Beruflich Dokumente
Kultur Dokumente
INTRODUCTION
The Scuderi Split Cycle (SSC) engine concept was first conceived by Carmelo Scuderi, who filed the first patent for the concept on July 20, 2001 which was granted in 2003. The SSC concept divides the four strokes of a conventional combustion cycle over two paired cylinders. The first cylinder, referred to as the compressor (Figure 1.1), provides intake and compression strokes. The second cylinder, referred to as the expander, provides power and exhaust strokes. For the SSC prototype engine, the compressor top dead center (TDC) is phased 20 later than the expander TDC (TDC-e). Throughout this paper, crank angles are referred to in terms of degrees after TDC of the expander cylinder (ATDC-e). The SSC architecture can utilize any combination of normally aspirated, turbocharged, spark or compression ignition combustion strategies.
Figure 1.1. Principal Elements of the Scuderi Split CycleEngine The two cylinders are connected by a crossover port, through which the high pressure gas is transferred from the compressor cylinder to the expander cylinder between the compression and power strokes. This crossover (Xovr) event must quickly transfer all the working fluid in typically 35crankangle (CA) duration around TDC-e, essentially at compressor pressure, before combustion can commence. Outwardly opening Xovr valves
(opening away from the pistons) are employed to enable this high mass flow while avoiding contact between the valves and pistons. The combination of valve opening period and engine rated speed results in a valve event duration of 1.25 milliseconds (ms). This, together with the required valve peak lift means very challenging valve acceleration values.
Figure 1.2 compares the Scuderi Xovr valve events to the intake valve lift events of a standard production gasoline engine and of a Formula 1 engine. The comparison is done by calculating the pseudo acceleration by dividing the peak lift by the square of the event duration. To meet these challenging targets required a careful, CAE led design of the valvetrain and its components, together with experimental validation of the analysis tools and durability testing of the parts. This paper describes the architecture of the Xovr valvetrain, its analytical development, and experimental verification.
Figure 1.2. Typical valve operating parameters for various engine types at rated speed
Table 2.1.1 Summary Specification of the Scuderi Xovr Valvetrains From the start the driving approach to this valvetrain design was to deliver a highly reliable system which would enable the team to concentrate on combustion development work on the research engine without concerns for the valvetrain durability. The need for a high geometrical compression ratio (>30) in both the compressor and expander cylinders was recognized early on. This, together with the need to time the valve opening events near TDC, drove the architectural choice to open the valve outwardly, i.e. away from the piston. This therefore avoids altogether the risk of valve to piston contact as well as removing the need to machine pockets on the piston top surface, or to recess the valve seats in the cylinder head. As a result the valve head sealing surface faces inward toward the
piston and the corresponding surface of the valve seat insert faces outward, as shown in Figure 2.1.1
Figure 2.1.1 Xovr valve actuation elements for initial research engine
The selection of outwardly opening valves led to the adoption of a center pivoted rocker with a forked end to accommodate the valve stem and engage a valve adaptor from underneath. The valve adaptor is connected to the valve via standard design collets. The high valve acceleration rates created by the target lift and periods drove the requirement for minimum moving masses together with a stiff kinematic chain capable of transmitting the motion from the cam profile to the valve and to overcome the loads generated by the high differential pressures on the valve head with the minimum amount of vibrations. One obvious choice was to use titanium valves. The adoption of a pneumatic spring in place of a more conventional high strength steel coiled spring achieved both the objective of very low mass and low vibrations The pneumatic spring piston sits directly on top of the valve stem, resulting in a direct load path giving the most lightweight solution. Pneumatic springs also offer the ability to modulate the spring pressure as a function of engine speed allowing lower pressures to be used at low engine speeds, resulting in reduced friction. The requirement for high reliability drove the choice, for the camshaft, rocker, and valve adaptor, of adopting high strength steel (UTS > 2000 MPa), capable of withstanding
high contact stresses in operation. The rocker and valve adaptor were also coated with a diamond-like coating (DLC) layer to reduce the dry friction coefficient and improve their tribological behavior in boundary elasto-hydro-dynamic (EHD) lubrication experienced during operation. Another striking feature of this valvetrain in its current design is the large size of the cam lobe. The base circle radius of the cam lobe is in excess of 40 mm. The combination of the 2- stroke operation, short valve opening periods, high accelerations and low rocker ratio (for high stiffness), conspire to the need for the cam in this valvetrain to be larger than conventional automotive engines. This is mainly because of high contact stresses at the cam nose due to small radius of curvature, as well as small radii in the concave part of the cam profile which impose manufacturing challenges and high values of pressure angles. The reliability first approach also led to the adoption of a cam of diameter larger than necessary, which results in substantial safety margins on lubrication and on the contact stresses the materials are capable of withstanding. This left open the possibility in the future to design profiles with more aggressive accelerations without the need for major design changes. The rocker shaft is mounted eccentrically in its supports to enable lash adjustment. Oil ports are provided to the rocker shaft to supply pressurized oil to the rocker bore.
