II E> R.AR.Y OF THE UN IVLR.SITY OF ILLINOIS 620- 1 123 1 1 59/ v. 3-5 cop. 2 The person charging this material is re- sponsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for disciplinary action and may result in dismissal from the University. To renew call Telephone Center, 333-8400 UNIVERSITY OF ILLINOIS LIBRARY AT URBANACHAMPAIGN CCQ 02 HI81 L161— O-1096 Digitized by the Internet Archive in 2011 with funding from University of Illinois Urbana-Champaign http://www.archive.org/details/inelasticdesigno35univ / ? WADD TECHNICAL REPORT 60-580 Part III INELASTIC DESIGN OF LOAD CARRYING MEMBERS Part III The Significance of an Inelastic Analysis of Eccentrically-Loaded Members O.M, Sidebottom Theoretical and Applied Mechanics University of Illinois WRIGHT AIR DEVELOPMENT DIVISION WADD TECHNICAL REPORT 60-580 Part III INELASTIC DESIGN OF LOAD CARRYING MEMBERS Part III The Significance of an Inelastic Analysis of Eccentrically-Loaded Members O.M. Sidebottom Theoretical and Applied Mechanics University of Illinois September 26, 1960 Materials Central Contract No. AF 33(616)-5658 Project No. 7351 Wright Air Development Division Air Research and Development Command United States Air Force Wright Patterson Air Force Base, Ohio CZ0.IIZ3 ,■3 ~ 6 ENGINEERING LIBRARY, • FOREWORD 7-yoo ^3 This report was prepared by the University of Illinois under USAF Contract No. AF 33(616) -5658 . This Contract was initiated under Project No. 7351 "Metallic Materials", Task No. 73521, "Behavior of Metals". It was administered under the direction of the Materials Central, Directorate of Advanced Systems Technology, Wright Air Development Division with Mr. R.F. Klinger acting as the Project Engineer. This report covers work conducted from November 1, 1959 to October 31, 1960. The work was conducted in the Department of Theoretical and Applied Mechanics in the Engineering Experiment Station, University of Illinois, Urbana, Illinois. Professor O.M„ Sidebottom was the Project Supervisor. WADD TR 60-580 Pt III ABSTRACT The author has worked with others on ten investigations, sponsored by Wright Air Development Division, which have considered the theoretical and experimental inelastic analyses of eccentrically-loaded tension and compression members. In all cases good agreement was found between theory and experiment for numbers tested at room temperature and at elevated temperatures . This in- vestigation was undertaken to consider the significance of an inelastic analy- sis of eccentrically-loaded members. If the inelastic deformation can be con- sidered time independent, a choice has to be made between an elastic and an inelastic solution. A study was made of the effect of several variables on the ratio of the load necessary to produce a specified inelastic deformation to the maximum elastic load. If the inelastic deformation is time dependent (creep), the only choice is an inelastic solution. PUBLICATION REVIEW This report has been reviewed and is approved, FOR THE COMMANDER W .J . Trapp Chief, Strength and Dynamics Branch Metals and Ceramics Laboratory Materials Central WADD TR 60-580 Pt III iii TABLE OF CONTENTS Page No. I . INTRODUCTION 1 II . THEORETICAL APPROACH 2 III. ELASTIC VERSUS INELASTIC ANALYSES 5 WADD TR 60-580 Pt III iv LIST OF FIGURES Figure Page No . No . 1. Idealized Stress-Strain Diagram for Time Independent In- 12 elastic Deformation 2. Constant Depth of Yielding Interaction Curves for Rectangular 13 Section Member ( oL = 0) . 3. Family of Curves for Rectangular-Section Member Giving Rela- 14 tion Between A, q, and K. 4. Family of Curves for Rectangular-Section Member Giving Rela- 15 tion Between P , q, and K. 5. Configurations of Eccentrically-Loaded Tension Members and 16 Columns . 6. Cross-Sections for Eccentrically-Loaded Members. 17 7. Ratio of Collapse Load to Maximum Elastic Load for Rectan- 18 gular-Section Eccentrically-Loaded Columns. 8. Ratio of Collapse Load tc Maximum Elastic Load for Circular- 19 Section Eccentrically-Loaded Columns. 9. Ratio of Collapse Load to Maximum Elastic Load for Angle- 20 Section Eccentrically-Loaded Columns. 10. Ratio of Collapse Load to Maximum Elastic Load for T-Section 21 Eccentrically-Loaded Columns. 11. Ratio of Collapse Load to Maximum Elastic Load for I-Section 22 Eccentrically-Loaded Columns. 12. Ratio of Collapse Load to Tangent Modulus Load for Eccentri- 23 cally-Loaded Columns Made of a Material that Creeps. WADD TR 60-580 Pt III I. INTRODUCTION Modern design, particularly in the aircraft and missile fields, requires that load carrying members be used to the limit of their capacity. In many applications the peak loads are applied only a small number 01 times so tnat fatigue is not a problem. For the design of these members, the engineer has to choose between an elastic ana an inelastic analysis. Because of its ease of application and the confidence which the engineer has in its application, the engineer wouia prefer to use the elastic solution. However, the elastic solution is too safe in many applications, and the engineer must use an in- elastic design to utilize the increase in load carrying capacity which is pos- sible if small, inelastic strains are allowed. The percentage increase in load above the maximum elastic load depends upon such variables as type of member, shape of cross-section, shape of stress-strain diagram of the material, and the amount of inelastic deformation. In the case of design lor an ele- vated temperature of sufficient magnitude to produce creep, an inelastic anal- ysis is always required. The difficulty of an analysis depends upon the type of relation used to represent the stress-strain diagram for the material. Linear elasticity uses one of the simplest relations, namely, a straight line. If the material has a yield point and the member is designed for strength rather than deformation, a fully plastic analysis may be used; the resulting solution may be easier to apply than an elastic solution. *'A fully plastic analysis cannot be used in a large number of instances if any one of the following conditions are present: 1. The member is made of a material which does not have a yield point. 2. It is necessary to know the load-deformation relation for inelastic conditions . 3. The member may become unstable before reaching the fully plastic con- dition as is the case of an eccentrically-loaded column. In most cases, the loads on the members are assumed to maintain their same re- lative magnitude and to increase in magnitude so that a nonlinear elastic analysis can be used. In deriving the theoretical relations, the stress-strain diagram of the material is represented by one non-linear function or two func- tions. At room temperature, two functions are usually used to represent the stress-strain diagram for metals, and the resulting solution is called an elas- tic-plastic solution. Because of the ever increasing interest is the inelastic analysis of load carrying members, it is necessary that theoretical inelastic analyses be de- veloped for all types of load carrying members. Furthermore, it is necessary that these theories be checked experimentally for the following reasons: 1. Frequently simplifying assumptions are used in order to reduce the complexity of the theory so that the theory can be readily applied by the design engineer. It is necessary to determine the influence of these simplifying assumptions on the agreement between theory and experiment. Manuscript released by author September 26, 1960, for publication as a WADD Technical Report . WADD TR 60-580 Pt III 1 2. It is logical that the engineer does not have the same confidence in an inelastic analysis which he has in an elastic analysis . Experimental verification of inelastic theories helps promote confidence in those theories . Some of the investigations (1 through 10) sponsored by the Wright Air Development Division have considered the theoretical and experimental inelas- tic analyses of eccentrically-loaded tension and compression members. These investigations have considered a large number of variables, namely: 1. Several types of ductile metals such as S.A.E. 4340 steel, 7075-T 6 aluminum alloy, type 304 stainless steel, 17-7FH stainless steel, and Ti 155A titanium alloy. 2. Several types of cross-sections such as rectangular .angle , and T- sections . 3. Several end conditions for the columns such as fixed ends and pivot ends with equal and unequal end eccentricities of various magnitudes. 4. Several slenderness ratios for the columns, 5. Both room temperature and elevated temperatures. In all cases good agreement was found between theory and experiment indicating that the theory was sufficiently exact to be used in the design of eccentri- cally-loaded members. Since the theory was found to be reliable, was the in- crease in load carrying capacity of a member (which resulted from the inelastic deformation) of sufficient magnitude to justify the increase in labor necessary to complete an inelastic analysis? Answers to this question vary from yes for some cases to no for other cases. In order to obtain definite answers to the question, the theory will be reviewed so that the effect of various variables on the increase in load carrying capacity of eccentrically-loaded members can be determined . II. THEORETICAL APPROACH In deriving theoretical, load-deflection relations for eccentrically- loaded members, the procedure is to equate the internal load and moment at every section of the member to the applied load and moment. For elastic con- ditions, Timoshenko (11) has presented closed solutions for rather general loading conditions. Closed solutions have not been found for inelastic condi- tions. Several analyses requiring a trial and error solution have been pre- sented (2,3,5,6,7,8,9,10,12,13,14). In each case, a family of interaction curves was constructed to give a relation between the internal load, moment, and curvature at any section of the member. A relation between applied load, moment, and curvature was obtained by approximating the configuration of the deformed member , * Numbers in parenthesis refer to correspondingly numbered entries in the Bibliography . WADD TR 60-580 Pt III 2 At least three different types of interaction curves have been used in the analyses of eccentrically-loaded columns in which the inelastic deforma- tion was primarily time independent. Chwalla (13) and Bijlaard (12) used a family of curves giving the relations between the average stresses of origi- nally straight, centrally-loaded columns and the post-buckling deflection. Galambos and Ketter (14) used dimensionless , moment versus curvature curves for specified loads. Sidebottom and Clark (2) approximated the stress-strain diagram of the material by two straight lines (see Fig. 1), and constructed dimensionless, moment versus load curves for constant depths of yielding. These curves are shown