Click for next page ( 23


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 22
20 This detail significantly improved the behavior of the con- of these gages was low. Often, only one or two gages sur- nection. The connection lasted for 133,000 cycles before fail- vived to failure. However, the gages that did survive suggest ing. This connection was also stiffer than the others tested that bar strains were lower in the embedded specimens than (see the finite element modeling analysis section, this chap- in the nonembedded specimen. ter). After the bars in the positive moment area fractured, the specimen was still able to hold the applied load of 32 kips with an end deflection of 1.2 in. FULL-SIZE SPECIMENS Failure was due to fracture of the bars in the positive moment connection and by the girders pulling out of the Description of the Specimens diaphragm. At failure, there was significant spalling of the dia- phragm concrete. After the test was complete, the spalled The full-size specimens were constructed of two 50-ft diaphragm concrete was broken out so the beams could be Type III AASHTO I girders joined with a 10-in. diaphragm. inspected. Cracking was found in the webs (see Figure 25), so An 8-ft-wide 7.5-in. thick composite concrete deck slab while this connection performed better than any other, the was cast on top (see Figures 26 and 27). As with the stub cracking in the beams at failure might be an undesirable result. specimens, the slab was not cast all the way to the end (see Figure 26). The girders were to be reused by turning them end-for-end and creating a new specimen. The last 12 ft of The Effect of Embedment the deck slab was not cast to create the second specimen without having to remove any of the deck slab. For reference, Four of the six specimens had the girder end embedded into the cardinal compass directions are shown in Figure 26. The the diaphragm. For the bent-strand connection, embedment girders are designated as "east" and "west." The girder faces seemed beneficial. The number of cycles to failure increased are "north" and "south." A discussion of the experimental set by a factor of three. A change in failure mode was observed. up and instrumentation is included in Appendix B. In the nonembedded bent-strand specimen, the girders sepa- Each end of the girder specimens was cast with both bent rated from the face of the diaphragm, but there was no dam- bar and extended strands so that either end could be used for age to the face of the diaphragm. In the embedded specimen, either type of connection (see Figure 28). Unlike the Type II diagonal cracks indicating a pull-out type of failure were girders used for the stub specimens, it was possible to install observed. some of the bars pre-bent and still close the forms; however, For the bent-bar specimens, the results were less conclu- some of the bars had to be installed straight and then field sive. The embedded bent-bar specimen (Specimen 4) failed bent later. at less than half the cycles of the nonembedded specimen A constructability issue arose during the fabrication of the (Specimen 2), but another embedded bent-bar specimen girders. When the strands were detensioned, some or all of (Specimen 5) failed at twice the number of cycles as the non- the wires in some strands unwrapped and deformed, creating embedded specimen. Bonded strain gages were placed on a "bird cage" effect, as is visible on most strands shown in four of the bent bars in each specimen, but the survival rate Figure 28. The strands were rewrapped, but some of the bird- caging remained. This occurred approximately 6 to 8 inches from the face of the girders, just about at the point where the strand would be bent. Since there is a possibility that this sit- uation might occur in real field applications, it was decided Diaphragm to proceed with using these strands to see whether there was an effect on connection strength. W S Girder N E Figure 25. Cracking of the girder in Specimen 6. Figure 26. Full-size specimen.

OCR for page 22
21 Figure 27. Cross section of full-size specimen. Full-Size Specimen 1 placed as symmetrically as possible, but still allowed for mesh- ing of the bars. This configuration was still slightly asym- Behavior During Construction metrical and resulted in a very congested diaphragm area. Tolerances in bending rebar result in bends that do not line The first specimen used a bent-bar configuration in the up (see Figure 30). This creates problems in inserting corner positive moment connection. As with the stubs, the connec- bars as detailed in Figure 29. A similar problem occurred tion was designed to provide a capacity of 1.2 Mcr. This with the bent-bar stub specimens as shown in Figure 16. required eight No. 5 hooked bars (see Figures 27 and 29). The first specimen used a partial diaphragm pour (see Fig- Since the stub specimen tests indicated that asymmetrical ure 29). This configuration is used by several states. In this connections were undesirable, the bars in the connection were connection, the bottom of the diaphragm is poured well before the deck slab is added. The idea of the partial dia- phragm is that the weight of the deck slab concrete will rotate the end of the girder into the partial diaphragm and compress Strand that has not the concrete. Then, any tension caused by positive moments been rewrapped will simply reduce the compression rather than cause tensile cracking. The partial diaphragm was poured when the gird- ers were 28 days old. According to the survey, the depth of the diaphragm pour is usually one-third to one-half of the Strand that has been rewrapped diaphragm depth. In this case, the diaphragm was poured to depth of 19 in. This was determined as the minimum depth that would allow the tails of the bent bars to be completely covered with the initial pour. Figure 31 shows the variation of the east-end reaction with time. The west-end reaction data are almost identical. From the time the partial diaphragm is cast to the time the slab is added 28 days later, the end reactions increase approx- imately 3 kips, creating a positive moment at the diaphragm Figure 28. End of the girders for the full-size specimens. of 150 k-ft. This positive moment is caused by a combination

