This action might not be possible to undo. Are you sure you want to continue?
BY COULOMB FRICTION by
O. KeightleyI SYNOPSIS
Infill wall panels made of concrete segments prestressed together are presented as a nondestructive means of consuming vibrational energy in structures. The panels conform to building frame distortion like a deck of cards distorting in shear, consuming energy through Cbulomb friction along the slip surfaces. Interfloor Coulomb forces are determined for two buildings to approximate maximum response during the EI Centro earthquake of the same buildings with 20% viscous damping. Coulomb forces chosen for lowest stories are 6%-8% of a structure's weight. Residual displacements after the earthquake are small, but extra loading on frame members at joints is significant. INTRODUCTION Building vibration amplitudes due to wind or earthquake can be significantly reduced if a large part of the kinetic energy of the structure relative to the ground can be dissipated in each cycle as the building vibrates. A building with 20% critical viscous damping, which dissipates 92% of its energy in one cycle of free vibration, might experience earthquake induced vibrations only 1/2 as large as one with 2 % damping, which dissipates only 22% of its energy per cycle. Unless some type of dissipation mechanism is present, during a strong earthquake amplitudes increase to the point where the material of the structure becomes nonlinear, and energy is dissipated through processes which gradually destroy the structural elements. This paper presents the concept of non,-loadbearing prestressed COD'ft' crete wall panels to dissipate energy through CoUlomb friction, while the elements of the structural frame remain linear during the design earthquake. A technique for digitally computing the response of Coulomb damped multistory buildings is outlined and response maxima of a 4-story and an 8-story Coulomb damped structure during the El Centro earthquake are compared with maxima of 20% viscous damped structures. PRESTRESSED SEGMENTED WALLS A segmented prestressed wall is shOwn activated by building frame distortion in Figure l(a}. Figure l(b) shows the forces required for activation. Theoretically the lateral force to cause slipping of a panel is FC
x PST j.l hll '
where j.l is the coefficient of friction, PST is the prestressing force in the panel, and h and I are panel dimensions. The panels should fit snugly at all times within the frame opening, touching only at the corners, but not be subject to vertical load other than the force PST. Figure l(c)
Professor of Civil Engineering and Engineering University, Bozeman, Montana, USA.
~hows how the panels cause extra load in the frame. for :L. FP. x 1 (y + U. For the first iteration the forces in the panels at the end of the interval were assumed equal to the forces at the start of the interval (FPT = FP). tending the'floor below was defined as: FA. was necessary to yield maximum responses wi thin 2 % of maxima computed by the exact method. the actual friction force FPTin a panel was set equal to FC if IFAT I ?. then reduced to one degree of freedom for each floor. caused' by the distorted shape of the framework (the framework was assumed e1as. TO model this phenomenon. 1 is the horizontal force in the 1+ panel above the floor./ Coulomb friction is a passive phenomenon not linearly related to relative displacement or velocity. 1) 11 • where FS.Loor over the interval was less than the force FC necessary to cause slipping.l8. To decide on the size of the time step and also to check the step-by-step code. signs corresponding to the direction I I 3044 . y is ground acceleration. an active force FA.M. For the undamped 8story building under El Centro excitation. it was decided that during the interval the ith floor was locked to the floor below. . but if they are at relative rest FP depends on the magnitude and sign of the active force tending to cause relative.• DIGITAL COMPUTING TECHNIQUE 'l':> compute the time history of the behavior of buildings with Coulomb dampers. The actual friction force FP can assume any value between positive FC and negative FC.motion. 1 is acceleration 1- of the floor below. Upon reversal of relJative velocity the forces of the panels on the frame act along the opposite diagonal . a step-by-step technique'based on an assumed linear variation of acceleration across the time step was used (1). FC. therefore with Coulomb friction it was necessary to iterate across the time steps as described below. or the relative velocity changed from zero to a nonzero value.l to FAT if FAT ~FC. but no axial deformations of beams or columns. and U. This decision was based on the passive nature of Coulomb friction: an active force FA less in absolute value than FC cannot reverse a relative velocity nor start from rest a mass already at rest relative to the mass below. or wasseteq1. If the floors were found to be locked together at the end of an interval. and if the computation showed that during the interval the velocityof the floor changed sign relative to the floor below.\'is the spring force on floor mass M. 1 1 to move the ith floor relative to = FS. The stiffness matrix was formed for each framework by permitting joint displacements and rotations. The techn. Masses were lumped at the floors. the panel below the ith floor.ique requires a knowledge of all forces acting on the masses other than those linearly related to displacement or velocity. The force FP e~als FC times the sign of the relative velocity of the masses.tic and the panels rigid except on the slip surface). + FP'+ 1 1l . An active force was computed for each floor at the beginning (FA) and at the end (FAT) of each time step. all quantities positive to the right.1. computed responses of undamped structures were compared with responses computed by an exact method (2). and can in theory change instantly from one value to another.005 sec. If the average active force on a f. a time step as small as 0.
