U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590
202-366-4000


Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

FHWA Home / Highways for LIFE / Technology Partnerships / Bridge Technology / Precast Bent System for High Seismic Regions: Laboratory Tests of Column-to-Footing Socket Connections

Precast Bent System for High Seismic Regions: Laboratory Tests of Column-to-Footing Socket Connections

CHAPTER 5. MEASURED RESPONSE

This chapter documents the responses measured by instruments mounted on and within the test specimens.

Material Properties

Material tests for the concrete and reinforcement used in the test specimens were conducted in the University of Washington Structural Research Laboratory. The specified 28-day concrete strength for both the columns and the footings was 4 ksi, and the reinforcement conformed to ASTM A706, with the exception that smooth wires conforming to ASTM A82 were used where a direct scaling resulted in a deformed bar smaller than No. 3. A detailed description of the concrete mix and the material test results are presented in appendix B.

Concrete Strength

Specimens SF-1 and SF-2 had segmental precast columns. The column segments were fabricated off campus at a construction site in the City of Redmond, they were cast on the same day, and they were delivered 1 week later to the University of Washington Structural Research Laboratory. In the laboratory the columns were embedded in cast-in-place spread footings. Specimen SF-3 was built entirely in the Structural Research Laboratory.

For both the precast column and cast-in-place footing pours, concrete test cylinders were prepared (4-inch by 8-inch cylinders for SF-1 and SF-2, and 6-inch by 12-inch cylinders for SF-3). For specimens SF-1 and SF-2, the cylinders were kept adjacent to the corresponding element, whereas the cylinders for specimen SF-3 were stored in a fog room. Cylinders were tested in pairs at 7 days, 28 days, and on the test days for specimens SF-1 and SF-2. Column concrete strength tests at 7 days were not performed because of a shortage of test cylinders. Cylinders for specimen SF-3 were tested at 7 days, 14, days, 21 days, 28 days, and on the test day. Table 5 summarizes the average concrete strength for both the precast columns and cast-in-place footings on the test day.

Table 5. Average concrete strength on test day.
Specimen Column Concrete Footing Concrete
Strength (psi) Days Strength (psi) Days
SF-1 4,995 113 6,505 91
SF-2 5,496 129 6,764 107
SF-3 7,935 120 7,905 113

Grout Strength

The grout used was an unsanded, silica fume, commercially available product, which is designed to be used as a flowable grout. The grout was mixed in a high-speed shear mixer, and water was added according to the manufacturer's instructions. The grouting operations for the precast column interfaces for specimens SF-1 and SF-2 were performed on the same day. During this operation, four standard 2-inch by 2-inch test cubes were taken, and two cubes were tested on each subassembly's test day. The ducts were grouted four days later, at which time, an additional four test cubes were prepared. Table 6 summarizes the average grout strength measured from those tests. The table shows that the grout strength used in the ducts of specimen SF-2 was noticeably lower than the strength in other locations. This is attributed to the fact that one cube was mishandled before the compression test. If that cube is excluded from the average, the grout properties for SF-1 and SF-2 were nearly identical.

Table 6. Average grout strength on test day.
Specimen Interface Grout Grout in Ducts
Strength (psi) Days Strength (psi) Days
SF-1 13,075 60 12,500 56
SF-2 13,025 77 11,175 73

Reinforcement

Tension tests were performed on No. 7, No. 6, No. 5, No. 4, and No. 3 mild reinforcing bars, as well as 2-gauge and 3-gauge wires. Load and strain were recorded using a LabView data acquisition system. Stress was calculated using the nominal bar areas. A laser extensometer was used to measure bar elongation over a gauge length of 1 inch for specimens SF-1 and SF-2. For specimen SF-3, elongations during the tension tests were monitored with an 8-inch gauge length mechanical extensometer.

Test coupons had definite yield plateaus for all of the bars except the wires and the No. 3 bars in specimens SF-1 and SF-2. In these tests, the strain hardening of the bars was measured. In specimen SF-3, the coupons were loaded until the steel began to yield and the load dropped. The extensometer was then removed to prevent damage to the equipment. The coupon was then loaded until failure, after which time the length was measured to obtain a strain at failure. Therefore, the line from yield to failure shown in the SF-3 plots does not represent a measured response. Instead, it is only there to provide a visual connection between the yield and failure point. To be consistent with previous tests, the yield strain was taken as 0.35 percent strain to avoid any subjectivity into the test results. Table 7 summarizes the average measured yield stress, ultimate stress, and nominal area of the reinforcement.

Table 7. Measured mild reinforcement properties
Bar Anominal

(in2)
Specimens SF-1 and SF-2 Specimen SF-3 Use/Location
fy (ksi) fu (ksi) fy (ksi) fu (ksi)
No. 7 0.60 N/A N/A 66.2 95 SF-3/footing flexural reinforcement
No. 6 0.44 61.6 86.1 59.2 88.2 All/column longitudinal bars
No. 5 0.31 63.3 91.4 66.1 108 SF-1, SF-2/footing flexural reinforcement

SF-3/shear friction steel
No. 4 0.20 63.7 90.4 65.9 91.6 SF-1, SF-2/shear friction reinforcement

SF-3/ties
No. 3 0.11 65.2 94.5 62.4 99.9 SF-1, SF-2/top footing flexural reinforcement
2 GA. 0.054 63.4 86.4 N/A N/A SF-1, SF-2/ties
3 GA. 0.047 59.4 73.6 N/A N/A All/column spiral

All A706 bars exceeded their minimum specified yield stress of 60 ksi except the No. 6 column bars in specimen SF-3. These bars had a yield stress of 59.2 ksi, only 1.3 percent below the specified value. In both the SF-1/SF-2 and SF-3 tension tests, the No. 6 column bars did not reach a yield stress of 68 ksi, which is the expected value that AASHTO Seismic Guide Specifications recommended to be used (for A706 reinforcement) in the calculation for the column overstrength moment. The wires did not meet the minimum specified yield stress of 70 ksi. This poor performance of the wires is attributed to the fact they had to be straightened out before tensile testing.

