Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

SAE TECHNICAL
PAPER SERIES 2007-01-0264

Design and Development of the Valve Train

for a Racing Motorcycle Engine
Phil Carden and Ken Pendlebury
Ricardo UK

Naji Zuhdi
Petronas Malaysia

Andrew J G Whitehead
Del West USA

Reprinted From: New SI Engine and Component Design and Engine Lubrication
and Bearing Systems
(SP-2093)

2007 World Congress

Detroit, Michigan
April 16-19, 2007

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-0790 Web: www.sae.org
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

By mandate of the Engineering Meetings Board, this paper has been approved for SAE publication upon
completion of a peer review process by a minimum of three (3) industry experts under the supervision of
the session organizer.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.

For permission and licensing requests contact:

SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: [email protected]
Fax: 724-776-3036
Tel: 724-772-4028

For multiple print copies contact:

SAE Customer Service

Tel: 877-606-7323 (inside USA and Canada)
Tel: 724-776-4970 (outside USA)
Fax: 724-776-0790
Email: [email protected]

ISSN 0148-7191
Copyright © 2007 SAE International
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE.
The author is solely responsible for the content of the paper. A process is available by which discussions
will be printed with the paper if it is published in SAE Transactions.

Persons wishing to submit papers to be considered for presentation or publication by SAE should send the
manuscript or a 300 word abstract of a proposed manuscript to: Secretary, Engineering Meetings Board, SAE.

Printed in USA
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

2007-01-0264

Design and Development of the Valve Train for a Racing

Motorcycle Engine
Phil Carden and Ken Pendlebury
Ricardo UK

Naji Zuhdi
Petronas Malaysia

Andrew J G Whitehead
Del West USA

Copyright © 2007 SAE International

ABSTRACT The development of high speed valve trains has a long

history and has many known problems, pitfalls and
This paper describes the design and development of a conflicting objectives (see References 1, 2 and 3). Even
direct acting valve train for high speed operation in a a simple direct acting valve train has a bewildering
racing motorcycle engine. At the outset of the project the number of design variables and constraints. Traditionally
engine speed limiter was set to 14000 rpm and this was valve train designs evolved following prolonged
eventually raised to 16000 rpm. The paper covers the prototype testing. The approach of the Engine
evolution of the design and includes descriptions of the Development Team was to rely on advanced analysis
components including camshaft, tappet, shim, retainer, tools and to support decisions with minimal testing. This
valve and valve springs. Valve train dynamic analysis paper forms a report on the valve train development.
software was used for the following tasks.

x Assessment of the influence of the changed parts on COMPONENT DEVELOPMENT HISTORY

valve train dynamics and durability
x Design of new cam profiles OVERVIEW
x Setting speed limit for each build level
x Investigation of failures The rules of World Superbikes did not limit the engine
speed for a 900cc 3-cylinder engine so the Engine
These activities are covered in this paper. Development Team hoped to increase engine power by
increasing the maximum operating speed of the engine.
The valve train was identified as the limiting factor for
INTRODUCTION engine speed and so the main objective of the valve
train development became the need for increased
Petronas began racing the FP1 motorcycle in the World engine speed. A target of 16000 rpm was set. This
Superbikes series in 2003. From 2004 until 2006 corresponds to a challenging mean piston speed of
Petronas and Ricardo worked together to improve the 3- 26.29 m/s for the FP1 with stroke of 49.3 mm.
cylinder 900cc engine of the FP1 despite the World
Superbikes rule change that permitted other teams to The rules of World Superbikes prohibit changes to major
use 1000cc 4-cylinder engines. Petronas decided castings and so it was impossible to change the valve
against changing the FP1 engine to 1000 cc due to train type from direct acting to the pad finger follower
costs of re-homologation and so the bore (88.0 mm) and (which has potential for lower effective mass and is used
stroke (49.3 mm) could not be changed during the on most Formula 1 engines). The use of pneumatic
project. springs was also prohibited. Thus it was necessary to
reduce the mass of all the valve train moving parts to the
During the project it became obvious that an increase in minimum required for durability.
engine speed was required to raise the power output
and to give a wider power band. The following sections summarise the changes made to
each component.
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

