PRELIMINARY DESIGN OF A

MECHANICAL CLUTCH
Presented by Hari Dileep Reddy (820331)
PART 1: Computation of the required axial force and
definition of the disk compliance
Definition of the clutch geometry
A generic dry friction clutch has the following configuration during the engagement phase:

In the first part of the project, we are going to model the main parts of the clutch responsible of
transmitting the torque from the engine to the transmission of the vehicle:
 the flywheel,
 the friction disk (or driven plate),
 the pressure plate.
2
According to the project data provided, we have decided to use the following disk dimensions:
 internal radius → Ri = 100 mm → Di = 200 mm
 external radius → Re = 128.5 mm → De = 257 mm

Computation of the axial force

We have to calculate the axial force acting on the disk in two different cases according to the
distribution of the normal pressure in different situations:

1. Constant pressure:
3*T*(Re2-Ri2)
F= = 3265.3 N
2*k*μ*(Re3-Ri3)
2. Constant wear:
2*T
F= = 3282.3 N
k*μ*(Re+Ri)

The first case corresponds to a new clutch, where the hypothesis of constant pressure in the contact
between the friction disk and both the flywheel and the pressure plate applies. On the other hand,
the second case corresponds to a worn clutch, where the pressure cannot be assumed constant
anymore in the contact surfaces, but it can be used the hypothesis of constant wear of the disk.

Model components
Defined the geometry, the following pictures show the axisymmetric sketches of all the parts of the
clutch:
 Flywheel

3
 New disk

 Worn disk

 Pressure plate

4
Assembled axisymmetric model
Considering the previous axisymmetric parts, we have approached the behaviour of the clutch using
a 2D model. Observe that we have constraint the motion of the flywheel to simulate the connection
with the engine shaft and we have applied the axial force on the pressure plate at the contact point
with the diaphragm spring.
Additionally, we defined an interaction of surface-to-surface contact between the pressure plate and
the disk surfaces, and also between the flywheel and the disk.

 Model with the new clutch

PRESSURE PLATE

DISK - SPRING
LININGS

FLYWHEEL

 Model with the worn clutch

PRESSURE PLATE

DISK - SPRING
LININGS

FLYWHEEL

5
Definition of the disk compliance
Once the model is completed must be assigned the stiffness of the spring.
Trying different values of Young’s modulus, we arrived to an acceptable value equal to:
E = 0.8 MPa.
For this value, the spring’s elasticity guarantees the compliance for having, not only the required
displacement in the linings (ensuring complete contact between surfaces), but also a pressure
distribution as constant as possible in both the new and worn clutch cases.

The figure above shows the case of E = 10 MPa in the worn clutch condition.
It doesn’t guarantee the contact of the whole surfaces and consequently it leads to a non constant
pressure distribution.

This second figure shows the case of E = 0.8 MPa in the worn clutch condition.
It guarantees the contact of the upper and lower linings respectively with the pressure plate and the
flywheel, leading to a good pressure distribution.
6
For the new clutch condition with again E = 0.8 MPa, it has been found the following result:

Before evaluating which are the real pressure distribution values, we must carry out the
convergence analysis on the model’s mesh realized with the Q8 elements.

Convergence analysis
We have meshed the part with Q8 finite elements.
Increasing the number of elements in the mesh of the component, we have obtained the following
relation between nodes and maximum contact pressure:

Convergence analysis for new clutch

Mesh # nodes CPRESS,max [MPa]
1 1263 0.2724
2 1643 0.2710
3 2755 0.3137
4 4068 0.3184
5 5520 0.3236
6 10891 0.3298
7 26623 0.3406
8 46251 0.3455

7
Convergence analysis for worn clutch
Mesh # nodes CPRESS,max [MPa]
1 976 0.4195
2 1223 0.4406
3 1585 0.4731
4 2408 0.4583
5 3557 0.5195
6 4672 0.5142
7 8895 0.5053
8 21089 0.5014

Observe that as we increase the number of nodes of the meshes the contact pressure converges to a
certain value, which is the closer one to the actual situation.
The convergence meshes we have used are, consequently, the last ones stated in the previous tables.

Convergence meshes
 Mesh for the new clutch

8
  Mesh for the worn clutch

Pressure distribution
Now we are going to evaluate the pressure distribution found along the contact surfaces, once set all
the design parameters explained before, both for the condition of new clutch and for the condition
of worn clutch.

