Content uploaded by Dmitry Pogorelov
Author content
All content in this area was uploaded by Dmitry Pogorelov on Dec 24, 2022
Content may be subject to copyright.
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
OPTIMIZATION OF WHEEL PROFILES BY MINIMAL WEAR CRITERION
Alexander Rodikov*, Dmitry Pogorelov, Vitaly Simonov
*e-mail: rodikov@umlab.ru
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Contents
Technique of prediction wear of wheel profiles in UM software
Application to optimization of wheel profiles by minimal wear criterion
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Approaches to wear simulation
Approaches to wear simulation
Parallel
The mutual affect of wear processes and vehicle dynamics is taken into
account, i.e. small changes in the profile geometry due to wear occur directly
in the dynamic integration loop.
Sequential
Does not take into account the effect of the change of the profile geometry
due to wear on the vehicle dynamics –the profile geometry updates after the
dynamics simulation. To obtain measurable wear, sequential simulation must
be performed quite a large number of times.
1
2
3
Scheme of parallel approach of wear simulation
wear step
1
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Models of wheel/rail contact 2
Are suited for wear simulation
1. Hertzian contact theory + FASTSIM
2. Kik-Piotrowski contact model + modified FASTSIM
3. CONTACT library
Example of the contact of worn profiles: Kik-Piotrowski model
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
x
y
Wear models 3
Fleisher model
where is the volume wear [m3];
is the wear coefficient [m3/J];
is the work of frictional forces [J].
where is the power of frictional forces [W].
where is the tangential traction [Pa];
is the sliding velocity [m/s];
is the contact patch area [m2].
While using the FASTSIM-like algorithms, the work of frictional forces is calculated
according to the following formula:
where is the longitudinal wheelset velocity [m/s];
is the integration time step size [s]; is the creepages [-].
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Wear models 4
Specht model
where is the critical power density of frictional forces [W/m2];
is the jump coefficient [-].
mild wear regime,
W [μg/ mm2]
severe wear regime,
A [J/ mm2]
100
200
300
0 10 20 30 40 30
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Wear models 5
Archard model
where is the wear coefficient [-];
is the normal force [N];
if the sliding distance [m];
is the hardness of the softest contacting surfaces [N/m2];
is the normal pressure [Pa].
Sliding velocity [m/s]
Contact pressure [GPa]
Wear coefficient map
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Wear control parameters of wheel and rail profiles 6
Sd is the flange thickness, measured from wheel tape-circle;
Sh is the flange height;
qR is the flange steepness;
Tw is the tread wear;
St is the flange thickness, measured from flange vertex;
dSd, dSt are difference of parameters Sd and St between worn and initial profiles.
is the vertical rail head wear;
is the lateral rail wear;
is the gauge corner wear
Note:
UM has tools for calculating of wear control parameters of profiles. The third-party software (e.g.,
MiniProf) for the analysis of worn profiles is not required.
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Profile-updating algorithm
before wear step
Histogram of
distribution of volume
wear along profile (w)
after wear step
wear step
1Accumulation of volume wear for every separate wheel of wheelset
Rolling direction
Integration
step
7
Wear mask
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Profile-updating algorithm
2Scaling volume wear according to the mileage assigned to one wear step
3Weighted summation of volume wear for each wheel
where , are the statistical weights of configuration and speed respectively;
is the accumulated volume wear for configuration at running speed ;
is the number of configurations;
is the number of -configuration speeds.
8
4Calculation of wear depth in the center of each circumferential strip
where is the smoothed distribution of the volume wear;
is the arc-coordinate of the center of the strip ;
is the wheel radius;
is the width of wear accumulation interval.
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Forward
route
FL
FR
RL
RR
Profile-updating algorithm
5Profile updating
where is the external normal for the wheel profile.
9
Averaging (symmetrizing) of wear
•Averaging wear of wheel of same wheelset
where is the wheelset index.
•Averaging wear of wheelsets
where is the number of wheelsets.
3*
Backward
route
RR
RL
FR
FL
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Profile-updating algorithm
Comparison of simulation results of complete and simplified sets of configurations. The
"simplified set" is the set of configurations with only right curves and forward routes,
but with wear symmetry is taking into account.
10
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Parallel computations on multi-core processors
Test set of configurations
Curve radius [m]
Speeds
[km/h]
Traction [-
]
323
25, 50, 65
0/0.6
581
25, 50, 65
0/0.6
1497
25, 50, 65
0/0.6
25, 50, 65
0/0.6
Total number of computations
24
Class 2ES10 electric freight locomotive.
The loco model includes 37 bodies with
110 DOF.
11
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Parallel computations on multi-core processors
Simulation of vehicle dynamic
on one wear step
Master thread
BarrierTask manager
Profiles update
Thread 1
Thread 2
Thread 3
Thread N
…
Fork-join model of parallel computation of wheel profiles wear
12
*Intel Core i7-12700KF CPU, 3.6 GHz, 64 GB RAM
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Contents
Technique of prediction wear of wheel profiles in UM software
Application to optimization of wheel profiles by minimal wear criterion
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Optimization of wheel profiles by minimal wear criterion
Objective function to minimize the wheel wear:
Constraints: derailment coefficient, maximum wheel-rail lateral forces, minimum critical
speed.
Iterative algorithm for selecting the optimal profile
1Selection and parameterization a start profile. Parameterization of the profile by circular
arcs is used.
L
, are starting and ending slopes;
, , are radii of arcs;
, are slopes of arcs;
is the tangent length.
Note:
Not all parameters are independent,
between some of these parameters there
are relationship. For example, are set
,,,,and values, then
and values are calculated.
Optimization region
13
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Optimization of wheel profiles by minimal wear criterion
2Set of profile-candidates is generated by varying the parameters in valid ranges.
3Compared of profiles-candidates in wear simulation. Optimal profiles with minimum
of the wear control parameters are selected.
4Checking selected profiles for constraints satisfaction.
5New iteration of profile optimization –go to step 2 and generation of the new set of
profiles with decrease increment of parameters.
14
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
Optimization of wheel profiles by minimal wear criterion
Standard profile
Optimal profile
56%
Critical value of
flange wear
15
COMPUTER SIMULATION IN RAILWAY TRANSPORT: DYNAMICS, STRENGTH, WEAR
V scientific and technical workshop, Bryansk, Russia, 4-6 October 2022
OPTIMIZATION OF WHEEL PROFILES BY MINIMAL WEAR CRITERION
Thank you for your attention!