Flexural Behavior Of Prestressed Split-Beam Composite Concrete . - PCI

Thin-wall steel tubing (electrical conduit) of 1-in. O.D. and As-in. wall thickness was located in the prestressed element of the split beams in the position specified for the prestressing tendon. The tubing provided a channel for the unbonded tendons. The tubing was fixed in position at the ends of the form and at the midspan. For the reference beams, instead of using tubing as in the split beams, the tendons were coated with a thick layer of high density lubricating grease and then wrapped in a double layer of 6 mil polyethylene sheeting. In the split beams, a thin sheet metal spacer served to position the tendons at midspan. The spacer was wedge shaped, tapering from a 4in. width at the level of the tendons to 1 in. at the bottom of the form. The tendons in the reference beams were held in place by one loop of tie wire fixed to the bottom of the form. Shear connectors were not provided in the split beams, thus the resistance to horizontal shear at the interface of the two beam elements was dependent solely on the bond developed between the two layers of concrete. Casting and Curing One 4-cu. ft. batch of concrete was mixed to cast each element of the split beams and two 4-cu. ft. batches were mixed to cast each reference beam. All beams were cast in a metal form of adjustable depth. The prestressing tendon and the spiral reinforcement were fixed in position and the concrete was placed in two equal layers. All freshly cast concrete was vibrated with a high frequency internal vibrator. In order to aid in the developJune 1965 ment of a good bond between the two elements of the split beams, the top surface of the prestressed element was treated in the following manner. Approximately 1 1/2 hr. after placing the concrete in the form, the excess material was screedec! off with a wood screed. The screed. ing was carried out only to the ex tent of providing a plane surface Following the screeding operation, the concrete was undisturbed foi approximately 2 hr. which was sufficient time for the initial set to take place. At this time a stiff wire hand brush was used to roughen the fur. face to such an extent that the harg est size aggregate was exposed. After curing overnight under wet burlap, the element to be pre. stressed along with the adjustable bottom was lifted up and supported on the top of the form. The element remained in this position wrapped in wet burlap and a single sheet oft waterproof building paper until the concrete developed the desired strength for prestressing as determined by test of 6- by 12-in. conixo cylinders. Following the prestress:inb operation described below, the element was placed back into the form and the concrete for the compressive element of the split beam was cast. After remaining covered overnight, the beam was removed from the form and placed on supports spaced 9 ft. apart. Wet burlap and waterproof building paper were applied to the split beam in the same manner as described above for curing. Prestressing Procedure The split beams were prestressed by post-tensioning a '3/4 in. diameter tendon that was threaded on each end. The tensioning force in the tendon was measured with a steel dynamometer attached to the 81

tendon at the end of the beam opposite to the jacking end. This force was distributed over the ends of the prestressed element, with 1 in. thick bearing plates. Heavy duty cold pressed stainless steel nuts bore against the dynamometer on one end and the bearing plate on the other end to maintain the prestressing force in the beam. The 1-in. diameter tendons in the reference beams were post-tensioned in a similar manner, except that the bearing plates covered the entire cross section at each end and were 11/Z in. thick. INSTRUMENTATION Gage lines for a 10-in. Whittemore strain gage were established on each side of the beams, parallel to the longitudinal axis, at the time of casting with gage plug inserts fastened to the inside of the form. There were four gage lines on each side of the split beams and two gage lines on each side of the reference beams. The gage lines were located at midspan approximately 1 in. from the top and bottom surfaces on all the beams and 1 in. above and below the interface on the split beams. Initial readings on all gage lines were made one day after casting each element of the split beams and the reference beams. The use of Whittemore strain gage instrumentation allowed observation of the strains in the concrete during the early stages of curing and served to reveal the nature of the stress block developed over the cross-section of the beams. Type A3 and A9 bonded wire electric resistance strain gages were used to measure longitudinal concrete strains in the beams during load tests. These gages were mounted on the specimens with a quick setting adhesive immediately before testing. Three A9 gages, with a gage length of 6 in., were placed end to end on the top (compressive) surface of the beam centered at the midspan giving a coverage of 18 in. The gages were connected in series to show the average strain. The A3 gages with a gage length of 1 %s in., were mounted on both sides of the beams at midspan. These gages were located at various points Fig. 3—View of Beam in Testing Machine 82 PCI Journal

