AN INVESTIGATION OF THE ONE-HINGED STEEL ARCH AND ITS COMPARISON WITH OTHER TYPES A THESIS presented to the Faculty of the Graduate School Cornell University for the degree of Dodtor of Philosophy by Nee Sun Koo, B.S., M.C.E., McGraw Fellow in Civil Engineering, 1920-’2 1 1921 Reprinted from the Cornell Civil Engineer, volume 29, pages 110, 129, 150, March, April, May, 1921 Digitized by the Internet Archive in 2017 with funding from University of Illinois Urbana-Champaign Alternates https://archive.org/details/investigationofoOOkoon AN INVESTIGATION OF THE ONE-HINGED ARCH AND ITS COMPARISON WITH OTHER TYPES By Nee Sun Koo, B. S., M. C. E. (1919). McGraw Fellow in Civil Engineering, 1920-21. An Abstract of a Thesis to be Presented to the Faculty of the Graduate School of Cornell University for the Degree of Doctor of Philosophy. PREFACE During the last thirty years much study has been given to the design and construction of steel arches by American engineers and investigators. Two- hinged and three-hinged arches seem to have met the greatest favor, while few no-hinged steel arches have been built in this country. One-hinged arches are practically unknown in America, although a few have been successfully constructed on the European continent. After considerable study, the author has been in doubt of the practicability and value of the one-hinged steel arch. He could see no logical or conspicuous reason against its adoption. A number of writers attack it bitterly while enumer- ating a few disadvantages, but fail to demonstrate or justify their statements. Others deem it un- necessary to give a full treatment, because it' has not come into popular use. The deficiency of theoretical knowledge may be the reason why the building of one-hinged steel arches has been at- tempted so rarely. With this idea in mind, the author has undertaken a special investigation of the one-hinged steel arch. While this work is undertaken purely for the purpose of discovering and publishing some un- known facts, some contributions made by previous investigators will be mentioned. One of the noblest and most remarkable preliminary designs which have ever been made in bridge engineering was con- tributed by Charles Worthington in competition with other designs for the famous Quebec Bridge. It was a one-hinged steel arch with a span of 1,800 feet, more than .twice as long as that of any arch bridge then existing in the world. One of the most noted American bridge engineers. Dr. B. A. L. Wad- dell, in his book called “Bridge Engineering,” praises the work of Mr. Worthington and calls the design ingenious and quite feasible. It is to be regretted that the scheme was not accepted by the Canadian Government and thus the complete treat- ment of the theoretical and practical problems in- volved was prevented. The work gives some indi- cation of what could be done with the one-hinged arch and done economically. Dr. Waddell’s com- ment is valuable on account of his extensive ex- perience and gives promise for the future. In view of these facts, the author felt encouraged to carry on this investigation. Special emphasis is laid upon two points, first, to make an extensive study of its behavior in carry- ing the load ; and second, to reveal its characteristics by critical comparisons with other types of arches. It is his hope that the work may be of value to the engineering profession in the future, if not at pres- ent, Since so little has been written in engineer- ing literature upon the subject, it is hoped that every bit of the original work in this thesis will appeal to the sympathetic interest of engineers and investigators. Acknowledgment is due to Profes- sor Henry S. Jacoby, chairman of the special com- mittee in charge of the author’s graduate work at Cornell University, for his aid and helpful sugges- tions. Reactions Solved by the Author’s Method of Symmetrical Deflection Equations. Two kinds of beams, either straight or curved, are used in bridge building; those which are statically determinate, and those statically inde- terminate. For the solution of beams of the first class, three statical equations are available: — = O ( a > EH- O (b) EM O (c) For the solution of beams of the second class, three elastic conditions are available, besides the statical conditions. These are Ay -f- f Pixels El Plds (d) (e) (f) where Ay is the horizontal deflection, Ay the ver- tical deflection, and jo the change of the slope be- tween any two points on the beam, E the modulus of elasticity of its material, I the moment of in- ertia of a section, and .1/ the moment at any point between a and b , where I is taken. By means of these six fundamental equations, all beams can be solved. (For the derivation of formulas (d), (e) and (/) see standard books on' Mechanics). The author lias used these equations in deriving general formulas of reactions for no-, one-, two-, and three- hinged arches. Let an arch-rib with a span l and rise h be fixed at two ends a and b, and hinged at the crown c and be subject to a vertical load P at a distance kl from the left end. (Fig. 1). The load is sustained by the reactions H x , V x and M x at the left support, and by H 2 , V and M 2 at the right support. There are six unknowns and six conditions are required for its solution. The three statical equations furnish three conditions, while the hinge at the crown in- sures that the moment at that point is zero. It remains for us to find two more elastic conditions. In other words, we must choose any two equations from ( d ), (e) and (/) and apply them to the case of the one-hinged arch. In order to simplify the algebraic work, the fol- lowing notations are proposed by the author: — vertical line at the crown. (Fig. 1). Formulas (d) and (e) are used in finding the vertical and horizon- tal deflections between a and c and between b and c. As the supports are actually fixed, the vertical deflec- tion between a and i must be equal to that between b and c; thus giving the first condition. Also, the horiz- ontal deflections between a and c and between b and c must be equal in magnitude and opposite in di- rection, because what is lengthened in one-half of the rib must be shortened in the other. Thus, we obtain the second condition. For the left half of the rib with the origin at a the moment at any point ■between a and c is M = rij -t\£a? ~Hy -P(~c-fd) ••••(!) For the right half of the rib with the origin at b the moment at any point between b and c is M=Mzi-\t,3c-Hy (2) Substituting (1) in ( d ) and ( e ) and replacing the integral forms with the notation above given, we have the horizontal and vertical deflections be- tween a and c respectively, EA.ir - i- V x r ~/it -P(r 2 -/ilq 2 )' • (3) JEAy =7^/? i-lfs-Hr -P(p^-kCp^ • • (i Substituting (2) in the same formulas and making the same substitutions for the integral forms, we P=P, +P? q = + s divided into two parts at the crown by an imaginary y - Ifs -Hr -P(k> -k. ZpP) — -Hr 1 -h Ifr -Ht ~P(r 3 -ftlq e ) = ~(j% q +l£r-Ht) V (8) These two equations together with the three statical equations and the equation furnished by the hinge at the crown as shown below are sufficient to solve all the unknowns. M, .... (9) / T*L / /l • • • ( 10 ) j% -a* - m^ sk ) -° V+\£-P=o H, -H^ = o ... ( 12 ) Thus, by solving (7), (8), (9), (10), (11) and (12;, we have 4 B'.