in Fig. 2 for rectangular-section members made of a material with o^ = (see Fig. 1). In Fig. 2, P e = %& and M e = 0~ e I/c , It is the author's opinion that constant depth of yielding interaction curves are the easiest to construct and use. In case the inelastic deformation is time dependent (creep), Sidebottom, Clark, and Dharmarajan (6) approximated the isochronous stress-strain diagram of the material by an arc hyperbolic sine curve of the form 0~~ 0~ arc sinh -gr- (1) in which CJ and 6 are experimental constants. Since the stress distribu- tion in the eccentrically loaded member changes with time and OS 1.2 Curve e/'h 0.05 0.05 0.25 0.24 06 0.00 0.20 0.00 0.20 1.0 10 20 £ e WrY 30 40 Fig. 11 Ratio of Collapse Load to Maximum Elastic Load for I-Section Eccentrically-Loaded Columns WADD TR 60-580 Pt III 22 1.2 ■a cs o (0 3 rH 3 •o o s c 0) c a) H O ■p XJ 0) o V (0 a O o -p at 1.0 0.8 0.2 -Ul- k— 8t — -| .05 0.10 0.15 INITIAL ECCENTRICITY, e/h 0.20 0.25 Fig. 12 Ratio of Collapse Load to Tangent Modulus Load for Eccentrically- Loaded Columns Made of a Material that Creeps WADD TR 60-580 Pt III 23 WADD TECHNICAL REPORT 60-580 Part IV INELASTIC DESIGN OF LOAD CARRYING MEMBERS Part IV. The Behavior of Beam -Columns in the Inelastic Range B. B. Muvdi O. M. Sidebottom Theoretical and Applied Mechanics Department University of Illinois April , 1961 Materials Central Contract No. AF 33 (6 16) -7600 Project No: 7351 Wright Air Development Division Air Research and Development Command United States Air Force Wright Patterson Air Force Base, Ohio WADD TECHNICAL REPORT 60-580 Part IV INELASTIC DESIGN OF LOAD CARRYING MEMBERS Part IV. The Behavior of Beam -Columns in the Inelastic Range B. B. Muvdi O. M. Sidebottom Theoretical and Applied Mechanics University of Illinois April 1961 WRIGHT AIR DEVELOPMENT DIVISION FOREWORD This report was prepared by the University of Illinois under USAF Contract No. AF 33(6l6)-7600. This Contract was initiated under Project No. 7351, "Metallic Materials", Task No. 73521, "Behavior of Metals". It was administered under the direction of the Materials Central, Directorate of Advanced Systems Technology, Wright Air Development Division with Mr. R. F. Klinger acting as the Project Engineer. This report covers work conducted from October 1 1960 to Setpember 31, 1961. The work was conducted in the Department of Theoretical and Applied Mechanics in the Engineering Experiment Station, University of Illinois, Urbana, Illinois. Professor O. M. Sidebottom was the Project Supervisor. ABSTRACT Two theories were presented for constructing either moment -load or load- deflection relations as well as the collapse loads for beam-columns. In each case, trial and error solutions were required which used constant depth of yielding interaction curves. A "so called" exact theory was presented which gave results as accurate as desired; however, the theory was not -practical because of the excessive time required. An approximate theory was presented which gave results in close agreement with the exact theory and with experimental data. This theory required the elastic solution for maximum elastic conditions. The experimental part of the investigation included tests of rectangular - and T-section columns made of 2024-T4 aluminum alloy, SAE 1020 steel, and 17-7PH stainless steel. Several slenderness ratios were considered. In addition to the variable axial load, the columns wer subjected to a constant transverse load either at midspan of at quarter span which produced a bending stress of 0. 25, 0. 50, or 0. 75 C . PUBLICATION REVIEW This report has been reviewed and is approved. FOR THE COMMANDER: W. J. Trapp Chief, Strength and Dynamics Branch Metals and Ceramics Laboratory Materials Central in TABLE OF CONTENTS Page I. INTRODUCTION 1 1. Preliminary Statement 1 2. Purpose and Scope 2 II. ANALYSIS 4 1. Interaction Curves 4 2. Elastic Solution for the Beam -Column 7 3. Method of Solution 11 4. Sample Problem 12 III. MATERIALS AND TEST PROCEDURE 15 1. Materials and Specimens 15 2. Apparatus and Test Procedure 16 IV. RESULTS AND DISCUSSION 18 1. Symmetrically Loaded Beam -Columns 18 2. Unsymmetrically Loaded Beam -Columns 20 V. SUMMARY AND CONCLUSIONS 23 VI. BIBLIOGRAPHY 25 VII. APPENDIX A 27 IV LIST OF ILLUSTRATIONS Figure Page 1 T Section Member Subjected to Combined Loading. The 34 Sketch Shows the Stress Distribution When the Member is Partly Elastic and Partly Inelastic. 2 A Schematic Representation of Stress -Strain Relations in 35 Tension and Compression. 3 A Schematic Representation of Interaction Curves for 36 Constant Depth of Yielding . 4 A Sketch Showing the Beam -Column Studied in this Program. 37 5 Dimensionless Interaction Curves for Rectangular Sections 38 Made of 2024-T4 Aluminum. 6 Dimensionless Interaction Curves for T-Sections Made of 39 2024-T4 Aluminum . 7 Dimensionless Interaction Curves for Rectangular Sections 40 Made of SAE 1020 Steel. 8 Dimensionless Interaction Curves for T-Sections Made of 41 SAE 1020 Steel. 9 . Dimensionless Interaction Curves for Rectangular Sections 42 Made of 17-7PH Stainless Steel. 10 Test Apparatus Used to Apply Transverse and Axial Loads 43 to Beam Columns. 11 Knife -Edge Fixtures Used to Apply Axial Load to Beam 44 Columns. 12 Comparison of Moment Load Curves for Symmetrically 45 Loaded 2024-T4 and SAE 1020 Steel Beam-Columns Determined by Two Theoretical Methods and by Test. 13 Moment -Load Curves for Symmetrically Loaded, Rectangular 46 2024 -T4 Aluminum Beam-Columns. (M„/M = 0.25, e x = e 2 = 0) <* e 14 Moment -Load Curves for Symmetrically Loaded, Rectangular 47 2024-T4 Aluminum Beam-Columns. (M n /M = 0. 50; e = e 2 = 0) ^ e LIST OF ILLUSTRATIONS Figure Page 15 Moment-Load Curves for Symmetraically Loaded, 48 Rectangular 2024-T4 Aluminum Beam -Columns. (M Q /M e = 0. 25; e x = e £ = 0. 15h) 16 Moment -Load Curves for Symmetrically Loaded T-Section 49 Beam Columns of 2024-T4 Aluminum. 17 Moment-Load Curves for Symmetrically Loaded, Rectan- 50 gular SAE 1020 Steel Beam-Columns. (M n /M = 0. 50; e x = e 2 = 0) Q e 18 Moment-Load Curves for Symmetrically Loaded Rectangular 51 SAE 1020 Steel Beam Columns, (M Q /M = 0. 75; e, = e 2 = 0) 19 Moment- Load Curves for Symmetrically Loaded T-Section 52 Beam -Columns of SAE 1020 Steel. 20 Moment -Load Curves for Symmetrically Loaded Rectangular 53 17-7PH Stainless Steel Beam-Columns, (e, = e~ =0) 21 Moment-Load Curves for Symmetrically Loaded; Rectangular 54 17-7PH Stainless Steel Beam-Columns, (e. = e = 0. 15h) 22 Moment- Load Curves for Unsymmetraicaily Loaded, 55 Rectangular 2024-T4 Aluminum Beam-Columns. (M n /M = 0. 50; e x = e 2 = 0) g e 23 Moment-Load Curve for an Unsymmetric -Loaded T-Section 56 Beam -Column of 2024- T4 Aluminum. (M n /M =0.50; e x = e 2 = 0) ^ e 24 Moment-Load Curves for Unsymmetrkaily Loaded, Rectangular 57 SAE 1020 Steel Beam -Columns. on the tension side. The remaining depth, a, is elastic. The expressions for P and M as developed by Sidebottom and Clark (2), are given below without derivation. For a ^ t : a a b a -5- = 1 " -(c, - a.) -£ - -i-s- (1 - a,) — P a v 1 r + M Q (a) which applies to sections on either side of the transverse load Q. The symbols e 1 and e„ are the eccentricities of the load P at the left and right ends, respectively, y is the deflection of the centroidal axis of the member at the point considered, 4 is the length of the member and M n is the moment at the section in question due to the transverse load Q acting alone. Equation a may be written in dimensionless form by dividing both sides by the quantity M = P /y , in which y is a constant that depends on the properties of the cross -section. Thus M . P . M T V P ' e e y + e x -^(e^e^ + M„/M . Q e (b) It should be noted that Eqs. a and b are valid for either elastic or inelastic conditions The differential equations of the elastic deflection curve for the beam-column shown in Fig. 4 may be expressed as EI dy d7 Qd P(e x - e 2 > x - P(ej + y) x L (S. - d) (c) EI- dy Q( It -d) P(ej - e 2 ) x - P(e 1 +y) Q( ft - d) . . . . x ^ (1 - d) (d) where E is the modulus of elasticity and the remaining symbols are as defined 2 previously. Using the notation /3 = P/EI and rearranging terms, the above equations may be written as Pe, dy 2 dx 2 + „ 2 y=-A- EI Qd P(e i " e 2> x - ..... x ^ (i - d) dy 2 dx 2 + 2 y -A. EI QU-d) , P(e l " e 2 } T7 - T d > ...,x^({- d). EI Pe, EI (e) (f) 2 (Te l I In the case of a rectangular section, for example, M = CjA/I jjt- so that y is equal to 6/h. The general solutions of Eqs. e and f are given by the following equations y = C, cos fi x + C„ sin /3 x . . . . . x L (£ - d) Qd (e l - e 2 > PI " 1 x - e. y = Co cos (5x +C, sin /3 x + QU_Ld> . . x A (£ . d) . Q(I - d) (e l " e 2> Pi + 1 - e. T P From the conditions at the ends of the beam-column (i. e. , y = for x = and x = i), the constants of integration C, and C, are found to be C l = e l ' C 3 = " C 4 tan ^ * + e 2 / cos B * • (g) (h) (i) The constants of integration CL and C are determined from the conditions at the point of application of the transverse load Q. At this point, the two portions of the elastic deflection curve, as defined by Eqs. g and h possess the same values of deflection and slope. From these conditions, the values of C„ and C. are found to be C 2 = e 2 - e, cos p H sin/3£ e 2 - e cos /3 9. Q P£ sin p{i - d) tan/3£ cos p(i - d) Q sin p(JL - d) C 4 " sin/3£ " ~YJ tan 1 (J) Substitution of the values of the constants C. . . . . C. into Eqs. g and h, rearranging terms and simplifying leads to the following equations defining the elastic deflection curve of the beam -column shown in Fig. 4, y = e. sin /3(l - x) sin pi ix + e sin p x x 2| sin p S. ' I Q sin B d . _ Qd + ts-s — ■ n /i sin fix- ^t x P0sin/3£ H PS. x £ (i - d) (6a) y = e. sin j3(l - x) 9. - x sin £ /5 J sinjSx x ' e 2 sin/3£ " £ + g/if/iV d) sin^-x)- Q< £ D - d ^- x ) ....x^f-d) (7) (6b) X* x) „ V//) P/3 sin It smpi * * ; FT The elastic curvatures may be obtained by differentiating Eqs. 6 twice, thus j, = v „ = _ fl 2 sin /?(£ - x) _ 2 sin ff x * y " e l P sin 4 e 2 P liifpT " Q pt Sln // sin^x . . . . x^ (£-d) P sin /3 £ K v ' ri - v" ■■ P fl 2 sin ff (I - x) 2 sin fix * " y " e l^ sin pi e 2 P ^TnTl - Q Mi/fi\' d) sinMix)....xA (f -d). The bending moment at any section along the span of the member is then given by the equation M = -El(Jp ). (8) For a given member, the deflection, y , and curvature, ip , at the critical section may be determined for any value of P from Eqs. 6 and 7, respectively. Examination of the ratio ip /y revealed that, in general, this ratio did not vary greatly as the axial load P increased from zero to its maximum elastic value. This behavior led to the assumption that the ratio ip /y remains constant as the beam- column is strained inelastically. Essentially, then, a relation between curvature and deflection at the critical section in the inelastic range is assumed to be of the form ^ c =Cy c (9) where the constant C is determined as the rat io of curvature to deflection at the critical section when yielding is impending. Eliminating the deflection, y , between Eqs. b and 9. and solving for P/P , the required relation between axial load^moment, and curvature is obtained, namely 10 M/M - (M /M ) -^- (10) p e Y[V c + e i-V 1(e i- e 2>] where the subscript c designates conditions at the critical section. 