Friction Correction

The measured horizontal load was corrected using a friction model and the recommendations proposed by Brown.(22) This correction has been done previously in research at the University of Washington.(7) The test setup in the Baldwin Universal Testing Machine introduced a small friction force into the system by two mechanisms: rotation between the greased steel-to-steel spherical element in the swivel head bearing, and horizontal sliding between the bearing's top flat plate and the channel attached to the Baldwin head. The friction component in the channel was minimized by placing sheets of silicon-greased Teflon between the bearing and the smooth stainless steel plates in the channel. The model consists of a bilinear spring with a spring stiffness, of 60 kips/in., and has a maximum friction force, of where is the coefficient of friction and is the target axial load. The coefficient of friction can be approximated as shown in figure 24.

Mu subscript eff equals the sum of Mu subscript flat plus the product of Mu subscript curved times the quotient of R divided by L subscript total.

Figure 24. Equation. Coefficient of friction.

In this equation, μflat is the flat friction between the top swivel head bearing plate and the sliding channel, μcurved is the friction between greased steel-to-steel spherical element, R is the radius of spherical element, and Ltotal is the height between footing surface and top of bearing. When R« Ltotal, then μeffμflat.

Brown's model was implemented assuming μeff = 1.6 percent. For an axial load of 159 kips, the corresponding maximum resistance in the sliding channel was 2.54 kips, which was about 5 percent of the maximum applied load in all tests.

Moment-Drift Response

The moment at the base of the column was calculated using the equation in figure 25, for which the variables are illustrated in figure 26.

M equals the sum of the product h subscript 1 times H plus the product of the DELTA subscript 1 times the quotient of h subscript 1 divided by h subscript 2, end quotient, times P.

Figure 25. Equation. Calculation of the moment at the base of the column.

In this equation, M is the calculated base moment, h1 is the height from the top of the spread footing to the line of action of the lateral load, H is the corrected lateral load that takes into account both the friction force between the bearing and the sliding channel, and the greased steel-to-steel spherical element on bearing, Δ1 is the lateral displacement where the lateral load, H, is applied, h2 is the height from the top of the spread footing to where the axial load, and P is the axial load applied by the Baldwin Universal Testing Machine. Δ2 was not measured but estimated to be Δ1(h2/h1.

Figures 27 to 29 show the responses of the three test specimens in terms of moment vs. drift ratio.

This drawing defines forces and displacements acting on a column used in figure 25 to calculate moment response.

Figure 26. Diagram. Displacements and forces on test specimen used in figure 25.

Moment versus drift ratio plot for spread footing test specimen SF-1.

Figure 27. Graph. Specimen SF-1 moment-drift response.

Moment versus drift ratio plot for spread footing test specimen SF-2.

Figure 28. Graph. Specimen SF-2 moment-drift response.

Moment versus drift ratio plot for spread footing test specimen SF-3.

Figure 29. Graph. Specimen SF-3 moment-drift response.

All three specimens shared the same column geometry and reinforcing pattern. The measured responses of the three specimens were similar but not identical. The responses of both specimens SF-1 and SF-2 were stable to large drifts (at about 6 percent), at which point the lateral-load resistance dropped rapidly. By contrast, the strength of specimen SF-3 decayed more gradually, starting at 2.5 percent.

As shown in table 8, the stiffnesses of the columns (measured at the force corresponding to first yield of the reinforcement) were similar for all the specimens. However, the columns were slightly stiffer in the south direction of loading for specimens SF-1 and SF-2, which was the direction in which they were first loaded.

The maximum base moments were approximately 3,100 kip-in. for both specimens SF-1 and SF-2. Specimen SF-3 was approximately 10 percent stronger than the other two specimens. Material tests showed that the column longitudinal bars had virtually the same yield and ultimate strength for all three specimens (see table 7). In contrast, the concrete cylinder tests showed that the concrete compressive strength was about 50 percent higher (table 5) for specimen SF-3 than for specimens SF-1 and SF-2.

Table 8. Moments and drift ratios at maximum and 80 percent of maximum resistance.
Points of Interest Specimen SF-1 Specimen SF-2 Specimen SF-3
North Direction South Direction North Direction South Direction North Direction South Direction
Secant Stiffness at Initial Yield (kip/in.) 108 114 116 132 132 126
Maximum Column Base Moment (kip-in.) -3,073 3,091 -3,113 3, 065 -3,315 3,392
Drift Ratio at Maximum Column Base Moment (percent) -2.61 1.95 -1.45 1.38 -2.69 2.64
80 Percent of Maximum Column Base Moment (kip-in.) -2,417 2,473 -2,490 2,452 -2,652 2,714
Drift Ratio at 80 percent of Maximum Base Column Moment (percent) -6.79 7.09 -6.88 8.69 -8.26 10.5

As the columns were cycled, the longitudinal bars went through more inelastic strains and stress reversals. The definition of failure is often taken as the state at which the effective lateral load resistance measured drops below 80 percent of the maximum resistance obtained earlier in the test. In specimens SF-1 and SF-2, that state occurred at approximately 7 percent drift and corresponded to the onset of buckling of the longitudinal bars (+7.1 percent/-6.9 percent in SF-1 and +7.2 percent/-6.9 percent in SF-2). The 80 percent milestone occurred slightly later (at +8.7 percent) in specimen SF-2's south direction. Ultimately, the bars fractured in specimens SF-1 (at +10.6 percent) and SF-2 (at +10.6 percent), at which point the lateral-load resistance of the columns decreased abruptly.

