Application of a General Planar Kinematics and Multi-Body Dynamics Simulation Tool to the Analysis of Variable Valve Actuation Systems

Marcin Marek Okarmus and Rifat Keribar

Gamma Technologies Inc.

2010-01-1193
Published 04/12/2010

Edward Suh
Delphi Powertrain Systems
Copyright 2010 SAE International

ABSTRACT
The advantages of Variable Valve Actuation (VVA) in the aspects of improved engine performance, fuel economy and reduced emissions are well known in the industry. However, the design and optimization of such systems is complex and costly. The design process of VVA mechanisms can be greatly accelerated through the use of sophisticated simulation tools. Predictive numerical analysis of systems to address design issues and evaluate design changes can assure the required performance and durability. One notable requirement for the analysis and design of novel mechanically-actuated VVA systems is a general-purpose fast and easy-to-use planar mechanism kinematics analyzer with cam solution/design features, which can be applied to general mechanisms. This paper introduces a general simulation and design tool, which features general planar kinematics and multi-body dynamics analysis capabilities, as well as integrated hydromechanics and hydraulics to model devices such as lash adjusters and cam phasers. Application of the methodology to various mechanically-driven variable valve actuation systems is discussed, with focus on a specific system. The modeling process is broken down into multiple stages. First, the analysis of kinematic motion of valvetrain components along with the procedure to calculate the cam shape profile required to produce the desired valve lift is described. Second, a constrained-dynamics simulation of a rigid system is carried out in search of nominal (quasi-dynamic), inter-component forces, valve spring margin and cam-follower separation speed. Third, a complete multi-body dynamics analysis,

which considers the elasticity of valvetrain components and inter-component contacts, is employed to produce a wide array of detailed dynamic predictions. Ways of rapidly optimizing the key design parameters through the use of dedicated numerical analysis are briefly discussed.

INTRODUCTION
Variable Valve Actuation offers many opportunities to improve the performance of an internal combustion engine in the areas of fuel economy, power density and emissions. Evidence of this can be seen in the ever-increasing number of production engines incorporating VVA systems to vary the timing, duration and shape of the valve lift curve [1], [2], [3], [4]. In the last couple of decades, several kinds of VVA strategies have been researched and implemented, ranging from mechanically-driven valvetrains to camless systems that are actuated electrically, hydraulically and/or pneumatically. Whatever the method of variable valve control, the design and optimization of such systems remain difficult, timeconsuming and expensive activities. The amount of improvement in engine performance that can be achieved is clearly limited by the actual capabilities of each specific type of valvetrain and its actuation system and, often, requires multiple iterations and juggling of different design factors before an optimal set of operating parameters is found. The design process of VVA mechanisms can be greatly accelerated through the use of sophisticated numerical simulation tools. This paper gives an overview of such general-purpose simulation tool, which features general planar kinematics, multi-body dynamics as well as hydromechanics and hydraulics analysis capabilities. The tool

has been developed as part of the GT-SUITE simulation software, which is specifically designed for applications in the engine, powertrain and vehicle industries [5]. The solution methodologies implemented in the tool are presented in the context of application to variable valve control systems that employ a mechanical actuating principle. In particular, the swing-cam type, variable valve actuation mechanisms presented in [6] and [7] (a schematic of the mechanism described in [6] is shown in Figure 1 below) are discussed and multiple stages of a numerical modeling process of such valvetrains are described including: planar kinematics analysis of component motions combined with a specialized algorithm to compute the rotational cam and/or swing cam shape profile required to produce the desired valve lift constrained-dynamics simulations of the system, which ignores component elasticity, to obtain nominal (quasidynamic) predictions, i.e. inter-component forces, valve spring margin and cam-follower separation speed, etc. lumped-parameter model multi-body dynamics analysis, which considers the elasticity of valvetrain components and inter-component contacts. The scope of the paper is further expanded to describe a dedicated library of specialized valvetrain system analysis components, which utilizes all capabilities of the previouslymentioned general-purpose simulation tool, but significantly simplifies data input and model setup and, hence, speeds up the design process. Finally, the discussion is extended to include an example of integrated hydro-mechanical analysis. Several results from the simulation that are of high interest to a valvetrain designer are presented and ways of rapidly optimizing the key design parameters are discussed.

