Brake Roughness - Disc Brake Torque Variation, Rotor Distortion and Vehicle Response

8/13/2019 Brake Roughness - Disc Brake Torque Variation, Rotor Distortion and Vehicle Response

1/15

93 8 3

Brake Roughness - Disc BrakeTorque Variation, Rotor Distortionand Vehicle ResponseWalter Stringham, Peter Jank, Jerry Pfeifer, and Alex WangAllied Signal Automotiv

Reprinted fromABS/TCS and Brake Technology

International Congress and ExpositionDetroit, MichiganMarch 1-5, 199

SP-953

THIS DOCUMENT IS PROTECTED BY U.S. AND INTERNATIONAL COPYRIGHT

It may not be reproduced, stored in a retrieval system, distributed or transmitted, in whole or in part, i n any form or by any means.

Downloaded from SAE International by Universiti Teknologi Malaysia, Monday, October 29, 2012 09:49:43 PM


2/15

The app earance of theISSNcode at the bottom of this page indicates SAE's consentthat copies of the paper m ay be made for personal or internal use of specific clients.This consent isgiven on thecondition,how ever, thatthecopier paya$5.00 per articlecopy fee through the Copyright Clearance Center, Inc. Operations Center, 27Congress St.,Salem,MA 01970 for copying beyond that permitted by Sections 107or 108 of the U.S. Copyright Law. This consent does not extend to other kinds ofcopying such as copying for general distribution, for advertising or promotionalpurposes, for creating new collective works , or for resale.SAE routinely stocks printed papers for a period of three years following dateof publication. Direct your orders to SAE Customer Sales an d Satisfaction Department.Quantity reprint rates can be obtained from the Customer Sales and Sa tisfactionDepartment.To request permission to reprint a technical paper or perm ission to use copyrightedSAE pub lications in otherworks,contact the SAE Publications Group.

No part of this publication may by reproduced in anyform, in an electronic retrievalsystem or otherwise, without the prior written permission of the publisher.ISSN 0148-7191Copyright 1993Society of Autom ot ive Eng ineers, Inc.Positions and opinions advanced in this paper are those of the author(s) and notnecessarily those of SAE . The a uthor is solely responsible for the content of thepaper A process is available by which discussions w ill be printed with the paper ifit is published in SAE transactions. For permission to publish this paper in full or inpart, contact the SAE P ublications Group.Persons wishing to submit papers to be considered for p resentation or publicationthrough SAE should send the manuscript or a 300 word abstract of a proposedmanuscriptto : Secretary, Engineering Activity Board, SAE.Printed in USA 80 1203B/PG





3/15

930803Brake Roughness - Disc BrakeTorque Variation, Rotor Distortion

and Vehicle ResponseWalter Stringham, Peter Jank, Jerry Pfeifer, and Alex WangAllied Signal Automotive

ABSTRACTNoise and vibration related functionalcharacteristics of the disc brake has, for a longtime, been the nemesis of the design engineer'sexistence. New methods of measurement andanalysis techniques are providing informationwhich improves the practical assessment of a discbrake design and improves the basicunderstanding of noise and vibration operationalaspects. Utilization of these new techniques makeundesirable roughness prediction more feasibleand potential solutions more rapidly identifiable.Development of these new measurementmethods involved the measurement of in-stoptorque variation and rotor thickness variation (TV),as well as rotor total indicated runout. Detailedanalysis of the torque va riation signature of vehicleand dynamometer data indicates significantdifferences. These differences are shown to beinfluenced by vehicle suspension resonantcharacteristics and (in-stop) changes to both liningand rotor mechanical characteristics.

