STRENGTH OF MATERIALS Part II Advanced Theory and Problems By S. TIMOSHENKO Professor of Theoretical and Engineering Mechanics ‘Stanford University D. VAN NOSTRAND COMPANY, Inc. TORONTO NEW YORK LONDONNEW YORK D. Van Nostrand Company, Inc., 250 Fourth Avenue, New York 3 TORONTO D. Van Nostrand Company, (Canada), Ltd., 228 Bloor Street, Toronto 8 LONDON Macmillan & Company, Ltd., St. Martin's Street, London, W.C. 2 Copyright, 1930, 1940 by D. VAN NOSTRAND COMPANY, Ine. All Rights Reserved This book, or any parts thereof, may not be reproduced in any form without written per- mission from the author and the publishers. First Published, May 1930 Reprinted, Marck 1932, January 1986 February 1988 Second Edition, June 1940 Reprinted, October 1941, Fuly 1942 Fanuory 1944, August 1944, May 1945 May 1946, February 1947, August 1947 PRINTED IN TUE UNITED STATES OF AMERICA BY LANCASTER PRESS, INC., LANCASTER, PA.PREFACE TO THE SECOND EDITION In the preparation of the new edition of this volume, the general character of the book has remained unchanged; the only effort being to make it more complete and up-to-date by including new theoretical and experimental material repre- senting recent developments in the fields of stress analysis and experimental investigation of mechanical properties of struc- tural materials. The most important additions to the first edition include: 1. A more complete discussion of problems dealing with bending, compression, and torsion of slender and thin-walled structures. This kind of structure finds at present a wide application in airplane constructions, and it was considered. desirable to include in the new edition more problems from that field. 2. A chapter on plastic defor mations dealing with bending and torsion of beams and shafts beyond the elastic limit and also with plastic flow of material in thick-walled cylinders subjected to high internal pressures. 3- A considerable amount of new material of an experi- mental character pertaining to the behavior of structural materials at high temperatures and to the fatigue of metals under reversal of stresses, especially in those cases where fatigue is combined with high stress concentration. 4. Important additions to be found in the portion of the book dealing with beams on elastic foundations; in the chap- ters on the theory of curved bars and theory of plates and shells; and in the chapter on stress concentration, in which some recent results of photoelastic tests have been included. Since the appearance of the first edition of this book, the author’s three volumes of a more advanced character, “Theory of Elasticity,” “Theory of Elastic Stability,” and “Theory of Plates and Shells” have been published. Reference to these ifiv PREFACE TO THE SECOND EDITION books are made in various places in this volume, especially in those cases where only final results are given without a complete mathematical derivation. It is hoped that with the additions mentioned above the book will give an up-to-date presentation of the subject of strength of materials which may be useful both to graduate students interested in engineering mechanics and to design engineers dealing with complicated problems of stress analysis. SrepHen P, TimosHENKO Pato Auto, Cautrornta June 12, 1942PREFACE TO THE FIRST EDITION The second volume of Tue StreNctH oF Mareriats is written principally for advanced students, research engineers, and designers. The writer has endeavored to prepare a book which contains the new developments that are of practical importance in the fields of strength of materials and theory of elasticity. Complete derivations of problems of practical interest are given in most cases. In only a comparatively few cases of the more complicated problems, for which solutions cannot be derived without going beyond the limit of the usual standard in engineering mathematics, the final results only are given. In such cases, the practical applications of the results are discussed, and, at the same time, references are given to the literature in which the complete derivation of the solution can be found. In the first chapter, more complicated problems of bending of prismatical bars are considered. The important problems of bending of bars on an elastic foundation are discussed in detail and applications of the theory in investigating stresses in rails and stresses in tubes are given. The application of trigonometric series in investigating problems of bending is also discussed, and important approximate formulas for combined direct and transverse loading are derived. In the second chapter, the theory of curved bars is de- veloped in detail. The application of this theory to machine design is illustrated by an analysis of the stresses, for instance, in hooks, fly wheels, links of chains, piston rings, and curved pipes. The third chapter contains the theory of bending of plates. The cases of deflection of plates to a cylindrical shape and the symmetrical bending of circular plates are discussed in detail and practical applications are given. Some data regarding the bending of rectangular plates under uniform load are also given. vvi PREFACE TO THE FIRST EDITION In the fourth chapter are discussed problems of stress distribution in parts having the form of a generated body and symmetrically loaded. These problems are especially important for designers of vessels submitted to internal pressure and of rotating machinery. Tensile and bending stresses in thin-walled vessels, stresses in thick-walled cylinders, shrink-fit stresses, and also dynamic stresses produced in rotors and rotating discs by inertia forces and the stresses due to non-uniform heating are given attention. The fifth chapter contains the theory of sidewise buckling of compressed members and thin plates due to elastic in- stability. These problems are of utmost importance in many modern structures where the cross sectional dimensions are being reduced to a minimum due to the use of stronger ma- terials and the desire to decrease weight. In many cases, failure of an engineering structure is to be attributed to elastic instability and not to lack of strength on the part of the material. In the sixth chapter, the irregularities in stress distribution produced by sharp variations in cross sections of bars caused by holes and grooves are considered, and the practical sig- nificance of stress concentration is discussed. The photo- elastic method, which has proved very useful in investigating stress concentration, is also described. The membrane anal- ogy in torsional problems and its application in investigating stress concentration at reentrant corners, as in rolled sections and in tubular sections, is explained. Circular shafts of variable diameter are also discussed, and an electrical analogy is used in explaining local stresses at the fillets in such shafts. In the last chapter, the mechanical properties of materials are discussed. Attention is directed to the general principles rather than to a description of established, standardized methods of testing materials and manipulating apparatus. The results of modern investigations of the mechanical properties of single crystals and the practical significance of this information are described. Such subjects as the fatigue of metals and the strength of metals at high temperature arePREFACE TO THE FIRST EDITION vii of decided practical interest in modern machine design. These problems are treated more particularly with reference to new developments in these fields. In concluding, various strength theories are considered. The important subject of the relation of the theories to the method of establishing working stresses under various stress conditions is developed. It was mentioned that the book was written partially for teaching purposes, and that it is intended also to be used for ad- vanced courses. The writer has, in his experience, usually divided the content of the book into three courses as follows: (1) A course embodying chapters 1, 3, and § principally for ad- vanced students interested in structural engineering. (2) A course covering chapters 2, 3, 4, and6 for students whose chief interest is in machine design. (3) A course using chapter 7 as a basis and accompanied by demonstrations in the material testing laboratory. The author feels that such a course, which treats the fundamentals of mechanical proper- ties of materials and which establishes the relation between these properties and the working stresses used under various conditions in design, is of practical importance, and more attention should be given this sort of study in our engineering curricula. The author takes this opportunity of thanking his friends who have assisted him by suggestions, reading of manuscript and proofs, particularly Messrs. W. M. Coates and L. H. Donnell, teachers of mathematics and mechanics in the Engineering College of the University of Michigan, and Mr. F. L. Everett of the Department of Engineering Research of the University of Michigan. He is indebted also to Mr. F. C. Wilharm for the preparation of drawings, to Mrs. E. D. Webster for the typing of the manuscript, and to the D. Van Nostrand Company for their care in the publication of the book, S. TimosHENKo ANw Arsor, Mrcnroan May 1, 1930NOTATIONS Gz; 8y, 7z--.-Normal stresses on planes perpendicular to x, y and 2 axes. Cate .Normal stress on plane perpendicular to direction nn eesraeiete Normal stress at yield point. . Normal working stress t.. . Shearing stress Tay Tys Tex. Shearing stresses parallel to x, y and z axes on the planes perpendicular to y, z and x axes. . .. Working stress in shear . .- Total elongation, total deflection ...Unit elongation. . Unit elongations in x, y and z directions Pee feeeeeote Unit shear, weight per unit volume ..-Modulus of elasticity in tension and compression ...Modulus of elasticity in shear Poisson’s ratio ... Volume expansion ...Modulus of elasticity of volume +++... Torque Gee reeeeee eee Bending moment in a beam eerste Shearing force in a beam aee Cross sectional area ate Moments of inertia of a plane figure with respect to y and z axes ...Radii of gyration corresponding to J,, J. ...Polar moment of inertia . .. Section modulus .-. Torsional rigidity .. Length of a bar, span of a beam ..-Concentrated forces .- Temperature, thickness ixx NOTATIONS ....Strain Energy ..... Distance, arc length ....Load per unit lengthCONTENTS CHAPTER PAGE I. SpeciaL Prostems in Benvinc or Beams... . . I 1. Beams on Elastic Foundation... 2... . .. I 2. The Semi-infinite Beam on an Elastic Foundation. 12 3. Beams of Finite Length on Elastic Foundations . 15 4. Combined Direct Compression and Lateral Load . 25 5. Continuous Struts .. 1... . Peete tees Hae a be 6. Tie-Rod with Lateral Loading .. 2.2... 39 47. Representation of the Deflection Curve by a Trigo- nometrical Series... . . 4 44 8. Bending of Beams in a Principal Piane which is is not a Plane of Symmetry. Center of Twist. . . . 50 g. Effective Width of Thin Flanges .. 2... 1. 55 10, Limitations of the Method of Superposition . . . 59 Tl. Curvep Bars. 2 2. ee ee 65 11. Bending Stresses in Curved Bars... 2 1... 65 12. Particular Cases of Curved Bars... 2... 69 13. Deflection of Curved Bars. 2... eee 79 14. Arch Hinged atthe Ends... 1... 2.1, 94 15. Stresses ina Flywheel... 2... 2 2. eee 98 16. Deflection Curve for a Bar with a Circular Center Line . 2... 101 17. Deflection of Bars with a “small Initial ‘Curvature | 104, 18. The Bending of Curved Tubes... 2 2... 107 19. The Bending of a Curved Bar Out of Its Plane of Initial Curvature. 2.2... ee 12 TIL. Tur Puares anp SHELLS . 2... 7 eee 119 20. Bending of a Plate to a Cylindrical Surface . . . 119 21. Bending of a Long Uniformly Loaded Rectangular Place Hee eee ee eee Heese eee eee Hace 121 22, Deflection of Long Rectangular Plates Having a Small Initial Cylindrical Curvature... 2. - 126 23. Combination of Pure Bending in Two Perpendicular Bireceions see Hates a Hate stHe tes ars Het ita es Hae 129 24. Thermal Stresses in Plates... 2... 2 2 ee 133 25. Bending of Circular Plates Loaded Symmetrically with Respect to the Center... 1... +: 135 26. Uniformly Loaded Circular Plate... 2... 139 xiVI. CONTENTS 27. Circular Plate Loaded at the Center .. 1... 146 28. Circular Plate Concentrically Loaded... . . 149 29. Deflection of a Symmetrically Loaded Circular Plate with a Circular Hole at the Center... 154 jo. Bending of Rectangular Plates... . 2... 155 31. Thin-walled Vessels Submitted to Internal Pressure 159 32. Local Bending Stresses in Thin Vessels . . . . « 164 33- Thermal Stresses in Cylindrical Shells... . 174 34. Twisting of a Circular Ring by Couples Uniformly Distributed along Its Center Line 177 . Buckiinc or Bars, PLares ano SHELLS... . - 184 35. Lateral Buckling of Bars, Compressed within the Elastic Limit. . 2... 2. 184 36. Energy Method of Calculating Cr Roadsters HEE ae trees Hecate teteeetheb repr settat ist: 199 37- Buckling of Prismatical Bars under the Action of Uniformly Distributed Axial Forces... . . . 205 38. Buckling of Bars of Variable Cross Section. . . . 207 39. Effect of Shearing Force on Critical Load . . . . 209 40. Buckling of Latticed Struts .....--..- 211 41. Buckling of Circular Rings and Tubes under Ex- ternal Pressure. 2 2. 1. ee ee 216 42, Buckling of Rectangular Plates... 2... +e 224, 43. Buckling of Beams without Lateral Supports. . . 229 . Derormations Symmerrical Apour Axis... . . 236 44. Thick-walled Cylinder. 2... 0. ee ee 236 45. Stresses Produced by Shrink Fits... ....-. 24t 46. Rotating Disc of Uniform Thickness... . . . 245 47. Rotating Disc’of Variable Thickness... . . - 253 48. Thermal Stresses in a Long Hollow Cylinder . . . 258 TORSION 20 ee 265 49. Shafts of Non-Circular Cross Section... . . - 265 so. Membrane Analogy... . 2-0-0 ees 266 $1. Torsion of Rolled Profile Sections... .. . . - 274 2. Torsion of Thin Tubular Members . - . . + + + 278 $3» Torsion of Thin-Walled Members in which Some Cross Sections are Prevented from Warping . . 282 54. Torsional Buckling of Thin-Walled Compression Members . 2.2. / ee ee ee ee 294 55. Longitudinal Normal Stresses in Twisted Bars . . 298 36. Open Coiled Helical Spring ....--.. 304VIL. VII. IX. AUTHOR INDEX SUBJECT INDEX CONTENTS Srress CONCENTRATION «6 2 ee eee . «7. Stress Concentration in Tension or Compression = Members . 2. 2 ee ee ee 58. Stresses in a Plate with a Circular Hole... . 59. Other Cases of Stress Concentration in Tension Members teeie teeter eae 60. Stress Concentration in Torsion Members . . . 61. Circular Shaft of Variable Diameter... . . - 62. Stress Concentration in Bending .... 2... 63. The Investigation of Stress Concentration with Models . 2... ee ee ee ee ee 64. Photoelastic Method of Stress Measurements 65. Stresses at the Point of Load Application . 66. Contact Stresses in Balls and Rollers... . . . Derormartrons Bryonp Exastic Limir . . . . . 67. Pure Bending of Beams the Material of Which Does Not Follow Hooke’s Law . 2... 2.0: 68. Bending of Beams by Transverse Loads Beyond Elastic Limit. . . 2. RCE Hate eat 69. Residual Stresses Produced by Inelastic Bending - 70. Torsion Beyond the Elastic Limit... 2... 71. Plastic Deformation of Thick Cylinders under the Action of Internal Pressure... 2 2 2 1 ee Mecuanicat Properties or MATERIALS. . . . ss 72. Tension Test. 2 2 6 ee ee ee ee eee 73. Compression Test 74. Strain Hardening. . 75. Strain Hardening and Residu: 76. Types of Failures. 6 eee ee 77. Time Effect and Hysteresis . . . 78. The Fatigue of Metals... 2... eee 79. Various Factors Affecting Endurance Limit . . . 80. Fatigue and Stress Concentration. . . 2... . 81. Causes of Fatigue .. 2... ee 82. Mechanical Properties of Metals at High Tempera- abitres He ee eee ee eee eerie dee ee Pesta 83. Various Strength Theories... . 1... 005s 84. Working Stresses... 2... 0. He eioteseite xiii giz 312 313 318 324 329 335 342 346 352 355 362 371 379 383 389 396 396 403 406 412 418 423 428 436 442 459 473 482,CHAPTER I SPECIAL PROBLEMS IN BENDING OF BEAMS 1. Beams on Elastic Foundation.—Let us consider a pris- matical beam supported along its entire length by a continuous elastic foundation, such that when the beam is deflected, the intensity of the continuously distributed reaction at every section is proportional to the deflection at that section.! Under such conditions the reaction per unit length of the bar can be represented by the expression ky, in which y is the deflection and & is a constant usually called the modulus of the foundation. This constant denotes the reaction per unit length, when the deflection is equal to unity. The simple as- sumption that the continuous reaction of the foundation is proportional to the deflection is a satisfactory approximation in many practical cases. For instance, in the case of railway tracks, the solution obtained on this assumption is in good agreement with actual measurements.? In studying the de. flection curve of the beam we use the differential equation: * , £1.42 = 4, (a) in which g denotes the intensity of the load acting on the beam. 1The beam is inbedded in a material capable of exerting downward as well as upward forces on it. ? See S. Timoshenko and B. F. Langer, Trans. A. S. M. E., Vol. 54, P- 277, 1932. The theory of bending of a bar on clastic foundation has been developed by E. Winkler, Die Lehre v. d. Elastizitat u. Festigkeit, Prag, 1867, p. 182. See also A, Zimmermann, Die Berechnung des Eisenbahn-Oberbaues, Berlin, 1888. Further development of the theory will be found in the following publications: Hayashi, Theorie des Tragers auf elastischer Unterlage, Berlin, 1921; Wieghardt, Zeitschrift fiir ange- wandte Math. u. Mech., Vol. 2 (1922); K. v. Sanden and Schleicher, Beton und Eisen, 1926, Heft 5; Pasternak, Beton u. Eisen, 1926, Heft 9 and 10; W. Prager, Zeitschrift f. angewandte Math. u. Mech., Vol. 7, 1927, p. 354; M. A. Biot, Journal Appl. Mech., Vol. 4, p. 1A, 1937+ *See “Strength of Materials,” Part I, p. 137. 12 STRENGTH OF MATERIALS For an unloaded portion the only force on the beam is the continuously distributed reaction from the side of the founda- tion of intensity ky. Hence g = — ky and equation (a) becomes d* ELS = — ky. (a) Using the notation (2) the general solution of eq. (1) can be represented as follows: y = e(4 cos Bx + B sin Bx) + e-®(C cos Bx + D sin Bx). (4) This can easily be verified by substituting (4) in eq. (1). In particular cases the arbitrary constants 4, B, C, and D of the solution must be determined from the known conditions at certain points. Let us consider, as an example, the case of a single concen- trated load acting on an infinitely long beam (Fig. 1), taking the origin of coordinates at the point of application of the force, From the condition of symmetry, 7+ only that part of the beam to the right of ly fa) “the load need be considered (Fig. 1, 2). mcg In applying the general solution (4) to this case, the arbitrary constants must first be fopnd. It is reasonable to assume that Fro. 1. at points infinitely distant from the force P the deflection and the curvature are equal to zero. This condition can be fulfilled only if the constants 4 and B in eq. (4) are taken equal to zero. Hence the deflection curve for the right portion of the beam be- comes P e (o) y = e-*(C cos Bx + D sin Bx). (c) The two remaining constants of integration C and D must be found from the conditions at the origin, x = 0. At this point,PROBLEMS IN BENDING OF BEAMS 3 the deflection curve must have a horizontal tangent; therefore a) at dx =9, z=) or substituting expression (c) for y e-*=(C cos Bx + D sin Bx + C sin Bx — D cos Bx)2-0 = 0 from which c=D. Equation (c) therefore becomes y = Ce*=(cos Bx + sin Bx). (a) The consecutive derivatives of this equation are ® = — 26Ce~*? sin Bx, a = 26°Ce-**(sin Bx — cos Bx), (e) bed = 46°Ce~*= cos Bx. ” The constant C can now be determined from the fact that atx = o the shearing force for the right part of the beam (Fig. 1, 4) is equal to — (P/2). The minus sign follows from our convention for signs of shearing forces (see p. 72, Part I). Then a ady\ __P me = (FE) = ~ EB) or using eq. (f) El,4ec =F, from which 7 C= GRET,” Substituting this in eqs. (d) and (e), we obtain the following4 STRENGTH OF MATERIALS oe for the deflection and bending moment curves: e~**(cos Bx + sin Bx) ~ SET, a PB e*(cos 6x + sin Bx), (3) aay eee i ge EIS y= — pein Bx — cos Bx). (4) Both expressions (3) and (4) have, when plotted, a wave form with gradually diminishing amplitudes. The length a of these waves is given by the period of the functions cos 8 and sin Bx, i.e., = on J4EL. ome (5) To simplify the determination of the deflection, the bending moment, and the shearing force the numerical table below is given, in which the following notations are used: e**(cos Bx + sin Bx); a= lS g= y = — e**(sin Bx — cos Bx); (6) 0 = &-® cos Bx; ¢ = 6 sin Bx In Fig. 2 the functions y and y are shown graphically. lo} | 00 L 2 Px, ? oz a4! | 26 : | 40 EEE ° 1 & 3 e a Fig. 2.PROBLEMS IN BENDING OF BEAMS > Taste 1, Fuxcrions », y, 9 ap f be 7 o ¥ 6 t fee i 1.0000 0.00124 | 0.0245 | —o.o121 2 0.9003 0.0079 | —0.0210 | ~0.0131 on 0.8024. —0,0040 | —0.0177 | —0.0137 03 0.7077 —0.0008 | —0.0147 | —0.0140 04 0.6174 0.0019 | 0.0120 | —0.0139 0 0.5323 0.0040 | 0.0095 | —0.0136 ob 0.4530 0.0057 | —0.0074 | —0.0131 o7 0.3798 0.0070 | —0.0054 | 0.0125 08 0.3131 0.0079 | —0.0038 | —0.0117 09 0.2527 2.0085 | —0.0023 | —o.0108 no 0.1988 0.008) | 0.0011 | —o.0100 u o.1si0 0.2090 | 0.0001 | —o.co9t 12 o.10gt 0.0089 | 0.0007 | —0.0082 13 0.0087 | 0.0014 | — 0.0073 14 0.0084 | 0.0019 | —0.0065, Ls 0.0080} 0.0023 | —0.0057 16 0.0075 | 0.0026 | —0.0049 1 0.0069} 0.0028 | —0.0042 18 0.0064 | 0.0029 | —0.0035 19 0.0058 | 0.0029 | —0.0029 20 0.0082 | 0.0029 | ~0.0023 at 0.0046 | 0.0028 | —0.0018 a2 2.0041 | 