Figure 1. Swing-cam type, mechanically-actuated VVA mechanism analyzed in the paper

GENERAL PLANAR KINEMATICS ANALYSIS TOOLBOX

Having a tentative geometrical layout of the valvetrain along with the desired valve lift profile, the numerical modeling process typically begins with the kinematic analysis of the system. The aim of this simulation is to obtain idealized motions of all valvetrain components along with other useful kinematic predictions, i.e. component velocities and accelerations, motions of contact points, entrainment and scrub velocities at inter-component contacts, perfect rolling speed of the cam-follower roller, etc., but, primarily, to optimize the geometry and compute the cam shape profile required to realize the desired valve lift. The general-purpose planar kinematics analysis toolbox available in the simulation software and presented here is suitable for simulating single-degree-of-freedom, planar mechanisms. The swing-cam type, variable valve actuation systems mentioned above belong to the family of such mechanisms. The toolbox is essentially a library of kinematic operators (connections), which solve linear or non-linear relationships between translational or rotational kinematic nodes (components) shown in Figure 2. The assemblies of kinematic components and connections forming a mechanism are recognized by means of a recursive algorithm. Applying an angular or linear displacement (typically as a function of crank or cam rotation angle) to any component in the assembly results in the solution of motions of all other kinematic nodes in the cluster constituting the mechanism. Computational time required to complete the kinematic simulation is very low, usually on the order of a few seconds,

which makes the use of this tool feasible for rapid design and optimization studies.

Figure 2. Rotational and translational kinematic nodes For instance, if we denote the angular displacement of the rotating node as and the planar displacement of the translating node along the prescribed direction as , then the connection linking these two nodes solves the following relationship to compute the motion of one (output) component as a function of known displacement of other (input) component and mechanism geometry:

Figure 3. Four bar linkage mechanism defining a kinematic relationship between two rotating nodes

(1) Various types of kinematic operators are available in the tool including: binary links crank-sliders constant and variable gear ratios rack-pinion couplings planar contacts Selected examples are illustrated below in Figures 3 and 4. representing the Solution to the non-linear function relation between displacements of kinematic nodes is obtained either analytically, as is in the case of the planar four bar linkage mechanisms shown in Figure 3 where the angular displacement of the output link, , is computed as a Figure 4. Kinematics of the cam-to-translating circular follower contact

Vectors and in Figure 4 are position vectors of candidate contact points on body k (cam) and body l (translating circular follower), respectively measured with respect to inertial reference frame ( , ). Furthermore, vectors and are tangent and normal vectors to the surface of body k at common contact point. Similarly, vectors and are tangent and normal vectors to the surface of body l. Eq. (2),3,(4) below describe the kinematics of this planar cam-to-follower contact (as well as any other planar contact) and are solved simultaneously to obtain the displacement of output node as a function of displacement of input node and contact geometry. Multidimensional Secant Method (Broyden's Method) [8] is an algorithm employed to solve these non-linear equations. Notation indicates a

function of rotation angle of input link, , and mechanism geometry, or using an iterative numerical procedure to solve, for example, the kinematics of a cam-to-translating circular follower contact of Figure 4 and compute the displacement of the follower, , as a function of cam rotation angle, contact geometry. , and

transpose of a vector. Note that Eq. (4) defines the gap (lash), , between the contacting surfaces.

(2)

(3)

(4) An additional useful feature of the planar kinematics analyzer is an ability to account for lash at the contacting interfaces between kinematic components. Lash can be specified at any location in the system and its presence will influence the resulting motions of kinematic components. A sample application where this type of feature proves to be very useful is a study of loss of valve lift and change in area under valve lift curve as a function of valve-to-follower lash. The algorithm to account for lash is implemented as follows. First, Eq. (2) and (3) are solved simultaneously using the method described above to obtain the position vectors and of candidate contact points. The value of , is then computed using Eq. (4). In case instantaneous gap,

Figure 5. Graphical representation of the idea behind the cam lobe solution algorithm For the VVA mechanisms such as that shown in Figure 1, there are two locations of cam lobe surface profile, that of the rotational cam and that of the swing cam (see Figure 6) and the cam lobe solution algorithm can be applied to each of these locations. This means that: given the valve lift profile as a function of rotation angle of rotational cam and the surface profile of the swing cam as input, the algorithm can solve for the shape of the rotational cam given the valve lift profile as a function of rotation angle of rotational cam and the surface profile of the rotational cam as input, the algorithm can solve for the shape of the swing cam

, indicating separation (lash) between contacting surfaces, the displacement of output node is set to zero. , indicating interference between Otherwise, if contacting surfaces, Eq. (2),3,(4) are solved together again resulting in the displacement of the output kinematic node. Another novel feature of the planar kinematics analysis toolbox is a specialized cam lobe profile solution algorithm. The idea behind this algorithm is presented graphically in Figure 5. Given a displacement (angular or linear) of any component in the kinematic assembly as a function of rotation angle, , of another component in the assembly and the nodal location of cam lobe surface as input, the module enables computation of the cam lobe surface profile (expressed in polar form as cam radius vs. local rotation angle ) rigidly attached to a rotational kinematic node and required to produce the desired input displacement function.

