Cement grouted steel bars in glulam Buchanan, Andrew H 1, Moss, Peter J 1 and Eistetter, Sabine 2 ABSTRACT This paper describes an experimental investigation into the use of cement grout for bonding threaded steel bars in glued laminated timber. The objective of this research was to provide a fire resistant method of anchoring steel bars in wood, following the poor observed performance of epoxy in fire conditions. Unlike epoxy, cement grout has little or no adhesion to wood, so high strength can only be achieved with some form of mechanical anchorage between the grout and the wood. Transverse steel dowels and screws have been used successfully in this project. The results of these experiments show that high strength can be achieved if sufficient transverse mechanical anchorage is provided. The advantage of steel screws over dowels is that they can provide confinement of the timber cross-section, preventing the formation of longitudinal splits. This technology requires further refinement before being recommended for widespread use, but the results to date show excellent potential. INTRODUCTION Timber is a very popular building material in many countries of the world. It is not only used for smaller buildings like residential dwellings, but also for large industrial and commercial buildings. When aesthetics is of importance, timber is a particularly favoured material. There are many types of joint used in timber construction, a factor that sometimes makes it difficult to decide which type of connection to choose. The selection of fasteners is not only controlled by the load-bearing capacities of the structure and the direction of loading, but also the aesthetics, cost efficiency and the fabrication process which has to be considered when planning and designing a building. Glued-in bolted connections have been used in Europe for over twenty years and are becoming more and more accepted in other parts of the world. On-going investigations of epoxy bonded steel connections are providing more knowledge and experience about their performance. Very little research on the performance of timber connections exposed to fire has been carried out. Studies have shown which connections have a high fire resistance and which ones are not suitable when exposed to fires. An extensive review of the fire performance of connections in timber structures has been summarised by Carling (1989). One advantage of using glued laminated timber instead of other materials like steel is its better fire performance. This is because large timber members have a low charring rate which gives good fire resistance at no extra cost (Buchanan, 1999). The problem of a glulam structure with epoxy bonded steel connections is that the connection may fail prematurely when exposed to fire. Investigations by Barber (1994) on the fire resistance of epoxy bonded steel connections have shown that epoxy starts to lose its strength when heated above 50 o C with almost no strength remaining at 70 o C. This has been the motivation for the investigation of alternatives to glued-in bolted connections. Gaunt (1998) has experimentally studied connections with large screw fittings where no adhesive such as epoxy had to be used. The research described in this paper investigated a steel connection bonded by cement grout. Cement was chosen as it is a building material that is easy to handle, inexpensive and fire resistant. The cement bonded steel connection was formed by placing a threaded steel rod into an oversize predrilled hole in the glulam members and filling the space around the bars with cement grout. The holes were 10 to 20 mm larger than the steel rod. Different designs of the connection were tested, varying the embedment length, hole geometry, and 1 Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand 2 Ingenieurbüro für Prüfstatik, Stockach, Germany
reinforcement. It was found that the cement grout must have a fluid consistency to ensure the cement can flow through the cavities when injected under air-pressure. The major problem is the bond between the cement grout and the timber. Unlike epoxy, cement does not bond to timber and therefore a mechanical bond has to be designed. The objectives of the study were: To experimentally study the mechanical properties of cement bonded steel connections subjected to short duration tensile load. To develop a method of predicting the strength of the connections for various hole geometries and types of reinforcement. MATERIAL PROPERTIES Material property tests to determine the compression and shear strength of the timber, the compression strength of the cement grout and the yield strength of screws