OCR for page 22
22 Figure 29. Bent bar specimen with partial diaphragm. of creep, shrinkage, and average temperature effects in the slab, tensile strains ranging from 60 to 90 microstrain develop girders. The effect of daily temperature variation is seen as in the diaphragm. The variation seen is due to temperature. the waves in the graph. Prior to adding the slab, this effect of After the deck slab is added, this strain gage at the bottom daily temperature variation is approximately 1k. of the diaphragm shows an immediate increase in compres- The deck slab was added 28 days later. Figure 32 shows sive strain of 40 microstrain, which would correspond to a the results of a vibrating wire strain gage placed at the level stress of approximately 160 psi. Vibrating wire strain gages of the positive moment connection bar, in the middle of the placed 6 in. from ends of the girders (at the level of the bot- diaphragm (see Figure 29). The graph shows the time period tom row of strand) show an increased compressive strain of when the deck slab was cast to a few days after. During the 50 microstrain, which corresponds to a stress of approximately time from casting the partial diaphragm to adding the deck 250 psi. Therefore, some compressive stress develops at the joint. It is important to remember that the stresses cited are the instantaneous stressed caused by pouring the deck slab. At the time the deck slab was cast, the diaphragm was in ten- sion because of creep, shrinkage, and temperature effects in the girders. The compressive stresses created when the deck slab was poured were not enough to overcome this tension, and the diaphragm still shows a net tension even after the deck slab is added. After the deck slab is poured, the instruments show an unex- pected response. A few hours after the pour, the strain in the diaphragm increases (becomes more tensile) then decreases (becomes more compressive) over the next 2 days. After 2 days, the net strain is compressive. This is caused by thermal effects. In Figure 32, the second curve shows the temperature in the deck slab. The strains in the diaphragm are constant for a few hours, and then they follow the deck slab temperature curve. A similar response is seen in the reactions. Figure 33 shows Figure 30. Misaligned bars due to field bending. the east-end reaction at the time the deck slab is cast until

OCR for page 22
23 35000 All Forms Removed Pour Diaphragm 30000 25000 Reaction (pounds) 20000 15000 10000 5000 Pour Slab 0 0 20 40 60 80 100 120 140 160 180 200 Time - Days from beam fabrication Figure 31. Variation of east end reaction with time. 2 days after. There is an immediate increase in load of about expansion. The data show the girders camber upward 0.01 in. 10 kips when the deck slab is poured. This is consistent with Since the partial diaphragm provides some connectivity, the the girders still behaving as simple spans, so the partial dia- entire system appears to deflect upward with the girders. This phragm does not provide continuity. Over the next 8 h, the causes the tension at the diaphragm and the increase in the end loads increase an additional 4 kips. The peak load corre- end reactions. sponds to peak concrete temperature in the deck slab. During The final set of the concrete--the point at which there are the next 2 days, the end supports lose almost all the load measurable mechanical properties in the concrete--is usu- gained during the deck slab pour. The loss of load mimics the ally taken as just before the point at which the heat of hydra- deck slab temperature graph. tion graph peaks. At this point, the hardened deck slab begins These responses are caused by the heat of hydration in the to influence the system. When the deck slab begins to cool, deck slab. After the concrete is poured, the deck slab con- it will contract. Since the concrete and the reinforcing steel crete begins to heat up as the chemical reaction progresses. have almost the same coefficient of thermal expansion, the This will also heat up the top flange of the girder, causing an reinforcing steel expands and contracts with the concrete and 100 50 80 45 Slab Temperature 60 40 Temperature Degree C 40 35 Microstrain 20 30 0 25 -20 20 Cast Slab -40 15 -60 10 Diaphragm Strain -80 5 -100 0 56 57 58 59 60 Time - Days from beam fabrication Figure 32. Variation of strain at bottom of diaphragm with time.

OCR for page 22
24 40000 50 Cast Slab Slab Concrete Temp. 45 35000 40 Slab Temp (degree C) Reaction (pounds) 35 30000 30 25 25000 20 15 20000 10 End Reaction 5 15000 0 56 56.5 57 57.5 58 58.5 59 Time - Days from beam fabrication Figure 33. East end reaction after deck slab pour. offers little resistance to the thermal movements. Thus, the strain in the bottom flange and a negative (i.e., compressive) cooling of the deck slab results in a contraction of the deck strain in the top flange, as expected. Both the top and bottom slab, which mimics the effect of deck slab shrinkage. The flange strains stay reasonably constant until just before the models predict that contraction of the deck slab will cause a peak in the deck slab temperature, which is approximately the decrease in the end reactions, compression at the bottom of time of the final set for the concrete. The strains then approx- the diaphragm, and a downward deflection of the girders. All imately follow the temperature curve. It therefore appears that of these responses are seen in the data. at this point, the deck slab has hardened sufficiently to begin Figure 34 shows the placement of vibrating wire strain affecting the girder behavior. The peak strains in the flanges gages in the cross section. All gages shown in Figure 34 are occur slightly before either the peak deck slab temperatures placed at midspan of the girders. Figures 35 and 36 show the or the peak girder temperature. There is slightly better agree- strains in the top and bottom flanges of one of the girders. ment with the gradient (slab temperature - bottom flange tem- The initial values are the compressive strains due to prestress- perature), but the strain still peaks earlier. The reason for this ing and any creep or shrinkage that has occurred. The graph is not apparent. Clearly, the girder is subject to a complex also shows the temperature measured in the deck slab and in series of interactions involving the thermal contraction of the the top or bottom flange of the girder. When the deck slab slab and thermal strains in the girder. Completely understand- weight is added, there is a positive (i.e., tensile) change in the ing this behavior would require complex computer analysis beyond the scope of this project. Figure 37 shows the strain in the slab. The strain is taken as 0 at the point of final set, as determined from Figures 35 and 36. The deck slab strain follows the slab temperature. Tem- perature decreases cause tension as the deck slab attempts to contract but is restrained by the girder. Increases in temper- ature cause compression as the deck slab expands but is restrained by the girder. Again, this is the expected behavior. Monitoring after Construction The deck slab was wet cured for 7 days, and the forms were removed during the next 14 days. Form removal took some time because of the specimen geometry. In order to duplicate field conditions as clearly as possible, the specimen was supported by three load cells placed under each girder Figure 34. Placement of vibrating wire strain gages in (see Appendix B). Thus, all the support was near the center- cross section at midspan. line of the specimen and there was a possibility that the spec-

