Chapter 3 Experimental Test Program

An effective adhesive product suitable for fastening tension anchors in the notch connection was needed.  Two potential adhesive products were examined using connector withdrawal tests.  One of the two adhesives was then selected for incorporation in the construction of slip test and beam specimens. Findings from these glue tests are presented subsequently.  Withdrawal tests are also used to insure that the adhesive has properly set before adding a concrete layer to the connection, beam, and deck specimen.

Slip tests of notched/anchor specimens were then conducted to observe and quantify load-slip behavior.   The test method used was a single shear tension test using a specimen constructed of the mixed material layers and containing one anchored notch connection. The jig device used originally was devised by Debonis (Debonis et al. 1975) for use in testing nailed wood connection.  The device has been modified and used by many investigators (e.g. Pault et al 1977) on glulam bridge specimens [Chen on wood-concrete T-beam specimens] (Chen 1992, Chen et al 1992, and Gutkowski et al 1996).

The third phase in the testing was to experimentally quantify the degree of composite behavior of the layered beams.  This was done as a cost-effective approach to evolve an optimal connection for later studies of full-size deck specimens.  This was accomplished by measuring deflections of beam specimens subjected to a symmetric, two-point loading.

Originally it was desired to complete the slip testing prior to construction of the beams so the appropriate notch geometry could be chosen for the beams.  However, the project schedule prevented that timing.  A decision was made to construct all beams with pre-selected notch dimensions and place concrete on the same date for all connection and beam specimens. It also was advantageous to keep the concrete properties consistent between the slip and beam specimens.

The project's fourth phase was load testing on two full-scale deck specimens.  Displacements were measured under several individual point loads.  Efficiency of the composite action was observed.

Initial Withdrawal Test Specimens

The purpose of the withdrawal tests was to decide on a proper adhesive product for the notches of the slip and beam specimens. A Hilti adhesive (C-50 HIT WTR) had been used in prior slip specimens tests done by Thompson (Thompson 1974) and in some preliminary beam specimens tested prior to the current work.   This adhesive (only available in Europe) is expensive to obtain, has a limited shelf life, and is vulnerable to possible low temperature conditions during overseas air transport. In the project described here two alternative fast-setting glue products were investigated for possible use.

The first alternative adhesive investigated was a phenol-resorcinol laminating resin manufactured in the United States by Borden Chemical, Inc.  Its primary use is for the lamination of softwood glulam members used in wet-use and dry-use exposure conditions.  The manufacturer does not recommend the resin to be used with hardwoods and cautions its use when fire retardant and preservative treatments are applied. 

The second adhesive investigated was the Hilti (HIT HY 150) adhesive, a product manufactured in the United States by Hilti Inc.  The manufacturer advertises its use for seismic upgrading of masonry buildings, installation of roadway dowels, and other applications involving the anchoring of structural steel connectors to various base materials.

For the purpose of withdrawal testing it was decided to examine if the Hilti dowel hardware could be replaced with an easily obtainable, low cost, threaded rod of the same diameter and thread pattern.  A test was done on threaded rods and the Hilti dowel for comparison.

In the initial phase 70 withdrawal test specimens were constructed and tested (see Table 3.1).   Seven different configurations were used with 10 replicates of each.  Ten of the specimens were constructed using Hilti dowels and the remaining 60 were constructed using threaded rods. The threaded rod stock had an identical thread count, thread pitch, and diameter as the Hilti dowels.  For each of the two adhesives, threaded rods were investigated at three connector depths into the wood. Specimens constructed of the Hilti adhesive were tested at depths of 2.0, 2.5, and 3.0 inches.  In the case of the Borden adhesive, depths of 1.5, 2.0, and 2.5 inches were chosen. 

The specimen groups in the first column of Table 5.3 are represented by their group designation.  For example, the first specimen designation is Re-1.5.  "Re" represents the adhesive used (Re for the Borden resorcinol and Hi for the Hilti adhesive) and the "1.5" represents the connector embedment depth in inches.  The lumber used to construct withdrawal specimens came from the same source as the subsequent slip and beam test specimens.