I ---4 2250 6" I2 It BEAM A 6" A f4 - F Y7 000 le zzso x3Te It 2250 2" 335 597 -3T5 72 BEAM B 5I " F Y7.Y6ole 2463 2251 41 Y 042 12" BEAM C 1 t 1 F Y e,oso le 4 3 J 2314 MIDSPAN CROSS-SECTION STAGE STAGE 2 STAGE 3 Stage 1—Prestress plus weight of prestressed element Stage 2—Prestress plus weight of prestressed and compressive elements combined. Stage 3—Prestress plus weight of split beam plus effect of applied load producing zero stress at bottom surface of beam. Fig. 4—Midspan stress condition in the split beams computed for three stages of loading. The magnitude of the stress is shown on the stress blocks; minus signs indicate tension. F equals prestressing force. along the depth of the beam to show the distribution of strains over the cross-section. The deflections of the beams at midspan were measured with respect to tjie platen of the testing machine with 0.001-in. dial indicators. TEST PROCEDURE The beams were tested simply supported in a 600,000 lb. capacity hydraulic testing machine. A view of the beam set up for test in the testing machine is shown in Fig. 3. One end of the beam rested on a 1 in. thick steel bearing plate that extended across the width of the beam. This bearing plate was supported by a ball-socket assembly which provided longitudinal and June 1965 transverse rotational freedom. The other end of the beam was supported by a roller assembly. A '/4 in. thick strip of leather was placed between the beam and the upports. The prestress was adjusted and the load was applied at the third points through a steel loading beam supported on plate-roller assemblies. Load in increments of 2000 lb. -was applied until failure except between the loads of 10,000 to 14,000 lb. which was the critical cracking range where increments of 500 lb. were used in most cases. After the application of each load increment, the deflection of the beam, the force in the tendon, the strain in the concrete, and the extent of cracking were recorded. 83

RESULTS AND DISCUSSION The values of observed and computed characteristics of the split beams and reference beams are given in Table 1. The magnitude of prestress in the bottom fiber of the concrete section at midspan of 2250 psi was based on an assumed concrete compressive strength of 5000 psi at the time of test. The prestressing forces applied to the beams were 27,000 lb. for the type A beams; 27,980 lb. for the type B beams; 28,000 lb. for the type C beams; 54,000 lb. for the type RI beams; and 41,980 lb. for the type R2 beams. One of the variables for the three types of split beams was the magnitude of the prestress at the interface. The computed stress in the prestressed element at the interface of the beam at midspan corresponding to the effect of prestress and the weight of the beam (Fig. 4, Stage 2) was zero for type A beams, 547 psi in compression in the type B beams, and 381 psi in tension in the type C beams. The stress conditions at the mid- span for three stages of loading are shown in Fig. 4. The stress blocks illustrated are idealized omitting the effect of shrinkage and creep in the materials (shrinkage and creep were minimized by the curing technique previously described). Stage 1 in Fig. 4 represents the stress developed in the concrete due to the prestressing force and the weight of the prestressed element. Only the prestressed element of the split beam is involved at this stage. At stage 2 the weight of the compressive element is added to stage 1. As was pointed out earlier, the procedure for forming the split beam requires that the prestressed element be supported along its length in its cambered position while the compressive element is being formed. However, because of the short span length and the relatively light weight of the compressive element for the test beams in this study, initial stress allowances were made for the prestressed element to support the added weight of the compressive element. Stage 3 represents the stress condition in the DEFLECTION AT MIDSPAN Fig. 5—Observed load-midspan deflection relationships. Arrows indicate cracking loads estimated at points where curves departed from straight lines. 84 PCI Journal

