VEHICLE SEAT SUB-SYSTEM SQUEAK AND RATTLE TESTING AND REQUIREMENTS T. M. Fnuti 3. N. Fletcher Ford Motor Company Simulation Technologies AVT Seats and Interior Trim 7030 Ten Hill Road 15200 N. Commerce Dr. West Bloomtield, MI 48322 Dea~ban, Ml 48120 J. G. Grim Y. Gu Johnson Controls, Inc. Lear seating corporatim Automotive Systems Grp Adv. Tech. and Test Center 49200 Halyard Drive 21557 Telegraph Road Plymouth, MI 48170 SouthGeld, Ml 48034 rattle is usully one of the top five contributors to the overall vehicle squeak and rattle performance The question is, how is the best separation determined? what Xvindow should be selected such that the modes are not coupled? How are component seats tested to verify these seat modes and predict squeak and rattle perfo-ce prior to building prototype parts? To start llllwkg these questions. customs perceived effects such as squeak and rattle must be looked at. 1. INTRODUCTION The objective of seat modal testing is to understand it s dynamics in order to design the seat so that the resonances will not occw at the vehicle sbuctuial mmnance or at the fkqumcy of mad input forces *t the wheels. Tke methodology of modal testing is used to pmvide some insight and guidance for the design of new seats and the modifications of current seats. The m&l characteristics of a seat detaminez the seat vibration levels when excited by external forces. The physical shape, weight distribution, material composition, seat posture, and fastming condition all contribute to the sea s modal charactastics. Both the bequency content and the amplihtde of extemnl forces are critical to the seat vibration response level. Squeak and rattle phenomena are most likely to occur at high vibration levels mdtiq in anooying.wmd output If the seat s fore/aft and lateral resot&%nce s (See PIre Wl) are tuned such that they are not in the wheel hop frequency or flwpan bending range, the seat % ibratims and resulti squeak and rattles can be greatly reduced It is assumed that good modal separation between the seats, sucpeasion, and vehicle body is important in that it effects the seat system Noise, Vibration, and H&mess (NVH)dtigreal world vibration inputs. Moreover. the seat system squeak and Customers evaluate squeak and rattle of seats at relatively high acceleration (g) levels tiera. vehicle modal analysis and fmite element analysis (PEA) are done using very low g level inputs seat suppliers callnot test se& using input levels that we a.9 low as FEA levels and non-linearity problems wise when measuring mcdes at high (customer usage) g levels. However, supplier design analysis is uswdly done using these very low input FE4 levels. This creates a confusing web of design criteria and vaitication testing using three different (wy low, low. and high) input levels. To be able to successfully design B seat system to meet customer needs, an understanding of the &&ttships khueen these different criteria neuls to be deve1oped. This understandi can be thought of as links shown in Figure #2 that creates paths b&veal customer input, vehicle testing, seat ccmpxmt testing, and FEA design and prediction. The three LinlrsthatthispaperwillfocusonarelabcledinPigurr#3as p Jjn and c, I > nteselidcstillhavetok- so that methodologies for testing c mpxmt seats and pmdicuve tools for up Gont engineering can be developed. To da this, quanti@ng the customer s l-eqdmmts into measurable values needs to be developed. These req*ts will be in the form of m&l separations as explained in the next section. 2. DEVELOPING PREPERRED MODAL SEPARATIONS To develop preferred separations, a benchmarking study WBS done to compare customer Things Gone Wmng (TGW) data with 497
measured modal results. Customer squeak and rattle data was collected from six vehicles and experimenti modal analysis was conducted on each of these vehicles focusing on the seats, f?ont suspension, and underbody. The data from this modal analysis and from customer squeak and rattle TGW were collected and show in Figure #3. As can be seen, in general, vehicles with better modal separation between the seats and suspension exhibit better seat Squeak and Rattle pelfomlance. Other vehicles have been found that do not tit this simplistic rule. To obtain a better and more general&d correlation, other factors such as transmissibility, damping, bandwidth, etc. will have to be studied and are beyond the scope of this paper. Once it has been determined that a modal separation of some sort is important, the next step is to d&ermine what the target tidow should be. This has been done by looking at benchmarking data frmn the undabody and suspension systems and deftig a safe zone for the seat system. By lc&ing at good seat/vehicle systems, a specification can be drawn a3 to v&at would provide the best performance for the customer. Thus far, it has bexn show that a modal separation is important and a preferred range is desired for the seat. This defmes link A : The