 Pressure distribution for the new clutch

9
In the graph are represented the pressure curves in function of the radius of the linings for the new
clutch condition.
In particular, the blue line represents the contact pressure in the upper lining (between the lining and
the pressure plate), while the green line represents the contact pressure in the lower lining (between
the lining and the flywheel).
Looking at the graph above we can conclude that the pressure distribution is almost constant (or at
least constant within a good radius range) and it is lower with respect to the maximum contact
pressure that the lining’s material is able to stand (which is 2 MPa).
This shape is obtained because of the fact that all the surfaces of the disk, flywheel and pressure
plate are in contact, according to the selected Young’s modulus of the intermediate spring of the
disk.
We can also observe that the maximum pressure values are reached in the boundaries of the disk,
while in the internal part of it the pressure values are more or less constant.

 Pressure distribution for the worn clutch

The graph above represents the pressure curves in function of the radius of the linings for the worn
clutch condition.
Again, the blue line represents the contact pressure in the upper lining, while the green line
represents the contact pressure in the lower lining.

10
We can conclude that in this state, the pressure distribution is not as constant as in the case of the
new clutch, but it is still good enough.
In fact, the peak value is lower with respect to the maximum contact pressure that the lining’s
material is able to stand (2 MPa) and, moreover, the difference between the maximum and the
minimum contact pressure varies only around 0.3 MPa.
It is important to notice that the value of the maximum pressure is higher than the one obtained in
the new clutch condition, and this is because the linings are worn and the pressure able to guarantee
the contact between the parts must increase.
The maximum pressure is reached in the left side of the graph, which corresponds to the inner
diameter of the disk. This is because the thickness of the linings is higher in the inner part after the
wear process (the inner thickness of each lining is assumed 2.3 mm, while the outer thickness is set
to 2 mm).

Transmitted torque in both situations

It is requested to calculate the values of the transmitted torque in both the conditions of new and
worn clutch, using the data provided by the Finite Element analysis.

 New clutch
pmed = 0.221 MPa → average pressure in the contact surfaces
k=2 → numbers of surfaces in contact

(Re2-Ri2)
F = 2*π* pmed * = 4521.4 N
2

(Re3-Ri3)
T = 2*π*k*μ* pmed * = 415400 N*mm
3

 Worn clutch
pmed = 0.236 MPa → average pressure in the contact surfaces
k=2 → numbers of surfaces in contact

(Re2-Ri2)
F = 2*π* pmed * = 4828.3 N
2

(Re3-Ri3)
T = 2*π*k*μ* pmed * = 443590 N*mm
3
11
These are the real force and torque transmitted by the clutch in the new and the worn conditions.
They are higher than the ones from which we started, so we can conclude that the model is reliable,
because the clutch is able to transmit the desired torque (300 Nm) in all situations.

Axial equivalent stiffness

The value of the Young’s modulus E of the spring was previously defined when the disk
compliance has been assessed and it is equal to E = 0.8 MPa.

The area of the disk is equal to:

A = π*(Re2-Ri2) = 20459 mm2

The thickness of the disk (spring) is t = 2.5 mm.

Axial stiffness data

E [MPa] Re [mm] Ri [mm] A [mm2] t [mm]
0.8 128.5 100 20459 2.5

It is possible now to compute the axial stiffness as:

A*E N
Kaxial = = 6546.8
t mm

Torsional equivalent stiffness

The mean disk circumference is equal to:

L = 2* π*((Re+Ri)/2) = 717.8539 mm

The Poisson’s coefficient for a disk made in a soft material can be assumed equal to: ν ≈ 0.5.
The shear modulus is given by:

E
G= = 0.2667 MPa
2*(1+ ν)

The disk width is equal a = 28.5 mm.

The moment of inertia of the spring is given by:

J = (a*b3)/3 = 148.4375 mm4.

12
 Torsional stiffness data
E [MPa] a [mm] t [mm] ν G [MPa] J [mm4] L [mm]
0.8 28.5 2.5 0.5 0.2667 148.4375 717.8539

It is possible now to compute the torsional stiffness as:

G*J N*mm
Ktorsional = = 0.0551
L rad

13
PART 2: Design of the diaphragm spring
Definition of the diaphragm spring dimensions
In the second part of the project, we are going to design the geometry of the diaphragm spring.
First, we have to design the external radius of the spring so that it can fit the rest of the dimensions
of the clutch. In order to do that, we have to check the contact between the spring and the pressure
plate, as it is shown in the following picture.