Figure 2.2.1 Pressure differentials acting on XovrC and XovrE valves 4000RPM full-load
Figure 3.1. Simulation of coil spring dynamics. Top: Valve spring surge amplitude, Bottom: Maximum valve spring stress amplitude Pneumatic valve springs on the other hand have intrinsically high progressivity as shown by the spring frequency chart in Figure 3.2 because the gas pressure in them follows a substantially adiabatic transformation. This allows tuning the spring progressivity by varying the spring initial volume. Furthermore the extremely low mass of the DLC coated aluminum spring piston means that the critical damping of the pneumatic spring is circa 7 times smaller than the critical damping of the coil spring. The damping in the pneumatic springs is generated by the friction of the spring piston against the seal. The seal is made of a grade of
filled PTFE, a very low friction material. The sealing action is achieved by allowing the gas inside the spring to push the seal against the spring piston and this generates enough friction force to dampen any resonance of the pneumatic spring itself.
Figure 3.2. Pneumatic spring natural frequency. The curve indicates non-linear stiffness due to the non-linear nature of polytrophic compression The dynamic model was also used to determine the loads and stresses on components, help develop the valve lift profile and study the dynamic loads and motions, including the valve seating behavior. As more detailed information on component mass and stiffness became available during the design and testing process the dynamic model was updated and new simulations produced. The bending stiffness of the rocker was the controlling stiffness of this valvetrain. Its effect is felt especially during the valve seating portion of the valve opening event. Simulations were performed to define the value of this stiffness which would be needed to achieve acceptable valve seating velocities in the whole speed range. Figure 3.3 shows the effect on the valve seating velocity of a rocker with the design target stiffness and of another one with 30% lower stiffness. It can be seen that with low rocker stiffness the valve seating velocity becomes excessive at speeds just above rated (4000 RPM), giving very little margin for error in the overall valvetrain design.
Figure 3.3. Simulation of the effect of rocker stiffness on valve seating velocity
Figure 3.4 shows the relative contribution, for the design which was made and tested, of the mass and compliance of the individual valvetrain components on the natural frequency which is in the region of 4000 Hz. The natural frequency is influenced primarily by the flexibility of the rocker and masses of the rocker and poppet valve. The poppet valve constitutes the payload, whereas all the other components have the function of controlling the motion of the valve. The rocker shaft and the camshaft supports are not part of the moving elements of the valvetrain (although the hole for the rocker shaft does reduce the rocker bending stiffness) and were therefore sized to add minimum compliance to the overall valvetrain. Their target stiffness was chosen to be about twice the rocker stiffness. In the following section the design process for the valve adaptor and for the rocker will be discussed.
Figure 3.4. Mass and Compliance breakdown contributing to valvetrain natural frequency
Figure 3.5. Design features in the valve adaptor assembly The first objective was achieved by using titanium for the collets, a material with a relatively high friction coefficient, and to use a relatively small 4 cone angle for the collets themselves. The detailed design of the adaptor assembly tried to favor the capability of the joint to store strain energy during its assembly by making the titanium collets relatively thick and by keeping a small, constant wall thickness in the part of the adaptor in contact with the collects. This was to safeguard against the joint coming loose due to parts relaxing from general background vibrations or higher than expected impact forces. Furthermore a pressing operation was included in the assembly procedure where the top of the valve stem was pressed with a 5000 N load while holding the valve open by the bottom face of the valve adaptor. This load occurs naturally in operation during the early parts of the valve opening event at high engine speeds when the rocker imposes high valve accelerations through the valve adaptor assembly. This in a way makes the design self-locking. The pressing operation however enables presetting of the position of adaptor and collets, which in turns allows setting the nominal valvetrain lash without having to first run the valvetrain at high speed. The second objective was achieved by sizing the part of the adaptor in contact with the collets, so that its mean diameter was in line with the mean diameter of the bottom surface of
the adaptor touching the forked end of the rocker, and to have a thin connection between the bottom pad of the adaptor and the part in contact with the collets. This feature allows the bottom pad to deform in operation and to follow the deformation of the forked rocker avoiding localized point contact. This final configuration was arrived at with the aid of FEA simulations modeling the assembly and disassembling loads, as well as simulating the interaction between the rocker and the adaptor and checking the component for safe cyclic (fatigue) operation. Figure 3.6 shows how the bottom face of the adaptor follows the rocker as the operating load is applied.
