You are on page 1of 125
FORCES AND FORMS IN LARGE STRUCTURES Part I - Equilibrium in Structures Alfonso M. Albano Bryn Mawr College Part I] - New Forms in Long-Span Concrete Bridges William Case Grinnell College Part III - Los Angeles Freeway Bridges Newton H. Copp The Claremont Colleges MONOGRAPH SERIES OF THE NEW LIBERAL ARTS PROGRAM The New Liberal Arts (NLA) Program of the Alfred P. Sloan Foundation has the goal of assisting in the introduction of quantitative reasoning and concepts of modern technology within liberal education. The Program is based on the conviction that college graduates should have been introduced to both areas if they are to live in the social mainstream and participate in the resolution of policy issues. The New Liberal Arts (NLA) Program of the Alfred P. Sloan Foundation has led to significant, new courses and course changes in many colleges and universities. MIT Press and McGraw-Hill are jointly publishing an NLA series of books. These monographs are planned to provide teaching/learning materials for other educational developments. John G. Truxal, Feedback/Automation. Morton A. Tavel, Information Theory John G, Truxal, Probability Examples Newton Copp, Vaccines: An Introduction to Risk Joseph D. Bronzino and Ralph A. Morelli, Expert Systems: Basic Concepts Victor A. Stanionis and Hugh Berberich, Computer Music: Science and Technology of a New Art 7. David P. Billington and Alfonso M. Albano, Episodes in American Invention: The Steamboat and the Telegraph 8. Newton H. Copp and Andrew W. Zanella, The Electrification of Los Angeles: Engincering, Science, and History 9. Marian Visich, Jr., Bar Codes and Their Applications 10. Jonathan Hallstrom and George B. Todd, HyperCoustics: An Introduction to Sound and its Digital Representations (in disk form only) 11. Vincent H. Smith, The Economics of Technology 12. Warren Rosenberg, End Stage Renal Disease 13. David E. Henderson, Air Pollution and Risk Analysis 14, Alfonso M. Albano, William Case and Newton H. Copp, Forces and Forms in Large Structures era alae ca Copyright © 1990 Research Foundation of State University of New York, Stony Brook, NY 11794. All rights reserved. NLA Monograph Series Co-Directors: John G. Truxal and Marian Visich, Jr. Production Editor: Carol Galdi Design of this monograph J. Wayman Williams Associates, Basking Ridge, NI 07920 Printed by The Wexford Press, Copiague, NY 11726 (516) 842-0044. Table of Contents OVCWOID i eae iii Part I Equilibrium in Structures ............ceeeeeeeeeee eens 1 Part II New Forms in Long- Span Concrete Bridges ........... 37 Part III Los Angeles Freeway Bridges ... Foreword Forces and Forms in Large Structures is one of a series of monographs arising out of the Princeton Summer Seminars on "The Engineer's Experience and the New Liberal Arts" con- ducted in 1987, 1988, and 1989 by David P. Billington, Michael Mahoney, Robert Mark, and John M. Mulvey under theauspices of the Alfred P. Sloan Foundation. The seminars brought together faculty from liberal arts colleges for intensive work on the development of curricular materials in engineering to be integrated into courses at their institutions. Although the ma- terials reflected in the first instance the particular needs and opportunities of these courses, the seminar pursued several overarching themes in common: + Engineering (taken here to be the same as technology) has its own set of guiding ideas, different from those of the natural sciences. Although liberal arts college students can comprehend those ideas without prerequisite college-level study of science or mathematics, they should nonetheless carry out simplified numerical studies which convey the quantita- tive nature of modern technology. Moreover, because engi- neering aims ultimately at artifacts — at structures, machines, Processes, and systems — rather than words, students must see these objects of their study as well as hear and read about them. Hence, the seminars emphasized the development of visual materials in the form of slides, graphics, and models. + Engineering is an evolving social process conditioned by and in turn acting on politics, economics, and cultural traditions. Analyzing this process can best be approached through studies in the history of technology. Foreword + Engineering is a human activity, which often reflects the thoughts and feelings of individual engineers, whose lives and works deserve study. When set in the context of these themes, several of the resulting projects seemed to promise interest and effectiveness among students and faculty at liberal arts colleges in general, and with further support from the Sloan Foundation the au- thors devoted additional time in 1989-1990 to putting them in final form for distribution in the New Liberal Arts Monograph Series. As the product of successful classroom experience each monograph is designed to be used directly by students. In addition, the authors have tried to provide guidance for faculty who might wish to adapt only portions of it into their own courses or simply inform themselves about the subject. Arnold Pacey warned in The Maze of Ingenuity that “every technical innovation has social consequences — almost all technology is concerned with bringing about changes in soci- ety — and to train engineers in twentieth century universities as if engineering is simply a matter of pipes, electronics, con- crete and mathematical calculation seems little short of irre- sponsible." In the spirit of the New Liberal Arts Program, this monograph and its companions eloquently demonstrate the converse of Pacey's proposition, namely that to train liberal arts students as if pipes, electronics, concrete, and mathemati- cal calculation had nothing to do with changes in society is equally irresponsible. David P. Billington Michael S. Mahoney Princeton, New Jersey Preface The three teaching units comprising this monograph focus on bridge engineering. The three authors have each used this unit successfully at least several times in courses at his institu- tion. Because each has proven to be valuable in itself, we have kept each as a separate unit with its own integrity. But they are nevertheless closely related and in some contexts teachers will find it useful to use two or all three in one course. Therefore, we havethought it useful to put these three together within one cover. Alfonso Albano’s unit, used in an introductory physics laboratory at Bryn Mawr College, is the most basic and serves as a clear initial guide to the physical principles which are the primary elements of structural engineering. William Case’s unit, used as part of his full-term course on structures at Grinnell College, presents in simplified form the ideas that generated two of the most impressive prestressed concrete bridges of the past twenty years. Newton Copp’s unit, serving as part of a course on science for the non-scientist at the Claremont Col- leges, considers the more common case of highway overpasses especially in the context of the Los Angeles Freeway system; this applies well to the conditions around all major cities in the United States. The authors have designed these units for use in courses aimed at Liberal Arts students; but engineering students would benefit from them as well especially as introductory units in courses which proceed to develop sophisticated modern meth- ods of analysis for complex structures. Each author has used the Structural Studies developed at Princeton by Robert Mark and me. These units, extending and enriching those studies, are nevertheless independent and de- iii Foreword signed to be used on their own. However, a reference copy of Structural Studies is useful for the instructor and as a reference for students. These units, as well as the Structural Studies, introduce students to the simplified formulas essential to understanding the dramatic forms possible with the new material of pre- stressed concrete. The formulas require nothing beyond high school mathematics and allow students to make the connection between physical behavior of structures and their visual forms. For the problems described here computer programs are widely available and experienced instructors may wish to use some in connection with this monograph. It is our belief, how- ever, that simplified calculations which stress the principles of mechanics provide an essential introduction to engineering ideas both for liberal arts students and for those in engineering. A companion teaching manual, available from Princeton, uses personal computer programs to allow students to work with a wide variety of structural problems found in Structural Studies. We thank John Truxal and Marian Visich for their many suggestions which have improved the monograph. We are especially indebted to J. Wayman Williams for organizing these units intoa coherent whole and for numerous suggestions. on clarifying the content. By testing and revising these three units, each of the authors makes available here teaching materials that can enliven science and engineering courses in colleges and universities through- out the nation. David Billington Princeton, New Jersey Part I EQUILIBRIUM IN STRUCTURES Alfonso M. Albano Bryn Mawr College December 1990 Part I Equilibrium in Structures Table of Contents Othe Wistructor ee 5 Chapter1 MntoduchOn 7 Chapter 2 Some Preliminaries: Equilibrium Revisited ............./ 9, Chapter3 Rotational Equilibrium ........... 6.66. c cece eee ee eee 12 Chapter 4 Bending Moments and the Eiffel Tower ................ 17 Chapter 5 Cables and the George Washington Bridge .............. 24 Chapter 6 Putting It All Together: The Salginatobel Bridge ......... 30 Retetences 35 Cpl eee 36 Instructor To the Instructor The Course This module evolved from material developed for a three- week project in an introductory physics laboratory. It was motivated by the hope of introducing students to some aspects of structural engineering without taking time from the already overcrowded schedule of the lecture part of the course. It was inspired by, and draws heavily from, David Billington’s course on structures. In fact, it is designed to lead to the analysis of a number of structuresin Billington and Mark’s Structural Studies (bibliographical details are given at the end of the module). The material presented here is more than can be done in three three-hour laboratory sessions. Depending on the pace of the students and the inclinations of the instructor, various segments can be omitted without losing the flavor of the project. I have found, for instance, that Experiment #1 can be greatly abbreviated, and Experiment #4 omitted altogether. On the other hand, Hooke’s “Rule of the Arch” which prescribed using a loaded cable to determine the shape of the arch that would bear a similar load can be exploited to develop some interesting exercises which are not fully explored here. It could alsobe used as a good excuse to study in greater detail the carreer of Robert Hooke — most certainly one of the more colorful and more unappreciated characters in the history of physics and engi- neering. Ihave foundit useful to introduce this project with a lecture on Structural Art essentially summarizing the earlier chapters of Billington’s The Tower and the Bridge and illustrated with some of the slides from his course on structures. Showing the EQUILIBRIUM IN STRUCTURES Albano now classic film of the collapse of the Tacoma Narrows Bridge is, of course, de rigeur, if only to remind the students of what can happen if one does not do things right. In the preparation of this material, I have profited greatly from the advice and help of David Billington, Newton Copp, William Case, Toby Richards, Faruq Siddiqui and J. Wayman Williams. To them all, many thanks. Introduction CHAPTER 1 Introduction Understanding Technical Considerations This is a brief introduction to some of the simpler techniques used by structural engineers in the analysis and design of structures and structural components. Our aim is not to learn design. It is to understand some of the technical considerations that enter into these designs in the hope that this understanding will help us better to appreciate or to criticize structures. We will use physics as our starting point, but we will soon see that engineering is not just applied science. The two fields have different aims, are subject to different constraints, and their end products are judged by different criteria. Scientists seek to understand nature in terms of uni- versally applicable “laws”, if possible. Engineers build. The great aeronautical engineer, Theodor von Karman (1881-1963) is supposed to have said, “scientists try to understand whatis, engineers build what has never been”. Engineers use whatever scientific principles are applicable, making such simplifica- tions or approximations as they believe to be appropriate or necessary. Ifit becomes too cumbersome or too time - consum- ing to apply first principles to a specific design problem, they are more likely to rely on empirical observations using models of the planned structure, or a similar structure, used as a full- scale model. Financial and Aesthetic Considerations Technical considerations are often not the most strin- gent constraints on engineers. Equally important are financial and aesthetic considerations. Good engineering must be tech- 7 EQUILIBRIUM IN STRUCTURES Albano nically sound, inexpensive (or, at least, affordable), and aes- thetically pleasing (or, at least, not ugly). You can, no doubt, enumerate examples of engineering design and construction that violate one or more of these criteria just as you can name examples of bad science, or bad art. These counter-examples must be taken as reminders of the necessity of clearly enunci- ating the criteria and adhering to them. Sources for Specific Structures You will find an exposition and amplification of these ideas in David P. Billington’s The Tower and the Bridge: The New Art of Structural Engineering (Basic Books: New York, 1983; {paperback] Princeton University Press: Princeton, 1985), which also discusses the careers and the structures built by the mas- ters of this art. We will apply these techniques to analyze specific structures: the Eiffel Tower, the George Washington Bridge, and Robert Maillart’s Salginatobel Bridge. You will find technical analyses of these structures in D. P. Billington and R. Mark, Structures and the Urban Environment: Structural Studies (Princeton University, Department of Civil Engineer- ing, Princeton: 1984). Equilibrium Revisited CHAPTER 2 Some Preliminaries: Equilibrium Revisited Translational Equilibrium Every element of every structure is subject to a variety of forces. The shaded portion of the column shownin Figure 1a, for instance, is subject to the force, F,, exerted by the upper part of the column; toits weight, q, also directed downward, and to the upward force, F, , with which the lower part of the column supports it. Here q is the weight per unit length of the column. and we assume that we are dealing with unit length. If, forany reason, the lower part of the column is unable to exert an upward force equal to the sum of the downward forces, the column collapses. Thatis, the magnitudes of the forces must be related by the equation, F, = F, + q. If this condition is fulfilled, the shaded segment is said to be in translational equilibrium . (b) FIG, 1— Forces in acolumn EQUILIBRIUM IN STRUCTURES. Albano The relationships among the forces acting on the shaded segment of the column can be seen more clearly by imagining that we can separate the segment from the rest of the column as in FIG. 1b. We can then display just the forces acting on it. FIG. Ibis called a free body diagram, and uses the approximation that the weight of the segment may be represented by a single force acting at its center of mass (or centroid in engineering terminology) and that the other forces acting on it are exerted at the surfaces. Because the forces here tend to compress the element, it is said to be under compression. Internal Stresses It isusually notjust the actual force acting on an object that matters, but the object’s size as well. To take both factors into account, the notion of stress is used; stress being defined as force exerted per unit area. The shaded segment in FIG. 1 is subject to a compressive stress, fF. = F,/A. The shaded segment of the cable in FIG. 2a is subject to forces tending to pull it apart. For this reason, it is said to have (b) FIG. 2— Tensile stress 10 Equilibrium Revisited a tensile stress, f, = F,/ A, where A is the cross-sectional area of the cable but F, is now the upward force exerted on the shaded element by the part of the cable aboveit, and F, is the downward force exerted by the part of the cable below it. FIG. 2b shows the free body diagram. Sincestressis force per unit area, ithas units of pounds per square inch (psi) for (American) engineers and newtons per square meter or pascals for physicists. Compressive stress is traditionally considered negative, tensile stress positive. Thestresses obtained aboveare called internal stresses. In any design, the materials and the dimensions of the structural elements must be chosen so as to be able to withstand these stresses safely as well as efficiently. The strength of a material is defined as the stress needed to break a piece of the material. Itis a familiar fact that materials can withstand some kinds of stresses better than others. Good concrete, for instance, has a compressive strength (i.e., the compressive strength that will break it) of about 6,000 psi, but a tensile strength of only about 600 psi. Designers therefore try to make sure that concrete that is not reinforced or prestressed is subject only to compressive stresses since it cannot support much tension. Wood, on the other hand, may have a tensile strength of as much as 1,500 psi and a compressive strength of 4,000 psi, while commercial mild steel has both a tensile and compressive strength of about 36,000 psi. (These numbers are mainly from J. E. Gordon’s delightful book, Structures, or Why Things Don’t Fall Down ). 