3 Experimental Program
The hybrid joist design used in the experiments is intended to combine the benefits of prestressed concrete double tees and open-web steel joists but overcome their shortcomings. The hybrid joist was envisioned for use in office construction. A length of 32 ft and a tributary width of 8 ft were chosen for the initial design. Loads of 50 psf office live load and 20 psf superimposed dead load were assumed. All loads were assumed to be uniformly applied along the joist length. This resulted in a superimposed total uniform service load of 70 psf and an ultimate load of 113 psf.
The overall configuration of the joist is shown in Figure 1.1 The joist webs had a constant thickness of 6 in. Joist web ends were 10 in. deep; the depth of all other joist web elements was 6 in. Overall depth of the web was 24 in. Three openings were located along the joist length. Prestressing tendons were located in the top and bottom chords of the web. The prestressing tendon profile is shown in Figure 2. Six tendons were used, two straight and four draped. Figure 3 summarizes the web reinforcement of each of the beams. The cast-in-place concrete flanges of HJ-3 and HJ-4 had a thickness of 4 in. and width of 6 ft. The flanges of joists HJ-6 and HJ-7 were 4 ft wide. The slab was reinforced with welded wire fabric (WWF), 4 X 4-W4.0 X W4.0, placed at a height of 2 in. Detailed descriptions of each joist design are provided in Saleh, Brady, Einea, Tadros, and Decker (1997).
Four prestressed high-strength concrete tee-beams with integral web openings were tested. Two joists were used as control specimens. One control joist had insufficient shear reinforcement; one joist was properly reinforced, designated HJ-6 and HJ-7 respectively. The other two joists were repaired, HJ-4, or upgraded, HJ-3, with FRP to improve their shear performance. Joist designations are shown in Table 1.
Performance criteria were specified for the two joists to be repaired. It was required that their shear capacity be increased 15 kips over a length 3 ft-10 in. from each end and 10 kips over the following 4 ft. The two repaired beams were wrapped on three sides with Fyfe's TYFO S FibrwrapTM along the outer 8 ft of each end of HJ-3 and HJ-4, Figure 4. The FRP repair design was based on the following material properties:
fsj = 12 ksi (conservative estimate of allowable jacket stress)
fuj = 65 ksi (ultimate jacket stress, minimum)
Ej = 3250 ksi (modulus of elasticity of jacket)
_aj = 0.004 ( allowable jacket strain)
_uj = 0.02 (ultimate jacket strain)
uaj = 400 psi (allowable bond stress)
tj = 0.051 in. (jacket thickness per layer)
Standard structural engineering practice for shear designs was used to determine the jacket thickness. Calculations were based on controlling shear crack widths to maintain aggregate interlock and proper shear transfer through the concrete. The allowable jacket strain, _aj = 0.004, represents 20% of the ultimate composite strain. The calculations resulted in the requirement for two layers of SEH-51, with the main fiber strength vertical, over the extreme 4 ft. The next 4 ft required only one layer per the calculations, however, the Fyfe Co. recommended the use of a minimum of two layers (Gee 1996).
No additional anchorage system was used due to the potential interference with the prestressing tendons of the existing joist.
1. Concrete. The concrete mix used in the hybrid joist specimen webs was a high-performance concrete (HPC). It provided special performance requirements including ease of placement and consolidation without compromising strength, superior long-term mechanical properties, early high strength, volume stability, and long life in severe environments. The HPC concrete strength used was designed to have a strength of 12,000 psi at 28 days. Figure 5 shows the time versus strength curves for the concrete used in the webs. Ready-mixed concrete was used in the slabs of all specimens. The mix was specified to be 5,000 psi and consisted of Type I cement with a maximum aggregate size of 1.0 in. limestone. The mix corresponded to dry weight proportions of 1.0:3.0:2.6 (cement : fine aggregate : coarse aggregate). On the day of testing all cylinders were also tested. Compression tests were conducted in accordance with ANSI/ASTM C39-86.
2. Steel. The tendons used were manufactured by the American Spring Wire Corporation (26300 Miles Rd., Cleveland, OH 44146). These tendons were 1/2 in. diameter, 270 ksi, low relaxation. The stress-strain curve for these tendons is shown in Figure 6. The shear reinforcement in the webs consisted of bar reinforcement, Grade 60. A welded wire fabric mesh, Grade 75, was used as reinforcement for the cast-in-place slab.
