A UNIFIED INTERFACE FOR MOTORCYCLE SIMULATION IN LMS VIRTUAL.LAB

Transcription

1 The 3rd International Conference on Computational Mechanics and Virtual Engineering COMEC OCTOBER 009, Brasov, Romania A UNIFIED INTERFACE FOR MOTORCYCLE SIMULATION IN LMS VIRTUAL.LAB Nicola Cofelice 1, David Moreno Giner, Michal Manka 3 Jian Kang 4, Herman Van der Auweraer 5 1 Simulation Division, LMS international, Leuven, BELGIUM, nicola.cofelice@lmsintl.com Simulation Division, LMS international, Leuven, BELGIUM, david.morenoginer@lmsintl.com 3 AGH University of Science and Technology, Cracow, POLAND, mmanka@agh.edu.pl 4 Simulation Division, LMS international, Leuven, BELGIUM, jian.kang@lmsintl.com 5 Corporate Director RTD, LMS international, Leuven, BELGIUM, herman.vanderauweraer@lmsintl.com Abstract: Multibody dynamic methodology is an important tool for virtual prototyping of Powered Two Wheelers (PTW), where it can be very useful to analyze their dynamic behavior. However, the development of a detailed multibody model is often very time consuming, due to the complexity of the geometry of the motorcycle, to the set-up and to the unstable nature of the vehicle. With the aim of reducing the time in the simulation process, a new application has been developed in the multibody code LMS Virtual.Lab Motion. A graphic interface allows the user to easily create a motorcycle model by choosing among several front and rear suspensions and different tire models. The application provides the possibility to simulate the motorcycle behavior during normal riding maneuvers by coupling the model with a closed-loop controller. The interface and the main structure of the application are explained and first results of the dynamic simulations are presented. Keywords: Multibody dynamics analysis, motorcycle multibody modeling, virtual rider. 1. INTRODUCTION Driven by the development of the vehicle industry, in the last decades the urban area changed significantly its appearance (i.e. growing of the traffic area). This phenomenon led to an increasing development of motorcycle market, thanks to its intrinsic characteristics that made it an interesting alternative to cars (i.e. compact dimension, lightness, handling and low prices). One should bear in mind that PTW sector represents an added value of more than 1.5 bn EURO per year in the European market [1]. With the aim of studying the motorcycle dynamic behavior and reducing the time-to-market, multibody simulations can be really useful during the prototyping phase. However, due to its unstable nature, a closed loop controller is needed to make the motorcycle follow a certain path during maneuvers simulation. These aspects play an important role in the development of the motorcycle application presented in this article. This work has been developed within the Research Training Network MYMOSA (MotorcYcle and MOtorcyclist SAfety), project which has been launched by the European Community within the Marie Curie Actions []. The prime objectives of this RTN are to educate 10 Early Stage Researchers (ESR) in the partially unexplored field of Powered Two Wheelers' (PTW) rider's safety and to stimulate co-operation between researchers of 5 universities, 3 research centers and 6 industries through visits, secondments and training.. THE MOTORCYCLE INTERFACE The main objective of the motorcycle interface is to let the user create the desired motorcycle by choosing among several suspensions and two different tire models. It is also possible to modify the geometrical characteristics and the dynamic parameters of the suspensions, in order to make the application flexible and create most of the motorcycle models in the market. In fact the motorcycle classes available are manifold (i.e. sport bikes, touring bikes, cruise bikes, enduro bikes, etc); because of the wide range of design variations in the market, there are many factors that differentiate one motorcycle from the other, such as the geometry 157

2 (wheelbase, caster angle, trail, etc), the inertia properties, the materials and the different trim. By considering the motorcycle as a whole mechanism, every PTW can be divided into the same number of Sub-assemblies (or Sub-Mechanism, as term used in LMS Virtual.Lab) and afterwards assembled to create the full multibody motorcycle. The GUI (Graphical User Interface) developed allows the user choose among different models of front suspension (Telescopic Fork, Hossack/Duolever, Telelever, Bimota Tesi), rear suspension (Swinging Arm, Paralever) and of front-rear tires (as shown in Figure 1 and ). Figure 1: Front suspension and front tire models Figure : Rear suspension and frame models Those Sub-Mechanisms can be modified and then assembled using the interface; in fact, every geometrical component is defined by mean of some Hard Points that make the generation of the motorcycle totally parametric (Figure 3\left). In addition, the interface has predefined tabs where the user can customize the shockabsorber parameters, i.e. stiffness curve, damping curve and the free length. In order to perform a durability analysis, the interface is also extendable to create a motorcycle test rig by easily substituting the front and rear suspensions with different sub-assemblies (Figure 3\right). Four different actuators are implemented and their movements can be controlled by user-defined driving elements. Figure 3: GUI with the motorcycle model assembled and of the durability test-rig. Concerning the tire, two different models have been implemented: the first one is the LMS Complex Tire, which takes into account three components of force and three components of torque in the normal, longitudinal, and lateral directions. These reactions are not considered independent. The tangential reactions at the tire/ground interface depend on the normal force and the friction ellipse is used to limit the magnitude of the net tangential 158