0.0027 | ~0.0014 2.3 0.0036} 0.0026 | —o.cor0 4 0.0031 | 0.0024 | ~.0007 25 0.0026 | 0.0022 | —0.0004 2.6 0.0022 | 0,0020 | —0.0002 27 0.0018 0.0018 | +0.0001 28 cots} 0.0017 | 0.0003 29 0.0012 | 0.0015 | 0.0004 30 0.0009 | 0.0013] 0.0005 31 0.0006 | 0.0011 0.0006 32 0.0004} a.c010 | 0.0006 33 90,0002 0.0008 0.0006, 34 | ~0.0408 | —0.0237 | —0.0323 0.0001 | 0.0007} 0.0006 35 | —0.0389 | —0.0177 | 0.02836 STRENGTH OF MATERIALS Using the notation (6) and equations (d)-(f) we obtain P a; Pe } =Fe), B= -AF ree, & P. M = — E153 = 5 ¥(6"), @ ay P V= — El 7s = ~ 5 (6x). By using these equations together with table I, the deflection, the slope, the bending moment, and the shearing force for any cross section of the beam can be readily calculated. The maximum deflection and maximum bending moment occur at the origin and are, respectively, b= (de = 2B, ®) My = (Mona = 5 @) Using the expressicn (3) for a single load and the principle of superposition, the deflection produced in an infinitely long beam on an elastic foundation by any other type of loading can be readily obtained. e As an example let us consider the case of a uniform load distributed over a length / of an infinitely long beam (Fig. 3). Consider any point 4, and let ¢ and 4 represent the dis- tances from this point to the ends of the loaded part of the beam. The deflection at 4, produced by an element gdx of the load, is obtained by substi- tuting gdx for P in eq. (3), which gives Fic. 3. gax 3BET, e*(cos px + sin Bx). The deflection produced at 4 by the loading distributed over thePROBLEMS 1!N BENDING OF BEAMS 7 length / then becomes 3 i f EE e-t¥(cos pe + sin Bx) + f SMa eM eos bu + sin Be) = 7 (2 — e- cos Bb — e*™ cos Bc). (g) If c and 4 are large, the values e~* and ¢* will be small and the deflection (g) will be equal approximately to g/k, i-c., at points remote from the ends of the loaded part of the bar the bending of the bar can be neglected and it can be assumed that the uniform loading g is directly transmitted to the elastic foundation. Taking the point A at the end of the loaded part of the bar, we have ¢ = 0, 6 = /, e® cos Bc = 1. Assuming that / is large, we have also e~® cos Bb =o. Then y = /2k; ie, the deflection now has only one half of the value obtained above. In a similar manner, by using equation (4), the expression for bending moment at 4 can be derived. If the point 4 is taken outside of the loaded portion of the beam and if the quantities band c represent, respectively, the larger and the smaller distances from this point to the ends of the loaded part of the beam, the deflection at 4 is > ods y= [tae (cos 6x + sin Bx) = Sz €-* (cos Bx + sin Bx) o SET. = Ee cos pe — e® cos 82). (i) When ¢ =o and 6=/ is a large quantity, we obtain for the deflection the value g/2h, which coincides with our prev- ious conclusion. As the dis- tances 6 and ¢ increase, the deflection (4) decreases, ap- Proaching zero as 4 and ¢ grow larger. ba x—{ The case of a couple acting Fre. 4. on an infinitely long beam, Fig. 4a, can also be analyzed by using the solution (3) for a single load. The action of the couple is equivalent to that of the two forces P shown in Fig. 43, if Pe approaches Mo while e approaches zero.8 STRENGTH OF MATERIALS Using the first of equations (7), we find the deflection at a distance x from the origin: P y= lox) - ola + 90 — MB o(Gx) = ofBe +e] _ — MeBde | 2k e 2k dx Since, from equations (7), dee de = 7 28S we obtain for the deflection curve produced by the couple Mp the following equation: M y= va, (20) By differentiating this equation, we obtain a & = MP yas), M= ~ 1, = Me a(60), (10') Va £1 = — ME (es). Using these equations together with Table 1, one can readily calcu- late the deflection, the slope, the bending moment, and the shearing force for any cross section of the beam. We shall now consider the case of several loads acting on a beam. As an example bending of a rail produced by wheel-pressures of a locomotive will be discussed. The following method of ana- lyzing stresses in rails is based upon ‘the assumption that there is a continuous elastic support under the rail. This assumption is a good approximation,‘ since the distance between the ties is small in comparison to the wave length a of the deflection curve, given by eq. (5). In order to obtain the magnitude & of the modulus of foundation, the load required to depress one tie unit distance must be divided by the tie spacing. It is assumed that the tie is symmetrically loaded by two loads corresponding to the rail pres- sures. Suppose, for instance, that the tie is depressed 0.3 inch under each of the two loads of 10,000 pounds and that the tiePROBLEMS IN BENDING OF BEAMS 9 spacing is 22 inches; then 10,000, i k= og X 227 1,500 Ibs. per square inch. For the case of a single wheel load P, eqs. (8) and (9) are used for the maximum deflection and maximum bending moment. The maximum stress due to the bending of the rail will be @ where Z denotes the section modulus of the rail.® In order to compare stresses in rails the cross sections of which are geometrically similar, eq. (#) may be put in the following form: _P ANT fe : ome = Ge NE? (/) in which 4 is the area of the cross section of the rail. Since the second factor on the right side of eq. (j) remains constant for geometrically similar cross sections and since the third factor does not depend on the dimensions of the rail, the maximum stress is inversely proportional to the area of the cross section, i.e., inversely proportional to the weight of the rail per unit length. An approximate value of the maximum pressure Rmax on a tie is obtained by multiplying the maximum depression by the tie spacing Zand by the modulus of the foundation. From eq. (8) Pe, _ Pel_ P s[ ki Ross = age “2 ~ 2 NQEl, () It may be seen from this that the pressure on the tie depends prin- cipally on the tie spacing 7, It should be noted also that & occurs in both eqs. (j) and (&) as a fourth root. Hence an error in the determination of & will introduce a much smaller error in the magnitude of omax and Rmax. ‘See author's paper on “Strength of Rails,” Transactions of the Institute of Way of Communications, St. Petersburg, Russia (1915), and author’s paper in Proc. of the Second International Congress for Applied Mechanics, Ziirich, 1926. See also reference 2. © In writing eq. (i) it was assumed that the clementary beam formula can be used at the cross section where the load P is applied. More detailed investigations show that, due to local stresses, considerable deviation from the elementary eq. (4) should be expected.10 STRENGTH OF MATERIALS When several loads are acting on the rail, the method of super- position must be used. To illustrate the method of calculation we shall discuss a numerical example. Consider a 100-Ib. rail section with J, = 44 in. and with a tie spacing such that & = 1,500 Ibs. per sq. in.; then from eq. (2) gett ligeate tr sr “ 8 = Nonr,~ Vax go x 10 X aq agg? and from eq. (5) Qn 8 We take, as an example, a system of four equal wheel loads, 66 inches apart. If we fix the origin of coordinates at the point of contact of the first wheel, the values of Bx for other wheels will be those in the table 2 below. The corresponding values of functions y and y taken from the numerical table on p. 5 are also given. 272 in. Tapie 2 Loads | 1 | 2 3 | 4 eee are ree eee eee ° 152 3.05 457 ve. 1 =0.207 | —o.0s1 0.008 v. 1 ago | 0042 | —o.012 Now, after superposing the effects of all the four loads acting on the rail, the bending moment under the first wheel is, from eq. (4), M= ea — 0.207 — 0.051 + 0.008) = ons, ite., the bending moment is 25 per cent less than that produced by a single load P. Proceeding in the same manner for the point of contact of the second wheel we obtain fH P P Mz =~ (1 — 2 X 0.207 — 0.051) = 0.5352. 2 = 36 7 51) = 0.53545 It may be seen that due to the action of adjacent wheels the bending moment under the second wheel is much smaller than under the first. This fact was proved by numerous experimental measure- ments of track stresses. Using eq. (3) and the values in the lastPROBLEMS IN BENDING OF BEAMS. i jine of the numerical table 2, we find the following deflection under the first wheel: The deflections at other points can be obtained in a similar manner. It will be seen that the method of superposition is easily applied to determine the effect of a combination of loads having any arrange- ment and any spacing. ‘The analysis is based on the assumption that the rail support is capable of developing negative reactions. Since there is play between the rail and the spikes, there is little resistance to the upward movement of the rail and this tends to increase the bending moment in the rail under the first and the last wheels. Other elements enter into the problem and these may affect the accuracy of the analysis. Nevertheless, in general, the above theory for the bending of the rail, caused by static loading, is in satisfactory agreement with the experiments which have been made. Problems 1. Using the information given in Table 2, construct the bending moment diagram for the rail assuming that the wheel pressures are equal to 40,000 Ibs. Such a diagram should show that the moments are negative in sections midway between the wheels, which indi- cates that during locomotive motion the rail is submitted to the action of reversal of bending stresses which may finally result in fatigue cracks. 2. Find the bending moment at the middle of the loaded portion of the beam shown in Fig. 3 and the slope of the deflection curve at the left end of the same portion. 3. Find the deflection at any point 4 under the triangular load acting on an infinitely long beam on elastic foundation, Fig. 5. Answer. Proceeding as in the derivation of equation (g), p. 7, we obtain Fre. 5. y= Big Loe) — ¥(B) — 2810180) + 48e1.12 STRENGTH OF MATERIALS 2, The Semi-infinite Beam on an Elastic Foundation—If a long beam on an elastic foundation is bent by a force P and a mo- ment M, applied at the end as shown in Fig. 6, we again can use the general solution (4) of the preceding article. ot , _dince the deflection and the bending moment 7* approach zero as the distance x from the loaded be end increases, we must take 4 = B=o in 7 that solution and we obtain Fic. 6. y = €®(C cos Bx + DsinBx). (a) For determining the constants of integration C and D we have the conditions at the origin, i.e., under the load P: vy\ _ on (2) = — am #y El. (2 UE Substituting expression (2) in these equations, we obtain two linear equations in C and D, from which —V=P. __M “1@EL, I C= spiny, (P— BM); =D Substituting in equation (a), we obtain smart? cos Bx — BMy(cos Bx — sin Bx)] = 2 | Pocax) — amoiax) — (8x). 1) To get the deflection under the load we must substitute in (11) «=o. Then 3 = O)ea = appz? - BM). ar) The expression for the slope is obtained by differentiating eq. (11). At the end (x = 0) this becomes dy I : (2), > — 3RPEr (P — 28M). (12) Using these equations in conjunction with the principle of super- position, more complicated problems can be solved. If a uniformly loaded long beam on an clastic foundation has aPROBLEMS IN BENDING OF BEAMS 13 simply supported end, Fig. 7a, the reaction R is found from the condition that the deflection at the support is zero. Observing that at a large distance from, the support bending of the beam iR is negligible and that its de- pression into the foundation can x be taken equal to g/k,. we cal- culate the value of R by sub- | ta) stituting My =o and 6 =g/k “Bir into equation (11’). This yields the result: Z : : te, = 2. 4 at (13) Me ay A The deflection curve is now v obtained by subtracting de- Fie. 7. flections given by equation (11) for P = R, My = 0 from the uniform depression g/k of the beam, which gives ice Roos Bx = fa — Efe cos Bx). (14) Sar In the case of a built-in end, Fig. 74, the magnitudes of the reaction R and of the moment Mo are obtained from the conditions that at the support the deflection and the slope are zero. Observing that at a large distance from the support the deflection is equal to g/k and using equations (11) and (12), we obtain the following equations for calculating R and Mo: § eee — 47 ~ SBED (R + BM) and 1 © = apgy (R + 28Mo), from which M, = - E14, = Eff. (15) The minus sign of Mo indicates that the moment has the direction shown by the arrow in Fig. 75. SIn equations (11’) and (12), P = — R is substituted, since the Positive direction for the reaction is taken upwards.14 STRENGTH OF MATERIALS Problems 1. Find the deflection curve for a semi-infinite beam on an elastic foundation hinged at the end and acted upon by a couple Mo, Fig. 8. NM, Solution, The reaction at the y—* hinge is obtained from equation (11‘) substituting 6 = 0, which gives Fic. 8. P= gM. Substituting this value of P in equation (11) we obtain (16) By subsequent differentiation, we find dy _ 28M = 7 “¥(Bx), = Mo- (8), @) da = — ELSE = — 6Me- 9(6r). 2. Find the bending moment M and the force P acting on the end of a semi-infinite beam on an elastic foundation, Fig. 9, if the deflection 8 and the slope i at the end are given. aus Pi ly Fie. 9. Solution. The values Mo and P are obtained from equations (11’) and (12) by substituting the given quantities for 6 and (dylds)r» = i. 3. Find the deflection curve for a semi-infinite beam on an elastic foundation produced by a load P applied at a distance ¢ from the free end 4 of the beam, Fig. 10.PROBLEMS IN BENDING OF BEAMS 15 Solution. Assume that the beam is extended to the left of the end 4 as shown by the dotted line. In such a case equation (3) gives the deflection curve for x > 0 and at the cross section 4 of the ep fictitious infinite beam we have, Gh" to from equations (7), using the —4 —* condition of symmetry: vi on P P ly = gp 100» V =" (Bc). (c) To obtain the required deflection curve for the semi-infinite beam, pq (/Z free at the end 4, we evidently (6) must superpose the deflection of y the semi-infinite beam produced. Fic, 10, by forces shown in Fig. rob on the deflection of the fictitious infinite beam. By using equations (3), (21) and (c) in this way we obtain for * > 0: y= eae) + 2 trate +01 + BMo[a(x + ¢)] - BMRB +6)]} @) = re v(x) + ar fo@e)ola(e + ¢)] + WeROMLLe + 1] ~ ENE + ON. This expression can also be used for — ¢ cos Sg eee ®) Ja = Yo = R Sinh BYE sin Bl? # _ PB cosh B/ + cos Bl + 2 : J: GR sinh B+ sin BY ® For the bending moment under the load we find __ P cosh pl — cos pl . = Gpsinh pb sin Bf ® Fic. 14. The method used for the symmetrical case shown in Fig. 11¢ van be applied also in the anti-symmetrical case shown in Fig, 14a. Qo and My in this case will also represent an anti-symmetrical system as shown in Fig. 14c. For the determination of the properPROBLEMS IN BENDING OF BEAMS 19 values of Qo and Mp, the system of equations similar to equations (a), (2) and (c) can be readily written. As soon as Qp and Mp are calculated, all necessary information regarding the bending of the beam shown in Fig. 14a can be obtained by superposing the cases shown in Figures 14d and 14c. Having the solutions for the symmetrical and for the anti- symmetrical loading of a beam, we can readily obtain the solution for any kind of loading by using the principle of superposition. For example, the solution of the unsymmetrical case shown in Fig. 15a is obtained by superposing the solutions of the symmetrical and the anti-symmetrical cases shown in Figs. 154 and 1gc. The 7 % We 7, Y | | lise Y Ye Ti A Fic. 15. Fie. 16. problem shown in Fig. 16 can be treated in the same manner. In each case the problem is reduced to the determination of the proper values of the forces Q, and moments M, from the two equations (c). In discussing the bending of beams of finite length we note that the action of forces applied at one end of the beam on the deflection at the other end depends on the magnitude of the quantity 6/. This quantity increases with the increase of the length of the beam. At the same time, as may be seen from Table 1, the functions , ¥ and @ are rapidly decreasing, and beyond a certain value of @/ we can assume that the force acting at one end of the beam has only a negligible effect at the other end. This justifies our considering the beam as an infinitely long one. In such a case the quantities20 STRENGTH OF MATERIALS (82), $(8/) and 0(8/) can be neglected in comparison with unity in equations (4); by so doing equations (c) are considerably simplified. In general, a discussion of the bending of beams of a finite length falls naturally into the three groups: I. Short beams, 8/ < 0.60. II. Beams of medium length, 0.60 < g/ < 5. TIT. Long beams, / > 5. In discussing beams of the first group we can entirely neglect bend- ing and consider these beams as absolutely rigid, since the deflection due to bending is usually negligibly small in comparison with the deflection of the foundation. Taking, for example, the case of a load at the middle, Fig. 13, and assuming 8/ = 0.60, we find from the formulas given above for y. and y, that the difference between the deflection at the middle and the deflection at the end is only about one-half of one per cent of the total deflection. This indicates that the deflection of the foundation is obtained with a very good accuracy by treating the beam as infinitely rigid and by using for the deflection the formula _P TOE The characteristic of beams of the second group is that a force acting on one end of the beam produces a considerable effect at the other end. Thus such beams must be treated as beams of finite length. In the case of beams of the third group we can assume, in investigating one end of the beam, that the other end is infinitely far away. Hence the beam can be considered as infinitely long. In our previous discussion it was always assumed that the beam was supported by a continuous elastic foundation but the results obtained can also be applied when the beam is supported by a large number of equidistant elastic supports. As an example of this kind, let us consider a horizontal beam 4B, Vig. 17, supporting a system of equidistant vertical beams which are carrying a uniformly distributed load g.8 All beams are simply supported at the ends. Denoting by El; and / the flexural rigidity and the length of vertical beams, we find the deflection at their middle to be 5 ght Rit Y= 384 EN, ~ BET,’ a * Various problems of this kind are encountered in ship structures. A very complete discussion of such problems is given by I. G. Boobnov in his “Theory of Structure of Ships,” vol. 2, 1914, S. Petersburg.PROBLEMS IN BENDING OF BEAMS 21 where R is the pressure on the horizontal beam 4B of the vertical beam under consideration. Solving equation (/) for R we find that the horizontal beam 4B is under the action of a concentrated force, Fig. 17¢, the magnitude of which is 48EZ R= Soh - To ® Assuming that the distance @ between the vertical beams is small in comparison with the length / of the horizontal beam and replacing the concentrated forces by the equivalent uniform load, as shown in Fig. 17¢, we also replace the stepwise load distribution, indicated Abe 4 C0CT TS @) Fie. 17. in the figure by the dotted lines, by a continuous load distribution of the intensity an ky where eo| eat = BEN, Og qi b= aE 7 The differential equation of the deflection curve for the beam 4B en is dty EIT = 1 — by (m) It is seen that the horizontal beam is in the condition of a uniformly22 STRENGTH OF MATERIALS loaded beam on an elastic foundation. The intensity of the load and the modulus of foundation are given by the expressions (/). In discussing the deflection of the beam we can use the method of superposition previously explained or we can directly integrate equation (m). Using the latter method, we may write the general integral of the equation () in the following form: = B+ Ci sin Bx sinh Bx + Ca sin x cosh Bx } tn) + Ca cos Bx sinh Bx + Cx cos Bx cosh Bx. Taking the origin of the coordinates at the middle, Fig. 17¢, we conclude from the condition of symmetry that C=C =0. Substituting this in the solution (7) and using the conditions at the simply supported ends: @ O)s=u2 = 9, (2 we find asin Bl sinh pl Cesc eect +k cos B/ + cosh pl? 2cos a cosh 6 a=-% & cos wooBl-t cosh al ‘The deflection curve then is al. - asin & sun yk cos B/ + cosh 87 sin Bx sinh 6x 2.cos a cosh a ~ cos Bl £ cosh Bi °° Bx cosh Bx]. (0) The deflection at the middle is obtained by taking x = 0 which gives of - 2cos 7 cosh él ) Oder = M1 — Soar costa) * (2)PROBLEMS IN BENDING OF BEAMS 23 Substituting this value in equation (4), we find the reaction at the middle support of the vertical beam, which intersects the beam 4B at its mid-point. It is interesting to note that this reaction may become negative, which indicates that the horizontal beam actually supports the vertical beams only if it is sufficiently rigid; otherwise it may actually increase the bending of some of the vertical beams. Problems 1, Find a general expression for the deflection curve for the case illustrated in Fig. 12. Answer. __ 2P8 cosh Bx cos (/ — x) + cosh BU — x) cos Bx ae sinh g/ + sin B/ : 2. Find the deflections at the ends and the bending moment at the middle of the beam bent by two equal and opposite couples M,, Fig. 18. Fis, 18, Fie. 19. Answer. = y, = — 2M sinh Bl — sin pl vere & sinh 6/ + sin 6” : if yf J. pl sinh con + cook cin M. = 1M aT + sin Bi 3- Find the deflection and the bending moment at the middle of the beam with hinged ends and on an elastic foundation, the load being applied at the middle of the beam, Fig. 19. Answer. P8 sinh 6! — sin Bf 2k cosh B/ + cos pl? P sinh 6/ + sin 67 “4B cosh g/ + cos Bi”24 STRENGTH OF MATERIALS 4. Find the deflection and the bending moment at the middle of the uniformly loaded beam with hinged ends on an elastic founda- tion, Fig. 20. Answer. 