Modeling And Simulation For NVH Evaluation

Size: px

Start display at page:

Download "Modeling And Simulation For NVH Evaluation"

Dominick Harrison
6 years ago
Views:

1 Proceedings of the World Congress on Engineering nd Computer Science 007 WCECS 007, Octoer 4-6, 007, Sn Frncisco, USA ing And Simultion For NVH Evlution Wei Zhn Astrct A new methodology sed on modeling nd simultion is proposed for Noise, Virtion, nd Hrshness (NVH) nlysis for Anti-ock Brke Systems (ABS). First, correltion etween NVH mesurement dt nd simultion results need to e estlished. This reltionship llows engineers to focus on modeling nd simultion insted of NVH testing. The model cn e set up to do different level of simultion nd NVH nlysis. If some dt is ville from ctul testing, then the test dt cn e esily plugged into the model to replce the corresponding prt in the model. It is especilly useful when the design needs to e modified, or trde-off etween ABS performnce nd NVH is necessry. The model cn gretly reduce the time to mrket for ABS products. It lso mkes system level optimiztion possile. Index Terms Anti-ock Brke Systems, ing nd Simultion, Noise, Virtion, nd Hrshness, Vehicle Systems. I. INTRODUCTION NVH is one of the importnt mesures (see []) for mny utomotive systems such s Anti-ock Brke Systems. It is estimted tht the NVH relted cost to the utomotive industry is over one illion dollr every yer. In order to improve NVH, one must e le to effectively evlute NVH for given system. This is chllenging tsk since most of the NVH evlutions re done in sujective mnners (see []). Once n NVH evlution method is chosen, the evlution could e conducted mny times s design chnges re mde during the entire product development process. This is very time consuming nd inefficient wy of designing products. CAE is widely used for NVH study. Mny NVH simultors re ville for utomotive systems. However, the rke NVH modeling nd simultion efforts re mostly limited to foundtion rke (see [4-6, 9-]). For ABS systems, NVH is still eing studied nd evluted minly y testing. Usully the focus in the erly development stge is on the ABS performnce (see [3, 4]). The NVH spect is limited to simple theoretic nlysis. After the desired performnce is chieved, the engineers then try to improve NVH sed on feedck from customers nd test engineers. The vst mjority of the NVH work for ABS systems is done in NVH ls or vehicles. More often thn not, it is found tht there re conflicting demnds from ABS performnce nd NVH. Sometimes, the design process hs to e repeted to hve etter NVH result. Not only the process is time consuming, costly, ut lso the result is neither relile nor consistent. In this pper we propose new systemtic pproch for NVH study for ABS systems. First, extensive nlysis should e done using existing test dt. A good understnding of the reltionship etween system components nd NVH needs to e estlished. Second, modeling nd simultion should replce most of the ctul testing or sujective evlution. Third, NVH/performnce optimiztion/trde-off must e done t the system level, not t the susystem level. The dvntge of this new pproch is cler: The end result is optimized for the overll system; The time it tkes to complete the project is gretly reduced; The result is more consistent since the inconsistent humn fctor is tken out of the evlution process; The cost is frction of current prctice; This lso llows dvnced fesiility study efore ny hrdwre/test trck is ville for testing. We use ABS to illustrte the min ides. However, the principl of the new pproch cn e pplied to mny other NVH relted pplictions. II. MODEING A typicl hydrulic schemtic for ABS system is depicted in Fig. (Only one wheel is shown). When the driver pplies force to the pedl, the mster cylinder increses. This cuses the wheel to increse. If the wheel exceeds the limit tht is determined y the rod surfce friction coefficient, the wheel will e locked up. A locked-up wheel would not llow us to chieve optiml lterl nd longitudinl forces to mintin vehicle stility, steerility nd short stopping distnce. Once the ABS lgorithm detects this condition, the microprocessor would send commnd to switch the isoltion vlve to the closed position. This prevents ny rke fluid from flowing to the wheel. Next, the dump vlve cn e opened so tht the rke fluid in the wheel cn flow to the ow Pressure Accumultor (PA). This completes cycle for wheel regultion. In the next cycle, the isoltion vlve is pulsed open to llow smll mount of fluid to flow to the wheel. This strts the next regultion cycle. The rke fluid dumped to the PA needs to e pumped ck to the mster cylinder so tht there is enough rke fluid for the next regultion cycle. The pump is needed for nother reson: once the PA is full, then no wheel regultion cn e done (see [3]). ISBN: WCECS 007