The external radius has to be then greater or equal than 120.29 mm, so we have decided to use an
external spring diameter of 280 mm (Rspring,e = 140 mm).
On the other hand, we have a constraint from the project request of having a minimum internal
diameter of 65 mm, so we have set it to 70 mm.
Once we have defined these limits, we can proceed with the evaluation of the optimum dimensions
for the diaphragm spring.
To understand which are the affecting parameters and how we need to modify them, the following
graph can be used.
It represents the reaction force exerted by different springs as a function of the ratio ‘h/t’, where ‘h’
is the spring height and ‘t’ is the spring thickness.
Moreover, curves are parameterized with respect to the aspect ratio ‘h0/t’, being ‘h0’ the unloaded
spring height.

14
In addition, as long as the diaphragm spring is a revolution solid, we can divide it in several
portions of equal shape (called ‘fingers’).
In our case, we have chosen 18 fingers for the spring.

Shell model
The diaphragm spring can be modelled in by using just one finger (one portion of the spring), due to
the fact that all fingers have same dimensions and geometry.
In order to draw the finger, we have used ‘axisymmetric modelling space’ of type ‘deformable’ and
‘shell base feature’ in the panel ‘create part’.
Notice that we have created a partition that represents the region where takes place the contact
between the spring and the pressure plate.

Then, with the command ‘cut/extrude’ we have made a hole. It has been decided to use an ellipse
with vertical axis of 10 mm and a horizontal axis of 20 mm, which centre is placed at 100 mm from
the main axis of the clutch. The intermediate space between the two branches of the spring has been
set to 4 mm.
The following image represents what we have obtained:

15
At this step, we have assigned the material, which is a steel with yield stress equal to 1400 MPa,
and selected the thickness of the spring, which has been found to be suitable at 2.5 mm.
Additionally, we activate in the ‘step’ panel the option ‘NLGEOM’, corresponding to a non-linear
analysis. This is done because the diaphragm spring is characterized by a non-linear behaviour, so
the assumption of small displacements does not apply for its Finite Element analysis.
The interactions used for the model are the coupling between the internal diameter and a reference
point RP-1 placed on the clutch axis, and another coupling between a generic point RP-2 and the
partition line where the clutch is in contact with the pressure plate.

At the same time, in the ‘loads’ panel we have defined a vertical displacement of the point RP-1 of
10 mm downwards (constraining also all other possible motions), while RP-2 is constrained in all
directions (because that part of the spring will be fixed when mounted in the clutch).
We have also used the boundary condition ‘ZSYMM’ on the two oblique edges of the finger, which
blocks the motion in lateral direction:

16
For the mesh we have used again Q8 elements.
It has been reduced the dimension of the elements around the hole and near the internal diameter of
the spring in order to improve the stresses evaluation during the analysis ( because those zones will
be the most stressed ones).

Running the analysis, the following distribution of stresses is obtained:

17
Convergence analysis

Convergence analysis for shell model

Mesh # nodes STRESS,max [MPa]
1 385 1672
2 865 1659
3 2504 1651
4 5813 1658
5 11592 1662
6 17915 1664

After the convergence analysis, we can observe that the final stress (1664 MPa) is close, but greater
than the yield stress of the material (1400 MPa).
We have tried to find suitable parameters for the model in order to obtain an acceptable value of the
maximum stress (that means lower than 1400 MPa), trying also to maintain at the same time a good
force-displacement curve, but it has not been possible.
However, playing with the dimensions of the spring, we have realized that a good way to reduce the
stress could be to increase the external diameter as much as possible and/or reduce the unloaded
spring height ‘h0’.

Force – displacement graph

The relation obtained between the reaction force exerted by the spring and its displacement is:

18
TBN: Abaqus gives us ‘time’ as abscissa, but actually it is displacement
In order to verify that the torque is transmitted successfully, we have to divide the axial force found
in the first part of the project (for both the constant pressure and the constant wear conditions) by
the number of fingers (set to 18) and then verify if it is lower than the corresponding maximum
force exerted by the one finger of the diaphragm spring (from force-displacement curve shown
above).
The total variation of the displacement between the new linings and worn linings is 1.4 mm (each
lining in reduced of 0.7 mm thickness in the inner part). We select then two points that ensure the
transmission of the torque in each condition:

1. Constant pressure
Ftotal = 3265.3 N → Ffinger = 181.41 N
Time = 0.80 s (= 8.0 mm)* → Fgraph = 191.59 N > 181.41 N
*(Remember the total amount of displacement we applied to the inner diameter of the spring was 10 mm)

2. Constant wear
Ftotal = 3282.3 N → Ffinger = 182.35 N
Time = 0.66 s (= 6.6 mm) → Fgraph = 190.15 N > 182.35 N

Consequently, the force generated by the diaphragm spring is enough to ensure the correct
transmission of the torque in the different conditions of the linings.
Moreover, we have been able to find a force-displacement curve within the working range that is
almost constant and thus we are able to guarantee a pressure distribution almost uniform.

To conclude this part, here are shown some examples of other configurations we tried for the design
of the diaphragm spring. It has been changed the initial height and the thickness of the spring, and
the shape of the hole, until the ‘optimum’ solution explained before has been found.

19
20
3D model
The procedure to draw the 3D model of the diaphragm spring is very similar to the one of the shell
model already explained.
The main difference appears in the ‘create part’ panel, where we have selected ‘3D solid revolution
base feature’.

Another small variation from the shell model to the 3D model are the constraints, because now we
have to take into account the thickness. When we make the coupling with RP-1, we have to connect
it with the upper side of the spring (where the force is applied), while the coupling with RP-2 is
done in the lower side of the spring (where there is the contact with the pressure plate).

21
The obtained graph representing the reaction force exerted by the spring with respect to the
displacement is the following one, once we have finished the analysis:

Observing the force at the graph for the conditions of constant pressure (time=0.80 s) and constant
wear (time=0.66 s), we can see that the values have been reduced with respect to the shell model,
but the spring is still able to transmit the torque (183.23 N in the new clutch and 184.97 N in the
worn clutch).

On the other hand, we can see that the distribution of stresses provided by the analysis of the 3D
model (using brick elements), has its maximum stress at the internal diameter of the diaphragm
spring.
We have done the convergence analysis and we have found a value of stress very high with respect
to the yield stress.

22
 Convergence analysis for 3D model
Mesh # nodes STRESS,max [MPa]
1 991 1943
2 2537 2656
3 7868 6093
4 19052 8599
5 65221 13275
6 84974 15846
7 98275 16379
8 117865 16440

In this analysis, we have assumed that the axial force is transmitted with a coupling between a point
RP-1 on the revolution axis and the upper edge of the internal diameter of the spring, and that is the
reason why the stresses found are so high with respect to the yield stress.
We have to consider that in a real case the force is transmitted by the use of a release bearing, so the
force is not applied on the edge, but within a certain contact surface on the top of the fingers end,
providing a stress distribution with a lower peak stress. 

23
20/02/15 15.27 C:\Users\Marco\D...\Project_LASTVERSION.m 1 of 3

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%%%%%%%% PROJECT: Preliminary Design of a Mechanical Clutch %%%%%%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%% Hari Dileep Reddy - 820331 %%%%%%%%%%%%%%%%%%%%%%%%

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

clear all
close all
clc

% PART 1:

% Computation of the required axial force and definition of the disk compliance

T = 300000; % [N*mm] ->

minimum required torque

mu = 0.4; % ->

friction coefficient

k = 2; % ->

number of disk surfaces

R_e = 128.5; % [mm] ->

external disk radius

R_i = 100; % [mm] ->

internal disk radius

disp('Axial force (N) - Constant pressure hypothesis (new

clutch)') F_press = 3*T*(R_e^2-R_i^2)/(2*k*mu*(R_e^3-R_i^3))

disp('Axial force (N) - Constant wear hypothesis (worn

clutch)') F_wear = 2*T/(k*mu*(R_e+R_i))
% Convergence analysis of the new clutch model

C_new = [1263 0.2724

1643 0.2710

2755 0.3137

4068 0.3184

5520 0.3236

10891 0.3298

26623 0.3406

46251 0.3455]

figure
plot(C_new(:,1),C_new(:,2))
xlabel('Number of nodes' )
ylabel('Maximum pressure
[MPa]') ylim([0.26 0.35])