Figure 3.6. FEA results for valve adaptor assembly and operation: Vertical displacement of nodes on the bottom face of the adaptor
Torsional flexibility was designed into the rocker to accommodate a certain degree of misalignment between the cam lobe and the corresponding pad on the rocker which could arise in operation due to a combination of manufacturing and assembly tolerances. The misalignment was simulated in a FEA analysis where the rocker was pressed against a rigid plane skewed by 0.15 with respect to the rocker axis and compared with the case with no misalignment. Two levels of forces were used for the analysis: 2500 and 4000 N. The results are shown in Figure 3.9. The results were assessed comparing the normal pressure on the rocker surface between the case with and without misalignment. The increased contact pressure between the two cases did not exceed 10%.
Figure 3.9. Contact analysis showing tolerance of the design to misalignment of rocker shaft and camshaft centerlines. Von Mises stress on surface. Top row: Contact load = 4000 N, Bottom Row: Contact load = 2500 N. Left column: parallel loading, Right column: 0.15degrees misalignment loading
Figure 3.10. Details of the two valve lift profiles used on rig testing. Top left: Valve lift, Bottom left: Valve velocity, Top right: Valve acceleration, Bottom right: Valve jerk.
Figure 4.1. Xovr valvetrain test rig The Xovr valve used in the test rig had a threaded section added to its bottom face. This added negligible weight, and it permitted steel weights to be bolted to the valve as perturbation masses. These were used to correlate the valvetrain dynamic model. The locations of the instrumentation in the test head are shown in Figure 4.2. The valve lift was measured using a Micro Epsilon optoNCDT 1627-4laser. This unit has a 37 kHz bandwidth, which was marginal but sufficient to capture the valve bounce at high speed. The laser was mounted on a steel plate cooled by chilled water to isolate it from the temperature of the test rig. The camshaft angular position was measured using a 720 count optical encoder. A small volume port was designed into the test head and filled with compressed air at each revolution of the camshaft while the Xovr valve is shut. This was done to replicate the load the Xovr valvetrain experiences in the engine due to the differential pressure between the Xovr port and the expander cylinder. The port configuration was reached by simulating the valve opening event in the test head with GT-Power. Figure 4.3 shows that the simulated differential pressure across the Xovr valve for the engine and for the valvetrain rig are sufficiently close. To minimize the air consumption the Xovr port in the test head was designed with minimum volume. An adjustable orifice valve was also installed in the supply line to fine tune the dynamics of the gas flow.
Figure 4.2. Location of instrumentation in the test head of the valvetrain rig
Figure 4.3. Results of simulation of port pressure discharge during valve opening in the valvetrain rig: Differential pressure at XovrE Valve at 4000 RPM Piezoelectric pressure transducers were used to measure the dynamic pressures in the pneumatic spring and the Xovr port, while strain gauge type transducers measured the supply pressures to provide the pegging for the piezoelectric transducers. Bottled compressed air was used for the pneumatic spring and to pressurize the port to ensure consistent gas properties. When performing pressurized port testing the valve would open to atmosphere. It was important for the supplied gas to be dry to eliminate the possibility of icing of the valve seat, and air was used to avoid filling the test cell with another gas such as nitrogen and creating the risk of asphyxiation. The static stiffness of the valve and its stem were measured with the valve closed by adjusting the lash to a set amount, varying the air spring pressure, and measuring the lash as shown in Figure 4.4. This was done to compare to calculated values, and any deviation might indicate high friction in the pneumatic spring or improper seating of the valve. This was repeated at room
temperature, 57 C, and 90 C to ensure that there were no changes in component seating due to temperature.
As mentioned above two cam profiles were designed and tested in the rig. The first was representative of the profile to be used in the preliminary testing of the SSC experimental engine. The second was intended to drive the valvetrain to its dynamic limits at engine operating speeds, so that the limitations of the valvetrain dynamics could be explored and the assumptions made in the analytical modeling verified. The valve lifts measured at low speed resulting from the two cam profiles with negative lash (i.e. with the valve never touching its seat) are shown in Figure 4.5.