OCR for page 22
25 -700 42 Pour Slab Slab Temperature Bottom Flange Strain - Microstrain 35 Temperature - Degree C -725 28 Bottom Flange Strain 21 -750 14 7 Temperature Gradient -775 0 57 57.5 58 58.5 59 Time - Days from beam fabrication Figure 35. Bottom flange strain after deck slab pour. imen might tip over unless stabilizers were installed (see Fig- During the form removal period, the end reactions showed ure 26) as the forms were removed. Until the stabilizers were daily variation due to temperature, but the average reaction installed, the specimen was kept stable by formwork under did not change. This was unexpected because it was thought the diaphragm. As a result, it was not possible to accurately that these reactions would continue to decrease due to shrink- monitor the center reaction while the formwork was under age of the deck slab. A shrinkage specimen was made when the diaphragm. However, since the only formwork support the deck slab was cast. It was water cured for 7 days (the same was at the center, changes in the center reaction should result as the deck slab) and then allowed to shrink. The specimen in changes to the end reactions--for example, if the center was kept in a trailer near the beam. Shrinkage was measured reaction increases, the end reactions should decrease by the through an embedded vibrating wire gage. During the period same amount. Thus, during the form removal period, devel- of form removal, a shrinkage of approximately 350 micro- opment of moments due to time-dependent effects can be strain was measured. A similar shrinkage specimen had been determined from the end reactions. made for the girder; and, during the same period, the girder 0 42 Slab Temperature Pour Slab -50 35 Top Flange Strain- microstrain Temperature degree C -100 28 Top Flange Strain Temperature Gradient -150 21 -200 14 -250 7 -300 0 57 57.5 58 58.5 59 Time - Days from beam fabrication Figure 36. Top flange strain after deck slab pour.

OCR for page 22
26 120 45 Slab Temperature 40 100 35 Slab Strain - microstrain 80 Temperature degree C 30 60 25 40 20 15 20 Pour Slab 10 0 Final Set Slab Strain 5 of Slab -20 0 57 57.5 58 58.5 59 Time - Days from beam fabrication Figure 37. Deck slab strain after deck slab pour. concrete shrank only 45 microstrain. With a differential shrink- diaphragm would vary +250 k-ft per day. Given that the age of almost 300 microstrain, some development of nega- nominal cracking moment at the diaphragm was calculated tive moment should have been seen through a decrease in the as 420 k-ft and the positive moment connection capacity was end reactions. 504 k-ft, this daily variation is significant. After all the forms were removed, the system was moni- During the monitoring period, the average end reaction tored for 4 months. The most striking observation was the increases about 5 kips while the center reaction drops by daily change in the reactions due to temperature (see Fig- 10 kips, and there are not large differences in the girder-slab ure 31). The end reactions vary by approximately +5 kips per strains. Figure 38 shows that after the slab is cast, the deck slab day. This is a daily variation equal to approximately 20% of and top flange strains decrease the same amount (200 micro- the reaction. For this variation in reaction, the moment at the strain) while the bottom flange strain decreases slightly more 200 0 Deck Slab Strain -200 microstrain -400 Pour Slab Top Flange Strain -600 Strain at prestress transfer Bottom flange = -548 Top flange = 0 -800 -1000 Pour Partial Diaphragm Bottom Flange Strain -1200 0 50 100 150 200 250 Time - Days from beam fabrication Figure 38. Change in girder/deck slab strain with time.

