR contact stress. Firlik et al. [8] designed a wheel profile using GA for optimizing the wear index, derailment coefficient, and contact area. Another classical bio-inspired algorithm that has been used in wheel profile optimization is the particle swarm optimization algorithm (PSO) [15]. For instance, based on the Non-Uniform Rational B-Spline (NURBS) curve theory, Lin et al. [16] used PSO to design an LM thin flange wheel profile, in which the mean values of wear work and lateral force of wheelset were considered as objective functions, and the profile curve concavity and continuity were set as geometric constraints. The result showed that the average wear work of the first wheelset was reduced by 55.97% compared with the standard LM profile. Cui et al. [17] used PSO to design a new wheel profile for the CRH1 train in China, in which a weighted factor that considered ride comfort and wheel wear was introduced as the objective function, and the derailment-related indexes were considered as the constraints. The result showed that the new wheel profile designed by PSO could significantly reduce the wheel flange/rail gauge corner wear on the premise of ensuring the dynamic behavior of the vehicle.

### Target-based technique

Currently, there are four main kinds of target-based techniques: target RRD, target conicity, target contact angle, and target WR normal gap. Based on the target RRD, Shevtsov et al. [18] proposed a numerical optimization technique to optimize the wheel profile, in which the RRD function was used to characterize the superiority of different wheel profiles and was set as the optimization object. The results showed that a reasonable increase in RRD could improve the curve negotiation ability and decrease the wear rate. In [19], considering both RCF and wear, Shevtsov et al. achieved an optimized wheel profile. The results of the dynamic simulations showed that the use of the optimized profile resulted in a small increase in wear index and a small decrease in surface fatigue. This technique was also studied in [11, 20, 21, 22]. Similar techniques were presented in [9, 23], i.e., target conicity [9] and target contact angle [23], respectively. Concerning the target WR normal gap, Cui et al. [24] thought that a small weighted WR normal gap could make a conformal contact, thereby reducing the contact stress and decreasing the wear and RCF.

### Direct evaluation (wear model)

This approach is used to reduce wheel wear and can visually present the result. Ignesti et al. [25] used FASTSIM and USFD wear function to calculate the material loss under different wheel profiles. In their study, four different kinds of wheel profiles, CD1, DR1, DR2, and S1002, were investigated. The wear evolution of these profiles could be visually presented to help people select the optimal wheel profile. Similar studies were presented in [26, 27].

When designing a wheel profile, several of the listed strategies can be used simultaneously.

#### 1.1.2 Vehicle suspension optimization concerning wheel wear

From the perspective of wheel wear, Fergusson et al. [4] investigated the longitudinal and lateral primary suspension stiffness and the centre plate friction of a self-steering three-piece bogie. A large number of simulation results showed that these three parameters had a great influence on the wear number and dynamics performances including derailment coefficient, angle of attack, WR creepage, etc., in which three curves of different radii (300, 500, and 1000 m) were discussed. Through observation, an optimized combination of these three parameters was found, and it showed that a reasonable combination of these three parameters could reduce the wear number by up to 50%. Mazzola et al. [5] studied the influence of the wheelbase and suspension system on the running safety and wear of a non-powered high-speed car. In their work, the longitudinal and lateral stiffnesses of the primary suspension and the yaw damper coefficient of the secondary system were investigated. The results showed that different combinations of these three parameters had a different influence on the wear index. Finally, the optimal combination was calculated by polynomial interpolation functions. Bideleh et al. [6] explored the influence of the arrangement of the suspension system on ride comfort and wear number. In their work, GA was integrated into MATLAB/SIMPACK co-simulation to optimize the bogie suspension system, in which the ride comfort and wear were considered as objective functions, and the track shift force, stability, and risk of the derailment were set as constraints. The optimization results showed that the asymmetric suspension system revealed remarkable benefits in wear reduction when the vehicle operated on tracks with small radii. In [7] Bideleh investigated the effects of primary and secondary suspension stiffness and damping components on the wheel wear of a railway vehicle model with 50 degrees of freedom (DOFs) based on the multiplicative dimensional reduction method (M-DRM). It was found that the wear was most sensitive to the longitudinal and lateral primary springs. However, for the symmetric vehicle model, as the radius of curvature of the track increased, the effects of the longitudinal and vertical secondary springs became dominant. In the case of large curves and straight tracks, yaw dampers could also significantly affect wear. It was further stated that the sensitivity analysis results obtained in this paper could narrow down the number of the input design parameters for optimization problems of bogie suspension components and improve the computational efficiency. Ashtiani [28] studied the dynamic characteristics of the friction wedge geometry in the secondary suspension of a freight wagon with three-piece bogies. An optimization formulation was proposed to improve the performance of the wagon in terms of minimizing the extreme normal WR contact and vertical carbody acceleration. The optimal wedge geometry is then evaluated in terms of critical hunting speed and curving performance of wagon. The beveled geometry of the wedge with optimal angles presents an improvement in higher critical hunting speed and lower derailment coefficient. Although the relationship between the wedge geometry and wheel wear was not directly discussed in this article, it showed that the wedge geometry could affect the WR force, thus indicating that the wedge geometry may also have an effect on wheel wear. A similar study was presented in [29].

### 1.2 Motivation

Indeed, the aforementioned methods in Sect. 1.1.1 have great potential for wheel profile optimization if the models are accurately established and the strategies are correctly formulated. Most of these studies, however, have aimed at improving the running performance and/or reducing contact force/wear/RCF on small-section curves with ideal radii, losing the big picture of whole route [30]. However, the whole route usually consists of a lot of sections with different track layout parameters, such as superelevation, gauge, cant, and arc. Furthermore, some studies have shown that the track layout parameters have a significant influence on wheel wear. For example, Pombo et al. [31] investigated the influence of rail cant on wheel wear growth, in which the wheel profile S1002 and the rail profile UIC60 were used. The results revealed that the reprofiling intervals obtained when running on the track with a rail cant of 1/40 were larger than when traveling on a track with a rail cant of 1/20. Gao et al. [32] studied the superelevation setting for a 400-meter-radius curve of the China Shen-shuo railway line based on Hertzian theory and FASTSIM algorithm. The simulation results showed that the superelevation had a great influence on wheel wear. Therefore, it is necessary to consider the actual track layout parameters of the whole route when developing wheel profiles.

More importantly, with the expansion of urbanization, more and more railway vehicles shuttle on special lines, such as metro, light rail, and tram. In addition, due to some unique geographic or economic reasons, some vehicles typically operate on a designated line. For example, some CRH1A and CRH380A trains only run on the China Hainan Roundabout railway line because of the unique island geography of Hainan province. Some CRH380A trains mainly run on China Shanghai–Hangzhou special passenger line because of the heavy transportation tasks [33]. Furthermore, some vehicles that undertake special tasks often run on fixed lines, such as those coal wagons on the China Datong–Qinhuangdao railway line [34], the WLE beer wagons on German Warstein–München Riem railway line [35], and the lime wagons running on German Blankenburg–Rübeland railway line introduced in this paper [36]. For the aforementioned vehicles running on fixed lines, the traditional wheel profile or the profile designed by the approaches described in Sect. 1.1.1 may not be the optimal choice. In this sense, a novel wheel profile designing method, which considers the actual track layout parameters, for reducing wheel wear for these vehicles operating on fixes lines is introduced in this paper.

Regarding the influence of suspension optimization on wheel wear, the existing researches described in Sect. 1.1.2 have studied the lateral stiffness, longitudinal stiffness, and yaw damper characteristics of suspension systems, since these parameters have an obvious influence on wheel wear. However, as far as the authors know, there is currently little research on the relationship between the vertical suspension characteristics and wheel wear. Therefore, it is also investigated in this paper.

### 1.3 Proposed method

The basic idea of this paper is first to calculate the wheel wear under different initial wheel profiles and different vertical spring characteristics of the primary suspension of the Y25 bogie by the wear model and then establish the response model between these three parameters through the KSM. Finally, the optimal wheel profile is found and the relationship between vertical spring characteristics of the primary suspension and wheel wear is investigated based on KSM technique.

#### 1.3.1 A FaStrip-USFD based wheel wear model

The wheel wear calculation model is made up of two submodels: WR local contact model and wheel material loss model [37, 38].