Figure 4.5. Measured valve lift from two camshaft profiles tested in the rig The size of the orifice between the pneumatic spring and the supply line has to be chosen as a compromise between the mean and transient spring dynamics. The plot in Figure 4.6
shows the internal dynamic portion of the pressure in the pneumatic spring over one revolution at 1500 RPM and 20 bar air supply pressure for various orifice diameters. Increasing the diameter allows more gas to be forced back into the supply line as the piston rises and therefore reduced pressure is available to prevent bounce when the valve closes. After investigating the full speed and pressure range, a 0.2 mm orifice was chosen. The small orifice diameter keeps the dynamic pressure in the pneumatic spring high, but slows the response of the filling of the spring from the line. In the case of the SSC, this is tolerable, because of the relatively short valve event duration compared to a full revolution of the camshaft.
Figure 4.6. Influence of varying pneumatic spring orifice diameter The nominal lash for the valve lift profiles described above was 0.10 mm. According to the dynamic simulations this would leave enough closing ramp to obtain the correct valve seating behavior. To confirm this, a series of experiments was performed to verify the dynamic performance of the valvetrain at a variety of lash settings. Figure 4.7 shows the valve velocity at 0.05, 0.10, and 0.15 mm lash at 4500 RPM and 25 bar pneumatic spring pressure. This investigation showed that at this high speed, the opening and closing behavior show further degradation (increased number of bounces and higher seating forces) as the lash increases above 0.10 mm. If lash falls to less than 0.10 mm, performance will be acceptable as long as some lash is maintained. Lash less than 0.10 mm may be possible in the event of wear, manufacturing tolerances, and thermal gradients.
Figure 4.7. Valve velocity varying lash at 4500 RPM at 25 bar pneumatic spring pressure The valve lift at 0.10 mm lash and 20 bar spring pressure is shown in Figure 4.8 at 1000, 3500, 4000, and 4500 RPM. Also shown are the corresponding dynamic portions of the pneumatic spring pressure signal measured with the piezoelectric sensor. This signal was pegged to zero right before the start of the event. These curves illustrate several interesting aspects of the valve system behavior. The pneumatic spring pressure at low speed shows the tendency for the pressure to be pushed back through the orifice into the supply line at low speeds, hence the decrease in the pressure after the lift event. The remainder of the revolution provides sufficient time for the spring pressure to recover. This pressure reduction is much smaller as speed increases. The lift curves also illustrate the increase in peak spring force and the onset of closing bounce as the speed increases.
Figure 4.8. Valve lift measured experimentally at various speeds at 20 bar spring pressure showing onset of closing bounce
In the figure it can be seen that at valve opening the spring pressure starts to rise before and movement of the valve is detected by the laser and at valve closing spring pressure continues to drop after the valve has seated. This is because when the valve is seated the valve spring compresses the valve stem and therefore before the valve can be opened the compressive deformation on the stem must be relieved. This happens at all speeds; at high speeds however the inertia of the valve head adds to the phenomenon as can be seen in the figure by looking at the valve lift traces in the shaft angle interval between 180-190. At high speeds, the extra force required to get the valvetrain in motion has created additional elastic deformation in the valvetrain and as this deformation is released while the valve continues to open higher peak lifts are recorded, as is evident from the increased pressure and lift at the higher speeds. This release of elastic deformation occurs as an oscillation, which becomes evident when the valve velocity and acceleration plots are viewed. As mentioned above, the valve of the test rig was designed to accept a perturbation mass to provide more correlation points between the model and the experiments. Figure 4.9 compares the valve lift and velocity at 3000 RPM, 0.15 mm lash, and 20 bar pneumatic spring pressure without additional mass, and with 5 g and 10 g additional mass added to the end of the valve. The additional mass causes the valve to open later for similar reasons to those discussed above for speed. The change in oscillation due to the additional mass as the elastic energy is released becomes evident. The increases in mass also increase seating velocity, creating bounce at this speed where it did not previously exist. In this case, for the valve seating velocity to be brought down to more acceptable levels the height of the closing ramp would have to be increased.
Figure 4.9. Valve lift and velocity at 3000 RPM adding perturbation mass
Once the valve dynamics were understood, a portion of the testing was repeated with the Xovr port pressurized. Xovr port pressure increases the force required to open the valve.