2 Proceedings of the World Congress on Engineering nd Computer Science 007 WCECS 007, Octoer 4-6, 007, Sn Frncisco, USA Driver pplied Isoltion Vlve Mster Cylinder Pump low Wheel regulted ow Pressure Accumultor Dump Vlve Motor The noise is mostly cused y the wter hmmering effect when the vlves re opened nd closed nd when the pump moves rke fluid ck to mster cylinder. The virtion of the rke pedl is minly cused y the rke fluid movement due to the vlve nd pump ctivities. If there is more fluid flowing from mster cylinder to the wheel thn wht is eing pumped from PA to mster cylinder, then the driver will feel the pedl dropping. In the worst cse, the pedl cn drop ll the wy to the floor. On the other hnd, when the pump delivers more fluid to mster cylinder thn the mount of rke fluid tht flows to the wheel, then the driver will feel the pedl rising. During n ABS event, the pedl my move up nd down constntly. The pedl virtion or pedl feel, terminology commonly used y test engineers, is mesure for the qulity for this virtion. First we study wht elements of the ABS system ffect NVH. Bsed on test dt, the noise nd virtion for ABS is ffected y the following: () MC chrcteristics () Motor speed (c) Hydrulic design for the ABS unit (d) ABS control lgorithm (e) oction of the ABS unit (f) Noise isoltion cpility (g) Mounting of the ABS unit Items (e) nd (f) re usully determined y the OEMs. The other items re usully designed y rke suppliers. et us first nlyze the reltionship etween system components nd NVH. Noise for ABS is minly cused y the pulstion in the mster cylinder (MC). The MC is determined y the driver input, i.e., pplied pedl force, nd the Fig.. A Typicl ABS System pump nd isoltion vlves ctivities. The pump is driven y DC permnent mgnetic motor, which is in turn controlled y the ABS lgorithm. The vlves re lso controlled y the ABS lgorithm. The vlve commnds nd the MC determine the t the wheels. The PA cn ffect the wheel s in the sense tht when it is full, the wheel cn not e reduced. When the PA is not full, its impct on the wheel is miniml. The MC nd PA vlue determine the pump lod, which directly ffects the motor speed. The pedl virtion is determined y the MC nd the pump flow. The system level model is shown in Fig.. Next we define ech lock in the system model. To mke the equtions esier to red, we use only one wheel, one pump, nd one PA in the model. The ctul system of course hs 4 wheels, pumps nd PAs. Noise nd Brke Pedl A one time initil test is necessry to estlish the reltionship etween NVH nd some system prmeters tht cn used for further simultion nd nlysis. This cn usully e done using existing test dt. During NVH test, mny other vriles re recorded. Becuse of the significnt impct of mster cylinder hs on NVH, the trce is lwys recorded. We expect the fundmentl structure of the reltionship etween the, flowrte nd NVH to e the sme for different vehicle models. The only difference etween different vehicle models will e the coefficients in the model. ISBN: WCECS 007

3 Proceedings of the World Congress on Engineering nd Computer Science 007 WCECS 007, Octoer 4-6, 007, Sn Frncisco, USA Algorithm : Motor control Driver Input Motor Pump Mster Cylinder Noise Algorithm : Vlve control ow Pressure Accumultor Brke Pedl Brke Pressure Therefore, once the fundmentl structures of the noise nd virtion models re estlished for one vehicle, we only need limited numer of test trces to clirte the noise nd virtion model. This my seem to e s time consuming s the commonly used tril-nd-error method. But the dvntge of modeling nd simultion for NVH evlution is pprent when some prt of the system design is modified. In this cse, we only need to modify the model of the corresponding prt nd usully do not need to repet ll the NVH testing. One of the min contriutors to NVH is the pulstion due to the pumping of the rke fluid from the PA to MC driven y the motor. We propose to nlyze test dt to find the correltion etween the mgnitude nd frequency of the pulstion nd the noise level. The reltionship etween the pedl virtion nd hydrulic fluid movement cn e modeled using the test dt nd curve-fitting s follows N = f ( P, ) () MC F dd p r o ( F F ) Fig.. System for NVH simultion = π () V = f ( f, M ) (3) pedl pedl pedl where, P MC is the mster cylinder ; F is the frequency of the P MC oscilltion; D p is the pedl displcement; r is the rdius of the mster cylinder ore; F o is the pump output flowrte; F i is the flowrte to the wheel; V pedl is the mesure of virtion of pedl; N is the noise mesure; f pedl is the frequency of the pedl displcement nd M pedl is the mximum pedl displcement. Motor nd Pump Motor nd pump cn e modeled s follows i where di R = e T = K i m e = K d θ = K o i dθ T J m T J dθ i e (4) F = EK (5) = K i ; K i is the torque constnt, in N-m/A; K is the ck-emf constnt, in V/rd/sec; i () t is the rmture current, in A; R is the rmture resistnce, in Ω ; e ()is t the ck emf, in V; T () t is the lod torque, in N-m; Tm ()is t the motor torque, in N-m; θ is the rotor displcement, in rd; is the rmture inductnce, in H; e ()is t the pplied motor voltge, in V; J is the rotor inerti, in kg- m ; K is the speed to flowrte gin; F o ( is the pump output flow rte; nd E is the pump efficiency. Algorithm: Motor nd Vlve Control The ABS lgorithm clcultes the commnds for motor nd vlves in every lgorithm loop, typiclly of 5 to 0 milliseconds. The commnd for the motor cn e n on-off signl for the motor, which cn e result of some motor speed control lgorithm. We cn use the on-off commnd s n input or include the motor speed control lgorithm in the model, in which cse, the input would e speed request. The commnds for vlves cn e simply on or off. The commnd for motor is use y the motor model to control the motor speed. The commnds for vlves re used y PA nd Brke Pressure models to estimte t the wheels nd the volume in ISBN: WCECS 007