Figure 7. A graphical basis for describing the cam profile by means of theory of envelopes Figure 6. Locations of the rotational cam and swing cam for the swing-cam type VVA mechanism The basis for solution of the cam lobe shape profile is the theory of envelopes from calculus [9]. The theory states that the cam shape (inner envelope) that will produce the desired motion of the follower is obtained by fitting a tangent curve to the family of curves defining all possible follower positions. This procedure is presented graphically in Figure 7 for a cam-to-circular follower contact (the same procedure is used to solve the problem of a cam-to-flat-faced follower contact) and effectively boils down to simultaneous solution of the following equations for the X and Y coordinates of the cam shape profile: Application of the above cam lobe shape profile solution algorithm and other kinematic analysis capabilities to the actual swing-cam type VVA mechanisms will be presented in the following sections.

GENERAL MULTI-BODY DYNAMICS ANALYSIS TOOLBOX

Multi-body dynamics simulation of the valvetrain at the subsystem and system level has become necessary in reducing design cycle time. Using analytical models to perform this type of analysis at the concept design stage can identify potential problems and further aid in optimizing the design [10]. The general-purpose, multi-body dynamics simulation toolbox can be applied to model valvetrains including VVA's as well as other types of mechanisms in one, two and three dimensions (i.e. 1D/2D/3D). The description of the broad range of dynamic analysis capabilities available in the simulation tool and the underlying solution methodologies can be found in [5] and [11]. Here, the discussion will be limited to planar (2D) applications. Similar to the planar kinematics analyzer, the multi-body dynamics analysis tool is an object-oriented library of components and connections. These represent the physics of real parts/links/joints and serve as fundamental building blocks for mechanical systems. The internal solver architecture is based on the Finite Element Method (FEM). It also satisfies one of the distinguishing requirements of multibody dynamic analysis (e.g. when compared to finite element analysis), which is the requirement to consider various types of joints that impose constraints on the relative motions of the bodies in the system. The constraints are implemented using the Lagrange Multiplier technique [12]. This type of

(5)

(6) represents a family of curves that describe the Function planar positions of the follower. Components of vector define the X and Y coordinates of the cam profile. Angle , called the parameter of the family, distinguishes the member curves from one another, i.e. for a particular value of (5) defines one member of the family of curves. , Eq.

infrastructure, combined with Herting's modal reduction techniques [20], a sparse matrix (skyline) storage and solution scheme for non-symmetric matrices and an option to use several, different types of integration methods (both explicit and implicit), allow the user to create and efficiently analyze general mechanisms that may contain combinations of rigid and flexible bodies, joints and external loads. The solution procedure amounts to solving (and integrating) the following system of equations to obtain the states of structural components:

the constrained-dynamics simulation, include component separation speeds, valve spring margin, nominal contact Hertz deformation and stress, etc. Inclusion of component compliance and contact stiffness in the model leads to additional important predictions including component deformations, strains and stresses, loss of lift due to deformation, component vibration and surge, valve seating velocity and bounce, detailed Hertz and elastohydrodynamic (EHD) results and many more. The wide array of available outputs permits a thorough understanding of the mechanism's operation and provides means for fine-tuning the model. Application of the multi-body dynamics simulation to the analysis of the VVA systems will be described in the following sections.

(7) where is the configuration dependent, positive-definite is a vector that includes damping, elastic and is the displacement vector, is is a vector of Lagrange

mass matrix,

SPECIALIZED ANALYSIS OF VALVETRAIN SYSTEMS

Specialized, valvetrain system kinematics and multi-body dynamics analysis capabilities of the GT-SUITE software have been presented in numerous publications [13], [14], [15], [16]. This tool, which is specifically tailored to be used by valvetrain engineers, offers all of the planar kinematics and multi-body dynamics simulation features described above. It can be applied to various types of valve control systems including mechanically-driven VVA's. Furthermore, being part of an integrated simulation environment, it presents other features that prove to be very useful in rapid, concept level design and optimization of valvetrains [5]. These include a visual pre- and post-processor, an ability to use a direct optimizer and a DOE processor, distributed computing and, foremost, a formal integration of multi-body dynamics with other numerical modeling libraries, i.e. 1D flow simulation and hydraulics, electrical and magnetic components, control components. Vast valvetrain designer know-how is incorporated into the analysis tool making it easy and efficient to use. The input to the tool (i.e. valvetrain mechanism specifications) is preprocessed to automatically generate the underlying planar kinematic and multi-body dynamics models. The results from simulations of the underlying models are fed back to the valvetrain tool and further post-processed to create a wide array of specialized valvetrain outputs. Flow diagram in figure below illustrates the process.

