. . - ‘AN EXPERIMENTAL INVESTIGATION OF HIGH cotAcT RATIO GEAR TOOTH! BEHAVIOUR . = 4 . , ; - . Leslie Lajos Kenedi ‘ . : 7 : . : : A Thesis a e ; : 5 ge tn : oo ‘The Facult Q ee va : 0 7 of : +, Engineering = * . oe : ¢ . : - - ; os Presented in Parti Folfilinent of the Requirenents for the degree of Master of Engineering at “ Concordia University ; ~ baal * Montréal, Québec, Canada August 1962 "© ,twotte tajos Kenedt, 1982Ail EXPERDEENTAL LAVESTIGATION OF AIGH CONTACT RATIO GEAR TOOTH BEHAVIOUR —— este tijon Kenai, : . The objective of thig study is to examine experimentally the static behaviour of a pair of High.Contact Ratio Gears (HICRG). There’ ts recone tgrereat in outh suring vhich ateas fron tte poootbility of significant weight saving. Photoel. ticity was used extensively in the past for ‘ Anvestigating mbstly a single tooth. for HORG 2 gultiple tooth aodel ‘te necessary, as the interaction of the simultaneously engaged tooth pairs deteraines how\the Load ia distributed. Building guch photoelastic nodel 1s a novelty as extresely high precision is required. The fillet stres 49 ate observed in single and aultiple tooth pair engagenents, For a few selected fillet-Locations the stresses are ploteed ya'londing postion anagured on cha Man of accion. Thus che static stress excurbions of these points are obtained, allowing determination of the most danaging-stress range: 7 ‘reflection mesourénents are taken and nev conckpte are introduced for setting their reference datum. ‘The variable compliance ,of tooth paira af « fynction of load position 1s determined and the sharing of the Load {s explored in depth using several nev techniques. fs the structure ie indeterminate, the concept of “Stress Adwtttance” is ‘proposed. This is the stress due to the combined unte deflection of : s the critical Lgcation of the load in'wultiple tooah pair engagesent. pair of contacting teeth. It may be used for finding{ . DEDICATION =~ . -_ + i 0G o \ en . . eo. T dedicate this work to ay wife . fase, (tose enduring patience and Loving s . 7 €ncouragement made it all possible. ' : j > : . ° = Ts . . 7Corrieli-Heatervelt Dynaate Progr: Iam indepted to my thesis supervisor Professor S.V.Woa for ffort. “gulding and encouraging ov throughout thie ~ " Ewoutd ithe ge cereco ey apreciation to 1, bl. for inroducin. sd cotehing oe tn the application of Finttn Eleneats to che whject matter, and to A, Elkhcly mostly for his assistance at the input for the ‘T.gm grateful to all those colleagues at Pratt and, Wittney Aircraft of Canada Limited who have contributed.to this effort fa the’ form of advice, suggestion and inthe tikesone task of typing those equations and figures.: x, ‘ 7 TABLE OF CONTENTS 7 . ~ ° : ' . Page List of Synbols) : - 1 . List of Figures . 9 ist of Tabi 2 : CuPTER oF : 4 ; teropucrio® ; oe CHAPTER TWO as : LITERARY SURVEY. . 2.1 + stress Sensttivicy,- shore Bean Hode1 approach = "16 The Lewis Equation 5 a. edad “plan & Broghamer - Stréss Concentration Factor B, 7 213 The Agma Standard : 2b BB, Heywood - Gear footh .as‘a Projection “2s, M.A. Jacobson - Proximity Effects «| . 26 BW, Kellay and R. Pedersen - es - The Contenjorary State of the Art and a New Formula 2.17" Liy ALLison and £.J. Hearn - Effects of Friction ~ —27- 21:8 «RW, Cornell's ~ Expression for Fillet stresses tay 22 Stress Sensitiyity - Finite Element Method . 29 . 7 2a.’ | “ty Wilcox & He Colesan's Polynoatai Formula wo 2.22 G, Chabert, T. Dang Tran, Ri Mathis'e Expre: 3 for Tensile and Conpreesive Fillet Stresses . 23 Stress Sengitivity - sesso Wedge Theory ‘5 231 ade shbeterte 38 . Wedge Stresses and the Most Denaging Stress Rangeeee be : : . : 7 a 4 TABLE OF CONTENTS (Gont'd)~ . a 2 my Page . Reb Stress Sensitivity - Theory of Elasticity Solution, a” 1p ‘*. Using Complex Potentials and Contorpel . : . Transformation . Zobel y Te Alga and o-Teraueht - ” oat 2.442 GN, Baronet, G:V. Tordion's “Exact Stress a i a Distribution” ' rr——— _ , : Weber!s Seui-Elasuscity Solution : “3 . W.J. O'Donnell: * - “46 4, Mttessee and Deflections in Bullt in Beans - G, Chabert, . Dang Tran, Re Hathis - a oe Using Finite Elesents ! = ssa Freaithat, G.V. Tordion, Cf. Baronet - F te : using Coaplex Poteatials 7 | 295.5 Points of Simultaneous Contact 5.2 Sealing e. 5.3 loading ° = 5.3.1 Scheme of Loading Points 5.4, Stress Measurements 5.4.1 "The Photoelastic Method 2 oes Calibration: 5.4.3 Determjnation of Stresses 5.5, Pair—Deflection Measurements 4 ey “ Page 64 6 6 67 67 70 2 76 80 a5 as as 88 1 93 95 95 96 7 98 pte nnntnisnesntnenntn. oa- a 5 s : [oe ‘TABLE OF CONTENTS (Cost 'd) : - - ‘CHAPTER SIX a : AUTERNATIVE ANALYTICAL METHODS +6e1 ‘Application of the Kelly-Pedersen Formula 6.2 Application of the Alda~Tereuchi Foraula 6.3 Using the Comneli-Westervelt Dynamic Program 6.4 . An Application of Finite Elesente Element Network bebe? The Progr 6.4.3 Boundary ‘Conditions bebe Loading 6-445 \ Determination of Deflections — Using Novel Reference Data y 646 Determination of Fillet Stresses : CHAPTER SEVEN RESULTS ‘AND CORRELATION WITR ALTERNATIVE METHODS at Stresses - Single Tooth Pair Engagenent. 1.2 Stresses ~ Multiple Tooth Pair Engagenent 7.3 Deflections Page 100, 100 101 192 103- 103 110 _ Mo 4 8 120 im 121 134 MaTABLE OF CONTENTS (Cont'd) * a hia : . CHAPTER, EIGHT : : ANALYSTS OF RESULTS “ate ConpLiance,~ Experinental 8.2 ‘Conpltances from Finite Hlenents, Correlation of : Results ore XG ° 8.3 Sharing of the Load at Multiple Engagement 8.3.1 Determination of the Loads Using Gauge Point © Stresses, ss 8.362 Graphical Determination of the Loads Using Deflections \, saa teat tte etngHsty 8.4 Strea Sensitivity and Stree Adatttance «85 Equivalent Static Alternating Stresses . unpre ye AN ESTIMATE OF "THE. ERRORS 9.1 ‘Blas Due to Parasitic Loading 9.2 Random Errors ‘ cuarren Ten [ ConcLustons AND CONTRIBUTIONS tos1 "conclusions 10.2, ‘Gontrtbuttons to the Art List of References ¢ |S sAppendix 1 - Work Equivalent Loads Page 148 148 150 152. 155 158 158 162° 165 17% 1% 176 180 180 181 183 aL oa“variable (Wilcox § Coleman) oa. : u ‘ 7 ~ yg a ~ - can a sonent arn a2 : crosa, sectional area half width of contact constant (Wilcox § Coleman) : length of aide of equivalent atraight flanked beam face width of bean, thickness of model : distance of extrene fiber fron neutral axis constant (Wilcox & Coleman) compliance center dis equivalent compliance pair ~ complianct pair - compliance at pitch point dedendua constant (#ileex & Coleman) i vartable (Wilcox & Coleman) ‘ . : 1 the assemblance of symbols quoted from a variety of sources are often conflicting, To make the text gore readable the iotation of some of the quoted formulas vas changed, though an effort vas nade to retain as mich as possible of the original nomenclature.a a's LIST OF s¥wBoLS' (Cont'd) . half Length of the veakest section | Young's andulus Young's modulus of sugport ° material fringe constant [kPa - a], (pst-im) No distance from contact to tooth centerline along profile fa z size of gap shear modulus the normal to the total depth of tooth distance . the 4th number of a séquence . : Second area ‘noment geometry factor ~ dimensionless flexibility coefficient (Chabert) stress concentration factor +c stress correction factor (AGMA standard) constant, representing the effect of inaccuractes, on the stresses inthe model. “residual factor” (Wilcox § Cétenan) O'Donnell correction factor constant height from weakest section to lodd 7 . x height from weakest: aection to the load line at its intersection with che cooth’ edtteriine 4 and/or 4" with reference for maximum stress sensitivity=e Fe + hetght from veakests section to tip of the- tooth oF sympots' (Cont'd) : 41 with reference for maximum deflection height from fillet-flank blending point to load line and tooth centerline intersection Length module (mm) oad sharing ratio tip modification ratio \ . bending moment snteger nusber of teeth : aa fringe order “ nuaber of teeth in drivér (pinion) nuaber of teeth tn défvén gear (vheel) . contact pressute etreular pitch base pitch dianetral pitch : side length of wedge - compressive (Shotter) fillet radius ; oo . side length of vedge - tensile’ (Shotter) ii of curvature’ radius - polar coordinate |tr th. List of symmous! (cont'd) radius of the base circle” * radius of iighest Point fér Single Tooth Contact” pitch radius root radiue standard deviation vtandard deviation ta magertal chickae standard ‘deviation in material-fringe value standard deviation due to model inaccuracy taridard deviation in fringe’ reading etanderd devin in ‘stress aenateiviey Yjedtars cevtatton th toed distance along the Line of” action from pitch point, tve toward disengagement . normalized position along the Line of action 8/25 Outer most point on the line of action, where both of the contacting surfaces are unnodified true tovolutes ~ dteengagenent side : Stutlar to Sq - engagenent. side Reference length on the line of action, efther Soq oF Soe length of the disengagenent side of the Line of action Length of the engagement side of the line of action depth of the bean of the “weakest section” depth of the beam at the foundation “boundary” for maximum © deflection : depth pf the beam at the application of the load . flank blending point depth of the beam at thé fillet N\yp "EB yr LIST OF swmpois! (Cont'd) depth of the beam for maximm fillet stress depth of the beam at fillet angle y are tooth thickne complex coordtiiate in ¢ plane, real axis “unbalanced residual factor’ (Wilcox & Coleman) coxplex coordinate in ¢ plane, Laaginary axie complex variable on v axis atrei sensitavity [ma], (tn?) * normal Load deadweight or rig load ws normal load ~ model \ normal load ~ prototype radial component of the normal load shared” normal load “angential component of the normal load coordinate axis” atatence ~ Late Layout coordtoate exte eptecqeee tong Gta Lise of actions detection = bendiog congener deflection due to local compression deflection ~ contributed by the deformation of the fillet and the foundation region scomponent of y"p due to defornation in the fillet area component, of y'y due to deformation in the foundation area. : 6 ( LIST oF spots? (Cont'd) . y, «Lewis form factor Y modified form factor, ty i 2 compfx variable in the plane of rectangular coordinates Zz stress admittance [N/mm] (1b/in®) eo 4 - a polar angle with reference to the centerline of the tooth : cy polar angle to starting pint of the tnvolute ox base clrele | ¢ polar angle to the center of the fillet radius . | dp polar angle of the pitch point | ap q% polar angle for points D and M respectively 81% a5 wedge angles (SHottor) a tncLination of load to the normal at the point of saxiaum . utscteesses ees icie eieeeeareni fee eee mea te eee ce errr : Ce ee i centers of fillet circles on each aide of the tooth, ‘tve counterclockwise yp fillet angle for maximtzed deflection ~ Cornell method vp fillet angle at maximum tensile strei method Kelley & Peder YL — fillet angle at the tangency of the Lewis parabola Yo filet angle at fiLLet-flask Blending potst Ye ‘fillet angle-for maximized stress ~ Cornell method é width of a finite strip : A&,Ge email angular differences © roll angle of a point on the involute €ase¢ Toll angle for points A and C respectivelyList oF sywpous! (Cont'd) roll angle for pitch point fav (4) [4nvolute of 4] complex vartable in the plane of curvilinear coordinates 4 ‘Nectorial angle" of a potat on the iavolute (ep) also called ' fc . vectorial angle for point ¢ ' 8a rotation of the root of a built in beam (O'Donnell) : a angle of the normal to the profile at a point with reference to the normal to the tooth centerline , Xa Ag AprAy angle of normals at points A, C, D, M 1 respectively . , ¥ | Potsson's ratfo : “ coefftctent of frigtton : . 1) 12 73. dimensionless ratios” ° equivalent radius of curvature f ° stress - ay alternating component of stress ge equivalent alternating stress. by flee str Oe «mas compressive fillet stress - og ° “gauge stress “ (Chapter 8) a mean stress . a On normal stress ‘ Gp filet stress due to the radial component of : Orb the bending cotponent of o , ~ ora the dlzest compooet of oy je due to the tangential component. of the loadoon or Suc * ‘shared otre mix. tensile fillet stra: ultimate stres “List oF siMpous' (cont'd) . obtained in mltiple engagemént’ test in untaxial compression ultimate stress in uniaxial tension shear stress pressure pressure pressure pressure pressure pressure angle angle angle angle angle angle, = arbteegey reference at potats A and'¢ reapectivety at pitch point (normal préssure angle) at pitch potat “loaded side” at pitch point “unloaded side” angular displactnent transformation function2.12 : List of Figur ° : 21 Lewis Layout with Tip Loading [4] 2.2 Dolan & Broghaner's Set Up [11] . > 2.3 Radius of ‘the Wighest Point for Single Tooth CSntact (R'q) (or the driver gear) 2.4 — Heywood's Construction [16] 2.5 Kelley & Pedersen's Construction [18] 6 Stress Sensitivity - RW. Cornell [8] 2.7 Contour of Flaite Elenent Model, | Asyeneeric Tooth ~ Wiidox Coleman [31] 2.8° Tooth Layout ~ Wilcox & Coleman [31] 2.9 Layout of Fintte Element Model - Chabert et al. [7] 2.9 Layout, Wedge Principle, Shotter [27] 2.11, Mapping of the Tooth Form into a Straight Edged Half “Plane = Baronet & Tordion [5], Aida & Teraucht [1] Aida & Terauchi Layout [2] "2113 Pouphtance, Layout ~ Weber [29] 2,14 - Contact Deformation and Pressure Distribution — Weber [29] 7 2.15 Lond Sharing - Chabert et al. [7] 2616 Bea coapltance ~ Cornell [8] 2m vpeLLeb and Foundation Compliance ~ Cofnell [8], 2418 Local dompliance - Cornell [8] 3.1 Static Model for High Contact Ratio Tooth Engagement 3.2 Fatigue Diagram ’ 4A Tooth Form\- Pinton 4.2 toott’ orm } sheet 7 2 23 23 28 31 3 23H 36 38 40 4a rs 31 51 5h 59 59 66 666.3 6.4 65 6.6 6.7 6.9 6.10 Ta 12 Lst of Figures (Cont") Involute Tooth Profile Fillet Layout - Tip Modifiéation Profile Adjustment Set Up, Mounting of the Model Profile Adjustment Set Up, Guide for Abrasive Tool Experinental Set Up - Front#l View Experimental Set Up - Side View Polariscope and Straining Rig (Photograph) Matching Radit Points of Synultaneous Contact Loading Layout Load Sharing ~ Tentative Dynamic Program Finite Element Yodel ~ Wheel Finite Element Grid, Close Up - wheel Finite Element Grid, Close Up - Typical Loading Site Finite Elonent Yodel - rinton \* Finkte Element Grid, Close Up ~ Pinto Defected Shape ~ Pintan Tsostress Lines, tajor Principal étresses - Pinion Isostress Lines, Minor Principal Stresses Pinion Layout of Elliptic Load Distribution Fillet Stresses vs Angular Positio Wheel, $ Single =1,00 Fillet Stresses vs Angular Position, Single Wheel, s'= 0 Engagement ~ Engagement ~ 15 i 78 19° 81 82° 84 86 . \ 87 * eee | ' { 106 107. 