3 force. The second model is the TNO model ; the parameters have been provided by the homonym Dutch enterprise and it contains the latest implementation by Delft-tire of Pacejka s renowned Magic Formula tire model [3]. 3. CONTROL MODEL The virtual rider controls the direction of the motorcycle by means of both a torque on the handlebar and the movement of his body, modeled as an inverted pendulum [4]. An additional input is used to control the speed of the motorcycle. Its main objective is to make the motorcycle follow a predefined path with a given speed profile. However, the motorcycle is an unstable system [5] and the controller must also stabilize it. In order to achieve this purpose, a co-simulation between MBS developed in Virtual.Lab and SIMULINK has been established, by defining a closed-loop controller (as represented in the Figure 4): RW Torque Actual Point V FW Torque Look Ahead Point F Engine Sprocket Torque Vehicle Longitudinal Dynamics Control true On/Off Sprocket Torque <Longitudinal_Speed_U> Uv Point V data FW Torque in out <Lateral_Speed_V> Vv RW Torque plantout <Yaw_Angle_D> Psi_dot Point F data Tracking Algorithm (actual and preview points ) Vehicle Steering Torque Tau Look Ahead Point F Actual Point V Lateral Dynamics Control [PID] Figure 4: Scheme of the co-simulation Simulink-Virtual.LAB used for control the motorcycle model As shown in the picture, the virtual-rider can be divided in 4 different blocks: Tracking Algorithm (implemented in Matlab\Simulink): in order to make the motorcycle follow a predefined path, it is necessary to know its relative position with respect to the path for each integration step. Basically, the position of the motorcycle is described by a set of curvilinear coordinates s, n and θ [6]; s is the travelled distance along the reference path, n is the lateral error and θ is the relative angle between the path and the motorcycle. In the Table 1, the set of equation and the scheme of the tracking algorithm is presented, where ψ& is the yaw velocity of the motorcycle, κ( s) is the curvature of the reference path defined as a function of the travelled distance s and u the longitudinal speed of the motorcycle. Table 1: Equations of the tracking algorithm s θ κ ( s) = ψ& t t (6) s [ 1 nκ( s) ] = u cos( θ) vsin( θ) t (7) n = v cos( θ) + u sin( θ) t (8) 159

4 Longitudinal Dynamics Control (implemented in Matlab\Simulink): this algorithm is a PID controller with look-ahead, used to control the speed of the motorcycle. Here the term look-ahead means that the control action (i.e. the thrust or braking force) needed to follow a predefined speed profile is computed based on the difference between the actual speed of the motorcycle and the desired speed at a future point (i.e. the look-ahead point) which travels a distance L A ahead of the motorcycle. L a0+ a1u+ a u = f (u, κ) = A 1+ b + (4) b 1 κ κ This distance is used to anticipate the response of the system so it starts accelerating/braking before the motorcycle reaches the reference point, which is always moving ahead of the motorcycle. The output of the longitudinal controller is a single signal and, before introducing it to the multibody model, it must be split in three different components since the motorcycle has three inputs for the longitudinal control: thrust, front brake and rear brake. Lateral Dynamics Control (implemented in Matlab\Simulink): the main objective of the lateral dynamics controller is to stabilize the motorcycle so that it follows a certain path. In order to avoid the fall of the motorcycle, the controller should be able to balance the motorcycle s centrifugal and gravitational forces. For this purpose, the roll angle derived form the steady-state equilibrium condition [5] will be used as the main reference of the controller (as show in the eq. 5). mac u κ( s) φ reference = arctan = arctan + φrider mg g (5) The steering torque applied to the handlebar has been evaluated by using a specific PID controller (as described in the Table ). Table : steering torque terms Equation Physical meaning φ [ ( )] ( t) τ 1 = Pφ φ reference φ t + Dφ (6) Correct roll angle error t n ( ) ( ) ( t) τ = Pn n t + I n n t dt+ Dn (7) Correct path error t δ( t) τ 3 = Dδ (8) Stabilize wobble mode t The leaning angle of the rider s upper body (α), which has one DOF with respect to the main frame of the motorcycle, was computed by means of two different terms as shown in eq. 9. The first term of this equation makes the rider lean proportionally to the roll angle of the motorcycle and the second acts when the curvature changes, for instance, when entering or leaving a curve. = Pφ t + f & s (9) α α ( ) ( ( )) rider κ MBS model (implemented in Virtual.LAB): this block represents the multibody model of the motorcycle. The outputs are represented by the states of the motorcycle and the inputs are evaluated by the Matlab\Simulink controller. 4. DYNAMIC SIMULATION In this section the multibody model of the motorcycle and the result of one maneuver are analyzed. As it is shown in the Figure 5, the motorcycle is composed of seven rigid bodies: the front wheel, the lower part of the fork, the upper part of the fork (including the handlebars), the front frame (including the engine and the fuel tank), the swinging arm, the rear wheel and the rider s torso (not represented in the Figure 5). The selected suspensions are Telescopic Fork for the front axle and Swinging Arm - Monoshock for the rear axle, while the Complex model has been selected for both front and rear tire. With this configuration, the motorcycle model has a total of twelve degrees of freedom [7] that can be decomposed in the following way: six from the front frame (3 coordinates of the centre of mass together with the roll, pitch and yaw angles), one from the steering angle, two corresponding to the rotation of the wheels and 160