The WR local contact model consists of a WR normal contact model and a WR tangential contact model [39]. In our work, the Hertzian contact model [40] is applied as the WR normal contact model since it can balance the calculation efficiency and accuracy [41]. Concerning the WR tangential contact model, it includes Shen–Hedrick–Elkins (S.H.E) [42], Polach [43], FASTSIM [44], etc. Among them, The S.H.E theory and the Polach method cannot obtain the shear stress distribution and sliding velocity of the contact patch, which are not suitable for wheel wear calculation. The FASTSIM cannot ideally solve the shear stress distribution and stick–slip division issues, which may cause significant errors in wear prediction [45]. Recently, a novel WR local contact model, FaStrip [45], has been proposed. This method is based on Strip theory, Kalker linear theory, and FASTSIM, and can obtain high accuracy on both shear stresses and relative slip velocities. Therefore, it is applied in this paper.

Archard model [49, 50], where the material loss (\(V_{w}\)) is proportional to the normal force (\(N\)) and the sliding distance (\(S\)) divided by the material hardness (\(H\)), i.e., \(V_{w} = {kNS} / {H}\).

\(T - \gamma \) [51], which assumes that the material loss (\(T\gamma \)) is proportional to the frictional energy dissipated in the contact patch. It is expressed as the sum of the products of creep forces and creepages for the lateral, longitudinal, and spin components, i.e., \(T\gamma = T_{x} \gamma _{x} + T_{y} \gamma _{y} + M_{z} \omega _{z}\). USFD function [52], as a kind of \(T \mbox{--} \gamma \) method, is applied in this paper.

In this work, the FaStrip algorithm and USFD function are combined to calculate the wheel material loss. This method is called as FaStrip-USFD.

#### 1.3.2 KSM

Simulations involving parametric studies are often based on repetitive modeling, which greatly increases the calculation amount. For example, in the case of the vehicle multibody dynamics simulation (MBS) model built in this research, running the simulation takes more than 600 CPU hours. The calculation amount of such simulations is so large that the railway industry is reluctant to do many deterministic analyses. In addition, the test process involves steps such as numerical calculation, analysis, etc., involving a large number of uncertainties. It not only consumes much human labor but also may increase errors [53]. Therefore, it is of interest to find a high-efficiency and reliable method to simplify the simulation procedure.

Recently, the KSM technique [54] has begun to be used in railway engineering[55], since it can overcome the above two shortages. This technique uses a small number of sample points that fit a specific sampling strategy to construct a simplified mathematical model that approximates the original complex model. Therefore, it can replace the original analytical model and simplify the calculation process while maintaining high calculation accuracy. In our work, this technique [56] is introduced to build the relationship between the wear area, wheel profile adjustment factor and vertical outer primary stiffness, since it is an unbiased estimation model that fully considers the spatial correlation of variables.

### 1.4 Contribution and structure of this paper

- (1)
For those vehicles that mainly operate on fixed lines, a FaStrip-USFD-KSM based model for optimizing the wheel profile and vehicle suspension to reduce wheel wear is proposed, where the German Blankenburg–Rübeland railway line with detailed track layout parameters and an Sgnss wagon are established.

- (2)
Based on the S1002 profile, an adjustment factor is proposed to optimize the wheel profile. This method simplifies complex curve design problems by using classical transition curves.

- (3)
The influence of the vertical primary suspension characteristics on wheel wear is studied, and a suggestion for the arrangement of the vertical primary spring stiffnesses of the Y25 bogie is given.

The rest of this paper is structured as follows. In Sect. 2 the German Blankenburg–Rübeland railway line and an Sgnss wagon are modeled, where the detailed track layout parameters are considered. In Sect. 3 the theories of the FaStrip, USFD wear function, and KSM are briefly described, and a FaStrip-USFD-KSM based optimization method for reducing wheel wear is proposed. In Sect. 4 a case study of the Sgnss wagon running on the Rübeland–Blankenburg line is presented. In Sect. 5 quasi-static and dynamic tests for acceptance of the optimized wheel profile according to the standard EN 14363 are presented. Discussion and conclusions are briefly drawn in Sect. 6.

## 2 Operational line and vehicle model

### 2.1 Operational line

The experiment data (radius, superelevation, rail cant, gauge distance, vehicle speed) used in this paper were supported by the DynoTRAIN project, which was funded by the European Commission. This project started on June 1, 2009, and ended on September 30, 2013. The authors’ group, Fachgebiet Schienenfahrzeuge, TU Berlin, was one of the partners in this project. More information about this project can be found in [58].

- (1)
Lubricants significantly affect the wheel wear [62], in our work it is not considered since the freight wagon we used was not equipped with the lubrication device, and the WR friction coefficient used in the whole simulation is 0.35.

- (2)
In the subsequent wheel wear calculation, the used vehicle speed is corrected based on the actual test speed (Fig. 3(e)), which simply considers the influence of the train’s traction and braking on wheel wear, rather than a comprehensive consideration. This will affect the calculation accuracy to some extent, but fortunately, the braking and traction distance is not long relative to the total mileage.

More information concerning the speed can be found in [36, 60].

### 2.2 Vehicle model

The vehicle model introduced here is an Sgnss wagon, which is modeled in SIMPCAK 2020.2. The MBS model is made up of one car body, two bogie frames, four wheelsets, and eight axleboxes. It has 55 DOFs in total. Since one of the purposes of our work is to study the effects of vertical suspension characteristics on wheel wear, it is important to accurately establish the suspension system.

Date of primary suspension

Parameter | Outer spring | Inner spring |
---|---|---|

Vertical stiffness (kN/m) | \(k_{\mathrm{co}} = 498\) | \(k_{\mathrm{ci}} = 808\) |

Lateral stiffness (kN/m) | 240 | 240 |

Rest length (m) | 0.26 | 0.234 |

Stop length (m) | 0.1425 | 0.1674 |

There is no real secondary suspension on the Y25 bogie. The secondary suspension system consists of a spherical center pivot, only allowing three rotational DOFs, and two side bearers (Fig. 4(b)). The center pivot is modeled using a constraint element for preventing translational motions between wagon body and bogie frame as well as four frictional force elements. The side bearers are mounted outboard of the bogie pivot, providing friction damping in yaw and restraint in roll. They each consist of horizontal friction plates mounted on twin vertical coil springs. Each side bearer is modeled as a mass element and two force elements. One connects the bogie frame and the mass element, and has the summed stiffness of the two coil springs, taking into account the bump stop, where the characteristics of the vertical, longitudinal and lateral forces are shown in Figs. 5(d), (e), and (f), respectively. The other one represents the frictional force between the wagon body and the side bearer, whose normal force stems from a constraint element between the wagon body and the mass element. Moreover, the accuracy of these parameters was verified by using the actual test data [53, 64].

In order to make the simulation results closer to reality, the track irregularities are also considered. In this model, the PSD of the typical European spectrum (ERRI B176) defined in SIMPACK is applied as the track irregularities. The track is modeled using a discrete model including a tie with three DOFs (lateral, vertical and roll) placed under each wheelset. More information concerning the MBS model, including the modeling steps and the detailed parameters, can be found in the authors’ previous work [65, 66].

Primary parameters of the lime wagon

Parameter | Value | Unit |
---|---|---|

Vehicle frame mass (\(m_{v}\)) | 70 000 (load), 10 600 (unload) | kg |

Bogie frame mass (\(m_{b}\)) | 1887 | kg |

Axlebox mass (\(m_{a} \)) | 152 | kg |

Wheelset mass (\(m_{w}\)) | 1121 | kg |

Sleeper mass (\(m_{s}\)) | 330 | kg |

Carbody roll moment of inertia (\(I_{\mathit{cxx}}\)) | 7399 (load), 4327 (unload) | kg m |

Carbody pitch moment of inertia (\(I_{\mathit{cyy}}\)) | 1 998 266 (load), 299 140 (unload) | kg m |

Carbody yaw moment of inertia (\(I_{\mathit{czz}}\)) | 2 418 154 (load), 366177 (unload) | kg m |

Bogie frame roll moment of inertia (\(I_{\mathit{bxx}}\)) | 1188 | kg m |

Bogie frame pitch moment of inertia (\(I_{\mathit{byy}}\)) | 1484 | kg m |

Bogie frame yaw moment of inertia (\(I_{\mathit{bzz}}\)) | 2582 | kg m |

Axlebox roll moment of inertia (\(I_{\mathit{wxx}}\)) | 2.18 | kg m |

Axlebox pitch moment of inertia (\(I_{\mathit{wyy}}\)) | 7.57 | kg m |

Axlebox yaw moment of inertia (\(I_{\mathit{wzz}}\)) | 6.71 | kg m |

Wheelset roll moment of inertia (\(I_{\mathit{wxx}}\)) | 592.9 | kg m |

Wheelset pitch moment of inertia (\(I_{\mathit{wyy}}\)) | 96.34 | kg m |

Wheelset yaw moment of inertia (\(I_{\mathit{wzz}}\)) | 592.9 | kg m |

Pivot distance (\(2 a^{*}\)) | 14.2 | m |

Wheelset base (\(2 a^{+}\)) | 1.8 | m |

Wheel rolling circle diameter ( | 920 | mm |

Contact damping (\(c_{c}\)) | 10 | kNs/m |

Secondary suspension damping (\(c_{sx}\), \(c_{sy}\) and \(c_{sz}\)) | 10 | kNs/m |