3. FRP. The FRP was specified as TYFOTM S Fibrwrap System and manufactured by Fyfe Co. L.L.C. of San Diego, CA. The TYFOTM S epoxy is a two-component, solvent-free, moisture insensitive epoxy matrix material. It is a high elongation material which gives optimum properties as a matrix for the TYFOTM fiber system. The epoxy has no offensive odor and maintains its properties up to 140 oF. Table 2 lists the epoxy properties. The TYFOTM fiber system is a plain weave, predominately warp unidirectional fabric comprised of a warp (0 degree orientation) of E-glass roving and a weft (90 degree orientation) of aramid, E-glass, and Thermoplastic Adhesive. The ratio of warp fiber to weft fiber is 17.5 to 1 by weight. Table 3 lists the yarn properties and Table 4 the fabric properties. Two layers of the TYFOTM S Fibrwrap System were used. Table 5 lists the composite laminate specifications and Table 6 the composite properties. The system has been tested and develops an allowable shear stress of greater than 350 psi without anchors.
The webs of the joists were prestressed and cast horizontally, i.e., on their sides as shown in Figure 7. Hold-down devices were used at the draping points to position the tendons and resist the prestressing forces. The concrete mix was placed in the forms and vibrated to ensure consolidation of the concrete. The specimens were covered with wet burlap that was kept moist for the first 3 days. The specimens cured at room temperature for 7 days. Cylinders measuring 4 in. diameter by 8 in. tall were cast and cured with the joists under the same conditions. The concrete strength was monitored by compression testing of cylinders to assess when the required release strength was achieved. When the strength reached 7000 psi the tendons were released by alternately torch cutting a tendon on each face at the joist ends. Casting and release dates for each specimen are shown in Table 7. The webs were then turned vertically and stored in the lab. The webs were then positioned vertically upright and level. The slab forms were then constructed around them. After concrete placement, the forms and test cylinders were covered with wet burlap followed by plastic sheets. The burlap was maintained moist for 4 days following casting. After 7 days the forms were stripped. Figure 8 shows the final shape of the joists.
Prior to application of the composite overlay the joist surfaces were prepared. This involved removing the paint on the outer 8 ft of the webs, rounding the corners at the bottom of the beam web to a minimum radius of 1.5 in., and removing trowel marks and smoothing out rough areas using an electric grinder. Once completed, creases in the web left by the concrete form lining were filled with a rapid strength repair mortar. After the mortar was cured, the surface of the beams was again ground and then cleaned using methyl ethyl ketone to remove any excess dust. Cracks in the concrete of HJ-4 created during pre-loading were ignored since they were less than 1/16 in. wide. The two part epoxy TYFOTM S Tack Coat was mixed and troweled onto the surface of the beams where the repair/upgrade was to be applied.
While the tack coat began setting up, the reinforcing fabric was cut to the proper length using scissors and infused with the TYFOTM S two part epoxy. This was done by laying the fabric out flat and evenly spreading the resin on the fabric by hand to saturate the fabric. The fabric was then laid up around the end of the joist from just beneath the slab, around the web and up to the slab/web intersection again, Figure 9. The material was placed vertically (main fibers vertical) in bands of 52 in. (1.2m) on the sides of the joist. Adjacent bands were placed with a 4 in. butt splice. In regions of taper, the bands were applied as four pieces, two per side ensuring that main fibers remained vertical on joist faces. The material was carried under the joist and the excess cut off.
Because of lack of Cab-O-SilTM in the tack coat, the system applicators had great difficulty getting the FRP system to adhere properly to the concrete prior to curing. The cure time was also slow because of high humidity. Upon cure it was noted that the FRP had slipped down on both HJ-3 and HJ-4. A gap, uncovered by FRP, existed beneath the bottom of the slab on the web. In most locations the gap was not significant; however, on the north end of HJ-3 the gap was observed to be 1.25 in., Figure 10. After curing, voids between the composite and the joist were filled with epoxy, Figure 11.
Test specimens HJ-3, HJ-4, HJ-6, and HJ-7 were instrumented with displacement potentiometers, strain gages, and linear variable displacement transducers. Tables 8 and 9 summarize the instrumentation plans for HJ-3 and HJ-4. Figure 13 shows the layout of internal strain gages for HJ-3 and HJ-4. Internal strain gages were located so as to measure strains in both prestressing tendons and reinforcement. Once the composite was cured, strain gages were placed on the external surface at the locations of the most dramatic shear cracks, other previous shear failure areas, and at the FRP lap joints to monitor strain in the composite. Gages were symmetrically placed at each end of the joists. Figure 13 shows the location of these gages for HJ-3 and Figure 14 shows the gage locations for HJ-4. LVDT locations were the same for all joists (Figure 15). Displacements were measured by potentiometers at the center of the joist, beneath one web post and a distance 25% of the span length from a support along the inclined portion of the joist, Figure 16. All recorded potentiometer displacements were absolute, measured with respect to the laboratory floor. Displacement measurements were also taken manually on the west and east faces of the slab at each joist end, and along the east slab face at the center and beneath each actuator.