-iZ-rMlL-ff- ( A) v _ „ Ip-kLpx-s -ss (B) * Ip -as (C) Mj = - Vi fj + &(j-zk) (D ) M 2 ~ Ml 1-Jfl PI (l-fi ) (E ) It must be remembered that these formulas hold good only when the load is on the left of the crown. They are applicable to a one-hinged arch with any form of arch-ribs. 1 F/gJ v. c 0 $ \Yz K- A- f — 3 fir Mis FO t i i F/g.2. These formulas again hold good only when the load is on the left of the crown with its direction toward the left. They are also applicable to any form of arch-ribs. Reactions Solved by Author’s Cantilever Method Another interesting method! is derived by the author in securing the two elastic equations for solv- ing the reactions of a one-hinged arch. It is called the cantilever method, because the main feature lies in separating the arch into two separate cantilevers free at the crown and fixed at the supports. The vertical and horizontal deflections at the free end of each can- tilever are then found and equated 1 so as to furnish the two necessary equations. The method is very simple, because the deflections at the free end's of can- tilevers can be easily obtained. Thus, the following deflection table includes all the data needed for the solution : Table 1. Deflection Table for Curved Beams Load/ng D/agra/7? Ax. at c Ay at c- Vert Lootof of M from <=> VLAR. - # H- ^ hor Load at k! from o ( -jr (dkfi-r) Vert Loacf of free enof _ |C -f ~ ^ (ip-ri) Hor Loo of at free esief !/< £ o\ i__i + (hfi- r) e-.J.s. If instead of a vertical load P on the span, we place a horizontal load P at the same position acting towards the left and at a distance k'h from the line ab, (Fig. 2) the problem can be solved in a similar way. In this case the direction of II. > and V 2 are reversed in comparison with those under the vertical loading. The following formulas are derived for the horizontal load on the span with the same process : «• _ d shg-hk'g,-t~t fj, 3 y ^ V -knpi as> -ip m 2 =Mj -jpfih ....(F) ....(G) ....(H) .•••(I) The expressions in this table can be easily derived by using formulas (d) and (e). / In order to find the elastic conditions, let us sepa- rate the arch at the crown. Since there is no load on the right half of the rib (See Fig. 3) and since the right reaction line must pass through the crown, the two forces H and V 2 must act at the free end of the right cantilever with their directions as shown. Therefore we can consider the left cantilever as sub- jected to two external forces H and V 2 at the free end. On the left half of the rib there is a load P. But since action must equal the reaction, the forces acting at the free end of the left cantilever must be H and V 2 . Let ns find the vertical and horizontal deflections between a and c when the loads H, V 2 , and P are act- ing on the left cantilever. Evidently, the vertical and horizontal deflections at c are contributed by three factors, those due to P, H and V 2 respectively. From table 1 these factors can be summed up in the ex- pressions, J2jk.p = -f-H(hp-r) —p(klpi -Sjj -( 13 ) =\^(Lq-r) (14) Similarly, the vertical and the horizontal deflections at c of the right cantilever are contributed by two fac- tors, those due to H and V 2 . From table 1, we have the expressions, 5 Table 2 — Reactions and End Moments of a One-Hinged Ar h with Various Forms of Arch-Ribs. 1 k/ P 0-2)1 Si* / -w /a < r l '' One-hinged Arch m'V K-K* n* 3z- 3 r* 0 = 3tt-8 oc=sin'(zk-/j g = r-h cosec- ° T7- r- +i£±l M th 1 * I 0 Sec 3 E * Constant P r= fiityds n-r^yf 6 S.jfxVs s, s A\^ K = zarfr- §)(” - oc)+l(f-izr-ahg) d « L2r*l-l 3 -z4r z g(j£ -oc) NOTE- / Notations as shown above.. Z. Assumptions, 1 = 1° Sec e. E = constant, shear nea- lected 3. Formulae under Bending F)omenf ho/d good, when the toad is on tn> of the span. CURVE: 15 I « § CQ 4j K led e left ha If b <& 5 § | u> i: ^ u H Vi 14 Mi Me H, Uz V, Vz Mi Me n, Mi Hi Ml GENERAL p kiy-'i . 2(h 9 -t) P bApi-s, Lp-ZS -V,±+tih-rQ(,- Z k) M,+V,L-Pl(l-k.) peho-k’hg.-t-t' Z(hq-t) pJdhOi-t, pJL - . h'fr p, ZS - Lp . p n - Ftp , Zs-lp vd + h,h-Ph(l-k) M,+V,L - Pk'h Set" l z^-tyin' ^ Eet°lh if. ZLL- Z(hq-t)* -ln 2 - sL ZChq-t) shL 2(h ? -t) PARABOLIC #f (***-**) pl [« ) 4Ph(k-k - 7k) /3k? 4k) -4Ph(3k-VV-t4K) ± iSFe t'lo zh*+i5n 3 zp 15 E24c'+2(c-*c)]\g+r6tni \6lrg(/-4K)sinoc' pjsiRecV^ sinZoc] [fek z sinoc'--§KJ U +d,h-Ph(l-k) M,fti,l~Pk'h ± /z Eet'll* K+izln* T izket°ll°h K + IZlr,* ! 2 sl I , y. IZslhL H Table 3 — General Formulas of Reactions and End Moments of an Arch with the Regular Form of Arch Rib. P ,ft*p r + S - ft 4^ds Jo l p*-ft ¥ 6 <1- find? sz Mj ^ iMffyds t -/j q.- £nds tpjf'gtds n.Jf^s 1 * I 0 sec e r 2 .fi*qds £ ■ ConsEarrb NOTE- /. Formulae for one and three hinged Arches under the term Bending Moment hold good, when the load is on the left of C. 2 . B/se or Fall of Temperature has certain effect upon M of the three hinged. The formula is not derived, as its effect upon the truss can best be found bu the graphical method, d — ) 3. Notations are shown above, r, - radius of Gyration s =■ unit compressive stress in Pib-shortening Formulae. (Average stress) ARCH NO-HI NGEzD ONE-HINGED TWO-NINGLD THEE L -HINGE D £ % ? 