2 cosh ee a | 9 Fe eee eee &U 7 Cosh B+ cos Al J? Bl / sinh sin 2 2 * * 268 cosh B/ + cos By” 5. Find the bending moments at the ends of the beam with | built-in ends and on an elastic foundation. The beam is carrying a uniform load and a load at the middle, Fig. 21. inn cin aaa: q_sinh 6/ — sin 6/ P ~ Gsinh 6/ + sin B/~ 26° sinh @/ + sin 6” Mo = 6. Find the deflection curve for the beam on an elastic founda- tion with the load applied at one end, Fig. 22. P ”, WTA ; IG TT y y Fic. 22, Fra. 23. Answer. 2P8 [sinh 8/ cos Bx cosh BU — x) (sinh? Al — sin? BD) — sin B/ cosh Bx cos BY — x).PROBLEMS IN BENDING OF BEAMS 25 7. A beam on an elastic foundation and with hinged ends is bent by a couple Mo applied at the end, Fig. 23. Find the deflection curve of the beam. Answer. 2M, = ieokiay fe op ay eosh a sin Bx sinh a — 3) — cos ff sinh Bx sin B(? ~ x)]. 4. Combined Direct Compression and Lateral Load. Let us begin with the simple problem of a strut with hinged ends, loaded by a single force P, and centrally compressed by two equal and opposite forces S, Fig. 24. Assuming that the Fro, 24. force P acts in one of the principal planes of the strut, we see that the bending proceeds in the same plane. The differ- ential equations of the deflection curve for the two portions of the strut are: a Pi EIS, = — Sy, (a) & Pl — EI oa — Sy = Pur 9G —*). (2) Using the notation S 2 EI P) (17) we represent the solutions of the equations (a) and (4) in the following form: y = Cy cos px + Cy sin px — Ps, (c) Cy cos pe + Ccsin ps PU 9G ay, (a) ~ tl26 STRENGTH OF MATERIALS Since the deflections vanish at the ends of the strut, we con- clude that Ci =0, C3 = — Cy, tan pi. The remaining two constants of integration we find from the conditions of continuity at the point of application of the load P, which require that equations (c) and (d) give the same deflection and the same slope for » = / — ¢; we obtain: Cy sin p(J — c) = Ci{sin p(/ — c) — tan pl cos p(/ — c)], Cxp cos p(/ — c) = Cup[cos p(/ — c) + tan pl sin p(/ — c)] + a from which co, -Psinpe og, 2 _ Psinpd — «) 2 Sp sin pl’ oc Sptan pl” Substituting in equation (c) we obtain for the left portion of the strut: Psin pe . Pe y = Span pi PX — B® (18) From this, by differentiation, we find: dy _ Psin pe Pe de ~ 'Ssin pl ©°8 P* — $7? dy _ Ppsinpe . G9) da ~ ~ Sin pl 0 P* The corresponding expressions for the right portion of the strut are obtained by substituting (/ — x) instead of x, and (J — ¢) instead of ¢, and by changing the sign of dy/dx in equations (18) and (19). These substitutions give: _ P=6) = Psin pl = 0) gin pa — x) Fee =, (20) Sp sin pl yPROBLEMS IN BENDING OF BEAMS 27 dy _ _ Psinp(/—¢) Ha Pil —c) dx Ssinpl 8 pd 8) +a (1) dy _ _ Ppsinp(?—c) |. He : Pan eiein peat pl — +). (2) In a particular case, when the load P is applied at the middle, we have ¢ = //2, and, by introducing the notation SP _pP iy gel 4 uy (23) we obtain from equation (18) (max = (ect = asp 7 9) PR ti —4u = Geer ae (24) The first factor in expression (24) represents the deflection produced by the lateral load P acting alone. The second factor indicates in what proportion the deflection produced by P is magnified by the axial compressive force S. When S is small in comparison with Euler load (S, = EI7?/I*), the quantity ~ is small and the second factor in equation (24) approaches unity, which indicates that under this condition the effect on the deflection of the axial compressive force is negligible. When S approaches the Euler value, the quantity 4 approaches the value 7/2 (see eq. 23) and the second factor in expression (24) increases indefinitely, as should be expected from our previous discussion of critical load (see p. 244, Part I). The maximum value of the bending moment is under the load and its value is obtained from the second of equations (1g), and ete dy — pyr PP ya P! _ Pi, tanu Manz z1( 3) ae = Ertan =F BH. (as) Again we see that the first factor in expression (25) represents28 STRENGTH OF MATERIALS the bending moment produced by the load P acting alone, while the second factor is the magnification factor representing the action of the axial force S on the maximum bending moment. Having solved the problem for one lateral load P, Fig. 24, we can readily obtain the solution for the case of a strut bend by a couple applied at the end, Fig. 25. It is only necessary Fic. 25. to assume that in our previous discussion the distance c is indefinitely diminishing approaching zero, while Pc remains a constant equal to Mp. Substituting Pe = Mo and sin ke = ke in equation (18), we obtain the deflection curve: = An sees pl -7) i" (26) from which My (ae - /): “S \sinpl 77 The slopes of the beam at the ends are (2)_.-9'(stn-7) Mil I I GET 6(52 sin aa ~ Tea) en 1 1 - 45 aaa) Again the first factors in expressions (27) and (28) taken with proper signs represent the slopes produced by the couple. oe PROBLEMS IN BENDING OF BEAMS 29 M, acting alone (see p. 158, Part I), and the second factors represent the effect of the axial force S, Considering equations (18) and (26), we see that the lateral force P and the couple My occur in these expressions linearly, while the axial force S occurs in the same expressions in a more complicated manner, since p also contains S (see eq. 17). From this we conclude that if at point C, Vig. 24, two forces P and Q are applied, the deflection at any point may be obtained by superposing the deflections produced by the load Q and the axial forces 8 on the deflection produced by the load P and the same axial forces. A similar conclusion can be reached regarding couples applied to one end of the beam. The conclusion regarding superposition can be readily generalized and extended to cover the case of seve’ al loads, Fig. 26. For each portion of the strut an equatio’: similar to Fic. 26. equations (a) and (4) can be written, and a solution similar to those in (c) and (d) can be obtained. The constants of integration can be found from the conditions of continuity at the points of load application and from the conditions at the ends of the strut. In this way it can be shown that the deflection at any point of the strut is a linear function of the loads Py, P2, +++ and that the deflection at any point can be obtained by superposing the deflections produced at that Point by each of the lateral loads acting together with the axial force S. Let us consider a general case when 7 forces are acting and m of these forces are applied to the right of the cross section for which we are calculating the deflection. The expression for this deflection is obtained by using equation (18) for the forces Pi, Pa, +++ Pn and equation (20) for the30 STRENGTH OF MATERIALS forces Pini1; Pmy2,*** Pa» In this way we obtain the required deflection: = S2P*_ Fp. sin per - EPs Y= Spsin pi & Pesin per — gy Paes 1 xX Pisin p/ — ¢:) femtt sin p(/ — x) + Sp sin pT } cm ar Pl — es). (29) tomt1 If, instead of concentrated forces, there is a uniform load of intensity g acting on the strut, each element gdc of this load, taken at a distance ¢ from the right end, can be con- sidered as a concentrated force. Substituting it, instead of ef P;, in equation (29) and replacing summation signs by integra- tion, we obtain the following expression for the deflection curve: sinpx (i. x ye Span al j q sin pe de — al, qede sinp/~—-x) fi. i + er Af sin pI ode —5-® fg - ode. Integrating the above gives = Aon =x) (Go) and I Joux = (y)enye : (se le I id gh cos Rael aoe 3DPROBLEMS IN BENDING OF BEAMS 31 By differentiating equation (30), we readily obtain the ex- pressions for the slope and for the bending moment. The slope at the left end of the strut is zl dr) a (aa a gh _tanu —u ( ae ae ya” (3?) The maximum bending moment is at the middle where . ay Max = — a1(3) aal/2 (st) PETS a] gt 201 = 0s w) po 8 uw? Cos u 2 =EI + (33) S cos By using solution (26) for the case of a couple together with solution (29) for lateral loads, and applying the method of superposition, various statically indeterminate cases of bending of struts can be readily solved. —% 7 Taking as an example the case L, of a uniformly loaded strut Fro. 27. if built in at one end, Fig. 27, we find the bending moment M) at the built-in end from the condition that this end does not rotate during bending. By using equations (28) and (32) this condition is found to be _ gf tanu—u | Md ( 3 3 \= 24k ae + 3EI \autanan~ Guy) ~° “from which _ 9? 4 tan 2u(tan wv — 4) | 8 u(tan 2u — 2u) (34) M, = In the case of a uniformly loaded strut with both ends built-in the moments Mo at the ends are obtained from the32 STRENGTH OF MATERIALS equation: _ gf tanu-u 24ET hae from which sige eam a Mo = 12 $v? tanu (35) It is seen from expressions (34) and (35) that the values of the statically indeterminate moments in the case of struts are obtained by multiplying the corresponding moments calcu- lated from the beam formulas by certain magnification factors. All necessary calculations can be greatly simplified by using the prepared numerical tables for determining magnification factors.® When the maximum bending moment for a strut is found, the numerical stress maximum is obtained by combining the direct stress with the maximum bending stress, which gives | Moz Ss max = Gt (e) where 4 and Z are, respectively, the cross-sectional area and the section modulus for the strut. Taking, as an example, the case of a uniformly loaded strut with hinged ends, we obtain from equation (33): _ 8S, g@ xt —cosu) mx AT 82° w@oosu In selecting the proper cross-sectional dimensions of the strut it is necessary to consider that the right side of the equation (f) is not linear in S since the quantity w also de- pends on S, as may be seen from expression (23). Owing to this fact the maximum stress increases at a greater rate than * Various particular cases of laterally loaded struts have been dis- cussed by A. P. Van der Fleet, Bull. Soc. of Engineers of Ways of Com- munication, 1900-1903, St. Petersburg. Numerous tables of magnifi- cation factors are given in that work.PROBLEMS IN BENDING OF BEAMS 33 the force S. Thus the usual method for determining the proper dimensions by taking '° =o lomax| => (g) where x is the factor of safety, fails in this case. If the strut must be designed so that it will begin to yield when the forces S and g increase 7 times, the cross section must be selected so that [o|max will be somewhat smaller than oy,/n in order to satisfy the equation: Sup _ Sg? 24 — cos us) a A+ 82 ugcosm ° @) in which a; = nw. It is apparent that if we proceed in this manner we satisfy the requirement regarding the beginning of yielding; by multiplying both sides of the equation (4) by ” we find _ nS ngl? 21 — cos wu) i om = 4tRZ ucosm ” “ which indicates that the maximum stress reaches the yield point stress when § and q have been increased 7 times. Similar procedure in the design of struts can be applied in other cases of loading. We can conclude from the above discussion that to ascertain a factor of safety m in the design of struts, we must use instead of equation (g) a modified equation similar to equation (4), in which the parameter « is replaced by u1 = nu. Problems 1, Find the slope at the left end of a strut with hinged ends which is loaded at the middle by the load P. Tt is assumed that material of the strut has a protounced yield Point. 4 This method of design of struts was developed by K. S. Zavriev, see Memoirs of the Institute of Engineers of Ways of Communication, 1913, S. Petersburg.34 STRENGTH OF MATERIALS Answer. (2) _ Pi ~cosu_ PP 1—cosu emo dx Jno 2S cosu 16FT hu cosu 2. Find the slopes at the ends of a strut carrying a triangular load, Fig. 28. Fic, 28. Solution. Substituting in equation (29) gocde//, instead of Pi, and replacing summation by integration we find: sinpy (Fae. x (ge y p ‘sin pede = 5 “pac “‘SpsinplJ, 7 sinpU—x) (ge. 9, ., Gey “Spsinpl J Fn PU ~ Ode TO - ade. Differentiating this with respect to x, we find that dy _ 290 a = pete — 9) dy\ _ a (2). =~ Sper — > where and 8 are functions given by expressions (36) (see p. 36). 3. Find the slopes a the ends of a strut symmetrically loaded by two loads P, as show 1 in Fig. 29. Answer. #) te dy —P(cospt_\. Ax Jew ax Jon § pl coe 4. A strut with built-in ends is loaded as shown in Fig. 29. Find the bending moments, M, at the ends.PROBLEMS IN BENDING OF BEAMS 35 Solution. The moments M are found from the conditions that the ends of the strut do not rotate. By using the answer of the Fic. 29, preceding problem and also equations (27) and (28), the following equation for calculating Mp is obtained: Mo Mo P b sare + jar +5(“4)-*)=0 My = —2PEL_«_ (£08 pb i 0 = — ST tanu\ cosa * from which If £ = 0, we obtain the case of a load 2P concentrated at the middle. Ms My Maes Sa Se. Snot me hh (3 ne lan nor o al iG en 7 7 es Fic. 30. 5- Continuous Struts.—In the case of a continuous strut we Proceed as in the case of continuous beams (see p. 201, Part I) and consider two adjacent spans, Fig. 30. Using equations (23), (27), and (28) and introducing notations for the th span: "The theory is due to H. Zimmermann, Sitzungsb., Akad. Wiss., Berlin, 1907 and 1909.36 STRENGTH OF MATERIALS I PIR ET Ste] Oe OT Bun Sin 2u, (2uy)? J? (36) Becerra pe | aoe 2a, tan wa}: tan un — Un Pater rg arnt (37) We conclude that the slope at the right end of the th span, Fig. 30a, produced by the end moments M,-1 and M,, is Maly Mn-tln — Bn SET, ~ °* GET @ The slope produced at the left end of the 7 +1 span by the mo- ments M, and Mn is Masgalnss Milnes | Ot EE Tact + Bost Toa” 7 @ If there is no lateral load acting on the two spans under considera- tion, the expressions (2) and . must be and we obtain: ale FP Moa + 2(az + Ban =) at, + ena AY Mast = 0. (38) This is the three moment equation for a continuous strut if there is no lateral load on the two spans under consideration. If there is lateral load acting, the corresponding slopes produced by this load must be added to expressions (a) and (4). Taking, for example, the case of uniform load g, and ga41 acting on the spans mand + 1 in a downward direction, we obtain the corresponding slopes from equation (32) and, instead of expressions (a) and (4), we obtain: Maln Galo ~ Beer, ~ GET, ~ Va4k I? © Masini Malnss Insalngs® an Bice + Bas gerne t Yet ie @ Equating these two, expressions we obtain: nln Jat te. +o(a7 + Bar et ee Mays * ey, Gln? gngtlnss? Te TO Ts (39)PROBLEMS IN BENDING OF BEAMS 37 This is the three moment equation for a strut with a uniform load in each span. It is similar to the three moment equation for a continuous beam and coincides with it when S' = o and all functions a, B, y become equal to unity. For any other kind of lateral load we have to change only the right side of the equation (39), which depends on the rotation of the adjacent ends of the two spans produced by lateral loading. Taking, for example, the case of a trapezoidal load shown in Fig. 31 and dividing the load into two parts, uniform loads and triangular loads, we use for the uniform loads the terms which we already have on the right side of equation (39). Fie. 31. To these terms we must add the terms corresponding to the triangular loads. Using the expressions for the slopes in problem 2 of the preceding article, we find that the two terms which we have to add to the right side of equation (gg) in the case of the load shown in Fig. 31 are: (qn-1 = Gn)n 2(qn = Gnirvlnss TE 0 OR Ge DO in which a and Bris are defined by expressions (36). If concen- trated forces are acting on the spans under consideration the re- quired expressions for the rotations are readily obtainable from the general expression for the deflection curve (29). The calculation of moments from the three moment equations (39) can be considerably simplified by using numerical tables of functions a, 8 and y.!8 In the derivation of equation (39) it was assumed that the moment M, at the mth support had the same value for both adjacent spans. There are cases, however, in which an external moment M,° is applied at the support as shown in Fig. joc; in such cases we must distinguish between the value of the bending moment to the left and to the right of the support. The relation between these two mo- ments is given by the equation of statics: '* M, — M,° — My! = 0, ¥ Such tables can be found in the book by A. S. Niles and J. S. Newell, “Airplane Structures,” Vol. 2, 1938, see also writer’s book, “Theory of Elastic Stability,” 1936. “The direction of M,° indicated in the Fig. 3oc is taken as positive direction for an external moment.38 STRENGTH OF MATERIALS from which M,! = M, — M,°. oO Equation (39) in such a case is replaced by the following equation: Onde lass py, Me + bat M, + Buss ft Me! + omit Ts * Musa it Gnbnd gasalnss® = TT (49) If the supports of a continuous strut are not on a straight line, the additional terms, depending on the differences in the levels of the three consecutive supports, must be put on the right side of equation (39) or (40). These terms are not affected by the presence of the axial forces, and are the same as in the case of a beam (see p- 204, Part I). Problems 1. Write the right side of the three moment equation if there is a concentrated force P in the span x + 1 at a distance ¢y41 from the support 7 -++ 1. Answer. , in SPE (si im Paint -#) se 6P (gndest fe) : Smet \sin parilagt Inga ~ pastas (Sin Pasilngs Anat 2. Write the right side of the three moment equation if the 7th span is loaded as shown in Fig. 29, p. 35, and if there is no load on span n + 1. Answer. Using the solution of problem 3, p. 34, we obtain the following expression:PROBLEMS IN BENDING OF BEAMS 39 3. Find the right side of the three moment equation if the load is as shown in Fig. 32. Answer. — Ltn [£8 Pnbn 2 Pail Pn cos? 6. Tie-Rod with Lateral Loading.—If a tie-rod is sub- mitted to the action of tensile forces S and a lateral load P, Fig. 33, we can write the differential equation of the deflection 5 fs s = Fie. 33. —* curve for each portion of the rod in exactly the same manner as we did for a strut, Art. 4. It is only necessary to change the sign of §. In such a case instead of quantities p? and x? defined by expressions (17) and (23), respectively, we shall have — p? and — w*, and instead of p and x we shall have pv—T = piand uV—1 = ui. Substituting — 8, pi, and ui in the place of S, p and w in the formulas obtained for the strut in Fig. 24, we obtain necessary formulas for the tie-rod in Fig. 33. In making this substitution we use the known relations: sin wi =isinhu, cosui =coshu, tanui =i tanha. In this way we obtain for the left portion of the tie-rod in Fig. 33, from equations (18) and (19): _ Psinh pe Sp sinh p/ ay P sinh pe dx ~ ~ S sinh pl COS + @y _ _ Pp sinh pe dx? ~ ~ “S sinh pf sinh px. sinh px + a x, (41) (42)40 STRENGTH OF MATERIALS Similar formulas can also be obtained for the right-hand portion of the tie-rod by using equations (20)-(22). Having the deflection curve for the case of one load P acting on the tie-rod, we can readily obtain the deflection curve for any other kind of loading by using the method of superposition. Considering, for example, a uniformly loaded tie-rod and using equations (30) and (31) for a strut, we obtain: cosh (@- p) cosh ol y-F ~1 [+ 4x0 - 9, and the maximum deflection is Joox = (y)enty2 I we gee ee eee geea SU et5 gee stosit atu Ue ets Hegel = 5a, ET (lsat = 30g EL g(t), (43) 7 384 5 3 where we (y = Soha 1 FG eee (5/24)ut The slope of the deflection curve at the left-hand end, from equation (32), is dy _ gu —tanhu de Jeo” 2ET ae (44) The maximum bending moment, which in this case is at the middle of the span, is obtained from expression (33): _ @ | 2(cosh w — 1) _ a Mux = “Fs —Reoha = (45) where 2(cosh u — 1) : w? cosh u vila) = The deflection and the maximum bending moment are ob- tained by multiplying the corresponding expressions for aPROBLEMS IN BENDING OF BEAMS 41 simple beam by factors ¢:(#) and ¥(u), depending on the magnitude of the axial tensile force S. The numerical values of these factors are given in Table 3.% In the case of bending of a tie-rod by a couple applied at the right-hand end, the deflection curve is obtained from equation (26), from which _ Mo(x _ sinh px 9 98(7 - Sang) (46) If there are two equal and opposite couples applied at the ends of a tie-rod, the deflection curve is obtained by the method of superposition: = (5 - 25) 2 = 9 \7 7 sinh pl Mal s sohpl = 9] SLT 7 sinh pl See Jere (ntsc From this equation we find the deflection at the middle and the slope at the left-hand end of the tie-rod: ( — Mo coshu —1 _ Mol? coshu —1 Denti2 = “g coshu ~ SET" 3x2 coshu ’ dy) _ Mo, = Mol, tanh (48) dx Jeug SPAM = FEE The bending moment at the middle is a&y T (Man =~ EI($2) = Me eo Go) Having the deflection curves for a tie-rod with hinged ends bent by transerve loading and by couples at the ends, we can readily obtain various statically indeterminate cases of 5 Various cases of bending of tie-rods are investigated in the papers by A. P. Van der Fleet previously mentioned (see p. 32) and also in the book by I. G. Boobnov, “Theory of Structure of Ships,” vol. 2, 1914, S. Petersburg. From the later book the Table 3 is taken.42 STRENGTH OF MATERIALS bending of tie-rods by the method of superposition. Taking, for example, the case of a uniformly loaded tie-rod with built-in ends and using expressions (44) and (48), we obtain the bending moments M, at the ends from the equation: geo ou tanh w Mi tanhu _ agEl e+ 2EI aw from which vary leh Niel UK. pagina ci Mo= — 35 4 tanhu 1a #24), (50) where u —tanha we) = Tank a” The numerical values of the function y2(u) are given in Table 3. By using expressions (45) and (49) the bending moment at the middle, M,, is obtained: M, =. coshw — 1) _ gw — tanh x tore: w@coshu 12° inh w _ gf 6(sinhu ~ u) 24 #@ sinh u - Leo. (50 The deflection at the middle is obtained by using equations (43) and (48) which give I cosha 7+ 2 gh, coh TT Jax = (y)ent2 = 384 EL (s[aayae gi (u — tanh u)(cosh »— 1) _ gh ~ 16ET # sinh « 3eqer | °2) (52) ucosh w— 4 sinhPROBLEMS IN BENDING OF BEAMS 43 Taste 3. Dertectioxs ax Maximcm Benpisc Moment Constants 1 Lateratty Loapep ‘Ire-Rops ee ee eee «© | 1.000 | 1.000 ' 1.000 1.000] 6.5 | 0.084 | 0.197 | 0.047 | 0.391 | 0.139 0.5 | 0.908 | 0.976 0.905 0.972] 7.0 | 0.047 | 0.175 | 0.041 | 0.367 | 0.121 1.0 | o.7ir | 0.909 , 0.704 0.894 | 7.5 | 0.041 | 0.156 | 0.036 | 0.347 | 0.106 15 | 04523 | 0.817 ‘0.511 0.788 | 8.0 | 0.036 | 0.141 | 0.031 | 0.328 | 0.093 20 | 0380 0715 0.367 | 0.806 | 01673) 8.5 | 0.092 | 0.127 | 0028 | 0311 | 0.083 a.g | o.281 | 0.617 | 0.268 | 0.736 | 0.563] 9.0 | 0.029 | 0.115 | 0.025 | 0.296 | 0.074 3.0 | 0.213 | 0.529 | 0.200 | 0.672 | 0.467] 9.5 | 0.026 | 0.105 | 0,022 | 0.283 ! 