11 EQUILIBRIUM IN STRUCTURES Albano CHAPTER 3 Rotational Equilibrium The Cantilever FIG. 3a shows a cantilever, a beam that is held fast only at oneend and is kept horizontal by forces exerted only at that end. Consider the shaded segment, one end of which is the free end of the beam, while the other is at a distance, x, from it. The segment’s centroid (center of mass) is at a distance, x/ 2, from the free end, and has weight which we denote by qx . In order for this segment to bein equilibrium, an upward force, V =-qx, must be exerted on it by the rest of the beam. This force, called the shear force, can only be exerted at the boundary, QQ’. The conventions for assigning signs to shear forces can be quite complicated, but we need not bother with them here. For our FIG. 3 — Forces in a cantilever 12 Rotational Equilibrium purposes, since we represent the (downward) weight of the segment by the quantity, qx, then the upward force needed to balance it must be equal to —qx. When a force such as qx acts on an extended object such as the shaded portion of FIG. 3, there results a tendency for the object to rotate. This tendency is measured by what is called torque by physicists and moment by engineers. In the case il- lustrated in FIG. 3, the moment of the force, qx, about the point Bis simply the product of the magnitude of qx and the perpen- dicular distance from B to the line along which qx acts. This distance is called the moment arm of qx about B. The moment of qx about B is thus M = qx (x / 2), and this moment tends tomake the shaded segment rotate clockwise about the point B. For rotational equilibrium, a counterclockwise moment of magnitude M, also about B must be exerted on the segment by the rest of the beam. Again, this moment can only be exerted compression ce stress compressive stress } FIG. 4 — Tensile and compressive stresses 13 EQUILIBRIUM IN STRUCTURES Albano by forces acting at QQ’ (see the free body diagram, FIG. 3b). The Bending Moment A note on how the beam exerts a moment: The weight of the segment causes the beam to bend slightly as in FIG. 4a, causing the upper part to be put in tension and the lower part to be put in compression. The resulting compressive and tensile stresses, which are shown schematically in FIG. 4b, then provide the needed moment. This moment, which is a reaction that arises from the bending of the beam, is called the bending moment. Experiments in Rotational Equilibrium The experiments will be done on a 2’x3' wooden drawing board mounted vertically from the edge of a laboratory table with vise clamps (See FIG. 5). The board is provided with means for attaching a peg and a pulley for use in Experiment 1, as well as hooks for attaching chains or spring balances for the other experiments. FIG. 5 — Setup for experiment #1 14 Rotational Equilibrium Experiment #1: A “moment meter” This experiment is meant as a warm-up to familiarize you with the equipment. It should not take you too much more than half an hour. FIG. 5 shows a simple set-up which may be used to measure moments (torques). Thebar is mounted so thatitis free to rotate about a pivot through its center of mass (centroid). Choosing a pivot through the centroid makes it possible to neglect the mass of the bar in the rest of the experiment. The string connecting the bar to the spring balance may shift the location of the centroid slightly. This can be easily remedied by placing a paper clipora piece of string atan appropriatelocation on the other side of the pivot. spring balance pulley FIG. 6 — The moment meter The bar is subdivided into painted and unpainted seg- ments of equal length, |, with holes bored at intervals of 1/2. For our purposes, the precise unit of length is not crucial, and we mightas well use as our unit of length. In FIG. 5, as in the actual ‘bending moment meter’, L=51. 15 EQUILIBRIUM IN STRUCTURES Albano Weights such as P, and P, in FIG. 6 may be attached to the bar to simulate forces acting on a structural element. The moments of these forces may be obtained by measuring the force exerted by the spring balance to keep the bar horizontal, and the moment arm, L, of this force. 1. Suspend weights from a variety of locations on the bar. Adjust the string that attaches the spring balance to the hook (see FIG. 5) as necessary to make the bar horizontal. Record the weights hung, their distances from the pivot (in units of 1) and the reading of the spring balance. 2. Draw a calibration curve by plotting the moment calculated from the spring balance reading and its moment arm against the corresponding moment due to the weight(s) whose moments you were measuring. For this calibration, it is not necessary to express your moments in standard units. You may, for instance, use (weight hung in pounds) x (moment arm in I’s) or (weight hung in grams) x (moment arm in I's) to get something in units of “pound I's” or "gram I's”. You can always convert to proper units later. 3. How reliable is your “moment meter”? How seriously does friction affectit? Howmay youminimize frictional effects? What other factors (if any) affect its reliability? 16 Bending Moments and the Eiffel Tower CHAPTER 4 Bending Moments and the Eiffel Tower Cantilever Bending Moments Wenoted earlier that one important design consideration is that the materials, shape, and dimensions of structural ele- ments must be such that they can safely and efficiently support the internal stresses required by the loads that they bear. In the case of the cantilever (FIG. 3), this means thatata distancex from the free end, the beam must carry the shear force, -qx, as well as the internal bending moment, M,. Experiment #2 —Measuring a Bending Moment In this experiment, we devise a model of a cantilever, implement the model using the equipment used in Experiment #1, and measure M, as a function of x. Consider a cantilever with length L, weight W, and con- stant cross-sectional area. We subdivide the cantilever into n equal elements, each of length L/n and weight, W/n (see FIG. 7a). For our physical model, we represent L/n by 1, the length of a section of the bar, and W/n by the weight of a 50-gm mass hanger which we denote by q. Suppose that we wish to determine the internal bending moment, M,, about B in FIG. 7b,a distance, x = 3] from the free end. This is equal to the bending moment due to the three hatched segments to the right of B. If we could ignore the rest of the cantilever and just measure the total bending moment due to these three segments about B, then we would have what we want. This procedure is exactly what our “moment meter” (FIG. 7c) does. If we represent the point B on the cantilever by the pivot, B, of the meter, and represent the weight of each shaded element of the cantilever by a 50-gm weight hanger, then the bending moment we wish to measure is merely that of the three 17 EQUILIBRIUM IN STRUCTURES Albano (a) to spring balance pivot FIG. 7 — Acantilever and "moment meter" weight hangers hung from the three holes nearest and to the tight of B in the moment meter as shown in FIG. 7c. Thus, to measure the bending moment of k elements from the free end of the cantilever, we merely need to hang k 50-gm weight hangers on the “moment meter” and measure the moments of these weights as in experiment #1. 18 Bending Moments and the Eiffel Tower Measurements 1. Measure the bending moment, M,, exerted by the cantilever at a distance x =k] from its free end. 2. PlotM, vs. xand try to obtain an equation for M, asa function of x. This plot is called the bending moment diagram for the cantilever. On deducing equations from plots: If the plot of one set of experimental data (y) vs another (x) is a straight line (a linear plot), then its equation is just the familiar y = mx +b, where y is the ordinate, x the abscissa, m the slope, and b the y-intercept. If, however, the graph is not straight, you might try plotting y against some function of x and, if you are fortunate enough to get a straight line, then, as the mathematicians say, you have reduced your problem to the previous case. That is, suppose you plot y against some function, f(x), of x and you get astraight line then, f(x) is now the abscissa and the equation of your line is y=mf(x) +b. For your cantilever data, plot M, vs x and see if you get a straight line. If you do, determine the equation of the bending moment as a function of x’. The procedure may, of course, be used for other functions of x. Another technique for trying to get the equation of your graphis to try plotting it on semi-log or log- log graph paper. Log-log paper is appropriate if you suspect a power law, y = ax’, in which case, the plot of log(y) vs log(x) should be linear. An Analysis The bending moment about B in FIG. 7b can also be mod- elled by considering the three segments to be a single element of length 31 and weight 3w acting at the centroid of the 3- element segment - i.e., a distance 31/2 from B. Generalize this to the case of k elements and show that the bending moment, M,, at a distance x = kl from the free end of the cantilever is given by, M, = (1/2) qx (1) 19 EQUILIBRIUM IN STRUCTURES Albano where q = w/I, the weight per unit length of the cantilever. How does this compare with your experimental result? Form and Function: Bending Moment Diagram as Silhouette A structure that uses the least amount of materials to support the load it is supposed to bear, and do so safely, is an efficient structure. To attain efficiency, designers try to shape and arrange the elements of a structure so that they are more or less uniformly stressed close to the maximum values allowed by applicable building codes. In FIG. 4b, we saw how stresses may be distributed in a solid beam that is slightly bent. It is clear that there, the beam is notuniformly stressed. The top and the bottom parts are subject to much greater stresses than the middle. It is also clear that the smaller stresses in the middle exert much smaller bending moments. The material in the middle is thus not fully utilized, and greater efficiency may be achieved by removing as much of itas safety allows. One way of doing this is to replace the solid beam with an “T-beam” as shown in FIG. 8a. It consists essentially of two parallel slabs connected to each other at their center lines by a slab perpendicular to the first two. Another is shown in FIG. 8b where the two slabs are connected by perpendiculars and diagonals to keep the beam’s form rigid as well as to withstand shear forces. In structures such as those shown in FIGS. 8a and 8b, the bending moment is borne primarily by the forces acting on the parallel slabs. In FIG. 8c, for instance, the top slab is subject toa tensile force, T, while the bottom slab is subject toa compressive force, C. Since there are no other horizontal forces acting on the shaded portion, T = C, and the bending moment of these two forces is: M,=Th), (2) where h(x) is the depth of the beam at the location, x, as shown. 20 Bending Moments and the Eiffel Tower FIG. 8 — Tensile and compressive forces A pair of forces such as T and C in HIG. 8c, which are par- allel but not collinear, which have equal magnitude, but are oppositely directed are called a couple. Themoment exerted by a couple about any point in its plane is, as in Equation (2), the product of the magnitude of either force and the distance between the forces. Equation (2) says that if the slabs bear the same force throughout their lengths, then the bending moment supported by the structure at any point is proportional to its depth at that point. Thus the least amount of material is needed, and the Sreatest efficiency is achieved, if the shape of the structure imitates its bending moment diagram. This result is of more general applicability, and we will makemuch use of it in the rest of this project. 21 EQUILIBRIUM IN STRUCTURES. Albano The Eiffel Tower Read the appropriate sections of “The Tower and the Bridge” and “Structural Studies”. Because of its height, the Eiffel Tower is subject to consid- erable wind loads which turn out to be more significant than the structure’s weight (dead load) in influencing its form. (See photo by J. Wayman Williams FIG. 9 — The Eiffel Tower by Gustave Eiffel “Structural Studies” Ppp. 38-39). Horizon- tal wind stresses ex- ert forces perpen- dicular to the tower’s axis with effects quite analogous to those of weight on the cantile- ver westudied above. Although wind stress increases with height, the force per unit length on the tower is uniform to first ap- proximation because it is tapered. Since the tower is es- sentially a vertical cantilever, consider- ations of efficiency suggest a parabolic shape. In his design, Gustave Eiffel stayed close to the parabolic silhouette and at the same time created a graceful form which Bending Moments and the Eiffel Tower has become so identified with Paris that it is now practically impossible to imagine the city without the tower. Exercises “Structural Studies” gives the weight of the tower (dead load) as 18.8 x 10‘ Ib (or 18.8 x 10° kips, 1 kip (abbreviated as k) = 1 kilopound = 10° Ib). The height of the tower is 984 ft. and it is 328 ft. wide at its base. 1. What vertical force (in kips) does each of the four supports exert to bear the dead load? 2. The effects of a relatively high wind may be approximated by a constant force per unit height of 2.6 k/ft (“Structural Studies”, p.45). What bending moment [in foot-kips (ftk)] is exerted at the base of the tower in this situation? 3. With a wind as in Exercise 2, what is the force exerted by a leg on the tower’s windward side? leeward side? EQUILIBRIUM IN STRUCTURES Albano CHAPTER 5 Cables and the George Washington Bridge Cables The graceful shape of the cables in a suspension bridge is determined largely by the manner in which it supports its load and the fact that cables can support only tensile stresses. In a suspension bridge, the roadway is supported by the main cables by means of vertical cables called suspenders connecting the main cables to the roadway. The suspenders are usually equally spaced horizontally and exert equal downward forces on the cables. The shape assumed by a cable supporting only its own weight is called a catenary. When it supports a load that is uniformly distributed horizontally as in a suspension bridge, it assumes another shape. In this experiment, we seek a mathe- matical description of that shape. Experiment #3 — Uniformly Loaded Cable 1. Suspend a length of light plumber’s chain from two hooks about a meter apart. Use enough chain so that there is a reasonable amountofsag (FIGS. 9 and 10). Tapeoneormore pieces of large graph paper between the chain and the board and lightly trace the shape of the suspended chain. The resulting shape is a catenary. Observe how this shape is changed as the cable is loaded in the rest of the experiment. 2. Subdivide the (horizontal) distance between the mounts into an even number of intervals and hang 50-gm weight hangers from the chain at the end points of each of the intervals. The weights must be equally-spaced horizon- tally. Make sure that the weights are hanging freely. Make rough sketches of the shape assumed by the loaded cable for each set of weights that you use. 24 Cables and the George Washington Bridge FIG. 10 — Chain supported at two ends 3. Choosing an even number of intervals means thata weight hangs at the mid-point of the chain. This is not usually done in a real bridge, but will make later analysis easier. L FIG. 11 — Weights (q) at ten equal intervals EQUILIBRIUM IN STRUCTURES Albano 4. We can determine an equation that approximately de- scribes the shape assumed by the loaded chain by mea- suring the x- and y- coordinates of the points on the chain to which the weights are attached. To do this, take the mid-point of the chain to be the origin and mark appropri- ate points on the graph paper you had taped on to the board. If you wish, you may read off the coordinates of the points and re-plot the curve on a regular piece of graph paper. Find an equation, y = f(x), that fits your points. 5. Does using more weights, spaced more closely together, make a difference? Horizontal Force, Span, and Sag. The stresses ona suspension cable obviously depend on the load that it must support. For a given load per suspender, the stresses also depend on the span, distance L between the supports, and the sag, or the depth at mid-span, d. FIG. 12 — Spring balances at each end 26 Cables and the George Washington Bridge For a given length of cable and a given loading, the sum of the vertical forces exerted by the supports must equal the weight of the cable and the load, regardless of the values of L and d. It follows then, that only the horizontal components of the forces at the support would change as L and d are changed. In this experiment, you will determine the relationship between the horizontal force, H, exerted at each support, the span, L, and the depth, d, of the cable. Instead of using the loaded plumber’s chain you used in Experiment 2A, it is more convenient to use a short length of heavy chain. In this case, the chain’s shape is a catenary, but this does not change the results. Experiment #4 — Horizontal Force 1. Attacha spring balance to each end of the heavy chain using a piece of string, leaving a couple of inches of string between the balance and the chain, and mount them on the board as in FIG. 12. 