1 Q § -4 O K 0) s H Bl(brk.l<} 2 )-l kloi — n Zfrq-t) a /7 + k.lQi. z-t P>k.l ~zK V, + F-Q-k) f=> IfO -kl)0,-S-S* lp-zs P(L-k) pQ-k) Vz P-V, p kip, -3, lp—2s Pk Pk Mi 4^0 Qlq - 3 d)-Pkl(/-kf -V/L-r Hh + P) Q-Zk) o o Mz pr, + Kl -Pl(/-k) M, +-V/L- PlQ-k) o o MOEUZON TA L LOAD n dfa-k'hqj-qlzq,- s S\ If k'UQ-te ) 1 > 2 L it-z(klp, -s,)' - v ^kp- r ) -H(hq-t) (18 ) from which the formula for H can be obtained di- rectly. From the statical conditions and the condi- tion furnished by the hinge at the crown, we can obtain all necessary formulas for the reactions and the bending moments at the ends. In the case of the horizontal loading, the above method applies equally well. The expressions for the vertical and horizontal deflections at the free end of a cantilever under a horizontal load P are also given in table 1. Influence of Temperature Changes in temperature cause changes in the values of and II i but do not affect the vertical reaction l 7 ^ It. is usually specified that an arch shall be designed to be subject to a certain variation in temperature from a standard value. Let 2 be the coefficient of expan- sion and r the rise of temperature, then the spau length will be increased by %t°l provided one end is free to move. As both ends are fixed in position when the supports do not yield, equal and opposite reactions and end moments are produced. The values of II, and 1/j must be such as to prevent the horizontal dis- placement 2t°£, which is due to the effect of bending and thrust. The horizontal deflection due to bending and thrust are given by the following expressions: in which A is the area of cross-section. Equating the deflections, and substituting H^h-y) for M and solving for H 1 and M u we have ■/. Eet°l (K) (L) where Io is the moment of inertia of the section at the crown and )\ is the radius of gyration of the same sec- tion. For a rise in temperature use the positive sign for II 1 and a negative sign for M\. For a fall in tem- perature, the reverse is true. Rib-Shortening The direct effect of the thrust along the axis is to shorten the axis of the arch-rib. It would also shorten the span provided one end were free to move, but as this is not the case it will develop equal and opposite negative reactions H 1 and positive moments M\. The effect of displacement due to II 1 and M\ must equal that due to rib-shortening. Let us use S as the aver- age compressive stress on the rib. The shortening of the span is given by A. Si (21) E which must be equal to ( 22 ) Equating (18) to (19), substituting H^h-y) for M in the equation, and solving for H u we have r = _ SI a(hg - 1) Shi a(hg-t) (M) (N) Application to Parabolic, Segmental Circular, and Elliptical One-Hinged Arches The general formulas derived above have been ap- plied to three forms of arch-ribs with parabolic, seg- mental circular and elliptical curves respectively. The equations of these curves all have their origin at the left support a, with the X-axis passing through the supports a and b. The equations of these curves are as follows: (y -f-r-hf = ) • ( 24 ) An important assumption is made in applying the general formulas, that is I=lo sec e where 7 is the moment of inertia of the cross-section at any point and 7 o the angle of inclination of the axis of rib at the point where / is taken. This assumption simplifies the work materially and should be dose enough for deriving the formulas to be used in the preliminary designs. Also, ds=sec a dx, while the modulus of elasticity is assumed to be constant in the general formulas. Table 2 gives a comparison of the formulas for the reactions and end moments of the one-hinged arches with parabolic, elliptical and segmental circular arch ribs. These formulas have been derived after a tre- mendous amount of work. Formulas for tempera- ture effects and rib-shortening are also given in the table. The Reaction Locus Fig. 1 shows a one-hinged arch under a vertical load P. As the ends are fixed, it is unknown where the reaction lines must pass the supports. Let us assume that they pass the left and the right supports at points distant y 1 and y 2 below and above the points of support respectively. The right reaction line must pass through the center hinge. By similar triangles, we have the relation in which q is the ordi- kl H nate to the curve at any point But IJt = > therefore fj filVi -/-I'll ^ III ( 0 ) which is a general formula of the reaction locus of the one-hinged and the no-hinged arches applicable to any form of arch-rib. It is not applicable when the load is on the right half of the arch rib,. The reaction locus must be symmetrical about the center, so it is not necessary to d'erive the formulas for the load on the right of the span. For parabolic, elliptical and segmental circular arches we have by substitution the following formulas : h(/4 -/3k) 9 h(/+/3k) .(P) 3(/ik)~ for the load on the left and right of the crown hinge respectively. Equation Q* Equation Rf Formulas for Reactions of Three — ,Two — And No-Hinged Arches General formulas for reactions of three-, two-, and no-hinged arches have been derived in order to obtain a comparison between them. They have also been applied to these types of arches with parabolic, elliptical, and segmental circular ribs. Table 3 shows a comparison of the general formulas of the four types of arches, while those for a comparison of the reactions of parabolic, segmental circular and elliptical forms of arches are not here reproduced. The Envelop of Reaction Lines The position of live load for the maximum posi- tive and negative moments and the maximum posi- tive and negative shears of the one-hinged arch can be found graphically by means of the reaction locus and the envelop of the reaction lines. Since the ends of the one-hinged arch are fixed, we cannot tell where the reaction lines pass through the sup- port for a load at a certain point of the span. By means of the reaction envelop and the reaction locus, this can be easily done by drawing the tangent line to the reaction envelop from the point where the load line intersects the reaction locus. The equa- tion of the reaction envelop of the one-hinged arch was found by the author to be an equation of the third degree. To plot the curve from the equation involves a tremendous amount of labor. It is easier to draw the reaction envelop from the computed values of ordinates than to plot it from the equation. The envelop passes through the hinge and intersects the Y-axis at a certain distance below the origin. To determine the points of division of the live load for the maximum positive and negative moments, tangent lines are draAvn to the envelop from the point on the arch-rib we wish to investigate. The intersection of the tangent lines with the reaction locus gives the required points of division. To de- termine the points of division for the shear, a tangent line to the envelop is drawn perpendicular to the normal line of the section. The intersection of this tangent line with the reaction locus forms one division point, while the section itself is another division point. Study On Designs The design of a one-hinged arch can only be made with a series of approximations. Stresses due to temperature and rib-shortening play an important q=h^/+8(37r-3) (/—2ft) ft 3 "1 4(4 k e -3- ft -P3)-)k-ff-h3(/-2K)(37r -3a ) J in which Ct^cos t 1-h 4 Ik *(/-2./t) h[sr 3 &i7z 3 c*'j-3r s l(/-£fi)(stvi £o<'-/- 2 ci- 2 c(')- 8 g 3 ^3l^(j -3 ft +4ft 2 ^ in which Cf = r~h tx'— siii~ I (c±k.