108 109” m 12 .4| 43 ug | we | |3.3. 14 15 1.6, wT 7.8 1.9 7613 Tels . Ls of Figures (Cont'd) \ Fillet Stresses vs Angular Position, Single Engagement ~ Wheel, S'='1.33 Fillet Stresses vs Angular Position, Single Engagement - Pinion, 8'= -1.00 ’ Fillet Stre vs Angular Position, Single Engagement - a + Pinion, $= 0 Fillet Stresses vs Angular Position, Single Engagéaent ~ . Pinion, S'= 1.33 Peak Tensile Stresses and thetr Angular Location vs 5? (Normalized Position Along the Line of Action), Single Engagenent ~ ue ‘ B (xy /p) (2/3)P : 2x : Letting the form factor yy =—— Eq. (2-1) 18 obtained. 3p 2 Dolan & Broghasér - Stress Concentration Factor (12 It was soon discovered that the Lewis formla gave grosely inaccurate results. Following a few early investigators such as S. Timoshenko [39], R.V. Baud [33], R.E. Peterson [34], P.H. Black [35], Dolan and Broghamer made the first extensive study to isolate those parameters which influence the fillet stresses mst. Eleven models were ande of bakelite, incorporasing a systenatic variation of load position, number of teeth, pressure angle, addendum ‘proportions, fillet size, and fillet shape. Each model consisted of a middle tooth and tw halves of the neighboring teeth, as in Fig. (2.2). They were held’ in a rigid frame (1). Deadvetght loading was transmitted by a half wedge (2), through, a, thin cellulose strip (3) protecting the model. * After careful measurenents and analysis the parameters were organized {nto the shape. of the vell known stress concentration factor formulas forthe tensile ffllet, For nom{ial pressure angles of: bp RET Km 22 + (e/g) 062 (0/2 )004 (2.3) op 720° Ks 18 F coreg 15 (e/2)0-45, (2.4) __ The factors*above include the direct effect of the radial load and increase significantly the nominal stresses obtained by the Levis formula, The use of these expressions is discouraged for angles other than indicated. ‘; ‘| | | ‘ FIG. 2.2; DOLAN & BROGHAMER'S SET-UP [17)These. stress concentration factors constitute an iaportant feature of the "Yodified Levis formula”, which along vith ‘other considerations foras the basis of the AGMA standard described below. 241.3 The AGMA Standard [4] . This standard as-received wide acceptande. It stipulates a series of service factors providing a guide for design. Modifications were introduced to the Levis formula, which tn its revised fore constitutes the core of the standard. Me Pe oe = 2 (25) Bd where Pg the dianetral pitch has replaced the ctreular pitch, and Ja “geonetry factor” was introduced. Y Je - (2.6) Kem where Ke. is @ stress correction factor, essentially Dolan % Broghamer's stress concentration factor. tq 18 the loed sharing ratio. For conservative estimates tip Loading is assumed and it is taken as unity. The .new form factor Y is defined by the expression: 1 Ye cH | Neos 2 (2 2) , eosty t . where x; & ¢ is read from the layout as in Fig. (2+1) The second term in the denominator accolints for the direct effect of the radial coaponent of the load, as it was proposed:by Merrit [38] (1938). The monent gffect of the same 1s ignored.x) sovaNoo utoos atoms wos snroa SSMHDIH SHE a0 S0TeH . ‘ z NN (avao WaaTwa-aHL woa) i co % uys. ys ee ora Suse ts tj =e chee 0%] s «444 006 = oan >: 2 j Vien the designer is confident thet manufacturing can kebp i profile errors at very low values, he Vay assume that load sharing does exist at each and every mesh. Then the critical load may be placed to the "Highest Point for Single Tooth Contact” (HPSTC). -The term tacitly tnfepes LOW gearing. The radial position of thts point way be found for the driver gear using Fig. (2.3). Essentially the length of one base pitch is measured forvard froa point E, the engaggnent end vf the line, of action. To find Ry, the position of the HPSTC for the driven gear, point D the end of the disengagement branch is used as datu The t form factor is then determined simtlarily as before except the load is located at the HPSTC. 7 _ For standard tooth forss che values for ¥ are’ available tn tabular fora: For non-conventional designs however an accurate layout or computation is still necessary. es The accuracy of the “Modified Lewis formula” is far from being satisfactory for the nev industrial demands. The following researchers are all atteapting to tmprovd the accuracy using vartous approaches. 2.1.4 BeBs Heywood = Gear Tooth as a Projection 41948) 17 This author used photoclastictty in his investigation of the behaviour of loaded projections. The scope of Mis work ts quite general. Tt deals with Screw threads and other prottyding structural eleseats, symetric or skewed, bounded by flat flanks of any practical i For the extession of his analysis to gear teeth the curved flank of the tooth was transformed to an equivalent straight sided projection “show tn Fig. (2.4).Equivalent straight sided projection FIG. 2.4 wswiaop's coistavctroy [18] « Equivalent straight sided projection Lewis tangencs - lewis parabola: \ 5 S . ft Im the analysis of the gear problem feet {¢ te explained thet Fy the Lewis’ approach is unsound. The simple bending formula was intended for long parallel beaus, not for short tapered shapes. Alsé the load is auch t00 close to the fillet so its “proximity” cen not be ignored. Lr ——C—_ true stresses, anda formula we constructed to fle the caulte. ti ie Weitere of opeetiatey ivan (opievel enitislgtventattcs (econ Cocatin the second bracket below. ‘he aaxtmum tensile fillet stre: eV. a, 7.36yt7 5: 2 lag (Begiece] o« The stress concentration factor depends upon a single ratio, it . was found independent of the load height, cf Eq’s. (2-3 & 2.4). No attempt was made to construct # formula for the compressive stresses. 21.5 Meds Jacobson = Proxinity Effects (1955: This author eaphasizes the importance of careful experimental practices. His usin Contetbution to the art was the deaonstration of the proximity effects. He shows that as the load approaches the fillet area, the tensile fillet stresses start to increase again. This happens in spite of the disinishing bending monent.+3 ‘| 2.1.6 BsW. Kelley and R..Pedersen ~ The Contemporary Sgate of the Art i anda New Formula (20. These authors have analysed and compared the afove mentioned , methods and test results, including their ‘wn, and an improved formula was proposed. : : Their original pbjective vas to find a stress concentration factor formula for 25° pressure angle, sinilar to those established by Xe wes s000 found that so unique constant would satisfy Leese and sani tooth ayers gully. Ths toad eh authors to" serutintze te etded pcojection, Also he shape of the f4ilete was wore cloely examined. . ~ |" The radius of curvature of trocholdal f11lets, which are produced by rack type tools ina generation process, has its minimum value as, tt Dlends with the root radius. Fron this point {t increases gradually cea eee) ee orem ee ee fee calculations the actual fillet radius at the location of maxima stress. i Te was observed, that aa the Load soved coward the ti}, the | Location of the peak: stress shifted toward the nvolute. the parabolte | construction in the Levis layout does act give the right location, but § moves {t in the proper direction with the variation of: a) load position, b) beam depth and c) the “tapes shap ‘The Heywood construction 1s not sensitive to this “stress shift". Once the taper of the equivalent straight side 1s determined, a26 comGtant 30° ts taken for the location of the aax- stress frou the potat of tangency,~a9 shown in Fig. (2.4) p 23. . A tybetd process wis proposed. The Levis parslola was chosen as the basis of the construction, as in Fig. (2.5) p 23. The location of the aaxioun stress was deterained by the “stress sh{ft" expression Eq. (2:11) below. the structure of the stress formula reseables that of Heywood. Kelley and Paderaey have observed ther che ongle between the levis tangent and the tangent to the fLLlet at the polat of the aaximun stress my be closely approximated by ‘ . exp = 25° = yn (2D) where exp is the angular shift of the maxisus tensile stress vhich is the angle between the above tangents. ‘1 the angle between the Lewis tangent and the ces of the tooth. ‘The maximm fillet stress ts: sin B (0.85 + 2e (bye) | (12) where 6 1s the acute angle between the direction of the load and the set of principal streases at the “erttical polne”, (Figs 2.5), p 23+ The proxtalty tems are somewhat different froa those of Heywood, Neither of thea are based on the lave of physics, they both are purely empirical but Kelley and Pedersen's terus f1t the results better. It is claimed that the formula may be applied also to concave sided internal gears. i i { | 'a 27 ‘They conducted also soe fatigue tests in support to their formla and found excellent correlation at stress levels near the endurance Linit. The significance of the “range of stress” is nentioned > with respect to residual stresses. In the carburized layers large :com— pressive stresses are locked in by quenching. These superiwposed onto the gervice stresses give an entirely sey set 9f fatigue condition noe 27 SoM, AlLtgor and: EJ. Hearn ~ iffects of frtetion [3 - The article deals with a wide range of conditions effecting the ending strength of gear teeth, However the role of friction should be mentioned here. Zt was shown that frictional forces at the contact may alter the stress pattern radically. Gear perforsince {s,only slightly affected « since Lubrteation usually keeps u sen )- | won th) IEG es = (2.13) 2 3 4 5 . ) el where factor (1) is fhe stress concentration factor + : tera (2) to the bean cantilever stress (9) fs the proximity stress due to. the bending (4) is the proximity stress due to the radial load (5) ds the direct stress due to the radial Load The location of the aaximun fillet stress {s considered to be a vartable. It is determinedin teras of fillet angle Ys in a computerized process which saxiatzes, the fillet stress. After ¥5 18 found ty, tg and t's can be calculated. The regults were compared with stresses obtained by other close form merle tests and by “analytic transformation” (5] and good correlation was found. 2.2 Stress ‘Sensitivity - Pinte Element Method ‘This 1s the most widespread tool of stress analysis for continua fm the eonputer ages Tygrefore sts application to the gear problem ts of primary importance. : Ths method does aot yleld a general solution, tt gives rather a - specific answer to a specific question. It is however extrenely suitable for parametric studies so data may be: ccunulated economically for building up a closed form empirical foroula. In this context the Finite element aethod may be regarded as “numerical experimentation’| Colenan's Pols (1 Formula (1973) (/31 + These authors developed enpirical exptessions for standard addeadue symmetric and asymmetric involute tooth profiles) from data * generated by finite elements. The variation of pressure angles and tool-edge adit gave nine symmetric and six asyanetric models. The position of the load was varied from tip to pitch line. ‘The “equation developed is similar in structure to that of Kelley and Pedetsen °[20]. . Ge = K Sop Kj for the symmetric and (2.14) eK Sg Ufc the asymmetric cases. (2.15) where Scp is the basic beam bending stress. The’ stress concentration factor: Jo. 236, : Ko. of i Ga : FA . = ’ K, and U are the “balanced and unbalanced fesidual factors” for symmetric and account for radial, shear, Heoxtatsy ete. effects are cast into these vneric tofen form respectively. 45 ee aed coefficients. 2 iI , “ky de expressed with \ power series of 24 degree in terms of a . \ 492 Ry = Bp) + atte] + aol] (2.17) ‘ ry he . | t, J nondimenstonal losd hetghe, 4s in Fig (2.8). : ” | By, Cy and Dy are functions +f pressure angle and are assumed to Have - the form: 3 A(tp) 7h +e a(= (2.18)| poureeineen, surface - both sides. £ FIG. 2.7 CONTOUR OF FINITE ELEMENT MODEL ASYMMETRIC TOOTH ~ WILCOX & COLEMAN [31] Se2 where $9 is an arbitrary choice of pressute angle, to have the expression nondimensional. The constants b, ¢), d, are detetmined fron the finite element’ results and solving aystens of (three) equations. 