externally applied loads,

the constraint Jacobian matrix,

is the vector multipliers associated with constraints, containing terms associated with the differentiation of constraint equations and respect to time. represents the derivative with

The multi-body dynamics solution procedures described above can be employed in the planar analysis of variable valve actuation mechanisms. In addition to the geometrical layout of the system, which by this time has been optimized based on results from the simulation of mechanism kinematics, the input required to the model also includes the mass-elastic properties of the valvetrain parts along with stiffness, damping and frictional characteristics of intercomponent contact, bearings, guides, etc. The study of VVA system dynamics appropriate at the concept design level can be carried out in multiple steps starting with a fast but simplified, rigid, constraineddynamics simulation, which ignores component elasticity and only considers inertial effects, friction and valve spring force, and proceeding to a full-blown dynamic model, which also accounts for elastic compliance, component separation and impact, valve spring surge, etc. The term rigid constraineddynamics refers to a simulation mode, in which all components are modeled as rigid bodies and all intercomponent contacts and joints (i.e. guides, pivots) are handled with constraints. The ability to eliminate high frequency response associated with high stiffness of components and contacts from the dynamic solution enables to obtain clean, nominal (quasi-dynamic) predictions for forces and moments acting on the valvetrain and, at the same time, significantly decrease the computational time required to complete the analysis. Other results that are of high interest to valvetrain designer, which can be extracted by means of

Figure 8. Flow diagram illustrating valvetrain system analysis process

APPLICATION OF SIMULATION TOOL TO THE ANALYSIS OF VVA SYSTEMS AND VALIDATION OF SIMULATION RESULTS
The simulation tool was applied to the analysis of a swingcam type, variable valve actuation mechanisms. Kinematic and dynamic predictions were validated by comparing results from other simulation programs. The first system selected for analysis was the Continuously Variable Valve Lift (CVVL) valvetrain developed by Delphi Corporation [17]. Geometrical layout of this mechanism is shown below in Figure 9. The system is composed of a rotational (input) cam, which actuates an oscillating/swing cam (also referred to as a rocker or output cam). The swing cam, in turn, drives two roller finger follower (RFF)-valve assemblies. Variable valve lift is realized by changing the position of the control shaft and effectively modifying the location of the swing cam pivot point. The pivot point moves along a section of a circular path, which is centered at the RFF roller. More details on the principle of operation of this and similar mechanisms can be found in [1] and [6].

Figure 9. Continuously Variable Valve Lift (CVVL) valvetrain A model of the CVVL mechanism was constructed from the library of specialized valvetrain components based on the information provided by Delphi. Geometrical layout of the system along with mass-elastic properties of valvetrain parts were imported into the object-oriented interface of the simulation tool. Figure 10 below illustrates the main building blocks of the simulation model. <figure 10 here>

PLANAR KINEMATICS ANALYSIS

First, a planar kinematics simulation of the valvetrain was carried out. In addition to geometrical specifications of the system, the input to this simulation included the rotational and the swing cam shape profiles. The model was analyzed at several (total of 17) positions of the control shaft, corresponding to the angular displacement of the swing cam pivot point from 0 deg. (max. lift) to 16 deg. (min. lift) in the clockwise direction, to produce a family of valve lift curves along with many other kinematic predictions. Total time required to complete the simulation (all 17 cases) was reported at about 15 seconds on a 2GHz, dual-core machine. Selected results from the simulation of planar kinematics are presented below in Figures 11,12,13,14. Figures 11 and 12 show the kinematic lift and acceleration of the valve, respectively, plotted vs. rotational cam angle. Furthermore, Figures 13 and 14 show the angular displacement of the swing and a finger follower roller pressure angle. All results were normalized through division by a maximum value in the range analyzed.