Specimen SF-3 maintained its strength above 80 percent of the maximum to a drift ratio 8.3 percent when being pushed in the north direction ("valley") and 10.5 percent drift when pushed to the south ("peak"). The column bars did not fracture in specimen SF-3. Instead, the column punched through the footing before bar fracture.

Effective Force

The effective force acting on the specimens was calculated by dividing the moment at the base of the column by the height from the top of the spread footing to the line of action of the lateral load. The equation in figure 25 is divided by to give the equation shown in figure 30.

F subscript eff equals the sum of H plus P times the quotient of DELTA subscript 1 times h subscript 2 divided by h subscript 1 squared.

Figure 30. Equation. Effective lateral force.

Figures 31 through 33 show similar plots to figures 27 through 29, but they are expressed in terms of the effective force and displacement.

Effective force versus displacement plot for spread footing test specimen SF-1.

Figure 31. Graph. Specimen SF-1 effective force-displacement response.

Effective force versus displacement plot for spread footing test specimen SF-2.

Figure 32. Graph. Specimen SF-2 effective force-displacement response.

Effective force versus displacement plot for spread footing test specimen SF-3.

Figure 33. Graph. Specimen SF-3 effective force-displacement response.

Table 9 summarizes the maximum effective force (MEF) resistance, 80 percent of the maximum, and their corresponding displacements.

Table 9. Effective force and displacement at maximum and 80 percent of maximum resistance.
Points of Interest Specimen SF-1 Specimen SF-2 Specimen SF-3
North Direction South Direction North Direction South Direction North Direction South Direction
MEF (kips) -51.2 51.5 -51.9 51.1 -55.3 56.5
Displacement at MEF (in.) -1.57 1.17 -0.87 0.83 -1.61 1.58
80 percent of MEF (kips) -40.3 41.2 -41.5 40.9 -44.2 45.2
Displacement at 80 percent of MEF (in.) -4.07 4.25 -4.13 5.21 -3.57 6.33

Distribution of Column Curvature

Average column curvatures were determined from the column rotations at various heights. These were obtained from the relative displacements between the ends of threaded rods embedded horizontally in the column at about 2, 7, 12, and 18 inches above the footing surface. The average curvature within a column segment between rods is plotted at the midpoint of the segment. The curvatures were calculated using the equation in figure 34.

Phi subscript i equals the quotient of Delta subscript i,N minus Delta subscript i,S, end of sum, that sum divided by L subscript I, that quotient divided by H subscript i.

Figure 34. Equation. Calculating the average curvature.

In this equation, φi is the calculated average curvature, δi is the displacement of the curvature on one side (north or south) at particular height above the spread footing surface, Li is the horizontal length between north and south potentiometers, and Hi is the initial vertical distance between the curvature rods.

Figures 35 to 37 show the average column curvature for selected drift ratios versus height above the spread footing surface.

Column height versus average column curvature along the height of the column for various drift levels.

Figure 35. Graph. Specimen SF-1 average column curvature.

Column height versus average column curvature along the height of the column for various drift levels.

Figure 36. Graph. Specimen SF-2 average column curvature.

Column height versus average column curvature along the height of the column for various drift levels.

Figure 37. Graph. Specimen SF-3 average column curvature.

All potentiometers in specimens SF-1 and SF-2 worked well up to a drift ratio of 2 percent. At larger drifts some potentiometers measured only a small deformation, which suggests that the measurements were no longer accurate. At no time during the tests were potentiometers noted to lose contact in a way that might explain this behavior. This behavior was not observed for specimen SF-3, which was instrumented with new potentiometers. For SF-1 and SF-2, curvatures above a drift ratio of 2 percent are not reported in this report because of these discrepancies between potentiometers.

The curvature distribution was similar for all three specimens. The behavior of the column-to-foundation connection in all specimens was similar to that expected in cast-in-place construction; the deformation was concentrated at the base of the column. As expected, the distribution was super-linear, even though the moments were distributed linearly. This behavior reflects the fact that the response is non-linear in the plastic hinge region near the column base and essentially linear above that.

Column Splice

The precast columns for specimens SF-1 and SF-2 were constructed segmentally, whereas the column in specimen SF-3 was cast in a single piece. For the segmental columns, the column splice was located 20 inches above the spread footing surface. This location was where the moment was expected to just reach the yield moment, My, when the overstrength moment, Mpo, was reached at the column base. This location was determined for the smallest axial load expected on the prototype connection, corresponding to a scaled-down value of 87.5 kips. The splice would have been closer to the footing with a higher axial load; about 18 inches for an axial load of 159 kips. For SF-1 and SF-2, one potentiometer was mounted on each side of the column to monitor potential crack opening, as shown in figure 38. The measured splice interface openings are plotted in figures 39 and 40.

Photo shows a potentiometer located on the north side of the precast column intended to measure potential crack opening in the column splice region.

Figure 38. Photo. Crack opening measurement pot.

Splice opening versus drift plot. A solid line represents a north face splice opening, and a dashed line represent a south face splice opening.

Figure 39. Graph. Specimen SF-1 splice interface opening.

Splice opening versus drift plot. A solid line represents a north face splice opening, and a dashed line represent a south face splice opening.

Figure 40. Graph. Specimen SF-2 splice interface opening.