A schematic of a typical withdrawal specimen is shown in Figure 3.1.  The construction process for the withdrawal specimens began by cutting one-foot sections from 12-foot long nominal 4x4 boards.   The threaded rod stock was cut into pieces of about six-inch lengths.  Holes were predrilled to the appropriate depth in the geometric center of each timber piece.

Table 3.1 - Summary of withdrawl test specimens

Specimens constructed of the Borden adhesive were drilled with one-half inch diameter holes.  This diameter allowed adequate space for adhesive to occupy the space between the dowel and the wood.  Since application of the adhesive was atypical, no usage recommendations were available.  The Hilti specimens were drilled with 9/16-inch diameter holes as recommended by the manufacturer.  A dispensing gun obtained from the Hilti Company was used to insert the adjacently-packaged components of the Hilti epoxy.  Application of the Borden adhesive was achieved by mixing the components in containers and then pouring the appropriate amount in the predrilled fastener holes.  The predrilled holes were filled approximately three-quarters of the way prior to placing the connector.

Conduct of the Withdrawal Tests

The withdrawal test specimens were loaded using an Instron Universal Testing Instrument.  An apparatus was configured such that a tensile load could be applied to the mechanical anchor.  Potentiometers were attached to the base of the steel dowel and to the centroid of the wood member.  Displacement of the steel dowel relative to the timber member was determined from the difference in the two displacements. The span of the timber member was approximately 11 inches.  This was deemed sufficient to avoid any local bearing effects by keeping the reactions a reasonable distance from the dowel anchor location itself.   The tests were controlled at a strain rate of 0.5 inch per minute.

Additional Withdrawal Test Specimens

As described subsequently, initial withdrawal tests (Brown 1998) indicated that the phenol-resorcinol adhesive had a superior behavior to that of the Hilti HIT HY 150 adhesive.  It also was subsequently learned (after the beam tests) that European practice was to tap the holes before putting in the HIT HY 150 adhesive. Tapping the holes increases the glued surface and suppresses the large slip surface between the wood and adhesive, and therefore increases the strength of the connection.  Thus, prior to the deck tests, additional withdrawal tests were conducted on the HIT HY 150 using pre-tapped holes.  Ten withdrawal test specimens were constructed and tested.  The specimens included a threaded rod with a 0.07 in thread pitch and a 0.47 in diameter.  Only one connector depth of 2.5 inches was investigated and the results were compared with those obtained by Brown (Brown 1998) for the Borden glue and the same Hilti product at the same connector depth but, without tapping the holes.  The lumber used was of the Hem-Fir species group and Standard and Better grade.

The construction process for the additional withdrawal specimens also began by cutting one-foot sections from 12-foot long nominal 4"x4" boards.  Holes were predrilled to the appropriate depth in the geometric center of each timber piece.  They were drilled with a 5/8" diameter drill bit as recommended by the manufacturer. The holes were tapped to 3/4" using a common steel tap tool. The predrilled holes were filled approximately 3/4 of the way prior to placing the connector.  All the specimens were tested seven days after being glued. A potentiometer was attached to the steel dowel to determine its axial displacement. The tests were subjected to a strain rate of 0.5 inch per minute.

Slip Test Specimens

The purpose of the slip tests was to quantify interlayer slip resistance behavior of the notch connection. It was desirable to determine to what degree an increase in the size of the notch influences behavior of the connection. Fig. 3.2 illustrates the general geometry of the specimen.  Three different notch sizes and two lumber configurations associated with the two beam cross-sections were investigated.  The angle a was 15 degrees for all specimens. A perpendicular cut would minimize the upward tensile load transferred to the connector, however, it was avoided due to the increased stress concentrations, which may have resulted at the base of the notch.  Sixty slip test specimens were constructed and tested.  Table 3.2 lists the dimensions used for each notch type.

The lumber configurations were chosen because their overall thickness of six inches was similar to a typical concrete slab.  The first configuration consisted of a single nominal 4x4 piece of lumber and the second consisted of three nominal 2x4s.  The 4x4 slip specimens were constructed from a single one-foot section of board.  The 2x4 specimens were made from three one-foot pieces of wood nailed to each other at two locations.  The cross-sections are shown in Figure 3.3.