2. Using the spring balances, measure the forces exerted at the ends of the chain and the angle @ that the strings make with the horizontal corresponding to a series of at least five values of L and d. spring balance string H=Wcos@ Xu heavy chain FIG. 13 — Determining the horizontal force EQUILIBRIUM IN STRUCTURES. Albano 3. Makesure that the L you measureis the horizontal distance between the ends of the chain, and not the distance between the spring scales. Also make sure that the chain is suspended symmetrically so that both spring balances give approximately the same readings for T and that the angles which the strings make with the horizontal are approximately equal. 4. Record readings of the spring balances as well as the angles made by the strings on both sides of the set-up, and use their averages in subsequent calculations. 5. Calculate the average horizontal component, H, of the forces exerted by the spring balances (see FIG. 13) for each set of values of L and d. In a suspension cable where there are many more sus- penders than you can model here, His related to L and d by the equation, H-=ql?/8d 8) where qis the weight per unit (horizontal) length supported by thecable — i.e.,W/Lwhere Wis the weight of the chain. Devise a graphical way of presenting your data so as to check if your measurements satisfy Equation. (3). The George Washington Bridge Read the appropriate section of “Structural Studies” and, for a lesson on how not to design a suspension bridge, watch the film loop, “The Collapse of the Tacoma Narrows Bridge”. Structural Studies gives the following data about the bridge’s center span with L = 3500 feet, d = 327 feet, and Total dead + live load 47 k/ft Number of cables (2 on each side) 4 Cross sectional area per cable 800 in? 28 Cables and the George Washington Bridge photo by J. Wayman Williams FIG.14 — The George Washington Bridge by Othmar Ammann Exercises 1. Calculate the horizontal and vertical components of the force on each main span cable at a tower support. 2. Calculate the resultant force and the tensile stress on each main span cable at a tower support. Show that this is the maximum stress exerted on a cable in the main span. 3. The efficiency of a structure is defined by, efficiency = —actal stress _ Y= allowable stress where the allowable stress is a value deemed to be safe, and so defined by building codes. The allowable stress for the steel used in this bridge is 82 ksi (kips/in?). Whatis the efficiency of a main span cable? (The side span cables are actually subject to greater stress, so the structure is actually more efficient than the results of the above calculations would indicate. See Structural Studies, p. 106 for more detail). 29 EQUILIBRIUM IN STRUCTURES Albano CHAPTER 6 Putting It All Together: The Salginatobel Bridge A Most Beautiful Structure In this section, you need not perform any experiments. Instead, you will do a number of exercises making use of what you have already learned to understand the design of Robert Maillart’s Salginatobel bridge (FIG. 18), certainly one of the most beautiful bridges in the world. [see Structural Studies, The Tower and the Bridge, D. P. Billington, Maillart’s Bridges: The Art of Engineering (Princeton University Press, Princeton: 1979), and DP. Billington, Robert Maillart and the Art of Reinforced Concrete (MIT Press, Cambridge: 1990)]. In experiments #3 and #4, you showed that a uniformly loaded cable assumes a parabolic shape that supports only tensile stresses. That is, there are no compressions and no bending moments in a uniformly loaded cable. It follows from symmetry considerations that if this shape were inverted to form a parabolic arch which supports a uniform load, then the arch would be subject only tocompressive forces: no tension,no bending moments. It is basically this shape that Maillart used for theSalginatobel bridge. Note thathis use of vertical members, uniformly spaced horizontally, transferred loads from the roadway to the arch ina way that mimics the use of suspenders in a suspension bridge. The configuration also takes advantage of the fact that concrete can resist compressive forces very well, but tensile forces hardly at all. If the arch were to support only the uniformly distributed load of the roadway, then an arch with uniform thickness would have sufficed. The use of a loaded flexible cable to determine the (in- verted) shape of the arch that would support a similar load 30 The Salginatobel Bridge without tension and without bending moments goes back to the great and controversial English scientist and inventor, Robert Hooke (1635-1703). He was brilliant and prolific, but had the misfortune of being a contemporary of Newton whosescientific reputation overshadowed his, in their time as well as in succeeding centuries. Just as he did with the famous law of elasticity that bears his name, Hooke published his rule for the use of cables to determine the shapes of arches in an anagram. The solution of the anagram, first published two years after Hooke’s death is, “As hangs the (loaded) flexible cable, so, inverted, stand the touching pieces of an arch”. An account of Hooke’s con- tributions to the study of arches may be found in H. J. Hopkins’ erudite book, “A Span of Bridges”. Live Loads The bridge must support not only its dead load (its weight), butalso such live loadsas traffic, wind, and in this case, since the bridge is high up in the Swiss alps, snow. In the following, we consider only live traffic loads. Live traffic loads introduce bending moments. As you saw in the case of the Eiffel Tower, one can resist bending moments efficiently by making the profile of the structure mimic the bending moment diagram. This way, thereisenough material to resist stresses where stresses are high, and excess material is not present where it is not necessary. It is not possible to anticipate all possible traffic loads, so one tries to design for those situations which would give rise to the largest stresses. One such situation is when live loads are applied to the quarter points of a span. FIG. 15 showsa schematic diagram of the arch together with its dimensions, subject to pointloads, P, at each of the quarter points. “Structural Studies” uses quarter point loads, P = 55 k. The dead load is 1680 k. Note that by using Hooke’s rule for determining theshape of the arch that would support these quarter point loads, you 31 EQUILIBRIUM IN STRUCTURES Albano merely need to hang equal weights from a cable at its quarter points. You probably do not need to perform the experiment to conclude that the resulting shape is an isosceles trapezoid. Compare this with the result of the calculations in part 2(b) below. Another interesting feature of Maillart’s design is the use of hinges at the bases and at the crown, C, of the arch. These hinges make it possible for the arch to adjust in response to slight motions of the mountainsides to which it is anchored, and ensures that only horizontal forces are exerted at the crown. L = 295ft FIG. 15 — Loads on left half of bridge Exercises 1. Consider the left half (segment AC in FIG. 15) of the bridge. The horizontal and vertical reactions to the assumed quarter-point loads at the support, A, are labeled H and V respectively. The force exerted on this segment at the crown, C, of the bridge is only horizontal because of the crown hinge. By imposing the conditions for equilibrium on this segment, calculate the reactions, H and V, assum- ing that P=55k. 2. The bending moments that results from the quarter point loads may be calculated by separately calculating the moments due to H and those due to the vertical forces, V and P, and then adding the two together. 32 The Salginatobel Bridge FIG. 16 — Horizontal bending moments In the following, take counter-clockwise moments to be positive. (a) Bending moments due to H: Imagine cutting the segment, AB, from the arch where B is the point, (x,y), using the coordinate system shown in FIG. 16. Let M3, be the bending moment which must be exerted about point B in order to balance the moment due to H acting at A. Since you know the shape of the arch, Myx may be obtained for each x graphically from the shape of the arch. Plot My}, as a function of x. FIG. 17 — Vertical bending moments. 33 EQUILIBRIUM IN STRUCTURES. Albano FIG. 18 — The jn Salginatobel Bridge by Robert Maillart (b) Bending moments due to the vertical forces: Again, imagine cutting the segment, AB, from the arch and let My, be the bending moment that must be exerted about Bin order to balance themoments due to the vertical forces (See FIG. 17). Note thatifx < L/4, then V is the only vertical force acting on the segment, but that if x > L/4, then both V and the load, P, at the left quarter point act on it. Plot Myx. (c) Graphically (qualitatively) add My, and Mj, and plot the net bending moment as a function of x. 3. Compare the bending moment diagram you obtained in (2) with the actual profile of the Salginatobel bridge. Notations and References REFERENCES David P. Billington, “The Tower and the Bridge: The Art of Structural Engineering”, Basic Books: New York, 1983 (in paperback, Priceton University Press: Princeton, 1985). This is the book in which the term “structural art” is used. The tenets of structural art are exposed and illustrated by the careers and the designs of the masters of the art. David P. Billington and Robert Mark, “Structures and the Urban Environ- ment: Structural Studies”, Princeton University Department of Civil Engineering: Princetion, 1984. Detailed but simplified analyses of a number of historic structures. The laboratory exercises here are really designed as an introduction to the structural studies of the Eiffel Tower, the George Washington Bridge, and the Salginatobel Bridge contained in this volume. David P. Billington, “Robert Maillart’s Bridges”, Princeton University Press: Princeton, 1979. The life and the bridges of one of the masters of structural art. David P. Billington, “Robert Maillart and the Art of Reinforced Concrete” MIT Press: Cambridge, 1990. Illustrated with many beautiful new color photos of Maillart structures. JE. Gordon, “Structures, or why things don’t fall down”, Da Capo Press: New York, 1978. A delightful book on materials and structures, natural as well as man made. H. J. Hopkins, “A Span of Bridges”, Praeger Publishers: New York, 1970. An erudite and profusely illustrated history of bridge building. It is, however, not always smooth reading. 35 EQUILIBRIUM IN STRUCTURES. Albano EPILOGUE “The best engineering structures of the past 200 years, since the introduction of industrialized iron, have been de- signed to achieve both efficiency and elegance, however, effi- ciency does not always result in elegance; there are too many efficient but ugly structures to support such a simple-minded idea. Efficiency is a necessary but not sufficient condition for engineering elegance. These studies emphasize how elegant structures can result from allowing inherent structural prop- erties to influence the form and how such structures are often the most efficient and economical.” — From the Preface to “Structural Studies” Part I NEW FORMS IN LONG-SPAN CONCRETE BRIDGES — A Structural Study of the Felsenau and Ganter Bridges William Case Grinnell College December 1990 Part II New Forms in Long-Span Concrete Bridges Table of Contents Chapter 1 Tntroduction 2. The Felsenau Bridge The Ganter Bridge Weferences se 6. Problems 39 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case photo by William Case The Ganter Bridge by Christian Menn 40 Introduction CHAPTER 1 Introduction The Cantilever Construction Method Some structures are so striking that they immediately appeal to our aesthetic sense and curiousity. There is an elegant rightness about them that demands an explantion and deeper understanding. The Felsenau and Ganter Bridges of Christian Menn are two such structures. The Second World War left many European cities in ruin. The ravages of war had also demolished many bridges, especially those along the Rhine River. Faced with such a massive rebuilding task new methods of construction were developed. One major development was cantilever construction. This method originated with the German engineer Ulrich Finsterwalder. He built several bridgesin the 50’sand 60's using cantilever construction, including the 200 meter span over the Rhine at Bendorf, Germany. This last structure remains one of the longest concrete spans in the world. In the cantilever construction method one begins by con- structing the foundations and piers in the usual way. Once these are completed the deck is built segment by segment, each segment being supported by the previous one. In this way the loads are carried back to the pier. The concretesegments may either bemade elsewhere and lifted into place or poured into forms supported by the earlier sections. In either case, once in place, steel prestressing tendons are passed through the added segment and it is com- pressed into the earlier sections. The partly completed deck extends out asa cantilever, from which the method gets its name. Construction continues until the cantilevers from adjacent piers meet at midspan. ‘The earlier method of construction would use falsework built up from the ground. With the new method, much less falsework is used, especially if the bridge is high. Since the cost of 41 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case falsework represents about 30% of the cost of the bridge, these savings are very significant. Another advantage of this method is thatitleaves theregion underneath the construction siteopen. This is very important if the bridge is to cross a road or river carrying heavy traffic. Along with these advantages this daring method of con- struction clearly presents large stresses which must be analyzed carefully. We will do this through the study of two particular structures, Christian Menn’s Felsenau Bridge of 1974 and his Ganter Bridge of 1980. The main goal of this section is to present the technical design considerations relevant to these structures. Fora discussion of these structures from a different perspective the reader is referred to the paper by D. P. Billington in Civil Engineering. 1 photo by William Case The Felsenau Bridge by Christian Menn 42 The Felsenau Bridge CHAPTER 2 The Felsenau Bridge Bridge Design Competition When Christian Menn was invited to enter a bridge design competition for a bridge to carry a 6 lane expressway over the Aare River Valley in the small Bern suburb of Felsenau, he developed a 17 span structure with a total length of slightly over 1 kilometer. This design is shown in FIG. 1. i [ree FIG. 1 — Seventeen spans cross the Aare River Valley The longest spans, two of 144 meters and two of 96 meters, were to be executed via the cantilever method while the others, ranging from 38 to 48 meters, were to be built using traditional methods. Although the design owes much to the earlier work of U. Finsterwalder, it contains many elements which mark it as one of the most finely developed structures of its type. Six lanes is a very wide roadway and would normally require the building of two bridges side by side, each carrying three lanes. Menn chose to carry all six lanes with a single box deck, as shown in FIG. 2. With this approach, one has a single bridge, rather than the clutter of two bridges standing side by side. The cross section of thesingle box is deeper over the piers than atmidspan. Itis also strongly tapered vertically, making a smooth meeting between the box and the pier while offering a relatively wide support where the box meets the upper partof the deck. Thebox is one monolithic structure along its entire 1116 meter length. As such it is quite likely the longest single piece of concrete that 43 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES. Case Cross Section at Pier 26.2m Cross Section at Midspan 7 FIG. 2 — Cross sections of the single box deck for 6-lane superstructure does not rest on the ground as a road. This innovative design won the competition and the bridge was executed according to Menn’s design. Analysis of Stresses During Construction As one can sense intuitively, great care must be taken with the analysis of the stresses during construction as well as after completion. One critical point occurs just before the two halves meet at midspan. Each half is a cantilever at this stage. The cross section of the box changes along its length in a rather complicated way, for reasons which will be presented later. In order to simplify our analysis we will assume that the box has aconstant weight per lengthof q = 41 tons per meter. Following the dimensions used in the design of the structure we will use tons for force (1ton = the weight of 1000 kilograms = 9800 newtons = 2.2 kips) and meters for length. The vertical forces acting on the dead load of the hollow box cantilevers are shown in FIG. 3. Balancing these forces we have a vertical upward force provided by the pier and equal to 44 The Felsenau Bridge peep yy dd 4 FIG. 3 — Vertical force from pier qL= 41 (144) = 5900 tons. This force (qL in FIG. 3) the upward force of the pier on the deck and must be carried by compressive stress within the pier. Since concrete is strong in compression, this force is easily carried by the pier. This force is proportional to the length L. Other forces examined below increase more rapidly with length and will require more care. We determine the internal forces of the box section by