-/') (Q) (R) 8 part. These can not be exactly determined without knowing the area of the cross-section of the arch- rib. Yet the latter again depends upon the total stresses to be carried. The method of trial is the only way to secure the right section for the one- hinged arch-rib. Three comparative designs were made by the author in order to study their compara- tive merits: (1) The moment of inertia at any section of the arch-rib is assumed to vary as the secant of the angle of inclination of the section, with the effect of rib-shortening neglected; (2) Same assumption as in the first case, with the effect of rib-shortening included; and (3) the true moment of inertia of the section is used. A one-hinged arch of twenty panels, with a span of 258 feet and a rise of 26' was adopted for investi- gation. The dead load is assumed 59 kips and the live load 18.5 kips per panel load. The depth of the rib at the center is assumed 5 feet. An allowance of 75° F is made for the rise or fall of temperature. The Modulus of elasticity is taken as 26,000 kips and the unit-stress for the gross section of the flanges, 15 kips. The arch chosen for investigation has been greatly studied by former graduate students of Pro- fessor H. S. Jacoby when containing three, two or no hinges. The reason for using this arch by the author is, of course, to compare the design of the one-hinged arch with that of the other types. Study On Deflections The deflections of the one-hinged arch w T ere stud- ied in three ways: (1) Vertical and horizontal de- flections under the vertical loading ; (2) vertical and horizontal deflection under the horizontal load- ing; and (3) maximum and minimum deflections. The same arch used in the design was used for the study of deflections. The deflections of this arch having three, two and no hinges respectively were investigated by Dr. P. H. Chen, a graduate of Cor- nell University. It gives a good opportunity for critical comparisons. (See The Cornell Civil Engi- neer, Vol. 26, Page 229). RESULTS OF INVESTIGATION Discussion On Formulas So far as the formulas for the reactions of the one- hinged arch are. concerned, those for the parabolic ribs are the simplest in form, while those for the segmental circular form require a vast amount of labor for their derivation and application. Formu- las for the elliptical form are intermediate in these respects. Neglecting the effect of the axial thrust and using a ratio of the rise to the span for a segmental cir- cular arch equal to 1-8, the relative effect of tem- perature upon the three forms of the one-hinged arch is as follows : Seg. Parabolic Elliptical Circular Coefficient of Eet°Io /; 2 7.5 8.4 6.0 Ratio 125 140 100 The temperature effect in the parabolic form is 125% as great, and that in the elliptical form 140% as great as that for the segmental circular form. The effect upon the elliptical form is 112% as great as that upon the parabolic form. Thus, the tem- perature effect upon the elliptical form is the larg- est, while that upon the circular form is the small- est. It is to be noted that in the segmental form the ratio of the rise to the span is assumed as one- eightli. This is only a particular case. If we as- sumed another ratio, the results would be different. The relative effects of rib-shortening upon the horizontal reaction and the end moments of the three forms of one-hinged arches have the same re- lations as the temperature effects. There is an interesting fact about the number of equations required to analyze no-, one-, two-, and three-hinged arches. Three-hinged arches re- quire three conditions; two-hinged arches, four; one- hinged arches, five ; and no-hinged arches, six. Thus, we see that the more hinges an arch has, the less are the number of the conditions required to solve its unknowns. In addition to the three statical con- ditions, the three-hinged arch requires no-elastic condition for its solution ; the two-hinged arch re- quires one ; the one-hinged arch, two ; while the ne- hinged arch requires three. Therefore the three- hinged arch is said to be statically determinate. The formulas for reactions of three-hinged arches are exact, while those for no-, one-, and two- hinged arches are subject to many imperfections and assumptions. First of all, the elastic limit is not assumed to be passed. If very large loads should ever be applied which cause the stresses to exceed the elastic limit of the material in any mem- ber, the theory of elasticity fails and it is impossible to predict the degree of safety of the structure. The moment of inertia of any section of the rib is as- sumed to vary as the secant of the angle of the inclination of the arch-rib. The modulus of elastic- ity is assumed to be constant throughout the span. Shear is neglected in the derivation of the formulas. No- and one-hinged arches are similar in one re- spect; that is, they are both subject to vertical and horizontal reactions and have end moments under either vertical or horizontal loads. Two- and three- hinged arches are all subject to vertical and horizon- tal reactions, the vertical reactions being the same for those two types of arches under either vertical or horizontal loads. Temperature and rib-shortening cause horizontal reactions and the end moments in both the one and no-hinged arches, while in two- hinged arches they cause horizontal reactions but no end moments. It is generally supposed that a three-hinged arch is not subject to stresses due to a change in temperature. Strictly, however, such stresses will occur, for a fall in temperature causes a deflection of the crown hinge, and as the span does not change, the horizontal thrust will be in- 9 creased. The stresses produced, however, are very small. Neglecting the effect of axial thrust upon the rise and fall of temperature, the relative importance of temperature effects upon the horizontal reactions and end moments of no, one-, tw r o-, and three- hinged arches are as follows : Arch No- One- Two- Three- Hinged Hinged Hinged Hinged ELLIPTICAL 160/8 67.2/8 12/8 0 M x 125.6/8 67.2/8 0 0 PARABOLIC H x 90/8 60/8 15/8 M x 60/8 60/8 0 0 Thus, the temperature effect upon the no-hinged arch is found to be the greatest of all. Referring to H in the parabolic form, it is one and one-half times as large as for the one-hinged arch, and six times as large as for the two-hinged arch. In the elliptical type it is 2.38 times as large as that for the one-hinged