Figure 10. Simulation model of the CVVL valvetrain

Figure 11. CVVL kinematic valve lift

Figure 12. CVVL kinematic valve acceleration

Figure 13. CVVL kinematic swing cam angular displacement

Figure 15. Validation of kinematic results - CVVL valve lift comparison

Figure 14. CVVL kinematic RFF roller pressure angle Kinematic predictions presented above were validated by comparing results from another simulation tools. Here, the Unigraphics motion analyzer, UG Motion, was used as a reference program for comparison. Figures 15,16,17 below show comparisons of simulated kinematic valve lift, velocity and acceleration (referred to as Simulation) to the reference results (labeled as Reference). For clarity of presentation, three simulation scenarios, corresponding to three distinct locations of the control shaft, i.e. 0 deg (max. lift), 8 deg (middle lift) and 16 deg (min. lift), were selected and used for comparison. As seen below, the plots show very good agreement between Reference and Simulation with a maximum percent difference of 1.2% (in the selected regions of valve acceleration) and an average percent difference of 0.1%. These small percent differences in higher derivatives of lift can be associated with the different numerical algorithms used for computation of function derivatives.

Figure 16. Validation of kinematic results - CVVL valve velocity comparison

was used as input to the calculation and the result was the shape (excess radius) of the rotational cam. Again, the comparison between the Reference and Calculation rotational cam lobe profile is shown in the figure below.

Figure 17. Validation of kinematic results - CVVL valve acceleration comparison Once validated, the kinematic model of CVVL valvetrain was further exercised in order to test the novel cam shape profile design features. This time, the input to the model included the rotational cam shape profile as well as the maximum target valve lift defined as a function of rotational cam angle. The goal of this exercise was to compute the swing cam shape required to produce the desired valve lift. Figure 18 shows the comparison between the Reference swing cam lobe excess radius (excess radius is defined as the radius above the base radius of the cam) and the one obtained using the Calculation. Results were again normalized through division by the maximum value of the Reference signal.

Figure 19. Calculation of the CVVL rotational cam lobe profile - comparison The above mentioned cam shape profile solution algorithm can be efficiently applied in rapid design of lobe profiles of both rotational and swing cams and, effectively, in optimization of valve lift curves. Subject to a set of design criteria, the fast planar kinematics solver can be integrated into an optimization loop in order to compute the cam shape which produces the desired valve lift and, at the same time, meets all the required design specifications. The design criteria mentioned here typically include parameters such as a maximum valve acceleration limit, an allowable range of cam radius of curvature, cam-to-roller follower contact pressure angle limit, limit on the kinematic prediction of roller perfect rolling speed along with contact scrub and entrainment velocities, etc. For illustration purposes, a sample study was set up where a Reference shape of the rotational cam was used as input to the simulation and the magnitude of Reference target valve lift (also input to the analysis) was varied in 2 steps. The output from this simulation was the shape of the swing cam corresponding to each of the input valve lift profiles. Input valve lifts and resulting swing cam shape profiles were compared to their Reference values and are shown below in Figures 20 and 21, respectively.

Figure 18. Calculation of the CVVL swing cam lobe profile - comparison In a similar fashion, the rotational cam shape profile can be back-calculated by the same algorithm. In this case, the swing cam angular displacement corresponding to the maximum valve lift scenario and defined versus rotational cam angle

Figure 20. Study - calculation of the CVVL swing cam lobe profile - input valve lifts Figure 22. Continuously Variable Valve Lift and Duration (VVLD) valvetrain Since the main purpose of this exercise was to showcase the generality of the simulation tool and due to the lack of true geometrical specifications of the valvetrain, the dimensions of system's components were approximated to resemble the VVLD mechanism in question. Similarly, rotational and swing cam shape profiles were borrowed from a different valvetrain model. Planar kinematics simulation was carried out at 11 different positions of the control shaft. The resulting family of valve lift curves is shown in Figure 23 below.

Figure 21. Study - calculation of the CVVL swing cam lobe profile - computed lobe profile As previously described, the kinematics solver is quite general and can be applied to the analysis of various types of planar mechanisms. To illustrate this, the solution methodologies were applied to the study of another swingcam type VVA mechanism, namely the continuously Variable Valve Lift and Duration (VVLD) valvetrain described in detail in [7] and pictured in Figure 22. This design concept, which realizes a relatively high lift with small duration by virtue of its structure and can operate at considerably high engine speeds through its high rigidity [7], is somewhat similar to the CVVL mechanisms analyzed above. The key difference is the presence of an additional oscillating component, i.e. control arm, between the swing cam and the rocker arm. Variable valve lift is realized by turning the control shaft, which subsequently modifies the location of pivot point of the control arm.

Figure 23. VVLD kinematic valve lift It can be observed that the shape of valve lift profiles does not look realistic. As mentioned above, this is a result of the lack of detailed information about the geometry of this VVA mechanism. Nevertheless, this application supports the main premise, which was to emphasize the generality of the simulation tool. The VVLD system described above was

analyzed in the forward mode with the rotational and swing cam profiles as input. However, the same cam lobe shape solution procedures used in the case of CVVL valvetrain can also be applied to this as well as other, mechanically-driven, conventional valvetrains and variable valve control systems.

average way below this number. These small differences in the results can be attributed to the uncertainty in the values of contact stiffness and damping parameters used.