Table 3.3 indicates the number of replications of each specimen, notch types and wood members used for each specimen.  For example, specimen 2x4 - A used 2x4s and notch detail A.  All specimens used the Borden adhesive and 1.5 inches dowel depth into the wood.

Table 3.2 - Notch dimensions

Table 3.3 - Summary of slip test speciments

Construction of the slip test specimens began by cutting 12-foot long nominal 2x4 and 4x4 boards into 1-foot lengths.  For the specimen involving 2x4s, the three 2x4s were nailed together using three-inch galvanized nails at mid-depth and spaced at eight inches.  For the specimens involving 4x4s, a single member was used so no nailing was involved.  After the timber portion of each slip specimen was constructed, notches were cut into the wood. Notching of the specimens was achieved by cutting inclined surfaces with a circular saw.  The blade was set at an angle of 15 degrees with respect to the vertical plane.  The remaining material was then removed with the use of an ordinary wood chisel.  Holes were predrilled at a depth of 1.5 inches and the Hilti dowels were installed using the Borden adhesive. Concrete formwork was constructed by placing  notched wood sections next to each other with a thin slat placed between them.  Wood boards were screwed in place around the group of specimens to complete the formwork. Finally, the concrete was placed.  Concrete consolidation was achieved by means of a hand held vibrator.

Conduct of Slip Tests

Slip test specimens were loaded with an Instron Universal Testing Instrument at a rate of 0.5 inch per minute.  A jig (see Figure 3.4), which attaches to the Instron testing machine, was employed to support the specimen. The jig is comprised of four aluminum plates, two on each side.  Front and back sections are forced to slide up and down relative to each other by the action of four straps attaching four bearings functioning to hold the plates together. The specimen layers also are forced to slide relative to each other by being attached to the aluminum plates with clamps and pins.  The interlayer slip motion was measured using potentiometers.  A potentiometer was hooked to the wood and the concrete sides of the aluminum frame.  An additional potentiometer was attached to the wood directly.

Layered Beam Specimens

Twenty layered beams were constructed and tested. End views of the two configurations used are shown in Figure 3.5.  Ten of the beams were built using three nominal 12-foot 4x4 boards connected along the length with structural adhesive, resulting in an overall width of 10.5 inches. Ten additional beams were constructed the same length using eight 2x4 boards nailed vertically with three-inch galvanized nails spaced 16 inches on center; resulting in an overall width of 12 inches.

An elevation view of the beam specimen is shown in Figure 3.6.  Each beam had four notches along its length.  Notches were not cut in the beam segment between the load application points because no shear is present if the two loads are equal.  All beam specimens had an overall length of 12 feet (144 in.) resulting in a clear span of 11.5 feet (138 in.).  The notches were spaced at 13 inches on center between the each load point and the adjacent support.  Notch B (see Table 3.3) was used for all beam specimens and each notch included two Hilti dowel connectors spaced at the third points across the width of the cross-section (3.5 in. for the 4x4 specimens and 4 in. for the 2x4 specimens) at a nominal depth of 1.5 inches.   Wire fabric, with a two-inch square grid pattern was placed in the beams to prevent cracks resulting from the drying and hydration processes.  The quantity of mesh used satisfied requirements set forth by Section 7.12.2.1, of the ACI 318-95 concrete building code (American Concrete Institute 1995).  Concrete formwork was constructed by attaching six-inch wide plywood strips around and between groups of four beams.  Concrete was placed and consolidated with a hand held vibrator.  Specimens were moist cured for a minimum of 28 days.

Conduct of the Layered Beam Tests

Beam specimens were loaded symmetrically with two concentrated loads, spaced at five feet apart.  Two separate 100-kip capacity actuators were used to apply the two point loads.  Dual overhead steel frames supporting the actuators were connected rigidly to the laboratory floor.  Each beam end was supported on a concrete filled barrel.  Simple supports were realized by placing a circular pin between two slotted plates.  Vertical beam deflections were measured by placing string potentiometers at mid-span and directly below each of the load points.

