Railway Vehicle Dynamics: Computer-Aided Analysis of Multibody Dynamics (Part 2)

Computer-aided analysis of
multibody dynamics (part 2)
Railway vehicle dynamics
Paul Fisette
([email protected])
GraSMech – Multibody – Part II
Contents
Part I : Wheel-rail contact in railway dynamics
¾ Contact geometry
¾ Contact forces
¾ Wheelset dynamic behavior
Part II : Railway dynamics - multibody approach
¾ Multibody representation pter

9
Cha
Kluwer Academic Publishers, 2003
¾ Wheel/rail contact – independent wheel model
¾ Flange contact model
¾ Validations - Applications
¾ Other topics
1
Contents
¾ Contact geometry
¾ Contact forces
¾ Multibody representation
¾ Other topics
Contact geometry
Wheelset with conical wheel on « knife-edge » rails
Rolling radii :
( In practice : δ0 ~ 1/20e … 1/40e )
Equivalent conicity :
Rolling radii difference (left-right) :
Roll angle :
2
Contact geometry
Wheelset kinematic motion (~ first order kinematics)
Motion of a wheelset initially centered on the track
Hypothesis : pure rolling motion (no slip)

¾ Kinematic relations (first order)
Combined lateral-yaw motion of a wheelset
¾ Constraints (in y and ψ) :

Lateral (y) :
Yaw (ψ) :
Contact geometry
Wheelset kinematic motion (next)

2 d.o.f. kinematic equations (= constraints at « velocity level »)
Periodic Solution :
¾ frequency : ¾ Wave length :

(Klingel formula, 1883 !)
Wheelset kinematic yaw (under pure rolling assumption)
3
Contact geometry
Bogie : yaw kinematic motion (similar reasoning)
l : half wheelset base

Wave length :
l0 : half wheelset length
Wheel/rail with circular profiles (as a matter of interest)
More complex kinematics
¾ Roll angle :
¾ vertical disp.:
4 bar-mechanism
¾ equiv. conicity :
Contact geometry
Wheel/rail contact : yaw angle influence on contact geometry

=> Longitudinal shift of the contact point (Ù « shift angle » θ) due to wheelset yaw + wheel conicity
Insignificant for the tread contact point (angle ~ 0.05 rad)

Fundamental for the flange contact point (angle > 1 rad) : see latter on
4
Contact geometry
Real wheel profiles : numerical somution

Profile definition (analytical form or numerical measurements)
¾ ex. UIC 60 rail (piecewise circular), S1002 wheel (piecewise polynomials)
Numerical determination of the contact(s) point(s) = f (y, ψ)

¾ Pre-computation (look-up tables => f (y, ψ) )
¾ In-line computation (Newton-Raphson, dichotomic method, … see Part II)
Computation of the left/right rolling radii r and contact angles δ :
∆r ∆δ
y y
Contact geometry
Double (flange) contact point

Tramway Ù « clear » double contact (tread + flange)
Train Ù the contact point « moves » from tread to flange (with possible jump !!)
Always present
intermittent
5
Contact geometry
Double (flange) contact point

Tramway Ù « clear » double contact (tread + flange)
Train Ù the contact point « moves » from tread to flange (with possible jump !!)
Always present
Contact geometry
Contact point « jump »

Exists on theoretical « new » profiles (ex. S1002 wheelset on UIC60 rail)
on produces :
ROBOTRAN results
Exists also on worn profile (but their location change with wear !) (IAVSD benchmark #1, 1991)
¾ New rail :
¾ Slightly worn rail :
6
Contents
¾ Contact geometry
¾ Contact forces
¾ Other topics
Contact forces
Coulomb’s law :
No slip =>
Slip =>
Creepage (« between slip and pure rolling »)
Longitudinal creepage :
Lateral creepage :
Spin creepage :
¾ with : Point of contact assumption !!!!!
7
Contact forces
Creepage computation (not MBS approach !)
Lateral creepage :
Spin creepage :
Contact forces
Contact surface
Non uniform pressure distribution
Contact ellipse (a, b) ÙHertz (but + creepage)
A, B, m, n depend on
contact curvature radii
8
Contact forces
Kalker’s theory*
Linear theory
with :
Heuristic modeling of saturation :
Non linear theories : Duvorol, Fastsim (Kalker), Contact (Kalker)
Contents
¾ Contact geometry
¾ Contact forces
¾ Other topics
9
Wheelset dynamic behavior
Linear model
Hypotheses :
¾ « suspended » wheelset
¾ Linearized equations of motion :
Solution :
Im (λ)
=> « Linear » Critical speed VLim:
« Hunting » eigenmode
¾ Eigenvalue λ versus speed :
or.: Jashinski thesis (Delft)
Real (λ)