arch, and 13.33 times as large as for the two-hinged arch. The temperature effect upon H of the one-hinged arch of the parabolic form is four times as large as for the two-hinged arch, while for the elliptical form it is 5.6 times as large. The tem- perature effect upon M of the no- and one-hinged arches is the same in the parabolic form. In the elliptical form, the former is 18.7 times as great as the latter. In the segmental circular form, the tem- perature effect upon H in the one-hinged arch is three times as great as that for the two-hinged arch. With respect to the effect of rib-shortening the re- lative ratios above stated apply to different types of arches equally well. So far as the formulas for the reactions are con- cerned, there is a decided advantage gained by those for the parabolic form over all other types of arches because of their simplicity. Discussion on Reaction Influence Lines In order to study the variation of the reactions and the end moments as a single load moves over the span, reaction influence lines are drawn and compared. (See diagrams 2, 3, 4, 8, 9, and 10). Only a few of the numerous diagrams are repro- duced in the abstract of the thesis. Under vertical loading, the elliptical form of the one-hinged arch has greater horizontal reactions and end moments than the parabolic and segemental circular forms. The last two forms cause less dif- ference in the H t and M x . The vertical reactions are all the same in the one-hinged arch for the three forms of ribs. There is not much difference shown under the horizontal loading. The circular form seems to have the greatest advantage for the one- 10 hinged arch judging from a comparison of the curves. However, the curves for the circular form are only drawn for a single case with a ratio of rise to span equal to 1-8. If other ratios were used, the results may be different. On the other hand, the parabolic form has the advantage of low reactions besides that of simplicity in formulas. Therefore, the author comes to the conclusion that the para- bolic curve is the best form for the neutral axis of the arch-rib to be used for a one-hinged arch. Under vertical loading the reaction influence line of H for a three-hinged arch consists of straight lines ; for a two-hinged arch, a curve resembling a parabola ; for a no- and a one-hinged arch, curves symmetrical at the center. Curves for one- and three-hinged arches break at the crown while those for no — and two-hinged arches do not. This is be- cause the hinge at the crown of the one- and three- hinged arches breaks the continuity of the rib. Curves of no-, two-, and three-hinged arches are very close to each other, while that for the one- hinged arch has a great variation, being lower at the supports and higher near the crown. This brings out one disadvantageous fact for the one-hinged arch. When the load is at the crown, the value of H is about two times as much as that for the other types. This means that the supports have to resist a greater horizontal thrust, and extra care has to be taken in making the end supports secure. The end moment of the one-hinged arch has a greater varia- tion than that of the no-hinged arch. of fk&cfar? Zoc/ S//s/?Acr&/ & fnx/Arr The one-hinged arch has an advantage over the no-hinged arch with regard to the effect of tempera- ture and thrust, but it is not so favourable with re- gard to the reactions and the end moments under vertical loads. For reactions due to temperature, thrust and vertical loading, the one-hinged arch is not as advantageous as the two — and three-hinged arches. So far as the simplicity of the formulas for fc‘srr / a?/-*sas? of~ Ahfefao Zoc/ Os?&, /mo < 5 * Ztsae /? styes' asx/er & Mr*/ /&&/ Ctsrya SA&e/ 2 / the reactions is concerned, the same statement holds true. However, attention will be called subsequent- ly to some decided advantages of the one-hinged arch. Discussion On Reaction Loci In order to compare the reaction loci of different types of arches, the equations of the reaction loci for no-, one-, two-, and three-hinged arches are plotted for different forms of arch ribs. They are shown in diagrams 20 and 21. All reaction loci for the one-hinged arch with dif- ferent forms of arch-ribs pass through the crown hinge. When the load is at the crown, the reaction lines must pass through the crown hinge, for other- wise there would be rotation at the crown. All re- action loci break their continuity at the crown but the two halves are symmetrical. The parabolic one- hinged arch has the highest ordinate when the load 11 is near the support, while the segmental circular form has the least ordinate. For a parabolic arch-rib, the reaction locus of the three-hinged arch is a straight line for each half of the span; that for the two-hinged arch, a curve like a parabola; that for a no-hinged arch, a horizontal straight line at an ordinate of 1,2 h; and that for the one-hinged arch, a curve breaking at the center. Curves for no- and two-hinged arches are contin- uous, while those for the other two break at the center. Curves for no- and two-hinged arches do not pass through the crown, while those for the one- and three-hinged arches must do so. At the support, the three-hinged arch has a maximum- ordinate of 2.00h; the two-hinged arch, 1.60 h; the one-hinged arch, 1.40 h; and the no-hinged arch,. 1.20 h. At the crown the curve for the two-hinged arch has the highest ordinate. Similar statements can be drawn for the reaction loci for the elliptical forms of arches. The ordi- nates are highest in certain cases and lower in others, while the general forms are about the same. Discussion On Reaction Envelop By means of the reaction locus and the reaction lines, the position of live loading for the maximum positive and negative moments and maximum posi- tive and negative shears for the one-, two-, three, and no-hinged arches can be found graphically. In the three- and two-hinged arches, the reaction lines can easily be drawn as soon as the position of the live load is given; because the hinges at each sup- port fix the direction of the reaction lines. In the one- and no-hinged arches, the reaction lines cannot be so easily drawn, if the position of the load is only given ; for the fixing of the supports makes the di- rection of the reaction lines uncertain. In order to avoid the difficulty, reaction envelops are required for both the one- and no-hinged arches. The reaction envelop of the no-hinged arch was found by A. V. Saph, a graduate of Cornell Uni- versity, to be two hyperbolas symmetrical and tang- ent to each other at the center