5 another two from the suspensions. As explained in the previous section, the virtual rider has one degree of freedom with respect to the front frame. This approach is valid for the most common maneuvers. Figure 5: Scheme of the tracking algorithm and equation implemented in Simulink With the closed-loop controller, several maneuvers have been simulated [8]. Figure 6 shows the closed-loop results for the reference path shown in Figure 6.a. The simulated maneuver is a eight-maneuver, that allows studying the steady state behavior of the system motorcycle-rider. As it is shown in Figure 6.a, the controller makes the motorcycle follow the trajectory with satisfactory results. The first graph on the right shows the response of the steering angle during the maneuver. Two important things can be observed: the presence of two regions in which the motorcycle angle reaches the steady-state (i.e. the roll angle becomes constant) and the peaks of the steering angle signal before those regions (counter-steering maneuvers). In the second graph, the roll angle of the motorcycle is shown. Figure 6: Closed loop result without considering the rider s movement Figure 7 shows a comparison between the results obtained without considering the movement of the upper part of the rider s body (red lines) and those obtained considering it (blue lines). As can be observed, the movement of the rider facilitates the maneuver; the required torque is smaller when the rider is active. 161

6 Figure 7: Closed loop result considering the rider s movement 5. CONCLUSION The research contents of this article include two parts. The first part describes the motorcycle toolbox developed in Virtual.Lab Motion, which allows a rapid generation of multibody motorcycle model. Every component is parameterized, in the same way as the suspension dynamic behavior. The second part of this research is to apply the control algorithm on the vehicle and the rider. The controller has been coupled with a multibody model of the motorcycle generated with the toolbox. First results of simulated maneuvers have been presented, paying close attention to the effect of the introduction of the rider s torso (simulated as an inverted pendulum). This application can be improved in the future by comparing the dynamic behavior of different motorcycle models and by introducing the human limit for the steering torque. Currently, the research is mainly focused on the development of the test-rig reverse load identification [9], an application that will led to the drive file of the dynamic actuators. 6. ACKNOLEDGEMENTS The work presented in this paper was performed within the 6th framework of the Marie Curie Host Fellowships: MYMOSA project (MRTN-CT ). The authors would like to thank the European Commission for the grant received and we also would like to thank TNO for the Magic Formula tire model. REFERENCES [1] ACEM - Association des Constructeurs Européens de Motocycle, Brussels, 009 [] Marie Curie Actions 6th framework project- Brussels, 008 [3] TNO MF-Tyre & MF-Swift 6.1 User manual 008, The Netherlands, 008 [4] Moreno Giner D.: MYMOSA Towards the simulation of realistic motorcycle maneuvers by coupling multibody and control techniques, ASME 008, Boston [5] Vittore Cossalter, Motorcycle Dynamics 00, Greendale, WI: Race Dynamics. [6] Vittore Cossalter, Mauro Da Lio, Roberto Lot and Luca Fabbri. A General Method for the Evaluation of Vehicle maneuverability with Special Emphasis on Motorcycles, Vehicle System Dynamics, 1999, Vol. 31, pp f. [7] Dick Kading, Multibody Dynamics theory and Modeling issues, LMS internal training course script, 00 [8] Claudio Brenna, David Moreno, Ioannis Symeonidis, Gueven Kavadarli, Steffen Peldschus. MYMOSA Towards a virtual motorcycle rider for realistic simulations of motorcycle maneuvers. International Motorcycle Conference, October 008, Cologne (Germany). [9] Joris De Cuyper, Linear Feedback Control for Durability Test Rigs in the Automotive Industry, PhD Thesis 006, Katholieke Universiteit of Leuven. 16