0.066 3.5 | 0.166 | 0.453 | 0.153 0.614 | 0.386 | 10.0 | 0.024 | 0.096 | 0.020 | 0.270 | 0.060 40 | 0.132 | 0.388 | 0.120 0.563 | 0.320 | 10.5 | 0.021 | 0.088 45 | 0.107 | 0.335 | 0.097 | 0.519 | 0.267 | 11.0 | 0.020 | 0.081 5.0 | 0.088 | 0.291 | 0.079 | 0.480 j 0.224 | 11.5 | 0.018 | 0.075 5.5 0.074 ae 0.446 | 0.189 | 12.0 | 0.016 | 0.069 5.0 | 0.063 | 0.233 0.055 | o4t7 0.162 i 0.018 | 0.259 | 0.054 0.017 | 0.248 | 0.050 0.015 | 0.238 | 0.045 101g | 0.229 | 0.042 All these functions are equal to unity at 2 = 0, i.e., when only a transverse load is acting. As the longitudinal tensile force in- creases, each function decreases, i.e., the longitudinal tensile forces diminish the deflections and the bending moments in laterally loaded tie-rods. Some applications of the above table will be given later in discussing the bending of thin plates (see p. 122). Problems 1, Find the maximum deflection and the maximum bending moment for a tie-rod loaded at the middle. Answer. PR 4 —tanhu (Y)max = BAe? aaa Pi tanh uw 4. 2. Find the bending moments M at the ends of a tie-rod with built-in ends symmetrically loaded by two forces P as shown in Fig. 29. Solution. The bending moments at the ends are obtained from the equatio: P{ _ cosh pb\ | Mol tanh « i 2El ou cosh _,3-, Find the bending moments at the ends of a tie-rod with built-in ends loaded by a triangular load as shown in Fig. 28. Hint. Use solution of problem 2 on p. 34 together with equa- tion (46).44 STRENGTH OF MATERIALS 7. Representation of the Deflection Curve by a Trigonomet- tical Series.—In discussing the deflection of beams, it is sometimes very useful to represent the deflection curve in the form of a trigo- nometrical series.!°_ This has the advantage that a single mathe- matical expression for the curve holds for the entire length of the span. Taking the case of the beam with supported ends shown in Fig. 34, the deflection at any point e—4P may be represented by the following : iy series: o 7 am ess Soon te) y = a sin-F + ay sin ia taysiney (a) a V5 I Geometrically, this means that the Fic. 34. deflection curve may be obtained by superposing simple sinusoidal curves such as shown in Fig. 34 (2), (©), (@), etc. The first term in series (a) represents the first curve, the second term, the second curve, etc. The coefficients a1, a2, a3 of the series give the maximum ordinates of these sine curves and the numbers 1, 2, 3, +++ the number of waves. By properly determining the coefficients m, a2, +++, the series (2) can be made to represent any deflection curve ® with a degree of accuracy which depends upon the number of terms taken. We will make this determination ote coefficients by a considera- tion of the strain energy of the beam (eq. 188, p. 297, Part I) as given by the equation ay 2 Jo The second derivative of y, from (a), is dx. ® a in gyat gin eee gaat esi as Sysi 4225 sin =| a3 de Equation (4) involves the square of this derivative, which contains 1 See the author's paper, “Application of General Coordinates in Solution of Problems on Bending of Bars and Plates,” Bulletin of the Polytechnical Institute in Kiev, 1909 (Russian); see also H. M. Wester- gaard, Proc. Amer. Soc. Civ. Eng., Vol. 47, pp- 455-533» 1" For other cases analysis becomes too complicated for most prac- tical purposes. See Byerly, “Fourier Series and Spherical Harmonics,” §§ 19-24. See also Osgood, “ Advanced Calculus,” 1928, p. 391.PROBLEMS IN BENDING OF BEAMS 45 terms of two kinds: nh. amx wna nex . max an? sin? and 2am —F— sin sin —F— + By direct integration it may be shown that a a + Wax i . ATK. MEX ff sce ae =! and f sin sin dv = 0, lo Z 2 lo i where nm. Hence, in integral (4), all terms containing products of coefficients such as 4,4» disappear and only the terms with squares of those coefficients remain. Then EIn* 4h U= (tea? + 2a? + ZhaP be) = ar x nia. (53) In a previous discussion (see eq. 4, p. 339, Part I) it was shown that if an elastic system undergoes a small displacement from its position of equilibrium, the corresponding increase in the potential energy of the system is equal to the work done by the external forces during such a displacement. When the deflection curve is given by series (a), small displacements can be obtained by small varia- tions of the coefficients a1, a2, az, «++. If any coefficient a, is given an increase da,, we have the term (a, + dap) sin (wmx//) in series (a) instead of the term a, sin (zxx/l), the other members remaining unchanged. This increase da, in the coefficient a, represents an additional small deflection given by the sine curve day sin (2mx//), superposed upon the original deflection curve. During this addi- tional deflection the external loads do work. In the case of a single load P, applied at a distance c from the left support, the point of application of the load undergoes a vertical displacement day, sin (nxc/l) and the load does the work: day ( sin) p, G) Let us consider now the increase in the strain energy, given by eq. (53), due to the increase dap in an, aU EIv* dU = Fo ding = Eran (a)46 STRENGTH OF MATERIALS Equating this to the work done (c), BT hay = Psin™, from which 2? 1. nxe = ET Krom this we can determine each of the coefficients in the series (a) and the deflection curve becomes 2PP( . me. me 1, ome, nx sin sinF + -sin“Fsin + o> Ete Tee 7 Petey 7 = aad, t sin sin 7 (54) From this the deflection may be calculated for any value of x. For example, the deflection at the middle when the load is at the middle, ¢ = x = //2, will be 2Pp 1 I b= Oeu=ga(sththe-). By taking only the first term of this series, we obtain Comparison with eq. (90), p- 143, Part I, shows that we obtained 48.7 where the exact value was 48, so that the error made in using only the first term instead of the whole series is about 1} per cent. Such accuracy is sufficient in many practical cases and we shall have other examples where a satisfactory accuracy is obtained by using only one term in the series (a). From the solution for a single load (eq. 54), more complicated problems can be studied by using the method of superposition. For example, take a beam carrying a uniformly distributed load, of intensity g. Each elemental load gde at distance c from the left support produces a deflection obtained from eq. (54), with P = qdc, Integrating this with respect to ¢ between the limits ¢ = 0 andPROBLEMS IN BENDING OF BEAMS 47 ¢ = 1 gives the deflection produced by the entire load (55) Taking the first term only, the deflection at the middle of a uni- formly loaded beam is we find that the error in taking only the first term was less than 2 per cent in this case. The trigonometric series (a) is especially useful when the beam is submitted to the action of a longitudinal compressive or tensile force in addition to lateral loading. In the problem shown in Fig. 35, the hinge B approaches the fixed hinge 4 during deflection Fro. 35. by an amount equal to the difference between the length of the deflection curve and the length of the cord 4B. For a flat curve this difference is (see p. 177, Part I) rat f (ZY ae (56) With y as given by series (2), the square of its derivative contains terms of the two forms: UTX MTX d peered ee an 2dnam Pe cos 7 cos 7 By integration it can be shown that f 1 nme i nie mine fi cov a = J, 608 cos de = 0, nm. 0 Z 2 7 i ** Longidutinal contraction due to the axial force can be considered as constant for small deflections.48 STRENGTH OF MATERIALS The equation for \ then becomes ne yea (57) To calculate the coefficients a1, @, a3, +++ in the series (2) we con- sider the work done by the external forces during a small displace- ment da, sin (nnx/1) from the position of equilibrium, In the case represented in Fig. 35 both the force P and the longitudinal force S do work during such a displacement. The displacement A, due to the increase da, in the coefficient ap, increases by an amount an re =< da, =n ay 3a," a” anlan. ‘Then the work done by the force S is me Lary Ss a” Adan. This is added to the work (c) done by the lateral force and the sum is equated to the increase in the potential energy (eq. @). This gives us the following equation for determining any coefficient an: nm EIn ap Pain dan + 8 5 Mayday = — Mandan, from which Tf the ratio of the longitudinal force to the critical value of this load (see p. 27) be denoted by « = S?/EIn?, we obtain He Ee I Pu o = Ele re — a) 7 Substituting in the series (a), the deflection curve is 2Ph I. me. me I . me. mx (in Main 4 agli Man E+.) _ 2PR I ; ATC. nex 8) ~ ETB E mea MTS Comparing this with eq. (54) for the case of a lateral force P only, we sce that the deflection of the bar increases due to the action of the longitudinal compressive force §. We have seen that the firstPROBLEMS IN BENDING OF BEAMS 49 term in series (2) represents a good approximation for the deflection; hence the increase of the deflection produced by the longitudinal force will be approximately in the ratio 1 : (1 — a). This conclusion holds also if there are several transverse loads of the same direction or if there is a continuous load acting on the beam. Denoting by 6) the maximum deflection produced by lateral load acting alone we can assume with satisfactory accuracy that under the combined action of compressive forces Sand lateral load the maximum deflection is b Ia. é (59) This expression for the maximum deflection can be used also for an approximate calculation of bending moments in a strut. For ex- ample, in the case of a uniformly loaded strut with hinged ends the maximum bending moment can be calculated from the following approximate formula: Ps nae = pO I-a 60) If the longitudinal force is tensile instead of compressive, the method discussed above still holds, with —a instead of w in the expressions for the deflection curve (58). Taking only the first term in this expression, the approximate formula for the deflection at the middle becomes by ome ) where 8 denotes the deflection produced by lateral loads only. It must be noted that in the case of longitudinal tensile forces a can be larger than unity, and the accuracy of the approximate equation (61) decreases with increase of «. Taking, for instance, a uniformly distributed lateral load, the error in eq. (61) at a= 1 is about 0.3 per cent. At a = 2 the error is 0.7 per cent and at @ = Io the error is 1.7 per cent. ‘ In the case of a bar with built-in ends an approximate equation, analogous to eq. (61), may be derived for calculating the deflec- tion at the middle, which gives Sears (62) in which é is the deflection at the middle produced by lateral loads acting alone and & has the same meaning as before.50 STRENGTH OF MATERIALS The applications of these approximate equations will be shown later in considering the deflection of thin rectangular plates. The method of trigonometric series can be extended also in analyzing beams of variable cross section.” 8. Bending of Beams in a Principal Plane which is not a Plane of Symnietry. Center of Twist.—In the discussion of pure bending (see p. 93, Part I) it was shown that the plane of the deflection curve coincides with the plane of the bending couples provided these couples act in one of the two principal planes of bending. This does not hold however in the case of bending of a beam by a coplanar system of transverse forces. If the plane in which the forces are acting is not a plane of symmetry of the beam, such bending is usually accompanied by torsion of the beam. The following = discussion will show how this ac 7 z_ torsion can be eliminated and a simple bending established bh @ iy () Yq bya proper displacement of the plane of the acting forces parallel to itself. We begin with simple examples in which the cross section of the beam has one axis of symmetry (z axis), and the forces are acting in a plane perpendicular to this axis, Fig. 36. Let us consider the case shown in Fig. 36a and determine the position of the vertical plane in which the transverse loads should act to produce simple bending of the beam in a vertical plane. From our previous discussion of distribution of ver- tical shearing stresses r,., see p. 109, Part I, we may conclude that practically the whole of the shearing force V will be taken by the flanges alone. If we consider the flanges as two separate beams whose cross sections have moments of inertia I,’ and J,” respectively, then their curvatures and their deflections in bending will be equal if the loads are distributed 2See paper by M. Hetényi, Journal of Applied Mechanics, 1937, vol. 42 A~49.PROBLEMS IN BENDING OF BEAMS 51 between them in the ratio 7,’: J," The shearing forces in these flanges also will be in the same ratio. This condition will be satisfied if the transverse loads act in the vertical plane through the point O (Fig. 36, 2), such that I _ Ie" Ay Ti? where 4; and fy are the distances of O from the centroids of the cross sections of the flanges. In this manner we find that for the case of flanges of small thickness the point O is dis- placed from the centroid C of the cross section towards the flange whose cross section has the larger moment of inertia. In the limiting case, shown in Fig. 36, 4, in which one of the flanges disappears, it can be assumed with sufficient accuracy that the point O coincides with the centroid of the flange and that the transverse loads should act in the vertical plane through this point in order to have simple bending. The point O, through which the plane of loading must pass to eliminate torsion, is called center of twist. Let us now consider the channel section (Fig. 36, ¢) and determine the position of the plane in which vertical loads must act to produce simple bending with the z axis as the neutral axis. For this purpose it is necessary to consider the distribution of the shearing stresses over the cross section in simple bending. To calculate the vertical shearing stresses tyz for the cross section of the web, the same method is used as in the case of an I beam (page 10g, Part I) and it can be as- sumed with sufficient accuracy that the vertical shearing force V is taken by the web only. In the flanges there will be hori- zontal shearing stresses which we shall denote by 7.2. To find the magnitude of these stresses let us consider an element cut from the flange by two adjacent cross sections dx apart and by a vertical plane mmm; parallel to the web (Fig. 37). If the beam is bent convex downward, the upper flange will be _ 4 The effect of shearing force on deflection of flanges is neglected in this consideration.52 STRENGTH OF MATERIALS in compression and the compressive forces N and N + dN acting on the above element will be numerically equal to ve - Of yaa (wes i) and =N+dN=— co J 4a, where the integration must be extended over the shaded portion of the cross section of the flange. The integral repre- sents the moment of the shaded area with respect to the z axis. The difference of the compres- sive forces Nand N +dN must be equal to the sum of the shearing stresses 7,, acting over the side mnmyn, of the element. Assuming that these stresses are uniformly distributed over this side and denoting by ¢ the thickness of the flange, we obtain the following equation for calculating rzzt dM dx ratdy = dN = — Sf yd, from which ta = — Fp f yd. (a) The moment of the shaded area is proportional to the distance u from the edge of the flange; hence r,: is proportional to w. As we have shown before (see p. 111, Part I), shearing stresses 722, equal to rz2, must act horizontally at points along the line #7, in the cross section of the flange. Hence the stresses rz are distributed non-uniformly over the cross section of the flange but are proportional to the distance u. At the junction of flange and web the distribution of shearing stresses is complicated. In our approximate calculation we shall as- sume that eq. (a) holds from u = otou = 4. Then denotingPROBLEMS IN BENDING OF BEAMS 53 by 4 the distance between the centroids of the flanges and observing that the moment of the cross section ¢ of the flange with respect to the axis z numerically is d/(4/2), we obtain, from eq. (a), (res)nae = (Pas)aae = SF o The resultant R of the shearing stresses r., distributed over the cross sectional area d¢ of the flange is : pn Lik VBI . The sum of the shearing stresses 7.. over the cross section of the lower flange will evidently be an equal and opposite force. Thus the shearing stresses over a channel section reduce to the forces shown in Fig. 38. This system of forces is statically equivalent to a force V applied at a point O at a distance from the center of the web: _ Rh _ bE ea ii aia From this it is seen that, in order to obtain simple bending with z the neutral axis, the vertical plane in which the trans- verse loads act should pass through the point O, which is called the cen- J Bu ter of twist. At any other position of h this plane, bending of the beam will 2 je |_. be accompanied by twist, and the fed [ stresses will no longer follow the simple : law in which o, is proportional to y by R and hence does not depend entirely iia ie upon the coordinate x. In the case of an angle section (Fig. 39) the shearing stress T at points along mz will be in the direction shown, and will be equal to # ™ The same method of calculating these stresses as in the case of channel sections is used.54 STRENGTH OF MATERIALS Vv re-7 f ya, in which the integral represents the moment of the shaded area with respect to the z axis. These shearing stresses yield a resultant force in the direction shown in Fig. 39, 4 equal to _ Vet © 3LN2 A force of the same magnitude will also be obtained for the lower flange. The resultant of these two forces is equal to V and passes through the point of intersection of the middle lines of the flanges O, which is therefore the center of twist in this case. I¢ is | ££ vt y Fi. 39. Fic, 40. In the case of a “L section, Fig. 40, assuming simple bending in a vertical plane and proceeding as in the case of an |_] section, we find that the shearing forces R in both flanges have the same direction. Their resultant goes through the centroid C: By geometrically adding this resultant to the vertical shearing force Y, we obtained the direction of the inclined plane in which the transverse forces must be applied to produce simple bending of the beam in the vertical plane. Point C is the center of twist in this case. Assuming that the cross sections which were discussed above belong to cantilever beams fixed’ at one end and loaded by a concentrated force P at the other end, we may conclude that if the load P is applied at the center of twist, it producesPROBLEMS IN BENDING OF BEAMS 55 only bending of the cantilever without any twist. By using the reciprocal theorem (see p. 330, Part I) it can at once be concluded that a torque applied at the same end in the plane perpendicular to the axis of the cantilever and producing torsion of the cantilever will not produce any deflection of the center of twist. Hence during such torsion each cross section of the cantilever is rotating with respect to the axis passing through the center of twist and parallel to the axis of the beam. The method of determining the position of the center of twist shown above on several simple examples can be gener- alized and extended to embrace cases of non-symmetrical cross sections of thin-walled members, provided the thickness of the material is so small that the distribution of shearing stresses over the thickness can be taken with sufficient accuracy to be uniform A further discussion of this problem is given in Art. 53 (p. 292). When all the dimensions of a cross section are of the same order, the problem of determining the center of twist becomes more complicated; exact solutions of this problem exist in only a few cases. 9. Effective Width of Thin Flanges.——The simple bend- ing formula (see Eq. 55, p. 90, Part 1) shows that bending stresses in a beam are proportional to the distance from the neutral axis. This conclusion is correct so long as we are dealing with beams, the cross-sectional dimensions of which % The problem of determining the center of twist has been discussed by several authors. See, for cxample, A. A. Griffith and G. I. Taylor, Technical Reports of the Advisory Committee for Aeronautics, England, Volume 3, p. 950, 1917, R. Maillart, Schweiz. Bauz., vol. 77, p. 1973 vol. 79, p. 254 and vol. 83, p. 111 and p. 176. C. Weber, Zeitschr. f. angew. Math. u. Mech., vol. 4, 1924, p- 334 A. Eggenschwyler, Proc. of the Second Internat. Congress for Appl. Mech. Ziirich, 1926, p. 434- In recent time the problem became of importance in airplane design. The review of the corresponding literature is given in a paper by P. Kuhn, Techn. Notes, Nat. Adv. Comm., no. 69. 