3. Method of Solution The purpose of this article is to show how the various relations established in the preceeding article may be used to determine the moment -load curve or the load- deflection curve as well as the load P that causes complete collapse of a beam-column when it is subjected to a constant value of the transverse load Q. The value of the axial load P that would initiate yielding in the member may be determined, by trial and error, using Eqs. 7 and 8. The procedure consists of first assuming a value for P and determining the quantity /3 = A P/EI. The value of x locating the critical section is now determined by maximizing Eq. 7. The maximum moment is then given by Eq. 8 and the maximum stress at the critical section by the equation (M ) c. P = max 1 + - . (11) max. j A This process is repeated until the stress determined by Eq. 11 becomes equal to the compressive yield stress of the material, a With the value of P that initiates yielding known, the constant C may be easily found from Eq. 9, in which, the values of y and ip are determined by the use of Eqs. 6 and 7, respectively. Furthermore, this value of P may be used to locate the intersection of the moment-load curve and the interaction curve defining the beginning of yielding (k = 1), see point B in Fig. 3. The solution is then carried into the inelastic range by means of the inter- action curves along with Eqs. 5 and 10. In applying Eq. 10, the assumption is made that the critical section does not move as the member deforms inelastically. This is equivalent to saying that x and (M„/M ) retain the same values they assumed at initiation of yielding. With members which are symmetrically loaded (i. e. , d = 1/2 and e. = e 2 ), x does retain the same value, namely H/2, in the inelastic as well as in the elastic range. However, with members which are not symmetrically loaded (i. e. , d f 1/2 and/or e 1 f e„), x does not, in general, retain the same value throughout the loading process. Test data indicated that the above assumption did not introduce a serious error into the theory. 11 Now consider the problem of determining the value of P/P that would produce a given depth of inelastic penetration (i. e, , k = k. = const), see Fig. 3. The condition that has to be satisfied is that M/M be of such magnitude that the value of P/P given by Eq. 10 be equal to that given by the relation P/P = f(M/M ) as expressed by the interaction curves. To this end some value of M/M is assumed and the curvature ip is determined by Eq. 5. Equation 10 is then used to obtain a value of P/P which is compared to that given by the interaction curve corresponding to k = k. . This process is repeated until the two values of P/P are the same or very nearly so. This usually requires not more than three trials. Thus the problem is reduced to the solution of two simultaneous equations in the unknowns M/M and P/P , The deflection y , if needed, may now be obtained from Eq. 9. Once the correct value of P/P is established for k = k 1 , a point is located on the appropriate interaction curve. Other values of k are then considered and the same process repeated to establish a set of points such as C, D, E, F, and G in the inelastic domain, see Fig, 3. The location of point A on the M/M axis is dictated by the value of the transverse load Q, A smooth curve is then constructed through the points A, B, .... G to give the moment- load curve. The collapse load for the beam-column may be obtained from the moment load curve and in dimensionless form, is the maximum value of P/P . e If additional points are needed in the elastic range (i. e. , between points A and B in Fig. 3) they may be easily determined from the elastic solution, using Eq. 6 and the equilibrium condition expressed by Eq. b. 4. Sample Problem In this article, a sample problem is solved to illustrate the method used in deriving the moment-load curve for a given beam-column. (a) Given Data For a Rectangular SAE 1020 Steel Beam-Column: A = b x h = 0. 500 x Q625 = 0. 3125 in 2 ; I = 0. 01017 in 4 ; it = 13. 52 in(4/r = 75) M = 0. 5 M g ; d = 0. 25 i ; e = 0. 15h = 0. 09375 in e =0; a = a =30, 300 psi; P = 9470 lb. 2 e l e 2 -4 e e = € = 10. 1 x 10 in/in. e l e 2 (b) Problem: To determine the moment -load curve for the beam-column specified above. 12 (c) Solution: Q 16 " 81(7 a I e Q = 3 tc = 194. 49 lb. 1 From the given conditions it is evident that the critical section occurs either under the load Q or slightly to the left of this position Maximizing Eq. 7 for x ^ (£ - d) leads to the condition that the location of the critical section is defined by the equation tan fix = r c Q sin fl d - e. fl P cos fl I + e„ fl P e, fl P sin fl £ Since in this problem e„ = 0, Eq. k becomes tan fix = ■ H c e. fl P sin fl £ (k) Q sin fl d - e fl P cos fl £ (i) By Eqs. 7 and 8 and using the condition e = 0, the maximum moment, M , is given by the equation EI e, r 2 M max sin jr sin m ' x c> EIQflsinfld + sinflx Psinflf c The maximum stress, a , will then be given by Eq. 11. (m) The computations for the axial load that would initiate yielding are given in the following table. Trial P - lb fl-l/in x -in. ^max °max lb-in. psi 1 3000 0. 0992 10. 14 654. 12 29, 700 2 3200 0. 1024 10. 14 668.03 30, 770 3 3100 0. 1008 10. 14 661. 17 30, 240 4 3110 0.1010 10. 14 661.93 30, 290 13 Thus, to initiate yielding P/P = 3110/9470 = 0. 328. By Eq. 8 for x ^ (I - d), y = 0. 0308 in, for P = 3110 lb. By Eq. 9 for x ^ (£ d), ip = 21. 69 x 10" 4 l/in,for P = 3110 lb. C = if) /y =0. 0704 1/in y = 6/h = 9.6 1/in If) 14 At x c = 10. 14 = jj^fi ft = 0. 75 i , (M„/M e ) c = 0. 50 Therefore, the equations that apply in the inelastic range are r c 32. 32 x 10 -4 M/M M /M u e (5) M/M - 0. 50 P/P e = 9. 6 ( 14.2 ip + 0.0234) - (10) and the functional relation P/P = f(M/M ) as expressed by the interaction curves shown in Fig. 7. A sample calculation is shown for k = 0. 7 in the following table. Trial M/M (Assumed) ^c Eq. 5 P/P e Eq. 10 P/P e Fig. 7 y c Eq. 9 1 2 3 0.830 0.832 0.831 34.21 x 10" 4 34. 30 x 10" 4 34. 25 x 10" 4 0.478 0.480 0.479 0.480 0.478 0.479 0. 0487 Thus to produce 0. 3 depth of inelastic penetration (k = 0. 7), P/P = 0. 479, which locates a point on the moment-load curve. Other points, corresponding to other values of k may be similarly located. 14 III. MATERIALS AND TEST PROCEDURE 1. Materials and Specimens In order to examine the validity of the analysis discussed in Section II, tests were performed on beam-columns made of three materials, namely 2024-T4 aluminum alloy, SAE 1020 steel and 17-7PH stainless steel. The 2024-T4 aluminum samples were tested in the as-received condition The SAE 1020 steel was annealed, prior to machining, by soaking at 1600 F for 3 hours and furnace cooling. The 17-7PH stainless specimens were precipitation hardened after machining. The SAE 1020 steel samples were machined from a 3 1/2 in. diameter bar, while the 2024-T4 aluminum and the 17-7PH stainless steel specimens were machined from 1/2 in. plates. The various beam-columns tested in this program with their physical characteristics are shown in Table I. Either one or two beam -columns per condition were tested in this study. Standard tensile and compressive tests were performed on the aluminum alloy to determine the various properties needed for the analysis. However, only com- pressive tests were performed on the mild and stainless steels. Results of previous tests by Sidebottom and associates (21) on the last two materials have indicated that the difference between tensile and compressive properties was negligible for all practical purposes. The various properties used in the present analysis are shown, for the three materials, in Table II. These values represent an average of at least six tests. Using the various properties indicated in Table II, and Eqs. 1 through 4, interaction curves for the three materials and the two types of cross -section were developed. These interaction curves are shown in Figs. 5 to 9. Figures 5 and 6 represent the 2024-T4 aluminum interaction curves for rectangular and T-sections, respectively. Figures 7 and 8 show the SAE 1020 steel interaction curves for rectangular and T-sections, respectively. And Fig. 9 illustrates the interaction curves for rectangular sections made of 17-7PH stainless steel. In all instances, k decreased in increments of one tenth to a minimum value of 0. 4. This heat treatment consisted of soaking at 1400 F for 90 minutes, cooling to 60° F in 60 minutes, soaking at 60° F for 30 minutes, soaking at 1050° F for 90 minutes and finally air cooling. 15 2. Apparatus and Test Procedure The testing apparatus used in this program is shown in the photograph of Fig. 10. A Riehle testing machine having a capacity of 120, 000 lb. was used to apply the 2 axial load through fixtures provided with knife edges as illustrated in Fig. 11 The position of the knife edges may be adjusted with respect to the centroid of the specimen by means of set screws. The transverse load was applied by means of dead weights through the pulley system shown in the photograph of Fig. 10. The deflection of the specimen was measured by means of a 1/1000 in. dial indicator which may also be seen in the photograph as well as in Fig. 11. The specimen was properly placed in the test apparatus and a small value of axial load was applied to maintain the specimen in the proper position. The pulley system was then raised or lowered so as to apply the transverse load at the desired position along the span of the beam-column. The dial indicator was subsequently placed in positon and the desired transverse load applied. This transverse load was maintained at a constant value throughout the test. The axial load was subsequently increased, and readings of this load and the corresponding deflection were taken at predetermined intervals of load until the predicted value of the collapse load was approached. Beyond this point, judgment was exercised in spacing the readings at reasonable intervals of deflection. Sufficient readings were taken to make it possible to construct the moment-load curve well beyond the collapse load. Once the member was loaded beyond the elastic limit, sufficient time was allowed for equilibrium conditions to be reached (i.e. , for the axial load to reach a steady value). In the case of the aluminum and the stainless steel samples, a period of approximately two minutes was sufficient for the axial load to stabilize. In the case of the mild steel specimens, due to the presence of an upper yield point, appreciably more time was required for steady state conditions to be reached. In 3 most instances, up to 20 minutes elapsed before a reading could be taken. It was discovered, however, that this delay time could be shortened, without appreciably 2 A complete description of these fixtures may be found in a paper by Sidebottom et al (22). The two end fixtures added a total of 1. 