When axial loads were applied to the specimens, the columns shortened. The recorded average shortening within the splice regions for SF-1 and SF-2 were recorded over a gauge length of 2.75 inches, which was the center-to-center distance to the aluminum plates mounted on the specimens (see figure 38). The applied axial loads and the corresponding measured average axial strains are reported in table 10. For a load of about 159 kips, the measured average axial strains were about 0.001. Assuming that the elastic modulus can be approximated as 57,000 , the expected strains would be about 0.00012, differing by a factor of eight. For a load of about 240 kips, the measured axial strains ranged from 0.0011 to 0.0012, in comparison with the expected axial strains of about 0.00019. It is possible that the approximately ½-inch grout layer was more deformable than the concrete.

Table 10. Axial load and strains across and near splice interfaces.
  Specimen SF-1 Specimen SF-2
Loading 1.0DL+1.0OT* 1.25DL+1.75LL 1.0DL+OT* 1.25DL+1.75LL
Axial Load (kips) 159.4 241.7 159.2 242.2
Measured Axial Shortening (Averaged LVDTs) (in.) 0.0030 0.0033 0.0025 0.0029
Average Axial Strain 0.0011 0.0012 0.00091 0.0011
Calculated Axial Strain 0.00013 0.00019 0.00012 0.00018
Ratio of Measured/Calculated Axial Strain 8.5 6.3 7.6 6.1
Measured Average Axial Strain in Bars near Interface 0.00027 0.00035 0.00029 0.00038
Ratio of Measured/Calculated Axial Strain 2.1 1.8 2.4 2.1
*OT = overturning

For the lateral-load tests, the maximum extreme compressive strain calculated from the measured data was 0.002 on the north side of SF-2. In both tests, cracks appeared at the splice early, but they were small and closed after each cycle. The largest opening was 0.045 inches in SF-2 and corresponded to a rotation at the splice of about 0.24 percent. This value corresponded to about 2.1 percent of the total rotation at the location where lateral load was applied. There were no indications of damage at the interface between the column segments. At the end of cyclic testing, the maximum axial shortening measurements with these gauges were about 0.0028 inches for specimen SF-1 and 0.0052 inches for specimen SF-2, indicating that the splice interfaces had closed.

Strains in Column Longitudinal Bars

The longitudinal bars in the columns were gauged as described in chapter 3. In all specimens, symmetry of the column longitudinal bars was utilized; therefore, only the N-NE and S-SW longitudinal column bars were strain gauged. Gauges were attached on the bars in pairs at five locations in specimens SF-1 and SF-2:

  • 18 inches above the spread footing surface.
  • At the column-to-footing interface.
  • 7 inches and 15 inches below the footing surface to verify reinforcement development length.
  • 20 inches below the footing interface by the longitudinal terminators to estimate the extent to which the anchors were engaged.

Specimen SF-3 had a shallower spread footing than specimens SF-1 and SF-2 and did not have a column splice, so the strains in SF-3 were only monitored in two locations:

  • At the column-to-footing interface.
  • 7 inches below the footing surface.

Strain Profiles along the Height of Specimen

The strain distributions over the height of the S-SW bar at various drift levels are shown for all three specimens in figures 41 to 43. In general, the reported strains for SF-1 and SF-2 at each location correspond to the average of the strain readings on each side of the bar at a particular location. Strains reported at the column-to-footing interface (0-inch height) in specimens SF-1 and SF-2 are from one gauge only because the data acquisition system capped the readings at a strain limit of 0.011. Similarly, the strain readings were not averaged for specimen SF-3, because the gauges facing the cover of the column did not produce reliable strain gauge readings. Strains are plotted up to 3 percent drift, and this drift is less than the maximum drifts reached by the specimens.

The strain measurements indicate that the bars at the interface yielded at a drift ratio of about 0.5 percent for all three specimens.

The strain profiles for specimen SF-1 and SF-2 were nearly identical. The strains were largest at the column-to-footing interface and decreased down into the spread footing and up in to the column, as expected. By contrast, specimen SF-3 showed much lower strains at the interface and significant post-yield strains at the bottom end, next to the terminators. First yield at the terminators occurred at 0.49 percent drift. Furthermore, gauges were placed on the bars in pairs, yet at some locations one gauge showed compression while the other showed tension.

Column height versus average longitudinal bar strain for various drift values.

Figure 41. Graph. Strain profiles in S-SW bar in specimen SF-1.

Column height versus average longitudinal bar strain for various drift values.

Figure 42. Graph. Strain profiles in S-SW bar in specimen SF-2.

Column height versus average longitudinal bar strain for various drift values.

Figure 43. Graph. Strain profiles in S-SW bar in specimen SF-3.

These measured bar strains in specimen SF-3 are inconsistent with the observed behavior. The column was seen to undergo many cycles of inelastic bending deformations just above the interface before it finally failed by combined punching and moment transfer in the connection region. A possible explanation is that the gauges were mislabeled or connected wrongly to the data acquisition system. Results of individual strain gauges for all three specimens are reported in more detail in the section regarding the column longitudinal bar strain histories within the footing.

Residual strains became apparent at about 0.8 percent drift in the strain gauges at the column-to-footing interface. The S-SW bar in specimen SF-3 did not reach yielding at the interface throughout the test. After 2.4 percent drift for specimen SF-1, 2.6 percent drift for specimen SF-2, and 0.98 percent drift for specimen SF-3, strains began to exceed the reading capacity of the data acquisition system (25000με for specimens SF-1 and SF-2, and 12000με for specimen SF-3). Maximum strains that the system was capable of reading ranged from +/-0.011 to +/-0.025. Before the strain gauges had delaminated from the reinforcement, strains would be capped at these limits and the readings would be within the recording range again once strains were lower.

Strain Histories for Bars near Splice

Figure 44 shows the strain histories for the strain gauges located 2 inches below the segment splice (18 inches above the spread footing). These gauges indicate that the gauged N-NE bars started to yield at 0.8 percent drift for specimen SF-1 and 1 percent drift for specimen SF-2. For the S-SE bars it was about 3.2 percent drift for SF-1 and 1.2 percent drift for SF-2.