4 “Cy(bp) and Di(¢p) are determined Ym a sintlar manner, a quadratic equation of nine terns was obtained’ U che unbalanced residual plays a sin{lar roll for ssymetric tooth foras. A cubic equation of (16) terms deteraines this factor. the sondtnenttonal prtssure angle Nereis expressed ag SPY orn + épu where the indices PL and PU refer to the loaded and “unloaded” sides. Good agreement is found with the photoelastic results of Dolan & Broghamer [12] and Kelley ahd Pedersen [20]. Tt remains to be seen that the formula Tould apply to modified addendum cases, as the constant® were found by ising standard addendum proportions. The coristants are published in the article, thus they may be readily adapted to a programable hand calculator. : The tavestigatora used constant strain type rectangular elenents. In the low gradient ‘regions the size of the elenents was approximately 10% of tooth height. In the fillet area this size yas “reduced to 28. To obtain reasonable Wetace stressgs numerical averaging and hyperbolic extrapolatton was applied. In a later paper [32] they report having used isoparametric elenents. Substructuring was enployed to handle the Large mnber of elements in acceptable. computation time. .7 ° a 2.2.2 G, Chabert, Ts Dang Tran, Rs Mathis's (1974) ‘Hprasuies) for Tensile aad Daptesstie Fillet stresses (7 de article which extenda ehls survey to the eontinedtal furope. The scope of this investigation 4s narrower than that of ref. [31]. 20° nominal pressure angle, standard addendus symetrié profile of the European aodule system was exanined, (Fig. 2.9). The tool edge radius, mumber of teeth and the load positions were varied in the finite ’ : element experiment. From the results closed form expressions were developed for the peak tensile and slso for the maxtmu compressive fillet stres 1 -w a= |a+— fe» ] (2.19) ¥ Baa for 12. | This solution is credited to Mitchell J.H. (approximately 1900) | and sey be found in Dinette & Goodter'e Thaoty of Hlastieity (28). 2.3.1 Beds Shotter's Wedge Stresses and the Hoot Danaging Stress Range (271 : The inadequacy of the bean approach is even nore true for high Ve strength tooth fofas, such as the Wildhaber § Novikoff circular are 4 profile, than for involute shapes. It was found that when for aanufadeuring conventeace the fillet depth was increased on-such tooth shapes, the stresses have decreased unexpectedly. ‘ Apparently the close proxiaity of the contact surpassed the bending moment in gereseing the £Lllet. Therefore the elasticity solution of an Asotzoplc wedge vas applied. For an arbitrarily orfented 1 toad the maximua tensile and compressive stresses are: 2H ay cos a) ~ sin a3 cosa aan oe 7 eB a3” - sin? a3 . : 2H a3 cos ay ~ sin a5 cos a2 (2.22) a3? - sin? a3 : where rp & re are the radial distances fromthe point of load respectively, as shown in Fig. (2.10). i | | application to the tangencies on the tensile and comptessive filletsFIG., 2.10 LAYOUT, WEDGE PRINCIPLE ~,sHOTTER 27] 36ee anes eee fatlure. {t is emphasized, that the maximum tensile stress is not sutlotet for the dateraination afte tia Le ofthe ent, ever St 16 vad iam precisely. the fat dtotrbution of the steesse around tie FiLlete are needed for 4 whole load cyce to find thelr wost danaging stress range. 3.4.0 stcoss Sensitivity - Theory of Elasticity Solution, Using Complex Totentiat Funct Lans-and*Canforaal: Trassforeation The thédry of elasttetty dertves the stresses and the deflect ons fou the Atry stress furctlon. The complex potential theory relates the stress function to certain analytic functions, the so called complex potentials. Hend& the stresses may be expressed in general terns of these complex potentiais. the curved edge of the tooth profite (9 eransforsed to « stratghe edge of an inftntte half plane using conforaal mapping (Fig+ 211). The houndery condtttons on this steaight edge allov the coaplex potentials to be specified. As the stresses te the plane‘of curved contours are expressed in term of the complex jotenttals aot with spectfie functions, a solution ts obtained. , debel Te Alda 6 Ys Tezaueht These authors de credited with being the first to apply this evo dimensional élasticity solution using curvilinear coordinates to the— = a Rann nameinaammaanannnemneecceeeeeemee een eet 38 -[r] toavess-vary *(6] soravou-saNowya “:wos3 uaqes SNVId S1VH GHOdE LHDTVULS V OLN HUOSHLOOL BNL AO" ONTAAVH TT*Z, “OLE apened < (2) ™ = 2 worsoung wxozsuvaz > ane ‘quersuos A aeqeuezp opTsynogear problen. The derivation and even the application 1s quite | Involved, its proper presentation is beyond the scope af this study, ~ ' ‘Aida and Teraucht have real{zed, that the coapléx popential method {s not suitable for every day practice, therefore they which gives stresses reasonably close to } : | constructed a utility formuls ( : those obtained previousty. i The loads were resolved into tangential and radial components and “npminal stress elenents” were selected corresponding to, the load components. The layout and nomenclature is shown in Fig (2-12). The noninal bending stress is: 6H’ cob A oy -——— > (2.23) ' Be’ : whéie Bis the face width of the tooth. The nominal direct compressive stress, including the effect of tea moment’ about the centerLing of the tooth: a rin) (evey sin o, =~ ———-——__,, (2.24) Bt Be? : and the nominal shear cosponent: W cos 3 ts (2.235) ae ‘These nominal stress components were compared with the stresses, . obtained under the respective loads by the complex potential method. ‘The process has lead to the establishment of weighting factors for each stress component: and an expressi8 for the gtress concentration factor. The latter is the quantity in the first parenthesis below. The maximus tensile stress ts: yFic. 2.12 ArDaqTeRavcet Lavour (2]t oe -f 0.8 =) 1-66 of + -40 (op? + 36x2)t + 1.15 of] (2.26) = The rational structure of the formula, suggests that it may be ezployed in a wore general sense. This observation was used later by the wrteér to extend the usefalloese of the formula for esapressive peske, and for analysing frletlonal effeces. ‘Aida and’ Tetauchi Investigated the accuracy of Hofer's method for nding the “weakest secelon", vhich 1s the saae 0° eangency fr00 the tooth centerline as it was done by Heywood (18). Te was found that the location of the tensile peak varies fron 24° to 168°. It is clataed that the chotce of a constant 30% leads to on underestinate of cuax, by cot woe than St. : Thy cleo pesforsed a caries of photoslastis ceste to verify whe forasla Good agreeaent wat found for chose shapes which reseable tenutce tooth Fors, even for lov‘dedaniim loading cases. fo obtats « better representation of tye gear shapes, it was realized that more Secor tee tee te) eee by the authors of the vext article. 2.4.2 CoN. Baronet, G.V. Tordion's “Exact Stress Distribution” wmpp ‘Simultaneously with Aida and Terauchi's paper [1], V.L. Ustinenko [40] published 2 similar solution using five terms in the mapping function. The calculations were performed for the standard European tooth form. Baronet and Tordion used a similar transform function ‘and computed the stress for American Standard teoth shapes. Using thesfive terms a straight Line may be mapped alabst perfettly into a gear tooth préfile, as in Fig.(2-11) p 38. Only the discontinuity at the tip is rounded off with Little effect on the /atresses. ‘The eight constants for the trandfora Function were deterained by trtal ad error. They ate given in tabblar form for 20° and 25* , pressure angles and standard addendum propottions for six different tooth musbedt at the end of the article. The stresses are expressed by the "Modified Levis foraul: 4 WePD an Cmax *§ ——— | 227) ‘ B Xe or’e) where Jy and Jg are geometry factors for the teneile and coupressive sides respectively and are given in tabular form. Je {s unfortunate that the results are coupared only with one source which was also obtained by computation. Soae supporting evidence from a photoelastic experiment would be desirable.43 2.5.0 Survey ~ Investigating Tooth Compliance To find the dynamic, loading on e structure it is necessary to "know tts stiffness. For a pair of geat teeth in mesh the reciprocal term.compliance is ore convenient. The displaceuents are considered in the direction of the normal load, i.e. along the line of action. : 2.5.1 Weber's Semt-elasticity Solution (29) The deflection of a tooth may be {dealized to comprise of three discrete elenents. : a) The Hertzian deformation of the contact’ and of the underlying region. b) The bending and shear deformation of the tooth as a beau. we) “The displacenent at the load due to the deforiation at the ‘ foundation of the tooth. : The Hertzjan deflection 1s derived from theory of elasticity peinctples, using a stress function of complex varjable. The elliptial “Load atsertbution given by Hertz 4s adopted (Fig. 2.140). : Po Paax (1 - Gxt/b)?]4 (2.28) y Where bts the half width of the contact strip between the jodutucting cylinders,’ as ta Fig-(2-14e). tf chey are made of the saae material: ° a yr vt 5 . ee ee ina (2.29) mB ort. E on & ty and ty are the radii of the cylinders. The length “h” over which the deformation of tHe contact and the underlying region is considered, is delimited to the centerline of the tooth. The apex of the shear parabola in a rectangular bean is at theBEG. 2.13 COMPLIANGE LavouT - weBER [25] " elliptic pressure distribution FIG. 2.14 CONTACT DEFORMATION AND PRESSURE DISTRIBUTION ~ weaER [25] ( 44. 45 centetline. Hence the main flow of the forces once they have reached the centerline of the tooth istransattted through shear, which is” accounted for in the “beam deflection” term. Therefore the Hertzian deflection fpr one tooth is: woivt— my . yor — [» = 4 [ste]? (2.30) ery eenp es where h is the distance from the contact to the centerline of the tooth, (ig. 2.13). ‘ and for a wating pair uade of the sane aterial: wit ~ mm y yer + y¥'eg 2 ——>. Jin ~-—| {ste}! 31) BE > Iv. The other tvo components of the deflection were found using energy sethods. The strain energy in the tooth was equated with the work done by producing the displacement at the application of the load Chliy"). The bean deflection te given as: wf ray? santa) ay Ytp = — cosa j12, f———— dy +3. fl + —| (2.32) EB Oxy? Ba J g) 2x. where 2x, the section width, is a function of y, as in Fig. (2.13) The first term is due to the bending moment, the second, constant term is"a factor for the shear, and the third term represents the effect of the radial coaponent of the load. The writer could \R.W. Cornell {8} found the last ter to be arrive only to the latter result. 20»)46 ee eee foundation of the tooth is: w are ork tan? : rennet b.2(=) (= aes fis | (2.39 eB t t 3.1 The first term {a due to the behding moment imparting root a) che (qoaeat effecting root dtsplacenent, b) the shear force producing root rotation. The constant in the last parenthesis tepresents the shbar displacenent and the last tera is due to the radial component of the load. - . Saaple caleulations are tabulated, and are compared with the experimental data of Walker (41] finding good ‘agreement. The ,theory developed in this work 1s quoted répeatedly in the Literature. \ 2.5.2 WeJ. O'Donnell: Stresses and deflections in Built in Beams (22) This article focuses on displacements which are due to the rotation at the foundation of the bean under a bending moment. tt is shown, that for short beams like a gear tooth, if this rotation 1s not accounted for, it leads to serious errors. It 1s equally true, and’ admitted by this author in the subsequent discus ton, that for gear tooth foundation the “other” displacement elements dealt with by Heber [29] in fq. (2.33) mist also be included. . Nevertheless reasonable solutions were obtained expressed in a tabular hand book style. The nomenclature used {8 Identical to that used by Roark [24] so that the insertion of ’a correction factor K, permits ready appMcatton.a ‘The rotation, the built in end: 16 \gy * plane stress solution , on * 7 for narrow beans (2.34) wEy (t+ 15 re)? B : 16.67 m (1 ~ v4) plané strain solution 6, = —_____—_—_ for wige beans 2.35) H Ey (t +15 rg)? B winere Eg is Youngs aodulus for the support saterial. The correction factor is given as: Ky + Oy EL (2.36) 2.5.3 Ge Chabert, T. Dang Tran, R. Mathis ~ Deflections, Using Finite Elenents 7 These authors used the finite elenent set up as described tn parageaph 2.2.2 and Fig. (2-9) p 34. At the outset the boundary of the structure was determined. The cuter contour ABCDEF was fixed first, thea the displacesents along Lines PQ and RS vere computed. As tt vas found, that these displacements exceeded novhere 3.5% of the deflection at the loading polnt M, it was concluded that the boundary may be safely reduced to PQRS, where zero Atsplacements were taposed. ‘The author's rationale Yor this step vas to ascertain that a load on one tooth did not affect significantly the deflections on the next one.. Consequently the flexibility coefficients were used directly to compute the sharing of the load for two pairs of teeth in, contact. . Deflection values were generated for tip) pitchpoint and “Lowest, Point for Single Tooth Contact” loadings. From the results son-dimenstonal flexibtlity coefficients were computed, which were defined by the equation: r1 ye 7 7 (2.37) BE where y 1s the deflection the normal load 3B the face width E {s Young's modulus The Flexibility coefficients were then plotted vs Load position and it was found, that they depend only on the geonetry of the tooth and fon the position of the load. The shariag of the load was determined by considering the two teeth in contact as springs in serte the pairs as springs in parallel, as in Fig. (2152)! Hence the two branch forces equal the total force: wth aw (2.38) The two branch deflections are the same. Cera + Hap) Wi = kaa + key) We : (2.39) The loads per unit face width were chen plotted vs position on the Line of action as in Fig. (2-15b)- Te 1s clear that when all the load ts carried by a single pair of teeth, 1 fe the total sormal force. However for double engagement the load variation of a gingle pair béing “strictly linear” is questionable. Some additional evidence would be helpful to-support such claim. The process where the boundary was reduced mst be examined mote closely (Fig. 2.9) p-36- The proxtaity of the outer boundary ABC of zero displaceueats to line PQ has controlling influence on the, displace- ments along that line. This very fact ay invalidate the conclusion, which ruled out significant interaction between adjacent. pairs.flexibility coefficients Gear "b" (Driven) a) Double Singh Double Pair Contact b) re, 2.15 toaD shanise - cuseer, et ALC 4950 2.5.4 As Premtihat, G.V; Tordion, C.N. Baronet ~ Deflections Using Complex Potentials (23 The complex potential functions obtained ty G.V. Tordion and G.N. Baronet [5] are ‘used to find the deflections. Using plane strain analysig, the displacenents of the median line of the tooth along. x,y coordinates in the original z plane are obtaine: Due- to’ the singularity at the potnt load, the deflection at the potnt of intersection between the tooth centerline and Line of action ts couputed. To find the total deflection one may add the contact defornation as'per Eq. (2-30) and footnote te the reported values. | The dteplacenents of the tooth centerline are compared with Weber's [29] resulés graphically. They show similar trends, Webers values being sonevhat higher. The difference is itkely to be reduced 1 the contact deformations are added to the deflection of the centerline. 