20 in. to the machined length of the specimen. 3 This time-sensitivity of mild steels was investigated previously by Clark, Corten, and Sidebottom (20). 16 influencing the load-carrying capacity, by prick punching one row of shallow indentations along the span of the member. These indentations were spaced approximately one- half inch apart. In a few instances, two tests were performed for the same conditions, one with and one without indentations. The results indicated the difference between the load -carrying capacities of the two specimens to be insignificant. 17 IV. RESULTS AND DISCUSSION In this section, a comparison is made between the results of the proposed approximate theory, and those of the experiments. The comparison is made on the basis of moment Toad curves (i. e. , M/M vs. P/P ), This method of presentation was selected over the more conventional load -deflection plots because it conveniently indicates the value of M^/M to which the beam-column was initially subjected before the axial load was applied. The various tests that were performed in this program may be conveniently classified into two types as follows: 1. Symmetrically Loaded Beam -Columns This class of members include those subjected to end loads with zero or equal eccentricities and to a transverse load at midspan. They are characterized by the fact that the maximum moment is located at midspan and possess symmetrical deflection curves throughout the loading process. In order to examine the degree of accuracy attainable by the approximate method, a comparison was made between moment-load curves obtained by this method and those developed by the so-called exact method (see Appendix A). In view of the fact that the latter method involves rather lengthy computations, particularly with unsymmetrically loaded members as indicated in Appendix A, the comparison was limited to two cases only, namely one 2024-T4 aluminum member and one SAE 1020 steel member. Both members were symmetrically loaded with zero end eccentricities, had a slenderness ratio of 75 and were subjected to transverse loads at midspan such that M n /M was equal to 0. 50. The results of both methods are illustrated in Fig, 12 together with the values exhibited by test. In the case of 2024 -T4 aluminum, the two theoretical moment -load curves were nearly identical and were found to compare very well with the test values. In the case of SAE 1020 steel, however, a slight difference of about 5 per cent is observed between the collapse loads as predicted by the two methods. While this difference was in favor of the exact method, it is too small to justify the added labor required by this method. Furthermore the approximate theory was conservative. The actual moment -load curves for symmetrically loaded members are presented in Figs. 13 to 21 in comparison to curves developed by the approximate method. Figures 13 to 16 illustrate the moment-load curves for members made of 2024-T4 aluminum alloy; Figs. 17 to 19 show the moment-load curves for members made of SAE 1020 steel; and Figs. 20 and 21 indicate the moment-load curves for members made of 18 17-7PH stainless steel. In the case of the 2024-T4 aluminum and SAE 1020 steel, both rectangular and T- sections were examined. Furthermore, four different slenderness ratios ranging between 50 and 100 were investigated. In addition, for each of these two materials, two different values of M n /M were analyzed, namely 0. 25 and 0. 50 for 2024-T4 aluminum, and 0. 50 and 0. 75 for SAE 1020 steel. In the case of 17-7PH stainless steel, however, testing was limited to members with rectangular-sections, having slenderness ratios of 50 and 75, and subjected to values of M^/M equal to 0. 25 and 0. 50. Examination of Figs. 13 to 21 leads to several interesting conclusions. It is observed that, in all instances, the predicted curves compare very well with the test data. In general, the discrepancy between the predicted collapse load and that exhibited by test is less than about 7 per cent, as shown in Table HI by the ratio of actual collapse load, P , to theoretical collapse load P . This discrepancy is not excessive when considered with respect to the scatter that is normally encountered in materials testing. In general, when the beam column is symmetrically loaded, the theory always predicts values of collapse load which are either equal to, or slightly lower than those exhibited by test. In other words, the theory tends to predict values which are conservative, as may be seen from Table III by the ratio P /P . Ccl C L It is also observed that, in every instance, there is a considerable increase in the load carrying-capacity above that associated with the beginning of yielding. This increase in the load carrying capacity is represented in Table III by the ratio of actual collapse load, P , to the actual load that initiates yielding, P . Evidently, the increase in the load carrying-capacity (i. e. , P /P - 1) depends upon several Cci y^- factors which include, among other things, the slenderness ratio, magnitude of the transverse load, end eccentricities, material and type of cross -section. While the tests that were performed are insufficient to evaluate the influence of all pertinent factors, it is possible to determine qualitative trends regarding the effects of the slenderness ratio, magnitude of transverse load, material and end eccentricities on the ratio P /P . ca ya Examination of Table III indicates that for a given material and for zero eccentricities the ratio P /P increases with increase in the magnitude of the trans - ca ya & verse load (i. e. , with increase in the ratio M n /M ), and with decrease in the slender- ly e ness ratio. Furthermore, a comparison of the data corresponding to M n /M = 0. 50 v e and zero end eccentricites indicates that P /P is slightly, but consistently, higher ca ya 19 for SAE 1020 steel than for 2024-T4 aluminum. In addition, based on the two tests that were performed on 17-7PH stainless steel with zero end eccentricities, the ratio P /P * ca ya for this material was found to be considerably less than that for the aluminum and mild steel alloys. These trends lead to the conclusion that one of the factors influencing the ratio P /P is the strain at yielding, namely e . Apparently, the larger the value of ca ya e e the less is the ratio P /P for a given slenderness ratio. e ca ya ° The influence of end eccentricities on the increase in the load-carrying capacity may also be seen from Table III. A comparison of the values corresponding to M n /M in the case of 2024-T4 aluminum, indicates that P /P attain higher values for ca ya & members provided with end eccentricities than for members subjected to concentric axial loads. 2. Unsymmetrically Loaded Beam -Columns This type of members include those subjected to unequal end eccentricities and/or to a transverse load not at midspan. They are characterized by the fact that the maximum moment may not occur at the same section of the member throughout the loading process, and the deflection curve is, therefore, unsymmetrical. The actual moment-load curves for beam-columns of the type described above are presented in Figs. 22 to 27 in comparison to theoretical curves. Figures 22 and 23 illustrate the moment-load curves for members made of 2024-T4 aluminum; Figs. 24 to 26 show the moment -load curves for members made of SAE 1020 steel; and Fig. 27 indicates the moment-load curve for a member made of 17-7PH stainless steel. In the case of 2024-T4 aluminum and SAE 1020 steel, testing was performed to include both rectangular and T-sections, four different values of slenderness ratio ranging between 50 and 100, and two different values of the ratio M n /M , namely 0. 50 and 0. 75. In the case of 17-7PH stainless steel, only one test was performed on a member of rectangular cross-section having a slenderness ratio of 75 and subjected to a trans- verse load such that M„/M was equal to 0. 25. In this series of tests, as in the previous series, very good agreement is observed between the predicted and the actual test values. The discrepancy between the theoretical collapse load and that given by test, was found not to exceed about 5 per cent, as shown in Table IV by the ratio P /P . ca ct However, in this series, unlike the previous series, the tests did not consistently yield values of collapse load which were either equal to, or slightly higher than that predicted by theory. This lack of consistency may be seen in Table IV as the ratio 20 P /P varied from a minimum of 0. 95 to a maximum of 1, 05 depending upon the Ca c r material, slenderness ratio and loading conditions. This inconsistency in the observed trends may be explained as follows: At the beginning of loading (i. e. , when only the transverse load, Q, is acting and the axial load, P, is zero or very nearly so), the maximum moment occurs at the point of application of the transverse load. When the axial load P reaches a certain value that; depends upon the material, the slenderness ratio and the cross -sectional area, the position of the maximum moment (critical section) begins to move away from the point of application of the transverse load, Q . The axial load, P, at which movement of the critical section begins, may occur either within the elastic or within the inelastic range of the material depending upon the conditions stated previously. The conditions investi- gated in this program were such that movement of the critical section was always from the point of application of the transverse load Q towards midspan. In the elastic range, the influence of the migration of the critical section can be easily included in the anslyses. However, there is no simple method known by means of which the positon of the critical section may be determined at any stage of inelastic action. This difficulty was circum- vented in the present analyses by making the assumption that the critical section, during inelastic deformation, retains the same position it assumed at initiation of yielding, as explained previously. Obviously, this assumption cannot be expected to be equally approached by members made of different materials, possessing different physical characteristics and subjected to different loading conditions; hence, the lack of a con- sistent trend observed in the actual behavior of members when compared to theoretical predictions. However, this lack of consistency does not appear to be detrimental, since for all cases examined in this program, the theory predicted values of collapse load which were in close agreement with those exhibited by test. As in the previous series of tests, an appreciable increase was observed in the load-carrying capacity beyond that corresponding to initiation of yielding. This increase may be seen in Table IV by the ratio P /P . Unfortunately., not all the para- meters that may influence the increase in the load-carrying capacity can be examined on the basis of the few tests that were performed. However, the remarks made The dependence of the position of the critical section on the axial load, P, in the elastic range may be seen from Eq. k. 21 previously with regards to the influence of slenderness ratio, material and end eccentricities on the ratio P ca /P ya apply equally well in the case of unsymmetrically loaded beam-columns. 