post-tensioned bar were sheared off while the beam was being loaded. Beam SG-2 failed by complete separation of the interface of the tensile and compressive elements in the shear zone under load.

The expression Asfsyd is here shown to be a valid. index for an ordering of the overall load response for the beams in this investigation. It can also be seen in table 1 that the moment index (Asfsyd) values are in good general agreement with the respective measured ultimate moments for all the beams failing in tension. Therefore, the moment index is a means by which a comparison can be made of the tendon sizes required to produce equal load response and capacity for split-beams and monolithic beams. The moment index for the monolithic beams (RU-1, RU-2) is 755 in-kips and the tendon size is 0.37 sq in. With these values as references, the required tendon size for equal performance by the split beam is,

With the type of steel used for the post-tensioned beams, the tendon size for the split-beam is found to be 0.30 sq in. This is a reduction of approximately 20 percent below the size for the monolithic beam.

An analysis of the principal properties of the stress block was conducted to evaluate the performance of the concrete at ultimate load. It has been demonstrated in laboratory tests [3] that the shape of the stress block at the ultimate capacity of a beam varies with the strength of the concrete. The shape of the stress block varies from nearly trapezoidal for low strength concretes in the 2000 psi class to nearly triangular for high strength concretes in the 7000 psi class. The stress block has been found to be nearly parabolic for 5000 psi concrete. However, in determining the ultimate strength of beams, the exact shape of the stress block is not as important as the magnitude and location of the internal compressive force. This force can be defined and located in terms of three

parameters [4], kı, k2, and kз. The parameter k1 is defined as the ratio of the average compressive stress to the maximum compressive stress of the concrete in the compression zone of the beam at ultimate. The parameter k2 is defined as the ratio of the depth to the line of action of the resultant compressive force to the depth to the neutral axis. Parameter kз is defined as the ratio of the maximum compressive strength of the concrete in flexure to the cylinder strength.

The assumed stress conditions at ultimate load are shown in figure 12. The expression for the ultimate resisting moment is:

T=As su

FIGURE 12. Stress conditions at ultimate load.

measured steel stresses at failure. The results are the plotted points shown in figure 13. The curve shown by dashed line in this figure was developed from the results of a study by Janny, Hognestad, and McHenry [5] with rectangular beams covering five types of reinforcement. This curve represents the relationk2 ship in eq 3 for a value of equal to 0.52. The kikз plotted points in figure 13 are in close agreement with the curve. This indicates that the basic characteristics of the stress block at ultimate load for the Tee beams in this investigation are similiar to those for the rectangular beams studied by Janny et al. It also indicates that the unusually low steel ratios, tendon location, and initial stress gradient discontinuity in the split-beams had no adverse effect on the ultimate load performances of these beams over a wide range of concrete strengths in the compressive elements (flange sections).

As would be expected for beams failing in tension, the strength of the concrete in the compression zone had little if any effect on the ultimate capacity of the beams. In this study only a rather general trend may be noted in a comparison of concrete strengths and ultimate moments for beams in the 925 in-kip moment index group (SU-5 thru SU-14). However, when the results for the beams in the 650 in-kip index group (SU-1 thru SU-4) were considered together with those of the 925 index group, no direct linear correlation of concrete strength and ultimate moment was found.

The principal differences between the split-beams and the conventional monolithic beams in this study were in the initial prestress parameters allowed by the methods of construction. However, it was found that even with different initial prestress parameters, beams having the same ultimate capacity showed essentially the same overall flexural response to loading. The index Asfsyd, which was used to categorize the beams in relation to a scale of load response and capacity is essentially a measure of the internal resistant moment capacity for under-reinforced beams. The very close agreement between the index values and the respective measured ultimate moments for the Tee beams can be explained by considering the factors in the moment index expression. The factor d is approached within 10 percent by the actual moment arm at ultimate and the actual stress in the steel is somewhat greater than the yield strength of the steel, fsu, by a similar difference but opposite in direction to that for the d factor. Consequently, the two depar tures from actual conditions conveniently compensate for each other.

Split-beams that had compressive elements with concrete strengths around 2000 psi showed erratic performances under loading. The failure mode for these beams was unpredictable and included brittle failures in the constant moment region and in the shear span for different beams. In order to avoid the

risk of these type of failures, concrete strengths below 3000 psi should not be used in the compressive element of split-beams.

Due to the manner of construction and the design of the precompressed section, the split-beam enjoys the advantage of a reduced amount of reinforcing steel for the same overall flexural characteristics as compared with the conventional monolithic beam. However, it should be emphasized that the comparison here is between a composite beam and a monolithic beam. Also, no tensile stresses were allowed in the stages of construction. It may be better to evaluate the split-beams in relation to other composite beams. For example, it was stated earlier that the split-beam is a special case of composite construction where the construction joint is designed to coincide with the neutral axis of the composite section. Figure 14

shows the cross section of a conventional composite beam where the construction joint is located at the intersection of flange and web. The dimensions of the cross section are the same as for the other beams in this investigation. To include this section in a comparison with the other beam sections in this study, the basic prestressing parameters for the three types of beam construction features were computed with respect to the common moment index Asfsyd = 755 in-kips. These values are presented in table 2.

From the standpoint of performance, it is apparent that nothing is gained, in comparison with the splitbeam design, by locating the construction joint above the neutral axis of the cross section. In fact, for the same flexural characteristics, the required area of the reinforcing steel will increase with the distance of the construction joint above the neutral axis. Con

versely, when the construction joint is located below the neutral axis the required area of the reinforcement will decrease as the distance of the construction joint to the neutral axis increases. The limiting distance of the construction joint below the neutral axis will be affected by several practical considerations. Among these considerations are: (1) the minimum cross section needed for prestressing to a desired value; (2) the degree to which tensile cracks will be tolerated in the zone between the construction joint and the neutral axis within the working load range.

Although a strict economic evaluation for the practical use of split-beams was not within the scope of this study, the experience gained in preparing these specimens raises a serious question as to the balance between materials savings and the added cost of form work and construction handling.

6. Conclusions

Placing the construction joint at the neutral axis. of a prestressed composite beam allows for an efficient prestress distribution over the cross section with no adverse effect on the performance of the beam.

Stirrups should be provided throughout the span length for these beams to prevent the development of extensive horizontal cracking just above the neutral axis in the region of maximum moment, and to serve as reinforcement against possible interface separation.

The product of the factors Asfsyd was found to be a satisfactory index and very close indication of the