at a point 2-3 h above the level of the supports. The reaction envelop of the one-hinged arch was found by the author to consist of two curves very similar to those for the no-hinged arch in form yet quite different in their equations. The curves are symmetrical and meet each other at the crown hinge. The reaction en- velop is a plane curve of the third degree ; because the formulas of the reactions of the one-hinged arch involves the k’s in the fourth power. In the case of the no-hinged arch, the k's in the formulas of the reactions are of third power; hence a second-degree curve is obtained for the envelop. The method of finding the positions of live loading is the same for the four types of arches, except that the reaction envelops of the no- and one-hinged arches have the function of the end hinges in the case of the three- and two-hinged arches. A com- parison of the reaction envelops of the no- and one- hinged arches is given in diagram 28. Discussion On Design With the Moment of Inertia Varying As the Angle of Inclination of the Arch-Rib There is no marked difference in the procedure of designing the arch-rib for the two-, one-, and no- hinged steel arches with the assumption that I varies as sec b . The procedure may be generalized in the following heads: (1) calculation of reactions and end moments from the formulas for a unit load at each panel point; (2) to find the position of live loading for maximum positive and negative mo- ments, by either algebraic or graphical methods; (3) calculation of dead load and live load moments for each section; (4) calculation of dead load and live load thrust; (5) calculation of moments and thrust due to temperature effect by assuming a cer- tain moment of inertia at the crown; (6) calculation of moments and thrust due to effect of rib-shorten- ing by means of the assumed moment of inertia ; (7) design of the flange area of the crown section using the maximum moments and thrusts so obtain- ed; (8) test the moment of inertia of the crown section, and see if the assumed value of the moment of inertia at the crown section agrees with the value obtained; (9) after the right value of I 0 is obtained the flange areas at other sections can be calculated by using the relation that / = I t sec Q ; (10) the position of live load for maximum positive and nega- tive shear is obtained by either graphical or alge- braic method, the maximum shear at each section is secured by combining the dead load, live load, tem- perature and rib-shortening shears, and the webs are designed accordingly. The design of the three- hinged arch may be carried out in the similar or- der with the exception that no trial is required in securing the sections. There is an interesting fact found in combining the moments and thrusts caused by the dead load, live load, temperature and rib-shortening to produce the maximum stress on the sections of the one- hinged arch. The live load may be considered in three ways; (1) loading for maximum positive mo- ments; (2) loading for the maximum negative mo- ments; (3) loading for maximum thrust. In the design with the effect of rib-shortening neglected, case (1) controls for the section 0-4 and case (3) controls for the section 5-10. In the design with the effect of the rib-shortening included, case (2) controls for the section 0-4, while case (3) still con- trols for the sections 5-10. The reason is quite clear; for in the sections near the center, the mo- ments are comparatively small, while the thrusts are large. This means that a full live load is re- quired to design the sections near the center. As there exist large bending moments in the sections near the support, naturally the moments control 12 the design for those sections. The effect of rib- shortening is to produce the negative moments and thrusts. This is why the negative moments control the end sections when the effect of the rib-shorten- ing is included. The design of the one-hinged arch is different from that of the two-, and no-hinged arch in two respects. The design of the center sec- tions of the one-hinged arch is controlled by the thrust while that of the two- and no-hinged arches is controlled by the moments. The reason is be- cause the no-hinged and two-hinged arches have large moments at the middle sections, while the one-hinged arch has large thrusts. The signs of the moments and thrusts produced by the effect of temperature and rib-shortening are the same for all sections of the one-hinged arch, while those for the no-hinged arch are the same for the sections below 2-3 h of the rib, and opposite each other for the sections above 2-3 h of the rib. The signs of the moments and thrusts produced by the effect of tem- perature and rib-shortening are all opposite for the two-hinged arches. In designing the sections at the crown, only thrust is used for the one-hinged arch. This makes the design of the one-liinged arch easier, because the right value of the moment of inertia can be secured immediately. Positive shear governs the design of the web for all types of arches, whether the effect of rib-short- ening is included or not. The latter which tends to produce negative thrust, tends to increase the posi- tive shear in all sections for the one-hinged arch. Under maximum loading the comparative effects of dead load, live load, temperature and rib-short- ening on the flanges are shown in the following table and also on curve sheet 22 : Dead Live Tempera- Rib- Section load load ture shortening Per Cent. Per Cent. Per Cent. Per Cent. 0 31.5 29.7 24.2 14.6 1 38.8 24.2 23.0 14.0 2 47.4 17.7 21.7 13.2 3 58.2 10.6 19.3 11.9 4 72.7 1.3 16.1 9.9 5 75.8 23.7 20.6 —20.1 ' 6 75.9 23.8 15.5 —15.1 7 76.0 23.8 11.4 —11.1 8 76.0 23.9 8.5 — 8.4 9 76.0 23.8 6.6 — 6.5 10 75.8 23.8 6.3 — 5.9 It is seen from the curve that the dead load has the greatest effect of all, and its effect on the sec- tions near the center is greater than those near the ends. In the first place, this is because the dead panel load is comparatively great, and naturally it takes greater stress. In the second place, the effect of temperature and rib-shortening near the center sections is greater than their effect on the sections near the ends. This makes the percentage of the stress carried by the dead load gradually decrease in the sections near the ends. The live load is the next important factor in causing flange stresses. The effect of temperature and rib-shortening can never be omitted. They have about the same effect. Temperature has the greater effect in the sections near the ends than in those near the center. The same is true for the rib-shortening. The negative signs of the rib-shortening in the sections near the. ends are explained by the fact that both negative thrusts and moments are used. Under maximum loading the comparative effects of moments and thrusts on the flanges is shown in the following table and also shown on curve sheet > : — Section Moment Thrust Per Cent. Per Cent. 0 73.1 26.9 1 67.4 32.6 2 60.3 39.7 3 51.8 to (Sections) X '/-to .