22 V. SUMMARY AND CONCLUSIONS In order to examine the inelastic behavior of beam -columns, two theories were developed. An exact numerical procedure as outlined in Appendix A, would yield results as accurate as desired. However, in view of the fact that this method may become rather lengthy and sometimes prohibitive, an approximate, but relatively simple method was developed. Essentially, the approximate method requires a trial -and-error solution of two independent relations betwen the variables axial load, moment and curva- ture. The first of the two P-M-ip relations is established from considerations of the equilibrium of internal forces and moments, and is represented graphically by dimension- less interaction curves along with a moment-curvature equation. The second P-M-$ relation is developed by considering the equality between the internal and external moments, as well as the deflection characteristics of the member. The proposed method makes it possible to determine the magnitude of the axial load (concentric or eccentric) at any stage of inelastic deformation including that at which collapse takes place, for a beam -column which is initially subjected to a constant transverse load. To examine the validity of the various assumptions made in developing the approximate theory, tests were performed on members made of three different materials, namely 2024 -T4 aluminum, SAE 1020 steel and 17-7PH stainless steel. Several parameters were investigated within each material. These included the type of cross section, slenderness ratio, position and magnitude of transverse load, and eccentricity of the axial load. Dimensionless moment-load curves were constructed from the test data and compared to those obtained analytically. Furthermore, in two instances, a comparison was made between the approximate method and the so-called exact method. The various results lead to a number of significant conclusions which may be summarized as follows: 1. The results obtained by the proposed approximate method compared well with those obtained by the exact method. The differences observed between the results of the two methods are too small to justify the added labor required by the exact method. 2, Very good agreement was observed, in all instances, between the moment -load curves predicted by the approximate method and those obtained by test, particularly up to the collapse load. While close agreement was generally encountered beyond the collapse load, there were a few instances in which the discrepancy between the predicted and actual curves became pronounced. 23 3. The difference between the predicted and actual values of collapse loads was, in general; much less than 7 per cent. Thus the proposed approximate method was found to be sufficiently adequate for all practical purposes. 4. In general, inelastic action was found to induce a considerable increase in the load-carrying capacity beyond that associated with initiation of yielding. The increase in the load-carrying capacity was observed to range from a maximum of 159 per cent for SAE 1020 steel, to a minimum of about 2 per cent for 17-7PH stainless steel. 5. The increase in the load-carrying capacity was found to vary directly with M„/M , and inversely with £/r and e . 6. Members with end eccentricities of 15 per cent of the section depth attained a much higher increase in load -carrying capacity than members with zero end eccentricities. 24 VI. BIBLIOGRAPHY 1. Seely, F. B. , and Smith, J. O. , Advanced Mechanics of Materials , John Wiley and Sons, Inc., New York, 1952, pp. 511-581. 2. Sidebottom, O. M. , and Clark, M. E. , "The Effects of Inelastic Action on the Resistance to Various Types of Loads of Ductile Members Made From Various Classes of Metals, " WADC Technical Report 56-330, Pt, I, 1956. 3. Sidebottom, O. M. , and Clark, M. E. , "Theoretical and Experimental Analysis of Members Loaded Eccentrically and Inelastically, " University of Illinois Engineering Experiment Station, Bulletin No. 447, 1958, 4. Von Karman, T. , "Untersuchungen uber Knickfestigkeit, " Mitt, uber Forschungsarbeiten des V. D. I. , H. 81, Berlin, 1910. 5. Chwalla, E. , "Die Stabilitat zentrisch und exzentrisch gedruckter Stabe aus Baustahl, " Akad. d. Wissenschaften in Wien, math. -naturwiss. Klasse, Sitzungsberichte Abt. II a, Bd. 137, 1928, pp. 469-512. 6. Chwalla, E. , "Theorie des aussermittig gedriickten Stabes aus Baustahl, " Der Stahlbau, Vol. 7, 1934, pp. 161-165, 173-176, 180-184. 7. Osgood, W. R. , "The Double -Modulus Theory of Column Action, " Civil Eng. , Vol. 5, No. 3, Mar. 1935, pp. 173-175. 8. Shanley. F. R. , "Inelastic Column Theory, " Jour, of Aero. Sci. , Vol. 14, No. 5, May 1947, pp. 261-267. 9. Wang, C. T. , "Inelastic Column Theories and an Analysis of Experimental Observations, " Jour, of Aero. Sci., Vol. 15, No. 5, May 1948, pp. 283 292. 10. Cicala, P. , "Column Buckling in the Elastoplastic Range, " Jour, of Aero. Sci. , Vol. 17, No. 8, Aug. 1950, pp. 508-512. 11. Jordan, W. D , "Inelastic Behavior of Eccentrically Loaded Columns, " Doctor's Thesis, Dept. of Theo. and Appl. Mech. , University of Illinois, 1952. 12. Bijlaard, P. P. , "Buckling of Columns with Equal and Unequal End Eccentricities and Equal and Unequal Rotational End Strains, " Proc. 2nd U. S. National Congress Appl. Mech. , 1954, pp. 555-562. 13. Galambos, T. V. , and Ketter, R. L , "Columns Under Combined Bending and Thrust, " Proc. ASCE, Vol. 85, EM-2. 1959, p. 1990. 14. Costello, G. A. , "The Creep Buckling of Columns Made of Canvas Laminate. " Doctor's Thesis, Dept. of Theo. and Appl. Mech. , University of Illinois, 1959. 15. Sidebottom, O. M. . Pocs, E. , and Costello, G. A. , "The Effect of End Conditions on the Collapse Loads of Columns, " Dept. of Theo. and Appl Mech, Report No. 585, University of Illinois, 1960, 25 16. Timoshenko, S. , Theory of Elastic Stability, McGraw-Hill Book Co. , New York, 1936, pp. 1-63. 17. Osgood, W. R., "Beam -Columns, " Jour, of Aero. Sci. , Vol. 14, No. 3 Mar. 1947, pp 167-170. 18. Ketter, R. C. , "Stability of Beam-Columns Above the Elastic Limit, " Proc. ASCE, Vol. 81, No. 692, Oct 1955. 19. Clark, M. E. , Corten, H. T. , and Sidebottom, O, M. , "Inelastic Behavior of Ductile Members Under Dead Loading, " University of Illinois Engineering Experi- ment Station, Bulletin No 426, 1954. 20. Morkovin, D. , and Sidebottom, O. M. , "The Effect of Non-uniform Distribution of Stress on the Yield Strength of Steel, " University of Illinois Engineering Experiment Station, Bulletin No. 372, 1947. 21. Sidebottom, O M. , Dharmarajan, S. , Gubser, J. L. , and Leasure, J. D. , "The Effects of Inelastic Action on the Resistance to Various Types of Loads of Ductile Members Made From Various Classes of Metals, " WADC Technical Report 56-330, Pt. XIL 1959. 22. Sidebottom, O. M. , Clark, M. E. , and Dharmarajan, S. , "The Effects of Inelastic Action on the Resistance to Various Types of Loads of Ductile Members Made From Various Classes of Metals." WADC Technical Report 56-330, Pt. VIII, 1958. 26 VII. APPENDIX A In this appendix, a brief outline is given of another method of approach to the problem of beam-columns. This method is capable of yielding extremely accurate results, and is referred to here as the "exact method". Essentially, the exact method differs from the approximate method described in this paper only in the way of relating curvature to deflection in the inelastic range. For simplicity in presenting this approach, a symmetrically loaded beam -column with zero end eccentricities will be considered, as shown in Fig. 1A. Fig. 1A The method of relating curvature to deflection in the inelastic range is similar in principle to that used by Chwalla (6) in his solutions of column problems. It consists of subdividing the member into a number of equal segments, X , as shown in Fig. 1A. The assumption is then made that within each segment, the curvature remains constant (i. e. , within each segment, the deflection curve is a circular arc). The deflection at any section of the. member may then be found by the relation 27 A 2 3X 2 ^-1, N-l. 2, 3 etc. z o (1A) where ip^, designates the curvature at the Nth section along the member, and y is the deflection at midspan. For the beam -column shown in Fig. 1A, Eq. b becomes MP M H- - y^)Y + m:^) (b) e e e where M is a factor varying linearly from unity at the center section to zero at the end of the member. Eqs. 1A and b thus provide one P-M-^i relation. The second P-M-V relation is obtained using Eq. 5 and interaction curves of the type shown in Fig. 3. The solution of the problem is effected by means of the above relations in the following manner. As in the previous solution, point A (see Fig. 3) is determined by the magnitude of the transverse load Q, and point B is determine from the elastic solution. For a given depth of inelastic penetration (k = k.), some value of P/P is assumed and the corresponding value of M/M determined from the appropriate interaction curve. With M = 1, the value of y is then determined from Eq. b. Equation 5 is mow used to determine the curvature, # , at the center of the member. The deflection y may now be determined by Eq. 1A. Using the value of y and the appropriate value for M in Eq. b, the dimensionless moment, M/M , corresponding to the section of the 1 e member at y may be established. Then the curvature lj> . is found by Eq. 5. The values of M /M and k for use in Eq. 5 are found from the interaction curves by inter- polation. This process is continued until the deflection at the end of the member is determined. The assumed value of P/P is correct if the computed value of deflection at the end of the beam -column is zero or very nearly so. Thus to determine one point on the moment-load curve, this numerical procedure may have to be repeated three or four times until the boundary condition is satisfied. In determining the value of M/M from Eq. b for any given section, the procedure consisted of adding the maximum value due to P, to the average value due to Q. Thus for example, (M/M^ = 7