PLANAR, MULTI-BODY DYNAMICS ANALYSIS

The focus of the following section is on the simulation of dynamic behavior of the VVA systems. As explained above, this type of analysis can be broken down into multiple stages according to the level of model complexity and resulting physical phenomena captured by numerical simulation. It typically starts with a simplified study of rigid and constrained dynamics of the valvetrain and extends to a more detailed analysis of components' elastic response and dynamic vibrations.

RIGID AND CONSTRAINED MULTI-BODY DYNAMICS

In this analysis, the elasticity of all valvetrain components is ignored. The components are modeled as lumped inertias that are free to move in 2-D space. Further, planar motions of these inertias can be limited through the application of constraints, e.g. revolute joints, prismatic joints, sliding joints, etc. The inter-component contacts can either be assumed to be very stiff but still allow separation or additional contact constraints can be prescribed to prevent lash and force valvetrain parts to follow kinematically determined motions. The simulation tool was applied to the study of a rigid dynamics model of CVVL valvetrain. Frictional effects were ignored in the analysis. Valve spring and the swing cam torsion spring were modeled as massless links. Effective moving mass of the valve spring was added to the mass of the valve. Given the material properties and geometry of mating surfaces, the contact stiffness was calculated based on the Hertz model [19]. High contact damping coefficients were used to attenuate the high frequency noise associated with contact stiffness. Selected results from the simulation of the valvetrain run at 6000 RPM were again compared to the reference results generated using another simulation program. The ADAMS multi-body dynamics simulation tool from MSC Software [18] was used as a reference program for validation. Comparison of normalized valvetrain dynamics predictions (Simulation) to the Reference results is shown below in Figures 24,25,26,27,28,29 for three distinct locations of the control shaft, i.e. 0 deg (max. lift), 8 deg (middle lift) and 16 deg (min. lift). An average CPU time required to complete a single engine cycle was 3 seconds on a 2GHz, dual-core machine. Again, a very good agreement between the Reference and Simulation results can be observed with the highest percent difference of 4% and an

Figure 24. Validation of rigid dynamics results - CVVL valve acceleration comparison

Figure 25. Validation of rigid dynamics results - CVVL valve pallet contact force comparison

Figure 26. Validation of rigid dynamics results - CVVL valve pallet contact scrub velocity comparison

Figure 28. Validation of rigid dynamics results - CVVL rotational cam torque comparison

Figure 27. Validation of rigid dynamics results - CVVL swing cam contact force comparison

Figure 29. Validation of rigid dynamics results - CVVL rotational cam contact Hertz stress comparison In addition to the results presented above, the rigid, constrained-dynamics analysis can be used to obtain other predictions that are of high value to valvetrain designers. Examples include the swing cam-roller and rotational camroller separation speed (i.e. speed at which the magnitude of normal contact force drops to 0 N) or the valve spring and swing cam torsion spring margin. The spring margin is a nondimensional quantity that determines if the spring is capable of producing a force, which will be sufficient to keep valvetrain components in contact and prevent separation during the region of negative cam acceleration. It is computed as where is an equivalent spring

is the force (measured at cam-follower contact) and effective valvetrain inertia force. The spring margin results of CVVL valvetrain springs (both valve and torsion) computed

at 6000 RPM for different locations of the control shaft are plotted in Figures 30 and 31. Both the valve spring and the torsion spring margins were found to be adequate for the CVVL design at the engine speed analyzed.

components and the elasticity of inter-component contacts. It can also account for friction resulting from relative motions of contacting surfaces. Valuable predictions from such model, in addition to those obtained via rigid dynamics, include: component deformations, component vibrations and resonance, valvetrain tribology and frictional power loss, loss of valve lift (as compared to idealized, kinematic lift) due to deformations, contact forces, component separation and impact, valve seating velocity and others. For illustration, the dynamic model of Delphi's CVVL system was modified to account for elasticity of valvetrain parts. In general, mechanically-driven VVA mechanisms are designed for high rigidity. The compliance of valvetrain parts can be obtained either experimentally via a strain gage measurement of component's deformation due to an applied load or by means of the static finite element (FE) analysis. In absence of true stiffness values, the compliance of the swing cam and the roller finger follower parts was approximated and modeled as a single torsional stiffness value. A massless valve spring connection was replaced by a more physical mass-elastic model of a helical spring in order to capture the inertial effects associated with motion of spring coils as well as coil clash. Coulomb friction model was applied to account for friction at contacting surfaces. The simulation was carried out at the engine speed of 6000 RPM. The use of implicit integration scheme (3-stage Radau algorithm) to integrate the dynamic equations of motion ensured that the simulation time step was not constrained by the presence of high frequency modes (e.g. stiff spring vibration modes) and that the model remained computationally efficient. An average simulation time was found to be about 30 seconds per engine cycle on a 2GHz, dual-core machine. Selected results from the dynamic simulation of the valvetrain at 6000 RPM for control shaft positions of 0 deg, 8 deg and 16 deg are shown below in Figures 32,33,34,35.