Linear model : example
ROBOTRAN/MBsysLab model :
http://www.prm.ucl.ac.be/FDP/Tutorial/Medium/Bogie/Introduction.html
=> Comparison between linear – non linear model
10
=> Behavior at low and high velocity
y ψ

=> Evolution of the hunting mode versus velocity

(via successive modal analyses)
stable
unstable Top view (Robotran)
Rear view (Robotran)
11
Non-linear model
Notion of limit cycle (ex. van der Pol differential equation
« damping term »
ω0 , ξ > 0
• small y => negative damping Ù the system stores energy

• large y => positive damping Ù the system looses energy
=> Limit cycle : oscillatory solution in which the 2 energies compensate
Application : limit cycle of a « suspended » wheelset : .

Lateral wheelset displacement y yÙy
ROBOTRAN results ROBOTRAN results
Contact geometry
Critical speed Vcr
Limit cycle amplitudes versus system velocity (system = wheelset, amplitude = y)
speed
Numerical determination of the critical speed Vcr and VLim

¾ Time simulation from high speed (with lateral « safety bumps ») to low speed
y
speed ROBOTRAN results (IAVSD benchmark #2, 1991)
12
Conclusions (part I)
Key points
Contact geometry : a difficult problem to solve
Contact forces : idem Ù thanks ! Mr Kalker (et al.)
Dynamic behavior of a wheeset on a straight track : critical speeds
Other aspects
Curving dynamics : « crab-wise » bogie behavior
Derailment criteria (Y/Q wheel force ratio)
Wear next lecture (N. Docquier)
Track irregularities => noise, comfort, …
Full train : different morphologies, tilting trains, pneumatic suspensions, …
Contents
¾ Contact geometry
¾ Contact forces
¾ Other topics
13
The « BAS2000 » bogie : a general case
The « tram2000 »
Six « 3D » kinematic loops
Four independent wheel/rail contacts
=> No more wheelset, even artificial (left-right)
MBS : wheel or wheeset - does’nt matter!
In relative coordinates, a wheel is a leaf body to which forces apply
Independent wheels Wheelset = (axle + 2 « locked » indep. wheels)
14
The « BAS2000 » bogie : a general case
A wheel = a leaf body which is constrained on a rail (inertial body)
The BAS2000 cannot be modeled as a bogie

… but as a mechanism with wheels ! => relative coordinates approach suitable
Contents
¾ Contact geometry
¾ Contact forces
¾ Other topics
15
The « wheel/rail » joint »
=> 5 dof + 1 « normal » constraint
Recall : coordiante partitioning reduction
Coordinate partitioning :
ODE !
32 GraSMech – Multibody – Part II
16
A Single Wheel on a Rail : 5 d.o.f.
w j j
G
k
i
β(w) ρ (w) 2 Auxiliary variables :
h
x_
j ¾ w : lateral position
θQ
w
θ
¾ θ : Shift angle
...
3-D Contact Constraints :
1. Point Q belongs to Rail Surface
2. Wheel / Rail Surfaces Tangent at Point Q
=> Lagrange Multipliers Technique
Wheel on rail: Constraints solution
Robust solution of contact constraints

rail profile
λ2 = 0 !!
λ3 = 0 !!
A dichotomic method “embeded into” the Newton-Raphson algorithm
17
Wheel on rail: Constraints solution
requires
= 0 when the constraints are satisfied