INTRODUCTIONVehicle Brake Roughness is the generic termused to describe undesirable tactile vibration feltduring the braking event. The term BrakeRoughness oversimplifies the complex vibratory

environment which occurs during braking;therefore, the goal of this investigation was to gaina better understanding of the braking forces andtheir role in brake roughness. The goal was metby application of multiple measurement andanalysis techniques of data acquired duringextensive vehicle and inertial dynamometertesting. Two (2) new measurement techniqueswere developed which provided real time,non-intrusive torque measurement and in-stoprotor thickness variation (TV), and rotor totalindicated runout (TIR) data. As a result of thisstudy the following benefits were realized, namely:

That a relationship between objectivebrake torque variation measurements andsubjective roughness evaluations doesexist and that it can be successfully utilizedas a developmenttool. The determination that utilization of thebrake torque variation measurementmethodology can be successfullyemployed as a means to benchmarkvarious caliper designs, compare liningmaterial formulation, as well as provide ameans to assess the vehicle's role inbrake roughness .

That multiple test and analyticalmethodologies must be employed inconcert to successfully gain an improvedunderstanding of the complex vehiclevibratory environment called brakeroughness .

111





4/15

B CKGROUNDThree (3) test measurement techniquesand several analysis techniques will be discussedwhich resulted in a better understanding of thebraking forces and their role in brake roughness.Measurement techniques:

Non-intrusive torque variationmeasurement for both dynamometer andvehicle Non-intrusive rotor thickness variationmeasurement. Vehicle suspension and steering systemfrequency response function measurement.Analysis techniques: A variety of time and frequency basedanalysis techniques were used to developthe interrelationship of torque variation and

Steering/Suspension Systems.NON-INTRUSIVE TORQUE VARIATIONMEASUREMENTS

Since the ultimate goal of this investigationwas to gain a better understanding of brakingforces and their role in brake roughness,measurement of torque variation during vehicletesting was required. Commercially availablewheel end installed torque transducers are ameans to measure a torque variation duringvehicle testing. However, torque wheels influence(to some degree) the operational characteristics ofboth the disc brake and suspension system. Thisinfluence stems from the fact that the torque wheelinstallation alters the torsional inertia, unsprungweight, airflow,scrub rad ius, etc.In an effort to compare dynamometer andvehicle torque variation data and to avoid theintrusiveness of the instrumented torque wheel, itwas necessary that the following two (2) issues beaddressed, namely:1. A single torque variation measurementtechnique had to be implemented. This singlemeasurement technique had to be used for bothdynamometer and vehicle testing.2. The measurement of torque variationhad to be non-intrusive to both caliper and vehiclesuspension functional operation.

To fulfill these two (2) requirements, amethod of strain gaging the disc brake anchorplate was developed. The location of the straingages was on the trailing end load bearingmember and at a point between the shoe reactionpoint and anchor plate to spindle mounting bolt(Figure 1). It is at this location that virtually all thebraking forces generated by the inner/outer padsare transmitted to the suspension knuckle. Thebenefit of this particular location is that it is asclose as possible to where braking forces aregenerated. The general sequence to implementthis torque variation measurement techniqueincluded anchor plate preparation, strain gageinstallation, dynamometer calibration and, finally,vehicle installation.

Anchor plate preparation involved surfacepreparation in the area where the strain gageswere to be applied. The as cast surface of theanchor plate beam had to be machined to providethe necessary flat and uniform surface finish forproper gage adhesion. Material removal was keptat an absolute minimum to avoid changing thebeam's stiffness. Strain gage installation was theconventional full bridge bending beam techniquewith adhesive selection based on an upper surfacetemperature of 93C.

112





5/15

DYNAMOMETER CALIBRATION OF THEINSTRUMENTED ANCHOR PLATE - Thedynamometer calibration procedure consisted ofperforming a series of ten (10) stops from thesame speed at a prescribed fluid pressure andinitial brake pad temperature. The prescribed fluidpressure and brake pad temperatures had four (4)test points (Figure 2) and were selected becausethey are similar to the four (4) test conditionstypically used during a subjective brake roughnessevaluation:

During each stop, fluid pressure, anchor platebridge output, lining temperature,once-per-revolution tach pulse and dynamometertorque were simultaneously recorded utilizing atransient data recording system. For allsubsequent data processing tasks, the acquireddata was downloaded into the analysis software.The following data processing sequence waseasily accomplished via the software's worksheetand macro capabilities.1. Perform seventh order polynomial fit onboth dynamometer and instrumented anchor platebridge output data. Figures 3 and 4 reflect theseventh order fit and time history relationships.Since data acquisition was triggered on a pressurethreshold of 68.9 kPa (10 lb f/in2), the time historydata as shown does not originate from zero. Thisprocedure was employed to provide a repeatablestarting point for data acquisition from stop to stop.2. Relating both dynamometer torque and anchorplate bridge output via x-y plot.3. Determining the torque-to-bridge outputconversion factor via the slope of the linearregression performed on Item 2 above.