4 See paper by M. Seegar and K. Pearson, London, Roy. Soc. Proc. (ser. A), vol. 96, 1920, p. 211, and the writer’s paper, London Math. Soc. Proc. (ser. 2), vol. 20, 1922, p. 398. See also “Theory of Elas- ticity,” 1934, p. 3or.56 STRENGTH OF MATERIALS are small in comparison with their length and so long as we are considering points at a considerable distance from the ends. In practical applications we sometimes use beams with Pies siigh tet NI ‘ i t i asl i } @) ¢t i 7 7 q @) Ih 1 wide flanges, to which the ele- mentary beam formula cannot Lael 1 be applied with sufficient ac- i curacy. Take, as an example, r 4 b>] be x ——+| the case of a beam consistingof arib and a wide flange shown in Fig. 41. Assuming that the beam is simply supported at the ends and loaded in the middle plane xy, we observe that there are shearing stresses acting between the flanges and the rib at the surfaces of contingency mm, Fig. 414, and directed as shown in Fig. 414. It is seen that these stresses tend to reduce the deflection of the rib, to make it stiffer. At the same time they produce compression of the flanges. Con- sidering a flange at one side of the rib as a rectangular plate submitted to the action of shearing forces along one edge, Fig. 41c, we see that the compressive stresses will not be uniformly distributed along the width of the flange, and a rigorous analysis shows * that the distribution will be such as is indicated by the shaded area, the maximum stress in the flange being the same as in the utmost fibers of the rib. From oy 7 Fis, 41. %The discussion of the rigorous solution, obtained by Th. von Karman, is given in the “Theory of Elasticity,” p. 156, 1934. See also W. Metzer, Luftfahrtforschung, vol. 4, p. 1, 1929. K. Girkmann, der Stahlbau, vol. 6, 1933, p- 98; H. Reissner, Z. angew. Math. Mech., vol. T4, 1934) p- 312; E. Reissner, Der Stahlbau, vol. 7, 1934) p 206; E. Chwalla, Der Stahlbau, vol. 9, 1936, p- 73-PROBLEMS IN BENDING OF BEAMS 37 this non-uniformity of stress distribution it can be concluded that in applying to the beam in Fig. 41@ the simple beam formula for maximum bending stress we must use a somewhat reduced width 2dz instead of the actual width 24 of the two flanges, in order to obtain the correct value of the maximum stress. This reduced width, usually called the effective width, can be calculated if the compressive stress distribution, shown by the shaded area in Fig. 41¢, is known. It is only necessary to make the area of the rectangle, indicated in the figure by the dotted lines, equal to the shaded area. Its magnitude, 2X2, usually varies along the span of the beam, for it depends on the proportions of the beam and also on the shape of the bending moment diagram. In the particular case when the width of the flange is very large, say 24 = /, and the bending moment diagram is given by the sine curve: < 7x M = M,sin 7? (a) the reduced width becomes constant and equal to iiggieeeieer (etree ~ rt +H) = #)” where y is Poisson ratio. For » = 0.3 we obtain Ie = 0.363/. (63) Hence, in this particular case the actual beam can be replaced by an equivalent T beam of a constant cross section and with the width of the two flanges equal to 0.363/. Applying to this beam the simple beam formulas, we obtain the same maximum stress and the same flexural rigidity as the actual beam has. Dde In a general case of transverse loading, the bending moment diagram can be represented by a sine series: nex Mz = 2M, sin] (2) in which the coefficients M, can be calculated, in each particular case, from the known formula: % * See Article 7.58 STRENGTH OF MATERIALS M, ee + MTX 3 f Mein “= de, © In the case of a uniform load, for example, we have ae le *) and formula (c) gives p Mu = 3, @ where 7 = 1, 3, 5, °¢'- Having the coefficients M, in the series (4), we obtain the effective width from the rigorous solution, which, in the case of a large width of the flanges, gives L M aye ee ea are eel (64) M,, sin —— x Z mab UE at gue in which 8 = é//dh is the ratio of the area / to the cross-sectional area of the rib, and _ G+HG3-») 4 Taking, for example, the case of a uniformly distributed load and substituting the expression (¢) for M, in formula (64), we find that for various values of the ratio 8 the variation of the effective width along the length of the beam is as shown in Fig. 42. Itisseen that in the middle portion of the span the effective width varies very little and is approximately the same as for a sinusoidal bending moment diagram (see eq. 63): When the effective width is Fo. 42. found from formula (64), the maximum stress and maximum deflection are found by applying simple beam formulas to the equiv- alent beam. We discussed the case in which the flanges of the beam have a very large width. There are also rigorous solutions for the case in which the flanges are not so very wide and for the case of a long k = 0.878 for p = 0.3.PROBLEMS IN BENDING OF BEAMS 59 rectangular slab, reinforced by a system of identical and equidistant ribs. In all these cases the problem is reduced to that of calculating stresses and deflections in an equivalent beam.” 10. Limitations of the Method of Superposition —In dis- cussing the bending of beams (see p. 146, Part I) it was shown that the calculation of the deflections can be greatly simplified by using the method of superposition. This method can always be used if the bending of the beam does not introduce changes in the action of the external forces. For instance, small deflections of beams by lateral loads do not change the bending moment diagrams for these loads, and superposition can be successfully used. But if we have bending combined with axial tension or compression, the deflection produced by the lateral loads changes the action of the axial forces, and the latter produce not only axial tension or compression but also some additional bending. In such cases, as we have seen (see Art. 4), there are some limitations of the method of super- position; we can use this method only with regard to the lateral loads, assuming that the axial force always remains constant. There are other cases in which small deflections of beams may introduce considerable changes in the action of forces. In such cases the method of superposition fails. Some examples of this kind will now be discussed. As a first example let us consider the bending of the cantilever 4B, Fig. 43, if during bending it comes gradually into contact with a rigid cylindrical supporting surface 4C having a constant curvature 1/R and a horizontal tangent at A. It is seen that as long as the curvature of the beam at the end 4, as given by the formula I M Pl 7 ET, ETD 7 *' These rigorous solutions found some application in specifications for concrete slabs reinforced by ribs. In airplane design the fact of non-uniform stress distribution in wide flanges is taken care of by using an approximate theory, the discussion of which can be found in papers by P. Kuhn, National Adv. Committee for Aeronautics, Reports No. 608, 1937, No. 636, 1938. See also H. Ebner, Luftfahrt-Forschung, vol. 14, 1937, p. 93 and vol. 15, 1938, p. 527.60 STRENGTH OF MATERIALS is less than the curvature of the support 1/R, the cantilever will touch the surface 4C only at the point 4, and the de- flection 6 at the end B will be given by the known formula: Pr aS BEL. © ® From the equation I Plt 7” EI,” R © we can obtain the limiting value of the load P, for which the beam begins to come into contact with the cylindrical sup- porting surface beyond the point 4. Let P: = EI,/R be this limiting value of the load; then for P > P; a part AD of the beam will be supported as indi- cated in Fig. 43 by the dotted line. The length x of the un- supported portion of the canti- lever is obtained from the con- dition that at D the curvature, 1/r, of the beam is equal to the curvature of the supporting surface; hence Fro. 43. Pet EI, R’ and we obtain = Be The total deflection at the end B of the cantilever consists of three parts: (1) deflection of the portion DB of the beam as a simple cantilever, which is . _ Pe _ (EL) “= 3ET, 7 3PRP &) 6 (2) deflection owing to the slope at D,PROBLEMS IN BENDING OF BEAMS 61 and (3) deflection representing the distance of the point D from the horizontal tangent at 4, which is 7 — x)? a = Loe" = ( ® Summing up these three parts, we obtain the total deflection: P 1 (EI,)* b= tht a= Te. (2) This expression for the deflection must be used instead of equation (6), if P is larger than thelimiting value P; = EJ,//R. Note that the deflection is no longer proportional to P. If, in addition to P, there is a load Q applied at the end B of the cantilever, the total deflection will not be equal to the sum of the deflections produced by P and produced by @Q if both are considered to be acting alone. Hence the method of super- position does not hold in this case. Ms AH 7 7 2 Lag @ 4] @ Fic. 44. As a second example let us consider the case of a uniformly loaded beam with built-in ends, as shown in Fig. 44. It is assumed that during bending the middle portion of the beam is supported by a rigid horizontal foundation so that along this portion the deflection is constant and equal to 6. It is seen that if the deflection at the middle is less than 6, we have an ordinary case of bending of a beam with built-in ends. The limiting value q1 of the load is obtained from the known equation: 1 gilt 384 ET, = 2) * For an intensity of the load larger than qi, a portion of the62 STRENGTH OF MATERIALS beam will be supported by the foundation as shown in the figure. This part remains straight; there is no bending moment acting in the portion CD of the beam, and the load is balanced by the uniformly distributed reaction. At the ends C and D, however, concentrated reactions X may act on the unsupported portions of the beam. The length a of the unsupported portions of the beam and the magnitude X of the concentrated reactions can be obtained by considering the portion 4C of the beam as a cantilever with a uniform load q and with a concentrated load X at the end. Observing that the cross section at C does not rotate during bending and using equations (94) and (100) from Part I (see p. 147 and p. 149), we obtain ge _ Xd 6EI, ~~ 2ET,’ from which xf. j 3 ) Another equation is obtained from the condition that the deflection at C is equal to 6. Using the known formulas for the deflection of the cantilever, we obtain ga Xa SEI, 3El. 6 (&) Solving equations (j) and (&) we find a= 7? X = 18/95ET. 7. It is immediately apparent that the reaction X is not propor- tional to the load. The numerical maximum of the bending moment, which is at the built-in ends, is obtained from the equation: ®PROBLEMS IN BENDING OF BEAMS 63 which gives eee Cee M. = & = V36ET. (m) Again we see that the bending moment does not increase in the same proportion as the load. Hence the method of superposition cannot be used. Problems 1. Find the deflection of the cantilever shown in Fig. 43 if, instead of force P, there is a uniformly distributed load g. 2. Find an expression for the deflection at the middle of a beam, supported by two identical cylindrical surfaces of the radius R and loaded at the middle, Vig. 45. Fro. 45. Solution. As the load P increases, the points of contact of the beam with the supporting surfaces move inwards and the span diminishes; hence the deflection increases in a smaller proportion than does the load P. The angle a, defining the positions of the points of contact, is found from the condition that at these points the deflection curve is tangent to the supporting surfaces; hence, for small values of a, = P= 2Ra) | Os EI Having a, we obtain the deflection at the middle from equation: _ P= Ra)? , Ret 48El, 2 _3: Solve the preceding problem assuming thar the beam built-in at the points 4 and B.64 STRENGTH OF MATERIALS 4. Solve problem 2 if the load is not at the middle of the span 4B. 5. A long, uniformly loaded beam 4) is supported by a horizontal, rigid Me foundation, Fig. 46. Find the angle a of rotation of the end 4 and Fro. 46. the length » which will be bent by the moment Mp applied at the end. Solution. The length x is found from the equation: coe eee 24El 6ET The angle of rotation at the end 4 is My gx? © SEI” 24ETCHAPTER II CURVED BARS 11. Bending Stresses in Curved Bars.—In the following discussion itis assumed that the center line} of the bar is a plane curve and that the cross sections have an axis of symmetry in this plane. The bar is submitted to the action of forces lying in this plane of symmetry. Let us consider first the case of a bar of constant cross section in pure bending, produced by couples applied at the ends (Fig. 47). The stress distribution for this case is obtained by using the same assumption as in the case of straight bars, namely, that transverse cross sections of the bar originally plane and normal to the center line of the bar remain so after bending? Let ab and cd 1 The center line is the line joining the centroids of the cross sections of the bar, ? This approximate theory was developed by H. Résal, Annales des Mines, 1862, p. 617, and by E, Winkler, Der Civilingenieur, Vol. 4, 1858, P. 2323 see also his book, “Die Lehre von der Elastizitat und Festig- keit,” Prag, 1867, Chapter 1s. Further development of the theory was made by ¥. Grashof, “Elastizitit und Festigkeit,” 1878, p. 251, and by K. Pearson, “History of the Theory of Elasticity,” Vol. 2, part I, 1893, P. 422. The exact solution of the same problem was given by H. Golo- vin, Bulletin of the Institute of Technology at St. Petersburg, 1881. See also C. Ribiare, C. R., Vol. 108, 1889, and Vol. 132, 1901, and L. . Prandtl in the paper by A. Timpe, Zeitschr. f. Math. u. Phys., Vol. 52, 1905, p. 348. The above approximate theory is in good agreement with the exact solution. See “Theory of Elasticity,” p. 58, 1034. 6566 STRENGTH OF MATERIALS denote two neighboring cross sections of the bar and let de denote the small angle between them before bending. As a result of bending, the cross section cd rotates with respect to ab, Vet Adg denote the small angle of rotation. Due to this rotation the longitudinal fibers on the convex side of the bar are compressed and the fibers on the concave side are extended. If n-n denotes the neutral surface, the extension of any fiber a distance ® y from this surface is yAdy and the corre- sponding unit elongation is ydde ¢=Goyae? @ where r Cenotes the radius of the neutral surface and the denominator in eq. (a) is the length of the fiber between the adjacent cross sections before bending. Assuming that there is no lateral pressure between the longitudinal fibers,’ the bending stress at a distance y from the neutral axis is os = puAde () ( — y)do It will be seen that the stress distribution is no longer linear as in the case of straight bars, but that it follows a hyperbolic law as shown in Fig. 47 (c). From the condition that the sum of the normal forces distributed over the cross section is zero, it can be concluded that the neutral axis is displaced from the centroid of the cross section towards the center of curvature of the bar. In the case of a rectangular cross section, the shaded area (Fig. 47, ¢) in tension must equal that in compression; hence the greatest bending stress acts on the concave side. In order to make the stresses in the most remote fibers in tension and in compression equal, it is necessary to use sectional shapes which have the centroid nearer the concave side of the bar. 8 y is taken positive in the direction towards the center of curvature of the bar. 4 The exact theory shows that there is a certain radial pressure but that it has no substantial effect on the stress o, and can be neglected. The lateral pressures in direction perpendicular to the plane of curvature may be of importance in the case of bending of thin shells (see art. 20).CURVED BARS 67 Equation (4) contains two unknowns, the radius 7 of the neutral surface and the angle Ady which represents the angular displacement due to bending. To determine them we must use the two equations of statics. The first equation is based on the condition that the sum of the normal forces distributed over a cross section is equal to zero. The second equation is based on the condition that the moment of these normal forces is equal to the bending moment M. These equations are: Ede [yd foda 2 a7 () Eddy ydA fovea = i fS-™ @ The integration in both equations is extended over the total area of the cross section. The integral in eq. (¢) may be simplified as follows: ef (o- a jaa ee = - frase f2%. © The first integral on the right side of eq. (e) represents the moment of the cross sectional area with respect to the neutral axis and the second, as is seen from eq. (c), is equal to zero. Hence "dA 474% “ in which ¢ denotes the distance of the neutral axis from the centroid of the cross section, Equation (d) then becomes Eddp _M “de ~ de (65) and eq. (4) gives _ __ My oz = Aer — yy x (g) The stresses in the most remote fibers, which are the largest68 STRENGTH OF MATERIALS stresses in the bar are Mh Mh (¢2) max = Tea and (¢2)min = — gece (66) in which 4; and #e are the distances from the neutral axis to the most remote fibers and a and ¢ are the inner and outer radii of the bar. The radius 7 is determined from eq. (c). Several examples of such calculations are shown in the next article. If the depth of the cross section is small in comparison with the radius R of the center line of the bar, y may be neglected in comparison with r in eqs. (c) and (d). Then, from (c), we obtain foza =o, i.e., the neutral axis passes through the centroid of the cross section. From eq. (d) Eddy de Substituting this into eq. (4), My 77 Hence, in the case of a relatively small depth 4, the distribution of the bending stresses ¢, approaches a linear one, and the same equation as used for straight bars can be used to calculate them. From eq. (4) we obtain for a thin bar MRde _ Mas Ade = EL “ETL? (6 5= M. (4) ro o, in which ds denotes the element of the center line between two adjacent cross sections. This equation is analogous to eq. (a), p. 145, Part I, for the straight bars, and is often used in cal- culating the deflections of thin curved bars. Tn a more general case when a curved bar is submitted to the action of any coplanar system of forces in the plane ofCURVED BARS 69 symmetry of the curved bar, the forces acting upon the portion of the bar to one side of any cross section may be reduced to a couple and a force applied at the centroid of the cross section. The stresses produced by the couple are obtained as explained above. The force is resolved into two components, a longitudinal force N in the direction of the tangent to the center line of the bar and a shearing force V in the ‘plane of the cross section. The longitudinal force produces tensile or compressive stresses uniformly distributed over the cross section and equal to N/A. Due to these stresses the center line of the bar undergoes extension or contraction and the angle dy between the two adjacent cross sections changes by the amount ‘ds Ande = ME. 5. (68) The transverse force Y produces shearing stresses and the distribution of these stresses over the cross section can be taken the same as for a straight bar 12. Particular Cases of Curved Bars.—It was shown in the previous article (eq. 66) that bending stresses in curved bars are readily calculated provided the position of the neutral axis is known. In the following examples the calcu- lation of the distance e¢ of the neutral axis from the centroid of the cross section is given for several particular cases. Rectangular Cross Section.—The magnitude of the radius r of the neutral surface is determined from eq. (c) of the previous article, from which dA Jeet @ If we denote by v (Fig. 48) the radius of a shaded element @4, then ver-y or yurr-o ® This assumption is in a good agreement with the exact solution for a narrow rectangular cross section; sce “Theory of Elasticity,” p. 73» 1934.70 STRENGTH OF MATERIALS Substituting into eq. (2), f a a4 a pane from which fo __a =—70° (69) 7 In the case represented in Fig. 48, A = bh,dA = bdo and integration is ex- tended from» = atov = c, where aand ¢ are the inner and the outer radii of the curved bar. Substituting in equation (69), we obtain ail TT — Nr Re 3 | = aR’ (7) 14i(4 mR By using three terms of the series (4), a second approximation is obtained: R hy? em ael +(e) | ©) we obtainCURVED BARS a7 It can be seen that the distance e of the neutral axis from the centroid decreases as the ratio 4/R decreases. For small values of this ratio the distance ¢ is small and a linear stress distribution, instead of a hyperbolic one, may be assumed with good accuracy. In the table below, the values of the maximum stress obtained by assuming a linear stress distri- bution are compared with those obtained on the basis of a hyperbolic distribution (eq. 66). Tawie 4, Companisox or Hyrernoric axp Linear Sraess Disraiautioxs | Error in omax Hyperbolic Linear due to assuming stress distribution stress distribution Freleng | Tmax Smin_ max ] min Ry | ome u/ M/AR M/AR M/AR | M/AR . 1 92 = 44 6 -6 | 35 2 144 =103 2 1 17 3 20.2 =16.1 18 —18 | 10.9 4 26.2 22.2 34 734 92 10 62.0 —58.0 60 —60 32 This indicates that for R/A > 10 a linear stress distribution can be assumed and the straight beam formula for maximum bending stress may be used with suffi- cient accuracy. Trapezidal Cross Section. The length of an elemental strip at distance » from the axis 0-0 (Vig. 49) is cro = ht hi — bs) ©. Substituting into eq. (69), we obtain Fic. 49. 4 -jelbe ene (73) ac dic be Feb tog f= bs)72 STRENGTH OF MATERIALS When 4; = 2, = 4, the above equation coincides with eq. (70) for a rectangle. When J; = 0, the case of a triangular cross section is obtained. -L Cross Section. In this case eq. (69) gives (Fig. 50) — if. ‘a Ah at bof (74) I LE Fic. £0. 