Figure 30. Rigid, constrained-dynamics results - CVVL valve spring margin

Figure 31. Rigid, constrained-dynamics results - CVVL swing cam torsion spring margin

FLEXIBLE MULTI-BODY DYNAMICS

Rigid, constrained-dynamics simulation presented in the previous section is very useful in quickly investigating the behavior of a valvetrain and obtaining a set of nominal (quasi-dynamic) predictions. However, a true dynamic response of the system can be closer analyzed by means of a flexible multi-body dynamics simulation. A lumpedparameter, linear-elastic model, although a bit more computationally expensive as compared to a rigid model, is still feasible at a concept level of valvetrain design. Such a model takes into account the compliance of valvetrain

Figure 32. Flexible dynamics results - CVVL valve lift plotted together with kinematic lift

Figure 34. Flexible dynamics results - CVVL roller finger follower torsional deformation

Figure 33. Flexible dynamics results - CVVL valve pallet frictional power loss

Figure 35. Flexible dynamics results - CVVL swing cam contact force

INTEGRATED HYDROMECHANICAL SIMULATION OF VVA SYSTEMS

The multi-physics simulation environment of GT-SUITE software enables direct integration between the multi-body dynamics simulation tool and a 1-D fluid flow dynamics and hydraulics analysis. The ability to combine the dynamic and hydraulic models in a single simulation is essential in understanding the true physical phenomena that are taking place in the system. This, in turn, allows for rapid predictive analysis of design issues and evaluation of design changes. Examples of application of integrated simulations to the study of hydro-mechanical interactions in valvetrains can be found in [15] and [16]. A possibility to couple with hydraulics

significantly expands the scope of application of the valvetrain simulation tool, which now covers components and sub-systems such as hydraulic cylinders, cam phasers, hydraulic lash adjusters, etc. To showcase the ability to combine multi-body dynamics and hydraulics models in a single, integrated simulation run, a model of the hydraulic lash adjuster (HLA) was added to the flexible dynamics model of the CVVL valvetrain. In this setup, the main function of the HLA was to support the pivot end of the RFF and automatically control the finger arm position in order to eliminate lash and keep the finger in contact with the swing cam and the valve. A hydromechanical model of the HLA assembly consisted of the following components: high pressure chamber, which represents the outer shell/ housing of the HLA plunger, which is hollow and contains an orifice through which oil is pumped from the supply gallery into the main chamber check valve assembly comprised of a ball-shaped valve, spring and spring retainer cage main HLA spring More details regarding the hydraulic lash adjuster model, including the equation governing the evolution of pressure in the high pressure chamber, can be found in [16]. A combined hydro-mechanical analysis of the CVVL valvetrain was, again, executed at 6000 RPM at several positions of the control shaft. It required 40 seconds of CPU time per engine cycle on a 2GHz, dual-core machine. Figures 36,37,38 show various interesting results extracted from the coupled model. Figure 37. Integrated hydromechanics results - CVVL HLA chamber pressure

Figure 38. Integrated hydromechanics results - CVVL HLA leakage flow rate

SUMMARY/CONCLUSIONS
Novel features and solution methodologies of a generalpurpose planar kinematics and multi-body dynamics simulation tool were presented in the context of application to mechanically-driven, swing-cam type, variable valve actuation systems. In the same fashion, the tool can be applied to the analysis of any planar mechanism. Multiple stages of the VVA valvetrain analysis process feasible at the concept design level were described including: Figure 36. Integrated hydromechanics results - CVVL motion of RRF pivot analysis of kinematic motions of valvetrain components along with a procedure to calculate the rotational and swing cam shape profiles required to produce the desired valve lift

 rigid, constrained-dynamics simulation to extract nominal (quasi-dynamic) predictions flexible, multi-body dynamics analysis to account for elasticity of valvetrain components and inter-component contacts, frictional effects, etc. A specialized and dedicated valvetrain system analysis tool, which simplifies and accelerates the model building process, was described and used to construct VVA valvetrain models. Numerical simulation procedures were applied to analysis of the Continuously Variable Valve Lift mechanism. Results from planar kinematics and rigid dynamics simulations were validated through comparison to results from other simulation programs. A very good agreement between simulation results was shown. Furthermore, various kinematic and dynamic predictions that are of high interest to mechanical VVA system designers were presented. The generality of the simulation tool was further demonstrated by applying it to another, swing-cam type (VVLD) valvetrain. The tool's ability to include models of hydro-mechanical components and to perform integrated analysis of valvetrain dynamics and hydraulics was also demonstrated.