% Convergence analysis of the new clutch model

20/02/15 15.27 C:\Users\Marco\D...\Project_LASTVERSION.m 2 of 3

C_worn = [976 0.4195

1223 0.4406

1585 0.4731

2408 0.4583

3557 0.5195

4672 0.5142

8895 0.5053

21089 0.5014]

figure
plot(C_worn(:,1),C_worn(:,2))
xlabel('Number of nodes' )
ylabel('Maximum pressure
[MPa]') ylim([0.41 0.53])

% Actual transmitted force and torque in the new clutch

p_med_press = 0.221; % [MPa] ->

Average contact pressure for constant pressure

disp('Actual transmitted force (N) - new clutch')

F_press_actual = 2*pi*p_med_press*(R_e^2-R_i^2)/2

disp('Actual transmitted Torque (Nmm) - new clutch')

T_press_actual = 2*pi*k*mu*p_med_press*(R_e^3-R_i^3)/3

% Actual transmitted force and torque in the worn clutch

p_med_wear = 0.236; % [MPa] ->

Average contact pressure for constant wear

disp('Actual transmitted force (N) - worn clutch')

F_wear_actual = 2*pi*p_med_wear*(R_e^2-R_i^2)/2

disp('Actual transmitted Torque (Nmm) - worn clutch')

T_press_actual = 2*pi*k*mu*p_med_wear*(R_e^3-R_i^3)/3
% Axial equivalent stiffness

E = 0.8; % [MPa]

A = pi*(R_e^2-R_i^2); % [mm^2] ->

area

t = 2.5; % [mm] ->

thickness of the spring

disp('Axial equivalent stiffness')

K_axial = A*E/t % [N/mm]

% Torsional equivalent stiffness

nu = 0.5; % ->
20/02/15 15.27 C:\Users\Marco\D...\Project_LASTVERSION.m 3 of 3

Poisson's coefficient for the spring

G = E/(2*(1+nu)) % [MPa] ->

Shear modulus

a = 28.5; % [mm] ->

disk width

J = a*t^3/3 % [mm^4] ->

moment of inertia

L = 2*pi*((R_e+R_i)/2) % [mm] ->

mean spring circumference

disp('Torsional equivalent stiffness')

K_torsional = G*J/L % [N*mm/rad]

% PART 2:

% Design of the diaphragm spring

% Convergence of the shell model

C_shell = [385 1672

865 1659

2504 1651

5813 1658

11592 1662

17915 1664]

figure
plot(C_shell(:,1),C_shell(:,2))
xlabel('Number of nodes')
ylabel('Maximum stress [MPa]')
ylim([1650 1675])
% Convergence of the 3D model

C_3d = [991 1943

2537 2656

7868 6093

19052 8599

65221 13275

84974 15846

98275 16379

117865 16440]

figure
plot(C_3d(:,1),C_3d(:,2))
xlabel('Number of nodes')
ylabel('Maximum stress
[MPa]') ylim([1500 17000])
20/02/15 15.28 MATLAB Command Window 1 of 3

Axial force (N) - Constant pressure hypothesis (new clutch)

F_press =

3.2653e+03

Axial force (N) - Constant wear hypothesis (worn clutch)

F_wear =

3.2823e+03

C_new =

1.0e+04 *

0.1263 0.0000

0.1643 0.0000

0.2755 0.0000

0.4068 0.0000

0.5520 0.0000

1.0891 0.0000

2.6623 0.0000

4.6251 0.0000

C_worn =
 1.0e+04 *

0.0976 0.0000

0.1223 0.0000

0.1585 0.0000

0.2408 0.0000

0.3557 0.0001

0.4672 0.0001

0.8895 0.0001

2.1089 0.0001

Actual transmitted force (N) - new clutch

F_press_actual =

4.5214e+03

Actual transmitted Torque (Nmm) - new clutch

T_press_actual =

4.1540e+05

Actual transmitted force (N) - worn clutch

F_wear_actual =
20/02/15 15.28 MATLAB Command Window 2 of 3

4.8283e+03

Actual transmitted Torque (Nmm) - worn clutch

T_press_actual =

4.4359e+05

Axial equivalent stiffness

K_axial =

6.5468e+03

G =

0.2667

J =

148.4375

L =

717.8539

Torsional equivalent stiffness

K_torsional =

0.0551

C_shell =

385 1672

865 1659

2504 1651

5813 1658

11592 1662

17915 1664

C_3d =

991 1943

2537 2656

7868 6093

19052 8599

65221 13275

84974 15846
20/02/15 15.28 MATLAB Command Window 3 of 3

98275 16379

117865 16440

>>