1 Cross Section. From eq. (69) (Fig. 51) bifi + bof + bafs caesar eee ep EERE (75) i logn 5 + Os logs + bs loss It may be seen that a suitable choice of proportions in the cases of 1 and I sections will so locate the centroid that eqs. (66) will give the same numerical values for Gmax and omin. Such propor- tions are desirable if the material is equally strong in tension and compression, as, for instance, structural steel. In the previous discussion the distance ¢ from the neutral axis to the centroid of the cross section was determined as the difference R-r. As h/R decreases, this difference becomes small. To de- termine it with sufficient accuracy, r must be calculated very accurately, In order to avoid this difficulty and to obtain e directly, the following method may be used. If we let y; denote the distance of any point in the cross section from the axis through the centroid parallel to the neutral axis, then y; = y + ¢ and eq. (a) for deter- mining the position of the neutral axes becomes (n= edd -f dA Ae R—-y R-y ‘J R=” @CURVED BARS 73 The first integral on the right side represents a modified area of the cross section and can be represented as follows: nda Roy @ where m denotes a number to be determined in each particular case. The second integral on the right side of eq. (¢) can be transformed as follows: dd 1 n ,__4 Reyna (tet) et etm: 6) Substituting (d) and (e) in eq. (¢), we obtain ae m4 ey from which m eo Rh 78) In calculating m from eq. (4) the factor 1/(R — yx) can be expanded in the series See Byway). R= AG rt Rt ) Then g4-if( Bares ceeereeas S573 tt et Rt joa ne ww, mage f (1434+ ) naa. (77) As an example, for a rectangular cross sec- tion 4 = oh, dd = ddyy, and substituting in G7)» 1 pee yt m=Te (14 2484) nan M2 AR X (AV 424) NZ 3\2aR +; aR +7 ry) amine This series converges very quickly and mcan o be calculated very accurately. Substituting m in eq. (76), the distance ¢ is obtained. For a circular cross section (Fig. 52), dd = Wi — yildys. ) and74 STRENGTH OF MATERIALS Substituting into eq. (e), ie VI sao hy R- a i an(e a from which, using the serivs Le as 24R (is) 4 ~#(ze) ae we obtain 1fkAYW tf 4M, 5 A Ne ae wie) tice) +E GR) to This is a rapidly converging series, from which m may easily be calculated. Substitution of m into eq. (76) gives the position of the neutral axis. It can be seen that in calculating m from equation (e) the mag- nitude of m does not change if all elements ¢4 are multiplied by some constant, since in this way the integral on the left side of equation (e) and the area 4 on the right side of the same equation will be increased in the same proportion. From this it follows that the value (78) of m obtained for a circular cross section can be used also for an ellipse with the axes 4 and /i, since in this case each elemental area (/) obtained for a circle is to be multiplied by the constant ratio /y/A. The calculation of the integral on the left side of equation (¢) can sometimes be simplified by dividing the cross section into several parts, integrating for cach part, and adding the results of these integrations. Taking, for example, a circular ring cross section with outer diameter 4 and inner diameter /y, and using equation (g) for the outer and inner circles, we find for the ring cross section: m= an {#[i(&) + i(z) +~ ‘| ; ~ wef (4 ae (4) +--]}- (9) In a similar manner we can develop formulas for the cross sections shown in Figs. so and 51. When m is calculated, we find e from equation (76) and the fatm =CURVED BARS 75 maximum stress from equation (66). But we can also proceed in a somewhat different way and continue to use distance y: from the centroidal axis instead of distance y from the neutral axis. Sub- stituting y, — ¢ instead of y into equation (g) (p. 67), we obtain =Mmr-o Miyn_, o = Ae(R— yi) AR\ no ; (80) where v is the distance of the point from the axis through the center of curvature of the axis of the bar, Fig. 48. Equation (d) is the basis of a graphical determination of the quantity m for cases in which the shape of the cross section cannot be simply expressed analytically. It is seen that in calculating the modified area from eq. (d) every elemental area must be diminished in the ratio yi/(R — 91). This can be done by retaining the width of the elemental strips but diminishing their lengths in the above ratio (Fig. 53). In this manner the shaded area in the figure is obtained. R ay Fie, 53. The difference between the areas CDF and ABC gives the modified area m4. Knowing this, the quantities m and e can readily be calculated. The theory of curved bars developed above is applied in designing crane hooks.* In Fig. 54 is represented the working portion of a hook of a constant circular cross section. It is ©A theoretical and experimental investigation of crane hooks was recently made by the National Physical Laboratory in England. See the paper by H. J. Gough, H. L. Cox and D. G. Sopwith, Proc. Inst. Mech. Engrs., December 1934. The comparison of the theoretical stresses in hooks of a rectangular cross section with experimental results is given in the paper by K. Bottcher, Forschungsarbeiten, Heft 337, Berlin, 1931.76 STRENGTH OF MATERIALS assumed that the vertical force P is passing through the center of curvature O of the axis of the hook. The maximum bending stress occurs in the cross sec- tion perpendicular to the load P. Then, proceeding as explained in the preceding article, we find that on the horizontal cross section of the hook the tensile force P, applied at the cen- ter C of the cross section, and the bend- ing moment M = PRare acting. Com- bining direct and bending stresses and using equation (80) for the latter, we obtain ~P,_M(m _ Pn Fro. 54 w= Gt (1) - RE: Applying this formula to the most remote points, for which n= 2, we find that — P ree A tee . (a2)mox = 4° aan R= Ala § P h (2)min = — amR TDD” (81) It is seen that the numerically largest stress is the tensile stress at the intrados, which is obtained by multiplying the stress P/A by the stress factor: h k= SAR — als) (82) the magnitude of which depends on the ratio #/2R. Using expression (78) for m, we find that & varies from 13.5 to 15.4 as the ratio 4/2R changes from 0.6 to 0.4.7 7 At 4/2R = 0.6 the factor & has its minimum value.CURVED BARS 77 Problems 1. Determine the ratio of the numerical values of gmex and Gwin for a curved bar of rectangular cross section in pure bending if R= Sin. and # = qin. Solution. From eq. (66) the ratio is Ayc/iza. The distance of the neutral axis from the centroid, from eq. (72), is ees 4(4V¥]_ i em apt +2 (4) |= o270in Then fy = 2 — 0.278 = 1.722 in.; Ap = 2 + 0.278 = 2.278. The above ratio is 1.722. X7 gays xg 7 TS 1278 X 3 2. Solve the previous problem, assuming a circular cross section. Answer. Fwax = ins tend Toate e = 0.208 in; Gnin 2.208 3 1.89. 3. Determine the dimensions 4; and 4; of an I cross section (Fig. 51) to make omax and omin numerically equal, in pure bending. Givenfi: = Lin, fp = 2in., fp = Tin. @ = gin, dy = rin, dy + bs = sin. Solution. From eqs. (66), A _ he S== oor ac from which _ 2a _2X3X7_ peso eee 347 = 4.20 in, Substituting into eq. (75), eases eae snes Aree eee 4:20 = Flops 4/3 +1 log, 6/4 + (5 — Fy) log, 7/6” from which bloga 4/3 + (5 — x) logy 7/6 = ot — 1-loga 6/4, 0.2884, + 0.154(5 — d1) = 1.667 ~ 0.406 = 1.261 in, 4, = 3-67 in.; bs = § — 3.67 = 1.33 in.78 STRENGTH OF MATERIALS 4. Determine the dimension 4; of the 1 section (Fig. 50) to make ome and omin numerically equal in pure bending. A= Ving fe = jin, & = Lin, a= 3 in. Answer. by = 3.09 in. 5. Determine omax and comin for the trape- zoidal cross section mn of the hook repre. sented in Fig. 55 if P = 4,500 Ibs., a1 = 1 in., bp = Bin. a@=1} in. c= 5 in. Solution. From eq. (73), Ce r“iExXs Pee xi ESE EEE EeE EEE EEEE EEE Eee BAST tog, 5/14 — 8 — 3a Given 3-750 Sete — 375° _ £ bar 1.580 7 2373 in. be ce +} Tt Fie. 55. ‘The radius of the center line bitty hoo, = at TPR y= 27stin 2.373 — 1.250 = n.; de = 3.75 X 0.361 The bending stresses, from 1,123 in.; es 5 3 = 1.35; M = PR = 12,300 Ibs. ins. eqs. (66), are (c2)max = nea rre = 8,200 Ibs. per sq. in. (¢2)min = — are = — 4,800 Ibs. per sq. in. On these bending stresses, a uniformly distributed tensile stress P/A = 4,500/3-75 = 1,200 Ibs. per sq. in. must be superposed. The total stresses are Ginax = 8,200 + 1,200 = 9,400 Ibs. per sq. in. min = — 4,800 + 1,200 = — 3,600 Ibs. per sq. in. 6. Find the maximum stress in a hook of circular cross section if the diameter of the cross section is # = 1 in., radius of the central axis R = 1 in., and P = 1,000 lbs.CURVED BARS 79 1000 0.785 7. Find omex and min for the curved bar of a circular cross section, loaded as shown in Fig. 56, if k = 4in.,R = 4in.,¢ = 4 in., and P = 5,000 Ibs. Answer. Gaz = 10,650 Ibs. per sq. in., Gmin = — 4,080 Ibs. per sq. in. 8. Solve the preceding problem assuming that the cross section mn has the form shown in Fig. so with the following dimensions a=2in,d=3in,c=gin,&=4in, b= and P Answer. k = 13.9, Fmox = 13-9 = 17,700 lbs. per sq. in. in. e = 4in., 4,000 Ibs. Answer. Omax = 3,510 Ibs. per sq. in.; min = — 1,800 lbs. per sq. in. Fie, 56. 9. Solve problem 7 assuming that the cross section mn is trapezoidal, as in Fig. 49, with the dimensions @ = 2 in. c = 4-1/4 in., d) = 2in., dy = 1 in.,e = 0, and P = 1.26 tons. Answer. Omax = 3.97 tons per sq. in.; omin = — 2.33 tons per sq. in. 13. Deflection of Curved Bars.—The deflections of curved bars are usually calculated by the Castigliano theorem.’ The simplest case is when the cross-sectional dimensions of the bar are small in comparison with the radius of curvature of its center line? Then the change in the angle between two adjacent cross sections is given by eq. (67), analogous to 5 eq. (2), p. 150, Part J, for straight bars, -~@ and the strain energy of bending is P given by the equation (83) ~T in which the integration is extended along the total length s of the bar. Equation (83) is analogous to that (187) * See p. 308, Part I. | *'The case in which the cross-sectional dimensions are not small is discussed in problem 6, p. 87.80 STRENGTH OF MATERIALS (see Part I) for straight beams,’ and the deflection of the point of application of any load P acting on the bar in the direction of the load is aU 3B As an example, take a curved bar of uniform cross section ‘whose center line is a quarter of a circle (Fig. 57), built in at the lower end 4 with a vertical tangent and loaded at the other end by a vertical load P. The bending moment at any cross section mn is M = PRcos y. Substituting in eq. (83), the vertical deflection of the end B is a fo uekee aP J, ~2El. b= 1 aM EF. J, Mop ap Rae x PR -erf” PR cos! ede = = Fr If the horizontal displacement of the end B is needed, a horizontal imaginary load Q must be added as shown in the figure by the dotted line. Then M = PR cos ¢ + QR(t — sin g) and OM 3g The horizontal deflection is aU) _ a8 (*PMPRde _ aM oe (50)... i ol TET, ~ El. Jy Mae Rae. Q =o must be substituted in the expression for M, giving PR 2ET. = RU — sin ¢). i= arf” PR® cos o(1 — sin ey = Thin Ring.—As a second example consider the case of a thin circular ring submitted to the action of two equal and 3 The strain energy due to longitudinal and shearing forces can be neglected in the case of thin curved bars. See p. 84.CURVED BARS 81 opposite forces P acting along the vertical diameter (Fig. 58). Due to symmetry only one quadrant of the ring (Fig. 58, 4) need be considered, and we can also conclude that there are no shearing stresses over the cross section mz, and that the tensile force on this cross section is equal to P/2. The magnitude of the bending mo- ment My acting on this cross section is statically indeter- minate, and we shall use the Castigliano theorem to find it. It may be seen from the con- dition of symmetry that the cross section mn does not rotate during the bending of the ring. Hence the displacement cor- responding to My is zero and Fic. 58. mM, = (2) in which U is the strain energy of the quadrant of the ring which we are considering. For any cross section my at an angle ¢ with the horizontal the bending moment is * M = Mo— © RG - cos 0) (d) and aM. aM, ~ * Substituting this into expression (83) for the potential energy and using eq. (a), we find __d 7? MRde_ 1 tt) dM ona SES 7 ET. , Mana, Rde P -f [ My —= RG — cos o) | Rée, 4 Moments which tend to decrease the initial curvature of the bar are taken as positive.82 STRENGTH OF MATERIALS from which PR M = PR(1-2 = 0.182PR. «84) Substituting into eq. (J), M =? (cos ¢ -2). @ The bending moment at any cross scction of the ring may be calculated from this expression. The greatest bending moment is at the points of application of the forces P. Substituting g = 7/2 in eq. (c), we find M= oe — 0.318PR. (85) The minus sign indicates that the bending moments at the points of application of the forces P tend to increase the curvature while the moment M, at the cross section mn tends to decrease the curvature of the ring, and the shape of the ring after bending is that indicated in the figure by the dotted line. The increase in the vertical diameter of the ring may be calculated by the Castigliano theorem. The total strain energy stored in the ring is “2 MARde Gein f 2ET, ? in which M is given by eq. (c). Then the increase in the vertical diameter is -- -tS” Mop "oe pe (see i)e PR -(2-2 EI, moag ey (86)CURVED BARS 83 For calculating the decrease of the horizontal diameter of the ring in Fig. 58, two oppositely directed imaginary forces Q are applied at the ends of the horizontal diameter. Then by calculating (@U/0Q)q-0 we find that the decrease in the horizontal diameter is eH 3 7 Thick Ring—When the cross sectional dimensions of a curved bar are not small in comparison with the radius of the center line, not only the strain energy due to bending moment but also a that due to longitudinal and shearing forces N must be taken into account. The change \ bw in the angle between two adjacent cross sec- i tions (Fig. 59) in this case, from eq. (65), is v P i Mi Mag Mas dkEe~ AEeR and the energy due to bending for the element between the two adjacent cross sections is v Adg = Fic. 59. Meds 2AEeR @) The longitudinal force N produces an elongation of the element between the twe adjacent cross sections in the direction of the center line of the bar equal to Mds/4E and increases the angle dp by Nds/AER (2q. 68). The work done by the forces N during their applicatioa is N%ds/24E. During the application of the forces N the couples M do the negative work — MNds/AER. Hence the total energy stored in an element of the bar during the appli- cation of the forces N is dU, = 3Mddp = Neds MNds Ws = 5 7E ~ AER” © The shearing force / produces sliding of one cross section with respect to another, of the amount aVds/4G, where a is a numerical factor depending upon the shape of the cross section (see p. 170, Part I). The corresponding amount of strain energy is 2dG” (f)84 STRENGTH OF MATERIALS Adding (d), (e) and (f) and integrating along the length of the bar, the total energy of strain of a curved bar becomes - Me N? MN ab? a f (sais + 4B ~ GER T ve) ae) Let us use this equation to solve the problem represented in Fig. 57. Taking as positive the directions shown in Fig. 59, we have M=-PResy; N=~Pcosg; V=Psing, where R is the radius of the center line. Substituting in equation (88) and using the Castigliano theorem, we find the vertical de- flection of the point B to be dU — PR 7" ( Rcos® Pe did, (Este ~ cost e+ F sint e ) de aPR(R , ak = was ieGeit *): If the cross section of the bar is a rectangle of the width 4 and depth 4, using for ¢ the approximate value (71) and taking @ = 1.2, E[G = 2.6, PR Re a= RE +242); When A is small in comparison with R, the second term in the parenthesis representing the influence on the deflection of N and V can be neglected and we arrive at the equation obtained before (see p. 80). The above theory of curved bars is often applied in calculating stresses in such machine elements as links and eye-shaped ends of bars (Fig. 60). In such cases (q a difficulty arises in determining the load distribution over the surface of the curved bar. This distribution r( Q depends on the amount of clearance between the bolt and the curved bar. A satisfactory solution of the problem may be expected only by wn ® Fic. 60,CURVED BARS 85 combining analytical and experimental methods of investi- gation.” In arecent paper ™ the particular case of an eye-shaped end of rectangular cross section, Fig. 604, was investigated. In this discussion it was assumed that there are no clear- ances and that the bolt is absolutely rigid. The maximum tensile stress occurs at the intrados in the cross sections perpendicular to the axis of the bar, and its magnitude can be represented by the formula: 8P Bb? (g) Tmax = O in which P is the total tensile force transmitted by the bar, ais the numerical factor depending on the ratio 7./r; between the outer and the inner radii of the eye, and ¢ is the thickness of the eye perpendicular to the plane of the figure. For ra/r; equal to 2 and 4 the values of @ are respectively 4.30 and 4.39. The values obtained from formula (g) are in satisfactory agreement with experiments." Problems 1, Determine the vertical deflection of the end B of the thin curved bar of uniform cross section and semicircular center line (Fig. 61). 1 For a theoretical investigation of the problem, see H. Reissner, Jahrbuch der wissenschaftlichen Gesellschaft fir Luftfahrt, 1928; also J. Beke, Der Eisenbau, 1921, p. 233; Fr. Bleich, Theorie und Berechnung der eis. Briicken, 1924, p. 256; Blumenfeld, V. D. I., 1907, and Bau- mann, V. D. 1, 1908, p. 397 Experiments have been made by Dr. Mathar, Forschungsarbeiten nr. 306, 1928; see also D. Ruhl, Disserta- tion, Danzig, 1920; Preuss, V. D. 1., Vol. 5, 1911, p. 21733 M. Voropaeff, Bulletin of Polytechnical Institute, Kiew, 1910; E. G. Coker, “Photo- elasticity,” Journal of the Franklin Inst., 1925. 18H. Reissner und Fr. Strauch, Ingenieur Archiv, vol. 4, 1933, P- 481. 44 See G. Bierett, Mitt. d. deutsch. Mat.-Prif.-Amtes, 1931, sonder- heft 5. The photo-elastic investigation of the eye-shaped end was made by K. Takemura and Y. Hosokawa, Rep. Aero. Inst., Tokyo, vol. 18, 1926, p. 128. See also M. M. Frocht and H. N. Hill, Journal of Applied Mechanics, vol. 7, p. 5. In the latter paper the effect of clearance between the bolt and the hole is investigated.86 STRENGTH OF MATERIALS Solution. The strain energy of bending is U= f M*Rde -[ PPR — cos ¢)*Rdy e)Rde — 2EL, oer eee ‘The deflection at the end is dU _ PR 7 PR 6= B-Gr el (1 = cos g)'de ue. Fic. 61, Fic. 62. 2, Determine the horizontal displacement of the end B in the previous problem. Answer. _2PR “ET, 3. Determine the increase in the distance between the ends ./ and B of a thin bar of uniform cross section consisting of a semi- circular portion CD and two straight portions 4C and BD (Fig 62), Answer. aP 3-25 +K(Z p+ =e +R) | ? A eS bid L ? P f by ep ) 7 P Fo. 63.CURVED BARS 87 4. A link consisting of two semicircles and of two straight portions is submitted to the action of two equal and opposite forces acting along the vertical axis of symmetry (Fig. 63). Determine the maximum bending moment, assuming that the cross sectional dimensions of the link are small in comparison with the radius R. Solution. Considering only one quarter of the link (Fig. 63, 4), we find the statically indeterminate moment M, from the condition that the cross section, on which this moment acts, does not rotate. Then au dMo =o. Noting that for the straight portion M = M, and that for a curved portion M = My — (P/2)R(1 — cos y) and taking into considera- tion the strain energy of bending only, we find : dud iu ff meas 0 <2 dMy dMy' 20-1, PR y + aT, [oe - Fo — cos ») | Rdg ; from which PR oan 2 w+nR For / = 0, this coincides with eq. (84) obtained before for a circular ring. The largest moment is at the points of application of forces P and is equal to M = My = My". §. Solve the previous problem, assuming that forces P are applied as shown in Fig. 63 (c). Answer. The bending moment at points 4 is M, PR(w — 2) +2R/+ 2 ore aR + al : For / = 0, the equation coincides with that for a circular ring. For R = 0, M; = P1/4 as for a bar with built-in ends. 6, Determine the bending moment Ms and the increase in the vertical diameter of the circular ring shown in Fig. 58, assuming that the cross section of the ring is a rectangle of width 4 and88 STRENGTH OF MATERIALS depth A, which dimensions are not small in comparison with the radius R of the center line. Solution. If we use eq. (88) for the potential energy and eq. (4) for the bending moment, the equation for determining Mo is dU mie AL y aM” Jo (Ge- :) de =o, from which 2 zR Comparing this with eq. (84) we see that the third term in the parenthesis represents the effect of the longitudinal force and of the non-linear stress distribution. The magnitudes of the errors in using the approximate eq. (84) instead of the above accurate equation are given in the table below. Rlh= 1 LS 2 3 e[R= 0.090 0.038 0.021 0.009 Error in % = 15.8 6.7 3-7 1.6 It can be seen that in the majority of cases the approximate eq. (84) can be used for calculating My and that the error is substantial only when & approaches. ot becomes larger than R. The increase in the vertical diameter of the ring is obtained from equation dU a= Using eq, (88) for U and substituting in this equation PR P P. M=My—=-(— cose); N= 008 9% = -Fsiny, we find PRo{roo2 e ef 2 e © ma Ee 1 Fels 26 7 a) +826 -8) -3]+ Fae} Comparison with eq. (86) shows that the effect of the longitu- dinal and of the shearing forces on the magnitude of 3is usually very small.