OCR for page 22
27 (250 microstrain). It was expected that the center reaction Atr = area of concrete deck slab with transformed longitu- would increase and that there would be large differences dinal deck reinforcement, Ac + As (n - 1); between the strains due to differential shrinkage of the deck As = total area of longitudinal deck reinforcement; slab. Most analyses of this type of system show that if the n = modular ratio = Es /Ec; girders are older than 60 days when the deck slabs are placed, Es = modulus of elasticity of the steel; and a large increase in the center reaction and development of Ec = modulus of elasticity of the concrete. large negative moment occurs. The PCA tests (7 ) found an increase in center reaction with time. If the concrete modulus is low (as at early ages), the mod- The results of the current study are consistent with recent ular ratio increases and the effective shrinkage decreases. field studies of bridges. FHWA sponsored a program to build Thus, at early ages when the incremental shrinkage is high, high-performance concrete highway bridges in several states. the effect of bar restraint is also high. At later ages, the con- The results are compiled on CD-ROM (26). Strain gages were crete modulus increases and the effect of bar restraint is less placed in girders in three states: Louisiana, South Dakota, and pronounced, but the incremental shrinkages are also lower. Washington. In these projects, the girders were rather old BRIDGERM and RESTRAINT (the program developed when the deck slabs were placed (approximately 60 days for Louisiana, 200 days for Washington, and 300 days for South for this study) both incorporate an equation developed by Dakota), so the effects of the differential shrinkage should Dischinger (11). According to NCHRP Report 322, this equa- have been noticeable through large strain changes and down- tion is used to account for the restraining effect of the bar in ward cambers of the girders. However, the strains in the gird- the slab; what the equation actually does is attempt to correct ers remained almost constant (except for daily temperature for the fact that the modular ratio of the slab changes with variation), and there was little change in the cambers. Thus, it time because the concrete modulus changes with time. How- does not appear that large negative moments develop. This is ever, even with corrections for relative humidity and rein- also consistent with anecdotal field evidence that negative forcement, RESTRAINT still predicts that a negative moment moment distress has not been observed in these bridges. due to differential shrinkage should form. Clearly, more work The results found in the present study seem to agree with is needed on this aspect of the model. the studies by Ramirez et al. (14), which showed that the As previously noted, the end reactions did not decrease due models overpredict formation of negative moment due to dif- to differential shrinkage as was expected, but rather increased. ferential shrinkage. The reasons why bridges do not show the The cause of increase in end reaction is due to two effects. predicted negative moment development is not entirely clear; One was a slight settlement at the center support. In order to however, some possibilities can be offered. duplicate field conditions, bearing pads had been placed under The actual deck shrinkage potential may be overestimated. the load cells. These pads seemed to have creeped slightly, Analytical shrinkage values are usually based an assumed rel- and the center support showed a settlement of 0.06 in. An ative humidity. Real decks may be exposed to much higher analysis was run and it was determined that this settlement humidity, especially at early ages when shrinkage potentials would cause an increase in the end reactions of 1.8 k, about are greatest. In the case of the specimen reported here, the one-third of the observed change. specimen was monitored over the summer at a facility near the The remaining change is a combination of temperature Ohio River. The daily relative humidity frequently exceeded effects and creep/shrinkage effects; however, it is very diffi- 90%. Decks are also subject to frequent rewetting from rain cult to separate the individual contributions. What is clear is and snow. The models base negative moment development on dif- that over the monitoring period, the girder bottom flanges at ferential shrinkage, but usually use the free shrinkage values. midspan showed an average of 80 microstrain more compres- These are usually obtained from equations given in the sion than did the top flanges (see Figure 38). This strain gradi- AASHTO Codes or ACI 209 (12, 23). In reality, the deck is ent would have caused the girders to camber up and increase not free to shrink. Decks are usually heavily reinforced, and the end reactions. During the monitoring period, the center this reinforcement will restrain the shrinkage. deflection decreased by 0.02 in.; but, when the effect of the Because the deck shrinkage is restrained by the steel, the center support movement is removed, the girder actually shrinkage strain in the deck will not be the unrestrained shrink- cambered up 0.02 in. age, but rather an effective shrinkage strain. The effective During the monitoring period, cracks were observed at the shrinkage strain in the deck can be found from girder-diaphragm interface. During the monitoring period, these cracks opened an average of 0.015 in. when measured effective = sh (Ac /Atr) 3 in. from the bottom of the specimens, but there was varia- tion. The average crack openings were 0.007 in. on the south- where west joint, 0.015 in. on the northeast and northwest joints, sh = unrestrained shrinkage strain for deck concrete; and 0.020 in. on the southeast joints (see Figure 26 for direc- Ac = gross area of concrete deck slab; tion). Note that cracks on the east side of the diaphragm

OCR for page 22
28 opened more than did those on the west side of the diaphragm those on the opposite faces where the bars were further from and that the cracks on opposite sides of the same girder did girder faces. not open the same amount. The reason the cracks on oppo- Figures 39 and 40 show the variation of crack opening site sides of the girder do not open the same is because of the with time for cracks at the bottom of west girder. The crack bar placement (see Figure 27). Although it was desirable to openings appear to show some softening of the connection place the bars in a symmetric pattern, this cannot be done and during the monitoring period as the crack opening amplitude still allow for the bars to mesh. As a result, the bars had to be increases with subsequent cycles. No similar trend is seen with offset about 3/4 in. This places the bars closest to the northeast the reactions as the amplitude of the daily change remains and southwest faces, which had less crack opening than did fairly constant over the monitoring period (see Figure 31). 0.018 0.016 0.014 Remove All Forms 0.012 Crack Opening (in) 0.01 Pour Slab 0.008 Pour Diaphragm 0.006 0.004 0.002 0 -0.002 -0.004 0 20 40 60 80 100 120 140 160 180 Time - Days from beam fabrication Figure 39. Opening of the crack at the bottom flange of the girder, northwest joint. 0.018 0.016 0.014 Cast Slab 0.012 Cast Diaphragm Crack Opening (in) 0.01 All Forms Removed 0.008 0.006 0.004 0.002 0 -0.002 -0.004 0 20 40 60 80 100 120 140 160 180 Time - Days From beam fabrication Figure 40. Opening of the crack at the bottom flange of the girder, southwest joint.