Strain versus drift plots 2 inches below the splice interface in specimens SF-1 and SF-2 (SF-1 on the left and SF-2 on the right). The top row shows N-NE bar strains, and the bottom row shows S-SW bar strains.

Figure 44. Graphs. Strain-drift relationship 2 inches below the column splice interface.

All bars except the S-SW bar in specimen SF-1 yielded considerably near the column splice once spalling developed. This difference was attributed to the fact that the full spall height on the south side for specimen SF-2 was 12 inches while it was 8 inches for SF-1. However, on the north side for both specimens, the full spall height was 12 inches. This suggests that the splice bars were yielding.

Column Longitudinal Bar Strain Histories in Footing

Figures 45 through 47 show for all three specimens the strain-drift relationship at locations within the footing. In the figures, each strain gauge is plotted individually. The gauge marked "In" refers to the gauged facing towards the column core, and the one marked "Out" faced the column concrete cover. Strains were plotted until voltage spikes were detected frequently, at which point the gauges were deemed unreliable. Two strain gauges were damaged from the very beginning and therefore are not included in the plots.

Strain versus drift plots below column-to-spread footing interface in specimens SF-1 and SF-2 (SF-1 on the left and SF-2 on the right). The top row shows bar strains at the interface, and the bottom row shows bar strains 20 inches below the interface.

Figure 45. Graphs. Strains in N-NE bars in specimens SF-1 and SF-2 at various heights below the interface.

Strain versus drift plots below column-to-spread footing interface in specimens SF-1 and SF-2 (SF-1 on the left and SF-2 on the right). The top row shows bar strains at the interface, and the bottom row shows bar strains 20 inches below the interface.

Figure 46. Graphs. Strains in S-SW bars in specimen SF-1 and SF-2 at various heights below the interface.

Strain versus drift plots below column-to-spread footing interface in specimen SF-3. The top row shows bar strains at the interface (N-NE bar on the left and S-SW bar on the right), and the bottom row shows bar strains 7 inches below the interface.

Figure 47. Graphs. Strains in N-NE and S-SW bars in Specimen SF-3 at various locations below the interface.

Strains recorded in specimens SF-1 and SF-2 were similar. Pairs of gauges were attached at the interface and at 7, 15, and 20 inches below it. The gauges at 20 inches were adjacent to the terminators. The N-NE bars and the S-SW bars showed strains of similar magnitudes, confirming that the overall behavior of the two specimens was similar, as shown by the load-displacement curves. The largest strains recorded were at the column-to-footing interface, as was expected. Furthermore, at the interface, the readings of the two individual gauges differed, especially at higher drifts, implying the presence of bending as well as tension. The pairs of strain gauges located below the column-to-footing interface gave almost identical readings, implying pure tension. At 7 inches below the interface the bars almost yielded, at 15 inches the peak stress was approximately 10 ksi, and by 20 inches the peak stress was approximately 3 ksi. (The highest strain, for the "inside" gauge of specimen SF-1, is higher than this, but its value is an outlier and is inconsistent with the "outside" gauge reading at that location.) The values suggest that the bars were fully anchored by that depth and that the mechanical anchors served as a secondary anchorage.

The strains in the N-NE and S-SW bars in specimen SF-3 are somewhat erratic and do not follow the pattern that was expected from the observed behavior in the test. In specimen SF-3 the column underwent inelastic cyclic bending before the connection region failed in combined punching shear and moment transfer. That inelastic bending implies larger strains at the interface and smaller strains near the mechanical anchors. At larger drifts, the strain was expected to increase at both locations.

The measured strains in specimen SF-3 display three trends. First, the strains at the interface are general smaller than those near the terminators, 7 inches below the interface. This trend implies negative bond stresses, which are thermodynamically impossible. Other explanations must therefore be sought, and mislabeling of the gauges is the most likely one. Second, most of the gauge pairs show large differences between the two readings, implying significant bending of the bar. Such bar bending is plausible, but no independent verification measurements were available to confirm it. Third, the bending appears to be more severe near the terminator than at the interface. This is the opposite of the trend seen in specimens SF-1 and SF-2, and is difficult to explain, except by gauge mislabeling.

It was therefore concluded that the gages were most likely mislabeled, in which case the data cannot be used.

Footing Strain Corrections

Strains reported in the spread footings for specimens SF-1 and SF-2 are the mechanical strains derived from the measured strain values. The strains in the bottom mat were so small that they were affected strongly by thermal effects (see figure 48). This behaviour was not observed in specimen SF-3. In specimens SF-1 and SF-2 it was believed to have been caused by heat from the lights acting on the strain gauges lead wires. To determine the mechanical strain during cyclic testing, thermal strains needed to be accounted for. The mechanical strains were obtained by calculating the difference between the strain value measured at the peak of the cycle and the strain value when the moment in the column was zero. It was assumed that the change in thermal strain during that time (about 20 seconds) was negligble. The largest rate in thermal strain change occurred in specimen SF-1 between data counts 40681 and 41672 at an average rate of 0.083 microstrain/count. If this rate were applied to the period of loading (94 counts, going from "peak" to "valley"), it would imply a thermal strain change of 7.8 microstrain during that time. The calculated mechanical strain value was then added to the initial strain obtained when the specimen was loaded axially up to 159 kips before starting the cyclic test.

The chart shows displacement history together with measured strains in the footing. The chart illustrates strain changes due to thermal effects during temporary hold in loading.

Figure 48. Graph. Thermal effects in strain gauges.