Sleeper-ballast stiffness (\(k_{t}\)) | 150,000 | kN/m |

The friction coefficient between wheels and rails ( | 0.35 | – |

Poisson ratio ( | 0.28 | – |

Wheel profile | S1002 | – |

Rail profile | UIC60e2 | – |

## 3 FaStrip-USFD-KSM based optimization method

### 3.1 Theories of methods

#### 3.1.1 FaStrip-USFD

Currently, the most common wheel profile calculation method is based on the WR local contact model and wheel material loss model [67, 68, 69, 70, 71, 72, 73], in which, the WR local contact model consists of a WR normal contact model and a WR tangential contact model. The WR normal contact model originated from the Hertzian contact theory, i.e., Hertzian contact model, which is currently the most widely used in the WR normal contact analysis [74]. However, in the actual operation of the train, the WR contact may be a conformal contact and/or a non-Hertzian contact. The approaches for solving conformal WR contact problems include the finite element (FEM) method [75, 76], the CONTACT’s boundary element approach together with the numerical influence coefficients [77], a computer program called WEAR [78], etc. The approaches for solving non-Hertzian WR contact problems include the FEM, Kalker’s variational method [79], Linder method [80], Kik–Piotrowski (KP) model [81], Extended Kik–Piotrowski (EKP) model [82], Modified Kik–Piotrowski (MKP) model [83], Ayasse–Chollet model (STRIPES) [84], ANALYN [85], etc. These listed methods are generally more suitable for describing WR contacts than Hertzian contact model. However, these methods have a much higher computational effort than Hertzian contacts. Besides, in [41], three non-Hertzian contact models, namely KPM, STRIPES, and ANALYN, were compared to Hertzian contact model and the CONTACT code in terms of the normal contact solution, and the tangential contact solutions and wheel wear material loss were calculated by FASTSIM and USFD wear function, respectively. The results indicated that using Hertzian contact model to solve the WR normal contact problem in wheel wear simulation was a good choice from a compromise between the calculation efficiency and accuracy. Hertzian contact model, therefore, is applied to solve the WR normal contact in our work. More information on non-Hertzian contact models can be found in [86, 87] or the author’s previous work [83].

Concerning the WR tangential contact model, classical WR tangential contact models include S.H.E. [42], Polach [43], FASTSIM [44], etc. Among them, the S.H.E. theory and the Polach method cannot obtain the shear stress distribution and sliding velocity of the contact patch, which are not suitable for the wheel wear calculation. FASTSIM, currently, is the most widely used. In this theory, the shear stress in the stick area is assumed to increase linearly from the leading edge until the traction bound is reached. The traction bound is assumed as the product of the contact pressure and the friction coefficient. Although FASTSIM utilizes the Hertzian solution for normal contact solution, which yields an elliptical pressure distribution, the traction bound in FASTSIM is taken to be parabolic. Despite the reasonable error margin for creep force calculation, the combination of linear stress growth in the stick area and parabolic distribution in the slip area results in a considerable error in shear stress distribution. Kalker stated that FASTSIM is accurate up to 5% in a pure creepage case and up to 10% for a pure spin case. Errors up to 20% may be shown for combined lateral creepage and spin [88]. The error levels of FASTSIM stated by Kalker are based on the studies of circular contact. However, further studies show that these error levels do not hold for all possible contact ellipses. For instance, for contact ellipses that are narrow in the rolling direction, the errors are higher. In the case of pure longitudinal and lateral creepages, the error levels are about 5% for circular contact while it reaches 12% for elliptic contact with a semi-axes ratio of 0.2. In the combined lateral creepage and spin case, where the spin effect on creep force is opposing the one from lateral creepage, the error reaches above 25% for large spin values [89]. As stated in [90], these assumptions-induced errors may cause significant errors in wear prediction.

To improve the above shortages, FaStrip [45] has been recently proposed. It is based on Strip theory [91], Kalker linear theory, and FASTSIM. The original strip theory is only accurate for contact ellipses that resemble the rectangular plain strain contacts. In FaStrip, this original strip theory is improved and can achieve accurate estimations for all contact cases. Furthermore, it is combined with a numerical algorithm, like FASTSIM, to handle spin. This method can achieve higher accuracy on both creepages and shear stresses. This method has begun to be used to wear calculation and verification. For instance, in [89], this method was used to calculate the wheel wear of a heavy-haul locomotive, and the results showed that there was quite a good agreement between the measured and the FaStrip-based simulated results for the normal operational cases up to around 100,000 km. In [68], this method is verified by tracking test data of a CRH3 train running on China Wuhang–Guangzhou railway line. This method is also applied in our work. Concerning the detailed information of the FaStrip can be found in [45] or the authors’ previous work [92].

In FaStrip, the discretization of the contact patch has a large influence on the wear distribution. Sichani et al. [45] showed that the contact patch with 110 × 55 meshes could yield an error of less than 5% in the estimation of creepages. In FASTSIM, Polach suggested that the contact patch with 10 × 10 meshes could yield reasonably accurate predictions of contact force distribution [43]. Overall, the accuracy of prediction increases as the number of grids increases, and a dense discretization can provide a continuous wear distribution [93]. However, a dense discretization will increase the calculation amount. As a compromise between numerical efficiency and accuracy, the contact patch is divided into 50 × 50 meshes in our work.

#### 3.1.2 KSM

- (1)
Generally, the more sample points are selected, the higher the accuracy. However, when the number of samples reaches a certain level, increasing the sample points will increase the calculation amount instead of improving the accuracy of the model.

- (2)
It is necessary to ensure that the sample points can uniformly fill the whole space, thereby reducing the phenomenon that the local cannot be fitted.

Currently, the most widely used sample selection methods contain orthogonal experimental design [94]. Latin hypercube sampling (LHS) [95], uniform experimental design [96], etc. To facilitate the generation of the wheel profile curve, the uniform experimental design sampling principle is used in our work.

### 3.2 General architecture

- (1)
The rails are not subjected to wear and the profiles are kept to be constant.

- (2)
Discrete strategy to wear evolution, by dividing the whole simulation into a series of discrete steps and updating the profile after each step. In this paper, the wear depth-based updating strategy is used, and the threshold is set to \(\Delta d=0.01\text{ mm}\) referring to [92].

- (3)
The wheel profile is updated after each discrete simulation. A smoothing strategy is used to eliminate short-wavelength concavities, where a moving average filter with a window width equal to 5% of the wear band. The wheel profile used in the simulation is defined from −65 to 60 mm. This band is divided into 1201 nodes.

**Step 1**:Parameter settings: the corresponding wheel profiles and the suspension parameters are set.

**Step 2**:Run the short-term simulation in SIMPACK and generate the global contact parameters including normal force \(N\), global creepages (\(v_{x}\), \(v_{y}\)), spin (\(\varphi \)), half-length of the contact patch (\(a_{0}\), \(b_{0}\)), and contact position.

**Step 3**:Calculate local tangential stress \(p(x,y)\) and local creepage \(\gamma (x,y)\) within the contact patch using FaStrip in MATLAB.

**Step 4**:Calculate the wear distribution using the USFD function in MATLAB.

**Step 5**:Smooth the wear depth curve and feed the new wheel profile back to the short-term MBS model in

**Step 2**to implement the next iteration.**Step 6**:The total wear area of the eight wheels is calculated, and the relationship between the adjustment factor, vertical outer primary stiffness, and wear area is built by the KSM technique in MATLAB.

## 4 A case study for the German Blankenburg–Rübeland line

The selection of a wheel profile is a topic of big interest as it can affect running performances and wheel wear, which needs to be determined based on vehicle type, line, and usage. Practically, most of the selection methods of the wheel profile currently are determined according to the vehicle type, which may be due to the convenience of manufacturing. For example, in China, currently, LM_A profiles are widely used for CRH1, CRH2, CRH380A(L), CR400AF and CR400BF EMUs, LM_B (i.e., S1002CN) for CRH3, CRH380B(L) and CRH380D EMUs, LM_B-10 for CR400AF and CR400BF EMUs, and LM_C (i.e., XP55) for CRH5 EMUs [101]. In Europe, the most widely used wheel profile is the S1002 [102], including the case introduced in this work. However, the consideration of the line mainly stays on the ideal line, such as the ideal curves in the literature mentioned in Sect. 1.1.1. For the consideration of specific routes, as far as the authors know, there are few other studies besides [36].