Wheel on rail: Constraints derivative
{X̂}
{ Ŝ}
G {Ŷ}
w
cos α = _+ π/2 λ1 = FN
x
Q
{T̂}
Î 3
u α P
{R̂}
Î 1
Î 2 O
h(q) satisfied
P
18
Wheel on rail: Constraints 2d derivative
{X̂}
{ Ŝ}
G {Ŷ}
w
x
Q
{T̂}
u {R̂}
Î 3 P
Î 1
Î 2 O
P
Wheel on curved rail
19
Wheel on curved rail
Creepages : a « real » MBS computation
{ Ŝ}
{X̂} Lateral creepage :
G {Ŷ}
w
x
Q
{T̂}
Î 3
u
P
{R̂}
Spin creepage :
Î 1
Î 2 O
P
20
Wheel on rail: Kalker’s model
{X̂}
{ Ŝ}
G {Ŷ}
w
x
Q
{T̂}
u {R̂}
Î 3 P
Î 1
Î 2 O
P
Kalker (linear) :
Wheel on rail: flange forces (tramways)
21
Wheel on rail: flange forces
Influence of the shift angle
No wheel yaw With wheel yaw
For the flange contact force, in terms of tangent slip,

the tread contact (« creepage ») can be seen as a pure rolling motion (~ICR)
Contents
¾ Contact geometry
¾ Contact forces
¾ Other topics : « towards » multiphysics
22
Geometrical validation
UCL
Delft
Wheelset limit cycle
x x
UCL Delft
23
Bogie : « non linear » critical speed
ROBOTRAN results
(IAVSD benchmark #2, 1991)
Deraillement Cycle limite stable
The “BAS 2000” articulated bogie
24
“BAS 2000” : straight track behavior
« Ping-pong » effect on a straigth track

Yaw deformation
time
The “Tram 2000” tramway

The "Tram 2000"
Model :
Complete TRAM 2000 in Entry Curving :
- 4 Sub-Systems
- 74 Variables
- 32 Constraints
- 42 Degrees of Freedom
Carbody A Carbody C Carbody B
Front BAS 2000 Bogie
Conventionnal Bogie Back BAS 2000 Bogie

V
Velocity : 25 km/h
Initial Conditions : Tramway Centered on the Straight Track
25
“Tram 2000” tramway - results
Result :
Carbody A Carbody C Carbody B
Wheel / Rail Alignment in a Low Radius Curve (R = 50 m)
Front BAS 2000 Bogie
α
V
right rail
wheel Conventionnal Bogie Back BAS 2000 Bogie

V
Wheel / Rail Angle of Attack
Comparison of bogie behaviors
Bogies : BW BA
BW : Rigid bogie with wheeslets
BI : Rigid bogie with independent wheels
BA : BA2000 articulated « bogie »

BI
Situation : entry curving
26
Comparison of bogie behaviors
Steady state equilibrium : tread and flange lateral forces

- « Crab-wise »
- « Crab-wise » BW behavior
behavior
- Low longitudinal
tread forces
BI
BA
- Distorted, no crab-wise
- Dissipated power minimized
Contents
¾ Contact geometry
¾ Contact forces
¾ Other topics : « towards » multiphysics
27
Multibody and railway “electro” dynamics
Railway bogies actuated by three

phase induction motors
Cm
Fz
• Fz in the -motor to
chassis-joint
• electromechanical
torque Cm
Problem:
Torque oscillations lead to
fatigue stress in the chassis
=> Next lecture Fz Cm
Multibody and railway pneumatic suspension
Multibody system
Hybrid system
Pneumatic system
=> Next lecture
28

Railway Vehicle Dynamics: Computer-Aided Analysis of Multibody Dynamics (Part 2)

Uploaded by

Copyright:

Available Formats

Railway Vehicle Dynamics: Computer-Aided Analysis of Multibody Dynamics (Part 2)

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Railway Vehicle Dynamics: Computer-Aided Analysis of Multibody Dynamics (Part 2)

Uploaded by

Copyright:

Available Formats

Computer-aided analysis of

multibody dynamics (part 2)