OCR for page 22
29 During the monitoring period, there was virtually no opening Continuity was determined by the change in reactions and at the top of the joint. the changes in girder strains. If the girder is continuous, load- ing one span will have a known effect on the reactions and strains in the other span. In a simple-span case, loading one Testing after Monitoring span will have no effect on the other. If there is partial con- tinuity, the strains and reactions will be between the simple After the monitoring period was over, the girders were tested to determine continuity. It is thought that cracking at and continuous cases. the diaphragm will result in a partial or complete loss of con- Figures 42 and 43 show the results of loading the speci- tinuity. By loading the specimen one span at a time, the conti- men after the monitoring was complete. The test consisted of nuity of the structure could be determined. Load was applied one load cycle to establish a baseline. The east girder was to the specimens with a pair of hydraulic cylinders (see Fig- loaded first, and then the west girder was loaded. The x axis ure 41) placed in each span at 22 ft from the face of the dia- is time, and horizontal lines on the graphs show the anticipated phragm. Analysis showed that this was the most efficient posi- reactions and strains for both the continuous and simple spans tioning of the loads. cases. The reactions and strains are consistent with continu- The specimen was loaded as follows to create a negative ity, even though cracking exists at the connection. moment at the center support: After the initial testing, the cracks at the diaphragm were opened by simulating the positive moment that would develop from creep and shrinkage of the girders if continuity was 1. Span 1 was loaded with a downward load, established at an early age. To do this, a post-tensioning bar 2. Span 2 was loaded with a downward load, was placed through the bottom flange of the girder, and a 3. Span 1 was unloaded, and post-tensioning force was applied (see Figures 44 and 45). 4. Span 2 was unloaded. This would cause the girders to camber up--simulating the formation of additional positive moment. Note that the post- The loads were such that when both spans were loaded, the tensioning (called "PT" in the figures) bars only went through applied moment was equal to the negative live-load moment. the girders and did not go through the diaphragm (i.e., the bar This is consistent with design procedures where the maximum was dead-headed at the end of the girder where the girder was negative moment occurs when the design truck has at least one attached to the diaphragm). The bar was a 1.375-in.-diameter axle on either side of the support. This loading also allowed for Dywidag bar. It was tensioned to the maximum allowable checking the requirements of the current AASHTO LRFD force of 160 kips. This increased each end reaction by 8 kips, Specifications Art. 5.14.1.2.7c (12), which states: resulting in an additional positive moment of 400 k-ft or 0.96 Mcr. The cracks at the bottom of the joint opened as If the calculated stress at the bottom of the joint for the com- follows: southeast--0.01 in., northwest--0.008 in., and bination of superimposed permanent loads, settlement, creep, shrinkage, 50% live load and temperature gradient, if applica- southwest--0.004 in. The instrument on the northeast joint ble, is compressive, the joint may be considered fully effective. malfunctioned, so no data are available. The post-tensioning load was applied in four stages. After When one span is loaded, the applied negative moment is each stage of post-tensioning, vertical loads were applied to 50% of the negative live-load moment. the specimen. Downward loads were applied to the specimen as mentioned above to obtain the negative live-load moment. After the negative moment was applied and removed, posi- tive (upward) loads of 10 kips were applied in sequence in each span. This upward load placed the maximum positive live-load moment on the connection when both spans were loaded. Figures 46 and 47 show the crack opening at the bot- tom of the specimen for two cracks: northwest and southeast. The northwest crack appears to close under the initial load, even though the crack opening does not return to zero. This is not unusual with cracks in concrete--the surfaces tend to be rough and the rough cracks will often not close all the way back. The southwest crack does not appear to close under ini- tial load. By the time all the post-tensioning is applied, both cracks are opening and staying open on loading. However, the data from subsequent loading still show that the connec- tion remains effective because the girders act as if continu- Figure 41. Loading device. ous (see Figure 48).

OCR for page 22
30 18000 16000 End Reaction if Simple Spans End Reaction - Loaded Span - Continuous - One Span Loaded 14000 Change in Reactions (pounds) 12000 East Beam Loaded 10000 West Beam Loaded 8000 End Reaction Continuous - Both Spans Loaded East End West End 6000 Both Beams Loaded 4000 2000 0 -2000 End Reaction - Unloaded Span - Continuous - One Span Loaded -4000 0 4 8 12 16 Time (min) Figure 42. Change in end reactions: loading to negative LL moment before post-tensioning. 80 Strain - Loaded Span - if Continuous - One Span Loaded Strain if Simple Spans 60 Bottom Flange Strain (microstrain) East Beam Loaded West Beam Loaded 40 East Beam Strain if Continuous - Both Spans Loaded West Beam Both Beams Loaded 20 0 Strain - Unloaded Span - if Continuous - One Span Loaded -20 0 4 8 12 16 Time (min) Figure 43. Change in bottom flange strain: loading to negative LL moment before post-tensioning.