Deck Test Specimens

Two  layered wood-concrete decks were load tested in the laboratory and these are described here.

The first specimen, referred to hereafter as the "rectangular deck," was configured to simulate (at smaller scale) a right bridge structure. Figure 3.7 schematically illustrates the specimen.  The specimen had anoverall length of 141 inches resulting in a clear span of 136 inches.  All the lumber used for construction was from the Hem-Fir group species. The specimen consisted of 65 nominal 2"x4"x12' Hem-Fir longitudinal deck members. Two end bearing supports were constructed with steel plates and steel rods. One of those supports allowed the deck end to rotate and to move in the longitudinal direction.  The other support only allowed the deck end to rotate, but horizontal displacement was blocked.  Incidental gaps between the deck and the upper plates of the bearing supports were filled with a chemical mixture that quickly solidified after its application. The deck members were screwed together with 1/4" by 4" long screws at six-inch centers with two-inch offsets from members to members.

The second laboratory deck specimen, referred to hereafter as the "skewed deck," also was configured to simulate (at smaller scale) a skewed bridge structure. The specimen is shown in Figure 3.8. The deck measured 19'-2" long, with a clear span of 18'-6", by 8'-10.5" wide with a skewed angle of 43.6 degrees.  The deck members were of the Hem-Fir species group and standard and better grade.  All other timbers used for the construction of the reduced-size specimen were of the Douglas Fir species group. The specimen consisted of 75 nominal 2"x4" x 20' deck members.   The wood deck members were connected laterally by using 20d nails placed every 12 inches in staggered rows.  The end bearing supports were similar to those used for the rectangular deck.

Construction of the Deck Specimens

Figure 3.9 illustrates the notch detail.  Based on the slip and beam tests, the width and depth of the notch were set at 6.0 inches and1.0 inch, respectively.  Based on European experience and practice, the angle of aperture a was chosen at 10 degrees.  It was not believed to be a significant change from 15 degrees used in the preliminary tests.  Notching of the specimens was achieved by cutting the inclined surface with a circular saw.  The remaining material was then removed with the use of wood chisels. For placing of the Hilti dowels, each deck was drilled with 5/8" diameter holes at a nominal depth of two inches and then tapped with a UNC thread tap to a 3/4" diameter. The predrilled holes were filled approximately three-fourths of their depth prior to placing the connectors.

The thickness (2.5 inches) of the concrete layer was the same for both deck specimens.  This value was determined in such a way as to situate the neutral axis of the entire composite section at the interface between the layer of concrete and that of the wood.

To avoid shrinkage cracking in the concrete during the curing period, wire mesh reinforcement was used in each specimen.  The wire mesh employed 3/8" inches diameter rebar with a regular spacing of 8.5"x8.5".  The quantity of mesh used satisfied the requirements set forth in Section 7.12.2.1 of the ACI 318-95 concrete building code (American Concrete Institute 1995).   In addition, two 3/8" diameter rebar were placed at the bottom of each notch to increase the bending stiffness of the excess concrete around the notches and to improve the lateral distribution of the point loads.  The ends of these bars were shaped by hand in the form of a hook for anchorage to the concrete layer.

The concrete formwork was constructed by screwing plywood strips around the deck specimens.  The ready made concrete was obtained from a local vendor with a specified 28-day compressive strength of 4,000 psi. The concrete was placed and consolidated with a small, hand-held vibrator.

To restrain the dead weight of the wood and the concrete in composite construction, a wood shoring brace was placed at mid-span for each specimen.  The specimens were moist-cured for a minimum of 28 days.  Then the plastic caps of the Hilti dowels were removed and the connector nuts were tightened with a torque of 443 lb-in using a torque wrench.