Strains in Bottom Mat of Footing Reinforcement

The configuration of the bars in the footing and the locations of strain the gauges were shown in figure 13. The columns in all three specimens were designed to have the same flexural strength. Specimens SF-1 and SF-2 had the same spread footing depth of 22.5 inches, so the total amount of flexural steel was also the same in both specimens. However, it was distributed differently because specimen SF-1 contained slots in the bottom of the column to allow some of the bottom mat bars to pass directly beneath the column. In contrast, specimen SF-2 contained no slots, so the bars had to be moved laterally to pass by the sides of the column, where they were bundled with other bars already existing there. In addition, to satisfy AASHTO LRFD minimum bar spacing requirements and to provide crack control reinforcement, additional steel, of the same size and spacing as in the main mat (5 inches center-to-center), was provided in line with the column. Those bars were short and terminated at the column face.

Specimen SF-3 had a much thinner footing than the other two specimens (half of the column depth); thus, a great deal of flexural steel was required. The bottom mat was arranged to be similar to specimen SF-2, but modifications were needed due to the high amount of steel congestion. For example, the reduced footing depth resulted in a smaller effective width in which a much greater amount of steel was required. The consequence was the use of larger bars (No. 7 instead of No. 5) bundled at a spacing of 2.5 inches. Steel was placed in line with the column to satisfy AASHTO LRFD requirements. It was the minimum permitted and was terminated at the column face.

Strains in Bottom Bars in the North-South Direction (Loading Direction)

Longitudinal bars were gauged so that the distribution of strain across the footing could be determined. The northeast quadrant of the spread footings was equipped with gauges at a point 5.5 inches to the north of the column face, including some short bars. The gauges were placed on the side of the bar to minimize the effects of bending about a horizontal axis. They were all chosen to be on the east side of the footing because the response was expected to be symmetric in each principal direction. All strain gauges worked well except two—one located 17.5 inches away from column center in specimen SF-2 and the other located 27 inches away from the column center in specimen SF-3, which were damaged during casting. The bar strains at various drift levels are shown in figures 49 through 51.

. Strain versus location from column center for various drift levels. A black vertical line represents the effective width as defined by the AASHTO Seismic Guide Specifications.

Figure 49. Graph. Specimen SF-1 strain profiles in bottom mat of the footing.

Strain versus location from column center for various drift levels. A black vertical line represents the effective width as defined by the AASHTO Seismic Guide Specifications.

Figure 50. Graph. Specimen SF-2 strain profiles in bottom mat of the footing.

Strain versus location from column center for various drift levels. A black vertical line represents the effective width as defined by the AASHTO Seismic Guide Specifications.

Figure 51. Graph. Specimen SF-3 strain profiles in bottom of the footing.

The distribution of strains in the footing differed. However, the strains in all three specimens shared the common feature that they increased with drift up to about 1 percent drift, after which they were approximately constant. This occurred because the column yielded initially at about 0.5 percent drift and the force was nearly constant after 1 percent drift. The strain measurements in specimens SF-1 and SF-2 must be regarded as less reliable than those in specimen SF-3 because of the need to correct for thermal effects. The peak strains remained low up to 5 percent drift. In all specimens, the first yield in the column longitudinal bars was detected at about 0.50 percent drift.

In specimen SF-1, the strains consistently decreased with distance from the centerline of the specimen, as might be expected. After yielding of the first longitudinal bar, the moment introduced in the footing was nearly constant and therefore resulted a identical footing strain profiles after 0.5 percent drift for specimens SF-1 and SF-2. The biggest strain recorded at 5 percent drift was and was measured in the bars going through the slots. This strain value reflects the fact that the footing was essentially uncracked and remained elastic throughout the test. The lower strain in the bar 7.5 inches away from the column center is most likely due to the fact that it was a short bar that ended in front of the column. The strains increased in the set of bundled bars (12.5 inches away from center) and decreased rapidly in the remaining bars.

Strain profiles for specimen SF-2 did not have the two peaks that are present in the profiles for specimen SF-1. Recall that no bars passed beneath the column in specimen SF-2. The first two bars from the column center were short bars, and the following two were bundled. The strains in bars 7.5 inches and 12.5 inches away from the column center are three to four times smaller in specimen SF-2 than in specimen SF-1. This difference was already apparent during the application of the axial load. The strain profiles beyond one column diameter die out similarly in both specimens after that. The observation that the strains were all much less than the yield strain, and that the strains in short and long bars were almost the same, suggests that the concrete was not cracked and was carrying tension force.

The flexural reinforcement layout for specimen SF-3 was similar to the one in specimen SF-2, but the steel was heavier because the footing was shallower. In contrast to specimens SF-1 and SF-2, specimen SF-3's strain profiles consistently increased from the column center, and maximum strain was recorded outside the effective width, even though the all gauged bars were long ones. Strains increased constantly with drift ratio in specimen SF-3 which reflects the fact that the cracks were becoming more pronounced. At 5 percent drift, the maximum strain was about .

Implication of the Effective Width

The strain profiles shown in figures 49 to 51 allow the definition of the effective width to be evaluated. It is defined in the AASHTO Seismic Guide Specifications as the sum of the column diameter and two times the footing depth (beff = Dc+2Hf) and implies that only the bars inside it will experience significant tension strain and contribute to resisting the applied moment.(15) If the footing steel required for strength is more than the minimum, the concrete should be expected not to crack. Thus, it is reasonable to suppose that the definition of effective width is based on an uncracked section since specimens SF-1 and SF-2 did not display visible flexural cracks, and the lowstrains also suggest an uncracked condition. The evidence from their strains cannot be used to evaluate the AASHTO equation for beff. However, despite the uncertainty caused by the need for thermal corrections, the strain profiles displayed the expected reduction in strain with distance from the column centerline.