4 Proceedings of the World Congress on Engineering nd Computer Science 007 WCECS 007, Octoer 4-6, 007, Sn Frncisco, USA PA. These commnds cn e the input of the system model; it cn lso e the interfce etween this model nd n externl model such s ABS model tht includes the ABS lgorithm. ow Pressure Accumultor The volume in the PA is determined y the fluid coming from the wheel nd the fluid going to the pump. This cn e modeled s dv PA = Fo Fo( (6) Where V PA ( is the rke fluid volume in PA; F o ( is defined in (5); nd F o ( is defined in the rke model in (0). During development testing, the PA volume informtion is usully cptured. In this cse, the volume could e fed to the system model s n input. Brke Pressure The rke cn e determined using the vlve commnds nd the rke PV curve. P = PV ( V ) (7) V = V V (8) dv i ( t ) i o = C P P C (9) MC ( ) dv o t = C P Cdump( (0) where PV is the rke PV curve; C is the flow coefficient for the isoltion/dump vlves; V i ( is the volume flow into the rke nd V o ( is the flow out of the rke; C ppl ( nd C dump ( re the commnd for isoltion nd dump vlves (they re digitl signls equl to either 0 or ). During development testing, the rke is usully cptured. In this cse, the clcultion in (7) could e replced s n input. Mster Cylinder The mster cylinder is the min source during n ABS event. We cn ssume the drive pplies constnt force, rmp pply, spike pply, or pedl pumping. The mster cylinder model determines the hed tht the pump is working ginst. d x( m = F πr P () pedl PMC = PVMC ( VMC ) () VMC = Vi πr x( (3) where P MC is the MC, V MC ( ) is the volume in MC; r is the MC piston ore rdius; x( is the MC piston displcement. The PV curve PV MC ( of the mster cylinder cn e tested. ppl MC Evlute Impct of Motor Speed et s sy two different trces of the motor speed re otined from vehicle testing y clirtion engineer. Assuming other prt of the system re the sme, we cn esily evlute the NVH impct of the motor speed difference. In this cse, we hve no need to model the motor/control. Simply feed the motor speed to the pump input nd the output is the NVH result. Evlute Impct of Pump Size nd Efficiency If the pump dimeter or efficiency is chnged during the design process, we cn use the system model to reclirte the motor speed controller. The NVH result cn then e re-evluted using the simultion tool. Evlute Impct of ABS Algorithm Clirtion We cn use the system model to optimize the ABS lgorithm clirtion (optimize the control prmeters). This includes the vlve control nd motor control too. Typiclly, n ABS lgorithm goes through severl rounds of clirtion. We cn use the model to come up with NVH results for different clirtion sets. This would then llow us to do trde-off study etween ABS performnce nd NVH. Evlute Impct of Mster Cylinder Design Some design prmeters of the mster cylinder cn e vried to give suggestion for the optimized vlues. During the product development process, it is not uncommon tht the ctul hrdwre is chnged. The model cn e used to give quick estimtion of wht the impct would e. In ddition to the simultion tht cn e done using the system model proposed in this pper, we cn further improve the simultion cpility y including more components in the model. For exmple, the hrdwre durility is very importnt spect of ABS product. Only few ppers cn e found (e.g. [7]) tht ddress the performnce issue together with durility nd NVH. For ABS systems, durility is minly determined y the temperture t the semiconductor switches nd the current through the motor coil. The reltionship is not strightforwrd. We propose to use test dt to crete n empiricl model tht uses motor current s input nd durility performnce s the output. Eventully the whole vehicle model cn e dded using softwre such s Crsim or Mtl. Then we cn do virtul testing on the desk top for overll system evlution including performnce, NVH, nd durility. This will e very significnt progress. We cn lso do hrdwre in the loop simultion ([8]). The dvntge of these simultions is tht we cn design the test condition ritrrily. I. SIMUATION The model developed in the previous section cn e used in mny different wys for NVH nlysis for ABS systems. II. CONCUSION It is time consuming to try different comintions of hrdwre/softwre if we hve to do NVH testing for ech design. We propose new method for NVH nlysis using simultion. An ABS system is nlyzed. A set of equtions is ISBN: WCECS 007