and pins were carried out. These values were then be used in a later stage to obtain calculated strength values for each type of connection. Further details are given in Eistetter [1999]. TENSION TESTS A variety of geometries of the connection were chosen, to investigate their influence on the strength of the connection. The geometry of the connection includes the shape and roughness of the inner hole surface and the hole diameter. In addition, two different types of reinforcement were used: one comprised small diameter pins placed in holes drilled through the threaded steel bar and surrounded by cement grout while the other involved pairs of screws through the timber and the cement (close to the main bar ) as illustrated in Figure 1.. The screws provide a method to prevent the cement grout from pulling out as well preventing the wood from splitting around the connection. Eleven different designs were tested as illustrated in Figure 2. The test specimens comprised 90 x 90 x 750 mm timber with a 20 mm diameter, threaded, high strength steel bar bonded into a hole at one end. The bonding material was a standard mix as used for bonding prestressing strand in prestressed concrete beams, containing an expansive admixture. Since dry wood will suck water out of the cement grout thus resulting in low strength, it is necessary to wet the wood surface inside the hole before pouring in the cement mix. It is important not to leave the water for longer than 40 min. in the hole in order to prevent too much shrinkage and swelling of the wood. Sometimes small cracks parallel to the grain occurred after wetting the wood. After excess water was driven out by air pressure, cement grout was forced into the cavities surrounding the bar. Further details are given in Eistetter [1999]. TEST RESULTS The results of the tests are summarised in Table 1; more detailed results are given elsewhere [Eistetter, 1999]. The greatest scatter of results occurred in Figure 1 Sketch of typical test specimens (before grouting) Experiments 1 to 4, with the pinned and screwed specimens showing good consistency between specimens. In the case of the single pinned specimens (Experiments 5 and 6), the load increased rapidly and then either remained constant or decreased slowly. In the cases of the 2, 3 and 4 pinned specimens, the maximum load was reached at a pull-out deflection of about 6 mm regardless of the number of pins. In Experiments 10 and 11 where screws were used,
the specimens in the former case were able to maintain their ultimate load whereas the ductility was decreased in the latter case (i.e. the stronger connection). Experiment number Design Comment 1 A simple connection being a straight hole with a roughened surface 2+4 A definite key of 70 mm length with a 3 mm depth hole for Experiment 2 and 7 mm depth for Experiment 4 3 A tapered key 5 One 5 mm high strength pin through a hole in the threaded bar to prevent the cement grout from sliding out 6 As for Experiment 5 but with the inner hole surfaces roughened 7 As for Experiment 6, but with two high strength pins 8 As for Experiment 6, but with three high strength pins 9 As for Experiment 6, but with four high strength pins (two in each direction) 10 Screws were used to prevent the wood splitting around the connection. The inner hole surfaces were roughened. 8 screws were used 11 As for experiment 10 but with 16 screws Figure 2 Overview of the various hole geometries and reinforcement tested
Table 1 Results of the tension tests Test Number Embedment length mm Average ultimate load kn Failure mode 1 200 22.2 Slow bond failure with cement grout pulled out of timber 1 300 31.4 As above 2 300 22.8 Timber crushed around load bearing area of key, followed by withdrawal of cement grout 3 300 21.4 As 2 above 4 300 32.6 Local crushing of the timber around the key area, later followed by a tension failure of the grout and a bond failure between the cement grout and the timber 5 300 26.9 Bending failure of the bolts in the connection and local crushing of the grout 6 300 45.2 As 4 above 7 300 67.1 As 4 above 8 300 77.0 As 4 above 9 300 85.4 As 4 above 10 300 47.7 Bending of the screws, after yielding there was some local crushing of the timber followed by pull-out of the steel rod 11 300 89.0 As 10 above DESIGN EQUATIONS Connections with a distinct key Based on the results of the tension tests, the following equation can be derived for predicting the strength of these connections: F = f h1d π 2 2 ( D d ) + σ π dl where f hld = characteristic timber strength (MPa) d = hole diameter (mm) x = key size (mm) D = d+2x (mm) σ friction = friction shear strength being 0.15 MPa for a