Figure 17 is a functional block diagram of the instrumentation, data acquisition, and test control systems used at CERL. All of the transducer output signals were connected to a Hewlett Packard2 Model 3052A data logging system. The system was controlled by computer through an instrument controller interface bus. The record channels were scanned at a predetermined sampling rate, and the data were recorded in ASCII text files on the computer.
The loading system consisted of two CGS/Lawerence Model 307-50 electro-hydraulic actuators (controlled by closed-loop servo controllers) and a function generator. The actuators were operated in a displacement-control mode. In this mode, the function generator supplies a slowly changing command signal to the controllers. The controllers send a drive signal to each of the actuators, which causes the actuators to move until the displacement measured by LVDTs located inside each actuator is equal to the command signal. The actuators also include load transducers that measure the applied load.
The test setup on the CERL Structural Load Floor is shown in Figure 18. Each specimen was tested as a simply supported beam under two symmetrical point loads with a clear span of m (31 ft) and a shear span of m (11 ft - 3 in.). Vertical loads were applied by 50-kip hydraulic actuators suspended from a load frame. The actuators were centered directly over the web posts of the specimens. In testing at CERL the stroke of each actuator was calibrated to zero after making contact with the specimen; a small pre-load was associated with this positioning. Specimens were loaded at a constant rate to a specified stroke limit. The actuators were maintained at this stroke while the joist was inspected for cracks; these were marked. Measured readings of deflections were taken at selected locations and the deflection data were checked. Stroke was then further applied to the specimen until the stroke limit of the actuators was reached. The full stroke (i.e., full load) was then removed from the specimen. Steel plates were added between the actuator and the beam. The actuators were then moved into contact with the specimen again; this was associated with a small pre-load. The test was continued in the same manner until the specimen failed. Data were recorded during loading and unloading cycles.
Of the two repaired specimens, HJ-4 was damaged to a predetermined level defined subsequent to testing the control beams, which were unrepaired. The beam was then unloaded and repaired. HJ-3 was not loaded prior to upgrading it with FRP. After repair, the beams were loaded at a constant rate of 0.2 in./min. in increments of 1 in. At each displacement increment, measured readings of deflections were taken at selected locations and deflection data were checked. Loading of HJ-4 continued until the bottom of the joist was ¼ in. from the load floor. The joist was then unloaded. HJ-3 was loaded in the same manner as HJ-4. The joist was tested to failure.
The measured load and deflection, strains in concrete, steel rebar and FRP, and crack development and failure of each specimen are discussed. Results of the two repaired beams are compared with two control beams.
Table 11 summarizes principal test results, including cracking load, location of first crack, failure load, equivalent uniform superimposed (SI) load at failure for the test configuration, and type of failure. All load values in the table represent the sum of the two actuator loads. The experimental cracking load was determined at the time the first crack was observed. Joist HJ-4 was loaded to a peak of 55.2 kips. After the FRP repair, HJ-4 was reloaded to a peak of 56.6 kips, approximately 690% of the SI service load or 422% of the SI ultimate load. The upgraded joist HJ-3 was then tested, and failed at a load of 52.6 kips, 393% of the ultimate SI design load. The two control joists, HJ-6 and HJ-7, failed at 48.7 kips and 65.0 kips respectively. HJ-6 failed at well below the anticipated capacity but still 363% of the ultimate SI service design load. The premature failure was attributed to insufficient shear reinforcement.
Deflection parameters, including camber at tendon release and experimental deflections due to the applied loads are summarized in Table 12. For the 31 ft clear span and 6 ft tributary width, the experimental deflections at the load equivalent to live load (LL), 3.5 kips, and the load equivalent to SI dead load (DL) + LL, 4.1 kips, are much lower than the ACI 318-95 limitations of L/360 (1.0 in.), and L/240 (1.55 in.), respectively, for specimens HJ-3 and HJ-4. Similarly HJ-6, and HJ-7 with 4 ft tributary widths deflected much less than the ACI limitations under loads of 2.3 kips and 4.0 kips for (LL) and (SIDL + LL), respectively. HJ-4 with the FRP repair permitted a midspan displacement of more than 11.3 in. without failing. The test was stopped as there was a lack of space to further deflect the joist. HJ-3 was able to deflect 7.7 in. before failure was initiated.