CAMSHAFTS However the reduction in second moment of area

resulting from this 2.25 mm diameter hole in the 5.0 mm
The camshafts were machined from steel 16NiCr11. The diameter valve stem was just 3.5% so the hollowing
cam bearing journals and cam lobes were hardened to a procedure proved a reliable way to save 1.4 g.
depth of 1.6 mm and then carbonitrided to give a final
surface hardness of 40-42 HRc. The bearing journals The second design iteration on the intake valve was to
were then lapped. reduce the mass further by changing the form of the
back of the valve head from a 40 mm radius to an 11o
The baseline camshafts had outside diameter of 24.5 back angle. This gave a further mass reduction of 2.1 g.
mm between lobes and inside diameter of 16 mm. This technique does reduce the stiffness and strength of
During the project the inside diameter was increased to the valve and this can give problems with valve seat
17 mm and the timing gear mounting flange was wear or even valve failure. In addition the valve
scalloped as shown in Figure 1. These measures geometry change can affect the flow characteristics of
resulted in an 8.4% reduction in mass and a 10.3% the charge on entry to the cylinder. However, in this
reduction in rotating inertia of the camshafts. case the valves proved durable and the extra engine
speed resulting from the significant mass reduction
outweighed any loss in flow.

The basic dimensions of the valves are shown in Table 1

and these were unchanged except for a slight reduction
in length to suit the change from lash cap to lash disc.

Table 1 Valve dimensions

Parameter Baseline Final
Intake valve head diameter (mm) 36.0 36.0
Intake valve stem diameter (mm) 5.0 5.0
Intake valve length (mm) 93.7 93.2
Exhaust valve head diameter (mm) 30.0 30.0
Exhaust valve stem diameter (mm) 5.0 5.0
Figure 1 Camshaft end flange Exhaust valve length (mm) 95.0 94.5

VALVES TAPPET

The baseline valves (intake and exhaust) were made The tappets were made from a through hardened tool
from titanium alloy Ti 6242+0.2Si. The intake valves steel (H11), heat treated to 51-54 HRc, hard turned,
were solution treated and aged to give a tightly packed ground and then polished to around 0.05 Ra prior to
+ (alpha+beta) grain structure, exhibiting high application of a diamond-like carbon (DLC) coating
hardness (42-45HRc) and excellent fatigue and creep developed. The DLC coating was approximately 3 Pm
resistance at intake valve temperatures (up to 500qC). thick and was applied by a plasma-assisted chemical
The exhaust valves were heat treated and aged to give vapor deposition process following surface preparation.
a lamella alpha beta structure within prior beta grains, This coating was extremely hard (2500 Hv) and was
exhibiting fatigue and creep resistance up to 800qC. used to minimise friction and wear.

Both valves had a molybdenum alloy coating, plasma The baseline tappet was of traditional design and the
sprayed on the stem to minimize friction and prevent first design iteration involved the introduction of large
wear of the valve guide. The valve tip was protected by slots into the skirt of the tappet to reduce mass (see
a Cr2C3 NiCr cermet coating applied via a high velocity Figure 2). This concept had been previously subject to
oxy-fuel flame process. Further to these coatings a thin limited development but had not been used because
film plasma vapor deposition CrN coating was tested on most very high speed engines had switched to finger
the seat of the intake valve to address a valve seat follower valve trains for lowest mass. However, the
degradation problem. The coating performed well but it durability of the slotted tappet was proven quickly and
was determined that the seat wear was due to intake resulted in a significant mass reduction of 4.9 g.
debris being ingested and so this coating was not
adopted for race production parts.