%/oo $ P(Sections) Comparative Design of F/ange Ft re as Comparative Flange Freas of O-tt, /-H, 2-H, & 3-H Fashes Curves Nos. 22-26 Dl -Dead Load, LL = Live Load, T= Temperature PS = Rib -Shortening RN= Rib-shortening neglected , R I = Rib-shortening included. fvO/npa. Envelops •relive /Reaction of No- end One-h/’nged S/ee/ ftrches O-M /St SC) ’* dtrzf-tdhtx *ht* - O /-A. soco-ufA * r- ^-A $7 * robra-o ScbB^rOCbi *S-BJ J r Sc — O A— - # - fen it* pj * a-£ S-f S- aFc r / -dbcrdb’^F c-saAb -7 @ Reaction locus of o-A ® Reeclon locus of /-A © Reeclon Envelop oF o-A @ Reaction Envelop oF t-A 20/agram 20 case the moments and thrust produced by the rib- shortening always have the same negative sign in the case of a one-hinged arch. This naturally tends to decrease the area of section required, if the area of section is controlled by the positive moments and thrust, in the design, as is usually the case in the one-hinged arch. The comparative flange areas required for the three-, two-, one-, and no-hinged arches for the same form of the arch-rib under the same specifications are shown in the following table and on curve sheet 25. The areas given for the three-hinged arch are exact, while those for the two-, one-, and no-hinged arches are obtained from a design with the as- sumption / 5 seed with the effect of r ib-shortening included. Three- Two- One- No- Section. IIinged. Hinged. Hinged. Hinged. sq. in. sq. in. sq. in. sq. in. 0 68.0 68.4 153.7 119.5 1 73.0 72.8 127.0 93.1 2 78.2 78.1 103.6 75.9 3 82.4 85.0 84.0 72.3 4 84.4 90.7 67.7 65.0 5 84.1 93.1 67.7 67.9 6 81.2 94.9 66.8 72.8 7 77.7 94.8 66.0 76.9 8 71.0 92.3 65.4 78.4 9 65.0 89.3 64.9 78.3 10 64.0 89.0 64.7 77.3 tions are the largest of all. The two-hinged arch requires the largest area in the sections near the middle for the same reason. The one-hinged and no-hinged arches require the largest areas in the sections near the ends, because the end moment is the dominating factor in those sections. In com- paring the areas in flanges of the one- and no-hinged arches, we see that the one-hinged arch requires comparatively large areas in ihe sections near the ends and small areas in the sections near the crown. The areas near the crown of the one-hinged arch are rather uniform, because the thrusts at these points are large and about equal in magnitude. The comparative theoretical web areas required for the three-, two-, one- and no-hinged arches with the same form of the rib and designed under the same specifications are shown in the following table and also in curve sheet 26 : — Three- Two- One- No- Section. Hinged. Hinged. Hinged. Hinged 1 sq. in. 7.91 sq. in. 10.04 sq. in. 19.59 sq. in. 22.83 2 6.95 8.98 16.76 19.88 3 6.31 8.20 14.95 18.22 4 5.77 7.67 13.03 15.51 5 5.45 7.45 11.10 13.58 6 6.10 7.61 9.55 12.08 7 6.95 7.92 7.95 10.87 8 7.61 8.11 6.89 9.70 9 7.97 8.12 6.45 9.18 10 8.13 7.96 5.75 10.46 14 It is seen that the no-hinged arch requires the largest web area, while the three-hinged arch, the least of all. The one-hinged arch requires smaller web areas than the no-hinged arch, while the two- liinged arch requires greater areas than the three- hinged arch. The no-hinged and one-hinged arches require larger web areas in the sections near the ends and smaller areas in the sections near the crown. This is due to the fact that the shear in the no-hinged and one-hinged arches is far greater near the ends than in the sections near the crown. On the other hand, the web areas of the three-, and two-hinged arches are rather uniform, since the shears in the sections of the three- and two-hinged arches are rather uniform. The marked difference is caused by the fixing of the supports in the one case, and the hinging of the supports in the other. The following table shows the results of the de- sign based upon the assumption / 5 sec 0 Section. Composition Moment of Inertia. Ratio of I. See 0 . Proposed Assumption, of I. 0-2 6z.s 6" X 6” X 9/16" 3 Pis 14" X 9/16" 5 Pis 18" X 7/8" 1 Pis 18" X 3/4" 286,000 2.49 1.07 250% 2-4 6z.s 6" X 6" X 9/16" 3 Pis 14" X 9/16" 2 Pis 18" X 3/4" 1 Pis 18" X 13/16" 190,600 1.59 1.05 160% 4-6 6z.s 6" X 6" X 9/16" 3 Pis 14" X 9/16" 3 Pis 18" X 5/16" 123,800 1.03 1.03 103% 6-8 6z.s 6" X 6” X 9/16" 3 Pis 14" X 9/16" 1 Pis 18" X 1/4" 121,700 1.02 1.01 101% 8-10 6z_s 6" X6" X 9/16" 1 Pis 14" X 9/16" 1 Pis 18" X 3/16" 119,700 1.00 1.00 100% It is seen that the assumption I ' sec Q is about right for the sections near the crown, and is in great error for the sections near the ends. DISCUSSION ON TRUE DESIGN So far as the reactions are concerned, the effect of final design as compared with that of the pre- liminary design, assuming I z sec. e, is to increase the horizontal reactions when the load is near the center of the span and to decrease the same when the load is near the ends. There is no appreciable change in the vertical reactions for the two de- signs. The end moments have been affected so greatly that under the full loading, a negative end moment is produced in this design, which was not the case in the preliminary design. The latter has a serious effect in designing the sections, because the sign of the dead load moment is changed to negative, and the maximum moments at various sections are thereby changed. The preliminary design by assuming I z sec. o is far from being correct. It makes the flange areas at the various sections far different from the true values. The error is on the unsafe side. It is necessary that a revised design be made in de- signing a one-hinged arch. The preliminary design by assuming / ; sec. e is nearest to the true value for the two-hinged arch, while it differs most in the one-hinged arch. The relative error of the preliminary design from the true design of the no-, one-, and two-hinged arches with same dimensions and designed to carry the same load is shown below : Section No-hinged One-hinged Two-liinged per cent per cent per cent 0 —15.5 —30.7 5.2 1 —13.4 —31.6 —3.7 2 — 7.6 —33.1 —4.7 3 — 2.4 —34.4 —2.5 4 — 3.6 —40.5 2.4 5 8.7 —33.3 3.3 6 5.6 —27.0 7.6 7 4.1 —21.0 12.8 8 3.7 —16.1 12.1 9 2.5 —12.8 11.6 10 2.1 —11.7 13.7 Thus, we see that the error in the one-hinged arch is the largest of all. This is due to the presence of the large moments in the sections near the ends and the large thrust in the sections near the crown. The variation of the moments and thrusts in the sections is such that the moment of inertia in the various sec- tions required is not in accordance with the relation jz sece. As the assumption / s sec . o is far from being true, the I-Curve for a one-hinged arch with 1/10 rise is recommended by the author, by means of which a closer result can be secured in the approximate de- sign. A similar curve was recommended for the no- hinged arch by P. H. Chen in his thesis above named. The comparative values are shown below : Section One-hinged N o-h inged 0 Io + 185% Io 1 ~r 120% 1 Io + 135% Io 60% 2 Io + 100% Io _L 30% 3 Io + 70% Io 1 ~r 15% 4 Io + 45% Io 6 Io + 15% Io + 10% 8 Io + 5% Io — 10% 10 In I 15 {OQOOO V \ \ — i — i • - y — X Ch/ vrres :Ju ■ 0/7e */\ /9 re/? . .. . ( . 