In view of the above problem, this paper takes the Blankenburg–Rübeland line as an example to optimize the S1002 profile for reducing wheel wear. Besides, the influence of the vertical primary suspension characteristics of the Y25 bogie on wheel wear is also investigated.

### 4.1 Optimization principle and boundary condition

In the design of wheel profiles, the setting of the curve is a complicated problem. In order to avoid this problem, this paper introduces an adjustment factor \(a\) to optimize the traditional classical profile. Since S1002 is a classic and widely used profile, and the original profile of the introduced vehicle is also this profile. Therefore, it is introduced as a template. The specific optimization principles and boundary conditions are as follows:

### For the wheel profile

- (1)The tread base point (nominal circle contact point) is set as the origin of the profile coordinate, and the \(Z\) coordinate is multiplied by \(a\), as shown in Fig. 8. For convenience, the newly generated profile is called S1002A’XXX’, where XXX represents the adjustment factor multiplied by 100, such as S1002A094, S1002A105 illustrated in Fig. 8.
- (2)
Changing \(a\) will change the equivalent conicity (or RRD) of the wheelset. The increase of \(a\) will lead to an increase in the equivalent conicity, thus reducing the running stability of the vehicle. On the contrary, a small \(a\) results in a decrease in the equivalent conicity and an increase in lateral displacement of the wheel, resulting in the contact patch often appearing in region B (Fig. 1), which also increases the wear of wheel flange and gauge corner. In addition, a too small \(a\) will reduce the flange height and increase the risk of derailment. Based on the above considerations and the author’s practical experience, in our work, \(a\) is set between 0.94–1.05. This fine-tuning also allows the adaptability of the vehicle to operate on other lines.

### For the vertical primary suspension stiffness

- (1)
In order to ensure the load-bearing capacity of the vehicle, the total stiffness of the outer spring (\(k_{\mathrm{co}}\)) and the inner spring (\(k_{\mathrm{ci}}\)) is guaranteed to be constant, equal to the original total stiffness (\(k_{\mathrm{co}} + k_{\mathrm{ci}} = 1306~\mbox{kN}/\mbox{m}\)).

- (2)
The outer spring stiffness cannot be removed since it needs to be used to carry the sprung mass at an empty load. It is set between 298–798 kN/m.

### 4.2 Simulation result

One point that should be noted here is that the vehicle travels back and forth and is not turned at the end station. This means that the order of the wheelsets on the return trip is opposite to the order of the wheelsets on the forward trip. The wear distribution should also be arranged in reverse order. In this paper, the calculation amount is extremely large. Referring to [50], in order to shorten the simulation time, only the forward trip is considered, i.e., the wagon runs a one-way trip on the Blankenburg–Rübeland line, with a journey of 15.73 km. Finally, the wheel wear is calculated by the FaStrip-USFD method.

Total wear area under different adjustment factors and different vertical outer spring stiffness

Adjustment factor | Total wear area (mm | |||||
---|---|---|---|---|---|---|

298 kN/m | 398 kN/m | 498 kN/m | 598 kN/m | 698 kN/m | 798 kN/m | |

0.94 | 1.4406 | 1.5786 | 1.7526 | 2.0056 | 2.2242 | 2.3944 |

0.95 | 1.2646 | 1.4870 | 1.6844 | 1.9512 | 2.1896 | 2.3752 |

0.96 | 1.0192 | 1.1294 | 1.2296 | 1.3426 | 1.4524 | 1.5308 |

0.97 | 0.8390 | 0.9398 | 1.0714 | 1.2346 | 1.3560 | 1.4336 |

0.98 | 0.8192 | 0.9132 | 1.0328 | 1.1680 | 1.2620 | 1.3036 |

0.99 | 1.1640 | 1.2464 | 1.3666 | 1.5704 | 1.8038 | 2.0634 |

1.00 | 1.1661 | 1.2478 | 1.3801 | 1.5900 | 1.8246 | 2.1116 |

1.01 | 0.7444 | 0.7870 | 0.8462 | 0.9360 | 1.0648 | 1.1726 |

1.02 | 0.6240 | 0.6860 | 0.7450 | 0.8206 | 0.9154 | 1.0034 |

1.03 | 0.5726 | 0.6392 | 0.7122 | 0.7688 | 0.8496 | 0.9476 |

1.04 | 0.6352 | 0.6898 | 0.7876 | 0.8592 | 1.0050 | 1.1144 |

1.05 | 1.0430 | 1.0922 | 1.1870 | 1.3061 | 1.4906 | 1.7410 |

### 4.3 Model verification

Comparison of the calculated results and simulated results

Adjustment factor | Vertical outer spring stiffness (kN/m) | Wear area calculated by FaStrip-USFD (mm | Wear area calculated by FaStrip-USFD-KSM (mm | Error (%) |
---|---|---|---|---|

0.970 | 498.0 | 1.136 | 1.072 | −5.55 |

0.981 | 540.4 | 1.095 | 1.106 | 1.00 |

1.026 | 528.8 | 0.802 | 0.772 | −3.74 |

1.036 | 646.5 | 0.823 | 0.802 | −2.55 |

1.042 | 353.6 | 0.682 | 0.701 | 2.79 |

Generally, intelligent algorithms need to be applied for automatic optimization. However, the model established in this paper clearly shows that \(a \approx 1.03\) (S1002A103 in Fig. 10(b)) is the optimal choice (the red line in Fig. 11), with minimal wheel wear. Therefore, the optimization calculation is omitted in this paper. However, this step is necessary in the case of high-dimensional nonlinear relationships. How to use the particle swarm optimization (PSO) algorithm for optimization calculations can be found in the authors’ previous work [53].

### 4.4 Analysis of result

- (1)
The profile adjustment factor (\(a\)) has a large influence on the wheel wear and there is no definite trend, presenting a nonlinear relationship. As shown in Fig. 10(a), when \(a=1\), the flange wear is very serious, especially the flange face wear; when \(a\) is increased to 1.03, the flange wear is greatly reduced (Fig. 10(b)); and when \(a\) continues to increase to 1.05, the flange wear is serious again (Fig. 10(c)). The most obvious change is located on the flange face (the blue rectangles in Fig. 10). The reason for this phenomenon is that as \(a\) increases, the equivalent conicity of the wheelset increases, and when the vehicle is running on a curve, the WR contact patch tends to approach or occurs in region A described in Fig. 1. When running on a small-radius curve, the flange wear is reduced. However, if the equivalent conicity is too large, the lateral displacement of the wheelset will increase, and the area of the WR contact patch will decrease, and the spin-induced sliding velocity will also increase, thereby increasing the flange wear. More importantly, in reality, there are many nonlinear variables on the track (track layout parameters), such as arc curves, transition curves, superelevation, gauge, cant, etc., as well as vehicle speed. The position of the contact patch and the distribution of wheel wear will change with different combinations of these parameters. Finally, under the combined influence of these nonlinear factors, the relationship between the wear amount and \(a\) exhibits a nonlinear relationship. By modeling the actual line, a reasonable wheel profile for reducing wheel wear can be found. In the present work, an S1002A103 profile (Fig. 10(b)) is proposed for those Sgnss wagons running on Blankenburg–Rübeland line. Simulation results show that this profile can reduce flange face wear, and the total wear can be reduced by more than 50% in short-term running.

- (2)The vertical primary suspension characteristics have an influence on the wheel wear amount, which increases with the increase of the outer spring stiffness (Fig. 12). The reason for this phenomenon is that the vertical suspension characteristics can affect the angle of attack, thus affecting the WR contact properties and further affect the wear amount. To illustrate this phenomenon, a section, which consists of a straight line, a transition curve, and a circular curve, with a track gauge of 1.450 m and a cant of 1:40, is introduced here. The curvature and superelevation distributions are shown in Figs. 13(a) and 13(b), respectively. Then, three vertical primary stiffness combinations are set on the Sgnss wagon with a speed of 51 km/h to implement simulations, respectively. Since the first and third wheelsets are the guiding wheelsets, the angle of attack of these two wheelsets is shown in Fig. 13(c). It can be seen that when the outer spring stiffness is reduced (meanwhile, the inner spring stiffness is correspondingly increased), the wheelset attack angle on the transition curve and the circular curve is also reduced. The wear amount, therefore, will correspondingly reduce (Fig. 12). Based on this phenomenon, we recommend appropriately reducing the outer spring stiffness of the Y25 bogie while increasing the inner spring stiffness to reduce wheel wear.