problem is, what test is used to obtain this seat mode? Link A suggests that it is done in vehicle. Howeva, seat s~~~~krs need to test the sat alone on a hard plate or shaker Link 8 needs to be created to develop these requirements. The next section will discuss the testing that is medal to SuppQrt the ss-eation of this link. 3. HIGH Ih I VT USAGE AND LOW MPUT TESTS Johnson Controls, Inc. has hem doing wwk to develop an analysis methodology desired to ensure modal sepmatioa It illustrates the differaces that may occur during the duign and testing process. From design verification (DV), through pmduction validation (PV), a well coordimted effort between the OEM and the supplier is necessqly to ensure that component natund &equencies remain separated thus reducing the possibility of squeak and rattle issues. Table #1 is referenced throughout this case history. The colunm descriptions referred to in this section of the paper are as follows: Iii& Input Test: Hop and Tramp are reported at the highest respnse frequencies gaterated on the spindles during analysis of PSD data under the tiuence of high energy pmving grand input. The high input scat back test t?equencies are dctumined by selecting the highest amplitude frequency generated by a constant amplitude sine sweep (0.75 g Pk-Pk) as nteasured by a laser per O.E.M. specifications. High input test results are presated for a variety of l-awns. First, the suspension modes (hop and tramp) cannot be excited with the low cnagy inputs used during the automobile test. This failure to excite the suspension modes is due to the large stiffness and supporting masses of the springs and shock absorbas. SecondIy, vehicles are always subjected to high energy input conditions under customer use. Therefore, it is necessary to attempt to evaluate the difference between low input testing and high input responses that will be necessary to develop a consistent testing method. J.mv Input Test: Assumes that only the minimal amount of input energy ivas wed to accurately determine the mode shape and mhml trequency. The objective is to excite the struchue without inducing nonlinear responses that are the result of looseness (or chuck) between components. T%is automobile based testing is courtesy of Ford s NVH Group. only 2-3 lbs total RMS energy was used to excite the entire vehicle fmm 2 locations (front and rear) in a M&IO test. The seat back modal testing was paformed using an impact hammer with * very soti tip. DV/PV: Design Validation (DV) is the stage of testing that uses prototype built parts. Production Validation (PV) is the stage of testing that uses production twl built parts. The seat back and recliner udenvent design and tolerance changes in behveen DV and PV testing resulting in the markedly higher nahual t%quencies shown in Table #l. Only the DV tested parts are indicative of and can be compared to the FE model. 3.1 Eigh Input Testing vs. Law Input Testing Automobile: Conelation between high input and low input test results of the automobile were not conducted. In order to determitte the automobile test correlation, it would be necessary to study vehicle response to pmvitq ground input at each of the components (fltmqm, cab, instrument panel, steer@ tieel, etc..) and attempt to isolate which Gequencies twk the shape of the low input mode shapes. This would require an operating deflection analysis which was not in the scope of tbis paper. seat: The low input seat modal testing Cal be done at various input levels. An impact hammer test will give results that reflect closely the vehicle based modal testing. Ax the input level rises (as required by shaker table based testing) the modes dmp. Seat back low input vs. high input testing WBJ conducted in a hard motr~td (tixhtrcd to a solid steel table) condition. A drop of appmximately 15% occurred between the low and high input test results. Johnson Controls has wibxssed this drop in natural fhptettcy on other seat system. The predictabiity of this drop is not yet fully understo& but is critical in establishing ii 8 : Therefore, further study into the effects of performing low and high input test wrrelation should be developed before a seat high input test till be able to be compared to vehicle modal targas. 3.2 MOW Separation A lintha extension of this work is to lay out bow the automobile and seat modes are orgattiza-l Used early in a vehicle program, 498