All stops were analyzed in this fashion andthe average of all conversion factors thus becamethe anchor plate bridge output-to-torque calibrationfactor. ANALYSIS OF DYNAMOMETER TORQUE- In addition to providing the strain-to-torquecalibration factor, analysis of the dynamometertorque time history indicates that the in-stopmagnitude of torque variation increases in a

uniform manner (Figure 5) and that there are nodistinct peaks present. Generating Figure 5involved the extraction of the torque variationmagnitude per rotor revolution. Therefore, figure 5

113


It may not be reproduced, stored in a retrieval system, distributed or transmitted, in whole or in part, in any form or by any means.



6/15

is actually compiled revolution-by-revolution fromstart to finish with the peak-to-peak torquevariation value plotted versus revolution. Thelower trace in Figure 5 was typical of the stopsperformed at test point 1 (low pressure, lowtemperature), while the upper trace was typical ofstops performed at test point 2 (low pressure, hightemperature). The level of torque variation for thehigher temperature condition was 1.9 that of thelow temperature condition.

Further analysis of in-stop torque variationis illustrated in Figure 6. Figure 6 is athree-dimensional plot with the horizontal axisbeing frequency (Hz), vertical axis being torquevariation magnitude (N-m) and the axis into thepaper being time (seconds). Time starts at thebottom (first) trace (coincident with zero on thevertical axis) and increases upward toward the top(last) trace. The plot is made up of many tracesand each trace is the frequency spectrum of asmall time slice 1.024 seconds long. Therefore,the first trace is the frequency spectrum of the timeslice from 0 to 1.024 seconds, but the second (andall subsequent slices thereafter) are overlapped by97%,which means the second time slice starts attime 0.024 seconds and ends 1.048 seconds later.The benefit of overlapping the time slices is thatchanges in the frequency spectrum with respect totime are more easily distinguished. As can be

seen in Figure 6, there are numerous mountainranges and that these mountain ranges slopegradually to the left (i.e.,decreases in frequency).This decrease in frequency is directly proportionalto the decrease in speed which occurred duringthe braking event. This proportional relationshipbetween rotational speed and frequency iscommonly referred to as rotational order(s).Frequencies which do not change proportionatelyto speed are classified as stationary frequenciesand are indicative of a resonant characteristicbeing present. Therefore, the significance ofFigure 6 is in the information contained therein,namely: The first mountain range from the left isdirectly related to rotor rotation and is firstorder, the next mountain range to the rightis second order and so on. The magnitudeof first order clearly dominates over all

other orders. There are no stationary frequenciespresent in this frequency range, thusindicating the absence of any resonances.These two observations remainedconsistent throughout the dynamometer testingthat was performed at each of the four (4) testpoints which were listed in Figure 2.

114





7/15

The major sequential steps in this vehicle testprogram are:

115





8/15

VEHICLE TESTING -This investigation involvedmany tasks of which the major steps have beenitemized in Figure 7. Vehicle related testingincluded laboratory-based experimental modalanalysis and over-the-road brake applies.Experimental modal analysis was utilized todetermine suspension and steering columnfrequency response functions (FRF), along withidentification of dominant resonant frequenciesbelow 50 Hz. This upper frequency limit wasselected because it is consistent with thefrequency range generally associated with thestudy of brake roughness.Over-the-road testing involved performing 112-48km/h (70-30 mph) decelerations at each ofthefourtest points indicated in Figure 2. The rotor,inner/outer lining and caliper assembly utilizedduring dynamometer testing was installed on thesubject test vehicle at the right front location.Measured parameters included brake pressure,anchor plate bridge output, outer liningtemperature and a once-per-revolution tach pulse.Data acquisition instrumentation and dataprocessing software was also the same as thatused during dynamometer calibration of theinstrumented anchor plate. To extract the desiredinformation from the data, it was necessary toutilize four (4) different analysis techniques. Alisting of the types of analysis and what eachprovides is shown in Figure 8. The following two(2) examples provide the specific analysissequence which ultimately produced the torquevariation comparison (Figure 9) of two (2) different

lining materials and the torque variation vsroughness rating plot (Figure 10), respectively.

116





9/15

The analysis sequence which is shown above togenerate Figure 9 was repeated for ail ten (10)stops at each of the four (4) test points for two (2)lining materials and compiled. Two (2)conclusions are drawn from this information;namely: That both materials exhibit an increase intorque variation at elevated temperatures. That material A exhibits a significantlyhigher torque variation (at all test points)than lining B.

Next, the maximum first order torquevariation values are plotted versus their respectivesubjective roughness evaluation ratings toproduce Figure 10. The conclusion drawn fromthis analysis is that the first order torque variationmagnitude versus roughness rating has an inverselinear relationship.Figure 11 shows an overplot of vehicle anddynamometer first order torque variation at similartest point conditions.

Observations derived from this information are asfollows: the vehicle torque variation data has apeak at 86.9 km/hr (55 mph) and 12.5 Hzwhile the dynamometer data does not. the vehicle torque variation peak coincideswith the first major suspension resonanceof 12.5 Hz and 13 Hz, respectively. the magnitude of vehicle torque variation atthe peak is observed to be 3.5 times higherthan the dynamometer value.

NON-INTRUSIVE IN-STOP ROTOR THICKNESSVARIATION MEASUREMENTRotors and pads were removed fromvehicles that had exhibited varying degrees ofdriver perceived roughness and mounted on an

inertial dynamometer for further study. Althoughroom temperature static characteristics of rotorscan easily be measured, it is their dynamicproperties at braking temperatures that arerelevant for properly understanding roughness.Even though some uncertainty remains as towhether actual rotor/pad alignment from thevehicle can be reestablished in the dynamometersetup, some properties can still be measuredminus the interaction of the caliper/rotor/pad withthe vehicle suspension system. Although manyfactors can contribute to creating roughness in thevehicle, the primary concern here was theinfluence of rotor thickness variation onroughness.

An inertial dynamometer test procedurehas been established whereby in-stop variations intorque and pressure can be compared against

117





10/15

rotor properties such as runout, thickness variationand thermally induced deformations (coning,surface rippling). The in-stop rotor deformationsand thickness variation have been measuredusing a pair of non-contacting capacitancesensors. These fast responding sensors, one oneach side of the rotor, are mounted approximately0.5mm away from the rotor surface and exhibit alinear output over a +/- 0.2mm range. An x-ymachinist table, along with a micrometer stage,allows for translational adjustment and alignmentof the capacitance sensors with different rotorconfigurations. The outputs from the capacitancesensors, along with a variety of other in-stopvariables such as listed below, are collected w ith adata acquisition and analyzer system:

- Runout, inner rotor surface- Runout, outer rotor surface- Rotor Thickness Variation- Pressure- Torque- Rotor Orientation- Accelerometer outputs, mounted on brakecomponents- Microphone, noise studies- Tem perature, pad and or rotorAs an example, the results ofdynamometer studies on a pad/rotor coupleexhibiting roughness on a vehicle is presentedbelow. The stops were made from 1005 rpm

(equivalent to 113 km/hr or 70 mph) with constantpressure of 14 Kg(f)/cm2(200 psi), and at severaldifferent initial brake pad temperatures (IBT)The IBT for the first stop was 177C.Figure 12 shows the pressure vs. time trace forthis stop.