2.5.5 RW. Cornell's Expresaion for Pair Compliance [8} The deflection ‘of the tooth comprises of three component: a) The deflection of che tooth as a centilever bean b) The deFlection*at the load due to the deformations at the fillet and the foundation area, ’ c) The local deforaation at the contact and: the underlying region i delimited by the tooth centerliie- Computation Of the-beam deflection in general requires an integration. This operation 4s approximated by a summation process, which {s based on elementary strength of materials principles. In this form the ‘calculations can be readily performed on a progranable hand calculator. The layout and nonepelature.are shown in Fig. (2-16)+ a XMé BP: indation- regio, Rgfndacton (ite. 2.27 FILLET aN Fotsparton comPLiaNcE - comveLL [8] HY w32 ‘The deflectién of the tooth as a beam: vein W costa a ae [2.414 + tan2ay) ype ——— | by —— BE sl i (2.40) 1 f ro4 a toa where — = H— + — asd — = o— + — ae ta eal Ty ts the second azea soment of the ipy segment. ‘Ay 18 the, cross sectional area of the igh segment. dy ts the length of the that seguent. : 2 11 {a the distance of the segsent from the “eritical section". ‘The expression above assumes narrow teeth {.e. B/t C5. for wide teeth the values of I should te replaced by I/(I-v?). For the deflections at the Toad due to th deforsation at she foundation and £4let arese « aoditied'vereton of 0'Donsell's snsigete (22], was adopted. The built in end {s subdivided into two regions te. to foundation area or root and fillet area, see Figs (2417). The deflection at the load due to deformations at the root is: W costa [16.67 (2p)2 ty tant, y'gp * ——— { + xiv) | —|+ 1.5361 + —— 3E + Ute. tp. 2.4C149 plane stress solution for narrow teeth (3/T< 5) (241) W cos? 16.67 (tp)2 _ i-v-2v ty" tan’s ven oan EE (Ae sf] 5 * le re Ue, 2.4(14) plant strain solution for wide teeth (8/t>5) (2.42)"where the tera Ast 2nd ‘3rd 4th is due to the type of deforaation at the root rotation 1) ‘rotation! 44) dfeplacenent! deplacenent dfeplacenent th 33 caused by: moaent. shear radial component tp is the bean depth, which is obtained by maximizing the deflections. ‘Qty ts the distance fron the load-centerline intersection to tp~ ‘The deflection of the fillet area y'pp 1s calculated‘simd larily foundation and the fille y'e ty'ee + y'rp t area: y'g and the total deflection due to the flexibility of the (2.43) For local deflection Weber's [29] equation is adopted. Tf the mating gears are made of Mev?) ow ie ee rE B the save material: v 0-9) (2.44) ef (2-31) where hy and hy are the distances from the contact point to the respective centerlines of each tooth pong the Line of action. b 4s the half contact width (Fig. 218). In the dynamic program [9] the compliance of a mating pair ts considered. For this reason the deflection of the two menbers must be added, as they actin se! ries connection. are equal ters as per Maxwell's reciprocity theorem.hede, eds h rR sin ¢ ° Enlarged View of Contact a.) FIG. 2.18 LOCAL COMPLIANCE - CORNELL [§)‘These Lengthy expressions are calculated only for 5 Loading positions. Using the 5 results curve fitting is applied and a sth degree polynomial 1s generated, expre jedin terns of a special normalized position along the line of action. The pair compliance equation: where rebel Gg) Sq is a reference distance along the Line of action, '§, for the approach and Sog for the recess. Ss the directed distance of the contact frog the pitch point, positive toward disengagesent. ‘AyBysC] sDy are curve fleeting coeff tetents. Cpo ds the pair compliance at the pitch point.2-6 Conclusions of Survey, 7 The articles examined indicate a considerable development frow the Levis formula to the application of eciplex potentials in the art of calculating fillet str s+ Consequently there is a wide variety of formulas available for this purpose. ‘The Heywood [18] and KeLley-Pedersen [20] expressions have ; overcome most of the early difficulties and take into consideration the effects of stress concentration, radial component of load and the pects of “proximity” at lov dedendum loading. The Colenan-Wilcox polynosial equation covers a wide viciation of pressure angles, treats symmetric and asymmetric tooth: forms. The Baronet~Tordion tables [5} are expected to give accurate “form factors for the American standard tooth form, The Aida-Teraucht [2] formula retains the bean principle, accounts for all load coaponents. The - butlding block structure of the weighted stress coaponents has a. : rational appeal. : Te was indicated-ty hotter [27] and others that the sextmum tensile fillet stress, even tf it-~w accurately known 1s not suffietent “te peedie fatigue faiture, The strane éleeeibtion of thé sole q flee, observed over a full load cycle, can only reveal the most stress range. dangerou! Relatively few articles were found dealing with the compitance of gear teeth. The basic analysis 1s due to Weber [29] complenented by O'Donnell [22].' The knowledge of this property is vital for dynamic 4 analysis and for the deteraination of load sharing between stmuitaneously acting tooth patts. The antiysie applied by Gubert et7 al (7} to the simple case of LRG {s not cotvincing. For evaluating ~ HCRG this problem rises to primary importance: A full trestaent is given by Cornell and Westervelt [9][8]. It encompasses the determination of tooth proportions, compliance, dynamic loads afd their sharing, and of the saximum tensile fillet stresses. These theoretical developments are supported by experiments where only single teeth vere loaded. Some hard core experimental evidence 1é lacking on the interaction of simultaneously acting tooth pairs. Therefore tt ts deened, that by conducting tests on a high contact ratio photoelastic sodel with mltiple céntact and rolling ’ capabilities, our knowledge could be furthered concerning the following: » "How is the total load shared? . 26 Hoy are the fillet stresses affected by the :tnteraction of the ‘ > branch losds’ acting at sky point along the Line of action? : ‘To anaver these questions “the conventional photoelastic stress detgraination mst be auguented by deflection seasurenents fron uhich the coupliances snd the sharing of the Yond say be calculated. The, expertoestal reguits would be’ supported by finite elenent analyfhs. jure of the quality of the design a .new concept the Asa ne WALL be introduced, whfch ts expected to reveal the stress Matttanc ee ence era eceee such exngrinent aay yleld he range of stresses expeienced by a given fillet point during a loading cycle. This {nfornation may be used later for fatigue analysis.”CHAPTER THREE 58 FORMULATION OF THE CONCEPT OF STRESS ADMITTANCI AND FATIGUE CONSIDERATIONS The fillet stres: for HOR are itnear with respect to the load. Therefore the stress distettutton for' multiple tooth engageaest say be obtatned by superiaponing the atressen generated by the Indtvtdusl Losds. A stagle toad induces prteary stzeeses fe the adjacent filets. ‘Secondary atreeser appear on a aukecancial portion of dtefused stress {elds The eros effect of the loath results therefore in a complex stress distribution. The problem is further aggravated by the fact, chat the ind{vidual loads are not known. Bel Stress Sensitivity ' “Te was seen! that the “primary” fillet stresses due to a single load my vary greatly withthe shape of the ‘tooth. However once a particular desigi 1s chosen, the stress: depend Linearly on the aagnitude of the load, and in a anderately oonlinear aanner on the location, vhere it is applied. If this location ts identified along the “Line of action", a common paraseter for the pinion and wheel 48 obtained. The stress response of a single tooth due to a unit load: was referred to by Cornell [8] as the “Seri of the shape of the tooth and is a function of the position of the load. ° On tae where Vie the stress sensitivity GD J te che maximum fillet stress, tensile or compressive W ts the normal load - Sensitivity". Tt ts a property.’ Inietal position Fic. 3.2 FATIGUE DIAGRAM5 60 3.2 Load Sharing in Multiple Pair Engagement When a tooth acts in a group of simultaneously engaged paire of three theth, {t may be modelled for static analysis as in Fig (3-1). 5 The total load branches out fo three parallel paths using initially similar logic as tn ref's [7] and [9]: Nn Wath + of ’ (3.2) A pair of springs connected: in series represents a contacting patr of teeth. The parallel branches model the siaultanously act ing pairs. Intttal gaps are indicated for example in branches 1 and 3. These gap eventually aay be present due to manufacturing errors of profile modifications, which are Introduced deliberately to aintaize Iapact loading. 7 The total displacenent {s the sane for each pair: Pree syyey Qo, The total dtsplacenent in branch 3 1s typically the sum of the deflecttons of che teeth in that pair plug the efze of the gap gj, which must be closed before these teeth can deflect. y's *y'3a +¥'3p +83 ~ G4) ‘The compliance in general: ) y ee 7 (35) w The compliance for pair 3, using Eq. (2.3) y'3a + ¥"3b oy" > 83 C3 = Cag + Czy Ly Wy61 . giving the branch load 7 yin a yin 8 \ y= oe G.6), % oy : “The total Load from Eq's. (3.2 and 3.6) Hl i : ¢ Boy! = gt ‘ wer tL where nts the mmber of pairs simultaneously engaged ary, gy is the gap at the 1th pair. ‘This gap is taken relative to rr —. Solving for the relative deflections between the two gear bodter/ a at 7 yelwer =] oq (3.8) sl Gy + where Cy te the equivalent compliance of the n pairs connected in parallel {.e. the system compliance with the gear bodies congidered to be rigid. — - aol . Z er (3.9) Ce tel Cy If the gaps and coupliances are known, and the compltances do not depend uposi the load, the syaten deflection using Eq's. (3.8 and 3.9) say te caiculated, Then the branch loads Bay be found From 24. (5.6)« Adat Et : 3.3 Se ‘en from elenentary properties. . ‘The stress sensitivity of a tooth when acting as a menber in a group my mpt Indicate the moat dangerous location for the'losd. If7 5 “ applied at the ip {© induces the most adverse, stresses, but also causes the largest deflections. If the combined compliance of such tooth and its contacting matg@s more then the same for the adjacent pair, the bulk of the load would seek out the other stiffer path, thus leaving a snaller load for that more sensitive location. Therefore it 4s proposed chat for a tooth acting in a group the tatto of ite stress sensitivity over the combined compliance with ste contacting mite to be considered for ebtablishing the eritical location for the load. Froa Eq's. (341) and (346) the stress inleooth “3a Yaa 3a" M3 Vaq * (¥" ~ 83) — = Cy" ~ 83) Zaq 3.10) C3 Let Z be called “Stress Admittance” and be defined by the equation: v ze Get) cp Where V is the stress due to a unit load (stréss sensitivity) tance cp is the compliance f the tooth in question plus the co W of its mating pair. Or alternatively multiplying the numerator and the denominator by the load: Snax : 2-— (3.12) y'p qthere ax 18 the maximum fillet stress, tensile or compressive ytp 48 the combined deflection of the tooth in question’ with its mating patr. Both parameters to’ be generated by the same load. Z 1s a property of a subsystem, a pair.of contacting @eeth and is 4 function of the locatian of the load. At a given ‘point it hag fourEach particular point on the fillet contour experiences a ad values, each associated with one of the peak stresses in vheel and pinion, tensile and compressive. 3.4 Equivalent Static Alternating Stresses Fatigue analysis 1s beyond the scope of this study. However sone aspects of it are considered to demonstrate that the knowledge of the uaxfMmfensile fillet stress my not be suffictent to evaluaté the load carrg&gf capacity of a gear tooth. ferent range of stresses over a full load cycle. These ranges may be described by the mean of its extremes and the alternating amplitude about this mean. A comparable mpasure of the danage potential of these patra of stress components my be obtained by using the Goodman relation: oom . tei (3.13) xe Sut , where gg 1s the alternating component of stress og 48 the mean stress the ultimate stress in a untaxtal tension test the equivalent alternating stress wtth zero sean having * the same damage, potential og and o,.represent. ay, 1 Pigs (342) 9 the allovable maxims cggs Tt may be set at 60 of the endurance limit of the materiel. This method of evaluation was used by A. Elkholy [15], the diagram 42 shown in Figs (3-2), pe59- ti CHAPTER FOUR 6 : TEST EQUIPMENT =” . 