this can be a valuable twl for establishing targets. Following is a summary of some of the modes: power spectrum density @SD) is then measured Wrtb a lasaometer Frame Natural Frequencies: Suspension Natural Frequencies: Seat Back Natural Frequencies: (During DV Testing): (Ah Design Changes ad&d into PV Testing): 6-gH2 10-14 Hz 13-16Hz 15 _ 21 Hz 4.1 Low Input Test Results vs. FEA For the hvo seats tested, inspection of Table #1 reveals fairly good conelation (less than 10% error) between FEA and low input modal test results in most cases. The two exceptions are 1st Torsion Frame Mode and Cab vertical Bending Mode. Cab I Floorpan I Steering I IF Natural Frequencies: 23-35 Hz Conridaing this single sample, it appears possible to track and organize automobile natural Geqwencies into distinct groups that do not interfere with the allowable ranges of other groups. 4. MODAL ANALYSIS AND LOW JNPUT TESTS Given * method for suppliers to test canponent seats that will correlate with vehicle performance wants, the next step is to be able to predict this perfomutnce up front prior to building prototypes. Conceptually, this is lii c that requires data for a seat system modal analysis (Gem that same system using various input levels), and ftite element analysis @ PA) be collected and studied. Work is b&g done at both Lear Seating Corporation and Johnson Controls, Inc. to measure seat resonance Gequencies both on the shaker table (bard plate in lab with few level inputs). and during modal analysis (wy low level inputs). This data will show that there is general agreement between modal testing and low input lab testing. Tlds will help predict a seat s modal pcrformm~ce during the prototype testing which till veti@ that the seat will meet vehicle pexfommnce specifieatons. Again, referring to Table #l, an additional c&mm needs to be looked at: FEA: Computer models of the automobile and components are generated prior to building and testing parts. Ilwe models are usedtodetaminestragthandmcdalcltamcteristicspriorto prototype builds. FE modes assume linear responses for modal analysis. For seat A : the automobile modes are wwtesy of the Ford LTVC CAE Group and the seat back modes are lxwtesy of Johltsotl Cotltrols. For seat 8 : the testing provided by Lear Seating Corporation wmprises a seat that is bolted onto a bedplate that is driven by twolingsh&ux TheLMSsystemiswedfortbetestaodPCB acc&mmeters are used to measure the output responses. A six axis MTS table is used to excite the seat with a swept sine wave in the vutical, for&g, and lateral direction each in turn. The sinewsvesignalsweeps~2iizto2o~withanintavalof 0.25 Hz and an amplitude of O.lg. The seat back displacement Though this is only a sample of two seats, it shows promise that FE4 and low input modal testing can be well correlated (iii c in Figure #2), and that FEA will allow a prediction of the seat modes. This will allow seat suppliers to assess and modify designs early on in the process prior to prototype builds. 4.2 Hi Input Test Results vs. FEA other studies have shown that seat modes can change drastically with varying input levels. The ability to predict high g level performanceisanareawheremorewrkneedstobedonein order to understand the shifts that -. This work till be critical in that it will provide B link between supplier based prediction, design, and testing and real world cwtomer wage. With this, a seat supplier would be able to model and test a sat compone.nt, lwking for large for/aft and lateral nwvements at the seat back. Actions such as reducing the mass and rotational inertia of the seat back or stiffening the structurecanthenbetakentochangetheseat,sresonancetoa frequency range not easily excited by vehicle inputs. These resonauce s are critical to the resultant seat vibration magnitude witen it is driven by floorpan mounting forces. 5. RESULTS AND CONCLUSIONS TheworkonseatmodalanalysisisbssedonasaiesoflinLs. As highlighted in this paper, smne of Ute links are bxoming well &~~andsomenquirehrrtherstudybeforethey~bemadc. Aliihasbeendmwlxhveenc~st~mer~queakandmttle camems and m&l separation. For vehicle mntufactwers, this is the most important step because it defines subsystem pxfonwmce requirem~ts. However, for seat suppliers. the other Iii will alsoneedtobemade. The 1inkbetwgnhighblputtestingwilicll reflects customa wage low input testing (section 3.1) provides valuable aid in testing and verilj&g seats to meet the customa requirement. The link behveen low input testing and FEA (section 4.1) when taken with the above is important in that it will allow up Gent prediction of real world seat NVH performance. Seat &signers should consider seat dynamics a priority issue. A specigcation qwnti@ing Ezqwncy sepxaticn between the seat for/aft and lateral resommceandtieelhopneedstobedmfted basedontestresultstbatarclmdastaod andagwedtobzhveen OEM and suppliers. Only with the appropriate performance 499
I requirements and test methodologies specified, can a seat/vehicle system be designed and veritied to meet cwtomer requirements. 6. ACKNOWLEDGMENTS me authors wish to thank the following people for their aid in preparing this paper: Dr. Bijm Shahidi and Wei Ho of the Ford LTVC CAE Group Jerry Zimmerlee and Bala Jay-of Johnson c0nm1s. Dmin Burke of Ford s Squeak & Rattle Test and Engineaing section I FUelAft Lateral FIGURE x2 / LINKS REQUIRED TO RELATE CUSTOMER AND TEST DATA! LOW INPUT / VEHICLE / VEHICLETEST ~-, VEHICLE TEST ~! (DRIVE EVAL.) ~ (MODAL) 500
501