TRANSFER CASE CLUTCH TORQUE ESTIMATION IN AWD VEHICLES

Transcription

1 F6-VDCA-37 RANSFER CASE CLUCH ORQUE ESIMAION IN AWD VEHICLES Jung, Hojin * ; Choi, Seibu; Korea Advanced Institute of Science and echnology, South Korea KEYWORDS orque observer, ransfer case, Kalan filter, Wet clutch odel, Driveline odel ABSRAC AWD syste transfers soe portion of engine torque fro the ain drive shaft to the sub drive shaft. Recently, electronic otor actuated type AWD syste has been eerged to control the vehicle ore actively. Copared with a echanical type AWD, electronic type AWD enables the variable torque distribution toward the sub drive shaft with the wet clutch friction aterial. his paper proposes the clutch torque estiation ethod of the electronic type transfer case. o copensate the uncertainty of clutch paraeters, vehicle dynaics odel is integrated in the torque estiator to utilize vehicle inforation with the sensor signal which is obtained fro the vehicle can. By ipleenting carsi based siulation environent setup of AWD syste, torque estiation is validated with several driving conditions. INRODUCION AWD has a roll to distribute engine torque to drive both front wheels and rear wheels, which aes a difference to conventional WD vehicle. Recently, the application of AWD syste has been increasing in high perforance passenger car due to the iproveent of the engine power. Electronically actuated AWD syste which is introduced in this paper has different hardware constitution fro center differential 4WD and selective 4WD because it can control the aount of the distributed drive torque between the ain drive shaft and the sub drive shaft by controlling the engaging force which is applied to the wet clutch of the transfer case. In control aspects, estiated vehicle states lie slip ratio, slip angle, friction coefficient are coonly used as decisive factors in vehicle active safety syste lie ABS, ESC and AFS. For the vehicle state estiation, drive torque inforation is generally required in vehicle dynaics odel. In case of WD syste, engine torque which is obtained fro ECU directly applies to derive the drive torque and warrants the high accuracy. However, engine torque inforation is not enough in AWD syste because the drive torque of both front and rear shaft can be different depending on the aount of clutch engaging force. Variable clutch engaging force actuation of AWD is obviously different fro short transient phenoenon lie gear shifting, thus clutch torque estiation logic developent in AWD syste should be preceded for iproving electronic type AWD vehicle perforance by helping other principal vehicle states estiation in real tie. his paper suggests the novel estiation ethod of AWD clutch torque. Copared with the other previous wors to estiate the transitted clutch torque, vehicle dynaics odel is integrated in the clutch odel to utilize the vehicle sensor signal which helps the accuracy of the estiator as a copensation feedbac. Fro the siulation environent setup by ipleenting AWD syste as an external siulin odel in carsi, torque estiation is validated with various driving scenarios. SYSEM MODELING. Drivetrain Dynaic Model

2 AWD syste which is dealt in this paper can control the aount of torque by using ultiple disc plates of wet clutch. And vehicle dynaic states with AWD syste can change drastically depending on the aount of torque which is transferred to front or rear shaft. hus siulation description should be available of variable distribution of front shaft torque depending on clutch engageent force especially between clutch slipping state and clutch loc up state. Scheatic diagra of driveline odel fro transission output shaft to front and rear differential gear can be shown in the figure and atheatical odels are as follows: Figure : Scheatic diagra of AWD transfer case syste t c r Jtt () f F () c c, / i J (3) c f f c c f f c f bf c f (4) r r t r br t r (5) Where t is the transission output torque, c is the clutch engageent torque, f and r are the front and rear shaft torque each. Jt is the ain drive shaft inertia, Jc is the front drive shaft inertia and Fc is the clutch engageent force. ω is the clutch slip angular velocity. At the end of each drive shaft can be odelled as spring and daper syste to represent the distortion of shaft. ransitted clutch torque is expressed as a certain function of f (Fc, ωslip) and its detailed expression will be dealt in section.... Wet Clutch Model o ae the variable torque distribution by allowing clutch slipping for quite a long tie, wet clutch is generally used in electronic AWD by overlapping ultiple disc plates to transfer large aount of torque. o prevent the instantaneous disc wear, wet clutch absorbs the heat which is generated in friction plate by circulating oil continuously... Generalized Maxwell Slip (GMS) Model o guarantee the validity of siulation, appropriate friction odel for the wet clutch syste is iportant. GMS odel [,], which was proved for friction odel was selected for odelling transfer case syste. When clutch is fully loc up, differential equation of clutch transfer torque is as follows: dmi i p t dt, (6) Where i (p, t) is the function of clutch engageent force and transission output torque. When clutch is slipping, differential equation of clutch transfer torque is as follows: dm i M i sgn C i dt s, p (7)