* 1 A more accurate solution of the problem shown in Fig. 58 is given by the author; see Bulletin of the Polytechnical Institute in Kiew, 19105 see also Phil. Mag., Vol. 44, 1922, p. 1014, and “Theory of Elasticity,” p. 119, 1934-. This solution shows that the above theory, based on the assumption that cross sections remain plane during bending, gives very satisfactory results.CURVED BARS 89 7. Determine the bending moments in a thin ring with two axes of symmetry submitted to the action of a uniform internal pressure p. Solution. Consider one quadrant of the ring (Fig. 64), with semi-axes a and b, If Mp represents the statically indeter- minate moment at 7, the bending moment at any cross section C with coordinates x and y is M= Mo~ pala—x) +24 4 2% = My Oh PE BF, HM Sty t+ Z: ®) Substituting into equation dU/dMs = 0, we find (16-8). +2044) os in which s denotes the length of the quadrant of the ring, L= f yds and y= fi ads. 0 lo : My P+ 1). () Then If the ring has the shape of the link shown in Fig, 63, with a = R and b = 7 + R, we obtain sab-a4%; La ze ~ a + Eb — ap tie +20 — a)s = @- ae 4%. Substituting into eq (A), : My = PS attonl} (=a) Fatt 30%(b—a) + Zab—a)} The bending moment at any other cross section may now be ob- tained from eq. (g).90 STRENGTH OF MATERIALS For an elliptical ring the calculations are more complicated.!* Using the notations I, + I, = aa*b, Mo = — Spa’, the moment at B (Fig. 64) = My = ypat, and the values of the numerical factors a, B and +, for different values of the ratio a/4, are as given in the table 5 below: Taste 5. Constants ror Caxcvnatine Exurricat Rises alb = 1 0.9 8 °7 0.6 25 Og, 0.3 « 1.663 | 1.795 | 1.982 | 2.273 | 2.736 | 3.559 | 5.327 8 ° 0.057 | 0.133 | 0.237 | 0.391 | 0.629 | 1.049 | 1.927 Y : ° 0,060 | 0.148 | 0.283 | 0.998 | 0.870 | 1.576 | 3.128 8. A flat spiral spring (Fig. 65) is attached at the center to a spindle C, A couple My is applied to this spindle to wind up the spring. It is balanced by a horizontal force P at the outer end of the spring 4 and by the reaction at the axis of the spindle. Establish the relation between M, and the angle of rotation of the spindle if all the dimensions of the spring are given. It is assumed that the angle of twist is not large enough to cause adjacent coils to touch each other. Solution, Taking the origin of co- ordinates at 4, the bending moment at any point of the spring at distance y from the force P is M= Py. The change in the angle between two adja- cent cross sections at the point taken, from eq. (67), is Mas _ Pyds Ade = "Er, = Er,” Fro. 65. The total angle of rotation of one end of the spring with respect to the other during winding is (4) "See J. A. C. H. Bresse, Cours de Méchanique appliquée, 3d ed., Paris, 1880, p. 493. Sce, also, H. Résal, Journal de Math. (Liouville) (3), Vol. 3, 1877; M. Marbec, Bulletin de I’Association Technique Maritime, Vol. 19, 1908; M. Goupil, Annales des Ponts et Chaussées, Vol. 2, 1912, p. 386, and Mayer Mita, V. D. L, Vol. 58, 1914, p. 649. W. F. Burke, Nat. Adv. Com. Aeron., Techn. Notes, 444, 1933-CURVED BARS 91 The integral on the right side of this equation represents the mo- ment of the center line of the spring with respect to the » axis. This moment is obtained by multiplying the total length s of the spiral by the distance of its center of gravity from the x axis. In the usual case, it is sufficiently accurate to take this distance equal to r, the distance from the center of the spindle to the force P. Then, from eq. (4), Mos EL” ® If the end 4 is pin-connected, the turning moment Mo applied at ¢ produces a reactive force P at fixed end / of the spring. As long as the thickness of the spring is very small and the number of windings of the spiral is large and the coils do not touch, the above assumption that the force P remains horizontal can be considered as sufficiently accurate; hence eq. (/) holds.” 9. Assuming that the spring represented in Tig. 65 is in an unstressed condition and pin-connected at 4, determine the maxi- mum stress produced and the amount of energy stored in the spring by three complete turns of the spindle. ‘Take the spring to be of steel, } inch wide, #y inch thick and 120 inches long. Solution. Substituting the above figures into eq. (/), 120 X 40? X 12 Phrase jo X 1088 XF > from which My = 3.07 Ibs. ins. The amount of energy stored is "Mids Pt Pe srt v= [oe ~ 2S yas = e+ = 36.1 Ibs. in. The maximum bending stress is at point B, where the bending moment can be taken equal to 2Pr = 2Ms, then Omax = 3.07 X 2 X 40? X 6 X 2 = 118,000 lbs. per sq. in. 27 A more complete discussion of the problem is given in the book by A. Castigliano, “Theorie d. Biegungs-u. Torsions Federn,” Wien, 1888. See also E. C. Wadlow, “Engineer,” Vol. 150, p. 474, 1930, and J. A. Van den Brock, Trans. A. S. M. E., Vol. 53, p. 247, 1931+92 STRENGTH OF MATERIALS 10. A piston ring of a circular outer boundary has a rectangular cross section of constant width 4 and of a variable depth & (Fig. 66). Determine the law of variation of the depth 4 to obtain a ring which, when as- sembled with the piston in the cylinder, pro- y duces a uniformly distributed pressure on the cylinder wall. Solution. Let r denote the radius of the cylinder, and r + 6 the outer radius of the ring in the unstrained state. An approxi- petee: mate solution of the problem is obtained by using the outer radius of the ring instead of the variable radius of curvature of its center line. Then, by using eq. (67), the change in curvature due to bending is ddg 11M a> +b ET (a) The bending moment M produced at any cross section mz of the ring by the pressure p uniformly distributed over the outer surface of the ring is y = — aphrisint 2. (3) If we substitute this into (@) and take 4/8/12 for J and use 5/72 instead of (1/r) — 1/(r + 4) (for a small 5), then the following equa- tion for calculating 4 is obtained: Brat 22, © 2 from which Ba BE spe @ Letting g = 7, the maximum value of 4°, denoted by A:', is 24r4 wo BE g ‘The maximum bending stress at any cross section mn is M — 11pr* sin? (p/2) a ”) From (f) and (d) it may be seen that the maximum bending stress occurs at ¢ = 7, i.e., at the cross section opposite to the slot of theCURVED BARS 93 ring. Substituting 4 = Ao and gy = 7 in eq. (f), 12pr? ouae = EE, ® from which ho can be calculated if the working stress for the ring and the pressure p are given. The value of 6 is found by substi- tuting Ao into eq. (e). It may be noted that if two equal and opposite tensile forces P be applied tangentially to the ends of the ring at the slot they produce at any cross section mm, the bending moment — Pr(i — cos ¢) = — 2Pr sin? ¢/2, i.e, the bending moment varies with ¢ exactly in the same manner as that given by eq. (4). Therefore, if the ends of the open ring be pulled together, and in this condition it be machined to the outer radius r, such a ring will, when assembled, produce a uniform pressure against the wall of the cylinder." Determine, for example, 6 and fo for a cast iron piston ring if r = 10 in, ow = 4,200 lbs. per sq. in., p = 1.4 Ibs. per sq. in., and E = 12 X 108 Ibs. per sq. in. Substituting in eq. (g), we find Jy = 0.632 in. From eq. (¢), = 0.111 in. 11. Derive formula (87), given on p. 83. 12. Experience shows that the insertion of a stud in a chain link considerably increases its strength. Vind the bending moment M, at the points of application of the loads P and the axial com- pressive force 2H in the stud for the link shown in Fig. 63. Solution. Since in the case of a stud the horizontal cross section, Fig. 634, does not move horizontally and does not rotate, the statically indeterminate quantities My and H are found from the equations: ou a WU OM eee ce Are from which m= "Ba - a) and H= 18 This theory was developed by H. Résal, Annales des mines, Vol. 5 P- 38, 18743 Comptes Rendus, Vol. 73, p. $42, 1871. See also E. Reinhardt, V. D. I, Vol. 45, p. 232, 1901; H. Friedmann, Zeitschr. d. Osterreich. Ing. u. Arch. Verein, Vol. 60, 'p. 632, 1908, and V. D. 1, Vol. 68, p. 254, 1924.o4 STRENGTH OF MATERIALS where = MH Dnt + bmi + 124 = 2m + 48(n = 3)] mt + 4mm + 48m + 4mm +24 — 8)” pe y2(m + 2)C(r — 2)m + 2(4 — 7)] m* + anm® + 48m? + 242m + 24(n? — 8)? Pp ai maR 13. Find bending moment M) and tensile force H in the cross section 4 of the symmetrically loaded circular ring shown in Fig. 67. Answer. H = P7~—tana, 2h \ soe on Mo = sagee ee + sec a ani = (r— a) tana). 14. Arch Hinged at the Ends.—Figure 68 shows an arch with hinged ends at the same level carrying a vertical load. ‘The vertical components of the reactions at 4 and B may be determined from equations of equilibrium in the same manner as for a simply supported beam, and the horizontal com- Fic. 68. ponents must be equal afd opposite in direction. The magnitude H of these components is called the thrust of the arch. It cannot be obtained statically, but may be de- See paper by H. J. Gough, H. L. Cox and D. G. Sopwith, loc. cit. p. 75.CURVED BARS 95 termined by use of the theorem of Castiglano. In the case of a flat arch, the two last terms in the general expression (88) for the strain energy can be neglected and for usual proportions of arches the product eR can be replaced by the moment of inertia J, of the cross section. The equation for calculating His then dU_d f*{ M N? yaad, (si7,+ Fe) a=. (a) The bending moment at any cross section mm of the arch can be represented by M = My — Hy, ® in which M, is the bending moment calculated for the corre- sponding section of a simply supported beam having the same load and the same span as the arch. The second term under the integral sign of eq. (4) represents the strain energy due to compression in the tangential direction and is of secondary importance. A satisfactory approximation for flat arches is obtained by assuming this compression equal to the thrust H. Substituting expression (4) and N = // in eq. (a) we obtain ~ [apes (My — Hy)yds pe + [oe Hds _ from which * Moyds 7 (89) ee te EL, * Wor an arch of constant cross section, using the notation k = I,/A, eq. (89) becomes H= (90) [sas +R The second term in the denominator represents the effect of96 STRENGTH OF MATERIALS the shortening of the center line of the arch due to the longi- tudinal compression. In many cases it is small and can be neglected. Then Moyds fire (91) yds Take, for example, the case of a parabolic arch carrying a continuous load uniformly distributed along the length of the span with a center line given by the equation: H ye aie 8) ‘ @ Then My = Ex - 2). @ Substituting (c) and (d) into eq. (91), we obtain ip v : (e) The actual thrust, H, will be less than that obtained from eq. (e). To give some idea of the possible error AH, the ratios (AH)/H for various proportions of arches are given in the Table 6 below.” In calculating this table the whole expres- H= TaRLE 6 = 0.1771 | 0.0513 | 0.0235 | 0.0837 | 0.0224 | o.0101 | 0.0175 0.00444 | 0.00198 ‘ %See author’s paper, “Calcul des Arcs Elastiques,” Paris, 1922. Béranger, Ed.CURVED BARS 97 sion (88) for strain energy was used and it was assumed that for any cross section of the arch do _ Els cos yg” cos g’” = where 4) and Ely are respectively the cross sectional area and the flexural rigidity of the arch at the top, g is the angle between the cross section and the y axis and / is the depth of the cross section at the top. Equation (e) was used in calcu- lating the value of H in the ratio AH/I7. This shows that the error of equation (e) has perceptible mag- nitude only for flat arches of considerable thickness. As the supports of the arch are a fixed distance apart, a change in the temperature may produce appreciable stresses in the structure. To calculate the thrust due to an increase in temperature of ¢ degrees, we assume that one of the supports is movable. Then, thermal expansion would increase the span of the arch by /at, where « is the coefficient of thermal ex- pansion of the material of the arch. The thrust is then found from the condition that it prevents such an expansion by producing a decrease in the span equal to a/t. Using the Castigliano theorem, we obtain aU ad (*( NP ait = an, (ser, + syn) 4 = # Taking only the thermal effect and putting My = oand N = H, we obtain from (f) It i ds” (92) , EI. + J, AE H= A more detailed study of stresses in arches may be found in books on the theory of structure. 4 Johnson, Bryan and Turneare, “Modern Framed Structures,” Part II. See also Weyrauch, “Theorie d. Elastischen Bogentrager”; E. Mérsch, “Schweizerische Bauzeitung,” Vol. 47.98 STRENGTH OF MATERIALS 15. Stresses in a Flywheel.—Due to the effect of the spokes, the rim of a rotating flywheel undergoes not only extension but also bending. We take as the free body a portion of the rim (Fig. Fie. 69. 69, 4) between two cross scctions which bisect the angles between the spokes. Let R = the radius of the center line of the rim he cross sectional area of the rim he cross sectional area of a spoke moment of inertia of the cross section of the rim 2a = the angle between two consecutive spokes he weight of the rim per unit length of the center line he weight of a spoke per unit length w = the angular velocity of the wheel. From the condition of symmetry there can be no shearing stresses over the cross sections 4 and B and the forces acting on these cross sections are reducible to the longitudinal force No and the bending moment Mo. _ If X denotes the force exerted by the spoke on the rim, the equation of equilibrium of the portion 4B of the rim is : 2Nosina + X ~ 2RsinaZat = 0, from which ee for ~ 2sina” @ The longitudinal force N at any cross section mm is N = Nocos ¢ + Bar sins PER _ Xese £ £ 2sinaCURVED BARS 99 The bending moment for the same cross section ae M= Mom NoR(1~cos 9)-+ a sint f =m = Bint £ (.) 2 The force X and the moment a cannot be pie from the equations of statics but are calculated by use of the theorem of least work. The strain energy of the portion 4B of the rim is” __. (*MRde N'Rde uaa aEL +? J, 2Ed~ @) The tensile force Ni at any cross section of the spoke at distance r from the center of the wheel is ** =x4H F(R = r); hence the strain energy of the a is © The equations for calculating My and X are a aM (Ui + U2) = 0, f) a gy (Ui t U2) = 0. © Substituting (d) and (¢), we obtain, from eqs. (f) and (g), (93) (94) in which flo) = (22842), 1 sin2a |, « 1 fla) = h ( 4 +2)-53 2 Tt is assumed that the thickness of the rim is small in comparison with R and only the energy of the bending and tension is taken into account. ™ The length of the spoke is taken equal to R. In practice it wil be somewhat less than R,100 STRENGTH OF MATERIALS several values of the functions fi and fs, for various numbers of spokes, are given in the table 7 below. Taser 7 ” 4 8 Sila) 0.643 °, 1274 Sle) o.0068 | 0.00169 | 0.00076 From this table the force X in the spoke is determined from eq. (94) and the bending moment Mp from eq. (93). Then the longitudinal force and bending moment for any cross section mz of the rim may be found from eqs. (a), (4) and (c).2# Take, as an cxample, a steel flywheel making 600 r.p.m., with radius R = 60 in., cross section of the rim a square 12 X 12 sq. in., and with six spokes of cross sectional arca 41 = 24 sq. in. The rim is considered as a rotating ring which can expand freely; then the tensile stress due to centrifugal force is, from eq. (15), Part I, oo = 0.106 X w X R? = 0.106 X 62.8? X 5? = 10,450 Ibs. per sq. in. In the case of six spokes, a = 30°, fi(a) = 0.957, frla) = 0.00169. Then the force in each spoke is, from eq. (94), og WR 300 X 0.00169 + 0.957 $6 93 The longitudinal force for the cross section bisecting the angle between the spokes is, from eq. (2), The bending moment for the same cross section, from eq. (93), is : Re M=- oa EE. 4 The above theory was developed by R. Bredt, V. 1. 1. Vol. 45 (1901), p. 267, and H. Brauer, Dinglers Polytechn. Journ., 1908, p. 353, see also J. G. Longbottom, Inst. Mech. Eng. Proc., London, 1924, P. 433 and K. Reinhardt, Forschungsarbeiten, nr. 226, 1920. A similar problem arises when calculating stresses in retaining rings of large turbo-generators, see E, Schwerin, Electrotechn. Ztschr. 1931, p. 40.CURVED BARS 101 The maximum stress at this cross section is oma = = 10,780 Ibs, per sq. in. For the cross section of the rim at the axis of the spoke, eqs. (4) and (c) give 1a RE 2R2 (Nyon = 0.923 3 (M) ua = 0.476 ra The maximum stress at this cross section is Omex = 12,100 Ibs. per sq. in. In this case the effect of the bending of the rim on the maximum stress is small and the calculation of the stresses in the rim, as in a free rotating ring, gives a satisfactory result. 16. Deflection Curve for a Bar with a Circular Center Line.—In the case of a thin curved bar with a circular center line the differential equation for the deflection curve is analo- gous to that for a straight bar (eq. 141, p. 182, Part I). Let ABCD (Fig. 70) represent the center line of a circular ring after deformation and let denote the small radial displace- ments during this deformation. The variation in the curva- ture of the center line during bending can be studied by considering one element mn of the ring and the corresponding element m7, of the deformed ring included between the same radii (Fig. 70, 2). The initial length of the element mn and its initial curvature are dg doe I ds = Rdg; a ~ Rdg = RB . (a) For small deflections the curvature of the same element after deformation can be taken equal to the curvature of the element myn. This latter is given by equation: 1 dg +Adg 5 Ri 7 dsb Ads’ ()102 STRENGTH OF MATERIALS in which dp + Ady denotes the angle between the normal cross sections m, and m of the deformed bar and ds + Ads the ad \\L- te dts Fie. 70. length of the element mm. The displacement z is considered positive if towards the center of the ring and is assumed to be very smal] in comparison with the radius of the ring. Then the angle between the tangent at m, to the center line and the normal to the radius mo is du/ds. The corresponding angle at the cross section 7; is du, au at rr ds. Thea & Ady = F3ds, © In comparing the length of the element mm with that of the element mn, the small angle du/ds is neglected and the length myn, is taken equal to (R — u)dy. Then uds Ads = — udg = - er (d)CURVED BARS 103 Substituting (¢) and (d) into eq. (4), we obtain 1 _ dg au mn al ata from which ROR BY ae The relationship between the change in curvature and the magnitude of bending moment, from eq. (67), for thin bars is I I M RR Er f) The minus sign on the right side of the equation follows from the sign of the bending moment which is taken to be positive when it produces a decrease in the initial curvature of the bar (Fig. 47). From (e) and (/f) it follows that M etR= Er (95) This is the differential equation for the deflection curve of a thin bar with circular center line. For an infinitely large R this equation coincides with eq. (79), Part I, for straight bars. As an example of the application of eq. (95) let us consider the problem represented in Fig. 58. The bending moment at any cross section mm is, from eq. (c), p. 82, m=" (cos - 2), and eq. (9) becomes du, u PR(2 dst t Ri aEI ae OOS?104 STRENGTH OF MATERIALS Pu, _ PR . dg t "= TEIN a ~ O83? The general solution of this equation is RRA Re = doos¢ + Bsing + Fr — qEI? Sine or The constants of integration 4 and B here are determined from the condition of symmetry a =0, for g=o and for which are satisfied by taking PR* B=o; A=- eT Then PR PR PR a yz? sine — gE 8 © For g = oand g = 2/2, we obtain Wee=Fr(E-2)s Wen = FR(E-3). These results are in complete agreement with eqs. (86) and (87) obtained before by using the Castigliano theorem.” 17. Deflection of Bars with a Small Initial Curvature.—If a bar with a small initial curvature is bent by transverse forces only, the deflections may be calculated by the method used for a straight bar. The conditions are quite different, however, if there are longi- tudinal forces in addition to the transverse forces. A small initial curvature makes a great change in the effect of these longitudinal forces on the deflection. The solution of this involved problem may be greatly simplified by using trigonometric series for repre- senting the initial shape of the curve and the deficction due to % Differential eq. (95) for the deflection of a circular ring was established by J. Boussinesq; see Comptes Rendus, Vol. 97, p. 843, 1883. See also H. Lamb, London Math. Soc. Proc., Vol. 19, p. 365, 1888. Various examples of applications of this equation are given in a paper by R. Mayer in Zcitschr. f. Math. u. Phys., Vol. 61, p. 246, 1913.CURVED BARS 105 bending.