DEPARTMENT OF

UNIVERSITY OF HAWAII
DEPARTMENT OF CIVIL AND ENVIRONMENTAL ENGINEERING

Vertical and Lateral Load Tests of
A Nine-Panel Flat-Plate Frame

BACKGROUND INFORMATION:

Name Shyh-Jiann Hwang                                                             National Taiwan University of Science and Technology
Phone 886-2-2737658                                                                P.O. Box 90-130, Taipei
Email hwang@hp.ct.ntust.edu.tw                                                  Taiwan 10672, ROC

Name Jack P. Moehle                                                                  University of California, Berkeley
Phone (510) 642-0697                                                                1301 South 46th Street
Email moehle@ce.berkeley.edu                                                    Richmond, CA 94804-4698

SUMMARY

General

A test of a reinforced concrete flat-plate floor constructed at four-tenths of full scale is reported. The nine-panel model comprised a slab supported on columns without beams, drop panels, or slab shear reinforcement. A portion of the slab was designed for gravity and wind load in accordance with ACI 318-83, whereas the remainder was designed for moment redistribution not permitted by that code. Gravity load tests provide data on structural responses at the service load level. Lateral load tests provide data on behavior for loadings ranging from the service load level to the ultimate load level.

General Testing Program

Prototype Slab

The test slab geometry and details were derived from an idealized prototype flat-plate floor. The prototype is located at an intermediate level of a multistory office building and has three bays in each direction. The slab span is 15 ft (4.6 m) in one direction and 22.5 ft (6.9 m) in the other. Slab thickness is 8 inches (203 mm) and story height is 10 ft (3.0 m). Some columns have square cross section, whereas others have rectangular cross section.

Test Slab

The 0.4-scale test slab (Figure 1) was geometrically similar to the prototype slab. The test slab had thickness of 3.2 inches (81 mm). Center-to-center spans in the two principal directions were 6 and 9 ft (1.8 and 2.7 m). The columns extended 12 inches (305 mm) above the slab and 48 inches (1220 mm) below the slab. Reaction transducers were inserted beneath the columns to provide "pinned" connections. (In the prototype building, the major wind shear in an intermediate story is due to wind load applied to the stories above the intermediate floor, and inflection points are located near column midheights. It was not feasible to replicate these boundary conditions in the test slab. Instead, lateral loads were applied directly to the floor slab, resulting in slab membrane forces larger than those that would be typical in an intermediate floor. To reduce the resulting slab membrane forces and satisfy other experimental constraints, the lower columns were required to extend below the location corresponding to the idealized inflection point of the prototype. A consequence is that the columns are more flexible than they would be in a true replica model.)

Prototype Design and Specimen Design

Prototype design gravity loads consist of self weight, 10 psf (480 Pa) superimposed dead load, and 40 psf (1920 Pa) live load. The controlling design lateral loads are due to wind. Design moments at interior connections due to combined wind and gravity load effects are approximately 1.5 times those due to gravity load alone. Design seismic loads produce slab design moments less than those obtained from the design wind loading. (In regions of high seismicity, current design practice allows use of slab-column framing provided alternate structural elements are capable of resisting the entire design seismic loading and provided the slab is capable of sustaining gravity loads under the imposed lateral deformations.)

Material Properties

Concrete design compressive strength was 3000 psi (21 MPa) and steel design yield strength was 60 ksi (414 MPa).

Slab concrete (normal weight) had nominal maximum aggregate size of 3/8 in. (10 mm), with measured mean compressive strength of 3160 psi (21.8 MPa), splitting tension strength of 370 psi (2.6 MPa), and secant modulus of 2590 ksi (17860 MPa). Slab reinforcement was No. 2 deformed bars. The nominal Grade 60 slab reinforcement was from three different heats, and was placed systematically in three groups as follows: reinforcement with yield and ultimate strengths of 64.4 and 86 ksi (444 and 593 MPa), respectively, was placed in the column strips in the long direction; that with strengths of 66.2 and 101 ksi (456 and 696 MPa) was placed in the column strips in the short direction; and that with strengths of 70.7 and 102 ksi (487 and 703 MPa) was placed in the remaining locations.

Test Setup

The test slab was loaded with a combination of vertical and lateral loads. The test sequence is depicted in Figure 2. Details are provided below.

In the first test, SW, the shores were removed so that the structure was loaded by its self weight. Because the model was constructed at 40 percent of full scale, the internal stresses produced by this loading are expected to be only about 40 percent of those for the full-scale slab loaded by its self weight. This gravity loading was maintained for tests LAT1 and LAT2.

In test LAT1, two reversible hydraulic actuators applied loads to displace the structure in the North-South direction. The loads were applied through a yoke that applied compressive load to the slab edge at two points, one midway between frame lines a and b, and the other between lines c and d (Figure 1). In test LAT2, actuators applied load in the East and West directions with reaction points midway between lines 1 and 2 and lines 3 and 4. The displacement history involved two cycles in each direction at maximum drift ratio of 1/800 (0.125% drift), where drift ratio is defined as the lateral displacement divided by the column height measured from the pin supports to the slab mid-depth.