5 Proceedings of the World Congress on Engineering nd Computer Science 007 WCECS 007, Octoer 4-6, 007, Sn Frncisco, USA derived. A simultion model cn e uilt sed on these equtions. The purpose of the simultion method is to provide first cut estimtion of NVH results for ABS products. This method cn e used in different wys, such s comprison of different controllers or trde-off study for ABS performnce nd NVH. We re still in the preliminry stge of developing this new method. A lot of work still needs to e done in order for this new pproch to work. The effectiveness of this pproch needs to e vlidted y rel system. We re lso investigting the fesiility of extending this model to include the system durility nlysis. If it works, we cn perform system level optimiztion over ABS performnce, cost, NVH, nd durility. III. REFERENCES [] N. Otto, S. Ammn, C. Eton nd S. ke, Guideline for Jury Evlutions of Automotive Sounds, Proceedings of the 999 Noise nd Virtion Conference, [] J. Fry, P. Jennings, R. Willims nd G. Dunne, Understnding how customers mke their decision on product sound qulity, Proceedings of the 33 rd Interntionl Congress nd Exposition on Noise Control Engineering, pper 70. [3] Fennel H, Kole A, Honus K. Method And Circuit Arrngement For Controlling The Flow Rte Of A Hydrulic Pump. U.S. Ptent No. 5,704,766, Jnury 6, 998. [4] Nck, W., Joshi, A. M., Friction Induced Virtion: Brke Mon, SAE pper 95095, 995. [5] Misr, H., Johnson, E., Komzsik,., NVH Optimiztion on NEC Supercomputers Using MSC.Nstrn, Proceedings of the st Worldwide MSC Aerospce Users Conference, ong Bech, CA, June 7-0, 999. [6] Dunne,. W., MSC.Nstrn V68X A Tool for NVH Response Optimiztion, Proceedings of the 3rd Interntionl Conference on High Performnce Computing in the Automotive Industry, edited y M. Sheh, (pp 77-88), Ferury 997. [7] Deron ittlejohn, Tom Fornri, George Kuo, Bryn Fulmer, Andrew Moordin, Kevin Shipp, Joseph Elliott nd Kwngjin ee, Performnce, Roustness, nd Durility of n Automtic Brke System for Vehicle Adptive Cruise Control, 004 SAE World Congress, [8] M.W. Suh, J.H. Chung, C.S. Seok, nd Y.J. Kim, Hrdwre-in-the-oop Simultion for ABS sed on PC, Interntionl Journl of Vehicle Design, Vol. 4, No., pp ,000. [9] Brke Technology: ABS/TCS Systems, NVH, nd Foundtion Brkes, edited y Brin. Boyle, Dvid G. Eert, Pul S. Gritt, Jim Kuokw, nd Josef Mck, SAE, 000. [0] Blvedi, A. M., Gerges, S. N. Y. nd Tousi, S., Identifiction of Brke Squel Noise vi Sound Intensity nd Acousticl Mesurement, Proc. of INTER-NOISE 00, edited y Ahmet Selmet, Rjendr Singh nd George C Mling, Jr. (Derorn, MI, August 9-, 00). [] Ppinniemi, A., i, J. C. S., Zho, J. nd oder,., Brke Squel: literture review, Applied Acoustics, 63, (00). [] Dunlp, K. B., Riehle, M. A. nd onghouse, R. E., An Overview of Automotive Disc Brke Noise, SAE Pper (999). [3] S. Anwr, An nti-lock rking control system for hyrid electromgnetic/electrohydrulic rke-y-wire system, Proceedings of the 004 Americn Control Conference, 004 [4] Ptil, C.B. ongori, R.G. imroth, J., Control prototyping for n nti-lock rking control system on scled vehicle, 4 nd IEEE Conference on Decision nd Controls Proceedings, 003, pp ISBN: WCECS 007

TECHNICAL INSIGHT INTERNAL CLEARANCE - TYPES AND NORMS

TECHNICAL INSIGHT INTERNAL CLEARANCE - TYPES AND NORMS TECHNICAL INSIGHT PRODUCT AND APPLICATION ENGINEERING INFORMATION A PUBLICATION OF NSK AMERICAS INTERNAL CLEARANCE - TYPES AND NORMS Internl clernce is the distnce which the two rings of non-instlled bering