Anti-check bolts as means of repair for damaged split ring connections Quenneville, J.H.P. 1 and Mohammad, M. 2 ABSTRACT There are numerous large span timber hangars dating back to the Second World War. These structures were built with large green timbers and were normally assembled using split rings and/or shear plates. Due to shrinkage in the timbers in the joint area, most of the structural members show severe splitting. As a precautionary measure, the capacity of the connections is usually reduced. A typical repair method is to install anti-check (A/C) bolts running perpendicular to the longitudinal axis of the timbers. Repairing members using A/C bolts is more favourable over other methods, since A/C bolts are easy to install. However, no data is available on the effectiveness of this method. Five groups of full-scale specimens representing typical connections in Warren truss hangars were tested in this study. Specimens were fabricated with splits in the truss tension members and A/C to verify the effectiveness of A/C bolts as means of repair for split connection members in large span timber hangars. A/C bolts were installed either between the split rings or within the end distance. Results were compared to a previous study carried out on intact specimens (reference groups) without splits or A/C bolts (Quenneville and Sauvé, 1998). Comparison between results on intact specimens and those with tension splits and A/C bolts have indicated that the strength of damaged tension members, where splits run between split ring rows, was not significantly different from un-damaged tension members. Furthermore, A/C bolts significantly reinforced some damaged connections and could be used to reinforce un-damaged split ring connections and force an ultimate ductile failure mode. The simple repair method is described and its effectiveness is demonstrated. INTRODUCTION Warren timber trusses were used extensively by the Department of National Defense of Canada during and soon after World War II as hangars for aircrafts housing and maintenance and recreation halls. Two types of 34 m (112 ft.) span Warren trusses were built at the Canadian Department of National Defense bases. One type is a 7-panel parallel chord truss (shown in Fig. 1) and the other is an 8-panel pitched chord truss. A combination of split rings and bolts were used to join truss members together and to transfer loads from one member to another. Performance of existing timber hangars Due to the urgent wartime demand on the construction of these hangars (which were thought of at the time as being temporary structures), the large timber members could not dry out prior to fabrication. Not only that, but in some cases low grade timber was used as well. Rapid drying process in members with high moisture content following truss erection resulted in drying checks due to stresses caused by uneven shrinkage (CETO 1987). This is quite critical at the joints that hold truss members together. Shrinkage of the wood between the connectors tends to cause splitting of the wood member between the connectors in the direction parallel to grain (due to the restraints provided by the ring connectors). Also, shrinkage of timber around a single split ring connector may result in a check or a split passing through the area bounded by the ring. Inspection reports (CETO 1987) revealed that a large number of splits and checks occurred during the period immediately after erection when truss timber elements were drying. Generally, splits or checks due to shrinkage should not be considered as indication of member failure, unless they are located at the member joined end. A typical situation occurs in a truss joint where a split passes between the connector rings. An effect of the split may be the reduction in the load bearing capacity of one or more rings due to the assumed members having edge distances less than allowable. Repair methodology Traditionally, split Warren truss tension members have been replaced or repaired using clamps (usually used to stop crack 1 Associate Prof., Dept. of Civil Engineering, Royal Military College of Canada 2 Res. Associate, Dept. of Civil Engineering, Royal Military College of Canada
propagation), plywood gusset plates or A/C bolts. Repairing truss members using A/C bolts is more favorable over other methods, since A/C bolts are easy to install and involve little material and labor. This study was divided in two parts. The first part concentrated on the performance and effectiveness of A/C bolts installed between the split rings. The second part of the study was established to further investigate the performance of the A/C bolts in an attempt to enhance the effectiveness by installing the A/C bolts within the end distance. Top chord U0 U2 U4 U6 U8 U10 U12 U14 L0 L2 L4 L6 L8 L10 L12 L14 Bot t om chord Compression member Tension member Figure 1. Connections L2, L4 and L6 in Warren parallel chord truss used at the DND hangars. EXPERIMENTAL INVESTIGATION Five groups of 10 replicates