i/r = 50 */r = 75 i/x = 100 i/r = 60 2024-T4 Aluminum Alloy b = 0. 500 h = 0. 625 b = 0. 500 h = 0.625 b = 0. 500 h = 0. 500 t 2 = 0. 100 SAE 1020 Steel b = 0. 500 h = 0.625 b = 0. 500 h = 0.625 b = 0. 500 h = 500 t. = 0. 100 17-7PH Stainless Steel b = 0. 350 b = 0. 500 b = 0. 400 h = 0. 420 30 Table II. Properties of the Three Materials Used in this Investigation. Material E psi x 10 6 a e l 6 psi x 10 a psi x 10 e in/in x 10 e e 2 -4 in/in x 10 a l a 2 2024-T4 Aluminum Alloy 10.8 37.5 51.5 34.7 47.7 0.189 SAE 1020 Steel 30.0 30.3 30.3 10. 1 10. 1 17-7PH Stainless Steel 28.0 170.0 170.0 60.7 60.7 0. Ill 0.111 31 Table III. Values of P ca P /P, and P /P for the Various Symmetrically ca' ct ca ya ' y Loaded Beam -Columns Studied in this Program. P = Actual Collapse Load; P . = Theoretical Collapse Load; and P = Actual Yield Load, ct r ya Material Shape I/r M Q /M e e l e 2 P ca lb. P /P ¥ ca ct P /P ca ya 100 0.25 2,130 1.04 1.05 75 0.25 4,240 1.00 1.09 50 0.25 7,750 7,620 1.03 1.01 1.21 1.18 100 0.50 1,680 1.680 1.04 1.04 1.37 1.37 202 4 -T 4 Aluminum Rect. 75 0.50 3,580 3,510 1.05 1.03 1.45 1.43 Alloy 50 0.50 6,360 6,330 1.06 1.05 1.70 1.69 100 0.25 0.15h 0.15h 1,690 1.03 1. 17 75 0.25 0.15h 0. 15h 3,250 1,06 1.28 50 0.25 0.15h 0.15h 5, 130 1.05 1.50 T- 60 0.50 1,850 1.01 1.27 Section 60 0.25 0.15h 0.15h 1,550 1.03 1.21 100 0.50 3,410 3,400 1.07 1.07 1.44 1.44 75 0.50 5,370 5,400 1.05 1.06 1.49 1.50 Rect. 50 0.50 6,400 6,450 1.02 1.03 1.52 1.54 SAE 1020 100 0.75 2, 460 1.06 2.24 Steel 75 0.75 3 980 1.04 2.36 50 0.75 5,260 5,100 1.07 1.04 2.59 2.54 T- 60 0.50 2,450 1.02 1. 17 Section 60 0.50 0.15h 0.15h 1,640 1,680 1.00 1.03 1.30 1.33 75 0.50 4,770 1.00 1.17 17-7PH Stainless Rect. 50 0.25 13,200 1.01 1.10 Steel 75 0.50 0.15h 0.15h 3,890 1.00 1.30 50 0.25 0.15h 0.15h 9,100 1.04 1.27 32 Table IV. Values of P P /P and P /P for the Various Unsymmetrically ca ca' ct ca ya ' J Loaded Beam -Columns Studied in this Program P = Actual Collapse Ca Load; P = Theoretical Collapse Load; and P = Actual Yield Load, ct r ya Material Shape f/r M„/M Q e e l e 2 P ca lb. P /P . ca' ct P /P ca ya 2024-T4 Aluminum Alloy Rect. 100 75 50 0.50 0.50 0.50 1,740 1,710 3,610 6,480 6,490 1.00 0.98 0,96 0.96 0.96 1.22 1.20 1.32 1.58 1.58 T- Section 60 0.50 1,840 1,890 0.95 0.97 1.15 1.20 SAE 1020 Steel Rect. 100 75 50 75 75 0.50 0.50 0.50 0.50 0.50 0. 15h 0. 15h 0. 15h 3,660 5,680 6, 530 3,590 4,400 1.05 1.04 1.01 1.00 0.96 1.41 1.47 1.52 1.69 1.47 T- Section 60 0.75 2,210 1.01 1.28 17-7PH Stainless Steel Rect. 75 0.25 6,480 1.00 1.02 33 u CD 1 s X! c J3 CO en CD tt CO cd 5 m £ o X! CO J3 & CD J4 CO (I) X! H bj) c 3 cd o J T3 CD .3 X] . s o •— < O 4- U l/J o +-> •a - u ♦j CD Jh •T— > ■a CU CO T3 (U CO XI O 6 en S3 2 cv i| C O >■ M O M cu m CO Pl, H-2 EC 34 CO 03 « e e 2 J i e l i \ / ^Slope = a,E (Compression) — Slope = tt 2 E (Tension) 0) b b _ Slope = E r i ' Strain, e Fig. 2. A Schematic Representation of Stress -Strain Relations in Tension and Compression. 35 1.0 M/M k=0.5 0.6\. ^0.8 N. \. ^\ ^\ ^\ ^\ f\\ N, v\\\l\ \ >v \ \ J N> \ D A N. CV ^V \ Collapse Load \ 3i p/p 1.0 Fig. 3. A Schematic Representation of Interaction Curves for Constant Depth of Yielding. 36 " >•. 6 ed o W T3 W a s I bO .a w -G B '. k 1 y 1 _l 0. 2 .4 0. 6 0. 8 l.O p/p ' e Fig. 12. Comparison of Moment- Load Curves for Symmetrically Loaded 2024-T4 and SAE 1020 Steel Beam -Columns Determinec by Two Theoretical Methods and By Test. 45 1.6 0.2 0.4 0.6 0.8 1.0 P/P £ Fig. 13. Moment- Load Curves for Symmetrically Loaded, Rectangular 2024-T4 Aluminum Beam-Columns. (Mq/M s = 25; e x = e 2 = 0) 46 1.6 1,0 Fig. 14. Moment-Load Curves for Symmetrically Loaded, Rectangular 2024-T4 Aluminum Beam -Columns. (M n /M = 0.50; ei =e 2 =0) Q 47 1.6 1.0 Fig. 15. Moment- Load Curves for Symmetrically Loaded, Rectangular 2024-T4 Aluminum Beam -Columns (M Q /M = 0. 25; ej = e 2 = 0. 15h) ^ 48 1.6 1.4 12 1.0 M/M 0.8 0.6 0.4 0.2 e i =e 2 = M„/M = 0. 2 Q e S./t = 60 Theoretical Experimental e x = e 2 = 0, 15h M^ /M = 0. 50 Q- e 1/t = 60 \ \ X Q e l «- St/2 -* * IP e 2 '■ -« JE i ► , t \ Initiation of Yielding ) \ "V \ \ \ _\j 0.2 0. 4 6 P/P 0.8 1.0 Fig. 16. Moment- Load Curves for Symmetrically Loaded T-Section Beam-Columns of 2024-T4 Aluminum. 49 0.2 p p ' Q *- 1/2 -»- 1 - p 1 1 k y \ V V \ \ \ \J 0.2 0.4 6 0.8 1.0 P/P Fig. 17 Moment- Load Curves for Symmetrically Loaded, Rectangular SAE 1020 Steel Beam -Columns. e= E ' ( X> + <"jL> fl < 8 > I = | + XptP-V (9) e = | + Ka n F(t) (10) Numbers in parentheses refer to corresponding entries in the bibliography. Manuscript released by authors September, 1961, for publication as a WADD Technical Report. a . B = R 1 ( qh qh (23) It is convenient to define the cross sectional area A, by the relation: A - Dh 2 (24) and rewrite equations 19, 20 and 24 in dimensionless form as follows: P q Act M Act h o KD DK' h^ _ K b, B -b„ B + b„ B Ik 2 q-u K 3 q-.y K B 2lL k J b l C K 2 q-u + b. K q-v -b. K Vl K (25) (q-/3)P Act o (26) (27) It will be noted that the right sides of equations 25, 26 and 27 are functions of q and K and are independent of the properties of the material and of the magnitude of the relative dimensions of the cross -section. A trial and error solution is required if a dimensionless moment-curvature curve for a specified load is required. A program has been written for the IBM 650 digital computer to make the necessary calculations. A description of this program is given in Appendix A. The dimensionless moment -curvature curves for the rectangular-section 2 1 1 (b =b 4 =D,b =b „=0) are shown in Figure 2 and for the T- section (b.^j.b = -^-.b =0,b.=-g u= -r, v=0) are shown in Figure 3. Although an IBM 650 digital computer program was written for the general I- section, similar programs could be written for any cross-section made up of a finite number or rectangular elements. 3. Load-Deflection Curves The load-deflection curve for a given column or beam -column subjected to any general loading can be computed by a method of successive approximations (32) pro- vided moment -curvature curves for the given column cross -section are known. See Figures 2, 3, and 4. In order to illustrate the procedure, a point on the load-deflec- tion curve will be computed for the T- section beam -column used in this investigation. Consider the T- section beam-column shown in Figure 4, For convenience in the calculation of the deflection, the depth is assumed to be unity. The magnitudes of 2 1 11 the dimensionless coefficients in Figure 1 are b,= -^ , b = -y, ^>o = ^> ^4 = ~ft> u ~ 4' v= 0> and J3 =0.3393. For a slenderness ratio of 60. 2, its length is 17. 46 inches. The beam-column is subjected to a concentrated load Q at its midpoint and to an axial load P at an eccentricity of 15 per cent of its depth. Since the deflections are symmetric about the center, only half of the column need be considered. The beam-column has been arbitrarily divided into eight equal segments A = ft/8, the end- points of which specify station points a,b,c,d and f. The dimensionless moments due to Q, M n /cr Ah, have been computed for the specified locations and are shown in Fig- ure 4 For this example, P is taken of sufficient magnitude to make P/ffA = 0.7. The moments due to P depend upon the unknown deflections. The first step in de- termining the deflections is to assume a reasonable value for the deflection at each station point . The initially assumed deflections shown in Figure 4 are calculated us- ing the deflection formula for a linear elastic beam-column with E= u /e . The dimensionless moments M p /cr Ah are computed and added to M„/cr Ah to obtain the total moment M/cr Ah . For each dimensionless moment, the dimensionless curvature, h^/e Q , is obtained from the curve for P/Aj = 0.7, Figure 3. Since the curvature is equal to the angle change per unit length, the curvature diagram is some- times called the angle -change diagram. Using the conjugate beam method , deflections are calculated by loading the beam through stringers with the curvature diagram. The number of stringers used corresponds to the number of segments into which the beam is broken, in our case eight. By equations of statics, the concentrated curvature (i.e. finite angle change) at each station point can be calculated. This is analogous to the process of finding a reaction force on a stringer. Assuming the curvature to vary parabolically between any three station points, this gives: 7 € A # a = "H- < 7a + 6b - c) and e X ip h = -^-(a + lOb + c) where ip is the equivalent concentrated curvature, or finite change of angle, at the station point denoted by the subscript. Since the member is assumed to be made up of infinitely stiff segments of length a X = o- with concentrated angle-changes at each station point, the slope of each segment o is easily computed. The slope is zero at the center so that the slope of the segment from a to b is 22. 5 e A/12. The slope for the adjacent segment is equal to the slope of segment a to b plus the angle change at b. In this way the slope for each segment can be calculated. Beginning at the left hand reaction f where the deflection is zero, the deflection of the adjacent station point d is seen to equal the product of the seg- ment length and slope. The deflection of the next station point is equal to this deflec- tion plus the product of its segment length and slope, i.e. 6 = 126. 4e A 2 /12 + X (100.0 c A/12) - 226. 4e A 2 /12 c o o o Using the material properties listed in Figure 8, the deflections are computed and com- pared to the assumed deflections. The calculated deflections do not coincide with the assumed values so that another set of assumptions must be made. The calculated deflections may be used as a closer approximation to the real deflections and the computations repeated. However, the num- ber of trials can be greatly reduced by doubling the correction. For instance the de- flection at the center was 0. 