smooth hole and 0.50 MPa for a rough hole surface l = embedment length of steel bar (mm) 4 The characteristic timber strength was taken as the timber compressive strength of 45 MPa. The friction shear strength was obtained by comparing test results from Experiment 5 with one pin and a smooth hole surface with the results of Experiment 6 with one pin and a rough hole surface. The comparison of the calculated and measured strength is shown in Figure 3. The scatter of the results was possibly caused by the varying key size inside the hole due to the inability of visually controlling the actual key size while drilling on the lathe. Connection with pins The pins used were 4.5 mm diameter and had an average yield strength of 2075 MPa. The equation used to calculate the strength of the connection is based on Johansen s equation for steel-to-timber joints [Eistetter, 1999]. It was assumed that both the fastener and the timber were ideal rigid-plastic materials, and the pin forms a plastic hinge at the face of the main bar with the pin and the surrounding grout acting as a dowel-type fastener bearing against the wood. The frictional shear stress along the cement-wood interface and the reduced strength of more than one pin were also included in the equation: friction
Figure 3 Comparison of the calculated and measured strength values for connections with a distinct key F = f hld d b 4 M pl t1 2 + 1 2 n 2 ef f hld d b t1 + σ friction π dl where F = pull-out strength (N) f hld = characteristic timber strength (MPa) d b = diameter of cement block around the pin (mm) d bar = diameter of steel bar (mm) l b = length of pin (mm) t 1 = (l b d bar )/2 (mm) M pl = plastic moment of pin = 31,500 (N.mm) from bending tests [Eistetter, 1999] σ friction = friction shear strength being 0.15 MPa for a smooth hole and 0.50 MPa for a rough hole surface d = diameter of main hole (mm) l = embedment length of steel bar (mm) n = number of pins n ef = effectiveness factor = 1+(n-1) x 2/3 [Eistetter, 1999]. The comparison of results for Experiments 5 to 9 is shown in Figure 4. The characteristic strength of the timber was taken as 50 MPa based on the average compression strength of the timber [Eistetter, 1999]. Figure 4 Comparison of the calculated and measured strength values for connections with pins Connections with screws Since two types of screws were used having different yield strengths and bending behaviour, the failure type of each had to be analysed separately. Type 1 screws These had an outside diameter of 5.5 mm and an average yield strength of 1080 MPa. They exhibited ductile behaviour and Johansen s equation relating to bending in double curvature was used:
F = 2β 1 + β 2 M pl f hld d s 2 n + σ friction π dl where F = pull-out strength (N) f hld = characteristic timber strength (MPa) f h2d = characteristic cement strength (MPa) β = f h2d / f h1d d s = screw diameter + ½ thread = 4.65 (mm) M pl = plastic moment of type 1 screw = 9,860 (N.mm) from bending tests [Eistetter, 1999] σ friction = friction shear strength being 0.15 MPa for a smooth hole and 0.50 MPa for a rough hole surface d = diameter of main hole (mm) l = embedment length of steel bar (mm) n = number of screws Although the wood and cement grout were both tested, it was not clear which strength values should be used in the design equation. For this reason, several alternative values were tried in an attempt to obtain the best prediction of the test results. The influence of the ratio β on the calculated strength is shown in Figure 5. The numbers in the legend are the ratio of the timber strength and cement strength, i.e. the reciprocal of β. The test strength of 40 Mpa for the cement grout gives predicted strength values that are too low whereas predicted values based on a much higher grout strength give more accurate predictions. Figure 5 Comparison of the calculated and measured strength values for connections with type 1 screws
Type 2 screws F = f d t 4 β ( 2 + β ) M h1d s 1 pl 2 β ( 1 + β ) + β 2n + σ friction πdl 2 + β f h1d d st1 These screws had an outside diameter of 6 mm and were of higher strength (f y = 1390 MPa) than the type 1 screws, but more brittle with only limited ductility. Failure was in single curvature for which the following equation was developed: Where F = pull-out strength (N) f hld = characteristic timber strength (MPa) f h2d = characteristic cement strength (MPa) β = f h2d / f h1d t 1 = length of screw embedded in timber (mm) d s = screw diameter + ½ thread = 5.4 (mm) M pl = plastic moment of type 2 screw = 25,650 (N.mm) from bending tests [Eistetter, 1999] σ friction = friction shear strength being 0.15 MPa for a smooth