7/4 % 5(5 \ k' ' \ \ \ -- j c7 1 ~ I * • Fr&d/c - h . : | ' J ©> -<> \ 1 Z + 7o?. s-Cvrre 1 ' 1 f . ■i- L ! : i i..| U_, Zj. sec_ 1 1 . 1 1 '-t|— . j. Cos7sfe. -t- ... _L - *. _ J ■! - • i - ; o r . 5 . . V " ^ \ t : -i ’ fri-lrlrtirhiS - ; 1 z. i a f It is seen from the curves that the assumptions of I s sec.® and 1= constant are nearly correct in the sections near the center, while they are too small in the sections near the ends. The exceedingly large moments near the ends both in the no-hinged and one-hinged arches necessitate the use of the large sections and accordingly the values of the large moments of inertia. • DISCUSSION OF METHODS OF DEFLECTION COMPUTATIONS Deflections in the one-hinged arch-ribs are con- tributed by two factors, that due to thrust and that due to moment. The methods used by the author in finding these deflections are rather interesting and are here given. (a) DEFLECTION DUE TO MOMENTS— The formulas for finding the horizontal and vertical de- flections of a curved beam were given in equations (d) and (e) respectively. As the continuity of the one-hinged arch is broken at the crown by the presence of the hinge, the formulas could not be ap- plied directly to the entire span. A method of separating the arch into two cantilevers is intro- duced (Fig. 3). For example, when the load is at a certain point on the left half of the span, the re- actions acting on the left cantilever are H x and V 2 , while those on the right cantilever are H x and V 2 . lever are contributed by three factors, due to P, H x and V 2 , while those in the right cantilever are con- tributed by two factors, H x and V 2 . These factors can be easily found by considering each half in turn as a cantilever with one load on the arch. Thus, under the vertical load P on the left cantilever, we only need to find the deflections in the cantilever due to a horizontal load at the free end, a vertical load at the free end, and a vertical applied load. The deflections due to H x and V 2 in the left half of the arch are just the same in magnitude as those due to H x and V 2 in the right half of the arch. The final value of the vertical or horizontal deflections can be easily obtained by taking the algebraic sum of the vertical or horizontal deflections contributed by the various factors. If the deflection diagrams are required for a unit load at different positions on the left half of the span, the method is exceedingly simple. We need only to draw the deflection diagrams of the canti- lever with a unit horizontal load at the free end, and a unit vertical load at different positions on the cantilever corresponding to the positions on the arch. These are obtained as unit deflections. The values of H x and V 2 are then found for different positions of the unit applied load. The deflections due to II \ and V 2 are then found by multiplying the unit deflections by the values of H x and V 2 . Using proper signs of deflections contributed by each factor and taking the algebraic sum, the deflections at various points on the arch can be found for dif- ferent positions of the loading. The method has several advantages : first, it is simple as well as easy; second, it offers the oppor- tunity of studying the deflections contributed by each factor; third, it is easy to discover mistakes. By means of the graphical method, the deflection on the cantilever due to a unit vertical or horizontal load at any position on the beam can be easily cal- culated. By arranging the work in a systematic way, errors can be easily discovered by comparison. The method is applicable in any of the four cases : (1) vertical deflection under the vertical load, (2) horizontal deflection under the vertical load, (3) vertical deflection under the horizontal load, (4) horizontal deflection under the horizontal load. The only difference lies in the magnitude and direction of II x and V 2 and the unit horizontal or vertical de- flections to be found in the cantilevers. (b) DEFLECTION DUE TO THRUST— The general formula for the deflection due to thrust is given by Deflection = f*Tt in which T is the thrust due to the applied load P; while small t is the thrust due to a load unity P" applied at the point whose deflection is sought, the direction of P" being the same as the direction of of the deflection required. Let H' and V' be the hori- zontal and vertical shear (not in the normal section) immediately on the left of the section considered; and H" and V", those due to P" respectively. Then, by substitution, the general formula of deflection due to thrust becomes, Deflect /on =J' b H’H"cos 2 e^ +J^‘VV r "sjn *6^ which is applicable to the four cases above named, that is, the vertical and horizontal deflections due to the horizontal load. 16 p S 3 In order to simplify the numerical work( the above formula may be transformed into a simpler form for each of the different cases. For example, let us take the ease of the horizontal deflection under the vertical loading'. Let a unit load P' be applied at t and the horizontal deflection at 5 be found. (Figs. 59 and 60). The corresponding points on the other half of the arch are called s' and t' In examining the loading and reactions closely, we find that H' and H" in s-c and s'-c are correspondingly equal. H' in a-s is also equal to H' in b-s' under the load P ' ; while under the load P" , IP in a-s is equal to H 1 and H’ in b-s' is equal to H.,. Also V" in t-c and t'-c are correspondingly equal. V" in a-t is equal to V" in t'-b, while V" in a-t=V 1 and V” in b-t is equal to V 2 . Using the proper signs, the formula is given by H'Hcos e Oj{jz —J' (Hr )H coo P Q Ylt +f(v i ’V;') v s