The valve supplier had developed a technique to reliably

produce hollow valves with stem diameter as small as
4.5 mm. In an effort to reduce the mass of the intake
valve this method was applied to the FP1 intake valve.
Assuming the valve train was under control the main risk
of failure from this modification was valve stem bending.
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

Table 3 Spring dimensions

Parameter Baseline Final
inner outer inner outer
Wire diameter (mm) 2.90 3.95 2.90 3.7
Mean coil 17.9 24.75 17.1 23.7
diameter (mm)
Fitted length (mm) 32.6 32.6 28.75 28.75
Fitted force (N) 153 298 130 279

SPRING RETAINER, SHIM AND COLLETS

The spring retainers were machined from maraging steel

Figure 2 Slotted tappet skirt (C300) and nitrided to achieve hardness of 61-64 HRc
on the mating surfaces for the springs. A through
A second iteration with reduced crown thickness and hardened steel lash cap was located on the top of each
reduced wall thickness was introduced (see Table 2) valve stem. This design resulted in several problems as
and this resulted in a further mass reduction of 2.6 g. the speed of the engine was increased (see later
These parts were tested and proved to have adequate sections) and so the components were redesigned to
durability although they did occasionally exhibit some improve durability and to reduce mass.
signs of wear of the DLC probably due to flexing of the
tappet crown. The first design iteration involved a move away from the
lash cap concept to a lash disc held captive in a revised
Table 2 Tappet dimensions spring retainer (see Figure 3). This allowed for a
Parameter Baseline Final reduction in mass, an improvement in durability of the
Outside diameter (mm) 32.0 32.0 retainer and elimination of valve tip wear. C300 was
Overall length (mm) 23.5 23.5 used for the revised retainer and the lash disc was made
from cold worked die steel, through hardened to HRc 57-
Central crown thickness (mm) 3.3 3.0
61, then ground and polished.
Minimum wall thickness (mm) 0.8 0.7
Slots in skirt No Yes

SPRINGS

The springs used throughout the project were made

from steel. The spring manufacturers were very
protective regarding materials, surface treatments and
fatigue strength data so no further data can be provided.
The baseline springs had been designed for a rated
speed of 13500 rpm and problems soon emerged as
engine speed was raised (see later section).

Springs from several different suppliers were procured

and tested but eventually the spring supplier was
changed to NHK who provided a spring pack of their
design to meet a target maximum speed of 15500 rpm
for a cam profile with 12 mm peak lift and a target life of
3 million cycles. The outer spring was given a right hand
coil and the inner spring had a left hand coil. The coils Figure 3 Sections through baseline and final intake
were designed with a target interference of 0.1 mm and valve trains
the ends were ground and carefully chamfered. This
spring pack was used for most of the project but during The revised spring retainer had different diameters to
the final season a further optimized spring pack (see suit the final spring pack design. The position of the
Table 3) was designed to suit the cam profile with collet groove relative to the valve tip was changed to suit
reduced peak lift of 11 mm and to give a higher rated the new retainer design as shown in Figure 3 but the
speed of 16000 rpm. collets remained unchanged during the project.
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

COMPONENT MASSES

The changes made to component masses are

summarised in Table 4.

Table 4 Valve train mass

Mass (kg) Baseline Final
Intake valve 0.0243 0.0208
Exhaust valve 0.0222 0.0214
Tappet 0.0335 0.0260
Shim (cap or disc) 0.0010 0.0008
Collets 0.0002 0.0002
Retainer 0.0080 0.0072 Figure 5 Kinematic valve acceleration against crank
Outer spring effective mass 0.0115 0.0093 angle
Inner spring effective mass 0.0062 0.0051
Intake valve train 0.0847 0.0694 Table 5 shows some important kinematic parameters for
effective mass the baseline and final designs. The peak lift was reduced
Exhaust valve train 0.0826 0.0700 by 1 mm and the period was increased slightly. The
effective mass choice of peak valve lift and period was obviously a
compromise and the lift was reduced to enable the high
The final valve train is shown in Figure 4. speed operation. The final choice was guided and
supported by performance simulation and by extensive
engine performance testing on a series of cam profiles.
The reduction in peak lift entailed a corresponding drop
in intake valve L/D ratio but it was possible to slightly
increase the lift area integral (a non-dimensional
parameter defined as the area under the lift curve
divided by the theoretical maximum area under the lift
curve).