hole and 0.50 MPa for a rough hole surface d = diameter of main hole (mm) l = embedment length of steel bar (mm) n = number of screws The influence of various β values on the calculated strength of connections with 8 and 16 type 2 screws and their comparison with the measured values is shown in Figure 6. For 8 screws, the predicted strengths are all above the test values, whereas for the connections with 16 screws the predicted vales are significantly higher than the test results. Even though the type 2 screws have a plastic moment more than twice that of the type 1 screws, the full benefit of this strength cannot be achieved because brittle failure of the screws occurs before the full strength is reached. Figure 6 Comparison of the calculated and measured strength values for connections with type 2 screws
SUMMARY AND CONCLUSIONS Summary Epoxy bonded steel connections, as a method to construct moment resisting joints, gain increasing application in timber construction. Glulam timber is known for its low charring rate and thus good fire resistance, but previous research showed that this type of connection using epoxy has a low fire resistance. This is due to the fact that above a temperature of approximately 50 o C the strength of epoxy decreases rapidly, causing the connections to fail. An experimental study was carried out to investigate steel connections in glue laminated timber bonded by cement grout. Cement was chosen as an alternative to epoxy as it has good fire resistance and is available at an economical price. The experimental programme consisted of two parts: material property tests and tension tests of cement bonded steel connections. Compression and shear strength of the timber, compression strength of the cement grout and yield strength of screws and pins were tested. These tests were carried out to obtain strength values of the materials for further calculations. Ten various designs of cement bonded steel connections were tested. After testing specimens with straight embedment holes, it became clear that the major problem would be to provide a mechanical bond between the timber and the cement grout. Therefore, three methods of mechanically bonding cement to timber were tested. The first one was by means of a distinct key inside the main hole and the second by using small diameter pins drilled through the threaded steel rod and surrounded by cement grout. The third one was by means of pairs of screws through the timber and the cement (close to the main bar), preventing the timber from splitting and the cement from pulling out. The failure modes were investigated and equations were developed to predict the strength of the cement bonded steel connections with various designs. Conclusions 1. Straight holes do not give enough bond between timber and cement, even if they are roughened internally. A mechanical bond has to be provided. 2. A minor increase in connection strength can be achieved by a distinct key inside the embedment hole. 3. Strong connections can be made using pins or screws. The strength of these connections can be increased by an increasing number of pins or screws, but there is not a linear relationship between the number of pins or screws and the strength of the connection. 4. There is not a proportional relationship between increased screw strength and connection strength because high strength screws are less ductile. 5. Failure of all the cement bonded steel connections is by ductile pull-out of the cement-steel block. No wood failure occurred. 6. Initial splits along the specimens, due to wetting the hole before grouting, have no effect on the ultimate tensile strength of the connection, although the splits became larger as the bar was pulled out. REFERENCES Barber, DJ. 1994. Fire resistance of epoxied steel rods in glulam timber. Research Report No 94-1, Dept of Civil Engineering University of Canterbury, Christchurch, New Zealand. Buchanan, AH. 1999. Structural design for fire. Dept of Civil Engineering, University of Canterbury, Christchurch, New Zealand. Carling, O. 1989. Fire resistance of joint details in load bearing timber construction a literature study (translated form Swedish). BRANZ Study Report SR 18, Building Research Assoc. of New Zealand, Judgeford, New Zealand. Eistetter, S. 1999. Strength of cement bonded steel connections in glue laminated timber. Civil Engineering Research Report No 99-1, University of Canterbury, Christchurch, New Zealand. Gaunt, DJ. 1998. Joints in glulam using groups of epoxy grouted steel bars plus an alternative to epoxy bonding. Proc. 5 th World Conference on Timber Engineering, Montreaux, Switzerland, pp281-8.