each representing joints L2, L4 and L6 (Fig. 1) of a parallel chord Warren truss were tested in this study. All specimens were made of two Canadian Douglas-fir No. 2 grade wood members (representing tension members), sandwiching a Douglas-fir block (representing the bottom chord). Wood members were 80 mm in thickness. The two wood side members were sampled, cut and stored in a conditioning chamber to attain 12% equilibrium moisture content (EMC). All groups in this study were tested with a cut (1mm wide) made in the middle of each member in a direction parallel to grain at the loaded end of specimens (at the split rings location). The cut represents a split due to shrinkage in the tension member and extends 25.4mm (1.0 in) beyond the edge of the upper split ring (further up from the loaded end). Two or four 102 mm (4.0 in) split rings (depending on the specimen configuration) were used to connect each member to the inner block. Specimens configurations were in accordance with existing Warren truss joints. Connections dimensions and configurations are shown in Fig. 2. A typical test set up is shown in Fig. 3. A steel hollow section was used to apply a reaction to the lower sandwiched center wooden block. A monotonic tension load was applied through a wooden loading member at the top, which was attached to the two wood side members using shear plates and bolts. A universal loading machine (MTS) was used to apply the load. Four LVDTs were used to record the slip of the two side members with reference to the center member. A data logging system was used to record the machine load and slip from the four LVDTs. Test specimen was first set into the testing machine, the A307 A/C bolts (9.5 mm in diameter, 3/8 in) were installed across the specimen width and perpendicular to the narrow surface (Fig. 4). A/C bolts were installed either between (groups 1 to 3) or within the end distance (groups 4 and 5). Only one A/C bolt was used for each side member. To measure the loads carried by the A/C bolts, load cells were installed at the end of the A/C bolts (i.e. between the wood surface and the nuts). An initial preload of about 1.0 kn (225 Ib) was applied to all specimens. The initial load in the load cells was recorded before running the test as well. The test was displacement driven at a rate of 0.9mm/min (0.035 in/min) in accordance with ASTM standard D07.05.02 (ASTM 1994). Following the split rings failure, the loading rate was increased to 2.4 mm/min (0.95 in/min). Tests were stopped upon failure, either when the load dropped with no recovery or due to excessive deformation ( 35mm, 1.4 in). Three groups (L2, L4 and L6) were tested intact in another study by Quenneville and Sauvé (1998) following the same procedure and were considered to be the reference groups. Ultimate load values for the various groups of specimens tested are given in Table 1. It includes results of intact connection specimens (reference specimens), followed by results of specimens tested with splits and A/C bolts installed
between the split rings or within the end distance. The behaviour of the different types of connections specimens tested (L2, L4 and L6) was assessed through observations of the failure modes, their load-slip relationship and the ultimate strength. L6 L4 L2 Split 70 90 120 62 245 70 50 120 245 90 62 120 85 245 120 116 110 120 265 155 175 18 8 220 276 328 * All dimensions are in mm Figure 2. Truss tension members with split rings configurations. RESULTS AND DISCUSSION Failure modes Connections tested with A/C bolts reached their maximum load either due to the failure associated with the split rings or that of the bolts holding the two members together (as typical in bolted connections). Two dominant types of failures were observed in groups L2, L4 and L6 tested with A/C bolts installed between the rings. These were: shear plug and group tear-out. Few specimens failed in bearing and yielding of the split rings or in tension of the wood at the net section (caused by a load eccentricity and out-of-plane bending), but these failure modes were not as common. Shear plug of the wood was the most dominant mode of failure observed. Specimens that recovered their strength following the failure of the split rings failed ultimately due to shearing off the area underneath the bolt. The group tear-out failure mode was most evident in group L2 (4 split rings on each side member). In this case one of the cracks was the existing cut created artificially during fabrication. The existing split forced the connection to act like two separate connections, even with the presence of the A/C bolts, leading to the two parts of each side member acting more or less independently. A combination of two types of failures on one member was not uncommon for some specimens. For connections tested using A/C bolts installed within the end distance, bearing (mostly in group L6) and row shear-out of the area underneath the split connector (group L4) were the two dominant modes of failure, unlike specimens tested using A/C bolts between the rings, where failure was mostly due to row shear-out and to a less extent, group tear-out. Localized failures were observed in the load-slip envelope due to the failure of the split ring wood core. Load would usually pick up and a ductile behaviour was achieved. Many specimens in this group reached their ultimate strength in bearing before a sudden drop in the load occurred. The sudden drop in the load after reaching the ultimate load corresponded with the row shear-out of the wood area underneath the lower ring. One explanation for such behaviour would be due to the reduction in