Table 5 Kinematic parameters

Parameter Baseline Final
Peak kinematic valve lift L 12.0 11.0
(mm)
Inner seat diameter D (mm) 35.0 35.0
L/D 0.343 0.314
Figure 4 Section through final valve train Lift area integral 0.555 0.557
Period – top of ramp (deg) 307.2 310.0
Ramp height (mm) 0.20 0.20
VALVE TRAIN DESIGN ANALYSIS Ramp velocity (m/s) 0.432 @ 0.500 @
14000 rpm 16000 rpm
The valve train analysis software had a kinematics Valve acceleration on 29818 @ 33404 @
solver (used to design cam profiles and calculate opening flank (m/s2) 14000 rpm 16000 rpm
pseudostatic forces and oil film thickness etc) and a
Valve acceleration on cam 11530 @ 13305 @
dynamics solver (used to determine dynamic valve
nose (m/s2) 14000 rpm 16000 rpm
motion, dynamic forces and spring surge vibration).
Valve acceleration on 36962 @ 41554 @
The results of the analysis of the intake valve train, closing flank (m/s2) 14000 rpm 16000 rpm
comparing the baseline design with the final design, are Opening side acceleration 2.51 2.51
presented in this section. The larger valve and more ratio
aggressive, higher lift cam profile conspired to make the Closing side acceleration 3.21 3.12
intake valve train the worst case in almost every respect ratio
and so results are shown for intake only.
The ramp height was maintained at 0.2 mm and the
Figure 5 shows a comparison between the kinematic ramp kinematic velocity was also maintained at the
acceleration of the intake valve for baseline and final same value despite the increase in engine speed and
designs. the corresponding increase in actual ramp velocity
indicated in the table. The peak kinematic valve
accelerations (nose and flanks) were decreased by 13-
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

16% as shown in Figure 5 but the dynamic peak characteristic shape associated with direct acting valve
acceleration values were increased by 12-15% due to trains for baseline and final valve trains. The film
the increase in engine speed. Acceleration ratios were thickness is high on the flanks and low over the nose.
maintained at similar values. The lubricant film thickness at the nose is often
assessed by using the Deschler and Wittman number
The durability of the cam/tappet contact was assessed in (see Reference 2). This is usually in the range 0.15 to
terms of contact stress, lubricant film thickness and 0.25. For the final design this upper limit was exceeded
tappet edge clearance (the proximity of the cam/tappet slightly but no problems occurred.
contact line to the edge of the tappet). The results are
shown in Table 6. At the flank/nose transition the film thickness passes
through a very low region as the lubricant entrainment
The highest values of contact stress occurred at low velocity passes through zero. The lubrication in this
engine speed at the nose of the cams but racing engines region is assessed by consideration of the number of
spend very little of their lives at low speed. The final consecutive crank degrees at which predicted oil film
design had a larger nose radius and so correspondingly thickness is less than 0.1 Pm and this is usually
lower contact stress. At high engine speed this situation expected to be less than 10 deg.
was reversed (see Figure 6) and the final valve train
experienced higher contact stress than baseline at rated
speed.

The cam/tappet contact stress over the nose was also

calculated using dynamic forces.