The constant low frequency modulation observedhere is due to the dynamometer pressure controlsystem and is not from the brake; this is neglectedin the subsequent analysis. A minor variation inthe pressure fluctuation seen in the expandedmarker section, however, is due to the rotor.Figure 13 shows the torque signal for this stop; asevere fluctuation is noted that occurs once perrotor revolution (confirmed with rotor orientationsync signal). The runout of the rotor inner andouter surfaces recorded during the stop with thecapacitance sensors are shown in Figures 14 and15. As thermal energy is inputted into the rotorduring the stop, the rotor begins to cone or deflectoutward. The degree of coning is supported bythe theoretical thermal modeling done on the rotorgeometry.

The difference in magnitude between themeasured coning for the inner and outer surfacesis caused by the thermal expansion of the rotor inthickness during the stop.By taking the sum of the outputs from thetwo capacitance sensors, the relative variation inthe rotor thickness per revolution can be displayedas shown in Figure 16.The difference between the maximum andminimum readings per revolution gives the rotorthickness variation (RTV). A slight increase inRTV was observed to occur during this stop, asmarked in Figure 16.

118





11/15 9





12/15

Comparisons can now be made betweenthis 177C stop and the following stop startingfrom near room temperature. Figure 17 shows thetorque trace. The peak-to-peak torque variation is13 N-m (116 in-lb) as compared to 52 N-m(460 in-lb) which was observed on the highertemperature stop. In terms of percentage oftorque variation (delta torque/torque), thepercentage increased from 4 .1 % to 8.6% byincreasing the IBT to 177 c. Note that the stopstarting near 25 C took about 20.5 seconds to stopcompared to 11.4 seconds for the hotter stop.The coefficient of friction for the cooler stop wasmuch lower, hence the lower torque value andlonger stopping time. The runout shown in Figure18 is similar to the runout on the highertemperature stop. However, the rotor coning isabout half the value as observed on the hotterstop. This is consistent with having a smallerthermal gradient throughout the rotor. The moresignificant observation is that the RTV, seen inFigure 19, is smaller, 6.1 to 7.9 um (.24 to .31mils) than the 13.7 to 18.3 um (.54 to .72 mils)measured on the 177C IBT stop. The rotorthickness variation is increasing with temperature.This accounts for some of the increase in percenttorque variation.

Since runout and thickness variation areboth usually first order related and often haveabout the same phase, it is difficult to sayprecisely which is contributing to roughness. Oneevaluation method has been to do a frequencyresponse analysis on the torque, runout andthickness variation signals and compare theirsignatures. Figures 20, 21 and 22 show thefrequency response analysis for the 177 C stopdata. From the amplitude relationships of thevarious orders, it is apparent that the rotorthickness variation has a good match with thetorque variation.

12





13/15

From this series of tests performed withpad/rotors taken from vehicles exhibitingroughness, a linear relationship has beenobserved between in-stop rotor thickness variationmeasurements and percent increase in torquevariation. This relationship from experimentaldynamometer testing is shown in Figure 23.

TORQUE VARIATION, SUSPENSION,STEERING SYSTEM INTERRELATIONSHIP - Anobservation from vehicle testing is that brakeroughness (i.e., the vehicle response to braketorque variation) is maximum at certain vehiclespeeds. Maximum vehicle response occurs as thefrequency(s) of torque variation input (i.e., firstorder, second order, etc.), which is proportional tovehicle speed, becomes coincident with vehiclesteering and/or suspension resonant frequencies.At these resonant frequencies, the brake torquevariation is amplified .In an effort to identify the vehicle systemresonances, experimental modal analysis wasutilized. The vehicle was at curb weight with thebrakes locked. A shaker was positioned toprovide input into the right front suspension lowerball joint. The shaker axis was parallel to theground and perpendicular to the spindle axis. Inthis fashion, the shaker forces produced a torqueabout the spindle axis. This approach was used inan effort to simulate torque variation force inputswhich would have the propensity to excitesuspension torsional resonances (only). Thissurvey, therefore, was not intended to be a fullvehicle 3-axis modal survey, but rather to identifythose torsional resonances and their respectivefrequencies which could be excited by braketorque variation AND fall below 50 Hz (i.e., theroughness frequency range). Then , once havingidentified these resonant frequencies, determinetheir relationship to vehicle speed and rotationalfrequency.Figures 24 and 25 are the graphicalrepresentation of this relationsh ip. Both figuresare composed of two (2) graphs. The graph onthe right is the frequency response functionacquired during experimental modal analysis,while the graph on the left displays therelationships of frequency, tire rotational order andvehicle speed. The frequency vs. vehicle speedgraph was the result of calculations made usingthe tire static loaded radius (SLR) of the testvehicle at curb weight with OEM tires at nominaltire pressure. The usefulness o f both Figures 24and 25 lies in the fact that they show theapproximate speed at which to expect themeasured suspension and steering columnresonances to occur as well as which rotational