3 Where s(ω, p) is the stribec equation about torque, which is expressed as follows: s, p sgn M s M c M s e V s p (8) Wet clutch should consider viscous effect by the oil fil between disc plates. Especially for wet clutch, drag torque contributes soe part of transitted torque so that this should be odelled carefully to iprove the accuracy of transient response [4]. Drag torque of wet clutch is originated fro the shear stress between plates and can be odelled by fluid dynaic analysis [4-6]. hen, viscous friction torque equation is obtained as follows: M vis r 3 o r.94 h dr (9) r h. Re Where ri is a clutch inner diaeter, ro is an effective diaeter due to the centrifugal force, which is usually less than clutch outer radius. Reh is a Reynolds nuber of Autoatic ransission Fluid (AF), which is written as follows: Re h rh () Where μ is a viscous coefficient of AF, ρ is a density of AF and h is a clearance length of clutch plate. Finally, transferred torque when clutch is slipping is the su of the two effects: c Mi M vis () o guarantee the feasibility of wet clutch engageent siulation, clutch experiental friction coefficient data for coercial vehicle ipleentation was used. 3 ESIMAOR DESIGN 3. Vehicle Dynaics Model o iprove the accuracy of the estiated clutch torque given paraeter uncertainties, vehicle dynaics odel which can connect clutch torque state to other states can be used. Since conventional vehicle dynaics states can be copensated by vehicle easureent signals, clutch torque can also be corrected indirectly due to the relation of equations. By integrating vehicle longitudinal dynaics, lateral dynaics, yaw dynaics and wheel dynaics, clutch torque can be estiated by extended Kalan filter [7,8]. he equations of otion for the vehicle based on bicycle odel can be categorized by three dynaics. Longitudinal dynaics is: v cos sin sin x F xf F yf F xr air C d Av x g v y r () Where vx is the longitudinal velocity, is the vehicle ass, Fxf is the front longitudinal tyre force, Fxr is the rear longitudinal tyre force, Fyf is the front lateral tyre force, δ is the tyre

4 steer angle, ρair is the air density, Cd is the drag coefficient, vy is the lateral velocity, r is the yaw rate and θ is the hill angle. Lateral dynaics is Where Fyr is the rear lateral tyre force. Yaw dynaics is: vy Fxf sin Fyf cos Fyr vxr (3) r l F F l F I f xf sin yf cos r yr (4) z Where Iz is the vehicle yaw oent of inertia, lf is the distance fro the center of ass to the front axle and lr is the distance fro the center of ass to the rear axle. Another dynaics in which force ters are related is wheel dynaics that is expressed as follows: I d Re Fxi Re Rri (5) w Where ω is the wheel angular velocity, Iw is the wheel inertia, d is the wheel drive torque. In this paper, d is assued to be nown and does not consider braing situation. Re is the wheel effective radius. Fxi and Rri are the tyre longitudinal force and the tyre rolling resistance whose subscript i includes fl (front left), fr (front right), rl (rear left), rr (rear right) which is written throughout the paper. 3. Extended Kalan Filter Design o cobine vehicle dynaics odel which is entioned in section 3. with wet clutch odel which is entioned in section. to estiate the clutch torque, extended Kalan filter was adopted in this paper. he states for Kalan filter are defined as follows: Where x,,, ω,, (6) t vx vy r c fl fr, rl rr ω (7) Here, the aount of noinal clutch disc friction coefficient change was added as a state to deal with the uncertainty of clutch paraeter value. he easureents fro vehicle signals are as follows: Where ax and ay are related as: y t ax, ay, r, ω (8) a x F xf cos f F yf sin f F xr sin air C d Av x g (9)

5 cos a F sin F F y xf f yf f yr () o su up the above equations, the state space syste is obtained as follows: x t aircd Av x Fxf cos Fyf sin Fxr g sin vyr Fxf sin Fyf cos Fyr vxr vx v y l f Fxf Fyf lrfyr r I sin cos z fl fr i ff c ReFxf ReRrf f x t, u t Iw rl rr c t c i fr Re Fxr ReRrr I w c sgnc Fc rc n sgn Fxf Fyf Fxr aircd Avx g a cos sin sin x a y y t Fxf sin Fyf cos Fyr h t u t r x, ω r ω () () Please eep in ind that above state space for applies only when clutch is slipping. Viscous ter is not appeared in the equation but its effect is not doinant for estiation. When clutch is fully loc up, estiated clutch torque value which is obtained fro the above equation does not coply with the real value anyore so that clutch torque should be dealt with differently. In a rigid shaft, the aount of transferred torque at both sides of the shaft is proportional to load. hus the wheel vertical load contributes the transferred aount of torque in AWD vehicle. hen front and rear weight shifting was considered to deterine the aount of distribution torque. Shifted weight was calculated fro the following equation: l hcg l r f hcg Fzf g ax, Fzr g ax (3) l l l l hen distributed torque is deterined as follows: Fzf Fzr f t, r t F F F F zf zr zf zr (4) he algoriths of discrete tie extended Kalan filter can be expressed as follows [7]: xˆ xˆ f x, u d (5) t t P P F P P F LL d (6) t t K P H H P H R (7) h u t x ˆ x ˆ K ˆ, y x (8)