considering a length of the box as shown in FIG. 4. Balancing the vertical forces we find: V=qx v = 41 tons/meter FIG. 4 — Box section internal forces 45 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES. Case A graph of shear, V versus x, plotted with x increasing to theleft so as to match the figure of the structure, is given in FIG. 5. The bending moment M is given by: M=qx(X} 2960 tons ax(5) v x= 72 meters «x x-0 FIG. 5 — Shear force Agraphof M vs xisshownin FIG.6. This bending moment will lead to a force couple consisting of a compression along the lower region of the box and tension at the top. 106,000 ton-meters x= 72 meters x=0 =< « FIG. 6 — Bending moment 46 The Felsenau Bridge FIG. 7 — Force couple This is shown in FIG. 7. The magnitude of the force couple is given by: Since, for sections of the box well away from the free end x is much greater than d, we expect that T will be much larger than V. Thus we see that internal forces due to the bending moment will be larger than those directly due to shear and, in the following, we willonly consider those internal forces associated with the bending moments. A second key consideration is concrete’s inability to carry tension. In the cantilever construc- tion of concrete bridges the steel tendons which hold the added segments play the added role of prestressing the concrete. The steel is placed under sufficient tension so as to leave the concrete in the upper section of the box under compression. Weexamine the stresses in the box at the point where the box meets the pier. The box cross section at the pier is shown in FIG. 8a. Although the original calculations were carried out for the actual cross section®, we will analyze the simpler cross sec- tion shown in FIG. 8b. The total area in top plus bottom, the shaded areas in the figures, is the same for the model and the actual cross section and is equal to 18.6 square meters. The total area of our model and the actual cross section is 25 square 47 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case (a) Actual Cross Section nam em Total Area 25 m* . | Total Shaded Area 18.6 m | (b) Simplified Cross Section Total Area 25m? ‘ Total Shaded Area 18.6 m FIGS. 8a and 8b — Actual and simplified cross sections meters. Pure compressions applied to the cross section will be taken by the total cross section. We assume that bending moments are carried by a force couple in the top and bottom sections only with negligible contributions from the sides. The vertical distance between the centers of the top and bottom sections is 7.14 meters for the actual structure and for our simplified model. The bending moment determined above contributes a tension T in the upper section and an equal com- pression C in the bottom. Since, M=Td we find, —M _ 106,000 _ ee 14,800 tons We assume that this force is carried by a constant stress over each section, thus we find a stress of: f=T= ee 1590 tons /square meter A 93 as tension in the top section and an equal value for compression in the bottom. 48 The Felsenau Bridge The prestressing in the box adjacent to the pier consists of a combination of cables applying an initial total force of P=22,680 tons to the concrete.‘ The effect of the prestressing can be considered as a pure compression plus a bending moment equal to the prestressing force times the vertical distance from the center of the box section to the location of the cables ( the eccentricity). The pure compression results in a compressive stress of: _ 22,680 25 The cables are placed fairly high in the upper section, about 0.1 meter below the top surface of the cross section. Thus the bending moment applied by the cables is M = Pe = 22,680 x 3.9 = 88,500 ton-meters where e is the eccentricity. fs = 907 tons /square meter The resulting force couple induced in the concrete is, c=T= Me 88.500 = 12,400 tons We assume that this force is carried uniformly by the top and bottom sections of the box. Thus we have i uo = 1330 tons/sq. meter of compression stress at the top and an equal amount of tension stress at the bottom. The total stress is presented graphically in FIG. 9. (where + is tension and ~ is compression) +1590 Tim? -1330 Tim -647 Tim? + + 1590 Tim 7 -907 Tim? +1330 Tim? -1167 Tim ® Load Prestressing Total Fig. 9 — Tension (positive) and compression (negative) in the box cross section 49 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case Clearly all of the tension stress has been removed. The largest stress is 1167 tons/sq.meter at the bottom. This is well within the crushing strength of 4500 tons/sq.meter for the concrete used in this structure.> Analysis of Stresses on Completed Structure Once the box is joined at midspan, the wearing surface is added and this increases the load on the deck by 3 tons/meter. We also allow 11.4 tons/meter for maximum expected live loads on the bridge for a total of 14.4 tons/meter added after closure.® With the box closed and continuous, the internal forces can no longer be determined by simply balancing forces and bending moments; the loads can be carried by a variety of combinations of internal forces and bending moments. Con- sider the simple model of the completed box shown in FIG. 10. The deflection under the load is shown greatly exaggerated. FIG. FIG. 10 — Model of completed box under load Under theassumptions that (1) the load q is uniform along the box, (2) the cross section is constant along the length and (3) the box at the piers is at the same level and horizontal (i.e., no 50 The Felsenau Bridge rotation at the piers), the bending moment is shown in FIG. 11a. This system is called a “built-in” beam. 2 ql 24 FIG. 11a — Bending moment for built-in beam Although a careful derivation of this result is above the level of this presentation, we can see that itis consistent with our intuition and our earlier results. The graph indicates tension along the top of the box over the pier and along the bottom of the box at midspan as one would expect from Fig. 10. Before the box was closed at midspan we found the bending moment shown in FIG. 11b for a uniform load of q. qu 38 ue FIG. 11b — Bending moment for cantilever 51 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case If we consider a beam which is free to rotate at the piers (simple beam) we would obtain the bending moment given in FIG. 11c. FIG. 11¢ — Bending moment for a simple beam The corresponding structures are shown in FIG. 12. If we com- pare FIGS. 11a, 11b and 11c, we see that they all have the same shape with M,,,-M,,,,=qL7/8. Onecan be converted into the other by simply raising or lowering the curves. \-EEE EEE FIG. 12a — Built-in beam Fig FIG. 12b — Cantilever 52 The Felsenau Bridge FIG. 12¢ — Simple beam The difference in these three systems is the bending mo- ment applied at the piers which has the effect of adding a constant value to the expressions for M. Although all have the same value of M_,,-M.,,, the built-in beam has a significantly smaller maximum value of the magnitude of the bending moment equal to qLl?/12 rather than qL?/8. Thus the box in our built-in structure can be made more slender then for the other structures. Applying our analysis of the built-in beam, we find the bending moment diagram shown in FIG. 13 for the contribution due to the added 14.4 ton-meters. 25,000 T-m FIG. 13 — Bending moment from added loading The total bending moment is found by adding this result to that presented in FIG. 7, giving the total bending moment diagram in HIG. 14. This result is in close agreement with the result obtained by Menn and his colleagues by much longer and more careful means.” 53 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case 144m [ 12,500 T-m FIG. 14 — Total bending moment We again consider the stresses in the box at the point where it meets the pier, but this time for the completed struc- ture. We again make use of the simplified cross section shown inFIG.8b. As before, weassume thatall of thestresses are carried by the top and bottom sections and the sides carry a negligible amount. We find the compression and tension due to the total loads by, 131,000 C=T= i 18,300 tons This appears as a tension stress of, f= 130 = 1970 tons/sq. meter in the top sections and an equal amount of compression in the bottom. We also want to allow for a reduction of the prestress- ing forces due to creep of the concrete. A commonly used rule of thumb to account for this effect is to multiply the initial prestressing force and resulting stresses by 0.85. These stresses and their sum are shown in FIG. 15. 54 The Felsenau Bridge +1970 tons/sq.meter +1130 tons/sq.meter +69 tonsisq.meter + + = = = is _ “1970 tonsisq.meter —-771 tonslsqmeter 41190 tons/sq.meter 1610 tons/squmeter Loads Prestressing Total FIG. 15 — Tension and compression in the box cross section The compression at the bottom of 1610 tons/sq. meter is well within the crushing strength of 4500 tons/sq-meter. Our nu- merical result shows some tension in the top section butremem- bering the degree of our approximations this is consistent with zero stress. Assuming that concrete can take about 1/10 as much stress in tension as in compression, we conclude that 450 ton-meter tension is allowed and our value is well within that limit. In some cases a design will allow tension in a concrete structure under very unusual short term loads. Those tensions are then controlled with ordinary steel reinforcing. The cantilever method of construction leads to large internal forces in the deck near the pier due to thelarge bending moments. For the Felsenau Bridge, where the bending moment at the pier is a very large 131,000 ton-meters, this internal force is accommodated by making the box deeper. From FIG. 2wesee that the box at the pier is about 2'/, times the depth at midspan. Thus the internal force, which is given by F=M/dis reduced by a factor of 2/5. The compressive force is carried by the bottom portion of the cross section. This area is increased near the pier 55 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case by making the lower section 1.45 meter thick as opposed to the 0.20 meter thickness used toward the center of the span. In this way the stress, which is given by F/area of the bottom, is again decreased and the compressive stress is held to less than 1700 tons/sq. meter. Upon hearing that this 1000 meter 17 span bridge super- structure box acts acts as a monolithic structure, the reader may wonder how expansion and contraction due to temperature change is accommodated. For the spans toward the ends this is done by allowing the box to slide on the tops of the piers. For the center spans each pier consists of two slender, relatively flexible piers 12 meters apart. As expansion occurs these piers bend in a very slight S shape allowing expansion without damage to the structure. photo by Wiliam Case Pier for the main span of the Feisenau Bridge 56 The Ganter Bridge CHAPTER 3 Ganter Bridge An Engineering Challenge Roads which go over Alpine passes are notorious for their tight curves and the Simplon passroadis no exception. On its way to the crest of the pass, the road must pass the Ganter tiver and its steep walled valley. This area presents the civil engineer with a very severe challenge, in that both sides of the valley are unstable. The very steep North Side, whose surface tockis prone to slides does offer sound bed rock at a reasonable depth for the founda- tion. The less steep South side whose surface has been measured to move 0.5cm/year does not offer a satisfactory foundation. When Christian Menn was consulted about bridging this valley, heproposedan8span structure where the most difficult area would be covered by one 174 meter span and two side spans of photo by Wiliam Case Pier 3 of the Ganter Bridge is 124 meters high 57 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES. Case FIG. 16 — The Ganter Bridge 127 meters. This proposal is shown in FIG. 16. Piers 2 and 3 are fixed to bedrock and thermal expansion of the span between them is taken up in the flexure of the piers. Piers 4,5 and 6 on the slowly moving South side are free to slide at their bases. Due to the extreme height of the main piers, (124 meters for pier #3) itis unreasonable to construct such a structure using falseworkfrom the ground up to the deck. Thus we see that this is a perfect candidate for cantilever construction. As we have seen earlier, the bending moment near the piers during such constructions are very high and increase as the length of the span squared. Here, since the superstructure is smaller thanon the Felsenau, the road way carrying only two lanes, the weight per length will be less, but, with the narrower structure, the box cross section available to carry the loads is also reduced. Thus we see that, even for narrow bridges, long spans lead to high internal stresses. In order effectively to reduce the span, Menn suspended a section of the bridge on inclined cables. These act as “islands” suspended in space. This is shown in FIG. 17. The deck must now run from the pier to this “island”, then to the other “island” and finally to the next pier. The officials of the Canton were so impressed with Menn’s daring design, that they arranged to have it accepted without going through thedesign competition which would usually berequired 58 The Ganter Bridge FIG. 17 — Cable suspended "islands" for sucha large structure. Construction began in the summer of 1976 and the bridge was completed in 1980. The Ganter Bridge is currently the longest span in Switzerland. We will carry out our calculation of internal stress at three critical stages of construction. The first is the cantilever construction out to the point where the cables are attached at 36 meters. The second is just before the meeting at midspan and the third is for the completed structure. The cross section of the box section of the Ganter Bridge is not constant along its length, although it is much closer to being constant than that for the Felsenau. Again, in order to simplify our calculation, we approximate the weight of the box to bea constant of 16.73 tons per meter over its entire length. Construction Loads - First 36 Meters of Deck The first phase of construction of the main span consists of building 36 meter of box as a cantilever from the piers 3 and 4. This is before the section with the diagonal cables. As was the case with the Felsenau, the most severe demands on the struc- ture are associated with the bending moments. Thisis given by, -& Me"D which gives a maximum moment of M= Ee = 10,800 ton-meters 59 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case In order to reveal the relationship, FIG. 18 shows the entire deck with the completed section shaded. This will give large forces in the deck near the piers. 10,800 ton-meters FIG. 18 — Cable suspended "islands" The box cross section at the pier” is shown in FIG. 19a, and our simplified, symmetric cross section is given in FIG. 19b. Ne- glecting the stress in the sides, we find a compressive force of, —M _ 10,800 ea 465 = 2320 tons 60 The Ganter Bridge and an equal force in tension at the bottom. This gives a stress of: =C ~ 2320 — f 35 663 tons/sq.meter fq—— 10m —_______pJ} 5 ! 5m > Y Bess. Simplified Cross Section FIG. 18 — Actual and simplified cross sections The concrete used in the construction has a crushing strength of 4100 tons/sq.meter in compression, but only about 1/10 or 410 tons/sq.meter in tension. Thus we see that some- thing must be done to relieve the tension at the top of deck. As in the Felsenau, this was done with prestressing. A total of 26 cables pass through the upper part of the cross section, tens- 61 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case ioned to a total force" of 3520 tons. This force can be represented. asa compressive force of 3520 tons distributed uniformly on the entire cross section of 10.4 sq. meters plus a bending moment due to a force of 3520 tons acting above the center of the cross section. The uniformly distributed force gives a compressive stress of, = 3520. — f 104 338 tons/sq.meter The prestressing cables are located about 0.1 meter below the top surface or 2.4 meters above the center of the box section. Thus the bending moment is M = 3520 x 2.4= 8450 ton-meters This is carried by a compressive force of, c=M. 850 - 1810 tons in the upper section and an equal tensile force in the bottom. If the force at the top is distributed evenly over the top section we have an additional compressive stress of, f= 1810. = 517 tons/sq.meter in the top section. An identical calculation gives a tension of 517 tons/sq.meter in the bottom section. +663 tons/sq.meter -517 tons/sq.meter -192 tons/sq,meter +E + = ge Bae | 663 tons/sqmeter _-338 tons/sq.meter +517 tonsisqmeler —_-484 tons/aqmeter Load Prestressing Total FIG. 20 — Total stress in box cross section at pier 62 The Ganter Bridge The total stress is shown in FIG. 20. We therefore see that the tension has been removed and all of the values are within the 4100 tons/sq.meter limit of the concrete.” Construction Loads - Just Before Meeting at Midspan We now consider the internal stresses in the box at the construction stage where the box sections are about to meet at midspan. At this point the “islands” referred to above are in place and the box has been extended from the island by the cantilever method. We examine the stresses at two critical points; at the box cross section just to the right of the island, 34 meters from midspan, and at the box cross section at the pier. These points are shown in FIG. 21. Deck cross section Deck cross section at pier 34m from midspan FIG. 21 — Construction loads before midspan meets The determination of the stresses at the 34 meters from midspan section is almost identical to the previous calculation. 