beam corresponding to the applied load that reduced the initial prestress at the tensile surface to zero. The load-deflection curves for all beams are presented in Fig. 5. The deflections were measured with respect to the platen of the testing machine. Corrections for movement at the supports were unnecessary since the platen of the testing machine is extremely rigid under the loads applied in these tests. A comparison of the observed deflections with theoretical deflections showed very good agreement. The theoretical deflections were computed with a modified version of Maney's2 formula for the deflection of a reinforced concrete beam. The observed strain on the compressive surface of the beam was divided by the distance from this surface to the neutral axis giving a measure of EI. This value multiplied by the appropriate constant and the square of the span is equal to the deflection. For 1/3 point loading: 0 0.1065L2kd where: L span length e0 strain on the compressive surface kd distance from compressive surface to neutral axis of beam The load-deflection curves reflect the nature of the response of the beams to loading. As seen from the curves, the response to loading can be divided into two phases. In the first phase the beam exhibits a linear load-deflection relationship. In the second phase the beam has cracked and the propagation of the cracks through the cross section is indicated by a constantly increasing rate of deflection with load. June 1965 It is noted that the slopes of the load-deflection curves for both split beams and reference beams prior to cracking were very nearly equal. This would indicate that the addition of the electrical conduit to the split beams had no measurable effect on the rigidity of the beams up to the cracking load, even though the conduit increased the ratio of reinforcement by 42% in the split beams. The exact point in loading at which the first phase of the loaddeflection relationship ends and the second phase begins was not precisely determined in these tests. However, two different methods were used to determine approximately the cracking load. In one method the cracking load was determined as the point at which he load-deflection curve departed from a straight line. This point of departure was determined graphically. In the second method, the relationship between the applied load and the distance to the neutral axis, kd, was examined; the cracking load was defined as the load at which the value of kd departed from a constant value. Typical load -kd relationships are illustrated in Figs. 6 and 7 for split beams and reference beams respectively. These relationships were developed from the distributions of strains over the cresssection of the beams as illustrated in Figs. 8 through 11. Table 2 gives the approximate cracking loads determined from both the load-deflection relationship and the load -kd relationship. In general, the approximate cracking loads determined by the two methods were in good agreement. The computed value of the applied load for producing zero stress at the bottom fiber was 12,000 lb. for 85'

22.1( E40 2 i I 2 3 4 5 7 6 kE, 8 9 10 0 I 2 3 4 . 5 6 7 8 9 10 64, in. Fig. 6—Relation between applied load and distance to neutral axis, kd, typical of split beams. Fig. 7—Relation between applied load and distance to neutral axis, kd, typical of reference beams. all beams. Lower cracking loads cross section in these beams pracwere expected in the split beams tically undisturbed. than in the reference beams because Beam R2-1 developed several approximately 9 sq. in. of the botcracks in the top surface after pretom portion of the midspan cross stressing. These cracks were plainly section of the split beam was occuvisible and extended approximately pied by a thin sheet metal template 2 in. from the top of the beam. used to position the longitudinal Consequently, the prestress distrireinforcement. The longitudinal re- bution in this beam was somewhat inforcement in the reference beams different from that which was was held in place at midspan by planned, and higher initial stress one loop of tie wire fixed to the was obtained at the bottom surface bottom of the form which left the than in any of the other beams. A Table 2 Approximate Cracking Loads Beam A-1 A-2 B-1 B-2 C-1 C-2 R1-1 R1-2 R2-1 R2-2 Cracking Loads Estimated By Method 1 Method 2 Deflection D ata Initial Shift of N eu tral axis lbs lbs 12,000 12,200 12,100 13,000 13,200 12,500 14,500 12,500 14,500 14,500 ---- 12,500 11,500 13,000 13,000 12,500 13,500 13,000 ---- e 15,000 Erratic readings from electrical resistance strain gages on the concrete. This beam developed cracks on the top surface after prestressing. During the load test, the initial shift of the neutral axis was attributed to the closing of these cracks rather than the development of new cracks on the bottom surface. 86 PCI Journal

C C CS { NTERFACE CIRLLIE N AD IN INDICATE APPLIED LOAD IN INDICT CIRCLED NUMBERS INDICATE APPLIED LOAD IN LANDS O SR-4 GAGE MEASUREMENTS D WHITTEMORE GAGE MEASUREMENTS 600 400 200 TENSION d STRAIN 200 400 600 800 1000 1200 1400810-G 600 400 TENSION COMPRESSION 200 200 STRAIN 400 600 800 1000 1200 COMPRESSION Fig. 9—Strain Distribution in Split Beams of Type B Fig. 8—Strain Distribution in Split Beams of Type A iNTM.EAaE-------- D SR-4 GAGE MEASUREMENTS 0 WHITTEMORE GAGE MEASUREMENTS 3 M U A---------------------------- oAo D SR-A GAGE MEASUREMENTS D WHTTEMORE GAGE MEASUREMENTS 600 400 TENSION cc 200 1 STRAIN 200 400 600 800 1000 1200 COMPRESSION Fig. 10—Strain Distribution in Split Beams of Type C 140OX10-6 TENSION STRAIN COMPRESSION Fig. 11—Strain Distribution in Reference Beams R1 and R2 1400010-6