A more correct computational scheme would determine the exact instants. of course.of impending relative motion.:un earthquake responses at time steps one half as large changed in most cases by less than one percent.. to the sum of the masSes aBove it so that under increasing ground aocelerati. the values after 10 trials were accepted. At 0.on all panels would start to slip at the same instant. The Cou1oab . An energy balance.. Response of the buildings to the BJ.002 Fe from the previous iteration. Because the values of the FPT forces at the end of each time step must be known but usually cannot be predicted. and the nature of the ground excitation. TWO tests performed on the accuracy of the computational scheme used were 1) to lock all but one story with large Fe forces and to observe the decay of free vibrations. and 2) Coulomb damping of the proper magnitude to yield approximately the same ma. using lI¥:>dal superposition. approximately 10% of the oa1cu1ations did not converge. Beam shears at the joints are also increased significantly by the presence of the panels. The top 4 stories were also studied separately as a 4story building.. at which relative velocities of adjacent masses changed sign or changed between zero and nonzero values. the maximumresidual displacement is 1. and at 0.xiutum. Maximum top stOry shears (sum of shears in the columns only) and maximum accelerations in the Coulomb damped bui1dings are somewhat greater than in the viscous damped huildings. 3045 .1 inches.5 inch.1 inch. it is necessary to iterate.. 005 sec. If no FPT force changed by more than 0. With the 8-story structure •. using the digital technique described. the natural periods of the building. COULOMB AND VISCOUS DAMPING COMPARED Figure (2) shows 8 stories of a building which has appeared in the literature (3). within the time steps. Free vibrational decay of the SDP system agreed well with theory. This was not attempted here. column shear at the beam-column joint is increased by about 75%. amplitudes.0025 sec. maximum possible residual displacement with the Coulomb dampers shown is 0. approximately 4% did not converge. approximately 2 % did not converge. using the previously determined FPT forces as the basis for the next trial.for·ce PC ill each story was arhitraril. Accelerations at the end of the interval were computed from Newton's second law of motion. and 2) to recompute earthquake response with a reduced time step. at amplitudes larger than the design amplitude Coulombdampers appear less effective. Residual displacement of the top floor of the 4-story frame after 18 seconds of earthquake was 0. in which the difference between running totals of energy input and energy dissipated was compared with the sum of instantaneous kinetic and e1astically stored energy. time step. The fraction of calculations not converging depends on the size of the time step. which means that if the panel is located in the center bay. he used to reduce these maxima.00125 sec. Because Coulombdampers absorb energy only as the first power of vibration amplitude (viscous dampers absorb energy as the square of the amplitude). Centro earthquake was computed with I} 20%viscous damping in all modes. but if convergence was not reached in 10 trials. was satisfactory.y made· proportional. the calculation across an interval was accepted. In the 8-story frame the Coulomb force in the ground story is about 50%of the maximum story shear. For the 8-story frame. Other combinations of Coulombdamper values can. and maxinr.
1968).. 3. Pasadena.C. the digital computer work. New Jersey (1970). opposes relative velocity of the floors. Calif. Seis. e.F. N. When ground excitation was doubled in intensity. Digital Calculation of Response Spectra. Barnett. The force in the bracing. Tech. BIBLIOGRAPHY assisted with 1. Calif.and at smaller amplitudes more effective. Englewood Cliffs. 58. crossed diagonal braces containing Coulomb slip joints? These questions cannot be answered by analytical or digital computer studies alone.. "Maximum Response Ranges of Nonlinear Multi-Story Structures Subjected to Earthquakes. M.tures did not begin to move until more than one second of the earthquake had occurred. N. (1968). Society America. Earthquake Symposium. ACKNOWLEDGMENT David L. editor). "Earthquake Response of Structures. Nigam. Further investigation seems justified. Practical questions which must be ans.C. R. and P. Greater effectiveness at smaller amplitudes is shown by the fact that the Coulomb damped struc.wered regarding this technique of energy dissipation include: Can the prestressing force be maintained constant over the years and during an earthquake? Will a soft spring be necessary in series with the prestressing tendons? Will the coefficient of friction remain reasonably constant during an earthquake? Can the required close fit of the panels into the frame openings be attained and maintained? How can the extra shear forces at the joints of beams and columns be economically resisted? Is the entire concept economical when compared with other methods of design? Are there other forms of dissipation mechanisms more practical. pp 1639-55 (Oct.S. W. corresponding to about 5% viscous damping. 3046 .. CONCLUSIONS The prestressed wall panels described here act like bracing across one diagonal of the openings in a building frame. Research Lab.. student in Civil Engineering." in Earthquake Engineering (R. 5. Inst. V. an M. Jennings.. Wiegel. Prentice-Hall. response of the Coulomb damped structures roughly tripled. but only by constructing panels within frames and testing them." Bull. not relative displacement. 2.g. Digital computer studies have so far indicated that not unreasonably large Coulomb forces are required to limit maximum responses during a strong earthquake to the same maxima achieved with 20% viscous damping. Clough. limited by the Coulomb slip force. Giberson.