### 4.5 Long-term wear comparison between S1002A103 and S1002 profiles

- (1)
The wheel flange wear is greatly reduced;

- (2)
The wheel tread wear is more uniform.

In terms of \(S_{h}\), it reaches the limit value faster when using the improved wheel profile S1002A103. There are two reasons for this phenomenon: (1) S1002A103 is obtained by directly multiplying the adjustment factor \(a=1.03\) (Fig. 8), which increases the flange height. (2) Although the improved profile greatly reduces the flange wear and makes the tread wear more uniform, it also increases the wear depth at the nominal rolling circle to some extent.

In terms of \(S_{d}\) and \(q_{R}\), the use of the improved profile has significant advantages since the flange wear is significantly reduced, which makes these two values change slowly.

Considering that during the re-profiling process, severe flange wear often leads to deeper turning depth. The S1002A103 wheel profile, therefore, is more suitable for the case of this work than the standard S1002 wheel profile from the perspective of reducing wheel wear and extending wheel service life.

## 5 Quasi–static and dynamic tests for the optimized wheel profile according to EN 14363

Considering the allocation of dispatching or transportation tasks, the vehicles sometimes may be transferred to other railway lines, and thus the adaptability of the vehicle needs to be guaranteed. Therefore, the derailment and running safety of the optimized wheel profile should satisfy the requirements specified in EN14363 [98]. In this section, the quasi-static safety against derailment on twisted track and the running safety on straight track are presented.

### 5.1 Quasi-static safety against derailment on twisted track

a straight line of 6 m

a 100 m long transition curve with the end radius of 150 m

a 430 m long full arch with a radius of 150 m

a straight line of 30 m

Wheelbase distance \(2 a^{+} \leq 4\ \text{m}\): \(g_{\lim }^{+} =7\permil\)

Pivot distance \(4~\mbox{m} \leq 2 a^{*} \leq 20\ \text{mm}\): \(g_{\lim }^{*} = {20} / {2 a^{*}} +2=4.5\permil\)

Bogie: \(h^{+} = ( g^{+} - g^{0} ) 2 a ^{+} = ( 7\permil -3\permil ) \times 1.8\ \text{m}=7.2\ \text{mm}\)

Carbody: \(h^{*} = ( g^{*} - g^{0} ) 2 a ^{+} = ( 4.5\permil -3\permil ) \times 14.2\ \text{m}=21.3\ \text{mm}\)

For the bogie twisting test: \(d^{+} = {h^{+} b^{+}} / {(2 b_{A} )} = {7.2} / {1.45} =4.97\ \text{mm}\)

For the car body twisting test: \(d^{*} = {h^{*} b ^{*}} / {(2 b_{A} )} = {21.3\times 0.85} / {1.45} =12.49\ \text{mm}\)

The gaskets with thicknesses \(d^{+}\) and \(d^{*}\) are added separately under the primary and secondary suspensions, see Fig. 16(c). The used Sgnss wagon does not have secondary suspensions. Therefore, all the gaskets are added under the corresponding primary spring. The vehicle speed is set as 5 km/h.

### 5.2 Running safety—stability

Simulation results of RMS values of the track shifting force \(\sum Y_{\mathrm{RMS}}\)

Profile type | 132 km/h | 110 km/h | 91 km/h | Limit \(\sum Y_{\mathrm{RMS}\,\mathrm{lim}}\) |
---|---|---|---|---|

S1002A103 | 14.56 kN | 9.22 kN | 6.03 kN | 11.23 kN |

S1002A100 | 13.03 kN | 8.42 kN | 5.48 kN | 11.23 kN |

All in all, the quasi-static and dynamic test results show that the S1002A103 profile meets the criteria of EN14363. Tests for the suspension characteristics are omitted here. More information concerning the test can be found in the authors’ previous work [100].

## 6 Conclusion and discussion

- (1)
For those vehicles that mainly operate on fixed lines, a FaStrip-USFD-KSM based model for optimizing the wheel profile and vehicle suspension to reduce wheel wear is proposed. This method considers the track layout parameters, such as arc curves, transition curves, superelevation, gauges, and cant. In our work, an Sgnss wagon running on the German Blankenburg–Rübeland line is introduced as a case study, and an S1002A103 wheel profile is presented. Simulation results show that this profile can significantly reduce the flange face wear.

- (2)
Considering the allocation of dispatching or transportation tasks, the vehicles that primarily run on special lines, sometimes, may be transferred to other railway lines, and thus the adaptability of the vehicle needs to be guaranteed. The result shows that the S1002A103 profile satisfies the derailment and running safety requirements specified in EN14363.

- (3)
Based on the S1002 profile, an adjustment factor is proposed to optimize the wheel profile. This method simplifies complex curve design problems by using classical transition curves.

- (1)
The vertical primary suspension characteristics have an influence on the amount of wear, which increases with the increase of the outer spring stiffness. Based on this conclusion, we recommend appropriately reducing the outer spring stiffness of the Y25 bogie while increasing the inner spring stiffness to reduce wheel wear.

- (1)
Further field experiment verification is required.

- (2)
In this paper, the optimization of the wheel profile is to multiply the \(Z\)-axis of the wheel profile by an adjustment factor, which increases the flange height. The wheel profile generation method, as well as the mechanism analysis, will continue to be improved in the follow-up work.

- (3)
Vertical primary suspension characteristics have an influence on the amount of wear. The reason for this phenomenon is that the vertical suspension characteristics can affect the angle of attack, thus affecting the WR contact properties and further affect the wear amount. The authors will conduct an in-depth analysis on this part in future work, and the influence of the other suspension parameters on wheel wear will also be investigated.

## Notes

### Acknowledgements

Open Access funding provided by Projekt DEAL. This study is supported by the Assets4Rail project which is funded by the Shift2Rail Joint Undertaking under the EU’s H2020 program (Grant No.: 826250), and part of the experiment data concerning the railway line and the Sgnss wagon used in this paper is supported by the DynoTRAIN project, funded by European Commission (Grant No.: 234079). The first author is also supported by China Scholarship Council (Grant No.: 201707000113). The authors would like to thank Dr. Henning Schelle for the collection of the data. The authors would like to thank Dr. Boyang An from Southwest Jiaotong University for his valuable advice on WR contact calculation.

### Declaration of interests

The authors declare that they have no conflicts of interest.

### Replication of results

In this work, the 3D geometry of the bogie frame and axlebox was modeled in INVENTOR, noting that these geometrical appearances did not affect the dynamics calculations. The track model and whole vehicle model, as well as the vehicle dynamics analysis, were performed in SIMPACK. The wheel wear simulation was performed by SIMPACK/MATLAB co-simulation. The KSM was performed in MATLAB. The unlisted data about the track is confidential and the authors have no right to provide it. However, readers interested in the internal reports (in German) listed in this article, as well as the MATLAB codes, are encouraged to contact the corresponding authors by e-mail.