*® It is assumed, as before, that the curved bar has a plane of symmetry in which the external forces act and the bar is taken as simply supported at the ends. Let yo denote the initial ordinates of the center line of the bar, measured from the chord joining the centroids of the ends, and y, the deflections produced by the exter- nal forces, so that the total ordinates after bending are y= Jot ye (a) Let the initial deflection curve be represented by the series yas bysin Et by sin ® and the deflection produced by the load be 2 Te . TX y= asin pt sin + (c) In this case the same expression (53) for strain energy can be used as for straight bars. Taking the loading shown in Fig. 35, it is necessary in calculating the work done by the longitudinal forces § to replace the quantity \ (see eq. 56) by "Tdi yo P " (dyo\? worn df [Pa — 3 (BY a : oe 2 = FL sadn +E want). (96) wy et This represents the longitudinal displacement of one end of the curved bar with respect to the other during deflection. We proceed as in the case of straight bars (p. 48) and give to the bar an infinitely small additional deflection da, sin (#7x/I). The work done by the longitudinal forces S during this deflection is dQ — Xo) 7 da, The work done by the load P is Psin tat day = $7 (a, + bn)darn. oe l and the increase in strain energy, from eq. (53), is EInt apr Mandan. dan % See author’s paper, Festschrift zum siebzigsten Geburtstage A. Foppl, p. 74.106 STRENGTH OF MATERIALS The equation for calculating ap is 1 on ue Wandas = P sin dan + 55 (an + be)dany from which 2PR sin“ + Snail, a, = Elan’ — Six Substituting into expression (c) and using the notation, eontitt Ele © Cs sin 2 sin 22% appt Sn 7 Sn] Z Z YF ET as isin? desin = te pee +e J. (7) The first member of the right side of eq. (97) represents the deflec- tion of a straight bar (see eq. 58), while the second gives the addi- tional deflection due to the initial curvature. Take, for example, a bar which has an initial deflection Yo = Ssin (xx/l). The maximum deflection is at the middle of the span and is equal to 4. If only the longitudinal forces § act on the bar (P = 0), the deflection at the middle produced by these forces is obtained from eq. (97) by substituting P = 0; 4; = 4; b, = bg = +++ = 0, Then ab sin n=: @ I-a@ The total ordinates of the center line after bending are Pca ab sin ya nt y= oe tb sin = _ me zsin Fe. (68) Due to longitudinal compressive forces S the ordinates of the center line increase in the ratio 1/(1 — a), i.e., the increase of ordinates depends upon the quantity «, which is the ratio of the longitudinal force to the critical force. If, instead of compressive forces, longi-CURVED BARS 107 tudinal tensile forces act on the bar, it is only necessary to sub- stitute —q instead of ain the previous equations. In the particular case in which yo = 4 sin (xx//) the ordinates of the center line after deformation become cd a. y=Tpasiny (99) It is seen that the longitudinal tensile force diminishes the initial ordinates. Taking, for instance, a = I, ie., taking the longitudinal force equal to its critical value, we find y= sin, i.e., such a longitudinal force reduces the initial ordinates of the bar by half. 18. Bending of Curved Tubes.—In discussing the dis- tribution of bending stresses in curved bars (art. 11) it was assumed that the shape of the cross section remains un- changed. Such an assumption is justifiable as long as we have a solid bar, because the very small displacements in the plane of the cross section due to lateral contraction and ex- pansion have no substantial effect on the stress distribution. The condition is very different, however, in the case of a thin curved tube in bending. It is well known that curved tubes with comparatively thin walls prove to be more flexible during bending than would be expected from the usual theory of curved bars.*”_ A consideration of the distortion of the cross section during bending is necessary in such cases.”* Consider an element between two adjacent cross sections of a curved round pipe (Tig. 71) which is bent by couples in the direction indicated. Since both the tensile forces at the ® Extensive experimental work on the flexibility of pipe bends was done by A. Bantlin, V. D. 1, Vol. 54, 1910, p. 45, and Forschungs- arbeiten, nr. 96. Sce also W. Hovgaard, Journal of Math. and Phys., Mass. Institute of Technology, Vol. 7, 1928, and A. M. Wahl, Trans. Amer. Soc. Mech. Eng., Vol. 49, 1927. % This problem for the pipe of circular section was discussed by Th. v. Karman, V. D. 1, Vol. 55, p. 1889, 1911. The case of curved pipes of rectangular cross section was considered by the author; see Amer. Soc, Mech. Eng., Vol. 45, p. 135, 1923-108 STRENGTH OF MATERIALS convex side of the tube and the compressive forces at the concave side have resultants towards the neutral axis, the previously circular cross sections are flattened and become elliptical. This flattening of the cross section affects the strain of longitudinal fibers of the tube. The outer fiber ab takes some position a,; after bending; denote its displace- ment towards the neutral axis by 6. The total elongation of the fiber is ayb) — ab = aby — aye, — (ab — ayes). (a) The angle between the adjacent cross sections ac and 4d is denoted by dg, its variation during bending by Adg, the radius of the center line by R, and the radius of the middle surface of the tube by a. It is assumed that the ratio a/R is small enough that the neutral axis can be taken through the centroid of the cross section. Then, from the figure we obtain ab, — aye, = (a — d)Ady ~ addy. The total elongation of the fiber ab as given by eq. (a) is abdy — ide and the unit elongation is (Rado = The first term on the right side of this equation represents the strain in the fiber due to the rotation of the cross section dd with respect to the cross section ac. This is the elongation which is considered in the bending of solid bars. The second term on the right side of eq. (4) represents the effect of the flattening of the cross section. It is evident that this effect may be of considerable inportance. Take, for instance, R + a = 60in. and6 = 0.02in. Then 5/(R + a) = 1/3,000 and the corresponding stress for a steel tube is 10,000 Ibs. per sq. in. 2 The displacement 8 is considered as very small in comparison with the radius a.CURVED BARS 109 Hence a very small flattening of the cross section produces a substantial decrease in the stress at the outermost fiber ad, A similar conclusion may be drawn for the fiber cd on the concave side of the bend. A change in the direction of the bending Fo. 71. moment causes a change of sign of the normal stresses and as a result, instead of a flattening of the tube in the radial direction, there is a flattening in the direction perpendicular to the plane of Fig. 71 and the fiber ad, due to this flattening, is dis- placed outward. From the same reasoning as above it may be shown that here again the flattening of the cross section produces a decrease in the stress at the most remote fibers. It may therefore be concluded that the fibers of the tube farthest from the neutral axis do not take the share in the stresses which the ordinary theory of bending indicates. This affects the bending of the tube in the same way as a decrease in its moment of inertia. Instead of eq. (67) which was derived for solid curved bars, the following equation must be used in calculating the deflections of thin tubes: MRdp kEI,” in which & is a numerical factor, less than unity, which takes care of the flattening. This factor depends upon the pro- portions of the bend and can be calculated from the following Ade = (100)110 STRENGTH OF MATERIALS approximate formula~” k=1—-——? > (101) tosses (8) in which ¢is the thickness of the tube. This indicates that the effect of the flattening depends only upon the magnitude of the ratio ¢R/a*. As for the effect of the flattening on the stress distribution, Karman showed that, instead of the simple equation for normal bending stresses *' ¢ = My/I, in which y denotes the distance from the neutral axis, the following more complicated equation must be used: o “EE 1-65), 0) sae) sy The maximum stress, obtained from (c), is in which Cmax = kop (102) in which d is the outer diameter of the tube and is a numerical factor which depends upon the proportions of the bend. Several values of k, are given below: Taste 8 | 1.30 | 0.88 * See paper by Th. Karman, loc. cit. p. 107. Te is assumed that R is large in comparison with @ and that a linear stress distribution is a sufficiently accurate assumption.CURVED BARS lil It is seen that when /R/@? is small, the actual maximum stress is considerably greater than that given by the usual theory which neglects the flattening of the cross section. A theory analogous to the above may also be developed in the case of a tube of rectangular cross section.” For example, in the case of a thin tube of square cross section, the coefficient & in eq. (100) is found to depend upon the magni- tude of the ratio be Re’ in which ¢ is the thickness of the wall, R the radius of the center line of the bend and 4 the length of the side of the cross section. Then n I- + 0.0270n | k= (103) For instance, if 4/R = 0.1 and d/t = $0, we obtain 2 = 25 and, from (103), = 0.63. The maximum stress in tubes of rectangular section increases in the same proportion as the flexibility, ie., in the above example the distortion of the cross section increases the maximum stress approximately, by 60 per cent. If a cross section of a curved bar has flanges of a considerable width, again the question of distortion of the cross section becomes of practi- Fo, 72. cal importance. Such a problem we have, for example, when investigating bending stresses at a corner of a rigid frame of an J section, Fig. 724. Considering an element of the frame between the two consecutive cross sections mu and mm, we see that the longitudinal bending stresses o in the flanges give the components in a radial direction which tend to produce bending of the flanges, Fig. 726. This bending results in some diminishing of longitudinal Such a problem occurs, for instance, in the design of a Fairbairn crane, See reference 28.112 STRENGTH OF MATERIALS bending stress o in portions of the flanges at a considerable distance from the web. To take into account this fact, an effective width ab of a flange must be used in using formula (75) for an J section. Naturally the magnitude of the factor a, defining the effective width of the flange, depends on the flexibility of the flanges, which is expressed by the quantity: 4 B= 1.32 Te’ (a) in which ¢ is the thickness of the flange and r its radius of curvature. For the flange at the intrados r = a, and for the outer flange r = c. The calculations show that if 8 < 0.65, the bending of the flanges can be neglected and we can directly apply the theory developed in Art. 11. For larger values of B the formula 2-(j-25)? 0 can be used * for calculating the effective width of a flange. Assume, for example, that we have the width of the flange 6 = 6 in., the corresponding radius r = 8 in., and the thick- ness ¢=1in. Then, from formula (d), we obtain 8 = 2.80 and the effective width of the flange is 0.35 X 6 = 2.1 in. 19. Bending of a Curved Bar out of its Plane of Initial Curvature.—In our previous discussion we have dealt with the bending of curved bars in the plane of their initial curvature. There are cases, however, in which the forces acting on a curved bar do not lie in the plane of the center line of the bar.** * For derivation of this formula sce Dr. Dissertation by Otto Stein- hardt, Darmstadt, 1938. The experiments made by Steinhardt are in satisfactory agreement with the formula. ™ Several problems of this kind have been discussed by I. Stutz, Zeitschr. d. Osterr. Arch.- u. Ing.-Ver., 1904, p. 682; H. Miiller-Breslau, Die neueren Methoden der Festigkeitslehre, 2 ed., 1913, p. 258, and 4th ed., p. 265; and B. G. Kannenberg, Der Eisenbau, 1913, p. 320 The case of a circular ring supported at several points and loaded by forces perpendicular to the plane of the ring was discussed by F. Diister- behn, Der Eisenbau, 1920, p. 73, and by G. Unold, Forschungsarbeiten, nr. 255, Berlin, 1922. The same problem was discussed by C. B. BiezenoCURVED BARS 113 Then it is necessary to consi- Sree eect der the deflection of the bar in “7 7 4 two perpendicular planes and \, OST the twist of the bar. Asimple 4 problem of this kind is shown ace A in Fig. 73a in which a portion Neo. of a horizontal circular ring, ae built in at 4, is loaded by a 2 vertical load P applied at the end B.* Considering a cross section D of the bar and taking the coordinate axes as shown in Figures 736 and 73c* we find that the moments of the external load P with respect to these axes are M, = — PRsin(a— ¢), M, =0, = PRE — cos(@— g)]. (a) Fro. 73. By using these expressions the bending and torsion stresses can be calculated in any cross section of the bar. In calcu- lating deflection at the end B the Castigliano theorem will be used, for which purpose we need the expression for the strain energy of the bar. Assuming that the cross sectional di- mensions of the bar are small in comparison with radius R, by using the principle of least work, “De Ingenieur,” 1927, and Zeitschr. f. angew. Math. u. Mech, Vol. 8, 1928, p. 237. The application of trigonometric series in the same problem is shown by C. B. Biezeno and J.J. Koch, Zeitschr. f. angew. Math. Mech., vol. 16, 1936, p. 321. The problem is of a practical importance in design of steam piping. The corresponding bibliography is given in the paper by H. KE. Mayrose, Journal of Applied Mechanics, Trans. A. S. M. E., vol. 4, 1937, P- 89- See also the book by A. H. Gibson and E. G. Ritchie, “A Study of the Circular-Arc Bow-Girder,” London, 1914. % This problem has been discussed by Saint Venant; see his papers in Comptes Rendus, vol. 17, 1843, Paris. 86 Tt is assumed that the horizontal axis x and the vertical axis y are the axes of symmetry of the cross section and that the z axis is tangent to the center line of the ring at D.114 STRENGTH OF MATERIALS we apply the same formulas as we already had for a straight bar (see pp. 294 and 297, partI). Thus the expression for the strain energy of our bar is u= [(4e +) xe, ®) where C denotes the torsional rigidity of the bar? The required deflection 6 is then obtained from the equation: aU b= 3 Substituting for U expression (4), and observing that OM, A ep =~ Rsin(a— 9) and OM: = RL — cos (a — 9), we obtain bo BE Looe, + 9G Lr ~ cos(a— oF fe. (109) In the particular case, when a = 1/2, _ PRP, EL 30 bo trli te (%-2)]- © If the cross section of the ring is circular, C= GI, = 2G I; taking E = 2.6G, we obtain PR PR’ é= ELLs +13 (= - 2)| = 1.248 El (105) As an example of statically indeterminate problems, let us consider a horizontal semi-circular bar with built-in ends, loaded at the middle, Fig. 74a. Considering only small vertical deflections of the bar, we can entirely neglect any displacements i in the horizontal plane as small quantities of a % The calculation of C for various shapes of the cross section is discussed in Chapter VI.CURVED BARS 1s higher order. Hence there will be no bending of the ring in its plane and no forces or moments in that plane at the ends A and B. Considering the built- in end B, we conclude from the equilibrium conditions that there will act a vertical reaction P/2 and the moment M,, = PR/2. The moment M,, will also act, preventing the end section B from rotation with respect to 2 axis. The magnitude of this moment cannot be determined from sta- tics; we shall find it by using the principle of least work, which requires that Flo. 74. arn @ In deriving the strain energy of the bar we represent the moments applied at the end B by the vectors PR/2 and M,,, as shown in Fig. 744. Then the moments M, and M, at any cross section D are M, = - cos g — M,, sin g — = sin 9, (e) M, = 7B sin o + Mz, cos © — = (1 — cos 9), (/) and the expression for strain energy is uaa f(s + Me) Rae. ) Substituting this in equation (d) and observing that 6M, . OM. aM, ~ ~ sine aM, = 008 @116 STRENGTH OF MATERIALS we obtain 7 Er (78 sin? ¢ + M,, sin’ 9 PR. ait 7 — "Bain ecos ¢) de + ¢ [ PE sin v cos » 2 CI 2 + Mz, cos? g — ae (1 — cos ¢) cos | dg =o, from which M,, = PR? 2 a) = —o182PR. (106) The minus sign indicates that the direction of M,, is opposite to that shown in Fig. 74a. Knowing M., we obtain the bend- ing and the twisting moments at any cross section from the expressions (e) and (f). The maximum deflection is evidently under the load and we readily obtain it from Castigliano’s equation: 6U b= 3p @) Substituting expression (g) for U and observing that aM, _R : oP 7 3 (cos ¢ — sin ¢), ane = R (sin g + cos g — 1), (i) we obtain PR I é= oEL { @ - 0969) (%-3) El, + “le — 0.363) (E+ +4) +e nat 0.363 | | PR = 0.514 Ter (107) In the calculation of the partial derivatives (/) we disregarded the fact that the twisting moment M,, is not an independent quantity but the function of P as defined by expression (106).CURVED BARS 117 If we should take this into consideration, the right-hand side of the equation (4) should be written in the following form: au , 0U dM., j oP * 3M. dP a But the second term in this expression vanishes, by virtue of equation (d). Hence our previous procedure of calculating deflection 6 is justified. Problems 1. A curved bar with circular axis and with « = 2/2 (Fig. 73) is loaded at end B by a twisting couple M, = T. Find the deflec- tion of the end B in a vertical direction. Answer. Assuming EI,:C = 1.3, 5 = 0,506 T™. = 0500 FT. 2. Solve the preceding problem assuming that at end Ba bending couple, M@, = Mp, is applied in the vertical plane tangent to the center line at B. Answer. MR b= 1.150 Tr 3. A semicircular bar with the center line in a horizontal plane is built in at 4 and B and loaded symmetrically by two vertical loads P at C and D, Fig. 75. Find the twisting moments M,, at the built-in ends. Answer. 2 ™ . M., = —ZPR (2 ~ cosa - asin): 4. Solve the preceding problem for the case of a uniform vertical load of intensity g distributed along the entire length of the bar. Answer. m4 M, = — R(E = ) = — 0.32 gR- 7118 STRENGTH OF MATERIALS 5. The horizontal semicircular bar, shown in Fig. 75 and uni- formly loaded as in the preceding problem, is supported at the middle cross section F, Find the vertical reaction N at the sup- port F, Answer. N = 2gR.CHAPTER III THIN PLATES AND SHELLS 20. Bending of a Plate to a Cylindrical Surface.—Assume chat a rectangular plate of uniform thickness / is bent to a cylindrical surface (Fig. 76). In such a case it is sufficient to consider only one strip of unit width, such as 4B, as a beam of rectangular cross section and of the length /. From 4 ils the condition of continuity it may be * concluded that there will be no distor- < tion in the cross section of the strip during bending, such as shown in Fig. ie | erage 77h, p.89, Part I. Hence a fiber length- wise of the strip such as ss (Fig. 77) suffers not only the longitudinal tensile stress o, but also tensile stress o, in the lateral direction, which must be such as to prevent lateral contraction of the fiber. We assume, as before (see p. 88, Part I), that cross sections of the strip remain plane during bending. Hence the unit elongations in the « and z directions are Hr. 76. a * We have such bending in the case of long rectangular plates if the acting forces dc not vary along the length of the plate and if only the portion at a sufficient distance from the ends is considered. hg120 STRENGTH OF MATERIALS The corresponding stresses in x and z directions are then obtained as in the case of tension in two perpendicular direc- tions. By use of eqs. (38) (p. 52, Part I), =, 7, -He8 PD_ G= er? rs eer We proceed as in the case of bending of a bar and calculate the bending moment at any cross section of the strip. Then hI E +h] EI Me nya ouydy = G- ro ody = 12(r = 4)’ from which 1_M ea (108) where eee (109) r2(1 — p°) 9. This quantity is called the flexural rigidity of a plate and takes the place of EJ, in discussing bending of beams. Com- parison of eq. (108) for the strip mk S—x with eq. (56), Part I, for a bar rm TI aa : remeed i} shows that the rigidity of the strip pay “|, in the plate is larger than that of Fic. 78. an isolated bar of the same cross section in the ratio 1: (1 — p?). The experiments show that, in the case of bending of an isolated thin strip of considerable width 4, distortion of the cross section takes place only near the edges (Fig. 78, 2) and the middle portion aa of the strip is bent into a cylindrical form; ? hence eq. (108) is applicable in calculating deflections and the strip will prove more rigid than will be expected from the simple beam formula. For small deflections of the strip 4B (Fig. 77) the curva- ? Explanaticn ef this phenomenon is given by G, F. C. Searle, Experi- mental Elasticity, Cambridge, 1908. Sce also H. Lamb, London Math. Soc. Proc., Vol. 21, 1893, p. 7, and author’s paper in Mechanical Engi- neering, 1923, P. 259. mTHIN PLATES AND SHELLS 121 ture 1/r can be replaced by its approximate value d?y/dx* and the differential equation for the deflection curve of the strip is ay Dy =-M (110) The discussion of bending of a plate to a cylindrical surface involves the integration of this equation. A particular case in which the bending to a cylindrical surface is done by a uniformly distributed load is discussed in the next article. 21. Bending of a Long Uniformly Loaded Rectangular Plate.— If a rectangular plate whose length is large in comparison with the width is uniformly loaded, then it may be assumed that near the center, where the maximum deflection and stresses occur, the de- flection surface is nearly a cylindrical one and eq. (110) may be used to calculate these.? Let us consider this important problem ¢ for two extreme conditions: (1) the edges of the plate are simply supported and can rotate freely during bending and (2) the edges are built-in. In both cases it is assumed that there are no displacements at the edges in the plane of the plate. ‘Then an elemental strip such as AB in Fig. 76 is in the same condition as a tie-rod with uniform lateral loading (see article 6) and the tensile forces §. The mag- nitude of the forces & is found from the condition that the extension of the strip is equal to the difference between the length of the deflection curve and the length / of the chord 4B (Fig. 76). Simply Supported Edges. In the case of simply supported edges, a good approximation for S is obtained by assuming that the de- flection curve is a sine curve y=osin™, (2) where 6 denotes the deflection at the middle. Then by using eq. (56), p. 47, the extension of the center line of the strip is UC (eV, Be. if oY de = al ® If the length is three times the width for a supported plate and twice the width for a clamped plate, the solution derived on this as- sumption is sufficiently accurate. +A solution of the problem was given by J. Boobnov. See his “Theory of Structure of Ships,” Vol. 2, p. 545, St. Petersburg, 1914. The discussion of this problem, together with calculation of stresses in the hull of a ship, is given in “Theory of Plates and Shells,” 1940.