P P 8-19mm bolts St eel loading plat es 12-102mm shear plat es w/19mm bolt s Douglas-fir wood side members Wood loading block Loading block HSS 178x127x13 LVDT 2 or 4-102mm split rings w/19mm bolts A/C bolt Front view Side view Figure 3. Schematic for the test set-up. the triggering shear length as a result of the considerable crushing of wood underneath the rings. Once the shear strength is exceeded, the connection failed in a brittle manner. Group L4 tested, with A/C bolts installed within the end distance, exhibited similar modes of failure observed in group L6, except that row shear-out was more common. No splitting was observed in these tests. In few specimens (specimens 2, 5 and 9), following row shear-out in the rings, the connection exhibited an increase in the their load carrying capacity due to the bolts holding the members together. In other words, the connection was transferred into a bolted connection and the final failure was in a row shear-out of the wood area underneath the bolts. This was also observed in connections tested in part one, with A/C bolts installed between the split rings. However, the difference between the two groups, was in the amount of ductility observed in this type of connections. Connections tested without splits and A/C bolts always ruptured in a brittle manner, with some signs of localized failures (cracks and split ring core shear-out) (Quenneville and Sauvé 1998). Group tear-out, splitting and shear were the three dominant types of failure. In the connections tested without splits and A/C bolts, the ultimate load was reached when the specimens cracked severely either around the split ring closer to the end of the member (a shear plug or tear-out mode of failure) or entirely across one of the members (going through one of the split rings), especially if a defect was present. One of the first sign of rupture was a longitudinal split between the two split rings (or split ring rows in connection L2). The load would, however, keep on increasing. This agrees well with the findings from this study.
Art ificial cut Specimen Ant i-check bolt Load cell 2.5 mm (1.0 inch) Figure 4. Schematic for the split ring connection with the A/C bolt and the load cells. Connection behaviour Connection specimens tested with splits and A/C bolts exhibited similar load-slip curves at failure compared to those without splits and A/C bolts. However, the significant difference is that the repaired specimens carried a load beyond the point where the split rings failed. Typical load-slip curves for the type of connections tested in this study is shown in Fig. 5. These curves are characterized by localized small drops in resistance and subsequent recoveries. The small drops are believed to be due to the shearing of the split rings cores (Charron and Quenneville 1994, Quenneville and Charron 1996). Big and sharp drops in the load-slip curve usually correspond to the failure of the wood adjacent to the split rings. The first small drop of the resistance is associated with the failure of the lower split ring (the split ring closer to the end of the member) in one or both side members. As a result, the entire load get transferred to the remaining rings (in both side members). Most specimens (especially in groups L2 and L4) exhibited two or more small sharp drops in the resistance before a major failure occurred. Others, reached the ultimate resistance and failed in a sudden manner. Then the load picked up but never reached the previous high level. Further drops of load occurred later due to the failure of the remaining rings. This type of behaviour was observed mostly in group L6. In most specimens, failure occurred on both side members, except for few specimens in group L4. Looking at a typical load-slip curve for L6 specimens with A/C bolts installed between the rings (Fig. 5-a), it is obvious that the connection experienced distinct load drops following the lower split rings failure, until the maximum load was achieved (region A). Once the maximum load was reached, the load is assumed to be transferred by the bolts holding the split rings. At that point, the connections behaved very much like a bolted connection. This can be observed by comparison of the curve in region A (typical for split rings) and the amount of ductility the curve is showing in region B. Final failure was due to crushing of the