63 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case The bending moment due to the weight of the box, 16.73 tons/ sq-meter, is given by, 2 M= 7 16.13 x3 = 9670 ton-meters The box crosssection at this point is shownin FIG.22aand our simplified, symmetric cross section in FIG. 22b.¥ FIG. 22b— Simplified cross section This bending moment will lead toa compressive force at the bottom and a tensile force in the top. Neglecting the effects of the sides and assuming that the forces are carried by uniform stresses over the top and bottom sections we find, C=T= M = 9670 - 4240 tons 2.28 for the magnitude of the force and = 14240_ f= An 215 1970 tons/sq.meter 64 The Ganter Bridge for the tension stress in the top and an equal value for compres- sion stress in the bottom section. There are no prestressing cables within the box in this cross section, but prestressing is supplied by the diagonal cables" radiating from the upper section of the pier. A total of 16 cables, each tensioned to 219 tons, are attached to the box to the right of this cross section as shown in FIG. 23. 16 Cables attached in this section p—_—~—.__ midspan FIG. 23 —Loads from last 16 diagonal cables Although they meet the box over a distance of 20 meters and at a variety of angles, we will approximate their effect by a single force of 16 x 219 = 3500 tons acting at a point 24 meters from midspan and atan angle of 8° with the horizontal (note the bridge deck is at a small incline and not quite horizontal, but these effects will be neglected). The component of this force along the deck, 3500 cos 8° = 3470 tons, will act much as the prestressing forces seen in the earlier calculation. It will give a uniform 65 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case compression over the whole cross section of, =T-340- f= Ae 30 = 580 tons/sq.meter plus a bending moment since the force is not applied to the center of the cross section. Since these cables extend from near the top of the box section the distance from the point of applica- tion of the force to the center of the box section is 1.25 meters and the bending moment is given by, M=Fe = 3470 x 1.25 = 4340 ton-meters This results in a pair of forces in the top and bottom sections of, -TuM~4340.— C=T=4 338 1900 tons and stresses of == 1900. f An 215 884 tons/sq.meter This will give compression at the top and tension at the bottom. The vertical component of the force of the 16 cables, 3500 sin 8° = 487 tons, will also give rise to a bending moment in the cross section. The effect of this force will be reduced due to the weight of the portion of the concrete which encases the cables to the right of our cross section. This weightis given in thecalculations of the bridge as 66 tons and we will assume thatit also acts at the same point as the force of the cables.’ This gives a bending moment of, M = Fe = (487 - 66) 10 = 4210 ton-meters forces at the top and bottom of the box of, =T=M.4210_ C=T a 228 1850 tons and stresses of, in the top and bottom sections. This contribution will give 66 The Ganter Bridge additional compression at the top and tension at the bottom. The total stress at the cross section is combined in FIG. 24. +1970 tons/sq.meter -884 tons/sq.meter -860 1ons/sq.meter ~354 tonsisq.meter P cA ; = = = +1970 tonsisq.meter —-580 tons/sq.meter +884 tons/sqmeter +860 tons/sq.meter -806 tons/sq.metor + FIG. 24 — Total stress in the box cross section 334 meters from midspan As we can see there are no tension stresses in the cross section and the compression stress never exceeds the limit of 4100 tons/sq.meter in the concrete. Next we consider the additional stresses on the cross section near the pier due to the presence of the added construc- tion. This load must be added to those already found in FIG. 20. Inorder to determine the added loads we must first understand therole played by the cables radiating from the top of pier down to the box. We have already seen how 16 of these cables prestress the cross section 34 meters from midspan. Another 16 cables, also stressed to 219 tons each for a total of 3500 tons, are attached to the box between 36 and 53 meters from the pier. These form the island described earlier. After the cables are in place and tensioned as indicated above, they are encased in concrete and tensioned further to their final value of 274 tons each." In this way the concrete encasing will also be prestressed and under compression. This final encasing is performed at about the same time as the first side reaches midspan. Once completed the entire island consisting of cables plus encasing formsa very rigid structure compared to the box. To emphasize the relative flexibility, one can think of a hinge being in the box 67 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case at a point just to the left of the island at 36 meters from the pier as shown in FIG. 25. FIG. 25 —Loads and imaginary hinge Thus we see that bending moments applied by the box to the island from the right are carried by the island and not applied to the box at the left of the island.” This result is also supported by the more detailed calculations which show very low bending moments at our imagined hinge site. Although the hinge will not directly apply a bending moment to the section of box between the pier and the island it will apply vertical and horizontal forces. The horizontal component of all 32 cables, is applied to the hinge and thus carried to the cross section at the pier. We have already collected 16 cables and assumed them to act at a point 63 meters from the pier. For our purposes the remaining 16 cables will be considered as acting at a point 47 meters from the pier atan angle of 13°. This will givea horizontal component 68 The Ganter Bridge of 16 x 219 cos 13° = 3410 tons to the left and a vertical compo- nent of 16 x 219 x sin 13° = 788 tons upward. Note that the force used is that applied before the final prestressing of the concrete encasing. The final added prestress- ing is taken entirely by the encasing and not applied to the box. Anadditional load associated with theisland is the weightof the concrete encasing. Although some of this weight is supported by the pier directly, detailed calculations reveal that 408 tons are supported by the deck in the region of the island.* This is in addition to the weight of the section of the encasing to the right of the island considered earlier. One final load which will apply a vertical force to the hinge is the weight of the box to the right of the hinge which is given by 16.73 x (87-36) = 853 tons. All of these loads and forces are shown in FIG. 25. Combining these we find a total horizontal force of, 3470 + 3410 = 6880 tons When this is distributed uniformly over the cross section at the pier we have a stress of =~ 6880 _ f An 104 660 tons/sq.meter in compression. The total of the vertical forces, which result in a vertical force at the hinge, is given by, 487 + 788 - 853 - 408 - 66 = - 52 tons or 52 tons downward. One can see how carefully the tensions in the cables were chosen so as to almost balance the weights of the various components. This vertical force will give a bending moment at the pier of, M = 36 x 52 = 1870 ton-meters We now see the importance of balancing the vertical forces at 36 meters; they can give large bending moments. This bending 69 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case moment will give a force couple in the pier cross section of, =C=M-=1870 - 499 T=C 465 402 tons and a stress of tS a = 115 tons/sq.meter This represents a tension at the top and a compression at the bottom. All of the stresses on the box cross section at the pier are collected in FIG. 26. 192 tonsisq. meter +115 tonsisq.meter -737 tons/sa.meter {iii 338 — +998 wonsareer OE + = onseamotr ~484 tonsisq.meter — ~650 tons/sq.meter-115 tons/sq.meter -1259 tonsisq.meter ‘rom FIG. 20 FIG. 26 — Stresses in box cross section at pier Again we see that there is no tension in the cross section and the maximum compressive stress is below the 4100 tons/ sq. meter crushing strength of the concrete. Final Structure with Live Loads Inthe calculations carried out prior to the construction of the bridge, a variety of live load cases were considered. We will carry out the analysis for only one of these, a uniform load of 3.8 tons/meter over the entire deck. We will again examine the stresses at the box cross section 34 meter from midspan and the box cross section at the pier. We will also assume that due to creep of the concrete the internal prestressing has been reduced to 0.85 of their original values, the same figure assumed in the original calculation.” This factor should not be applied to the cables radiating down from the pier top since concrete creep 70 The Ganter Bridge will not reduce their force on the deck. Indeed, if the concrete in which these cables are embedded creeps, less of the cable’s tension will be carried by the encasing and more will be applied to the box. We will assume that their force on the box is unchanged. Thus the reduced prestressing will only effect the box cross section at the pier. We determine the bending moment at the 34 meters from midspan section by considering the span between the islands as a built-in beam. In our calculation for the Felsenau Bridge we gave the bending moment at the end of a build-in beam. For the present case this becomes 2 M= de 38x 58 1460 ton-meters This will give rise to a force couple of, =C=M.1460_ T=C a 228 640 tons and a stress of, $40 ~ 298 tons/sq.meter in the top (tension) and the bottom (compression) sections. The live load stresses are added to the stresses found earlier and displayed in FIG. 27. -354 tons/sq.meter +298 tons/sq.meter -56 tons/sq meter -580 tons/sq.meter + = -806 tons/sq.meter -298 tons/sq.meter -1104 tons/sq.meter from FIG. 24 Live Loads Total FIG. 27 — Total stresses with live loads 34 meters trom midspan 71 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case Again we see that there is no tension in the cross section although the compression has become very low in the top section indicating that full use is being made of the material. At the bottom the compression is a maximum but still well below the 4100 tons/sq.meter crushing strength of the concrete. Theadditional stresses in the concrete at the cross section at the pier are three fold. As already indicated, the prestressing forces in the internal prestressing will be reduced to 0.85 of their original value due to creep of the concrete. A second source of stress is the additional live load between the pier and 36 meters. We treat this section of the box as a built-in beam of length 36 meters and find an added bending moment at the pier of 3 12 This is a somewhat questionable application of the built- in beam model. An exact calculation, which is somewhat be- yond the scope of our treatment, yields a result which is half again larger, but this correction is still negligible compared to other contributions to the total stress in the box cross section. A third source of stress is the additional vertical force due to the additional weight over the 51 meters between midspan and 36 meters from the pier. This force is applied at the location of our imaginary hinge, and gives an added bending moment at the pier of, = 38x30 - 410 ton-meters M=qL = 36 x 3.8 x (87-36) = 6980 ton-meters When added to the moment found for the live load between 0 and 36 meters we find, M = 6980 + 410 = 7390 ton-meters This will give a force couple of =C=M.7390 Tc a 465 1590 tons and added stresses of, 500 = 454 tons/sq.meter The Ganter Bridge in the top (tension) and bottom (compression) sections. These contributions are added to those displayed in FIG. 20. and Fic. 26 to give the total stress in FIG. 28. -154 tons/sq.mater I I -947 tonsieq.metor Mea “1740 tons'sametor +454 tone/sqmoter 7 Ti pr Live Loads + “454 tonsisq.metor —/ + +115 tons/eq.meter “115 tons/sq moter Cables >and Dead Loads -660 tonsiaq:meter +439 tonsisq motor -439 tonsisqmetor Prestressing \ (reduced due to creep) “287 wneteg.mowr +663 tons/sq.metor il p-Dead Loads 663 tonsisqmeter top bottom FIG. 28 — Final stresses in box cross section at pier Again we see that all tension has been avoided and the maxi- mum compression of 1740 tons/sq.meter is high but well below the4100 tons/sq.meter crushing strength of the concrete. 73 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case Controlling the Bending Moments As was the case with the Felsenau Bridge, the main structural problem is the control of the bending moments. For the Ganter Bridge this is achieved through two mechanisms. The effective length of the span is reduced from 174 meters to about 68 meters by the use of the “islands” described above. Since the bending moment increases with the square of the span, this reduces the moment to about 1/6.5 of its original value. This type construction adds to the cost, so there is a price to be paid, but for such a long span it is justified. Theothermechanism used to control the bending moment is similar to that used on the Felsenau Bridge. As shown in FIGS. 19a and 22a the box section is deeper near the pier than in midspan by a factor of about 2. This reduces the size of the internal force couple. The internal compressive stress is further reduced by making the bottom portion of the box thicker, 0.45 meters at the pier compared with 0.18 meters toward midspan. Through these mechanisms the maximum compressive stress is reduced to about 1700 tons/sq.meter, approximately the same value we found for the Felsenau Bridge. The reader may wonder why, if the concrete specified has a crushing strength of 4100 tons/sq.meter , one uses such a small fraction of the maximum value. Part of the answer lies in the nature of the test by which the 4100 tons/sq.meter crushing strength is determined. In the test one takes a small cube of the concrete and compresses it on two opposite sides until it fails. If arectangular solid sample were used and compressed along its long dimension a lower value would be found. This is mainly due to the reduced influence of friction of the compressing faces on the concrete."* In our bridge box the concrete acts more like a long rectangle than a cube. The estimates for the reduced strength vary, butrange from 0.7 to 0.9, and a reduction to 0.6 of the cube crushing strength is indicated in the Swiss code.” A second reason for the conservative estimates of the crushing strength of concrete is the nature of concrete itself. It is a brittle material. If the structure fails due to tension one would expect the plastic character of the steel imbedded in the 74 The Ganter Bridge concrete to assist in maintaining the structure and to give some warning. If, however, the failure is due to compression of the concrete there would be no plastic flow and the collapse would be very fast and complete. Thus we see that, even though the cube crushing strength of the concrete is 4100 tons/sq.meter the maximum compressions of about 1700 tons/sq.meter rep- resent a full and efficient use of the material. Conclusion. For the reader who has worked through all of the details of this presentation, congratulations! Such a complicated un- dertaking would not be justified for one outside of engineering if it were not for the beauty of the structures. I would like to be able to tell you that there are airline tickets attached to the inside cover of this book allowing you to see these bridges first hand but unfortunately that is not the case. In any event, I hope that this treatment has allowed you to see some of the aesthetic as well as scientific choices thathave been made in designing these structures.” photo by William Case The Ganter Bridge at the Simplon Pass in Switzerland © : 75 NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case References 1. DP. Billington. Swiss Bridge Design Spans Time and Distance in Civil Engineering - ASCE Nov 1981 2. The value for the weight per length for the Felsenau Bridge was obtained by working backward from the bending moment for the cantilever construction phase, M,, (T= 0), given in Abb. 14 jn Die Projektierung des Uberbaues by H. Rigendinger and W. Maag found in Felsenaubriicke - Projekt und Ausfiihrung published by Gasser and Eggerling AG Chur, Switzerland. 3. The cross section of the box of the Felsenau Bridge near the pier is given in the FIG. between pages 16 and 17in thearticle Projektgrundlagen und Entwurfby C. Menn foundinFelsenaubriicke-Projektund Ausfiihrung published by Gasser and Epgerling AG Chur, Switzerland. The total area in the top and bottom sections of the actual cross section near the pier are estimated to be 7.9 sq.meters and 10.8 sq.meters respectively. This is then divided equally to give 9.3 sq.meters for the top and bottom sections of the simplified cross section. The area of the sidesis estimated to be 6.4 sq.meters giving a total area of 25 sq.meters for the actual cross section and for our model. The main reason for using this simplified model is to place the neutral axis in the center of the cross section. 4, The prestressing forceistaken from Abb.28 p48 of thearticle DieProjektierung a des Uberbaues by H. Rigendinger and W. Maag. See Reference 2 for a more complete citation. . Ibid. p32givesthecrushing strength @,,,,)=450kg/cm?=4500tons/sq.meter. 6. Theliveloads were estimated by using the loads used in calculations carried out for the Ganter Bridge (see ref. 9) and multiplying them by 3 since the Felsenau Bridge has 6 lanes rather than 2. The running surface weight per length was estimated by assuminga thicknessof4.5cmand using the density of concrete 2.5 T/cubic meter. 