017 inches greater than that assumed. For the next trial assume the center deflection to be : 6 A = 0.125 + 2 x 0.017 = 0.159 in. A Let the other deflections be increased in the same proportion, i. e. : 132 6 b = 07T42 x0 - 159in - The calculated deflections for the second trial are nearly identical with the assumed deflections so that only two trials are required. The above calculations locate one point on the dimensionless load-deflection curve shown in Figure 5. There are some features to be noted for the convergence of the iteration procedure. If the assumed deflection locates a point below the curve, the calculated deflection is less than the assumed. Above the curve, the calculated deflection is greater than the assumed. Therefore, if the assumed deflection is below or to the left of the actual de- flection curve, the numerical procedure will converge to the pre-collapse deflection. The procedure always diverges from the post-collapse deflection. Since the load-deflection curve can be presented in this dimensionless form, it is independent of the material. It is valid for any material whose isochronous stress - strain diagram is adequately represented by Equation 16. It can be used to determine the collapse load for any magnitude of time for which values of o and e are known. III. MATERIALS AND METHOD OF TESTING 1. Materials Used, Test Members, and Testing Procedure All test members were machined from a 1/2 inch by 2 inch bar of 17-7PH stain- less steel. The compression specimens had square cross sections, 1/2 inch on a side. The columns had rectangular or T -shaped cross sections with lengths and depths as indicated in Table 1. The T- section beam-columns were machined before being precipitation hardened, while the rectangular section test members were machined after the heat treatment. The specimens were heated to 1400°F. for 90 minutes, cooled to 60°F. in 60 minutes, held at this temperature for 30 minutes, then heated to 1050°F. for 90 minutes, and finally air cooled. All of the test members were loaded in a Riehle testing machine of 120,000 pound capacity. The machine was equipped with a load holder to maintain any desired load. Temperatures were measured by vertically spaced thermocouples along the test specimens. Two thermocouples were used on the short compression specimens, while three were used on each of the beam-columns. A piece of asbestos was used to cover each thermocouple as it was attached to the test member. An asbestos shield was placed between the test member and the heating coils, and baffles were placed in the furnace to prevent a chimney effect. Approximately one hour was necessary to bring the furnace up to temperature. The temperature was manually controlled and maintained at 972°F. + 2°F. Deformation readings were started as soon as the load was applied, and were taken every minute thereafter. 2. Properties of Materials Compression creep properties were obtained for the 17-7PH stainless steel at 972°F. The fixtures and furnace used in making the compression tests are shown in Figure 6. The deformations were measured with a Riehle dial-type high temperature creep extensometer with a 2 inch gage length. The compression creep curves for the material are shown in Figure 7. From these creep curves, corresponding values of stress and strain were obtained for zero time, for 30 minutes, and for 60 minutes to give the isochronous stress-strain diagrams shown in Figure 8. An arc hyperbolic sine curve (Equation 16) was fitted to the data for 30 minutes. The pertinent material properties for the material are listed in Figure 8. The heat treatment of the material used in this investigation was carried out at a temperature 15 to 30°F. higher than in previous investigations (33, 34). This is be- 10 lieved to account for a slightly higher value of u than before. Also, the data for zero time in Figure 8 indicates that some inelastic deformation occurred before the test load was reached, whereas in the previous tests on eccentrically loaded columns, the load deformation curve remained linear for zero time. 3. Method of Loading Beam-Columns The fixtures used in loading the beam-columns, the furnace and the 1/1000 inch dial indicator are shown in Figure 9. The transverse load was applied by means of dead weights attached to a wire, which passed over the pulley arrangement seen at the right of the figure. The axial load was applied to the beam-columns through the knife edge fixtures illustrated in Figure 10. The length of each knife edge was 2 inches. This fixture added 0. 60 inches to each end of the beam-column. In order to offset the effect of any initial crookedness in the beam column, the eccentricity was measured with respect to its midpoint. The error in measuring the eccentricity is believed to be less than plus or minus 0.002 inches. UNiVERsiTy of 11 IUIN0/S LIBRARY IV. ANALYSIS OF RESULTS A total of twenty-one beam-columns were subjected tq constant axial and lateral loading at 972°F. Four different slenderness ratios, and two types of cross-section were considered. Eighteen of the beam-columns had rectangular cross sections. These were split into three groups of six each with slenderness ratios of 50.0, 75.0, and 100.0. Half of each group were tested with the axial load applied through the centroid of the section, zero initial eccentricity, while the other half were subjected to an axial load with 15 per cent initial eccentricity. The deflection-time curves for these beam-columns are shown in Figures 11 and 12. The deflection-time curves for the three T- section beam-columns are shown in Figure 13. Each T-section beam-column had a slenderness ratio of 60. 2. As noted in Figure 15, two of these had zero initial eccentricity, and the other had an initial eccentricity of 15 per cent of its depth. The deflection of each beam-column is shown in dimensionless form in Figures 14 throughl7. The initial deflection, the maximum deflection, and one half maximum deflection are given. The theoretical load- deflection curves for zero time shown in these figures are based on the assumption that the material behaves elastically with E = a /e (37). The actual test data falls to the right of these curves since the ma- terial did not behave elastically while coming up to test load. The theoretical load -deflection curves for 30 minutes, shown in Figures 14 through 17 were constructed using the arc hyperbolic sine theory presented in Sec- tion II. Since the creep curves in Figures 11, 12 and 13 indicate that the deflection of the beam-column immediately prior to collapse was independent of the time to collapse, the experimental deflection may be compared with the theory for 30 minutes. It will be noted that the deflections of the beam-columns corresponding to one half the max- imum deflections are approximately equal to the deflections at the peaks of the theoreti- cal load -deflection curves. The theoretical deflection at the collapse load is not high- ly accurate while for loads appreciably below the collapse load the theoretical creep deflections are small and quite close to the experimental values. This discrepancy between the predicted and experimental values of the deflection at the collapse load is not considered significant, since it is the buckling load which is of primary importance. Experimental values of P/A are listed in Table 1 for all of the beam-columns along with the time to collapse. Also listed in Table 1 are computed values of P/A necessary to cause each beam-column to collapse in 30 minutes. The per cent differ- ence between the theoretical collapse load at 30 minutes and the theoretical collapse 12 load at other values of time was plotted as a function of time. The appropriate per cent difference was applied to the test load to obtain the adjusted load. Ratios of the theoretical to experimental collapse loads for 30 minutes are listed in Table 1 for each beam-column. The theory ranged from 6 per cent conservative to 12 per cent nonconservative, with an average of 2. 6 per cent nonconservative. None of the variables considered indicated any particular trend. The scatter was no more pro- nounced for the beam-columns than for the creep specimens. (See Figure 7). 13 V. SUMMARY AND CONCLUSIONS 1. Summary A method is presented for predicting the load- deflection relationship and the collapse load for beam-columns subject to creep. This method makes use of: a) a stress -strain-time relation which is obtained by fitting the arc hyperbolic sine equation to the isochronous stress-strain data. b) a set of dimensionless moment -curvature interaction curves. The IBM 650 digital computer was used to facilitate the construction of these curves. c) a method of successive approximations (32) for obtaining the load-deflec- tion relationship. To check the accuracy of the method developed, several tests were made on beam-columns at elevated temperature. In all, twenty-one 17-7PH stainless steel beam -columns were tested to collapse, at a constant temperature of 972°F. T-section beam-columns having a slenderness ratio of 60.2 and rectangular- section beam-columns having slenderness ratios of 50.0, 75.0, or 100.0 were used. Each beam-column was subjected to a concentrated load at midspan and a constant axial load having an initial eccentricity of zero or 15 per cent of its depth. 2. Conclusions 1. The arc hyperbolic sine function (Equation 16) adequately represents the isochronous stress-strain diagram for 17-7PH stainless steel at 972°F for a time dur- ation of thirty minutes. 2. Since the effect of stress redistribution with time is not directly considered in the theory, it is recommended that the experimental constant o be increased by ten per cent when calculating the collapse load or the load -deflection relationship. This empirical correction is in agreement with the results of previous investigations on eccentrically loaded columns (34, 36). 3. The IBM 650 digital computer can be conveniently used to calculate points for dimensionless moment -curvature interaction curves. A program is included in the appendix, which is valid for a general I-section. This includes the T-section and rec- tangular-section as special cases. Approximately five to fifteen minutes was required for the machine to make the computations for the particular cross- sections considered in this investigation. 4. Since the dimensionless load-deflection relationship is independent of the numerical values of a and e , these theoretical dimensionless load -deflection curves o o are independent of both time and material. 14 5. Test data from the twenty-one beam- columns indicated that the theory ranged from 6 per cent conservative to 12 per cent nonconservative in predicting the collapse load for thirty minutes. The theory averaged 2. 6 per cent nonconservative. No trend was noted for any of the variables investigated. The scatter was of approximately the same magnitude as that exhibited by the compression creep data. 15 BIBLIOGRAPHY 1. Bleich, H. H. & Dillon, D. W. Jr. , "Deformation of Columns of Rectangular Cross Section", Journal of Applied Mechanics, December 1959, pages 517-525. 2. Davis, E. A. , "Creep of Metals at High Temperatures in Bending", Journal of Applied Mechanics 5, A- 29. (1938). 3. Findley, W. N. and Poczatek, J. J. , "Prediction of Creep Deflection and Stress Distribution in Beams from Creep in Tension", Journal of Applied Mechanics, Vol. 77,. 1955, pages 165-171. 4. Finnie, I. and Heller, W. R. , "Creep of Engineering Materials", McGraw-Hill Book Company, 1959. 5. Freudenthal, A. M. , "The Inelastic Behavior of Engineering Materials and Struc- ture", John Wiley & Sons, 1950, First Edition, page 516. 6. Gerard, G. , "A Creep Buckling Hypothesis", Journal Aeronautical Science, Vol. 23, September 1956, page 879. 