## References

- 1.Pombo, J., Ambrósio, J., Pereira, M., Lewis, R., Dwyer-Joyce, R., Ariaudo, C., et al.: Development of a wear prediction tool for steel railway wheels using three alternative wear functions. Wear
**271**, 238–245 (2011). https://doi.org/10.1016/j.wear.2010.10.072 CrossRefGoogle Scholar - 2.Lewis, R., Olofsson, U.: Wheel–Rail Interface Handbook. CRC/Taylor & Francis, Boca Raton (2009) CrossRefGoogle Scholar
- 3.Iwnicki, S.: Handbook of Railway Vehicle Dynamics. CRC/Taylor & Francis, Boca Raton (2006) CrossRefGoogle Scholar
- 4.Fergusson, S.N., Fröhling, R.D., Klopper, H.: Minimising wheel wear by optimizing the primary suspension stiffness and centre plate friction of self-steering bogies. Veh. Syst. Dyn.
**46**(S1), 457–468 (2008). https://doi.org/10.1080/00423110801993094 CrossRefGoogle Scholar - 5.Mazzola, L., Alfi, S., Bruni, S.: A method to optimise stability and wheel wear in railway bogies. Int. J. Railw.
**3**(3), 95–105 (2010) Google Scholar - 6.Bideleh, S.M.M., Berbyuk, V., Persson, R.: Wear/comfort Pareto optimisation of bogie suspension. Veh. Syst. Dyn.
**54**(8), 1053–1076 (2016). https://doi.org/10.1080/00423114.2016.1180405 CrossRefGoogle Scholar - 7.Bideleh, S.M.M., Berbyuk, V.: Global sensitivity analysis of bogie dynamics with respect to suspension components. Multibody Syst. Dyn.
**37**, 145–174 (2016). https://doi.org/10.1007/s11044-015-9497-0 CrossRefGoogle Scholar - 8.Firlik, B., Staśkiewicz, T., Jaśkowski, W., Wittenbeck, L.: Optimisation of a tram wheel profile using a biologically inspired algorithm. Wear
**430–431**, 12–24 (2019). https://doi.org/10.1016/j.wear.2019.04.012 CrossRefGoogle Scholar - 9.Polach, O.: Wheel profile design for target conicity and wide tread wear spreading. Wear
**271**, 195–202 (2011). https://doi.org/10.1016/j.wear.2010.10.055 CrossRefGoogle Scholar - 10.Sadeghi, J., Sadeghi, S., Niaki, S.T.A.: Optimizing a hybrid vendor-managed inventory and transportation problem with fuzzy demand: an improved particle swarm optimization algorithm. Inf. Sci.
**272**, 126–144 (2014). https://doi.org/10.1016/j.ins.2014.02.075 MathSciNetCrossRefGoogle Scholar - 11.Santamaria, J., Herreros, J., Vadillo, E.G., Correa, N.: Design of an optimised wheel profile for rail vehicles operating on two-track gauges. Veh. Syst. Dyn.
**51**, 54–73 (2013). https://doi.org/10.1080/00423114.2012.711478 CrossRefGoogle Scholar - 12.Persson, I., Iwnicki, S.D.: Optimisation of railway profiles using a genetic algorithm. Veh. Syst. Dyn.
**41**, 517–527 (2004) Google Scholar - 13.Choi, H.Y., Lee, D.H., Lee, J.: Optimisation of a railway wheel profile to minimize flange wear and surface fatigue. Wear
**300**, 225–233 (2013). https://doi.org/10.1016/j.wear.2013.02.009 CrossRefGoogle Scholar - 14.Novales, M., Orro, A., Bugarín, M.R.: Use of a genetic algorithm to optimise wheel profile geometry. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**221**, 467–476 (2007). https://doi.org/10.1243/09544097jrrt150 CrossRefGoogle Scholar - 15.Kennedy, J., Eberhart, R.: Particle swarm optimization. In: Proceedings of IEEE International Conference on Neural Networks, vol. 4, pp. 1942–1948 (1995). https://doi.org/10.1109/ICNN.1995.488968 CrossRefGoogle Scholar
- 16.Lin, F., Zhou, S., Dong, X., Xiao, Q., Zhang, H., Hu, W., et al.: Design method of LM thin flange wheel profile based on NURBS. Veh. Syst. Dyn.
**114**, 1–16 (2019). https://doi.org/10.1080/00423114.2019.1657908 CrossRefGoogle Scholar - 17.Cui, D., Wang, R., Allen, P., An, B., li, L., Wen, Z.: Multi-objective optimization of electric multiple unit wheel profile from wheel flange wear viewpoint. Struct. Multidiscip. Optim.
**59**, 279–289 (2018). https://doi.org/10.1007/s00158-018-2065-5 MathSciNetCrossRefGoogle Scholar - 18.Shevtsov, I., Markine, V., Esveld, C.: Optimal design of wheel profile for railway vehicles. Wear
**258**, 1022–1030 (2005). https://doi.org/10.1016/j.wear.2004.03.051 CrossRefGoogle Scholar - 19.Shevtsov, I., Markine, V., Esveld, C.: Design of railway wheel profile taking into account rolling contact fatigue and wear. Wear
**265**, 1273–1282 (2008). https://doi.org/10.1016/j.wear.2008.03.018 CrossRefGoogle Scholar - 20.Markine, V.L., Shevtsov, I.Y.: Optimisation of a wheel profile accounting for design robustness. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**225**, 433–442 (2011). https://doi.org/10.1177/09544097jrrt305 CrossRefGoogle Scholar - 21.Jahed, H., Farshi, B., Eshraghi, M.A., Nasr, A.: A numerical optimisation technique for design of wheel profiles. Wear
**264**, 1–10 (2008). https://doi.org/10.1016/j.wear.2006.06.006 CrossRefGoogle Scholar - 22.Spangenberg, U., Fröhling, R.D., Els, P.S.: Long-term wear and rolling contact fatigue behaviour of a conformal wheel profile designed for large radius curves. Veh. Syst. Dyn.
**57**, 44–63 (2018). https://doi.org/10.1080/00423114.2018.1447677 CrossRefGoogle Scholar - 23.Shen, G., Ayasse, J.B., Chollet, H., Pratt, I.: A unique design method for wheel profiles by considering the contact angle function. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**217**, 25–30 (2003). https://doi.org/10.1243/095440903762727320 CrossRefGoogle Scholar - 24.Cui, D., Li, L., Jin, X., Li, X.: Optimal design of wheel profiles based on weighed wheel/rail gap. Wear
**271**, 218–226 (2011). https://doi.org/10.1016/j.wear.2010.10.005 CrossRefGoogle Scholar - 25.Ignesti, M., Innocenti, A., Marini, L., Meli, E., Rindi, A.: Development of a wear model for the wheel profile optimisation on railway vehicles. Veh. Syst. Dyn.
**51**(9), 1363–1402 (2013). https://doi.org/10.1080/00423114.2013.802096 CrossRefzbMATHGoogle Scholar - 26.Zakharov, S., Goryacheva, I., Bogdanov, V., Pogorelov, D., Zharov, I., Yazykov, V., et al.: Problems with wheel and rail profiles selection and optimisation. Wear
**265**, 1266–1272 (2008). https://doi.org/10.1016/j.wear.2008.03.026 CrossRefGoogle Scholar - 27.Molatefi, H., Mazraeh, A., Shadfar, M., Yazdani, H.: Advances in Iran railway wheel wear management: a practical approach for selection of wheel profile using numerical methods and comprehensive field tests. Wear
**424–425**, 97–110 (2019). https://doi.org/10.1016/j.wear.2019.02.016 CrossRefGoogle Scholar - 28.Ashtiani, I.H.: Optimization of secondary suspension of three-piece bogie with bevelled friction wedge geometry. Int. J. Railw. Technol. Transp.
**5**, 213–228 (2017). https://doi.org/10.1080/23248378.2017.1336652 CrossRefGoogle Scholar - 29.Ashtiani, I.H., Rakheja, S., Ahmed, A.: Influence of friction wedge characteristics on lateral response and hunting of freight wagons with three-piece bogies. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**231**, 877–891 (2016). https://doi.org/10.1177/0954409716647095 CrossRefGoogle Scholar - 30.Liu, B., Mei, T., Bruni, S.: Design and optimisation of wheel–rail profiles for adhesion improvement. Veh. Syst. Dyn.
**54**, 429–444 (2016). https://doi.org/10.1080/00423114.2015.1137958 CrossRefGoogle Scholar - 31.Pombo, J., Ambrósio, J., Pereira, M., Lewis, R., Dwyer-Joyce, R., Ariaudo, C., et al.: A study on wear evaluation of railway wheels based on multibody dynamics and wear computation. Multibody Syst. Dyn.
**24**, 347–366 (2010). https://doi.org/10.1007/s11044-010-9217-8 CrossRefzbMATHGoogle Scholar - 32.Gao, L., Wang, P., Cai, X., Xiao, H.: Superelevation modification for the small-radius curve of Shen-shuo railway under mixed traffic of passenger and freight trains. J. Vib. Shock
**35**, 222–228 (2018). https://doi.org/10.13465/j.cnki.jvs.2016.14.036 (in Chinese) CrossRefGoogle Scholar - 33.Ye, Y., Ning, J.: Small-amplitude hunting diagnosis method for high-speed trains based on the bogie frame’s lateral–longitudinal–vertical data fusion, independent mode function reconstruction and linear local tangent space alignment. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit., 095440971882541 (2019). https://doi.org/10.1177/0954409718825412 CrossRefGoogle Scholar
- 34.Rongju, T.: The development of China’s railway building. Transp. Rev.