Table 6 Cam/tappet durability

Parameter Baseline Final
Peak cam tappet contact 831 @ 764 @
stress at idle (N/mm2) 3500 rpm 3500 rpm
Peak cam tappet contact 400 @ 436 @
stress at rated speed 14000 rpm 16000 rpm
(N/mm2)
Figure 7 Lubricant film thickness at cam/tappet
Lubricant film thickness at 0.295 0.278
contact against crank angle
peak cam lift (Pm)
Deschler and Wittman 0.207 0.272 For the baseline cam profile with high lift the tappet edge
number at peak lift clearance was very low (0.3 mm) but as the lift was
Maximum number of 8.26 7.86 reduced this increased to a value higher than necessary.
consecutive crank degrees Reduction in tappet diameter was considered but in fact
at which oil film thickness is this was limited by the need for clearance between the
less than 0.1 Pm at rated tappet and the outer spring and so the potential benefit
speed was small.
Minimum tappet edge 0.30 1.90
clearance (mm) The valve trains were also assessed in terms of their
dynamic performance. This software used began as a
dedicated valve train dynamics package and although it
has expanded to have extensive multi-body dynamics
capability it retains exceptional speed and ease-of-use
for valve train dynamic analysis. The model used for
assessment of valve train dynamics is shown in Figure
8.

Figure 6 Pseudostatic cam/tappet contact stress at

high engine speed

The lubricant film thickness is shown plotted against

crank angle in Figure 7. This graph shows the
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

Figure 9 Valve seating velocity against engine speed

At low speeds the valve seating velocity was controlled

by the closing ramp on the cam but as engine speed
increased this control was progressively lost. The
Figure 8 Single valve line model baseline design showed an increase in seating velocity
from ~13500 rpm and a sharp transition to very high
The cam node was suspended on a stiffness element seating velocity at ~14800 rpm. During the project, as a
representing camshaft bending stiffness and camshaft result of engine testing and cylinder head rig testing, it
bearing support stiffness. The tappet top stiffness was was discovered that the titanium intake valves would fail
modeled as a function of eccentricity of the cam/tappet suddenly at the valve stem below the retainer if valve
contact. The valve stem was generally modeled as a seating velocity exceeded 4 m/s. This information, with a
single axial stiffness although multiple stiffness and suitable safety margin, was used to set the engine
mass models were tried. Valve seat and valve head speed limiter for each valve train build. For the final
bending were also modeled as a single stiffness. The design the valve seating velocity did not exceed 2 m/s
valve, spring retainer, shim and collets were modeled as until the engine reached speeds in excess of 16500 rpm
a single lumped mass. Each valve spring was modeled and did not experience a sharp transition to very high
as a series of lumped masses (8 per coil) connected by seating velocity at engine speeds below 17000 rpm. This
stiffnesses using a special macro element that accounts characteristic proved very significant as the valve train
for coil clash effects as the spring closes and loss of was able to survive the inevitable over-rev incidents that
contact between spring ends and mating parts. occur during racing. It was interesting to note that the
valve seating velocity was not very strongly dependent
The level of damping due to interference between valve on the assumed damping due to interference between
springs is intimately dependent on the fit between the the springs.
springs (which changes as the springs are compressed)
and so is very difficult to model explicitly. For this project Further insight into intake valve closing can be obtained
the approach taken was to make two analysis runs to by observation of Figure 10 which shows the valve
assess each design; one with high damping (assumed closing event for the baseline design at 9000 rpm, 13000
20% of critical damping) and one with very low damping rpm and 15000 rpm.
typical of independent springs with no interference (0.5%
of critical damping). The sensitivity of all aspects of
system dynamics to damping was thus considered at
every stage.

The software was used to calculate the dynamic

response of the valve train in the high speed range and
the results were mainly presented in terms of plots of
critical parameters against engine speed. Figure 9
shows a plot of valve seating velocity which was used to
assess the severity of the valve closing event.

Figure 10 Valve lift at closing against crank angle

 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

At 15000 rpm the valve hit the seat before the top of the Cam/tappet separation showed more sensitivity to spring
ramp, seating velocity was high and initial contact was interference damping than valve seating phenomena but
followed by a large valve bounce. The height of this first the general conclusions were not changed. The baseline
bounce can also be plotted against engine speed and design exhibited a significant loss of contact between
the results are shown in Figure 11. cam and tappet over the cam nose at speeds in excess
of ~14500 rpm while for the final design separation did
not exceed ~0.2 mm at speeds below 17000 rpm.