Railway vehicle dynamics

GraSMech – Multibody – Part II

 Part I : Wheel-rail contact in railway dynamics

¾ Wheelset dynamic behavior

 Part II : Railway dynamics - multibody approach

¾ Multibody representation pter

¾ Flange contact model

GraSMech – Multibody – Part II

 Part I : Wheel-rail contact in railway dynamics

¾ Wheelset dynamic behavior

 Part II : Railway dynamics - multibody approach

¾ Wheel/rail contact – independent wheel model

¾ Flange contact model

GraSMech – Multibody – Part II

Wheelset with conical wheel on « knife-edge » rails

( In practice : δ0 ~ 1/20e … 1/40e )

 Rolling radii difference (left-right) :

GraSMech – Multibody – Part II

Wheelset kinematic motion (~ first order kinematics)

Motion of a wheelset initially centered on the track

 Hypothesis : pure rolling motion (no slip)

Combined lateral-yaw motion of a wheelset

¾ Constraints (in y and ψ) :

GraSMech – Multibody – Part II

Wheelset kinematic motion (next)

¾ frequency : ¾ Wave length :

Wheelset kinematic yaw (under pure rolling assumption)

GraSMech – Multibody – Part II

Bogie : yaw kinematic motion (similar reasoning)

l : half wheelset base

Wheel/rail with circular profiles (as a matter of interest)

 More complex kinematics

GraSMech – Multibody – Part II

Wheel/rail contact : yaw angle influence on contact geometry

 Insignificant for the tread contact point (angle ~ 0.05 rad)

GraSMech – Multibody – Part II

Real wheel profiles : numerical somution

 Numerical determination of the contact(s) point(s) = f (y, ψ)

 Computation of the left/right rolling radii r and contact angles δ :

GraSMech – Multibody – Part II

Double (flange) contact point

GraSMech – Multibody – Part II

Double (flange) contact point

GraSMech – Multibody – Part II

Contact point « jump »

¾ Slightly worn rail :

GraSMech – Multibody – Part II

 Part I : Wheel-rail contact in railway dynamics

¾ Wheelset dynamic behavior

 Part II : Railway dynamics - multibody approach

¾ Wheel/rail contact – independent wheel model

¾ Flange contact model

GraSMech – Multibody – Part II

Creepage (« between slip and pure rolling »)

¾ with : Point of contact assumption !!!!!

GraSMech – Multibody – Part II

Part I : Wheel-rail contact in railway dynamics

Part II : Railway dynamics - multibody approach

Part I : Wheel-rail contact in railway dynamics

Part II : Railway dynamics - multibody approach

Rolling radii difference (left-right) :

Hypothesis : pure rolling motion (no slip)

More complex kinematics

Insignificant for the tread contact point (angle ~ 0.05 rad)

Numerical determination of the contact(s) point(s) = f (y, ψ)

Computation of the left/right rolling radii r and contact angles δ :

Part I : Wheel-rail contact in railway dynamics

Part II : Railway dynamics - multibody approach

Contact ellipse (a, b) ÙHertz (but + creepage)

Heuristic modeling of saturation :

Non linear theories : Duvorol, Fastsim (Kalker), Contact (Kalker)

Part I : Wheel-rail contact in railway dynamics

Part II : Railway dynamics - multibody approach

Application : limit cycle of a « suspended » wheelset : .

Numerical determination of the critical speed Vcr and VLim

Contact forces : idem Ù thanks ! Mr Kalker (et al.)

Dynamic behavior of a wheeset on a straight track : critical speeds

Derailment criteria (Y/Q wheel force ratio)

Wear next lecture (N. Docquier)

Track irregularities => noise, comfort, …

Full train : different morphologies, tilting trains, pneumatic suspensions, …

Part I : Wheel-rail contact in railway dynamics

Part II : Railway dynamics - multibody approach

A wheel = a leaf body which is constrained on a rail (inertial body)

The BAS2000 cannot be modeled as a bogie

Part I : Wheel-rail contact in railway dynamics

Part II : Railway dynamics - multibody approach