slight difference in the performance of this beam as compared with the other reference beams can be seen from the difference in slope of the linear portion of the load-deflection curve (Fig. 5) and in the applied load-compressive strairr relationship (Fig. 12). The linear distribution of strains over the cross-section of a beam under load is a positive indication of monolithic beam action. The linear distribution of strains observed in all split beams is illustrated in Figs. 8 through 10; the distribution of strain for the reference beams is shown in Fig. 11. One graph for each of the three types of split beams and the reference beams is presented as typical of the specimens. These graphs clearly show that the split beams responded to loading in accordance with the elastic theories of strain distribution, indicating that the abrupt changes of the strain gradients at the interface in the B and C beams prior to loading (see Fig. 4, Stage 2) had no apparent effect on the linear strain distribution of the beams under the applied load. Fig. 12 shows the relation between the compressive strain in the outermost fiber in the constant-moment span and the applied load. The strains are the average values over the 18-in, length covered by the three type A-9 gages. The compressive strains developed in essentially the same manner in all beams, except for the reference beam R2-1. It will be recalled that this beam developed several tensile cracks at the compressive face during prestressing resulting in greater apparent compressive strain during loading than in the other reference beams. All beams tested in this investigation failed by flexural-compression which is defined here as: crushing of the concrete in the region of constant moment above a flexural crack which has reduced the area available for resisting compressive stresses. The ultimate loads carried by the beams are given in Table 1. The average values of ultimate load Fig. 12—Relationship between compressive strain and applied load in the region of constant moment. 88 PCI Journal

for duplicate specimens of the A, B, C, RI, and R2 beams were 22,850 lb., 23,100 lb., 22,860 lb., 24,150 lb., and 24,650 1b. respectively. Although the differences between the reference beams and split beams were not considered to be of practical importance, it should be pointed out that the strength of the reference beams were all larger than those for split beams except for the tied results for beams A-I and R2-2. The magnitude of stress in the tendon at ultimate load was approximately 91,000 psi (0.80 f ., ) for split beams; 85,400 psi (0.69 f,) for the R1 beams and 78,500 psi (0.63 ff) for the R2 beams. At these stress levels the tangent moduli for the steels in the split beams, the RI, and the R2 beams were approximately 10.8 X 106 psi, 22.9 X 106 psi and 26.3 X 106 psi, respectively. These values were estimated from the stress-strain curves shown in Fig. 1. Thus, at loads near the ultimate, the stress in the tendons of the split beams was in the curvilinear range of the stressstrain curve while the stress in the tendons of the reference beams was essentially in the elastic range. The greater stiffness of the reference beams as compared with the split beams may be attributed to the larger ratio of reinforcement in the reference beams and the more nearly linear stress-strain curve for its tendon at loads approaching the maximum values. No attempt was made to gauge the effect of the electrical conduit in the split beams after cracking. Crack patterns typical of those in all beams are shown in Fig. 13. As was stated earlier, the first crack in all beams developed at or very near midspan and was shortly followed June 1965 by additional cracks symmetrically located about the midspan. During the loading operation close attention was given to the progress of cracks as they approached the interface in the split - beams to see if the cracks would propagate along the interface. However, the cracks in all cases proceeded to develop in the split beams with no particular regard to the plane of the interface. CONCLUSIONS The flexural behavior of the split beams and the reference beams as indicated by the load-deflection relationships and the strains developed over the cross section under load was similar up to the cracking load. Beyond the cracking load the reference beams exhibited a greater degree of stiffness than the split beams although the difference in the ultimate loads was very small. The abrupt changes of the strain gradient at the interface in the split beams due to prestressing had no measurable effect on the development of strains over the cross section of the beams under the applied load. The abrupt changes in stress were introduced by varying the location of the interface in the split beams with respect to the midplarte of the cross section while the location of the tendon in the split beam remained fixed. The procedure that was used for combining the two elements of flue split beams proved to be adequate for the development of sufficient bond for monolithic beam action throughout the tests. In all cases, cracks developed in the split beams with no particular regard to the plane of the interface. For the same working load capacity (load producing zero stress in bottom fiber of beam), the p

moved and the split beam attain its properties as a composite beam. SCOPE The objective of the investigation was to compare the structural prop-erties of split beams of different designs with conventionally pre-stressed beams referred to here as reference beams. Three sets of split beams were tested in duplicate and compared with two sets of refer-