401» The Prototype The subject of this investigation 1s one of several proposed shapes for a pair of spur gears, in the reduction, gearbox of a newly developed turbo propeller engine. The tooth proportions were designed uging @ computer program which 1s in the development stage and is based fon the articles of Cornell and Westervelt [8]. “High “Contact Ratio” tnvolute forms of esdtfied addenduas vere employed. 7 The gears are proposed to be made of AMS 6265 carburized gnd hardened steel, having a Young's modulus of E = 207 GPa (30,000 Ket). The gear data are given in Table 4.1. The tooth forms are shown in Fig’s. (4.1) and (4.2). ‘The various portions of the contour such as, true’ tovolute, modified tim arcs of fillet -and root circles are indicated. 4.2 The Model G A scale of 5X magnification was cliosen for. the-models. It gives a high resolution of position and serves to attenuate the .effect of asnufsctoring srrors. Five teeth were machined on each model, which were nade of . 6.35 mm (.25 in) thick, Photolastic PS¥-1 clear polyester plates of 305.x 254 and 350 x 305 mm (12-x 10 and 13 x 12) inch sizes. For sufficient diffusion ofthe applied forces the bodies of’ the models were reinforced by 9.5 mm (.38 in) thick polycarbonate plate cemented to both sides, see Fig. (4.8), p 81.65 i , Pinion Wheel Tooth vize ] a = 261m | (®p = 9.73 tnt) Nominal presoure angle | ro Number of teeth | 36 . 27, Pitch dqneter |< 93.977 357.635 [| c3t6e59 (14.0805) Outside diameter (nean) 1) fo2.21 362.81 (42024) C14 284 Root diameter (mean) 87.56 348.32 2 onary 13.713 ) | “true tnvoiute Form Dianeter™ 90.330 351.559 | (31556) (13,881) ["aoot fillet radius 1.331 ain 1.207. ain (010524 min)] (0.0675 ata) Circular are tooth thickness (mean)| 4.365 3.570 corm) (0.1406) Tooth to tooth composite error 0.010 | (0; 0004) 4 Center distance 225.806 08.89) Maden! 1.576 a 0.991 = Dedendus 1.228 a 1786 Contact ratio 2a Face Width 76.2 3:00) ‘m is the module in mm's. fe given in mn's, their inch equivalents are’ shown in. 66 wow se 7A 0004") — 7.71 mm (0.306") Modified tip True involute © saw oo 88s) fe Ciroular fillec RZhoot etrete : “, FIG. 4.1 TOOTH FORM ~ PINION 5.26 mm (.202") Pitch circle True tavolute \ j— Circular fillee 88°40" L . > Ny bRoot etrete FIG. 4.2 TOOTH FORM ~ WHEELs . es PSH-1 has a Young's modulus of 2.34 GPa (340 Ket), Pot: ratio .38, a nominal material fringe value of 7 KPa-m (40 psi-in). It te non-hygroscopic, therefore changes in humidity give no boundary stress, te, little of the so called tine edge effect: 2.1 Computation of the Profile The size of the sodel prohibited ite machining by generation. ‘The rematning wethods knowm.to the writer for fabricating the tooth profile required to calculate, the coordinates of the contour. All these computations vere performed on a Texas Instrument TI 59 programmable hand calculator. . : 4.2e1s1 ~The “Tnvoluce - ‘The equations for the involuté may be found in gear handbooks {6],(13f. For conventence they are repeated here in some detail. The involute is often described as the path of a point on a cord which te held taut, while it 1s unvound fron a cylinder, In Fig. (4.3) the cylinder 1s the base circle of radius % of @ gear, The cord is shown as line AE. Point A initially coincides with point B oa the base cfrcle, 1t passes the pitch ctrele at potnt P, ending its path on the gear at the outside dtaneter at point C. : Arc 3E has the sane Length as side AE in the right triangle AOB. ‘This key step gives the roll angle: cw-ag?t es = = —— = [(/g)*-1]t = tang (41) Be Rg Ry The pressure angle may be obtained directly: . % = cos™) (Rg/R) (4.2)y ¢ Are tooth thickne: So \e Piech efrele \ \ al ' Base * circle * 8s Oceeae ° P Hy | 0 FIG, 4.3 INVOLUTE TOOTH PROFILEThe vectorial angle # 1s also inom as the involute of ¢+ . Ome-$ = tang-¢ = invg (4.3) tn the destgn process the base radius Rp 1s volally obeatned a ee el ee ee ener oes Ry = By cos 4 (4.4) Are, tooth thickness T deteraines the position of the involute curve in relation to the centerline of the.tooth. It is measured oa the pitch circle, vhere the-pressure angle ‘s #p by definition. The” origin of the Javolute point Bis at polar angle ap with respect to the centeriise of the eaoeh. Bfom Fig. (6.3): ap = tp + 8p or T on =— + tandy ~ 9, (4.5) i pop The polar angle a for a general point A: a= ap-0 or ‘i a= + tangy = gp - 4 + tang : (4.6) 2, This 1s a polar-equation for the tnvolute curve as a tooth The polar angle a te implicitly the function of radial profil position R via Eq (4.1) or (4.2). The tie=in with the gear is through design paraneters T, Ry and ¢ps The normal to the involute with respect to the normal to the tooth centerline aay be read off Fig. (4-3). Amecan . (4.7)70 The Cartesian coordinates of a general point on the involute: x= Rcosa (4.8) y* Reina 4.2.1.2 The Fillet + At the time of designing the model, the exact manufacturing process for the prototype was not known, hence the actual shape of the fillet curve was uncertain. As an approxiaation circular fillets vere chosen for both aodels. The fillet curve was computed according to the geometry shown in Figs (4.4)«, . The center of the fillet circle lies on an involute parallel to the original profile at a normal distance of rg and at the same radial distance from the root circle. ree = Re tre and 4.9) rE . ' ten (4.10) Rp ‘The figure indicates that act agtse- 6c * agtéc-ectde oF Ry +rey? Jt ag = ag + coo? — (eS) “| +t aan at Bs s ‘The center of the fillet circle is determined by z Xq = (Ry, + Fg) sin ag : ed (4.12) yo = (Ry + tg) cos ag 7 e 5 The coordinates of @ general point F on the fillet at an angle yp my be found from * _ xp Xe - TE C08 YE (3) ye Yo > te sin ye .FIG: 4.4 FILLET LAYOUT noo n The fillet radii on the wodel wete increased by approxtnately St over the cified miiimum. This represents 1/4 of ‘the possible increase of fitlet radius for the pinion, without violating the “ rue Invblute Fora Dia" Mmit. The choice of this anount was arbitrary, the * change cae concur with the te of Kelley and Pederses”{20]. ‘The “radius of curvature of the’trochold, whieh, testhe actual shape of the Eilet produced” by a rack type tool, {9 substantially larger at” the regionof the peak stressds, than it ts adjacent to the.robt direle, where {ts minimum value is located. A moderately oversized radiue 18 thus a pore representative substitution. . 4.2.1.3 Tp Relief : m0 Was the gear tooth is an elastic structure it deflects underthe. loads, Consequently a loaded tooth lags behind in motion in relation to ah unlosddd togth néxt in ine. This tooth would then engage/4n a non conjugate contact with an ispact, the sharp edge of the tip scoring the mating face, causing early failure. Manufacturing-errors in the tooth form aggravate the problem. Renoving: spac iateriai frog the tip to snoothea the faitial contact is thg usual remedy. ‘The amount of tip relief may te calculated from the deflection of (jhe tooth to which the composite errér is added. If the transmitted torque ‘varies widely, a tip relief corresponding to the maximum loading e : may cause ,tmpact at lower torque values. = / Ae auch condition the deflection, of the loaded pati does not f11i the gap created by the tip andification. As the contacting patr aus , the next pair hes to jump the gap.m ' Hy A compronise aust be aade, the designer should be guided by good Judgement. The relief has its maxizum at the tip, then gradually decreases until it blends at a certain point into the involute. . Suggested optimua value#are given in two empirical formulas in D.W. Bear Handbook [13], Chapter 5, pe 23 Dudley’ } . ‘The maximum modification at thé engagement side WON). Senaxtam) = 5.08 x 10S (14) + a oe B (am) e . where eatérial ts reanved from the tip of the drtven gear. The sane et the’ disengagement sides : , vein Beanxlam] = 2.90 x 107> (4.15) 3 (nal | 4 . where daterial is renoved from the tip of the driver gear. | ® These formlds gave sax of .OMJ-m-(.0005") for the driven . wheel, and 0072 om (.0003") for the pinion. The designer of this drive “ . adopted 0102 am (.0006) for both gears, which was chen Lgcorporated | a ’ ; tnto the photoelastte models See I ‘ The departure from thé Lavolute aay: take a parabolically tner 18ng ‘form. Such scheme'1s given by R.W. Cornell and W.We ¥ A Westervelt [9]. The mdification is given in terms, of the position of the contact along the line of action “ . “$= Sef : a 8 * Snax |—— and ° . Soe ~ Se. 7 5 ef : (4.16) . 8-7 oT yen aoa . of - Vs | :’ 14 ote Soe and Soq are the end points on the line of action. Se and Sq are the respective values,where the tip relief starts with zero. At LORG usually ‘the "Highest Point for Single Tooth Contact” is chosen for starting the relief curve. For HCRG an analogue “Highest Point for Double Tooth Contact” may be adopted as the initial polst of tip modification, Replacing Pp in Eq. (2.8) by 2Pp and using Eq. (2.9) the respective Ry's and their corresponding positions on the Line of action were computed, obtaining: Ry = 180.4 mm (7,102") and Sp = 5.14 am (0.202")for the wheel. Riy = 49.8 mm (1.960") and Sq = 7.71 am (0.304")f0E the pinton. ec ne acetate mente This the inclination of the nordél to the contour at the check point: angle for the true tavolute using Eq. (4.7) and Fig. (4.5) 1 Aa A> 9B At the modified profile this notmal rotates with the rate of change of modification with riypect to radial’ position. a 5 —e an ay =a + aR Se dme/4R may be obtained from Eq. (4.16) and from i S = Rp sin $y ~ (R? -.ag2)t Bq, (544), p 92 " : see alag Fig. (5.1), p 86 dm day 45" S- SOR an + = 2 tex —— — ‘aR ds mk Soe =-Se (R? ~ By?)t 2 memax’ S- Se 5 Fein @ Soe = Se\ ‘. - B 15 Tooth ] center. f ‘ ‘Modified tip avoluce profile tangent at oa aca 1 | , ‘ / |, 1c. 4.5 Trp moprrzcatton : ; Menlo | : V | J; . 16 where § corresponds to radius R at point A. From,Fig. (4.5) it may be seen that 3 MAO = 90° - (aq +4) Trtankle AO Lends Lf to find the polar coordinates for point M. Ry = (RA? + mg? ~ Ry me win (oy + 24)]t \ tae ste (4.19) a . — cos (ay +44) (4.20) 8 4.2.2 Making the Models eee eee eee oe ie cae controlled machines (NC). ‘These systems accept discrete points as raw data, the “Best fie" curves are internally generated for servo control. 50X (prototype) comparator charts eeteae tooth with 2 spaces were dean on aylar vatog an “Interagtive Graphic Syates”. > The ceaplates for the models vere contoured on an.NO Onsru atlLing, machine. Unfortunately the’ latter produged a wavy surface. The templates had to be Hiptshed by hand, using the comparator charts as a gutde. ‘The contours were, transferfed from the templates to the sodels on 2 Photolastic HSR-L router. The-tw Mp router bit works under water, - thus stgnigteant residual stresses are avoided. . \ Tt was goon apparent, that the HCRG system is more sensitive to ichining errors,gthan the conventional LOR gearing. The accuracy so & a es ~ et far achieved, produced an un}dkely 'photoejastic picture. It was realized, that 1f any useful conclustor's are to be dravn from a multiple * contact test, the precision of the model should’ be one order higher than : that of the prototype. The SX enlargenent is helpful, but adequate(4), The gage blocks (5) tilt the model to a desired angle with high " : n nethods for machining and for inspection vere yet to be found, : The solution to the probles of inspection: cane first. A coordinate ssepectipn sachine DEA-Canma vas chosen. Ax the probe of the atching appronctise the target the computer registers the position of the probe at the Lostant of contact. It was clatsed, that this machine operates with a repeatability of .005 mm (.0002"). ‘The method which was devised for the adjustment of the profile nay te explained with the help of Fig.(4.6). The model (1) 1 clamped to the angle plate (2) with datum pins (3) held firmly against the side accuracy. Therefore 1f the direction of the normal to the profile at a selected potnt ts known this aornal can be vet exactly vertical, =~) Consequently an abrasive tool ich its motion restricted to a hérizontal plane aay touch the model prectecly at that point, s@ Fig. (4.7). The preloaded ball elide (1) gave a backlash free Linear aotion. , Te was aounced on a cross atlde (2), giving vertical adjustment by the means of its \éid screw (3). The sembly was bolted to an angleplate (4) oo that the top surface of the ball slide could * set ‘ortzontal within .005 mm (,0002)". A wedge nosed square tool be (S) with protruding front was attached to the top of the ball sltae, A f4ne (400 \ grit) enery paper (6) mounted on the nose with face dow} gave the abrasive aufface, . \ y With some caré this set up made it possible to Fenpve ae Paiele a .003 mm (.0001"); aquare to the plane of the aodel trod any convex tegion of the prdfiles The contours of the rolling surfaces wereeight gage vith indicator pat pin G) + Angie plate @ win ) Sine tarie@ € cage blocks G) Granite surface table Note: Clamps are not shown. PROFILE ADJUSTMENT SET-UP, MOUNTING OF THE MODEL9 «pos nt wind 7 Enery paper (400 grit) attached (6) by double coated aasking tape i verticat siice@ \ o | = | ' Lae sitde @) | 7 Angie plate @ Lt : fe Right angle block resting Wote: Clamps are not shown, on granite surfa plate ned FIG. 4.7 PROFILE ADJUSTMENT SET-UP, GUIDE. FOR ABRASIVE TOOL- 80 corrected point by point in several passes. The erve tavoluce wna evenevally epprpcinaced within .O13 a C0005"), The oveeth caeatbtitey whch var affect Aba by ether condteton auch an the Flatoean of ‘the plastic, the accuracy SB\e decun totes ete. tahibtced further teprovenent. __ Pe moe teva eran, witch affect the sharing of he Lad setvcen patze of feth to the devtation between thelr respective distances along the Line of action Le, the relative base pltch error, see Pig. (4.6). These dtstances may be seasured dteectly on the tticed set up described above. As a final step for refinement of the profile the base pitches for points in simultaneous contact for the planned loading positions were tuned within .008 am (,0003"). Ven the andelsArere installed into the rig the maximum gep tn the nvolute region was found to be Ol mm (.0004"). 