...:) 0 ~ u e.::l .. _J _J z: <: <: 0 LU W 0 u LL ..:) Po. ._ en _J c:o z: I- 0 ~ ...... r-I + H 4-10 or:: ~ u "0 .... . Cf... r:: <ll H OOQi -.... CO '-' <ll .0 :......o "0 r::"O r:: 0 H 0 C) c .." ~ 4-! ""t) Q)... CIl S H 0.... '-" .047 ..l -_ ~ U ~~~~ Po.0 _.!) 0 1- :::c ......jJ'-" H Q....c ~ r-I >t!) W _J a::: H 0... r-I O..jJ 0 s::: CIl .0 ~ U c.....-I CIl CIl..::l C)..c ~ :< U r-I r-I Z LU :..:: <ll4-1 CO 0.-I :< u S r-IH CO <ll CL c:: en en LU a::: 1en w <: z: 0 e:( _J :.a ..> a::: ~ c:: :::c ~ 0 r.:: a::: LU z: ~ j:Q -.- '~l '-' C) 3..... :..-1 .:...o A-S::N ..-1N .) <ll § S .0 <ll "0 H 0 H CO CO Qi ..~! r-I ..:) IJ.. :< I en z: 0 u 0 I- t!) en 0 Z _J ~.-1 <ll .
.-i -.. "c Cf...::t....-j 00 rz.:I '" fil 0 CIl til ........ 0 ~ C"'l r:... ~ -'-- i 000 r-i 0 0 ...:l .!l Z N H Q H 00 H ::::> f>:..< 0 ...o 0'" o.-j 0 .....-i IX: · r-i Cd Po.. rl r-. ""0 eo a- C .. C"') -< N CIl CIl .... 1>...... / .....-i r:r....0 f~~ 0 0'" ~ ~ ~ ~ ~ ~ 0 o-l 0 0 ::::> '''' N 0 0 u \0 t~ C"'l \0 0 r:r. r:... ::t:A 00 C""i r-...:lr-i ct! I-i (!) fil bObO ~ 0 til .. C C"'l ! r-i N CD +J 00 H (I) 0 fil 0 (I) r:: (J Lt"l rl ~ ~ c.. ..tOO I. ct! .-i N I ..... .048 .. 0 al (I) r:: 00 \0 Lt"l r- CIl CIl (J (J f>:. alo..:I -10) (J ::l "c Cf.. ........-i i o ....< ~~ -10)-10) .-i C1l bO.t r:r...... ~ >-0 . a- X Cd ct!"c 'H • ....l rz. CIl H I>.-i C0 :> 0 .-i 0 0 C1J C (!) (J . r: CIl -10) . O.. 0 ~ /(j .. 0 ....1>...-i r-i 0 H •..l+J CIl H Nr-l 0 0 0 0 0 0 r-i .::t N ~ ~ ~ i-- ~ ~ N N 0 ~ 0 -e 0 0 C"'l ~ 0 C"'l N 0 \0 0 C"'l 0 ~ 00 0 r:r.::t -... ..f!::..-i.-i U U 0 ::::> u I>.-i I ::l -.-i .. (I) (J -.-i r-i al Ul '" (!) 0 ......-i I +J o +J CIl 'H H c. 0 f>:.-i r-i ct! ...:/f:} ~ . It''\ ..-i 0 0 . 0 r:r.-i \ \ \ C ..< .....l I ::t:'-' • r-i al al"c X r-l r-l ..... Lt"l ~ ~ H ::s • .:l o .\ \ I' eo \ 0 ... 0 t ~ \0 t -e0 It''\ t ~ .-i "'''' . . t .... ~ tI.._-. . ><: ~t:l C1l al .......-... -..l Ol al ..l H CO ::> Q ~CIl ~ o. (J ..... Lt"l rl 0 N r-l r-i (I) (J (J ~ >: Cd • <: ~<: N ~ 00 r-l tI..l -e C1l C1J "c Ol ~ 0 a:> a(!) \O~ N ~ .. ..a- al .. -10) tI. I ~ Q ~ ~ :> .....::t r-i t:: ... \0 - 00..'M -Io)...... ....0 CIl ~ \0 U N Qj -10) ~ Cf...-i t:: 0 +Jr-i Ulo. • r-i ~ 0 .......::l >< . \ ..:l (J r:.. / It''\ rl CD 0 ~ ::. al ..... -10) I>..t 0 C"'l N \C " 3.!l Z 1-1 Cf....... ~ ~ CIl ~ r-l C""i 0 Z 0 Ul 0 til H ..-i 0 o 0 N • It''\ .l ~ .-i I r-i It''\ . . .
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.