The experimental load versus midspan deflection curves for joists HJ-3, HJ-4, HJ-6, and HJ-7 are shown in Figure 19. Initial stiffness (below 0.2 psf) of all specimens is similar. After this point the stiffnesses of HJ-3 and HJ-4 were less than for either control joist. HJ-3 displayed more flexible response than the damaged or repaired joist HJ-4. HJ-4 was not able to achieve the performance of the control beam HJ-7 which had sufficient shear reinforcement. All joists were able to achieve their peak load repeatedly for several loading/unloading cycles. HJ-4 achieved the peak load for 5 cycles before the test was stopped. Its stiffness did not change significantly from cycle to cycle, Figure 20.
Deflection profiles along the joist length were approximated using potentiometer data as well as manual measurements from the joists' top flanges. A deflection profile is shown for HJ-3 with respect to load increments of a single actuator in Figure 21. Figure 22 shows deflection profiles for HJ-4 prior to repair and after the joist was repaired with FRP. HJ-3 deflected more than either the original or repaired HJ-4 for comparable load levels up to 25 kips. It also deflected much more than HJ-6. Similar plots for the other tested joists are shown in Figures 23 and 24. HJ-4, while able to deflect significantly was not able to match the performance of HJ-7. The shapes of HJ-3 and HJ-4 are much like that of the control joist, HJ-7. The shapes reflect the constant moment between load points and the marked stiffness variation along the specimen length. The increased curvature with increasing load also reflects progressively greater cracking in the center section of the joists. The deflected shape of HJ-6 emphasizes the effects of insufficient shear reinforcement in the joist's inability to benefit from the prestressing and optimized shape.
Three types of strain readings were used in testing the family of hybrid joists: internal strain of reinforcement and external strain on FRP surfaces - both measured by strain gages - and displacement measured over a specified gage length on concrete surfaces by LVDTs. For the latter measurements cracks may have developed within the gage length, and the strain (displace-ment/displacement) may be greater than the maximum concrete strain range of 0.003 - 0.004 for compression or 0.0001 - 0.0002 for tension.
Strain distribution over section depth is shown in Figures 25 and 26 for three critical sections of HJ-3 and HJ-4 with FRP repairs. The distribution was approximated from concrete strain measurements near the top of the section and prestressing strand strains above and below the openings. Similar plots are shown in Figures 27 and 28 for HJ-6 and HJ-7, respectively. Strain along prestressing tendon length is shown in Figures 29 and 30 for HJ-3 and HJ-4 respectively. Figures 31 and 32 show strand strain measurements for HJ-6 through HJ-7.
Strains at the end and midspan of HJ-3 are similar in magnitude to those of HJ-6. It is apparent from Figure 25 that the full prestressing capacity could not be developed in these joists. This is further shown in Figure 29 where results of internal strain measurements along the strands for both top and bottom strands of repaired joist HJ-3 are presented. Strains in tendons were greatest in the shear span of this joist. In the constant moment region, strains are much less for both top and bottom tendons. From the strut section strain distribution, we can see that the neutral axis lies at a depth approximately 5 in. from the top of slab in HJ-3. From Figures 29 and 31 it is apparent that failure occurred before the full prestressing capacity could be developed in these joists.
Comparing Figures 26, 25, and 28, strain distribution in the repaired joist HJ-4 is quite different from that of either HJ-3 or HJ-7. The neutral axis indicated by the midspan strain is located at the member midheight. Peak strains in top and bottom prestressing strands of HJ-4 were greater than those in HJ-7, Figure 30 versus Figure 32. However, strain distribution over bottom tendon length is much more uniform in HJ-7 providing greater ultimate flexural capacity of this section. HJ-4 did not approach the load capacity of HJ-7.
To assess the stress in the strands, the strains shown must be added to the strain due to prestressing and related to the elastic modulus of the material. The strand was fully tensioned, so the effective strain due to the prestress is approximately 6705 micro strain [(fse/Es) = 0.75 (270) / (30,000) (106) = 6750 micro strain]. All strains were below the ultimate strand strain of 35,000 micro strain. Again, the lack of strain developed in the strand indicates the poor performance of HJ-6. During testing it was observed that the bottom chord of HJ-7 appeared to arch upward between the struts; this may be related to the larger strains shown at the struts than the midspan for some load levels.