43 The Rig i ‘The experimental setup is shom in Fig's. (4.8) and (4.9). The aodgle must be mounted on thelr respective conthrs to perait gesuine rolling. This would require an excessive size of photoelastic material fo sake a one piece wheel. A commescially available size of plastic was chosen and {¢ was consequentlf augmented by an aluminun plate, the arm (1), whtdh vas Joined to the plastic model at the. polycarbonate side plates (2) using press ft aluminum bushings (3). . Deadvetght loading pulls the ara upward by means of a 1.2 m (3/647) die flexible, stainless: steel cable (4), passing over evo pulleys (5),(6) equipped with ball bearings... Space between the side_~ straining rane @ / ray © Polariscope Loeding Disc = Upper @ Loading Unit - lower () ‘ Lever systen @ \~supporeing tench @ FIG, 4.9 EXPERIMENTAL SET-UP ~ SIDE. VEY ‘Travelling Microscope r. 82Vole plates yields access to the bottom side of the arm, to which the cable 83, is attached. ‘The circular arc’on the front end of this arm is furnished with a V-groove in which the cable is seated. This arrangement ensures’ { consistency in torque with respect to rolling and delivers the load to the vertical centerplane of the model. ‘The rotation of the pinion is restrained by the reaction Link (7). Tes length is adjustable, hence the. pofats of contact may be moved along the Line of action. This device 1s essentially a turnbuckle, made to accept hardened setting blocks (8) ensuring a repeatability of ting witha .025 am (001 tn.)- ‘The veight of the reaction Link ts balariced by deadveights via a second cable and a lever (not shown). ‘The center distance for the models exceeded the width of the straining frane (9). An outboard truss (10) was added to the structure to support the center of the wheel. Fine adjustuent of the center distance 1s accomplished by a differential screw (11), which shifts the i pivot hole block (12) within the arm. . The deadwetghts are seated on two discs hooked in tandem. The upper dise (13) carries a small load just to maintain a positive contact. The main load {s applied to the lower loading unit .(14). Rapid and sooth loading or unloading is achieved by a lever systen (15), which could engage of disengage the lower loading unit, see the side view Fig. (4.9). This device ia bolted to the supporting bench (16) of the polariscope and is operated from the viewing séat (not shown). ‘The experimentat set up is also shown on photograph Fig. (4.10). :_, DTH ONINIVULS ONY ZaOOSTEVIOd OT" y, “OLEo CHAPTER FIVE as \ TEST PROCEDURE al ——— Ay Sel Geonetrte Considerations -_ For proper correlation of Yata and for the planning pf the test tt to necessary to know in advance: yy ppint on the pinion that will pene with a given = WL, potnt on the wheels . be) “The points which are simultaneously in contact. S121 Theoretical Match-Potnts of Pinton and Wheel These points hay be found using Fig.(5.1)- The center distance: c= Rig) + Gn. where the prae refers to the pinion, the unprimed symbol to thé wheel. The angular posteton of the contact: (5.2) | | | 1 be tp oop > mie : Tle radius'on the rolltag surface of the pinion, whieh aatches the point’ \ \ at Ron the wheel a _ . R= + - 26 cos yt 6.3) The position of the conta on the-Line of action: oO | - Set sin ak ce 38 2 >t \ (3.4) SN 2, Roints of Simultaneous Contact Satake of Sipubceneone meas The polats which are 1% cdptact at the sane tine my be determined from\Fig. (5-2)- The' distance between tvo contacting potats is the base pitch Fy watch may be obtained from: /86 eanssead 30 oust edar| | | ae . 88 eS oo Py = — Rp 5.5) : N where’N is the number of teeth. Using the nomenclature as shown: ao : cos (90°F 2) = - sin 6 - (5.6) cos (90° 49) = sin ty (7) IE. the radius of the contact in the tooth on the middle ts known (fe), the contact on the higher. radius will be at: ‘ ; Ryn (Ry? +P? + 2RPp win golf = (5.8) See ee Ry = (Rp? + Py% - 2RQPQ sin gg)t (6.9) ¥ 5.2 Sealing / As the geometric proportions’ and material properties of the model: ‘ sce different ftom thowe for the prototype, “thete load, ‘stress and deflection relationships mat be established. Thetr sinilerity to remain valid, the distribution of the stresses cust be independent of Poisson's ratio (10],[14]- a ‘6 For plane stress Analysis of a simply connected structure jn a practically uniform gravitational body force field, the Potsgon's ratio does not influence the stress distribution. The presence of holes make * the modeis and the protM¥ypes mltiply-connected. The loads are reacted at these hoes hence thetr boundaries are not in self equilibrium. Due to this last condition the stre: jes do depend upon Poisson's ratio. However the fillet areas are sufficiently far from the holes, hence "X. staple connection my be safely assumed.For these reasons the relationship between model amd prototype saling factors. These may be determined can be expressed by constant from dimensionless ratios 1, formed of relevant patameters. Such ratio : for stress, using o = W/A.” ~ . Ne oat, . . ns W and, from uniaxial deflection y' - ye : ne : (5.11) 7 . where L length and y' displacement are dteigaces In the plane of Ue sade andof the gua eepecttely. : : B the face width of the gear or the thickness of the model are teanaverse distances. . W ts the noreal Load £ fa Young's moduli If for. and y' the same scaling ratio is chosea, and Bq. (5-10) + te divided by Eq, (5-11), 4 newedimensionles ratio 1 obtained, - ‘ (5-12) The similarity will be valid if the dimensionless ratios are the sane for the model and for the prototype. . 2 eal - es . then Gas” El Led, 5G 2s ; (5.28) s & 7. where m and p stand for aodel and prototype respectively. Using Eq's(5.10) ‘ang (5.14), the Load scale factor will be: 2 Fa a tay Ca ‘The scale factor for compliance from Eq.(3.5) and substituting L for y't 2m ye Cee ' The numerical values of the scale factors which energed ‘ss design choices arg: v'm ap : 6-35 om (.25 in) ty y By 1. elm Gin) BD [ 2nse ore cto tow) : % y 1 207 Gra (30,000 ksi) 2, on 1 88.2 The, load scale factor from Eq. (5-15): . wos a Wp (88.2)(12) 28 The scale factor for compliances from Eq. (5.16): a. = = (5)(21L8) = 1059 a 2g “This scaling scheme assumes plane stress conditions, which is correct for the model. The B= 76,2 mm (3 im.) wide and t= 5.36 om (.211 tn.) deep canttlever bean for the prototype is definately a “wide. Plane stre: bean! (B/e = 14,2) co which plane gergin, ansuiption fice betters. , Te problgn of ateennon in a wide team ogi addrenned by Wellner and’ Setreg [30], Roark [24], +134, Timoshenko and Goodier [28] in article 124. The aspect of deflections is dealt, with by-R.W. Cornell [a]. . From ‘these sources including some finite-elenent investigation by the writer, no confistent simple factor has emerged for the . ) tofiitde bean (plane transformation frow narrow beam (plane stre: strain) analysis. x wy It was recognized that this matter requires a thorough {nvestigation, hence it was excluded from the scope of this study. vas. assumed, with te understanding that the results do not reflect the more realistic’ plane strain condition for the, prototype. 5.3:0 Loading 7 - The prototype normal load {a given as 31,600 N, (7090 Ibs) hence the model normal Load: 50 ! ig = 31,600 N / 211.8 = 149 W (33.5 Ibs) . 5) From equilibriue of moneats for the rig the’ required’ deadweight he 227 8 (51.0 Tbe), (Figs 5.3). “As the wodel was aot completely free of residual str decided that the load would be applied at two levels. The fringes were’ measured accordingly and the diffeiaces in load and in fringe readings were considered, Thus the residual stresses and the slowly changing environmental effects. vere cancelled out, The reduction ia bt. / error was somevhat offset by an increas type tm precision errors due to the two readings per data point in place of one.Gl ass [sino (22:42 4a) yo i 'm (°33.5 tbs) \ - From ‘equilibrium of soments the necessary \deadveights are: 149 = 227 8 ¥51 tbs). . ee . . 93 - For the Multiple Contact: tests the folloving deadyetgh Loads were used: High level ‘load | . 239 N (53.7) Ibs) \, low Level load ~2N « : Yest load nN (51.0 1bs) 7 2 For the Single Contact ¢ te SOR of the test load was’ employed. * In the case of the deflection tests the low level load was reduced to 1.1 N (0.25 lbs) and the load required to. produce a given deflection |vas measured each tine. 56 : . 2 2 7 The! sacy of load transmission was chetked using a Ravere , : 5-500 N (1-10 1ba) ‘range Load cell. The ball bearing plllltés were found to pass 99.8% of the applied load through, 5.31 Selene of Loading Points “ bog ‘The loading points were firs't planned in terms of rad{al position Smee erin Geen G starting from the pitch point, For ons extreme,*the erttidhi tt . Loading, the point 127 mh (.005 in.) below the nominal outaide radiue was chosen, @ conservative eitinace of distance left intact on che profile by the actual etp rounding practice. The aatching radius on the wheel of euch fp point on the piston 1 gave the innermost loading potht. “When the locations of simultaneous contacts were considered as loading site . lost. The number of loading points turned Gut to be 19. As ‘tyo or the orderly increnents were . ® chee of ete consis size sive he seme tie, say dearsond the 8 loadtig stations, JJortes dy uate voraesor' © esavorpr Ghous oat Aaa sBuyaeyanp ane sonten Tena9e oy a10yx “porsTT exe aseaueD Jo Jypea TesTaeToRNT My eosoyquexed UT cee onde za" a z6Lt-s | att 78° (a89) cera) [(9r6z0) 14 ‘ ae eray | ever out yoy | srect o oe Teese) (0°99) é zay—| erect 6rsy | roar | se - + | (eros) | (27921) - Soon _ |estoet) or1s *| Ov9zt | cet ay Tat ee a “y7OBT 9° ce" (5708) a. : 9°05 «| C°9LT “Eat sly eee a cz" » * | @sroary sror | cr9ct | oor |] over | erect ° ° troy | weer | oer + |] seoy | eteer is | zee zt lee a s*8y 9°Let ts" 99 8°6LT y= . Cees Wqsooy| q yao} ,s D Wre0r | > ysoor is a weer | a ysoor | 4s vwou ,s. fone wt awyed 00 718 uo "y qwyod uO 1H uo y , | Whoa uoyI83S Buzpeor Burpe0r surpeor | | surpect G8) vorasy Jo ourT 242 BueTyuoTITs0a PaZTTEAzON 837 PUR (UOsUEd),A (TODUA)E I9>EIUCD Jo UOTaTeOE TeIPEE ‘saayos BuyPeOT US Tava- . . . ° 95 positions on the wheel (R), their matching radid on.the pinion (RY) See their normalized position (S') on the line of action. q - st = . “(5-17) 7 / where .$ is the position of the load on the line of action: Py is the base pitch The sign of ‘S' corresponds to the pinion being the driver, which reflects the condition on the engine. On the test this direction id “reversed, the load 4s applied through the wheel, which allowed the use of smaller deadvetghts. - . Se The Adentification of the teeth is shown on Fig. (4.8), p81. : 5.4 Stress Heasurement : «5.4.1 The Photoelastic Method” a he model ts made of « “temporarily birefringent” natertal, which to a property of soet transparent plastics. When a.potnt sn ach 4 mterial is in’ the state of two disenotonal stress and to tilylinaced by a pencil of polarized Light, the latter separates into tuo wive fronts. + “Bach fore Lo polarized in one of the two principal stress directions, and ptopagates with different velocities. whence the nané of birefringence. The change of velocity in each front 4s propértional to the corresponding princifal stress.’"The absolute velocities cannot be seasuied tn ovdianty photcolastictiy. However ao the fronts szaverse the model they accumulate 4 relative retardation, consequently they emerge with a phase shift. in the polariscope this phase shift 1s made ~ vistble, The interference phenomenon 1s utllized, resulting tn’ dark and bright fringes. With white Light tllumination one value. of relative f a ‘. . ie . 96 retardation torresponds to different phase dhifts for each wave lengths os A given str difference extincts ¢ partteular vive length but transmits the complétentary colours, resulting in a colourful * ‘ “cchrontcte™ fringe patteri. Such fringe is the Jocus of points where -the"ALeerence between, the principal stresses Je constant . arog = wE/B > (5.18) ref [10], chapter 7 where N' is the fringe: order B fs the thickness of the model £ 49 the terial fringe value” [KPa-u] (pei-in) which ts determined by calibration The isdchromatic fringes are clearly visible when the 1ighe extinction associated with the ortentation of the principal stresses 1s suppressed, which is achfeved by tnterposing the so called “Quarter wave + The plates.” In this mode the instrument is a’ “Circular. Polariscop plates are optteally removed in che “Plane Polartscope". ‘These ssttings of the tseramene are explained tn deta to cht. [10], chapters 6 & 7. In the Plane Polariacope additional fringes appear, the 40 called tsceLintes". Such fringe 1a the locus of potnts where the direction of ove sor of principal stresses cotncides with the anguléf t1Lt of the instrument. This position is read off a dial, thus the orientation of the principle stresses may be determined. . ‘The tranforaation, of the polariscope fron one mode to the other 4s simply rotating’ 45* the quarter wave plates by the neans of a lever. Sob Calibration Tks objective ts to relgte the fringe orders observed to the pee eee a: a c- : : on @ suitably loaded specimen for vhich the stresses aay be rellsbly ksiown, In a tensile sirtp loaded in unextal tension the stresses are — 4 -=— 3 . (5.19) where Wis the normal load - B Ss the thickness of the strip t te te wideh, substituting (5.19) into (5.18) gives the mterial fringe value: w i ee < (5.20) wr As the pinten and the wheel were cut from different sheets, one tensile strip vas made from each. The specimens were subjected to ‘imeremental loading, the fringe order N' was plotted ve the load. the “best fit" straight Line was determined using the method of,"Ledst Squares", An “teproved" i! was then calculated corresponding to a 445 N (100 Lbs) load. the foltoving material fringe values were found, using Eq. (5.20): . f= 7.10 kPatm (40.95 psi-in) for the pinion £ 6.95 kPa~m (40.07 psi-in) for the wheel. 