**4**(1), 27–42 (1984). https://doi.org/10.1080/01441648408716543 CrossRefGoogle Scholar - 35.Eisenbahn Freunde–Warstein. https://eisenbahnfreunde-warstein.hpage.com/willkommen.html
- 36.Schelle, H.: Radverschleißreduzierung für eine Güterzuglokomotive durch optimierte Spurführung. (Doctoral dissertation), Technische Universität Berlin (2014) Google Scholar
- 37.Tao, G., Ren, D., Wang, L., Wen, Z., Jin, X.: Online prediction model for wheel wear considering track flexibility. Multibody Syst. Dyn.
**44**, 313–334 (2018) MathSciNetCrossRefGoogle Scholar - 38.Ignesti, M., Innocenti, A., Marini, L., Meli, E., Rindi, A.: Development of a model for the simultaneous analysis of wheel and rail wear in railway systems. Multibody Syst. Dyn.
**31**, 191–240 (2014) CrossRefGoogle Scholar - 39.An, B., Wang, P., Zhou, J., Chen, R., Xu, J., Wu, B.: Applying two simplified ellipse-based tangential models to wheel–rail contact using three alternative nonelliptic adaptation approaches: a comparative study. Math. Probl. Eng.
**2019**, 1–17 (2019). https://doi.org/10.1155/2019/3478607 CrossRefGoogle Scholar - 40.Johnson, K.L.: Contact Mechanics. Cambridge University Press, Cambridge (2004) Google Scholar
- 41.Tao, G., Wen, Z., Zhao, X., Jin, X.: Effects of wheel–rail contact modelling on wheel wear simulation. Wear
**366–367**, 146–156 (2016). https://doi.org/10.1016/j.wear.2016.05.010 CrossRefGoogle Scholar - 42.Shen, Z.Y., Hedrick, J.K., Elkins, J.A.: A comparison of alternative creep force models for rail vehicle dynamic analysis. Veh. Syst. Dyn.
**12**, 79–83 (1983). https://doi.org/10.1080/00423118308968725 CrossRefGoogle Scholar - 43.Polach, O.: A fast wheel–rail forces calculation computer code. Veh. Syst. Dyn.
**33**, 728–739 (2000) CrossRefGoogle Scholar - 44.Kalker, J.J.: A fast algorithm for the simplified theory of rolling contact. Veh. Syst. Dyn.
**11**, 1–13 (1982). https://doi.org/10.1080/00423118208968684 CrossRefGoogle Scholar - 45.Sichani, M.S., Enblom, R., Berg, M.: An alternative to FASTSIM for tangential solution of the wheel–rail contact. Veh. Syst. Dyn.
**54**, 748–764 (2016). https://doi.org/10.1080/00423114.2016.1156135 CrossRefGoogle Scholar - 46.Auciello, J., Ignesti, M., Malvezzi, M., Meli, E., Rindi, A.: Development and validation of a wear model for the analysis of the wheel profile evolution in railway vehicles. Veh. Syst. Dyn.
**50**, 1707–1734 (2012). https://doi.org/10.1080/00423114.2012.695021 CrossRefzbMATHGoogle Scholar - 47.Sun, Y., Guo, Y., Zhai, W.: Prediction of rail non-uniform wear–influence of track random irregularity. Wear
**420–421**, 235–244 (2019) CrossRefGoogle Scholar - 48.Peng, B., Iwnicki, S., Shackleton, P., Crosbee, D.: Comparison of wear models for simulation of railway wheel polygonization. Wear
**436–437**, 203010 (2019). https://doi.org/10.1016/j.wear.2019.203010 CrossRefGoogle Scholar - 49.Archard, J.F.: Contact and rubbing of flat surfaces. J. Appl. Phys.
**24**, 981–988 (1953) CrossRefGoogle Scholar - 50.Jendel, T.: Prediction of wheel profile wear—comparisons with field measurements. Wear
**253**, 89–99 (2002). https://doi.org/10.1016/s0043-1648(02)00087-x CrossRefGoogle Scholar - 51.Pearce, T., Sherratt, N.: Prediction of wheel profile wear. Wear
**144**, 343–351 (1991). https://doi.org/10.1016/0043-1648(91)90025-p CrossRefGoogle Scholar - 52.Lewis, R., Dwyer-Joyce, R.S., Olofsson, U., Pombo, J., Ambrósio, J., Pereira, M., Ariaudo, C., Kuka, N.: Mapping railway wheel material wear mechanisms and transitions. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**224**, 125–137 (2010) CrossRefGoogle Scholar - 53.Ye, Y., Shi, D., Krause, P., Hecht, M.: A data-driven method for estimating wheel flat length. Veh. Syst. Dyn.
**304**, 1–19 (2019). https://doi.org/10.1080/00423114.2019.1620956 CrossRefGoogle Scholar - 54.Chowdhury, R., Adhikari, S.: Fuzzy parametric uncertainty analysis of linear dynamical systems: a surrogate modeling approach. Mech. Syst. Signal Process.
**32**, 5–17 (2012). https://doi.org/10.1016/j.ymssp.2012.05.002 CrossRefGoogle Scholar - 55.Nie, Y., Tang, Z., Liu, F., Chang, J., Zhang, J.: A data-driven dynamics simulation framework for railway vehicles. Veh. Syst. Dyn.
**56**, 406–427 (2017). https://doi.org/10.1080/00423114.2017.1381981 CrossRefGoogle Scholar - 56.Simpson, T.W., Mauery, T.M., Korte, J., Mistree, F.: Kriging models for global approximation in simulation-based multidisciplinary design optimization. AIAA J.
**39**, 2233–2241 (2001). https://doi.org/10.2514/3.15017 CrossRefGoogle Scholar - 57.Werner, R., Lauschmann, S., Ehms, H.: Betrieb mit 25 kV 50 Hz auf den Steilstrecken der Rübelandbahn im Harz. Elektr. Bahnen Verkehrssysteme
**107**(10), 413–425 (2009) Google Scholar - 58.European Commission. https://cordis.europa.eu/project/rcn/92249/reporting/en
- 59.Deutsche Reichsbahn: Strecke Blankenburg—Tanne. Entwurfs- und Vermessungsstelle der Deutschen Reichsbahn, Magdeburg (1965) Google Scholar
- 60.Pfeiffer, M., Hecht, M.: Concept & Decision Paper: force measurement method for ED-brake-monitoring (Electric Locomotive—BOMBARDIER TRAXX) incl. data analysis—Measurement campaign HVLE. (Report No. 11/2011), Technische Universität Berlin (2011) Google Scholar
- 61.Profillidis, V.A.: Railway Engineering. Avebury Technical, Aldershot (2000) Google Scholar
- 62.Olofsson, U., Sundvall, K.: Influence of leaf, humidity and applied lubrication on friction in the wheel–rail contact: pin-on-disc experiments. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**218**, 235–242 (2004). https://doi.org/10.1243/0954409042389364 CrossRefGoogle Scholar - 63.Kovalev, R., Lysikov, N., Mikheev, G., Pogorelov, D., Simonov, V., Yazykov, V., et al.: Freight car models and their computer-aided dynamic analysis. Multibody Syst. Dyn.
**22**, 399–423 (2009). https://doi.org/10.1007/s11044-009-9170-6 CrossRefzbMATHGoogle Scholar - 64.Hecht, M., Keudel, J., Friedrich, D.: Dokumentation der Messfahrten aus Juni 2006. (Report No. 12/2006), Technische Universität Berlin. (2006) Google Scholar
- 65.Hecht, M., Keudel, J.: Numerische Simulation eines Selbstentladewagens mit Y25-Drehgestellen. (Report No. 31/2005) (2005) Google Scholar
- 66.Hecht, M., Schelle, H.: Simulation von Kesselwagen mit Y25-Drehgestellen bei Gleislagefehlern. Technische Universität Berlin (2006) (Report No. 16/2006) Google Scholar
- 67.Enblom, R.: Deterioration mechanisms in the wheel–rail interface with focus on wear prediction: a literature review. Veh. Syst. Dyn.