OCR for page 22
31 Load Cell PT Bar Nut and Spacer Jack Figure 44. Position of post-tensioning duct. Figure 45. Post-tensioning device. Since the girders remained continuous after the post- actual bar yield was 83 ksi. The nominal concrete strength tensioning, the girders were loaded by increasing the positive was 4 ksi, but the actual strength was 6.7 ksi. Using the actual load (creating positive moment) to determine the load at properties, the capacity becomes 828 k-ft or 2.0 Mcr. There which, if any, the continuity was lost. Figures 49 and 50 show was no indication that the bars had yielded, nor was cracking the end reactions and bottom flange strains for a loading to observed in the diaphragm. the maximum negative moment (-38 kips at each point) fol- It appears from Figures 49 and 50 that the girders main- lowed by loading up to +40 kips at the east point and +42 kips tained continuity as the values of the reactions and strains at at the west point. This positive load would cause an addi- each load increment are those expected for a continuous sys- tional positive moment of 395 k-ft. Added to the 400 k-ft tem. It should be noted that when the west load was increased caused by the post-tensioning, the total positive moment was to 42 kips, the structure began to lift off the center supports, so 795 k-ft or 1.90 Mcr. This exceeded the nominal design the test was stopped at this point. Crack openings at the max- capacity of 1.2 Mcr, but that capacity was calculated using imum load were as follows: southeast--0.035 in., northwest-- nominal properties. The nominal bar yield was 60 ksi, but the 0.029 in., northeast--0.025 in., and southwest--0.015 in. 40 Load After Each PT Increment Load Before PT 20 Total applied load (k) 0 -20 -40 -60 -80 0 0.005 0.01 0.015 0.02 0.025 0.03 Crack Opening (in) Figure 46. Crack opening at bottom of girder load during post-tensioning, northwest joint.

OCR for page 22
32 40 Load After Each Increment of PT 20 Load Before PT 0 Total Load (k) -20 -40 -60 -80 0 0.005 0.01 0.015 0.02 0.025 0.03 Crack Opening (in) Figure 47. Crack opening at bottom of girder load during post-tensioning, southeast joint. 18000 16000 End Reaction - Loaded Span - Continuous - One Span Loaded End Reaction if Simple Spans 14000 12000 West Beam Loaded East Beam Loaded Change in Reaction (pounds) 10000 8000 East End West End 6000 End Reaction - Continuous - Both Spans Loaded 4000 2000 Both Beams Loaded 0 -2000 End Reaction - Unloaded Span - Continuous - One Span Loaded -4000 0 2 4 6 8 10 Time (min) Figure 48. Change in end reactions: Loading to negative LL moment after post-tensioning.

OCR for page 22
33 20000 East = -38k East = 0 East = -38k East = 0 West = 0 West = -38k West = 0 West = -38k 15000 East = 40k East = 0 West = 30k West = 40k 10000 East = 30k East = 10k West = 30k Change in Reactions (pounds) West = 0 5000 0 East = 40k West = 42k -5000 East = -38k West = -38k -10000 East = -38k East = 10k West = -38k West = 10k -15000 Applied Loads East = 30k Negative Load = Negative Moment West = 10k East = 40k Positive Load = Positive Moment West = 40k -20000 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Time (min) East Reaction West Reaction Figure 49. Change in end reactions: application of additional positive moment. Figures 49 and 50 also show that continuity is maintained for extended strands (see Figure 51) was chosen so that the strands negative moment after unloading. were symmetrical and there was as much space as possible between adjacent strands. This minimizes interaction between the strands and maximizes bond strength. Full-Size Specimen 2 This specimen was tested at 21 days, after ensuring that deck slab-diaphragm concrete strength exceeded 4,000 psi. After completion of the first full-size test, the diaphragm As with the first specimen, the loading mechanisms were and girders were cut apart. The second specimen was formed placed 22.5 ft from the face of the diaphragm. The second by turning the same girders end-for-end and forming a new diaphragm. When the slab was cast for the first specimen, the full-size specimen was loaded as follows: last 12 ft (3.7 m) of the slab was left off (see Figure 26). When the girders were turned around, there was room to cast 1. The east span was loaded to -38 kips. 12 ft of slab on either side of the new diaphragm so that a 2. The west span was loaded to -38 kips; at this point, the complete joint was formed. The new slab was connected to negative live-load moment was applied at the connection. the existing slab by bars extending from the existing slab. 3. The east span was unloaded. The slab section and diaphragm were poured as a single mono- 4. The west span was unloaded. lithic pour. 5. The east span was loaded to +20 kips; at this point, the The connection for the second full-size specimen was a positive live-load moment was applied at the connection. bent-strand connection (see Figure 51). This connection was 6. The west span was loaded to +20 kips; at this point, designed to a capacity of 1.2 Mcr , the same as full-size Spec- twice the positive live-load moment was applied at the imen 1. The strand length was kept at 26 in., the same as Spec- connection. imens 1 and 3. To get the proper capacity, a total of 10 bent 7. The east span was unloaded; at this point, the positive strands were needed. The girder had a total of 20 strands, so live-load moment was applied at the connection. only half of the strands were needed. The pattern for the 8. The west span was unloaded.