Figure 4.10 shows the results of two tests run at 0.10 mm lash and 20 bar pneumatic spring pressure at 4500 RPM. One test had the Xovr port pressurized at 20 bar and the other had no Xovr pressure. At these conditions, the Xovr port pressure delayed the opening of the valve by 1.5 due to the higher force required to lift the valve against the pressure. Once the valve has been lifted the pressure on both sides of the valve quickly becomes equal and the remainder of the valve travel is unaffected by the port pressure including the closing bounce shown. The pneumatic spring traces for the tests with and without port pressure lie on top of each other, indicating that the pneumatic spring travel was identical. The valve stem therefore stretched, and the contact pressure between the rocker and the valve adaptor increases with port pressure as predicted.
Figure 4.10. Effect of 20 bar Xovr port pressure on valve opening at 4500 RPM with 20 bar pneumatic spring pressure
adjustment is accomplished by placing the rocker bore eccentric with respect to the rocker shaft centerline, and rotating the rocker shaft. In the test rig, the rocker shaft is clamped at one end. It was originally assumed that valve actuation forces would push both ends of the rocker shaft to the bottom of the bores in the test head. The rocker shaft stiffness was modeled accordingly. In reality it was found that the rocker shaft behaved like a cantilever, supported only on the clamped side. The rig was therefore modified by splitting the rocker shaft and allowing it to be supported at both ends, which increased the stiffness and decreased the measured deflection. This would also better reflect the mounting of the rocker shaft in the research engine. Measurements on the rig also showed that the initial estimate for the camshaft support stiffness was 7.5% too high. The valve adaptor assembly stiffness used in the dynamic model was also slightly too high and was adjusted using the results of the FEA analysis. After taking these into account, the model correlated with the measured deflection within 20 m, which is well within the measurement capability. Understanding the deflection is important, as the timing of the valve events are so critical in this application, and the static and dynamic deflections must be known in order to determine the valve lift profile and the closing ramps needed to achieve the desired valve motion at all speeds.
Figure 4.11. Summary of static stiffness investigation showing total deflection of valve at peak lift as pneumatic spring pressure varied
After these corrections were made to the model, the measured valve lift correlated well with the simulated dynamic lift (Figure 4.12). The accuracy of the laser measurement was 60 m, which is insignificant compared to the experimental scatter which is shown as a shaded band in the figure. As can be seen the simulated lift profile lies within the measurement scatter band. The kinematic valve lift profile is also shown for reference.
Figure 4.12. Predicted and measured dynamic valve lifts at 4000 RPM, 20 bar spring fill pressure, 0.15 mm lash Good correlation was achieved between the predicted dynamic and measured valve velocities (Figure 4.13) and accelerations (Figure 4.14), and it was confirmed that the valve was seating with the appropriate velocity. What appear to be high frequency vibrations in these figures is actually noise in the laser signal. As noted above, the total inaccuracy in the laser measurement was 60 m, and roughly two-thirds of this was due to a combination of electrical and atmospheric noise. Changes in air temperature and air currents will affect the laser signal slightly, and electrical noise in a test cell with 3 kW heaters and 37 kW motors cannot be entirely avoided without high expense. The 60 m noise level was deemed acceptable, and is hardly apparent on the displacement curves. It becomes very obvious however when the displacement is differentiated to calculate valve velocity and acceleration. Filtering the signal would have attenuated the valve bounce amplitude, so the signals were averaged instead. The average of 300 valve lift events are plotted as the mean measurement lines in the velocity and acceleration traces for comparison with the simulation.
Figure 4.13. Predicted dynamic and measured velocities at 4000 RPM, 20 bar spring fill pressure, 0.15 mm lash
Figure 4.14. Predicted dynamic and accelerations velocities at 4000 RPM, 20 bar spring fill pressure, 0.15 mm lash
seating velocity to represent worst-case values, the pneumatic spring supply pressure was held to 20 bar 1 bar. Xovr port supply pressure was included and held to 40 bar 2 bar, and the test head temperature was held to 90C 2C. To obtain the 40 bar pressure difference across the valve, the cylinder side of the valve was open directly to atmosphere. This generated considerable noise so a noise abatement cabinet was built around the rig. This also consumed a lot of air, and sixteen16-packs of bottled compressed air were required to complete the test. The valve lift, seating velocity, and spring pressure were measured every fifteen minutes, and the lash and visuals were checked roughly every six hours. Figure 4.15 shows the valve lift and pneumatic spring pressure at 9, 16, 20, 25, and 30 hours into the durability test. The valve lift did not change measurably during the entire test, although the spring pressure did increase during the test. When this trend was noticed, the test was paused and the pneumatic spring was disassembled.