In specimen SF-3, the concrete cracked and the steel strains were larger, although they never reached yield. Thus, they are appropriate for evaluating the definition of beff. However, they show that the largest strains were in the bars outside the effective width, which is the opposite of what would be expected. Furthermore, specimen SF-3 was the only one of the three in which the effective width was significantly less than the total width, so the strains were expected to be strongly and unambiguously concentrated inside the effective width. The reasons for the anomalous behavior are unknown.

Strains in Bottom Bars in the East-West Direction

Strut-and-tie modeling suggests that compressive struts forming within the footing are more three dimensional in specimen SF-2 and specimen SF-3 than in specimen SF-1, which results in more tension in the transverse bars in the bottom mat.

In specimen SF-1, some of the longitudinal steel passed directly under the column, allowing the formation of a 2-D truss in the footing. In the other two specimens, no bars passed beneath the column, so any truss that formed would be necessarily 3-D and transverse bar forces would be necessary for equilibrium. Some transverse bars were gauged, as shown in figures 52 through 54, to evaluate the formation of such trusses.

Because the footing did not suffer visible flexural cracking in specimens SF-1 and SF-2, the bar strains were too small to permit a reliable evaluation. In each case they were approximately half the corresponding longitudinal bar strains, but they would were actually larger in specimen SF-2 than in specimen SF-1, contrary to what would occur if the expected truss had formed.

In specimen SF-3, the transverse bar strains implied a stress of approximately 10 ksi. This is large enough to confirm the existence of some 3-D behavior, but it is still only about one-quarter of the stress predicted by the truss model.

Strain versus negative drift plot at locations 8.5 inches and 18.5 inches from the column face.

Figure 52. Graph. Specimen SF-1 transverse strains in bottom mat of the footing.

Strain versus negative drift plot at locations 8.5 inches and 18.5 inches from the column face.

Figure 53. Graph. Specimen SF-2 transverse strains in bottom mat of the footing.

Strain versus negative drift plot at locations 8.5 inches and 18.5 inches from the column face.

Figure 54. Graph. Specimen SF-3 transverse strains in bottom mat of the footing.

Strains in Diagonal Bars

Diagonal bars were placed in both the top and bottom mats to play a role similar to that of "shear friction" steel, which in other situations would cross the interface between the precast and cast-in-place elements. The total amount of diagonal steel in specimens SF-1 and SF-3 was equal to the amount needed if the cohesion component of the AASHTO LRFD article 5.8.4 equation (5.8.4.1-3) for shear friction was ignored between the precast column and the cast-in-place footing. The difference between the two (SF-1 and SF-3) was that the longitudinal footing steel in specimen SF-3, running north and south, was accounted for when determining shear friction resistance. The amount of steel was reduced in SF-2.

In all specimens, two diagonal bars in the bottom mat and two diagonal trim bars in the top mat were gauged as shown in figure 13. The strains in the south bottom steel in specimens SF-1 and SF-3, and north bottom steel in specimen SF-3 as well are not reported because the gauges were damaged during casting. The strains in the diagonal bars are plotted in figure 55 through 57.

Strain versus negative drift plot for the diagonal steel placed around the base of the column for added confinement.

Figure 55. Graph. Specimen SF-1 strains in diagonal steel in footing.

Strain versus negative drift plot for the diagonal steel placed around the base of the column for added confinement.

Figure 56. Graph. Specimen SF-2 strains in diagonal steel in footing.

Strain versus negative drift plot for the diagonal steel placed around the base of the column for added confinement.

Figure 57. Graph. Specimen SF-3 strains in diagonal steel in footing.

Strains reported in specimens SF-1 and SF-2 are only the mechanical strains. The thermal strains were removed using the process described in the previous section on footing strain corrections. At 0 drift, the effect of only the axial load is noticeable. Each bar had one strain gauge on the side facing out from the column. When the column displaced northward in specimens SF-1 and SF-3 towards the gauges, the north-bottom gauges barely recorded any stress. However, strains increase for north-top and south-bottom bars as expected. The column was pushing out in those areas.

Overall, the strains measured in the diagonal steel were small. The maximum strain measured was in specimen SF-1 and it was about . However, the tension strains in the diagonal steel of specimens SF-2 and SF-3 were all less than 100 micro-strain, which corresponds to a stress of 3 ksi. The low strains in the steel, and the lack of cracking in the footings of specimens SF-1 and SF-2, imply that the shear-friction mechanism was never activated there. The column in specimen SF-3 punched through the footing, but the stresses in the bars were small enough that the diagonal steel contributed little to resisting the loads.

Strains in Footing Ties

Ties were provided in specimen SF-1 in accordance with the Caltrans recommendations and were reduced by half in specimen SF-2.(16) It is believed that the main reason for their existence is to permit a load path to exist in the footing when the column bars are bent outwards, as is commonly done in cast-in-place construction. However, this function was unnecessary here, because the column bars were straight and equipped with anchor heads. The ties, once in place, might serve any one of three functions: one-way "beam" shear resistance, two-way "punching" shear resistance, or joint shear resistance. For one-way shear, the footing depth was selected so the concrete resistance alone would just suffice, because WSDOT expressed a strong preference for that approach. The joint shear resistance is closely related to the internal force path, so the required strength is likely to be very different when headed straight bars rather than bent-out bars are used in the column. Thus, the expectations for stress in the ties were unclear.

The ties in specimen SF-3 varied greatly from both specimen SF-1 and specimen SF-2. Unlike SF-1 and SF-2, the ties were not placed to follow Caltrans' recommendations. Ties in the footing were strategically placed such that the one-way shear failure mode would be suppressed without inhibiting failure by combined moment transfer and punching shear, since the goal of the test was to investigate strength in that failure mechanism. This meant that the ties needed to be placed within the effective width of the footing but outside of the punching shear plane.