7. Hilton, H. H. , "Creep Collapse of Columns With Initial Curvature", Journal Aeronautical Science, Vol. 19, December 1952, pages 844-846. 8. Hoff, N. J. , "A Survey of the Theories of Creep Buckling", Proceedings of the Third U. S. National Congress of Applied Mechanics, pages 29-49. 9. Hoff, N. J., "Rapid Creep in Structures", Journal Aeronautical Science, October 1955. 10. Hu, L. W. and Triner, N. H. , "Bending Creep and Its Applications to Beam Columns", Journal Applied Mechanics, Transactions A. S. M. E. , Vol. 78, pages 35-42. 11. Kempner, J., "Creep Bending and Buckling of Nonlinearly Viscoelastic Columns", N.A.C.A, Technical Note 3137, January 1954. 12. Kempner and Patel, "Creep Buckling of Columns", N.A.C.A. Technical Note 3138, 1954. 13. Kempner, J. , "Creep Bending and Buckling of Linearly Viscoelastic Columns", N.A.C.A. Technical Note 3136, 1954. 14. Libove, C. , "Creep Buckling of Columns", Journal Aeronautical Science, July 1952, Vol. 19, No. 7. 15. Libove, C. , "Creep Buckling Analysis of Rectangular Section Columns", N.A.C.A. Technical Note 2956, June 1953. 16. Lin, T. H. , "Creep Deflections and Stresses of Beam Columns", Journal Applied Mechanics, March 1958, pages 75-77. 16 17. MacCullough, G, H. , "An Experimental and Analytical Investigation of Creep in Bending", Transactions A. S„ M. E, , Vol. 55, APM-55-9, 1933. 18. Marin, Joseph, "Creep Deflections in Columns", Journal of Applied Physics, Vol. 18, No. 1, January 1947, pages 103-109. 19. Marin and Hu, "Deflection of Members Subjected to Bending Accompanied by Creep", Proceedings of First National Congress of Applied Mechanics, 1952, pages 613-618. 20. Pao, Yoh-Han and Marin, J. , "Deflections and Stresses in Beams Subject to Bending and Creep", Journal Applied Mechanics, Vol. 74, 1952, pages 478-484. 21. Sanders, J. L. Jr. , "A Variational Theorem for Creep with Applications to Plates and Columns", N.A.C.A. Technical Note 4003, 1957. 22. Shanley, F. R. , "Weight -Strength Analysis of Aircraft Structures", Dover Book 1960, pages 369-384. 23. Tapsell, H. J. and Johnson, A. E. , "An Investigation of the Nature of Creep Under Stress Produced by Pure Flexure", Institute of Metals Journal, Vol. 57, August 1935, page 121. 24. Johnson, A. E. , Mathur, V. D. and Henderson, J. , "The Creep Deflexion of Magnesium Alloy Struts", Aircraft Eng. , 28: 419, 1956. 25. Odqvist, F. K. G. , "Influence of Primary Creep on Column Buckling", Journal Applied Mechanics, 22: 295, 1954. 26. Hult, J. A. H. , "Critical Time in Creep Buckling", Journal Applied Mechanics, 22: 432, 1955. 27. Hoff, N. J., "Creep Buckling", Aeronautical Quarterly, 7:1, 1956. 28. Lin, T. H. , "Creep Stresses and Deflections in Columns", Journal of Applied Mechanics, Transactions A.S.M. E. , Vol. 78, 1956, pages 214-218. 29. Shanley, F. R. , "Weight-Strength Analysis of Aircraft Structures", Dover Book 1960, pages 263-305. 30. Carlson, R. L.' and Manning, G. K. , "A Summary of Compressive-Creep Char- acteristics of Metal Columns at Elevated Temperatures", WADC Technical Report 57-96, 1958. 31. Kauzmann, W. , "Flow of Solid Metals from the Standpoint of the Chemical Rate Theory", Transactions American Institute of Mining and Metallurgaical Engin- eering, Institute of Metals Division, Vol. 143, 1941, page 57. 32. Newmark, N. M. "Numerical Procedure for Computing Deflection, Moments and Buckling Loads", Proceedings of A.S.C.E., Vol. 68, 1942, pages 691-718. 33. Sidebottom, O. M. , Costello, G. A. , Dharmarajan, S. , "Theoretical and Ex- perimental Analyses of Members Made of Materials that Creep", University of Illinois Experiment Station Bulletin No. 460. 17 34. Costello, G. A., Sidebottom, O. M. andPocs, E. , "Inelastic Design of Load Carrying Members", Theoretical and Applied Mechanics Report 178, "Effect of End Conditions on the Collapse of Load Carrying Members". 35. Dharmarajan, S. and Sidebottom, O. M. , "Inelastic Design of Load Carrying Members", Theoretical and Applied Mechanics Report 174, "Theoretical and Experimental Analyses of Circular Cross-Section Torsion-Tension Members Made of Materials that Creep". 36. Gubser, J. L. , "Theoretical and Experimental Analysis of Eccentrically Loaded 17-7PH Stainless Steel Columns at 1000°F", Unpublished Master's Thesis, University of Illinois, 1959. 37. Timoshenko, S. P. and Gere, J. M. , "Theory of Elastic Stability", Second Edition, McGraw Hill Book Co. , 1961. 38. Sidebottom, O. M. , Clark, M. E. and Dharmarajan, S. , "The Effects of In- elastic Action on the Resistance to Various Types of Loads of Ductile Members Made from Various Classes of Metals -- Part VIII Eccentrically- Loaded Tension Members Made of Two Stainless Steels Tested at Elevated Temperatures", WADC Technical Report 56-330, April 1958. 39. Schweiker, J. W. , "Creep of Thick-Walled Cylinders Under Internal Pressure and Axial Load", Unpublished Ph. D. Thesis, University of Illinois, 1961. 18 APPENDIX A The purpose of this appendix is to present a program for the IBM 650 digital computer for computing points on constant load moment -curvature interaction curves. Relations expressing dimensionless load, dimensionless moment, and dimensionless curvature as functions of the cross section (see Figure 1) and of the variables K and q are given by equations 25, 26 and 27. Before starting the program it was necessary to decide on limits for K and q and the dimensionless load P/Aa . Dimensionless moment- curvature interaction curves were constructed for dimensionless load incre- ments of 0. 1 as the dimensionless load varied from 0. 2 to 2. 0. At each load, a mo- ment and curvature were computed for each value of K which was applicable; the values considered were 1, 1.5, 2, 2.5, 3, 4, 5, 7, 9, 12, and 15. The magnitude of q was limited by considering only those values which would make 1/q greater than 0.60. Figure 18 shows the flow diagram for the IBM 650 digital computer program. It will be noted that there are three different dimensionless loads listed. P, /Aa is the b' o dimensionless load for which values of dimensionless moment and curvature are de- sired: P /Aa is the dimensionless load calculated for assumed values of K and a o 1/q; P /Aa is the dimensionless load calculated for previously assumed values of K and 1/q. The procedure followed in the computations can best be illustrated by consider- ing Figure 19 which is a plot of P/Aa versus 1/q for a rectangular-section. Con- sider the problem of calculating the first point on the moment -curvature interaction curve for Pj 3 /Aa=0.7. The load P,/Aa is computed for initial values of K and 1/q of 1 and 0. 6 respectively. Since this load is below the desired load, the next value of K is considered. This determines P„/Ao" . This load is above the desired 2 o load so that increments of 0. 12 are added to 1/q to calculate P /Aa , P J Aa , and /H 3' o 4' o P c /Aa which is below the desired load. This value of load is placed in the location 5 o r of P /Aa and an increment of 0. 04 is subtracted from 1/q to compute P,/Aer . Since this load is above the desired load, a linear interpolation between P,- & P, is used to calculate the desired 1/q. With K and 1/q known, the dimensionless mo- ment and curvature are calculated. Table 2 lists the entire program except for the square root (FLSQR) and log (FLLNX) subroutines. To describe important steps in the program, notes have been placed on the printed program. These notes correspond to the following commentary: NOTE 1. Three cards are read into the immediate access storage to locate constants 19 which are valid for all the cross sections. Card 1 Card 2 Card 3 Location Constant Location Constant Location Constant 9000 K = 1 9008 K = 9 9010 K= 15 9001 K- 1.5 9009 K = 12 9011 l/q = 0.6 9002 K= 2 9012 Al/q= 0.12 9003 K = 2. 5 9013 Al/q = 0.04 9004 K= 3 9014 P/Aa =0.2 o 9005 K = 4 9015 AP/Aa =0.1 9006 K = 5 9007 K = 7 NOTE 2. One card is read into the immediate access storage to locate constants which are peculiar to a given cross section. Card 4 Location Constant 9020 -u 9021 -v 9022 b l 9023 b 2 9024 b 3 9025 b 4 9026 D 9027 P NOTE 3. Initial values for load, 1/q and index registers A, B and C are set. Index register A is used to keep track of the four values of EL, and C„ in Equations 25 and 26. Index register B specifies the particular value of K being used in the calculations. Index register C is set to zero after each interpolation to indicate final calculations. NOTE 4. This is a subroutine to solve Equations 21 and 22 for the four values of N in Equations 25 and 26. 20 NOTE 5. Absolute values of N (except for N = K) are calculated and sent to the subroutine to solve Equations 21 and 22. (See NOTE 4.). NOTE 6. Equation 25 is solved. Calculations were made for the rectangular section at the first location in the program and for the I- section at the second location. NOTE 7. Index register C is checked for zero. If zero, goto NOTE 11. If not zero, the difference between P„/Acr and P k /Act is computed. (Go to NOTE 8 .) . a o do NOTE 8. The upper accumulator is checked for negative. If P /Act - qn H q£l 1}A d o •M 3 -Q ■rH Sh •M CO •rH Q •rH 3 u C/3 T3 CJ a a bD C BJ Vh J-i < s ■rH T3 cd o rt o ■rH o -«- * — ■* «— \ — ». <- A — »- A - 1/8 <— e = f p II ^ " ~~~t^--_ =r ^~^=rrr; 6 1 = . 15h M Q /a o Ah Assumed 6 A =5 + e M p /a Q Ah M/a Ah=(M„+M_)/a Ah o P Q o h ^ e o Equiv. Concentrated ij) Average Slopes Calculated 6 Calculated 6 f d c b a Common Factors ,0159 .0319 .0479 .0638 .050 .219 .113 .125 .150 .200 .219 .263 .275 .1050 .1400 .1673 .1841 .1925 .1050 .1559 .1992 .2320 .2563 1.49 2.22 2.93 3.52 3.96 26.4 35.0 42.1 22.9 6 A/12 126. 4 100.0 65.0 22.9 e A/ 12 126.4 226.4 291.4 314.3 e AY 12 o .057 .103 .132 .142 Assumed 6 064 115 148 159 A = 6+ e .150 .214 .265 .298 .309 M p /o- o Ah .1050 .1497 .1855 .2085 .2165 M/a Ah .1050 .1656 .2174 .2564 .2800 h$f€ 1.49 2.35 3.25 3.96 4.98 Equiv. Concentrated ip 28.2 38.8 47.3 25.9 e A/ 12 Average Slopes 140. 2 112 73.2 25.9 - e A/ 12 o ' Calculated 6 140.2 252.2 325.4 351.3 e A /12 Calculated 6 .064 .114 .142 .159 Figure 4. Sample Deflection Calculation for T-Section Beam -Column 30 6/he Figure 5 Dimensionless Load -Deflection Curve for T-Section Beam-Column (£/r = 60. 2, Q = 2a I/ci, e/h = 0. 15) 31 4) O 0) u D C o E u 4) Q. uo O- 0) O i_ U c o t/> 0) o U c "g « -£ E u u £ Q. < V5 _ c o w VI 01 a. i x c 01 a. 11 E 111 V 1- 1- h. -a o a> *^- o 01 > 3 01 X LL! UL LL 32 006 007 006 005 w c .004 a u w 003 002 001 30 Time, min. Figure 7. Compression Creep Curves for 17-7PH Stainless Steel at 972°F„ 33 70,000 CO CO CO 0) u en 60,000 50,000 40,000 30,000 20,000 10,000 ;^ cyb a / o -o A A A Stress -Strain Diagram Representation CD O o U 6 \ m \ \ o x o • \ \ X t^ — i \ \ '— ' X. II II \ \ " x^ 14 \ \w o CN O l> vO o 4-> CO H •rH fc 00 0) o" ■a s o 4-J u 0) ■l-l 3 Oh o £ ^H o u >> XI T3 o > u cd b^ 3 9 u cu U o CN ON —1