5.4.3 Determination of Stresses ‘The fringe orders along the fillets vere measured. As the normal stress on a fri surface is zero, these fringes give directly the tangential stiesses! Wnige Light 4 2uotnarton vis used, vote allows to recognize the integer fringe orders with e orders were determined using Tardi compensation, which is fully explained in ref. [10], Chapter 9. Boag . c f . The fractional fringe. . frye oo p 98 The accuracy of the location of the measurenent is Just as « taportant aa-it'ts for the fringe orders. Therefore a travelling Celeateroscope vas used quipped vith. stage sietoneters. The angular pootttors were found using 4 protractor reticle. "me fringe orders vere recirded at 10° positional Lherenents, the maxima vere determined to the nearest integer degree. The fringe . easuresents vere aubstisuted into 2. (5-18) thus the aodet ‘Ryesses giving the prétotype stresses. : For the Multiple Engagement tests two teflon shims were placed ++ between ‘each pair of contacting teeth. Thus friction was reduced to us0.1. Tn the Single Palr Eagagenent rune the shine sebved an additonal purpose. By fitting with the tefloas only che pate to be loaded, the unwanted contacts were removed. 5.5 The seatch for finéing a viable method lead to a tmprovel patameteri the "Relative Pair Deflection” "bef Lections of single palcs were coisidered. Single contact vas contact of the pair, the deflection of which was sought. At a snall imdedat load’ of 1-18 (.25 tbs) only the shinsed pair was engaged, the adjacent contacts vere separated. Ai the load was increased by saall increnests the shianed pair has eflected and the gap’ fe the adjacent pair’ gradually closed. F 2 + The closing of the” gap was determined optically. Even a very slight contact {s easily recognized gue to the comparatively high99 Hertzian strei e+g- Due to a Lead of 0.58 (0-1 ibs) a pressure gf 598 kPa (7B pot) tay develop at the contact. .THe brightness of the: corresponding fringe order of 0.50 sakes these slight contacts highly vistble ageinst the dark background 1a the polariscope. Firot the initial clearances were deternined, those created by tip modification and the siller oned due to sanufacturing errors. This was achieved y utlog so-called “onton papers", the flnest of stich measures a mere 0.005 mm (.0002"). 7 : x : For setting deflections in fine increnents the thickness of the Precision shim oy ranged from 0,025 mm (.001") to 0.254 mm (,010" The magnitude of the deflection was considered to be the width of the gap being closed, which vas equal to, tlle thickness of the shim plus - - a the initial clearancé @stablished previously by the ~t paper. ‘The corresponding load was taken as the differel betyeen the load at which thet second cgntact reappeared minus the initial load of Wg 11'W (25 Ibs), At each loading polnt 6 to 8 deflectiok incresents were measured and recorded with the corresyonding loads. This method gave the relative deflection” between the shined and observed contacts. As both’ contacts Ife on the Line of action the Loads and the deflections are all-in the required direction. No artificial reference datum 1s inosed, ‘the motions reflect the true action of the teeth. . , _ The measurenengs are sufficiently sensitive, the resolution of the loading is 141 N (.25 dbs) or 1% of the “single pair load level” x Wa = 113 8 (25.5 Ibeé ‘i | ' |, CRAPTER SIX 100 ALTERNATIVE ANALYTICAL METHODS ‘The objectives of this experimental investigation will be Fi . S analysed using other techniques also, which are described in the following. The gunerical results,will be compared in Chapter 7. 6.1 Application of ‘tha Kelly-Pedersen Formuhs (20) ’ This vas ¢hosén as a sample from the family of projection type For conventence Bq. (2-12) is repeated here: - ow [ (: Es stn 8 a) 5 — jl +26) — - 20 a [Te ae * Get ‘The constiuetion for this expression is based on the Lewis | a ayout, where the key step {s to find the point of tangency + between + o the parabola and the fillet radius as. in Fig. (2.1). The graphical, , ~, solution is cumbersome and inaccurate, therefore it was replaced by “| catculacton, The’ onenctdeure ts shown tn Fig (2.1)+ : 1 i : . my one ye con x6.) "then ah! + rg sta YL, (6.2) fl : © * Gg = te gs 11) cos YL ‘also! = — cot yy #. (6.3) > 7 . 2 in WL which may be reduced to: * 2 + xg cos yp - 2h! sin yy,” a Fj (6.4) : sin’ yy +1 a solving for yy locates point A. This may be done by retek and |’ error, using a prograsable hand calculator. The other paraneters in os Bq. (2412) may be read] off Pig. (25). : © ves to i 5 for the second tere using #40 (2.13):\ 7 : ig ’ ' 101 . : yee TYE + 25° ~ y4/2 = 25° * /2 (6.5) ue encode (6.6) . a, = Oh" + rg fin yep) con A “ (6.7) ’ ' 2 = Ue’ xp)? + Op.- 90 + Fe atm ee)Tt oo . ; (6.8) Bere -3k \ _ 6.9) 6.2 Application of the At chi Formula . _, The rational structure of the formula lends itself to include the ° galculation of the compressive maxima. Eq. (2.26) is repeated here in a slightly modified form to facilitate this latter application. The + nomenclature is shown in Fig. (2.12). - : 5 aa 2 er oe =[t + 0.8 =) ue) {#0-660p + -40 Cop? + 36 r2)4] + 1615 Copy * yd} : 2 (6.10) e where the 4ve sign applies to the maximum tensile, the -ve sign to the maximum compregsive fillet stresses. cs The “direct component” of the nominal Tadidl stress is; Wosind i 6-11) : a org =~ be The “bending component” for the sam 6 Way sind : oy t= 2S 2 be? . “oN The term in the bracket of Eq. (6.10) is the nominal stress due ee ae ee “eppostce forthe tensile apd compressive filets. agg As compre for both sides. However as the radial component of the load acts offi an fd ease the centerline of the tooth, opp {s compressive for the tensile fillet ‘and causes tension on the’ compressive fillet. . This formula may be used also to account for the effects of \ Superposition of this traction modifies the loading and the formula may frictional forces acting tangentially at the contacts. The be applied without change. Results may justify the extension of the formula. It has yorked for this pair of gedrs, but for full acceptance some further evidence swould be needed. . 6.3 Using the Cornell~Wegtervelt te Pro} A computer progran based on ‘the dynanie analysts tn ref's.(9] and [8] wae weiteen and vas nade available for testyg. By running it at ¢ very slos speed input (2 rpa) an essentially Sgatte. analysis vas obtained. ‘ : i ‘This progtam first optdmizes tooth proportions and calculates pate exapltonce, Thos it deteraises she haring of the dynaste tons * between a péssible maxinim of (4) pairs of teeth. Finally it coupytes the tenslon stress peaks produced by the previously obtained load * ‘Fractions individually on each of the respective teeth. Ther effects of speed, composite tooth etror, profile odification, backlash allovance, thermal expansion, manufacturing tolerances, center distance variation aré, 1 taken into consideration. For the computation of compliance, depending upon the ratio of face width over are toath thickness, optionally plane stress or plane strain calculation may be selected. As an additional option this . operation may be bypassed and the user may input its ovn compliance103 constants. The optimization process in the program may be nodified by fan input allowing an arbitrary increase in are tooth thickness “for one gear at the expense of the mating one; Im the output the gear data are listed in detail. Compliance properties are given tn the form ‘of constants fitting Eq.’ (2.45). Hertzian and maxinua tension stres are printed out. Tooth load sharfag, contact and peak root stresses ave also available in graphical form as.an option. Abample of load sharing curves 4s shovn in Fig. (6.1). The load ratio is the fraction of che individual beanch loads with respect to the total load. The horizontal axis extending from 0 to 1.00 represents one base pitch. The simultaneous \contacts are plotted as they occur giving @ cpupact presentation of the full length of the line of action. The vertiegl Lines were added onto the: computer generated graphs by hand. ‘They represent the experimental loading stations, with the simultaneously engaged loading positions identified by the typing. Thus the experiment@lly determined loads may be compared with those generated by the computer. 6.4 An Application of Finite Elements S 6.4.1 Element -Network Grids representing the photoelastic models were constructed using triangular elements. Thd/ vhree central teeth were fully modelled, the side teeth vere truncated leaving only the f1llets, as in Fig's. (6.2 wthru 6.6). The thickness of the elenents was unity except at the polycarbonate reinforcing plates show by the heavy lines, where it was 4 timed that value. A typical side of an element in a'general area ofwiwoowa OLAVNAG ~ ONTUVHS avOT "T"9 ‘OT . qeruewpasdxa x wer8oad opueukg 2 (GSWYON)’ NOTLOV 40 ANI7 SNOW AONVISTO 7 $or1 06" s. 03 sr. oc weinet zoe te v0 Number of eleaents: 882 w of prinary nodes: 555 FIG. 6.2 FINITE ELEMENT MODEL ~ WHEELsists Gast. FIG. 6,5 FINITE ELEMENT MODEL ~ PINION 108ames KS N> LOOKSi I | us the tooth was approxtzately 19% of tooth depth. The same for a Fillet element was 1.7% and for those carrying the distributed load was 0.7%. CS ——_=C nodes. For the pinion 870 eleaeats and 537 primary nodes were used. 6.42 The Program . A versatile two dimensional Finite Element program was used for the task. It has dyeeping capabilities such as handling plate, and axieyenetric problens under external, centrtfugel and thersal loading, with applying elastic, plastic and/or creep analysis. For this study only the elastic analysis of a plate under inplane external loads was needed. 7 Quadratic assumed displacenent fields are enployed. The corner nodes are specified by the user, additional midstde nodes sre generated ncernally. This allows ace particulars ta specifying boundary condttions and returas finer detail in the output. Plane stress or plane strain analysis aay be chécen. tn « standard output ‘external and” 4 o |isplacements at all nodes, coordinate stresses at the primary nodes and at the center of gravity of the elements are listed. reaction fore Optionally the principal stresses aay. be included. ‘The program may be linked up with a graphics prograa, which Aieplays the aetwork. This was instrumental in debugging the tnput and for printing Fig 19 treated sintlarily” (6-2 thru 6.6). The output giving samples of the deflected shape for the pinton, Fig. (6-7) and of the teostress Lines for principal stresses Fig's. (6-8 & 6.9). : . 644.3 Boundary Conditions Ae the photoelastic models were held in space by at pair of large“absolute” deflection: Isotropic point | | | % wpelative” deflection: yf, ~ yh « FIG. 6.7 DEFLECTED SHAPE - PINTON qu<< FIG, 6.8 ISOSTRESS LINES, MAJOR PRINCIPAL STRESSES = PINION : ° 12”C 4 +A, ' 7 7 6.9 ISOSTRESS LINES, MINOR PRINCIPAL STRESSES - PINIOS . s| 4 atane pins, the finite element models were also fixed at nodes representing these points. 6.4.4 Loading . It was intended. to reproduce the Hertzian displacenents as closely as feasible. Therefore the grid at the sites for the contacts were prepared for that purpose. ue co the on Linear nature of the Aisplacenent ag-in Eq. (2.44), incremental loading was planned. The spreadiig of the contact width as to reach. precisely a new pair of primary nodes at each increnent. ; aS The width of the rectangle ‘at the contact of two parallel cylinders pressed agatast edch other 1s proportional with the square root of the load as in q.(2.29).Hence 4f the subsequent load levels are WW, and Me . then the subsequent widths gf the contact Yectangles will be: by YT bg > and, be = /T Og” 6.13) Distributed loading is represented by so called “work-equivalent™ or “consistent” Loads acting at the nodes. These may be calculated for 4 parabolically distributed line load, spread over the edge of an element from: L . Wy = — L4py/+ 2p2 ~ pa) 5 : aa . : “hee . 7 Wg = — (2p, + 16p2 + 23) Bray 30+ , 3 : ‘ L Wy = — (py + 2p2 + 493) 30 : wyere Wj represent consistent nodal loadsPi, represent the intensities of the distributed line’ loading at the nodes 7 odes 1 and 3 are prisary, node 2 ts a aldside aods. # Eq'a. (6.14) are derived in Agpendix (1), which.vas based on * ref's [11] and [16]. The distances between the nodes were determined using the File set by Eq's. (6.13). * 5 An ellipete load distribution of a contact extending over five primary aodes-is shown in Fig."