Figure 4.15. Valve lift and spring pressure during durability test The first pressure increase was noted at 15 hours, and the assembly was disassembled at 22 hours. The cause for the increased pressure was that oil had leaked past the piston seal, and this oil reduced the spring volume, therefore increasing its compression ratio. The seal supplier was contacted and options to address this issue include altering the composition of the seal material, changing the pressure of the preload spring inside the seal, and changing the grooving geometry on the seal face. Resolution of this issue is considered part of the normal development process for a pneumatic spring system. The remainder of the components showed little to no wear after the test. The DLC coated parts including the rocker, valve adaptor, and pneumatic piston appeared as new. The cam journals also appeared as new. There was very light polishing of the valve guide, and the valve stem seals were in good condition. The lobe of the cam showed uniform light scuffing
and the profilometer trace indicated an average of 4 m wear at the most severe point on the lobe. The cam lobe after the test is shown in Figure 4.16.
The only parts with measurable wear were the valve and seat, shown in Figure 4.17. The wear was uniform around the valve and its sealing action was still intact by the end of the durability test. The seating surfaces were measured with a profilometer to determine the wear rates. The wear rate of the valve was 5.2 m/million cycles, and that of the valve seat was 6.4m/million cycles. The valve seat in the test rig was made of beryllium copper whereas the valve seat in the engine will be made of sintered hardened steel alloy (Bleistahl Como 406 FS) which will have a lower wear rate.
CONCLUSION
A fast acting valvetrain has been designed to manage the Xovr gas flow in the SSC engine. The application requires that a high mass flow of air be transferred from the compressor cylinder to the expander cylinder over circa 35 crank angle duration. This places considerable demands on the Xovr valves, which must open rapidly and close without bounce. The required performance was achieved with the adaptation of existing technology and by utilizing outward opening valves. The work demonstrated the need for substituting the traditional coil springs with lighter gas springs as used in racing engines. The geometry and the tribology of the rocker surfaces was optimized for the high loads experienced and so was the rocker stiffness-inertia tradeoff. A suite of analysis models was developed and calibrated to aid in the design of the valvetrain. The dynamic model of the valvetrain built using GTSuite was especially instrumental in balancing the requirements for low effective mass and high stiffness of the moving components, as well as in refining the design of the cam lift profile and its opening and closing ramps. An experimental rig was built to test the valvetrain under most of the conditions that it will experience when installed in the engine. This testing verified the dynamic capability of the Xovr valvetrain to achieve large accelerations and was also used to calibrate the analytical models. The additional loads on the valvetrain due to the differential pressure at opening of the Xovr valve were replicated when an extended durability test of the valvetrain was carried out. The test was completed successfully. The set of verified analytical models now available will be used to tailor the valvetrain design and performance as various Scuderi engine configurations are developed in the future
REFERENCES
1) Scuderi, C. J., Split Four Stroke Cycle Internal Combustion Engine, International Patent WO 03/008785 A1, January 30, 2003. 2) Phillips, F., Gilbert, I., Pirault, J.-P., and Megel, M., Scuderi Split Cycle Research Engine: Overview, Architecture and Operation, SAE Int. J. Engines 4(1): 450-466, 2011, doi:10.4271/2011-01-0403. 3) Pastorelli, S., Almondo, A., and Sorli, M., Performances of Cam follower Systems with Pneumatic Return Spring, Proceedings of the Bath Workshop on Power Transmission and Motion Control-PTMC 2005, Bath, UK, 379-394. 4) Pastorelli, S., Almondo, A., and Sorli, M., Effects of mechanical and pneumatic return springs on valve train dynamics, 8th Biennial ASME Conference on Engineering Systems Design and Analysis, 2006, Torino, Italy. 5) Pastorelli, S., Almondo, A., and Sorli, M., Mechanical Spring Replacement With Pneumatic Return Device in a Valve Train: Effects on Dynamics and Preload Tuning, ASME J. Mech. Des. 132: 011008-1-9, 2010, doi: 10.1115/1.4000640.
6) Gamma Technologies Inc., GT-Suite, GT-ISE, (Versions 5.2-7.0), Computer Software. 7) Mclaughlin, S., and Haque, I., Development of a Multi- Body Simulation Model of a Winston Cup Valvetrain to Study Valve Bounce, Proceedings Institute of Mechanical Engineers, Part K, J. Multi-body Dynamics, 2002, 216, 237-248. 8) Sczepanski, J.G., New Equipment and Methodology to Perform High Speed Valve Train Dynamics Testing and Analysis, SAE Technical Paper 2004-01-1720, 2004, doi: 10.4271/2004-01-1720.