In test LEAD (Figure 2), 378 lead weights (each weighing approximately 100 pounds (445N)) were stacked in two layers atop the slab (Figure 3). The total of the slab self weight and the lead weights was 118 psf (5650 Pa), which is the prototype slab self weight, 10 psf (480 Pa) dead load, plus twenty percent of the design service live load. This load was maintained throughout the rest of the test program. Tests LAT3 and LAT4 followed with the same lateral displacement history as LAT1 and LAT2.

In test CONSTR a 55 psf (2630 Pa) gravity load was added to individual panels, one panel at a time. This load intensity is in the range expected for construction loading. The load was removed for all subsequent tests.

Subsequently lateral load tests as described above were repeated for tests NS800 and EW800 (0.125% drift) through NS25 and EW25 (4.0% drift) (Figure 2), where the letters NS and EW refer to the loading direction (either North-South or East-West, respectively) and the numerals refer the reciprocal of the target drift ratio. The history involved two displacement cycles in one direction to a given amplitude, followed by two cycles in the transverse direction to the same amplitude. The testing then graduated to a similar set of cycles at double the displacement amplitude.

The final testing PF20 involved two displacement cycles in each direction to lateral drift of 1/20 (5.0% drift). Most of the connections had punched during previous load cycles; the purpose of test PF20 was to examine the vertical load carrying integrity of the slab following initial failure.

Instrumentation

Four horizontal displacement transducers were used in each principal direction to record the lateral drift of the slab mid-depth. The two actuators in each direction were controlled to avoid slab twisting about a vertical axis.

Column shears and axial loads in three orthogonal directions were measured by sixteen reaction transducers (one for each column). Each reaction transducer was a tripod consisting of a spherical bearing connector (resembling a refined trailer hitch) supported atop three inclined steel legs. The spherical bearing connector was designed to act as a three-dimensional hinge support, capable of restraining displacements in all directions. The sensitivity of reaction transducers was calibrated by applying skewed forces representative of those developed during the test. For transducers beneath exterior columns, the mean error of measuring horizontal shear was 0.06 kip (0.27 kN) and that of vertical force was 0.17 kip (0.76 kN). For transducers beneath interior columns, the corresponding errors were 0.12 kip (0.53 kN) and 0.14 kip (0.62 kN). The Appendix describes the data reduction procedure used to determine column reactions.

Local deformations were measured using an additional 65 displacement transducers. A steel grid was supported on the upper columns and served as a rigid reference frame for slab deformation measurements (Figure 3). The rotation of the upper column stubs relative to the steel grid was measured; this measurement is reported as the rotation of the slab-column joint relative to a horizontal plane. Vertical deflections of the slab relative to the steel grid were measured at the midpoint of each panel and at the midspans of each column line.

A total of 42 electrical-resistance strain gages were placed on top slab reinforcing bars around the perimeter of the test slab, with at least one gage at each exterior connection. An additional 26 gages were placed on bottom bars at midspans of the column strips and middle strips of the exterior panels.

Conclusions

An experimental study of a reinforced concrete flat-plate frame subjected to service level vertical loads and service and ultimate level lateral loads suggests the following conclusions of general interest to design of slab-column frames.

The design, which followed in part the requirements of ACI 318-83 (essentially the same as those of ACI 318-95), produced a structure that responded under service loads within commonly accepted limits for cracking and vertical deflections. Furthermore, half of the structure was designed for negative ultimate moments reduced by as much as thirty percent to account for possible moment redistribution; redistribution resulted in lower connection stiffness and higher reinforcement strains, but otherwise the performance was deemed acceptable. Neither the Direct Design Method nor the Equivalent Frame Method of ACI 318-83 accurately reproduced the moment fields of the slab under service gravity loads; however, either could have been used in design to produce proportions and reinforcement similar to those used in the slab. Hence, either would have produced a slab having acceptable service load performance.

Both the effective beam width and equivalent frame models overestimate the slab stiffness at interstory drift ratio of 1/400 (0.25% drift). If the beam stiffness of either model is reduced by a factor of one-third as has been recommended previously to account for service-level cracking, a conservative estimate of stiffness is obtained.

The ACI design strength of the slab was achieved under lateral loading. However, resistance at reasonable drifts [defined arbitrarily at drift ratio 1/100 (1.0% drift)] was less than the ACI design strength, and the resistance deteriorated markedly with few load cycles.

The test slab was deformed under reversed cycle lateral loads to drift ratio of 1/25 (4.0% drift) before occurrence of punching failures. The large drift capacity is attributed to the relatively low gravity load nominal shear stresses. The slab did not collapse following punching, apparently because the bottom slab bars continuous over the columns suspended the slab after punching.

A variety of slab edge details produced nearly identical behavior. Details included combinations of the following: (a) slab bars near the column and perpendicular to the slab edge being in the inner layer of the bar mat at some locations and in the outer layer at others, (b) slab bars near the column and perpendicular to the slab edge having spacings ranging from 5/8 to 2 times the slab thickness, and (c) middle strip top slab bars provided perpendicular to the edge in some locations and absent in others. The effects of such variations were not apparent at service or ultimate load levels.

Figures