7. The total bending moment for the finished structure is given as Mg in the 76 same FIG. cited in ref. 2. This is used in Menn’s calculation for the Ganter Bridge Vorprojekt - Ganterbriicke - Generelle Stat. Berechnung page 20. Most of the specifications and loads for the Ganter Bridge were taken from Menn’soriginalcalculation for the structure, Schweitz Nationalstrassen 10. 11. 12. 13. 14. 15. 16. DRAFT ~January 29,1991 References N9, Vorprojekt Ganterbriicke, Beilage 13, Generelle Stat. Berechnung. These are handwritten calculations carried out by C. Menn, H. Rigendinger and their assistants. Based on handwriting 4 different contributors can be identified. The weight per length useis taken from. page 5. Ibid. Drawings for the box cross section at the pier are contained at the front of the calculation. In order to make the cross section symmetric the top (.25 m) and bottom (45 m) thicknesses were averaged to give the top and bottom thickness of 35 m. This is taken from Beilage 5 of the Vorprojekt for the Ganter Bridge cited in ref. 9. These are prints which show the prestressing cables, their tensions and locations. The actual plans show 6 cables stressed to 185 tons, 6 cables stressed to 145 tons and 14 cables stressed to 110 tons to give a total of 3520 tons. In the text the average tension is used for the 26 cables. The crushing strength for the concrete used in the Ganter Bridge isgiven as, ,,>40N/mm‘in the article Il Ponte Sulla Valle del Ganter, per la Nuova Strada del Sempione in Svizzera by C. Menn, H. Rigendinger and W. Maag in L’Industrin Italiana del Cemento Vol 6 p 509 (1982). These are taken from the same drawings cited in ref. 10. Since we are considering a point closer to midspan than that given as “im Feld” the concrete encasing of the cables is not shown. For this cross section our simplified cross section and the actual cross section are very similar. The tensions in the cables as well as their placement are taken from the plans cited in ref. 11. The tensions given are the final tensions. From the article Prestressed Stays Stiffen Spans which appeared in Engi- neering News Record p 68 Dec. 18 1980 one finds that the initial prestressing was 80% of the final. Thus the values used in the text are 80% of the values given in the plans. Given on page 5 of the ref. cited in ref. 1. The actual situation is somewhat more complicated than that presented in the text. The bending moment in the box at this point is kept to a minimal value by therelative stiffness of the concrete encasing and the box, which is about 2.4 to 1, the sequence of construction and the precise order and tensioning of the diagonal supporting cables. The FIG. on page 11 of the ref. cited in ref. 1 clearly shows a large fraction of the bending moment carried by the concrete encasing. At the end of this reference, after page 154 one finds a graph of the bending NEW FORMS IN LONG-SPAN CONCRETE BRIDGES Case moment for the entire bridge superstructure. This reveals a very low, less than 1500 ton-meter, bending moment at the pier side of the island. 17. Ibid. p20 Here we see that V_(K), the prestressing tension at time = o is given by 85 Vo(K) where Vo(K) is the prestressing at time = 0. 18. F. Leonhardt, Prestressed Concrete -Designand Constructionp65 Wilhelm Emst and Sohn, Berlin and Munich (1964). 19. Ch. Menn, Briickenbau I p 34 (1979). These are notes for Prof. Menn’s course given at the ETH - Ziirich. 20. Ch. Menn, "Aesthetics in Bridge Design", Bulletin of the International Association for Shell and Spatial Structures, Aug. 1985, pp 53-62. Problems (1) Suppose the dead load of the Felsenau Bridge is in- creased by 1 ton /meter. Find the increases in the upper and lower portions of the cross section of the deck at the pier. What fractional changes do these represent? (2) Suppose the live load of the Felsenau Bridgeis increased by 1 ton/meter. Carry out the same analysis as indicated in problem (1) for this change. 78 Part I LOS ANGELES FREEWAY BRIDGES A Structural Studies Problem Newton H. Copp The Claremont Colleges December 1990 PART III Los Angeles Freeway Bridges Table of Contents Chapter1 Introduction .......... Chapter 2 Bridge Loading . Chapter 3 Bending Moment Calculations Chapter 4 Prestressing Cables Chapter 5 Stresses 108 Acknowledgments ...... Problems...... Guide to Notations Endnotes Selected Bibliography 81 photo by Newton Copp Interchange between the Santa Monica and San Diego Freeways photo by Newton Copp Prestressed section of the Southeast Connector 82 Introduction CHAPTER 1 Introduction Los Angeles Freeways Ina free-association exercise, nearly everyone in Los Ange- les would link “freeways” with “frustration”. Yet Southern California’s freeways also include some of the most beautiful large structures in the state. In our frenzy to get somewhere quickly, we often fail to appreciate the striking visual appeal of the bridges thatmakea network out of individual freeways. The elegance of modern freeway interchanges gradually emerged from the efforts of Caltrans engineers to improve both the function and the appearance of interchanges. This structural study examines the techniques that allowed these two goals to be achieved. The freeways in Los Angeles County would be nearly useless without interchanges. These huge structures distribute automobile and truck traffic among as many as four freeways without requiring large reductions in traffic speed. Engineers have tried to ensure safe, rapid traffic flow at interchanges by designing bridges with a large radius of curvature. As these graceful arcs sweep diagonally at shallow angles over the underlying multi-lane freeways, however, they must span long distances. Supports for the bridge must be spaced far apart so they do not impede traffic passing underneath. Simple beams made of reinforced concrete or steel, which had been used successfully in early freeway interchanges with short spans, proved insufficiently strong or too costly for the long spans of modern, high speed freeways. New solutions to the problem of achieving long spans have not only improved the function of the interchanges in Los Angeles freeways, they have also led directly to more elegant structures. LOS ANGELES FREEWAY BRIDGES, Copp FIG.1. Map of the freeway system in and around the city of Los Angeles. The site of the Southeast Connector is circled. The Southeast Connector between the Santa Monica (10) and San Diego (405) freeways illustrates a particularly notewor- thy solution to the span problem. This interchange, completed in 1963, is interesting because it distributes traffic between two of the busiest freeways in Southern California (FIG. 1) and in- corporates the first attempt in Los Angeles county to utilize relatively new techniques to increase spans without high costs. Caltrans (California Department of Transportation) engineers combined the techniques of casting-in-place (in which the con- crete beam is formed in its final position) and prestressing (in 84 Introduction which high-strength steel serves in the beam as a tension member). They also abandoned the simple beam for a continu- ous beam design in which the ends of the beam are restrained by being connected to the beams in the adjacent spans. This design and mode of construction became the standard for all freeway interchanges constructed in Southern California after 1963. The purpose of this structural study is to examine how these construction and design decisions helped Caltrans engineers develop spans that are long and thin yet safe and inexpensive. New possibilities for design opened by these techniques will also be dis- cussed. The Southeast Connector The Southeast Connector routes westbound traffic from the Santa Monica Freeway to the southbound lanes of the San Di- ego Freeway (FIG. 2). It is two lanes (34 feet) wide, 2863.8 feet long, and 66 feet above the roadway at the highest point. It curves with a ra- dius of 1088 feet. Theentirestructure includes 23 spans FIG. 2. Expanded diagram of a portion of the inter- change between the Santa Monica and San Diego freeways showing the Southeast Connector. Traffic flows along the connector from the Santa Monica freeway to the San Diego freeway. The main section of interest in this study is marked by lines across the connector. 85 LOS ANGELES FREEWAY BRIDGES Copp A B Cc D | | ee eee | 137 ft 192 ft 137 ft FIG. 3 Schematic elevation of a prestressed section of the Southeast Connector. The section represented here is the section marked in Figure 2. and weighs about 28 million pounds, not counting the support- ing columns. Most sections of the connector are made of precast, hollow-box, reinforced-concrete girders. The longest distance spanned by this type of girder in the Southeast Connector is 131 feet. Two sections of the connector that pass diagonally over underlying freeways, however, are made of cast-in-place, pre- stressed (post-tensioned), hollow box, concrete girders. The larger of these sections is 466 feet long and contains the longest span used in a freeway interchange up to that time, 192 feet (FIG. 3). This section passes over the westbound lanes of the Santa Monica freeway. The 192 foot span was not only longer than spans in preceding interchanges but less deep; the uniform depth is seven feet giving a span/depth ratio of 27. We will focus attention on this 192 foot span and, in particular, the role of the prestressing cable in allowing sucha thin structure to bear large loads. Possibilities for design will be considered in the context of the structural analysis. 86 Bridge Loading CHAPTER 2 Bridge Loading Loads The Southeast Connector must carry a variety of loads. In addition to its own weight (i.e. the dead load), the bridge must support automobile and truck traffic. The traffic loads are divided into a uniform live load, a concentrated live load, and an impact load. Unlike the dead load which can be calculated precisely from the amount of material used to build the bridge, the loads created by traffic will vary widely and must be estimated. The uniform live load is an idealized value intended to estimate the load created by filling both lanes on the bridge with traffic. The concentrated live load estimates the load experienced by the bridge when two trucks stand side-by-side at the middle of the span. These two live loads are estimated under the assumption that the vehicles are standing still. A moving vehicle adds another kind of load in addition to its own weight, however, because it bounces up and down slightly as it passes over irregularities in theroad’s surface. Thisis theimpact load. The total estimated load created bya vehicleis thusitsown weight multiplied by a factor that estimates its impact. Guidelines for Load Estimates Estimates for loads follow guidelines described in Standard Specifications for Highway Bridges compiled by the American Association of State Highway and Transportation Officials (AASHTO). People in the Los Angeles basin pride themselves on the absence of snow, so snow and ice loads areignored in the following study. Loads imposed by earthquakes are presumed to act only horizontally (see AASHTO) and will not be included in the following calculations. Our main concern remains therole of the prestressing cable in helping the beam to carry the vertical loads. 87 LOS ANGELES FREEWAY BRIDGES. Copp Itis interesting that the AASHTO guidelines do not specify the maximum loads that a bridge might experience, only the loads that a bridge is likely to experience. The freeway inter- changes in Los Angeles would probably collapse if loaded with tanks, for example. It would have been too expensive to build these interchanges to hold bumper-to-bumper tanks, so they were designed to hold lighter, more likely loads. These bridges must be able to support tanks, however, because the interchanges are part of the national highway network. Incase ofsuchanemergency, someone will bestationed at the entrance to the bridge to limit the number of tanks on the structure at any one time and to enforce speed limits to reduce the impact load. 1. dead load, q, = cross sectional area of beam (see Fig. 16) x weight of concrete per cubic foot (note that 1 kip = 1,000 Ibs): = 58.7 f? x .15 kips/ft =8.8k/ft Adding the guard rail increases the dead load to: qq = 9.7 k/ft 2. uniformliveload, — q, = load per lane x number of lanes: = .64k/ft per lane x 2 lanes q, = 1.28 k/ft This value must be adjusted for the impact load where “I” gives the proportion by which the live load must be increased: 1=50/(1+ 125) note: 1= length of span = 50/(192 ft + 125) =.16 The uniform live load thus becomes: q, = 1.28 k/ft x 1.16 =149k/ft 88 Bridge Loading 3. concentrated live load, q, = idealized weight per truck x number lanes: = 18k/lane x 2 lanes = 36k This load must also be adjusted for impact as before: q, = 36k x 1.16 =41.8k Responses to the Loads If the Southeast Connector is to remain standing, then the external forces represented by the loads must be exactly op- posed by internal reactions of the structure. In turn, the materi- als of the structure must be able to bear the stresses created by the internal reactions. In this study, we are concerned with the tendency of the concrete beam to bend under the dead and live loads imposed upon it. The concentrated load, for example, is assumed to act at the center of the 192 foot span. This section of the beam will tend to bend downward under this load. The bending will create compressive forces in the top flange of the beam and tensile forces in the bottom flange of the beam. Concrete is an excellent building material because it goes through a fluid phase, during which it can be molded to any desired shape, before hardening into a material excellent at resisting compression. (The concrete used in the Southeast Connector was designed to have a compression strength of 4500 psi.) Unfortunately, the strength of concrete in compression is not matched by its strength in tension. Indeed, AASHTO in- structs engineers to assume that concrete has no capacity for resisting tension forces. This means that the 192 foot span of the Southeast Connector will collapse under heavy loads unless the design of the bridge provides some means for carrying tension. The steel prestressing cable and the continuous beam design serve this function in the Southeast Connector. We will calculate the forces acting on the bridge, the bridge's reactions to these forces and the stresses that result from the reactions in order to understand how the Southeast Connector works. 89 LOS ANGELES FREEWAY BRIDGES Copp CHAPTER 3 Bending Moment Calculations Simplifying the Assumptions The spans in the Southeast Connector can be imagined as a series of levers rotating about the supporting columns. The tendency of a lever, a concrete beam in this case, to rotate about its support is indicated by the bending moment created by the loads on the beam. Like the force developed by a lever, the magnitude of the bending moment is the product of the size of the load (i.e. force) and the distance (from the support) at which the load is applied to the beam. (Please refer to Part1 by Alfonso Albano for a more complete discussion of bending moments). Bending moment, M = force x distance The equations used in the following analysis are variations on this simple form. Bending moment values can be calculated in a beam not only at the support but at each point along the beam to produce a bending moment diagram. Bending mo- ments created by loads on the structure must be exactly op- posed by moment resistance within the structure or the struc- ture will break. Calculating the actual bending moments for each point of a 466 foot beam using all of the actual dimensions of the beam is tedious and, fortunately, unnecessary for the purposes of this structural study. The engineers who designed the Southeast Connector, of course, calculated bending mo- ments for a number of sections of the bridge. Three characteristics of the Southeast Connector simplify the calculations considerably. First, the total beam (466 ft) behaves as a single unit independent of its neighbors at either end from which it is separated by expansion joints. Second, the structure can be treated asa single I-beam although itis actually made of four I-beams fused side to side and cast as one piece. Third, the beam has a uniform cross-sectional area along its length. 