Four ties were strain gauged in each of specimens SF-1 and SF-2. One tie was on the west side of the column, and the other three were on the north side. Because of the way that the ties were placed strategically in specimen SF-3, the ties were lumped together in four different corners. The six ties in the northeast quadrant of the footing were gauged.

Figures 58 through 60 show the measured strains in the ties for selected drift levels. Thermal stresses were accounted for in specimens SF-1 and SF-2 following the same process described earlier. The strains in specimens SF-1 and SF-2 measured in the ties were even smaller than those in the flexural and shear friction steel, and even with no thermal correction it never exceeded strain in tension. Many values were negative.

Strain versus negative drift plot for selected vertical ties.

Figure 58. Graph. Specimen SF-1 strains in ties.

Strain versus negative drift plot for selected vertical ties.

Figure 59. Graph. Specimen SF-2 strains in ties.

Strain versus negative drift plot for selected vertical ties.

Figure 60. Graph. Specimen SF-3 strains in ties.

The vertical ties in specimen SF-3 had much higher strains than those in specimens SF-1 and SF-2. Tie No. 2 reached yielded at 3 percent drift, whereas in the other two specimens the maximum strain never exceeded 4 percent of yield strain. This difference was expected since the footing was designed such that the ties would work to suppress the one-way shear. Although the gauged ties in specimen SF-3 were all within 1 foot of each other, it is likely that tie No. 2 was the tie closest to the punching shear failure plan, and therefore had the highest strain.

Axial Load-Response

Factored Axial Loading

The test specimens were loaded axially with three load combinations at three different times. Before cyclic testing, the columns were loaded with the biggest axial load expected on the socket connection scaled down from the prototype (1.25DL+1.75). After that, the axial load was reduced to the unfactored dead load plus overturning (1.0DL+1.0OT) while the cyclic lateral load was applied. After the lateral loading was complete, the column was subjected an "ultimate" axial load to failure. The last test was not conducted on specimen SF-3 because it had already failed by punching.

Specimens SF-1 and SF-2 successfully carried both of the first two load combinations throughout the tests. However, specimen SF-3 punched through the spread footing while being cycled, as was anticipated in the design. Table 11 summarizes the loading applied and the measured corresponding column vertical displacement. The socket shear stress is calculated assuming that the load is resisted uniformly with depth of the footing.

Table 11. Axial load combinations on the test specimens.
Specimen Load combination Axial Load (kips) Measured Deflection under the Column (in.) Socket Shear Stress (psi)
SF-1 1.25DL+1.75LL 241.7 0.022 175
1.0DL+1.0OT* 159.4 0.014 116
3.5(1.25DL+1.75LL) 842 0.082 611
3.8(1.25DL+1.75LL) 918 N/A 666
SF-2 1.25DL+1.75LL 242.4 0.0086 176
1.0DL+1.0OT* 159.2 0.0060 116
3.4(1.25DL+1.75LL) 819.5 0.079 595
SF-3 1.25DL+1.75LL 240.8 0.056 393
1.0DL+1.0OT* 159.2 0.031 260
1.4(1.25DL+1.75LL) 342 0.068 559
*OT = Overturning

Figure 61 shows column vertical displacement versus cumulative drift (P = 159 kips). Specimens SF-1 and SF-2 maintained the same vertical displacement throughout the test while specimen SF-3 gradually slid through the spread footing.

This chart shows column vertical displacement versus cumulative drift ratio for the three specimens.

Figure 61. Graph. Column vertical displacement vs. cumulative column drift.

Ultimate Axial-Load Capacities

Specimens SF-1 and SF-2 carried the full service axial dead load during the cyclic lateral loading. Subsequently, the column was loaded axially to induce between the precast column and the cast-in-place footing. In both cases, the column failed by crushing in the previously damaged plastic hinge region before any failure occurred in the column-to-footing connection region.

The axial load-deflection response is shown in figure 62. The load reached was about the scaled equivalent of 3.5 Pu, where Pu is the factored axial load given by 1.25DL+1.75LL in the prototype. The load was measured by a load cell in the Baldwin Universal Testing Machine, and the column vertical displacement was measured by the LVDT beneath the column. The potentiometer instrument itself was located just outside the column, and the motion was delivered to it by a lever mechanism that was treated as sacrificial in the event of the column slipping.

This chart shows column vertical deflection versus axial load.

Figure 62. Graph. Axial response of specimens SF-1 and SF-2.

Both specimens responded similarly to the axial loading. Specimen SF-1 was loaded up to 842 kips, and specimen SF-2 was loaded up to 819.5 kips (see figure 63). No sliding failure was observed between the precast column and cast-in-place footing during this test. The shear friction capacity of the connection region was at least that large, and it may have been much higher. In the absence of damage due to moment transfer, such as occurred in specimen SF-3, the shear friction capacity of the deeper footing is clearly sufficient.

This photo shows the underside of specimen SF-2 after being tested vertically to 817 kips.

Figure 63. Photo. Specimen SF-2 after axial load of 817 kips.

Strain gauge measurements in the bottom mat showed strains greater than concrete cracking strain (, but all were smaller than the yield strain of the reinforcement. The largest strain measured was and it was in specimen SF-1. The measurement was obtained from the bars going through the slots that allowed them to go under the column. In both specimens, strains in the bundled bars were approximately . There were no signs of permanent deformation or damage to either the specimen or the supporting hydrostone layer in the base. The deflections measured were attributed to the flexural and shear deformations of the footing.

Page last modified on June 27, 2016
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000