6 P I K H P (9) Where F and H are the linearization of equation () and () around step. P is the error covariance and L is the disturbance input atrix. 4 SIMULAION xˆ at each tie In this paper, carsi software was used to validate the AWD syste driving siulation. ransfer case developed by siulin was ipleented in carsi. Except for the transfer case part, carsi internal odel was used. o represent the hydraulic pressure response generated y otor pup, AMESIM odel was applied in the clutch engaging force input. Vehicle driveline is based on Front ship Rear wheel drive (FR) type. Estiator saple tie wat set to.s but easureent signal was updated by.s. At first, vehicle starts to ove fro a stop state with a steady throttle. And a step throttle is inserted between 5-5 seconds. Meanwhile, AWD is activated between 5-5 seconds to analyze the responses in gear upshifting and downshifting. AWD is also activated between 3-35 seconds for the validation for clutch loc up engageent. o consider various driving conditions, estiator was siulated at both high u and low u surfaces. o verify the robustness of integrated clutch torque estiation, % uncertainty of ass and clutch u paraeters were applied in the siulation. orque [N] otal output torque Front shaft torque Rear shaft torque gear* throttle* 3 4 ie [s] (a) ie [s] (b) Figure Clutch torque estiation in high u (.85) (a) Drivetrain response (b) clutch torque estiation result orque [N] otal output torque Front shaft torque Rear shaft torque gear* throttle* 3 4 ie [s] (a) Figure 3 Clutch torque estiation in low u (.3) (a) Drivetrain response (b) clutch torque estiation result Figure (a) and figure 3(a) show the response of driving situation which is described in the above paragraph. Figure (b) and figure 3(b) show the siulation result of AWD clutch torque estiation in each road surface. otal output torque eans engine torque ultiplied by transission gear ratio. Front and rear shaft torque are the aount of torque which is delivered to the each shaft. o grasp the driving situation easily, throttle and gear shifting response are also added in the figure. In both conditions, clutch torque was underestiated orque [N] orque [N] Clutch torque Estiated clutch torque Engaging force*. Clutch torque Estiated clutch torque Engaging force* ie [s] (b)

7 initially at certain engaging force, then followed real clutch torque value at low steady throttle region. At severe transient slipping region of rapid acceleration, high clutch slip caused spie-shaped estiation error and it was worse with strong engaging force. However, error was reduced as slip aount becoes low. As can be seen in high steady throttle region, estiated clutch torque didn t converge the real value since clutch dynaic odel is incoplete. Obviously, no discrepancy appeared in loc up condition at the range of 3-35s. 5 CONCLUSION In this paper, clutch torque estiation ethod using vehicle dynaics odel was suggested. o consider the feasibility of real tie application, siulation was conducted at both of high u and low u surfaces. With clutch u paraeter and vehicle ass uncertainties, clutch torque estiator showed robust perforance by copensating clutch u discrepancy. Proposed torque estiation algorith can be applied for the vehicle state estiation and tyre paraeter estiation, ultiately vehicle dynaics control area, for the logic developent of advanced AWD driving. However, estiator experiental validation is required for future wor. ACKNOWLEDGEMEN his wor was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea governent (MSIP) (No. -868). his research was supported by the MSIP (Ministry of Science, IC and Future Planning), Korea, under the C-IRC (Convergence Inforation echnology Research Center) (IIP- 6-H86-6-5) supervised by the IIP (Institute for Inforation & counications echnology Prootion) REFERENCES [] Agusian Partogi Opusunggu, Paul Sas, Hendri Van Brussel, Modeling and siulation of the engageent dynaics of a wet clutch syste subjected to degradation: An application to condition onitoring and prognostics [J]. Mechatronics, 3. [] Vincent Lapaert, Farid Al-Bender, Jan Swevers, A Generalized Maxwell-Slip Friction Model appropriate for Control Purposes [J]. IEEE Physics and Control Proceedings. International Conference, 3. [3] Jiwon J. Oh, Seibu B. Choi, Jinsung Ki, Driveline odeling and estiation of individual clutch torque during gear shifts for dual clutch transission [J]. Mechatronics, 4. [4] Hirou HASHIMOO, Sanae WADA and Yoshiai MURAYAMA, An Influence of Centrifugal Force on the Perforance of hero-hydrodynaic urbulent hrust Bearings [J]. Bulletin of JSME, 984, 7, No. 8. [5] Yiqing Yuan, Eysion A. Liu, Jaes Hill, Qian Zou, An Iproved Hydrodynaic Model for Open Wet ransission Clutches [J]. ASME Journal of Fluids Engineering, 7. [6] Youhei aagi, Yasunori Oano, Masatoshi Miyayaga and Nobuyui Katayaa, Nuerical and Physical Experients on Drag orque in a Wet Clutch [J]. Japanese Society of ribologists,, pp [7] Laura R. Ray, Nonlinear state and tire force estiation for advanced vehicle control [J]. IEEE RANSACIONS ON CONROL SYSEMS ECHNOLOGY, 995, 3, NO.. [8] Mooryong Choi, Jiwon J. Oh and Seibu Choi, "Linearized Recursive Least Square Methods for Real-tie Identification of ire-road Friction Coefficient" [J]. IEEE RANSACIONS ON VEHICULAR ECHNOLOGY, 3.