OCR for page 22
34 80 East = -38k East = 0 East = -38k East = 0 West = 0 West = -38k West = 0 West = -38k 60 East = 40k East = 0 West = 30k Change in Bottom Flange Strains (microstrain) West = 40k 40 East = 30k East = 10k West = 30k West = 0 20 0 East = 40k West = 42k -20 East = -38k West = -38k -40 East = -38k East = 10k West = -38k West = 10k -60 Applied Loads East = 30k Negative Load = Negative Moment at joint West = 10k East = 40k Positive Load = Positive Moment at joint West = 40k -80 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 Time (m) East Beam West Beam Figure 50. Change in bottom flange strain: application of additional positive moment. Figure 51. Full-size Specimen 2, bent strand.

OCR for page 22
35 Figures 52, 53, 54, and 55 show the results of the first load- In an effort to crack the joint, the specimen was loaded ing. The load and bottom flange strains generally exceed with a positive load in order to create additional positive those predicted by elastic analysis, especially when positive moment at the joint. When a positive load of 20 k was applied moment is applied and both spans are loaded. However, at both points, the first cracks appeared at the diaphragm. The since there was no cracking observed at the diaphragm, the total applied positive moment at the joint was approximately system should be continuous, so results of the load distribu- 445 k-ft or 1.07 Mcr. tion and the strains will be used as a baseline for comparison Additional positive load up to 40 kips/point was applied. with other results. The moment at the diaphragm reached 640 k-ft or 1.54 Mcr. As with the first full-size specimen, the girders were The average crack opening at the diaphragm was 0.011 in. As post-tensioned in four increments to simulate creep and with the initial load, the data showed that the responses gen- shrinkage. At the start of post-tensioning, the end reactions erally exceeded that expected from elastic analysis. How- were 33 kips. The post-tensioning should have increased ever, when compared with the base line (see Figures 54 and the end reactions by 8 kips. The first increment of post- 55), the responses indicated that continuity was maintained. tensioning increased the end reactions to 35 kips as expected; Table 2 compares the end reactions and midspan bottom flange strains for both the uncracked section under a load of but, after the first increment of post-tensioning was applied, +20 kips/point and the cracked section under a load + 40 kips/ the pump for the post-tensioning ram broke and repairs took point. Since the positive load is doubled, the ratio of the several days. When it was time to apply the last three incre- responses should be 2. As seen in Table 2, the average ratio ments of post-tensioning, it was found that the end reac- of the end reactions is 1.97. This is a reasonable agreement. tions had dropped 3 kips. This drop was mostly due to tem- The average ratio of the strains is 2.13, higher than 2, but rea- perature changes, although there was also a slight movement sonable considering that in some cases small numbers are of the west-end support (this support was monitored for the being compared. remainder of the test, and no further movement was found). When additional positive load was applied, the specimen As a result of this drop in the end reaction, the post- began to lift off the center supports, so it was no longer tensioning, when complete, increased the end reactions by possible to apply positive moment to the connection through 5 kips to a total of 38 kips rather than to the expected 41 kips. load. The research team had two choices: wait for the weather At this point, the applied positive moment was approximately to warm up (which would increase the end reaction) or artifi- 250 k-ft or 0.6 Mcr. As expected, no cracking was detected at cially increase the end reaction. It was decided to increase the the diaphragm. end reactions by jacking and shimming the end supports. 18000 End Reaction - Loaded Span 16000 Continuous - One Span Loaded End Reactions if Simple Span 14000 Change in Reactions (pounds) 12000 10000 East Beam Loaded West Beam Loaded 8000 East Rxn West Rxn 6000 End Reaction - Continuous Both Beams Loaded Both Spans Loaded 4000 2000 0 End Reaction - Unloaded Span Continuous - One Span Loaded -2000 -4000 0 5 10 15 20 Time (min) Figure 52. Change in end reactions: load to negative LL moment.