121





14/15


15/15

frequency is necessary to excite them. It shouldbe pointed out that while Figure 25 identifies threeresonant frequencies of 13.5, 21 and 24.5 Hz,subsequent frequency response measurementsmade on the steering column have identified thatonly the 21 Hz frequency is local to the steeringcolumn and that the remaining two (i.e.; 13 and24.5) are the result of the transmissibility betweenthe suspension, body structure and steeringcolumn(i.e.,they appear to be strongly influencedby the suspension resonances identified at 13 and23 Hz, respectively).Another significant attributes of Figures 24and 25 is contained within the information thetransfer function provides. In this case, thetransfer function trace is the ratio of outputresponse (measured in g's) to the input force(measured in kgf) over the frequency range of10-110 Hz. This ratio provides a relative measureof amplification, thus, the higher the magnitude ofthe ratio, the higher the amplification. For thisvehicle, tested in this manner, the suspensionresonance at 13 Hz has the highest transferfunction ratio within the roughness frequencyrange of 1-50 Hz. Likewise for the 21 Hz steeringcolumn resonance. The suspension transferfunction magnitude at 13 Hz was more than two(2) times higher than the steering column transferfunction magnitude at2 Hz.

SUMMARYThe findings presented throughout thispaper are based on extensive experimentaltesting. To arrive at these findings, three (3)separate test methodologies were employed,namely: Vehicle Testing dynamometer Testing Experimental Modal Analysis Testing

It was shown that each of the three (3) testmethodologies provided relevant pieces ofinformation and that when these pieces werecombined, yet another, larger picture (of brakeroughness) began to develop. Therefore, theconclusions drawn from the testing performedduring this investigation are provided in two (2)sections. The first section is a listing of

conclusions derived from each of the testmethodologies listed above. The later sectionaddresses the more global aspects of thisinvestigation.CONCLUSIONS FROM TEST METHODOLOGIES

The disc brake torque variation is anon-stationary function and its frequencycontent is a function of rotational speed. Torque variation signature is significantlyinfluenced by both rotor and lining physicalcharacteristics and that these physicalcharacteristics are further influenced by(increasing) tem perature. Different lining materials can producesignificantly different levels of torquevariation Rotor thickness variation increases linearlywith increasing temperature. Lining coefficient of friction increases withincreasing temperature (for the liningstested as part of this investigation). The disc brake torque variation signature isdominated by first order (once perrevolution) component and that rotorthickness variation (TV) is the primarycontributor. Vehicle brake roughness is a resonantresponse characteristic which is influencedby suspension and steering columnresonances and the degree of roughnessis associated with the resonanceamplification factor.

CONCLUSIONS OF A GLOBAL NATURE Multiple test m ethodologies must beemployed in the characterization and studyof vehicle brake roughness. The strain gaged anchor plate techniquecan be successfully utilized as anon-intrusive means to measure torquevariation and provides a link betweendynamometer and vehicle measurements.




Date post:	04-Jun-2018
Category:	Documents
Upload:	debisi14140
View:	218 times
Download:	0 times

Brake Roughness - Disc Brake Torque Variation, Rotor Distortion and Vehicle Response

Documents