**47**, 661–700 (2009). https://doi.org/10.1080/00423110802331559 CrossRefGoogle Scholar - 68.Tao, G.Q., Du, X., Wen, Z.F., Jin, X.S., Cui, D.B.: Development and validation of a model for predicting wheel wear in high-speed trains. J. Zhejiang Univ. Sci. A
**18**, 603–616 (2017) CrossRefGoogle Scholar - 69.Braghin, F., Lewis, R., Dwyer-Joyce, R., Bruni, S.: A mathematical model to predict railway wheel profile evolution due to wear. Wear
**261**, 1253–1264 (2006). https://doi.org/10.1016/j.wear.2006.03.025 CrossRefGoogle Scholar - 70.Luo, R., Shi, H., Teng, W., Song, C.: Prediction of wheel profile wear and vehicle dynamics evolution considering stochastic parameters for high-speed train. Wear
**392–393**, 126–138 (2017). https://doi.org/10.1016/j.wear.2017.09.019 CrossRefGoogle Scholar - 71.Enblom, R., Berg, M.: Simulation of railway wheel profile development due to wear—influence of disc braking and contact environment. Wear
**258**, 1055–1063 (2005). https://doi.org/10.1016/j.wear.2004.03.055 CrossRefGoogle Scholar - 72.Braghin, F., Bruni, S., Resta, F.: Wear of railway wheel profiles: a comparison between experimental results and a mathematical model. Veh. Syst. Dyn.
**37**, 478–489 (2002). https://doi.org/10.1080/00423114.2002.11666256 CrossRefGoogle Scholar - 73.Sun, Y., Guo, Y., Lv, K., Chen, M., Zhai, W.: Effect of hollow-worn wheels on the evolution of rail wear. Wear
**436–437**, 203032 (2019). https://doi.org/10.1016/j.wear.2019.203032 CrossRefGoogle Scholar - 74.Meymand, S.Z., Keylin, A., Ahmadian, M.: A survey of wheel–rail contact models for rail vehicles. Veh. Syst. Dyn.
**54**, 386–428 (2016). https://doi.org/10.1080/00423114.2015.1137956 CrossRefGoogle Scholar - 75.Laursen, T.: Computational Contact and Impact Mechanics, Fundamentals of Modeling Interfacial Phenomena in Nonlinear Finite Element Analysis. Springer, Berlin (2002) zbMATHGoogle Scholar
- 76.Nackenhorst, U.: The ALE-formulation of bodies in rolling contact. Comput. Methods Appl. Mech. Eng.
**193**, 4299–4322 (2004). https://doi.org/10.1016/j.cma.2004.01.033 MathSciNetCrossRefzbMATHGoogle Scholar - 77.Vollebregt, E., Segal, G.: Solving conformal wheel–rail rolling contact problems. Veh. Syst. Dyn.
**52**, 455–468 (2014). https://doi.org/10.1080/00423114.2014.906634 CrossRefGoogle Scholar - 78.Burgelman, N., Li, Z., Dollevoet, R.: A new rolling contact method applied to conformal contact and the train–turnout interaction. Wear
**321**, 94–105 (2014). https://doi.org/10.1016/j.wear.2014.10.008 CrossRefGoogle Scholar - 79.Kalker, J.J.: Three-Dimensional Elastic Bodies in Rolling Contact. Solid Mechanics and Its Applications, vol. 2. Kluwer Academic Publishers, Dordrecht (1990) CrossRefGoogle Scholar
- 80.Linder, C.: Verschleiß von Eisenbahnrädern mit Unrundheiten. Dissertation Nr. 12342, ETH, Zurich (1997) Google Scholar
- 81.Piotrowski, J., Kik, W.: A simplified model of wheel/rail contact mechanics for non-hertzian problems and its application in rail vehicle dynamic simulations. Veh. Syst. Dyn.
**46**, 27–48 (2008). https://doi.org/10.1080/00423110701586444 CrossRefGoogle Scholar - 82.Liu, B., Bruni, S., Vollebregt, E.: A non-hertzian method for solving wheel–rail normal contact problem taking into account the effect of yaw. Veh. Syst. Dyn.
**54**, 1226–1246 (2016). https://doi.org/10.1080/00423114.2016.1196823 CrossRefGoogle Scholar - 83.Sun, Y., Zhai, W., Guo, Y.: A robust non-hertzian contact method for wheel–rail normal contact analysis. Veh. Syst. Dyn.
**56**, 1899–1921 (2018). https://doi.org/10.1080/00423114.2018.1439587 CrossRefGoogle Scholar - 84.Ayasse, J., Chollet, H.: Determination of the wheel rail contact patch in semi-hertzian conditions. Veh. Syst. Dyn.
**43**, 161–172 (2005). https://doi.org/10.1080/00423110412331327193 CrossRefGoogle Scholar - 85.Sichani, M.S., Enblom, R., Berg, M.: A novel method to model wheel–rail normal contact in vehicle dynamics simulation. Veh. Syst. Dyn.
**52**, 1752–1764 (2014). https://doi.org/10.1080/00423114.2014.961932 CrossRefGoogle Scholar - 86.Magalhães, H., Marques, F., Liu, B., Antunes, P., Pombo, J., Flores, P., et al.: Implementation of a non-hertzian contact model for railway dynamic application. In: Multibody System Dynamics (2019). https://doi.org/10.1007/s11044-019-09688-y CrossRefGoogle Scholar
- 87.Marques, F., Magalhães, H., Liu, B., Pombo, J., Flores, P., Ambrósio, J., et al.: On the generation of enhanced lookup tables for wheel–rail contact models. Wear
**434–435**, 202993 (2019). https://doi.org/10.1016/j.wear.2019.202993 CrossRefGoogle Scholar - 88.Kalker, J.J.: Rollin contact phenomena—linear elasticity. In: Jacobson, B., Kalker, J. (eds.) Rolling Contact Phenomena. CISM Courses and Lectures, pp. 1–84. Springer, New York (2001) zbMATHGoogle Scholar
- 89.Hossein-Nia, S., Sichani, M.S., Stichel, S., Casanueva, C.: Wheel life prediction model—an alternative to the FASTSIM algorithm for RCF. Veh. Syst. Dyn.
**56**, 1051–1071 (2017). https://doi.org/10.1080/00423114.2017.1403636 CrossRefGoogle Scholar - 90.An, B., Ma, D., Wang, P., Zhou, J., Chen, R., Xu, J., et al.: Assessing the fast non-Hertzian methods based on the simulation of wheel–rail rolling contact and wear distribution. Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit. 095440971984859 (2019). https://doi.org/10.1177/0954409719848592
- 91.Kalker, J.J.: A Strip Theory for Rolling with Slip and Spin. TU Delft, Delft (1966) Google Scholar
- 92.Ye, Y., Shi, D., Krause, P., Tian, Q., Hecht, M.: Wheel flat can cause or exacerbate wheel polygonization. Veh. Syst. Dyn. (2019). https://doi.org/10.1080/00423114.2019.1636098 CrossRefGoogle Scholar
- 93.Johansson, A.: Out-of-round railway wheels—causes and consequences. (Doctoral thesis), Chalmers University of Technology, Sweden (2005) Google Scholar
- 94.E, J., Han, D., Qiu, A., Zhu, H., Deng, Y., Chen, J., et al.: Orthogonal experimental design of liquid-cooling structure on the cooling effect of a liquid-cooled battery thermal management system. Appl. Therm. Eng.
**132**, 508–520 (2018). https://doi.org/10.1016/j.applthermaleng.2017.12.115 CrossRefGoogle Scholar - 95.Roshanian, J., Ebrahimi, M.: Latin hypercube sampling applied to reliability-based multidisciplinary design optimization of a launch vehicle. Aerosp. Sci. Technol.
**28**, 297–304 (2013). https://doi.org/10.1016/j.ast.2012.11.010 CrossRefGoogle Scholar - 96.Xin, C., Lu, Q., Ai, C., Rahman, A., Qiu, Y.: Optimization of hard modified asphalt formula for gussasphalt based on uniform experimental design. Constr. Build. Mater.
**136**, 556–564 (2017). https://doi.org/10.1016/j.conbuildmat.2017.01.068 CrossRefGoogle Scholar - 97.BS EN 15313: Railway applications. In-service wheelset operation requirements. In-service and off-vehicle wheelset maintenance (2016). https://standards.globalspec.com/std/10011238/dsf-en-15313-pra1
- 98.BS EN 14363: Railway Applications-Testing for the Acceptance of Running Characteristics of Railway Vehicles-Testing of Running Behavior and Stationary Tests. CEN, Brussels (2005). https://standards.globalspec.com/std/13162966/en-14363 Google Scholar
- 99.Polach, O.: On non-linear methods of bogie stability assessment using computer simulations. Proc. Inst. Mech. Eng., F J. Rail Rapid Transit
**220**, 13–27 (2006). https://doi.org/10.1243/095440905x33251 CrossRefGoogle Scholar - 100.Ye, Y.G., Hecht, H.: Derailment safety and stability behavior tests of Y25-container wagon with wheel diameter decreasing from 920 mm to 550 mm. (Report No. 11/2018), Technische Universität Berlin (2018) Google Scholar
- 101.Zhang, Y., Zhou, S., Zhang, G., Chang, C., Zhou, Q., Wang, W.: Study and discussion of high speed wheel-rail hardness matching. China Railw.
**1**, 7–14 (2018). (In Chinese). http://www.cnki.com.cn/Article/CJFDTotal-TLZG201801002.htm Google Scholar - 102.Knothe, K., Stichel, S.: Rail Vehicle Dynamics. Springer, Berlin (2018) Google Scholar

## Copyright information

**Open Access** This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.