OCR for page 22
36 80 Change in Bottom Flange Strain (microstrain) Strain - Loaded Span Strain if Simple Spans Continuous - One Span Loaded 60 40 Strain - Continuous West Beam East Beam Loaded West Beam Loaded - Both Spans Loaded East Beam Both Beams Loaded 20 0 Strain - Unloaded Span Continuous - One Span Loaded -20 0 5 10 15 20 Time (min) Figure 53. Change in bottom flange strains: load to negative LL moment. The west end was jacked and shimmed approximately capacity of 1.2 Mcr. The cracks at the bottom flange of the 2.5 in., and then the east end was shimmed by a similar amount. girder had opened to an average of 0.07 in. (see Figure 56). With the combination of post-tensioning and shimming, the Unlike the bent-bar specimen, the crack openings were rea- end reactions increased to 58 kips or 25 kips above the orig- sonably consistent on all four faces. Figure 57 shows the inal reaction value, providing a total positive moment of opening of one crack as a function of applied moment. There 1,250 k-ft, which equals 3.0 Mcr or 2.5 times the nominal were cracks in the slab (see Figure 58) and at the bottom of 4000 End Reaction - Unloaded Span Continuous - One Span Loaded 2000 Change in Reaction (pounds) 0 Both Beams Loaded -2000 East Beam East Beam Loaded West Beam -4000 West Beam Loaded -6000 End Reaction - Continuous Both Spans Loaded End Reactions if Simple Span -8000 End Reaction - Loaded Span Continuous - One Span Loaded -10000 0 5 10 15 20 25 Time (min) Figure 54. Change in end reactions: load to positive LL moment.

OCR for page 22
37 20 Strain - Unloaded Span Continuous - One Span Loaded Change in Bottom Flange Strain (microstrain) 10 0 Strain - Continuous -10 - Both Spans Loaded West Beam -20 East Beam -30 -40 East Beam Loaded Both Beams Loaded West Beam Loaded -50 Strain if Simple Spans Strain - Loaded Span Continuous - One Span Loaded -60 0 5 10 15 20 25 Time (min) Figure 55. Change in bottom flange strains: load to positive LL moment. the diaphragm, and a small diagonal crack formed in the face maintained until the last load cycle. In the last load cycle, the of the diaphragm. This type of cracking was seen in the stub end reactions increased and strains increased significantly specimens and was usually a sign of impending failure. over the fully continuous case. Table 3 compares the end The jacking and shimming was done in increments, and reactions and bottom flange strains of the baseline case for the system was loaded to the negative live-load moment at negative moment at the support (see Figures 52 and 53). The various points during this process. Continuity, as measured important ratio is the case in which both spans are loaded by changes in the end reactions and strains in the girders, was because this case represents the full live-load moment at the TABLE 2 Comparison of responses for positive moment: full-size Specimen 2 Baseline Load = Load = +20k +40k Ratio End Reaction East/Load East (lb) 8,200 16,500 2.01 End Reaction East/Load Both (lb) 6,800 13,500 1.99 End Reaction East/Load West (lb) 1,150 2,700 2.35 End Reaction West/Load East (lb) 1,040 1,550 1.49 End Reaction West/Load Both (lb) 6,500 13,500 2.08 End Reaction West/Load West (lb) 8,075 15,250 1.89 Average 1.97 Strain East/Load East (microstrain) 34 73 2.15 Strain East/Load Both (microstrain) 30 63 2.10 Strain East/Load West (microstrain) 3 9 3.00 Strain West/Load East (microstrain) 5 7 1.40 Strain West/Load Both (microstrain) 26 56 2.15 Strain West/Load West (microstrain) 32 63 1.97 Average 2.13

OCR for page 22
38 As noted previously, the models predict that when positive moment cracks occur in the joint, the crack immediately prop- agates into the slab (see Figure 12). The experiments show this does not happen. The cracks start at the bottom of the joint and propagate upward under increasing load or under cyclic loading. When the crack finally penetrates the slab (as it did in this last case), the cracked section is now reasonably close to the cracked section used in the models and the expected drop in continuity is seen. Since the system was also to be tested for negative moment capacity, the positive moment testing was stopped at this point so that connection would not become excessively dam- aged. The ends of the girders were jacked up, and the shims were removed. The post-tensioning was also removed. The cracks closed back to openings of 0.005 in. Due to the tortu- ous nature of cracks in concrete, it is rare that the cracks close Figure 56. Crack at bottom of girder after post- all the way back when load is removed. The end reactions tensioning and shimming. also dropped to 25 kips because of the permanent deforma- tions at the joint and in the diaphragm (recall that the dia- phragm had also cracked). The system was subsequently support. The ratio here is about 1.3, indicating a 30% drop in reloaded with the negative live-load moment. Figures 59 and continuity. The system still shows some continuity, but it is 60 show the end reactions and strains measured from the ini- clearly reduced. tial loading sequence (before post-tensioning and jacking the In the section on the analytical studies, it was found that ends) compared with this final loading sequence. The results after cracking, continuity was predicted to drop by 20% or show full continuity was restored. more. This was not found in the experimental study until The tests show that the system maintains continuity even now, the point at which the connection was about to fail and if positive moment cracking occurs at the joint. Loss of con- the crack had propagated into the slab. Actually, this is the tinuity does not occur until the slab and diaphragm crack and point where the experiment became consistent with the model. the connection is near failure. 2000 Load/Unload after 1st Jacking West 2nd Jacking East Initial (pt) 1500 2nd Jacking West Applied Moment (ft-k) 1000 1.2 Mcr Nom Load/Unload after 2nd Jacking East 500 1st Jacking East Load/Unload After 1st Jacking East 0 -0.01 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 Load/Unload After Remove PT and Shims -500 -1000 Crack Opening (inch) Figure 57. Crack opening at northwest joint as a function of applied moment.