11. Gamma Technologies Inc.: Mechanics Theory Manual, 2009. 12. Amirouche, F.: Fundamentals of Multibody Dynamics. Theory and Applications, Birkhauser, Boston, 2006. 13. Keribar, R., A Valvetrain Design Analysis Tool with Multiple Functionality, SAE Technical Paper 2000-01-0562, 2000. 14. Gronlund, T., Larmi, M., Valve Train Design for a New Gas Exchange Process, SAE Technical Paper 2004-01-0607, 2004. 15. Okarmus, M., Keribar, R., Ham, R., Integrated Hydromechanical Simulation of a Cam-Rocker-Unit Injector System to Address Noise and Vibration Issues, SAE Technical Paper 2006-01-0887, 2006. 16. Okarmus, M., Keribar, R., Oliva, M., Tonin, N., Application of an Integrated Valvetrain and Hydraulic Model to Characterization and Retuning of Exhaust Valve Behavior with a DPF, SAE Technical Paper 2008-01-0292, 2008. 17. Delphi Corporation: www.delphi.com. 18. Adams Multibody Dynamics: www.mscsoftware.com. 19. Peterson, M., Winer, W.: Wear Control Handbook, ASME, 1980. 20. Herting, D.N.: A general purpose, multi-stage, component modal synthesis. Finite Elements in Analysis and Design, 1:153-164, 1985.

REFERENCES
1. Hannibal, W., Flierl, R., Stiegler, L., Meyer, R., Overview of Current Continuously Variable Valve Lift Systems for Four-Stroke Spark-Ignition Engines and the Criteria for their Design Ratings, SAE Technical Paper 2004-01-1263, 2004. 2. Dresner, T. and Barkan, P., A Review of Variable Valve Timing Benefits and Modes of Operation, SAE Technical Paper 891676, 1989. 3. Toyota develops continuously variable valve timing engine, WARDS Engine and Vehicle Technology Update, July 15, 1995. 4. Honda tweaks VTEC design, WARDS Engine and Vehicle Technology Update, July 15, 1995. 5. Gamma Technologies, Inc.: www.gtisoft.com. 6. Flierl, R., Hofman, R., Landerl, C., Melcher, T., Steyer, H.: The New BMW Four Cylinder Engine with Valvetronic. Part 1: Concept, Design and Construction, MTZ 62(2001)6, p450-463. 7. Variable Valve Lift and Duration Mechanism Designed for a High-Speed Engine, Engine Technology Progress in Japan, April, 2007. 8. Press, W., Vetterling, W., Teukolsky, S., Flannery, B.: Numerical Recipes in Fortran. Second Edition, Cambridge University Press, 1992. 9. Wilson, C., Sadler, J., Michels, W.: Kinematics and Dynamics of Machinery, Harper Collins Publishers, 1983. 10. Lin, Y., Ramachandra, P., Tanaka,Y., Tawata, K., Yano, Y., Sawada, R., Valve Train Dynamic Analysis And Validation, SAE Technical Paper 2004-01-1457, 2004.

CONTACT INFORMATION
M. Okarmus m.okarmus@gtisoft.com

ABBREVIATIONS
VVA variable valve actuation 1D one-dimensional 2D two-dimensional 3D three-dimensional FEM finite element method

EHD elastohydrodynamic CVVL continuously variable valve lift RFF roller finger follower RPM revolutions per minute VVLD variable valve lift and duration HLA hydraulic lash adjuster

The Engineering Meetings Board has approved this paper for publication. It has successfully completed SAE's peer review process under the supervision of the session organizer. This process requires a minimum of three (3) reviews by industry experts. 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. ISSN 0148-7191 doi:10.4271/2010-01-1193

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. SAE Customer Service: Tel: 877-606-7323 (inside USA and Canada) Tel: 724-776-4970 (outside USA) Fax: 724-776-0790 Email: CustomerService@sae.org SAE Web Address: http://www.sae.org Printed in USA