Figures 33 and 34 show load versus strain in the FRP material for HJ-3 and HJ-4 respectively. Strain gages along the beam web show elongation of transverse FRP with increasing load. In HJ-4 FRP strains do not begin to increase appreciably until the actuator load is approximately 12 kips indicating the widening of shear cracks in the concrete beneath the FRP and the developing shear resistance in the FRP. Strain in gages ES4 and ES5, closest to the beam center, reached a peak value greater than 0.005 in./in. This is above the allowable strain of 0.004 but much less than the ultimate strain of 0.02. The limited capacity of HJ-3 is shown by the much lower strain values of gages ES4 and ES5 than for HJ-4.
None of the joists cracked when the prestressing tendons were released. During handling, specimen HJ-7 developed a crack across the slab through its depth near the south strut. Cracks were marked on each of the joists throughout testing. Cracking and failure mechanisms resulting from testing of HJ-6 and HJ-7 were compared with those of the two hybrid joists upgraded or repaired with FRP. Early in the test series, limited cracking occurred in the bottom chord of HJ-6. As actuator stroke was increased, cracking in the shear spans became evident but the cracks in the bottom chord did not develop further. In HJ-6 an inclined crack developed near the support and progressed upward along the web/slab interface (Figure 35). This crack progressed into the slab and failure ultimately occurred in this North end of the joist (Figure 36).
Figure 37 shows crack development for HJ-7. Initial flexural cracks formed along the bottom chord at midspan. Cracks were regularly spaced, and they became more numerous and closely spaced as the displacement was increased. Near the end of testing, when the load was not increasing but the specimen was able to deflect significantly more, inclined cracks developed in the shear spans of the members. No actual failure was observed in specimen HJ-7. The joist continued to deflect after reaching an ultimate load capacity.
Cracking in HJ-3 initiated as for HJ-7 with flexural cracks in the web bottom chord. At an applied stroke of approximately 5 in. a crack began to develop along the edge of the FRP at the intersection between the joist web and slab, Figure 38(b). A gap of more than 1 in. of exposed concrete existed where the FRP had slipped down from the web/slab interface. The horizontal crack began near the point where the FRP lapped. As the horizontal crack progressed toward the North end of the joist, cracks also developed in the bottom of the slab perpendicular to the joist span as well, Figure 38(a). These were associated with popping sounds as if the FRP were debonding from the joist. A vertical crack in the FRP was observed at a distance approximately 6 in. from the north end of the joist. This occurred at a stroke of approximately 7.5 in. A maximum deflection of approximately 9 in. was achieved before complete collapse of the joist occurred by fracture of the top slab at a distance of approximately 56.75 in. from the north end. The FRP separated from the joist by buckling over the web depth at a distance approximately 41 in. from the joist north end. A third vertical break in the FRP was observed at 25.5 in. from the end, Figure 39 (a). These cracks in the FRP were accompanied by peeling of the top slab from the web at the construction joint, Figure 39 (b). Investigation of the failure revealed the concrete in the area of the FRP repair had completely broken up. The total length of crumbled concrete was approximately 50 in. Examination of the TYFOTM S Fibrwrap System showed it to be adhered to the perimeter concrete even at failure. Failure was in the concrete. This was precipitated by the weakness created by the gap in the FRP repair at the top of the web.
Initial testing of HJ-4 without FRP repair produced crack patterns similar to those for HJ-6, Figure 40. After repair testing began again, existing cracks between struts increased in size and additional cracks were observed to develop near the edge of the FRP repair area, Figure 41 (a) and (b). The test had to be stopped when there was no further vertical space between the web bottom chord and the floor for the joist to deflect. The joist did not fail. At the test conclusion, the FRP repair showed no signs of damage. The beam exhibited ductile response throughout the test.
HJ-4, while being damaged prior to application of the FRP repair, was able to deflect as much as HJ-7. However HJ-4 with FRP repair was not able to achieve the strength and stiffness levels of a properly reinforced specimen, HJ-7. The shear mode failure of HJ-3 was initiated by a gap on the joist web where the FRP had slipped during curing. Its performance was not improved over HJ-6.
1 All figures and tables are presented at the end of this report.
2 Hewlett Packard Co., 5301 Stevens Creek Blvd., Santa Clara, CA 95052-8059.