Spring surge is a well-known problem associated with

high-speed coil-sprung valve trains. The springs are
excited by the harmonic content of the cam profile and
may continue to vibrate following valve closure, possibly
affecting the motion during the next valve event. The
team quantified this phenomena by plotting the
amplitude of the residual vibration of the centre coil of
each spring just after valve closure as shown in Figures
13 and 14.

Figure 11 Valve bounce height against engine speed

The results show a very similar pattern to the valve

seating velocity results. Typically The team regarded
bounce height in excess of 0.1 mm as unacceptable and
this limit correlated reasonably well with the limit
established for valve seating velocity.

The next phenomena considered during dynamic

analysis was valve jump. This was characterized by loss
of contact between the cam and the tappet during the
valve event when inertia force (including vibration)
Figure 13 Outer spring surge against engine speed
exceeds available spring force. Figure 12 shows a plot
of maximum distance between cam and tappet against
engine speed.

Figure 14 Inner spring surge against engine speed

The spring surge amplitude for the baseline springs was

Figure 12 Cam/tappet separation against engine very high (~1 mm expected for production engines but
speed higher values can be tolerated for racing) and this
contributed to limited high speed dynamic performance.
The surge amplitude for the final design was greatly
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

reduced by improved matching between cam and spring

characteristics. The surge amplitude was moderately
sensitive to spring interference damping.

One of the problems associated with high spring surge

amplitude was spring seat hammering. The spring
vibration leads to loss of contact between spring and
seat followed by large impact forces as contact is re-
established (for example see Figure 15 at end of event).

Figure 17 Maximum force at bottom of outer spring

against engine speed

The final cam profile and spring design showed

significant improvement in valve spring seat force
characteristic.

The software was also used to calculate dynamic stress

in the springs. Stress at the worst case location on the
Figure 15 Force at bottom of outer baseline spring baseline outer spring is shown plotted against crank
against crank angle angle in Figure 18 and stress range is plotted against
engine speed in Figure 19. Although the spring fatigue
This problem manifested itself as breakage of spring end strength was not known the effect of valve train
tangs with some intermediate spring designs during this dynamics on spring stress was quantified and this
project (see later section) and so plots of minimum and proved useful during the project as a basis for
maximum spring force as shown in Figures 16 and 17 comparison between designs.
were made routinely to assess the risk of this problem.

Figure 16 Minimum force at bottom of outer spring Figure 18 Stress at worst case location in baseline
against engine speed outer spring against crank angle
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

Figure 20 Whole engine rotational dynamics model

Figure 19 Stress range at worst case location in

baseline outer spring against crank angle

Some results from spring analysis are given in Table 7.

Table 7 Spring cover factor and stress

Parameter Baseline Final
Pseudostatic spring cover 1.21 @ 1.39 @
factor at rated speed 14000 rpm 16000 rpm
Shear stress in outer 298 fitted 410 fitted
spring -pseudostatic 1065 max 1294 max
(N/mm2)
Shear stress in inner 356 fitted 291 fitted
spring -pseudostatic 1066 max 1199 max
(N/mm2)