(6+10). The load intensities were obtained from Eq. (2-28) repeated here for conventence: P™ Paax (1 ~ (x/d)? It . vere pags the vertical half axis of the ellipse, represents the aaximua load intensity. For normalizazion purposes the total load vas pee ete eer ee half ellipse, » Awd Puax bo 1 i @ (615) Then using q's. (2-28),(6-15) the pressure at each node was determined. Next the fractions of the consistent loads over each element were calculated eaploying q's. (6-14)., Finally the fractions of nodal loads were assembled as show in Fig- (6-10) for three load levels. The results are listed in Table (6.1).6 é : , 2 +k are primary, are uidside nodes. + For element " “For elenent ” irc q PL” PE at G27 Pe ‘ Ps * Pp sea a : G4ving nodal loads for each elenent using Eq's. (6.14): Ma 0 Mon Yo , Ma ns . Then the assembled nodal loads ar: We My =H é Wy ¥en rn S . Me Maa Mig 7 Mr We Mop a We 72 My by aymetey . ~ fro, 6.10 vomststext age C33) . 2 :16 Note Table 6-1 Non=dt Contact extending to: Node # 3 1 +001 1-2 , 031 ° 4 2 003] .027 2 2-3 +061 | .072 . 3 +032 | .080 | .072 ne +338} 276 | .233 4 +260:] 160} «128 ‘ “aes +338 | .276 | .233 5 +032} .080 } 072 5-6 +061 | .072 “6 003 } \ .027 6-7 oo +031 7 +001 Total | 1-000 } 1.000 1 Mideide nodes are indicated by the adjacent corner node numbers.r ug To determine the load which would extend the width of the contact to a predetermined distance, #q."(2-29) vas used., This is rearranged here: aH oye, It ew yt — P(E) ‘| rtm & | ode where: 2 . a(1-v?) 5 / (6.16) we “Oe : ‘ rite ” e- (617) rte : hence: 7 vn ’ We (6.18) KD! The face width 8 of the gears and constant Ky do not vary with + the position of the ¢odtact. The equivalent radius of curvature “p" was calculated for each igading point. ‘The required half width of the contact determined the magnitude of the total Joad over a given site. ‘The load fractions were then calculated and the group was orientated nto the noraal direction at the central primary node of the’ contact. 6.4.5 Determination of Deflections ~ Using Novel Reference Data The nin objective 6f his investigation Is to deterine the flexibility of these gear teeth. As everything deflects, the choice of a suitable datum such that ” 1 relevant diaplacenents but nothing else would be included is of paragount taportance. The dynamic progran considers tooth deflections only, but includes the effects of the elastic nature of the foundation region. 7 :e ug i ‘This datum ts determined ty maximizing deflections. A reasonably specific point may be found by entering photoelastic considerations. Ih the photographs of Fig. (7.11) 33 certain black areas appear in the : foundation zegion repeatedly. These spots nove with the position of the Lads Inside these areas there 1s an “Isotropic Point” which 8 untque e for a given loading. Its location may be determined by using the Asoclinics mentioned in Paragraph 5.4.1. i ! | i i * | The black colour indicates that 1 ~ og = 0. Therefore such 3 point is tna state of pure hydrostatic stri%s. The shear strain here 4s zero, hence no local rotation, only translation can take place. It represents the distortion free displacement of the rim. This property makes the isotropic point a suitable datud. Frow here on the displacement at a contact in reference to the isotropic point will be labelled as an “absolute” deflection, vhich excludes the , rigid. body displacements of the rim. These deflections are coaparable - with those generated by the dynaaic program In contrs it, the displacement at a contact telative to the / displacement at the cfatact site on the adjacent tooth is defined by the paraneter of “relativd deflection ‘The equivalents of the latter were measured experinentally. The Locations of the Ssotrople potate, each corredponifeg to a / ee eee | mode was acsigond £0 such auch pelut ha the Ethive aleenne sodet— tins / combined. Tall cases, first the wf dsdplacensate were coupseed andwd : 120 then, these were resolved’ into components along the line of action, and those perPindigular co At. The former gave the deflections, the latter represent slidin The versatility of the finite element sethod yields a comnon ground to compare the results from the dynamic’ progran with those from the physical model. . At each loading site the loads were applied in three increments, ‘Thus, tneluding zero Load, four points becane a\ LLable to expose the suspected non-linearity. 6.4.6 Determination of Fillet Stre Tia E4te azznouse de co single tooth looding vere exercced from the output of the third 1oad level computer runt. Thea Hey were pearly adjusted to che sundarized single fagigemmtt procotree load of 15:8 te 543 bef The poe was plotted and smooth Lines were drayn through the'data points, which constitute the final results. cen aee fete nin eee as eerie ope ee epee eee er ees Four group loadings vere applied to each model. :. XN , - ; CHAPTER SEVEN 121 RESULTS & CORRELATION WITH ALTERNATIVE METHODS, ‘The data are expressed in‘terms of the prototype. The reader may. relate to these results better than, for example, to the small stresses An the plastic. The scaling factors were described. in Paragraph 5.2. ‘The Fillet stresses are plotted vs angular position. ‘The sero polar posttton is defined by the Line connecting the centers of the _-tenstie andycompressive £L1let radii for ¢ tooth. Positive angles are C—O (741) the fillet circle blends into the shore segment of the root cttcles They are at slightly different angles for the plnton and for the wheel. . ' The posttton of the load is defined as “Normalized Position Along dia tine of ton’, rafervod to as! and ip obtetned by élvtdtoe ite directed distance’ from the pitch point over the ba pigchs Thus distances between contacts are integer nusbers and the full lehigth of the engagesent gives the onésct ratid. Such normalization provides conson grout for pinion amd wheel dats. “Best fie" Lines were drawn “by eye” aided by “French curves” though numerical curve fitedng could have been applied for this purpose. Isl Stresses — Single Tooth Pair Engagenent Sanples of fillet stress distribution vs fillet angles due to (TA), (7-2) and loads acting on dndividual teeth, are shown in Pig’ (7.3) for the wheel and in Fig'ss (744), (745) and (746) for the pinion. The tensile end the compressive peak stresses and thet angular location are plotted vs normalized position along the Line of action in Pig's. (747) thru (7-10).: * a2 Fier Serene at oo on ‘ 30 wo “ . 200 « 10 6 2 ~~ a0 Lecation of Cansile peak om scaiitece Coat tn Sale ste a “0 + | we ant : y 2 (a0 CEeeeunSase =e ‘ + Epprinneas Rae y Aidetesaeht a Parner Toeens 7 2 Finich tecence J, Bo A ey 7 peanmure ite ae atte FIG. 7.1 FILLET STRESSES VS. ANGULAR POSITION, SINGLE ENGAGEMENT — WHEEL, $= -1.00: 123 Fillet sirens iter Sere waa kat 400 a cal 300 40 coors . 200 a fitter cp 2 100 y 10 ° + ae ° 100 Vocation of ‘tenuile jean tooth, -20 load de ae s'= 1.00 200 Pant +t - 300 a c $0 400 — ° 2 40 son) + Raperteancad OCT ae e Abdantara 3 Kelley-redersan 2 Maite Elasence runes Boe P peetvare side FIG. 7.2. FILLET STRESSES VS. ANGULAR POSITION, SINGLE ENGAGEMENT - WHEEL, S'= 0124 rates se rani ie ‘ot : ar i a : PS see | y % © a “é Et » rate op Location of coupras te Tr on adjacest tooth, hes o * reas ioe » ae ~ = . fe : at ° ° o 7 wo” 100 ® . -200 : 5 . “0 =100 * ’ 30 #00, tr . ee wo ae ° ry 2 Repater Porseson= F oe eG F prannure ad Lites side FIG. 7.3 FILLET STRESSES VS. ANGULAR POSITION, SINGLE ENGAGEMENT ~ WHEEL, 5'= 1.33ae oo rattes scien : sane sea ‘wee wae i 7 » 4s 300 ‘ : ” . Q 30 200 100 100 40 200 “30 400 x -0 ° ee Mest fie" Lise ‘tooth: Aidectersuedt Kalley-Pederse Finite tlenene Faatetes as.8 ke rere i) : P prasmure side “FIG, 7.4 FILLET STRESSES VS. ANGULAR POSITION, SINGLE ENGAGEMENT ~ PINION, S'= -1.00 1s¢ ’ nO 126 rau erate Fsane saree, to 300 . ruulee cp 4 . 300 400 (eG. 7.3. Pitter studs vs. anouun rosrtiay, SINGLE ENGAGEMENT - PINION, S'= 0: * a me . . 7 rates sera raat fee 7 we : te . 400 . . 7 4 . ” ‘ 20 i . 3 : . 0 po = a ‘ * feces ) eee oe BD : Fic. 7.6, FILLET StlessEs VS. ANGULAR PostTrox, " vo" 0") SINGLE ENGAGEMENT — PINTON, S'=.1.33 \128 Fem : aS. ‘laa LNaHGOVONE ATONIS (NOIIOV 40 BNIT HHL ONOTV NOTLISOd GEZITVHUON) 1S SA NOTLYOO1 YVINONY UIAHL ANV SUSSHYLS ATISNAL AVE L*L “91d oc” sooatop oy orsuy? Srrate og 22TTa cot eer Ceaioccamem veiled a a re a Seat oun ance™ ecoameaces oO eer ty ts t6~ oz oc ov uaszepag-foT 19% Typne321~ep Ty os, squewoTs OITUTS Teauswprodsg 990225 29TTTa . s9933$ 3T129 ‘THIHA ~ LNENEOVONE STONIS +S “SA NOLLVOO1 UVINONY UIaUL ONY SUSSHIS AAISSH¥aHOD WEA O°L “OIA fos or soesten ete ety vo Syme oy foe atta eeteeT a a a ee a eet | oot wet Steet tga oor ove oz oor- - Twonezoy-wpy + swuewetg 937uFao : Tenemedeg 8 ° ° aad raH sbeaas 391TH 860325 2°TTTa130 ” < NOINId - INSRADVON STONIS + 4S !SA NOILVOOT UVINONV HIAHL GRY SASSHULS FTISNAL Wad 6°L “OTA y saax8ap A oz oc corr cet ze" on Ye sys- tL 00" 9 S21__oor ues ee Ota ty us 6 5 or EB 4 oot oz : pore ot yg | tnbas oc 002 ot uvexopad-AotToy x oot yyonexey-vpry + oc s2uowora oayuTd - v Tosuseyred 8 : oo 74 : ta soaaas 291tTd ~N seoaas a9rTTd 4 orev os 398Itra ¥ ro131 ~ Nomfita - inawiovoxa gious .S ‘SA NOILYOOT UVINORY UISHL GNY SaSSmMLS BALSSHVaNOD NVEd OTL ‘OTE saaa8ap & atauy 2eTTTd oc-| oz- : ‘FuOFeIaL-EP HY or- squamaTg 927Upd yeqwoupa9dxa ° 154 : 2aK 960335 ITTTa . 859225 JTTTd132 As say be expected, the tensile anf the compressive maxima tncrpase in magnitude a3 the load aoves tovard the tip. The reversal of this trend for tensile stresses at fow dedendum loading [19] may be observed for the pinton, but for the whesl, only the slope of the curve 1s reduced. Similarily as the'‘load moves toward the tip the location of the peak climbs toward the flank," with some rever'al at the dedendua. On figures, vhich display the peaks, the stresses obtained fron the photoelastic experiment are shown in full line. Those found on finite element analysis are connected with dashed line The stresses determined using the closed form equations appear as unattached points.” The locations, vhere the plotted stresses would superiapose on\ thq peaks at che adjacent teeth as “direct czons effect of stiesses™ on an assumed multiple engagenent, are indicated. +The finite element, photoelastic resules and those obtained by using the Aida~Terauchi formula in general correlate well. The Kelley and Pedersen stresses appear to be under estimates, however their angular podition fit the data neatly. The photoelastic peak stresses seen to fall behind the finite element aaxina as the load approachés the tips. The maximum deviations for the tensile and compressive peaks are respectively 6% and 9% for-the vheel, 11% and 16% for the pinion. Friction at the contact may be partially responsible for these discrepancies. Tats subject 1s discussed in more detail in Para. 9.1. Photograph Fig. (7+11) displays the {sochronatic fringe patter in black and whtte for a pair of teeth in single cagagenent. Six loading cases are shown. The cr cent ike fringes indicate the fille stresses, the isotropic polnts and che contacts are clearly recognisable. 4POOR COPY COPIE DE QUALITEE INFERIEURE134 y 7.2 Stresses - Multiple Tooth Pair Engagement Sample plots of fillet stresses vs angular position are show in Fig's. (7-12 theu 7.17). 8'goq Fefers to the load on the central tooth “ + The insert on the gYaphs illustrates the Location of the loads and the convention for the fillet angles. ‘The experimentally determined loads and the radii of contacts are also shown in the boxes. » ‘The Fillets are tdentified by two letters, The first of these stands for the tooth to which the fillet belongs, the index letter P or L refers to pressure or leeysides respectively. The finite elenent results are represented ogly by the peak values. These points are left unattached, they ate {dentified where they may be ambiguous ‘The highest stresses appear when only two pairs of teeth are engaged simultaneously. The magnitudes of the compressive stresses in general exceed those of the tensile stresses. The cross effect of the Loads on the adjacent teeth appreciably modifies the. stress distribution. This ts most apparent at the fillet bottom between two loaded tegth, where the tensile field die to one load and che compressive stresses oving to-the-other one are mutually reduced. The dominant Influence on each péak stress is still the load on the tooth to which the fillet belongs, the cross effect here ts small. ‘The aaxina vere plotted vs position along the Line of action in Fig's. (7.18) and (7.19). The experinental, finite elesent and dynamic program results and those which were calculated from "Stress Adaittances” (Para.8.4), all correlate well. The latter is based on data obtained from single engagement stress and deflection measurenents, and gave ‘an avgrage deviation of 6%. This value aay be used as a rough .: . . 135 7 Falter sereis— 7 eet \ 300 0 runet cy : 200 100 100 +300 0 2 4 6 atta 0 40 20 0 , Angular Position = y ~ dear! Min otheras Exparinentad FIG, 7.12 FILLET STRESSES VS. ANGULAR POSITION, “" MULTIPLE ENGAGEMENT - WHEEL, S'_.= 0alte serene cs 300 . Ne, 5 : jo = : < 0 09 om 0 @ wi 0 0, 20 0 snguiat setesen = y = dngreee . 8 Matte Heneces (peas ooly! 1 Teen "Serene Adsieeences” Tar othere! Eertaestat CE FIG. 7.13 FILLET «STRESSES VS. ANGULAR POSITION, MULTIPLE ENGAGEMENT ~ WHEEL, Soo .33 136137 anes sera at Osa ec ee te 00 =o 00 etna eco dogulay Position = 7 ~ dagreee FIG, 7.14 FILLET STRESSES VS. ANGULAR POSITION, MULTIPLE ENGAGENENT - WHEEL, S'oo= .82138 ister serees ates Serene ca ~ ir . Po wee ata) wo : ‘Sauter Foesion = y ~dearecn 2 pransure aude Laas tide FIG. 7.15 FILLET STRESSES VS. ANGULAR POSTTION, MULTIPLE ENGAGEMENT ~ PINION, S!o_= 0. © rather Seren raner serene 1 -200 -300 +00 ~ 0 VT 25 a6 60 FIG. 7.16 FILLET STRESSES VS. ANGULAR-POSITION, MOTTRLE ROACH ~ PRTON, gh 33 139