90 Bending Moment Calculations The calculations are further simplified by making a number of assumptions: (i) The weight of the beam is assumed to be uniform along its length, a reasonable assumption given the uniform cross-section of the beam; (ii) The horizontal curvature and tilt (banking) of the beam are ignored; (iii) The ends of the 192 foot section are assumed to be at the same level; and (iv) Only half of the beam needs to be solved because the beam is symmetrical about the center line (midspan B-C). Other simpli- fications are discussed below with the calculations. Bending Moments Due to the Dead Load The first step in this analysis is to determine the bending moments resulting from each of the various loads on the Southeast Connector. A total bending moment diagram reflect- ing all loads on the 466 foot beam will then be constructed although the focus throughout will be on the middle of the 192 foot span. The analysis would be straightforward if the 192 foot section was a simple beam, but a glance at FIG. 3 shows that this is not the case. The 192 foot section is not perfectly free to rotate over the supports, as specified in the definition of a simple beam, becauseitis attached to the supports, and the 137 foot sidespans restrict movements of the center span. Neither does the 192 foot span completely qualify as a built- in beam: The side spans are too long to completely restrict movement of the ends of the center span. You can imagine that the center span would more closely approach a built-in condi- tion, if the side spans were much shorter. (For a more complete discussion of built-in beams, see the Felsenau Bridge study by William Case in Part I.) The behavior of the 192 foot span falls somewhere between that of a simple beam and a built-in beam. The problem of discovering where along this spectrum the 192 foot span falls is not one of determining the shape of the bending moment diagram because the shape remains a parabola for uniform 91 LOS ANGELES FREEWAY BRIDGES Copp loads (see FIGS. 11a-cin the Felsenau Bridge study). The problem is in determining where on the Y-axis of abending moment plot the curve should fall, i.e. determining values for the bending moment at the center and the “ends” of the 192 foot span. The bending moment at the middle of the 192 foot span would be large if that section was a simple beam. For simple beams, the bending momentata pointa distance x alongabeam canbe calculated according to the formula for a parabola (where x = 0 at one end of the beam): M, = 1/2 qx(1-x) At midspan of a simple beam, where x = 1/2, then; M, = qh?/8 A potentially confusing variety of bending moments are calculated below. In an attempt to reduce the confusion to a minimum, I have given each value a notation that uses sub- scripted letters to signify first the load or conditions under which the calculation was madeand second the span or support to which the value applies. A complete guide to notations is provided at the end of this structural study. 1, If the 192 foot span behaved completely as asimple beam, the bending moment at midspan due to the dead load, M. abe would be (FIG. 4): M,y,.= 9.7 k/ft x (192 ft)?/8 = 44,698 kft —_ 44,698 kft FIG. 4 Bending moment diagram for the 192 foot span assuming simple beam conditions. A bending moment this large would create a tension stress well in excess of 2,000 psi at midspan, and the bridge would fail. A simple beam of concrete clearly will not suffice to span 192 92 Bending Moment Calculations feet. Making the beam deeper would only increase the dead weight and exaggerate the problem. A shorter span would require either moresupports ora smaller radius of curvature for the bridge which would reduce the speed at which traffic could flow safely. The continuous beam design and prestressing provide the solution. 2. The continuous beam design helps the 192 foot span behave, in part, as if its ends were rigidly fixed as is the case in built-in beams. If the 192 foot span behaved completely as a built-in beam, the bending moment at midspan would be 1/3 of the value calculated under simple beam conditions, ie. ql?/24 (FIG. 5; see also the Felsenau Bridge study, FIG. 11a): Magy. = 9-7 K/ft x (192 £t)?/2* = 14,899 kft 14,899 kft FIG.5 Bending moment diagram for the 192 foot section assuming built-in (ie. fixed end) conditions. The effect of the built-in condition is to reduce the bending moment at the middle of the 192 foot span. 3. The actual value for the bending moment at the middle of the 192 foot span lies somewhere between 44,698 kft and 14,899 kft, but where? Given that the bending moment diagram for the dead load retains the same parabolic shape under simple and built-in beam conditions, the bending moment at midspan willbe the maximum possible moment (i.e. 44,698kft) minus the bending moment over the supports (FIG. 6). There are two sources for the bending moment over support B, for example: (i) The load on the 192 foot span creates a bend- ing moment because it behaves, at least in part, like a built-in beam with fixed ends; and (ii) the load on the side span creates 93, LOS ANGELES FREEWAY BRIDGES. Copp M FIG. 6 Representation of the difference between the bending moments in the 192 foot span under assumed simple beam conditions (lower curve) and under actual conditions that approach a built-in state (upper curve). a bending moment at support B because the side span behaves as a beam free to rotate at one end (A in this case) and partially fixed at the other (B in this case). The moment over support B (M,,,) that results from the load on the 192 foot span under the assumption that the ends are fixed is given as follows: M,,= qV'/12 (see Felsenau study, FIG. 11a) = 9.7 k/ft x (192 ft? /12 = 29,798 kft The moment over support B that results from the load on the sidespan assuming its endis fixed atsupport Band free torotate at support A is given as: M,,,= 42/8 = 9.7 k/ft x (137 ft)?/8 = 22,757 kt 4. We have now calculated two different moments for the same point! This can not be. The actual bending moment must lie somewhere between 29,798 and 22,757, and again the ques- tion is where. The span that exerts the greater influence on the bending moment over support B must be the span thatis stiffer. Given the uniformity of the cross-sectional dimensions along the beam, the shorter span is stiffer than the longer by a ratio of -68/.32 (see endnote 1). The difference between the two bending moments is mul- tiplied by this proportion (either .68 or .32) and the result is either added to the smaller of the two moments (if 0.68 was 94 Bending Moment Calculations used) or subtracted from the larger moment (if 0.32 was used). The distribution is carried out as follows where M, , is the actual bending moment over support B produced by the dead load: M,, = M,,.- (M,,.- M,,,) x 32) = 29,798 kft - ((29,798 kft - 22,757 kft) x .32) = 27,545 kft 5. The dead load thus creates a downward bending moment at midspan B-C (M,,,.) that equals the maximum moment cal- culated earlier minus the upward moment at support B, or: May. = Ggl/8 - My, = 44,698 kft - 27,545 kft = 17,153 kft This value is much lower than obtained under the simple beam conditions indicating that the continuous beam design signifi- cantly reduces the moment at midspan although it increases the moment at the supports. One way to visualize the effect of the continuous beam design is to imagine a slab of foam rubber laid across four supports. A load (force) on the side spans will tend to lift up a load on the middle span. If the side spans are stiffer than the middle span, their compensatory effect will be even greater. 6. The dead load moment at midspan A-B (M, ,,) can be calculated in similar fashion but using 1/2M, , because oneend of span AB has a bending moment of zero: = q,0/8 - /2M,, = (9.7 k/ft x (137 ft)?/8) - (0.5 x 27,545 kft) = 8,985 kft 7. The bending moment diagram for the dead load cannow be drawn (FIG. 7). Remember that in a parabola the 1/4 and 3/4 points have values 3/4M,,0r3/4 M, ,,. By convention, the moment that would produce downward deflection in the beam 95 LOS ANGELES FREEWAY BRIDGES. Copp DEAD LOAD A B c D -27,545 +8,985 +17,153 FIG. 7 Representation of the dead load on the entire 466 foot continuous beam (upper diagram) and the resulting bending moment diagram (lower diagram). is indicated by a “+” sign and the opposite momentis indicated by a “—" sign. Bending Moments Due to the Uniform Live Load A safe bridge design is obtained if the uniform live load is assumed to act only on the center span (span B-C). The calcula- tions proceed as above with only slight modification. 1. Because this load is distributed uniformly, the bending moment at midspan BC assuming simple beam conditions becomes: M,, = ql/8 = 149 k/ft x (192 ft? /8 = 6,866 kft Theactual bending momentat midspan BC is less than 6,866 by an amount equal to the bending moment at support B. 96 Bending Moment Calculations 2. As above, the bending moment at support B is found assuming fixed-end conditions for spans BC and AB. = q,l/12 = 1.49 k/ft x (192 ft?/12 = 4577 kft 3. Sidespan AB is not considered to contribute to the bending moment over support A because in the worst case scenario that we are examining thereis no live load over span A- B, ie. m,ab = 0. 4. The difference between these two momentsis distributed as before to obtain the actual bending moment (M, ,) produced at support B by the uniform live load: Muy = Mae (My My,,) x32) = 4,577 kft - (4571 kft - 0 kft) x 32) = 3,112 kft 5. Themoment at midspan BC due to the uniform live load can be calculated as: =M,,.-M,, = 6,866 kft - "312 kft = 3,754 kft 6. The bending moment at midspan A-B is 1/2M,,. To understand why, visualize the moment diagram that will result from the uniform live load only on span B-C. 7. The bending moment diagram for the uniform live load can now be drawn (FIG. 8). Bending Moments Due to the Concentrated Live Load Again assuming a worst case scenario, the concentrated live load is treated as if it acts only at a point in the middle of the 192 foot span. The resulting bending moment diagram will not bea parabola. Imagine the shape taken by a string weighted in the of. LOS ANGELES FREEWAY BRIDGES Copp UNIFORM LIVE LOAD A B Cc D 3,112 43,754 FIG. 8 Representation of the uniform live load acting only on span B- C (upper diagram) and the resulting bending moment diagram for the entire 466 foot continuous beam (lower diagram). middle and held between two hands (see the section by Alfonso Albano in this document). 1. The equation for calculating the maximum bending mo- ment atmidspan BC under simple beam conditions and a point load is: M,,.= a4 = 41.8k x 192ft/4 = 2,006 kft Asense of this equation can be gained by imagining support A “pushing up” on the end of the beam rather than the load pushing down on the center of the beam. In this imaginary case the force at midspan becomes ql/4 because the support “acts” with a force of q/2 over a distance of 1/2.) The bending moment diagram for these conditions de- scribes a “V” with the two ends of the “V” at zero moment and the trough of the “V” at 2,006 k. 2. As in the discussion of the bending moments resulting from the dead and uniform live loads, the bending moment diagram for the beam under built-in conditions will have the 98 Bending Moment Calculations same shape but be “lifted up”. M,.= 18 = 1,003 kft 3. The sidespan AB does not carry any concentrated load in the worst case scenario, so it does not contribute to the bending moment at support B, i.e. M,,, = 0 4. Distribution of these moments yields the actual bending moment over support B due to the concentrated live load: May = Maye Mige- My, 3) ¥ 0-32) = 1,003 k - ((1,008 - 0) x 0.32) = 682 kft 5. The bending moment (M,,.) atmidspan B-C due to the concentrated live load thus becomes: Mare = Myc" May = 2,006 k - 682 k = 1,324 kft and M,,,=V/2M,, = 341 kft 6. The bending moment diagram for the concentrated live load can now be drawn (FIG. 9). CONCENTRATED LIVE LOAD A B c Oo -682 $1,324 FIG. 9 Representation of the concentrated load (upper diagram) and the resulting bending moment diagram in the entire 466 foot continuous beam (lower diagram). 99 LOS ANGELES FREEWAY BRIDGES Copp Bending Moment Diagram Resulting From All Loads The moments calculated above may be added to yield values from which a summary bending moment diagram can be drawn (FIG. 10). Dead Load -27,545, +8,985 +17,153, Unit, Live Load 3,112 3,754 Cone. Live Load . -682 +1,324 sum 31,339 +7,088 +22,231 FIG. 10 Summation of the bending moments contributed by the dead, uniform live, and concentrated live loads to the 466 foot continuous beam. (All moments are in kft) 100 Prestrassing Cables CHAPTER 4 Prestressing Cables Action of the Prestressing Cable The total bending moment diagram (FIG. 10) represents the forces that must be opposed if the bridge is to stand. In this particular design, the concrete by itself is not sufficient to resist the tension stresses created at midspan by the various loads on the structure. Prestressing, in concert with the continuous beam design, allows the bridge to carry these loads without reverting to a form with shorter, deeper spans. Prestressing cables are bundles of steel wire with very high tensile strength; 270 ksi (kips per square inch) for the wire in the Southeast Connector. When these cables are placed in the concrete, stretched, and then released against the hardened concrete, the elastic recoil of the steel tends to pull the ends of the beam together and create a powerful compressive force within the concrete (FIG. 11a). If the cables follow a path that does not coincide with the centroid of the beam, as in the Southeast Connector, the elastic recoil of the cable also creates a bending moment in the beam (FIG. 11b). The cable thus contributes a fourth “load” to the structure (in addition to the dead load and the two live loads). The magnitude of this load depends both on the magnitude of tension developed in the cable by application of the jacking force (7,000 k in this case) and the distance between the cable and the centroid (see endnote 2). This distance is referred to as the eccentricity. The bending moment due to the cable (M_) is a product of the force (i.e. tension in the cable, P) and the distance (i.e. eccentricity, e). M. = Pe 101 LOS ANGELES FREEWAY BRIDGES Copp FIG.11 A Action of a prestressing cable (line with arrows in top diagram) on a beam when the cable lies along the centroid. When the stretched cable is secured against the ends of the beam (shaded boxes), the beam is uniformly compressed (lower diagram). FIG.11 B Action of aprestressing cable on abeam when the cable follows a path below the centroid. When the cable is stretched then secured, an upward bending momentis created (represented by the bend in the lower diagram). There are actually 32 separate prestressing tendons distrib- uted among four conduits in the Southeast Connector, but the calculations will be greatly simplified by assuming that all of these tendons behave as a single, large cable. The actual path taken by the cable through the concrete beam is rather complex (FIG. 12). The problem of finding the bending moments due to the cable becomes difficult under these conditions. 102 Prestressing Cables FIG. 12. Longitudinal section of one half of the 466 foot continuous beam. The canter lina (right side) lies at the middle of the 192 foot span. The resultant of the cables’ paths is shown by the curved line. The path meets the centroid at the left end of the diagram. Horizontal scale: 1" = 52’. Vertical scale: 1" = 22', Fortunately, the cable can be analyzed as if its path is the sum of two paths as diagrammed below (FIG. 13). The upper component of the cable’s path creates downward forces in the beam that must be exactly opposed by the columns supporting the beam. The beam does not bear any stresses created by this component of the cable’s path, and this component is ignored in subsequent calculations. This leaves the lower component of the cable's path to be analyzed. The elements of this path do not quite forma simple parabola in each span. The calculations of loads will be carried out initially, however, under the assumption that the curves in each span do form a parabola. The difference between the bending moments calculated under this assumption and the 1 = 2.36 ft 02 =2.14 ft — —Ht eee = Yes 63 = 3.03 ft 4/2, e2 e2 =2.14ft SSS el + 122 = 3.43ft e3 + e2 = 5.17 ft FIG. 13 Illustration of one half of the 466 foot continuous beam showing the cable's path (top) and the component paths (bottom). e = eccentricity. 103 LOS ANGELES FREEWAY BRIDGES. Copp actual bending moments will be used at the end to arrive at more realistic stresses in the beam. The engineer's plans note that the tension force in the cable isabout7,000 kat theends but decreases to 5,200 kat themiddle of the 192 foot span largely because of frictional losses along the cable’s path. The following calculations are simplified by as- suming, however, that the tension in the cable is uniform with a value of 5,200 k. 1. As before, the first step is to calculate the load. Having assumed that the load due to the cableis uniformly distributed along the beam and that the cable's path is parabolic, we can calculate the load created by tension in the cable, pr aS follows: Since M, = Pe = 4.1/8 then q, = 8Pe/l? 2. The load (q, ,,) created in span AB thus is given as: (Note: The quantity (e, + 1/2e,) is taken from the diagram of the cable path components and represents the eccentricity of the cable along its “new” component path at midspan AB.) pa» = 8 X 5200 k x (e, + 1/2 e, ft)/P = (8 x 5200 k x 3.43 ft)/(137 ft? = 7.6 k/ft Similarly for span BC: ore = 8 x 5200 k x (e, + e,ft)/P = (8 x 5200 k x 5.17 ft)/(192 ft)? = 5.83 k/ft 3. If span BC is assumed to be a simple beam under a uniform load from the cable of 5.8 k/ft, then the bending moment at midspan is: M,,.= q/8 = 5.83 k/ft x (192 ft?/8 = 26,865 kft 104 Prestressing Cables Similarly, the maximum bending moment at midspan AB will be: M,,, = 17/8 = 7.6 k/ft x (137 ft)?/8 = 17,831 kft 4. The actual bending moments at midspan BC and midspan AB are less than 26,865 kft and 17,831 kft respectively by an amount equal to the bending moment at support B determined as before by first calculating the moment (M,,.) that would exist at the support if span BC was built-in (i.e. with fixed ends) and then calculating the moment (M,,,) at the support due to span AB acting as a beam free to rotate at A and fixed at B. M P12 ibe = Appe = 5.83 k/ft x (192 ft)?/12 = 17,910 kft tab = Fpavl/8 = 7.6 k/ft x (137 ft)*/8 = 17,831 kft 5. The difference between the moments at support B created by these loads is very small making distribution of this difference nearly pointless. Nevertheless, when the difference is distributed as before, then the moment over support B because of the prestressing cable (M.,) becomes: M.,= (M,,,- (M,,.