Positions of the Notches and the Hilti Dowels

Rectangular deck

Figure 3.7 illustrates the positions of the notches and dowels.  Since this specimen had a relative short clear span of 136 inches (345 cm), only three notches were cut.  One was cut at the mid-span and two others were placed at 17.5 inches (44.5 cm) from the deck ends. Based on work done by researchers in Switzerland (Natterer et al 1998), a ratio of 0.129 connectors per square foot (1.36 connectors per square meter) was used.  Eight Hilti dowels were used in the pattern shown, which resulted in a ratio of 0.09 connectors per square foot (0.9 connectors per square meter).  The exterior notches included three Hilti dowels.  One of them was placed in the middle of the cross section and the other ones at eight inches (20 cm) perpendicular from the edges of the deck.  The interior notch included two dowels.

Skewed Deck

Figure 3.8 illustrates positions of the notch cuts and the Hilti dowels for the skewed deck specimen.  Five notches following the skewed angle of the deck were cut.  One was placed at mid-span, two others were placed on both sides of the previous cut at 50 inches center to center, and the last ones were placed at 23 inches from the deck ends.  Thirteen Hilti dowels were used in the pattern shown, which resulted in a ratio of 0.08 connectors per square foot (0.82 connectors per square meter).  For the notches with three dowels, one dowel was placed in the middle of the cross-section and two others were placed at 8.25 inches perpendicular from the unsupported sides of the deck. For the notches with two dowels, each was placed at 24.8 inches perpendicular from the unsupported sides of the deck.

Deck Testing Procedure and Loading Procedure

One objective was to compare the transverse deck deflections before and after adding the concrete layer. Thus point loads were placed at the same positions on both the wood and the composite decks.

Rectangular Deck

Loading was achieved via an overhead steel frame rigidly connected to the laboratory floor.  A longitudinal steel I-beam spanned between the upper girders of the steel frame.  This beam was connected to the girders by a system of rollers and was able to move laterally across the specimen. One 100 kip capacity actuator was used to apply the point loads.  This actuator was attached to the longitudinal I-beam by rollers to be movable along its length. This configuration allowed the actuator to be positioned at different locations on the surface of the specimen.

Figure 3.10 shows four load positions used.   Two point loads were applied at mid-span of the specimen. One was placed in the middle of the deck and the other one at 5.75 inches perpendicular to the unsupported east side of the deck.  Two other point loads were positioned at the north quarter of the clear span at the same lateral locations as used for the mid-span loads.

A steel spacer block was used to apply the actuator load. This spacer block was constructed with a square steel shape and with a steel plate welded across each end.  The steel plates had a square section of 10"x10" in².

Vertical deflections of the specimen were measured at the mid-span and at both quarter point locations along the clear span.  They were recorded using a series of position transducers (potentiometers).  Figure 3.11 shows the 15 locations of the potentiometers used for collecting displacement data.  The potentiometers were fixed to the interior deck surface with small steel hooks and were uniformly spaced at 24 inches apart in the transverse direction.

Skewed Deck

Two overhead steel frames were used to support the ends of the deck specimen.  A longitudinal I-beam spanning the frames served as a crane girder spanning between the overhead frames.  A system of rollers allowed it to be moved laterally across the specimen.  One 55 kip capacity actuator was used to apply the point loads.  This actuator was attached to the crane girder by rollers to be movable along its length. This configuration allowed the actuator to be positioned at different locations on the surface of the specimen.

Because the deck had no symmetrical properties, the bridge specimen was tested with the actuator sequentially positioned at six different locations.  Figure 3.12 shows the load positions.  Two point loads were to be applied at the mid-span of the specimen.  One was placed in the middle of the deck and the other one at 5.75 inches perpendicular to the widest (east) side of the deck.  The other point loads were positioned at about the north and south quarter points along the clear span at the same lateral locations as used at mid-span.

A steel apparatus was configured to transfer the actuator load. This apparatus was constructed with a round steel shape welded to two steel plates at each end.  The steel plates had a square section of 10"x10".

Vertical deflections of the specimen were measured at mid-span and at both quarters of the clear span.  Figure 3.13 shows the locations of the potentiometers used for collecting displacement data.  The instrumentation included three lines of six potentiometers.  The potentiometers were fixed to the interior deck surface using small steel hooks and equally spaced 30.6 inches parallel to the skewed angle of the deck.