Pseudostatic spring stress levels were increased for the

final design but spring quality was such that spring
fatigue failures were not a significant problem. Figure 21 Comparison of intake valve seating results
from single valve model and whole engine model
An analysis model of the whole engine was created (see
Figure 20). Measurements of torsional vibration
displacement were made at each gear in the timing drive
and the values of stiffness, damping and clearance in VALVE TRAIN TESTING
the model were adjusted to give reasonable correlation
with measured data. The model was used to investigate A cylinder head test rig was used for durability testing of
the following. valve train variants during the later stages of the project.
In particular, testing of alternative valve spring packs
x Effects of crankshaft dynamics and timing drive proved very valuable (see later section). The rig was
dynamics on valve motions also used to measure valve train friction power loss.
x Dynamic loads on timing gears
x Dynamic torques at gear fasteners The idler gear on the timing drive was driven by an
electric motor via a drive shaft with a flexible element
A typical plot comparing intake valve seating velocity, as and the cylinder head was supplied with lubricating oil at
calculated using the single valve line model of the final 100qC and coolant at 80qC.
valve train, with predicted values for each intake valve
train from the whole model is shown in Figure 21. It is The rig was generally operated at steady state speeds
interesting to note that in this case (3 cylinder engine for predefined periods.
with geared timing drive) although the valve motions
were obviously affected by the timing drive dynamics the
magnitude of the effect was not very significant. It was
thus possible to use a single valve train analysis model
to set speed limits for valve train builds.
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

VALVE TRAIN FAILURE MODES

TAPPET BORE FAILURE

The first major problem to occur as the speed of the

engine was increased was the failure of tappet bores.
The cylinder head cracked and the tappet bore structure
broke off on several engines as shown in Figure 22.
Use of die penetrant on used cylinder heads revealed
that the cracks began at the fillet radius in the slot
machined to provide clearance for the cam lobe as
shown in Figure 23.

Figure 24 Stress concentration in tappet bore slot

VALVE TIP WEAR

The lash cap used to control valve clearance on the

baseline valve train resulted in sporadic examples of
high wear rate at the valve tip surface as shown in
Figure 25.

Figure 22 Tappet bore failure

Figure 25 Valve tip wear

This was thought to be due to inadequate control of

tolerances on the bore of the lash cap leading to variable
fit between the cap and the valve. Large diametral
clearance between these components led to tipping of
Figure 23 Crack in tappet bore the cap and edge loading of the valve tip. This problem
could probably have been solved by improving the fit
The software was used to calculate the pseudostatic and surface finish in the lash cap but it was eventually
contact force between the cam and the tappet and the eliminated by the change of design to the lash disc.
eccentricity of the contact point on the tappet top. These
values were then used to calculate the tipping moment
VALVE STEM FAILURES
acting on the tappet which in turn was used to calculate
the reaction forces between the tappet side and the
As engine speed was increased a series of failures of
tappet bore at the top and bottom of the tappet. These
the intake valve stems occurred. The stems typically
forces were applied to a local finite element model of the
broke just below the bottom of the spring retainer, at the
cylinder head in the tappet bore region (see Figure 24),
end of the collet contact area, as shown in Figure 26.
low safety factors were confirmed and various schemes
for improvement were investigated. Eventually the fillet
radius was increased and this failure was eliminated.
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

The transition to damaging level of seating velocity

occurred at ~300 rpm lower speed and the resulting
impacts were strong enough to break the valve stems.
The problem was resolved by changing the spring
supplier and introducing an improved spring design for
high speed operation. The relaxation of the final springs
was greatly reduced even though they were subjected to
higher stresses as shown in Table 7. Fitted force was
typically reduced by less than 3% if the engine was run
at less than 16000 rpm and this rose by up to 10%
following prolonged operation at 16000 rpm.

VALVE SPRING END TANG FAILURES

The most important valve train failure mode was valve

stem failure following spring relaxation (as described
previously) and with this in mind a procedure for
Figure 26 Valve stem failure evaluation of alternative spring designs on the cylinder
head rig was developed.
The failure surface indicated pure tensile fatigue so the
applied loads were probably far higher than expected. Springs were subjected to periods of constant speed
Previous experience suggested that loss of valve train operation followed by measurement of spring pack fitted
control could be to blame despite the fact that analysis force and further testing at increased speed as
suggested the dynamics should be acceptable at the illustrated in the typical test results graph shown in
failure speed. This led to a general re-investigation of Figure 28.
the valve train dynamics.