- M,,,) x 0.32) = 17,885 kft M, 6. The bending moment at midspan BC becomes: M.,.= M407 May = 26,865 kft - 17,885 kft = 8,980 kft 105 LOS ANGELES FREEWAY BRIDGES Copp 7. The bending moment at midspan AB is calculated slightly differently because there is no bending moment at one end (the cable comes to the centroid at the free end of span AB): M.4, =M,,,-1/2M,, = 17,831 kft - 8,943 kft = 8,888 kft When the cable is not assumed to follow a single parabolic curve in each span, the various bending moments take the following values (see endnote 3): M,,= -15,400 kft = +10,150 kft = +11,500 kft 8. The bending moment diagram for the cable can now be drawn (FIG. 14). ‘cab M ‘obe -8,888 -8,980 +17,885 FIG. 14 Summary diagram showing the cable’s path (curved line in top diagram) along the 486 foot continuous beam and the resulting bending moment diagram (bottom). 106 Prestressing Cables Total Bending Moment Diagram The total bending moment diagram for the Southeast Con- nector can be drawn from the sum of the bending moments contributed by the various loads (FIG. 15). Cable 8,888 -8,980 +17,885 -31,339 Total Load 422,231 Sum ~13,454 1,800 +13,251 Fig. 15 Summation of bending moments contributed by the cable and all other joads (‘Total Load’). 107 LOS ANGELES FREEWAY BRIDGES Copp CHAPTER 5. Stresses Center of Span The total bending moment diagram drawn above reveals a residual bending moment at the middle of the 192 foot span as well as at the supports. These bending moments will create stresses in the beam. The question now becomes whether the concrete from which the Southeast Connector is constructed can withstand the stresses created by the loads on it. Our main interest lies with the center of the 192 foot span because that is one spot where the bridge may fail. A simplified cross-section of the beam will be assumed for the purpose of calculating the stresses (FIG. 16). In this simplified version, the top and bottom flanges are treated as if they have equal areas of 18.5 ft? each with the total flange area remaining the same as in the actual beam. It is also assumed that the stresses are so concentrated in the top and bottom flanges of the beam that the intervening webs can be ignored. (See endnote 5 for calculations of stress completed without making these as- sumptions.) As described in the preceding studies, stress is the amount of force applied per unit area, ie.: Stress, f = force/area and force, F, is the bending moment, M, divided by the depth of the beam, d, from the center of the top flange to the center of the bottom flange (6.45 ft in this case): F=M/d Stresses resulting from the loads, the axial compression from the cable, and the bending moment contributed by the cable are determined separately then summed. 108 Stresses 2.67 ft 7H —————— Fig. 16 Actual (top) and simplified (bottom) cross sections of the 486 foot continuous beam. The total area in each case is 58.7 ft®. The actual areas of the flanges are 21.3 ft? (top) and 15.7 ft? (bottom). In the simplification, each flange is assumed to be 18.5 ft? 1. The dead load, uniform live load, and concentrated live load together contribute a bending moment, M, of 22,230 kft. The force created by this moment is: F = 22,230 kft/6.45 ft = 3,47 k This force creates a stress of: f = 3447 k/18.5 ft? = 186 k/f? That is 186 k/ft? of compression in the top flange and 186 k/ft? of tension in the bottom flange. 2. The elastic recoil of the prestressing cable against the concrete compresses the beam along its longitudinal axis. This compression can be treated as if it acts uniformly across the entire cross sectional area of the beam, 58.7 ft. 109 LOS ANGELES FREEWAY BRIDGES Copp f =5,200 k/58.7 f2 = 89k/f? 3. The bending moment created by the cable at the middle of the 192 foot span is 8,980 kft. The force that this generates is: F = 8,980 kft/6.45 ft = 1,392 k and the stress is: f= 1,392k/18.5 f =75k/f? or 75 k/ft of compression in the top flange and 75 k/ft? of tension in the bottom flange. 4. The magnitude and sign of the stresses can be visualized and summed as in FIG. 17. -186 -89 +75 -200 + + = +186 -89 -75 +22 Total Load S| sum Cable FIG. 17 Summation of stresses from all loads (i.e. Total Load), the approximate axial compression due to the cable (second from left), and the approximate bending moment of the cable. (+ indicates tension; - indicates compression). These calculated stresses may seem too large. The tension stress in the bottom flange appears especially worrisome. A tension of 22 k/ft translates into a stress of 153 psi. This value 110 Stresses exceeds the tension stress allowable in the concrete used in the Southeast Connector. The allowable tension stress, f,, is given as: f, allowable = 0.03 f, (where fis the maximum compressive strength of the con- crete) = 0.03 x 4,500 psi = 135 psi It appears from these calculations that the Southeast Con- nector does not satisfy the code because the stress in the bottom flange exceeds the amount allowed. Remember, however, that in order to determine the bending moments contributed by the prestressing cable, it was assumed that the cable described a parabola. This simplification led to an underestimate of the bending moment at midspan BC. If the more accurate value of 11,500 kft replaces the value of 8,980 kft in the calculations above (see endnote 3), then the stresses come out as follows: F = 11,500 kft/6.45 ft = 1,783 k f = 1,783 k/18.5 ft? = 96 k/f? The corrected stress diagram is presented in FIG. 18. The residual tension stress of 4 k/ft in the bottom flange translates into 28 psi, a value well below the allowable limit. AASHTO makes conservative estimates about how much tension is allowed in a prestressed concrete beam. The initially calculated tension of 153 psi in the Southeast Connector might lead to cracking in the concrete but the axial compression generated by the cable is sufficient to close these cracks and sustain the bridge. Ultimate Load and Safety Thebridge could havebeen designed from the total bending moment diagram shown above. Prestressed concrete bridges were, in fact, designed in this way in the United States largely 111 LOS ANGELES FREEWAY BRIDGES Copp -1 -89 +93 -182 + + = -89 -93 +4 +186 Total Load — Sum Cable FIG. 18 As in FIG. 17, but the stresses due to the bending moment created by the cable have been corrected as discussed in the text. because of the influence of Eugene Freyssinet, the French engi- neer who pioneered bridge construction with prestressed con- crete. Such designs reflect structures intended to receive only the loads as estimated in the calculations. What happens if some of the assumptions arenot correct? The concentrated load on the Southeast Connector may conceivably exceed the load created by two tractor trucks traveling side by side over the 192 foot span. The steel used in the cable may not quite meet specifica- tions for strength or diameter thus reducing the effects of prestressing. AASHTO now recommends that engineers design from more prudent estimates called the ultimate load. An ultimate load is first calculated for the dead and live loads: (the ultimate load can also be thought of as the ultimate bending moment, and so the symbol My is used) My = (1.4 x Mg) + (1.7 x Mp) where Mg is the dead load moment and Mj is the sum of the concentrated and uniform live load moments. The live load 112 Stresses moments are givena larger safety factor than the dead loads (1.7 vs 1.4) because they can be predicted with less certainty. My = (1.4x 17153 kft) + (1.7 x 5078 kft) = 32,647 kft An ultimate load or moment must also be calculated for the prestressing cable. In this case, the moment resulting from tension in the cable as calculated above must be reduced by a safety factor, 6, to allow for imperfections in the cable or its installation. OMy = O(Ag x fi)d where 0M, is the ultimate cable moment, Ag is the cross sec- tional area of the steel in the cable, f,is the maximum allowable tension stress in the cable, and dis the vertical distance between the center of the top flange and the centroid of the steel A, (6.17 ft). AASHTO specifies 6 as 0.9. The engineers that worked on the Southeast Connector calculated Ag by dividing the tension force in the cable by the maximum allowable working stress in the cable. Themaximum allowable working stress was specified in the plans as 60% of the tensile strength of the cable or: Max. stress = (.6)(270 ksi) = 162 ksi Thus Ags= 5200 k/162 ksi = 32.1 in2 The maximum allowable tension in the cable is specified by AASHTO as 80% of the tensile strength of the cable: fe =0.8x 270 ksi = 216 ksi 113 LOS ANGELES FREEWAY BRIDGES Copp Therefore: OM, =0.932.1 in2 x 216 ksi)6.17 ft = 38,502 kft @M, should always equal or exceed My. Indeed, 38,502 kft > 32,647 kft. Thus, enough cable and sufficient tension exists in the Southeast Connector to compensate for the dead and live loads with some room to spare. Thesmall residual tension left in the bottom flange can easily becarried by reinforcing bars; these were, in fact, includedin the structure. The estimated compression stress in the top flange can be translated into a value of 1263 psi. The allowable compressive stress, f, , is given as: f, allowable = 0.4 f', where f', is the compressive strength.of the concrete. Because 4500 psi concrete was used in the Southeast Con- nector: f, allowable = 0.4 x 4500 psi = 1800 psi The actual compression stress at the middle of the 192 foot span falls below this limit as does the compression stress at the supports. The concrete in the Southeast Connector appears tobe sufficient to bear the stresses induced in it. Efficiency The Southeast Connector is a safe bridge both in terms of its design and materials, but does it represent an efficient use of materials? The stresses in the bottom flange are near zero, so it makes little sense to calculate efficiency for that part of the structure. Regarding the top flange, however, the concrete has the capacity to bear a compression stress of 1800 psi, yet the 114 Stresses maximum compression stress at midspan BC will be approxi- mately 1263 psi. This gives an efficiency of: efficiency in compression = 1263/1800 =.70 It may seem from this number that too much concrete was placed in the bridge, at least at midspan BC. A look back at the total moment diagram will reveal, however, that the bending moments over the supports are slightly greater than atmidspan BC. The uniform depth of the beam may have been chosen to accommodate these larger stresses. 115 LOS ANGELES FREEWAY BRIDGES. Copp ACKNOWLEDGEMENTS. This study was completed as part of a seminar con- ducted at Princeton University from 1987 through 1989. The seminar, entitled The Engineer’s Experience and The New Liberal Arts, was sponsored by the Alfred P. Sloan Foundation. I am especially grateful to David P. Billington, Professor of Engineering at Princeton University, for his invaluable assis- tance in preparing this study and to Paul Askelson, Bridge Maintenance Engineer for the Department of Transportation of the State of California, for supplying essential information about the freeways of Southern California. PROBLEMS 1. The 466 foot section of the Southeast Connector has the same cross sectional area along its entire length. That is, the same amount of concrete was used at each point along the beam although some regions require less than others. Use the total bending moment diagram as the basis for proposing at least two alternative designs to the one used by Caltrans for the 466 foot section of the Southeast Connector. 2. a, What path would the prestressing cable have to take if the 192 foot span of the bridge described above was a simply supported beam rather than part of a continuous beam? Assume that all loads remain the same as described above. b. How large a prestressing force would be required to counteract the moments created by the loads on this simply supported beam? What diameter should the cable be to safely bear this new prestressing force? 116 Guide to Notations GUIDE TO NOTATIONS Loads — k (kip = 1000 pounds) q, = load due to prestressing cable .a)= load at midspan AB due to cable ey = load at midspan BC due to cable q,= uniform dead load q.= concentrated live load P = prestressing force q, = uniform live load Moments — kft (kip-feet) M, = moment created by cable M.,, = moment created by cable at midspan AB M,, = moment over support B due to cable M.,,, = moment created by cable at midspan BC M, , = Moment at midspan AB due to concentrated live load M,,, = Moment over support B due to concentrated live load Ma, = moment at midspan BC due to concentrated live load M, » = Moment at midspan AB due to dead load M,,, = Moment over support B due to dead load M,,, = moment at midspan BC due to dead load M,,, = “fixed end moment” at midspan AB due to load on span AB M,,. = “fixed end moment” at midspan BC due to load on span BC M, = ultimate bending moment M, » = moment at midspan AB due to uniform live load M,,, = moment at support B due to uniform live load M,,,. = moment at midspan BC due to uniform live load 117 LOS ANGELES FREEWAY BRIDGES Copp ENDNOTES 1. The distribution factors are calculated as the ratios of stiffness 118 (K) of the two sections, AB and BC. Kap = 3EIA, and Kp, = 2EIA, where: E = modulus of elasticity (the material stiffness in psi) I = moment of inertia (geometric stiffness in ft) 1, = length of span AB 1, = length of span BC But Eand Ihave the same values for the two sections of the beam, so in determining the ratio of the stiffnesses, these values drop out of consideration. Therefore; Kap = 3/137 ft and Ky. = 2/192 ft The distribution factor for AB = K,,/(Kap + Kc) = 68 The distribution factor for BC = Kye /(Kab + Kye) = .32 In Kap, the factor 3 accounts for the simply supported end, whereas the factor 2 in K,, accounts for the sym- metry of the continuous beam and its load so that Mpc = Mo, that is, the distribution produces a single value for the moment at the support. Endnotes 2. The centroid is determined according to the distribution of the mass of the concrete beam in cross section. Ax = area of (1) upper flange and railing (2) support (or web) (3) lower flange Y = distance from top of beam to center of (1) upper flange (2) support (3) lower flange Centroid, C = (A x Y)/Ay Ax Y AxY 214 315 675 2217 35 75.95 3143 6.67 95.38 totals 57.4 178.1 C=31ft 3. Because the actual cable profile is positioned above the simplified profile (i.e. the eccentricity is actually smaller than assumed in the simplification), the resulting mo- ment over the support is less than the calculated value of 17,850 kft. The correct fixed-end moment is found by calculating the product integral of the actual tendon profile. For copies of these calculations (performed by David Billington), contact Newton Copp. 4, The section modulus can be calculated for the portions of the beam above the centroid and the portion of the beam below the centroid as follows. Dimensions refer to the diagram A. 119 LOS ANGELES FREEWAY BRIDGES Copp y= bier3 _ (bi by) 23 : bac33 _ (b2_bw) c43 3 3 3 3 = 423.7 ft gs, = 1 3423.7 ~ 1367 £3 Cr ot sy =1 = 423.7 = 108.6 £3 C3 9 5. Stresses in the beam are more accurately calculated using the section modulus instead of thearea of the flange because the top and bottom flanges do not have equal area and the centroid is not in the middle of the beam. P/A = the axial compression stress created by the ten- sion in the cable (P is the tension force in the cable, 5200 k; A is the cross sectional area of the beam, 58.7 ft2) Pe/S =the stress generated by the bending momentdue to the cable (Pe; see calculations on pages 101-105) acting over the section modulus, S. Thesection modulus is calculated as the moment of inertia of the beam 120 Endnotes divided by the distance from the top of the beam to the centroid, and its units are given in ft3 (see endnote 4). Sy is the section modulus for the top flange (136.7 ft3) and Sa is the section modulus for the bottom flange (108.6 ft). MUS = sum of Mg pc/S, Mube/S, and Mei,pc/S where these values indicate stresses created by the dead load, uniform live load, and concentrated live load, respec- tively. Incalculating the total stress in the top flange, itmustbe remembered which of thestresses are from compression (and thus given a “~” sign) and those that represent tension (indicated by a “+” sign). Therefore: ftop = - P/A + Pe/Si- Mt/Si = —5200/58.7 + 8980/136.7 - 22231/136.7 = — 186 k/ft2 The total stress in the bottom flange is calculated in the same way except that the stresses due to the total load moments now are tension whereas the stress caused by the bending moment contributed by the cable is now compression: foot = — P/A - Pe/S2 + Mt/S2 =—5200/58.7 — 8980/108.6 + 22231/108.6 = +33 k/ft2 These stresses, 186 k/ft2 of compression at the top and 33 k/ft2 of tension at the bottom, must grade into each other through the web of the beam. A stress diagram can be drawn to represent this situation. Diagram B shows the two versions, the bottom one has been cor- rected for the cable as described in the text. 121 LOS ANGELES FREEWAY BRIDGES Copp EEE ar = momen “MC -186. Cio > T 433 Ii ps “AC - 169 ST — | Diagram B — Bottom stress diagram has been corrected as described in text SELECTED BIBLIOGRAPHY Beaufait, F. W. 1977. Basic Concepts of Structural Analysis. Prentice-Hall, Inc., NJ. Cerny, L. 1981. Elementary Statics and Strength of Materials. McGraw-Hill, NY. Ghali, A. and A. M. Neville. 1977. Structural Analysis: A Unified Classical and Matrix Approach (2nd ed.). Chapman and Hall, London. Gordon, J. E. 1978. Structures: or Why Structures Don’t Fall Down, Plenum Press, NY. 122

You might also like