Eventually the root cause was traced to the very large 18000 100
relaxation of the baseline valve springs. Spring
90
relaxation (or loss of spring force) following use is known
to be dependent on stress level, time and temperature 16000
80
and typically it results in a reduction in fitted force of
~5%. Measurement of spring force/displacement curves

Spring relaxation force (N)

70
before and after use revealed that the fitted force of the 14000
Engine speed (rpm)

baseline springs was reduced by up to 40% following period 1 60

high speed operation. Figure 27 shows the effect of a period 2
12000 50
20% reduction in spring fitted force on the valve seating period 3
velocity. Loss of control of the valve spring can also Fitted relaxation 40
result in application of bending loads on the valve stem
10000
and this may also have contributed to the failure. 30

20
8000
10

6000 0
0 100000 200000 300000 400000 500000 600000
Spring cycles

Figure 28 Spring relaxation chart

The final springs relaxed by just 7 N after 500000 cycles

(including 120000 cycles at 16000 rpm engine speed).
This increased to 33.5 N after a further 50000 cycles at
16500 rpm. On average this process revealed
interesting differences between the final spring pack
(that was eventually chosen) and the best competitor as
shown in Figure 29. The relaxation of the final springs
was the lowest of the springs tested.

Figure 27 Effect of spring relaxation on valve seating

velocity
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

140 x Success was achieved by making extensive use of

Competitor
dynamic simulation combined with minimal rig
120 testing.
NHK x The contribution of world class component suppliers
Averrage relaxation force (N)

100 to the success of the project was invaluable.

80
ACKNOWLEDGMENTS
60
The authors would like to thank senior management at
40 Petronas for providing permission to publish this paper.

20 REFERENCES

0 1. Valve train design for multivalve automotive gasoline

Fitted Compressed engines. Heath, A.R. SAE 885133, 1988
2. Optimal design of high speed valve train systems.
Figure 29 Comparison of spring relaxation following
David, J.W. Kim, D. Covey, J.A. SAE 942502, 1994.
final testing with 50000 cycles at 16500 rpm
3. The application of advanced simulation methods in
the design of Formula 1 valve trains. Di Paola, G.
The best competitor spring pack also suffered from
failure of the end tangs (as shown in Figure 30) when Smith, A. 19th International Vienna Engine
subjected to speeds in excess of 16000 rpm. Symposium, 1998, Vol. 1.
4. The design of cams for flat faced followers with
regard to elastohydrodynamic lubrication. Deschler,
G. and Wittman, D. MTZ March 1978.

CONTACTS

Phil Carden
Ricardo UK
Shoreham Technical Centre
Shoreham-by-sea
BN43 5FG
UK
Tel +44 1273 794959
e-mail [email protected]

Naji Zuhdi
Powertrain Technology,
Block E, PETRONAS Research & Scientific Services
Sdn. Bhd.,
LOT 3288 & 3289, Off Jalan Ayer Itam,
Kawasan Institusi Bangi, 43000 Kajang,
Selangor D.E., Malaysia
Figure 30 Failure of spring end tang Phone : +603-89244500
Fax : +603-89244548
This failure probably occurred due to loss of contact e-mail [email protected]
between spring and seat followed by impact force when
contact was re-established (see Figures 16 and 17). Andrew Whitehead
Del West USA
28128 W Livingston Av
CONCLUSIONS Valencia, CA 91355
USA
The design/development work on the valve train for the Tel +1 661 295 5700
Petronas FP1 had the following conclusions. e-mail [email protected]

x The final valve train had exceptional durability at

rated speed (speed limiter set to 16000 rpm) and
was able to survive overspeed events at up to 17000
rpm without failure.
 Downloaded from SAE International by Univ of Central Lancashire, Thursday, July 02, 2020

DEFINITIONS, ACRONYMS, ABBREVIATIONS Cam flank: Part of cam profile with positive acceleration

DLC: Diamond like carbon Tappet edge clearance: The tappet radius minus the
maximum eccentricity of the cam/tappet contact point
L/D ratio: Peak valve lift divided by inner valve seat
diameter

Cam nose: Part of cam profile with negative

acceleration