wood (bearing) or row shear-out of the wood underneath the bolts at high deformation (30mm, which is typical in ductile bolted connections). It should be noted that the load-slip curves of the specimens tested without A/C bolts did not exhibit the region B behavior (Quenneville and Sauvé,1998). Connections tested using A/C bolts installed within the end distance exhibited a different behaviour compared to those tested with A/C bolts installed between the rings (Fig 5-b)). Ductility was more obvious in connections with A/C bolts installed within the end distance. The ductility observed in this type of connections is attributed to the excessive bearing deformation of the wood underneath the rings. This is evident in the load-slip curves shown in Fig. 5. The reader can notice how big region A in Fig. 5-b) compared to region A in Fig. 5-a). This is a measure of the ductility that connections with A/C bolts installed within the end distance compared to the ones with A/C installed between the rings. The advantage of having the A/C bolt installed within the end distance is that it reinforces and confines the wood surrounding and underneath the ring connector, leading to the rings failing at higher loads. The result is a considerable bearing deformation prior to split rings failure and a ductile connection. As an example, the maximum load, for a typical L6 connection (shown in Fig. 5-b), was reached at a deformation of 18.0 mm, while for a similar connection tested with
A/C bolts installed between the rings, the deformation was 6.0mm, Fig. 5-a). 250 200 Failure of a split ring Behaviour of a bolted connection 250 200 Failure of a split ring Behaviour of a bolted connection 150 150 100 100 50 Region A Region B 50 Region A Region B 0 0 10 20 30 40 Slip (mm) 0 0 10 20 30 40 Slip (mm) a) b) Figure 5. Typical load-deformation relationship for connections with A/C bolts: a) A/C bolts installed between the split rings; b) A/C bolts installed within the end distance. The artificial split created in the specimen did not seem to influence the residual strength of the split rings connection. In fact, it was noticed that the cracks were closing in as loading progressed with little or no force being developed in the A/C bolts. Furthermore, the A/C bolts did not contribute very much to the connection strength in the first stages of failure (region A). However, once the ring connectors failed, and the connection transformed into a bolted connection, the A/C bolts started working and enhanced the connection resistance beyond the failure point of the split rings. Data gathered from the load cells attached to the A/C bolts indicated that as the connection is loaded, hardly any tension force got developed in the A/C bolt. The tensile force in the A/C bolts started increasing just before the ultimate strength was reached (at ~85% of the ultimate strength). Following the split rings failure, the tensile force in the A/C bolts increased considerably. Thus, the A/C bolts are more effective in region B, when most or all split rings have already failed and reached their ultimate strength. Connections ultimate strength The maximum load sustained by the connection following the split ring at failure due to row shear-out, group tear-out or bearing of the wood underneath the rings is referred to as. The ultimate values for all specimens in groups L2, L4 and L6 tested with or without splits and A/C bolts are given in Table 1. The mean strength values together with the 5 th % value for each group are presented in Table 2. The 5 th % resistance values were calculated assuming a two-parameter Weibull distribution with a 75% confidence level (ASTM 1994). Comparison of the results of this study with the ones in Quenneville and Sauvé (1998) shows that there are no significant differences for the ultimate capacities of split connections and intact ones. The ratio of average of split members to intact members is 1.10, 0.91 to 0.97 and 1.0 to 1.1 for L2, L4 and L6, respectively. Thus, one can conclude that split tension members (with a split running between the rings) do not have a significantly lower strength than intact ones. This is still valid for the 5 th % values with 75% and 95% confidence. However, the fact that almost all specimens with split tension members have exceeded the rings maximum load following failure indicates that the connection can be reinforced using A/C bolts. For L4 and L6 connections, the A/C bolts have no significant effect on the strength as installed. Groups L4 and L6 with A/C bolts installed within the end distance have higher mean ultimate strength compared to those with A/C bolts installed between the rings (192.4 kn compared to 180 kn for L4 and 197.0 kn compared to 185.1 kn for group L6). The 5 th % values agree well for connections type L4 (110.78 kn compared to 103.5 kn), however, for type L6, the 5 th percentile value was smaller for connections with A/C bolts installed within the end distance by a factor of 0.91. Clearly, installing the A/C bolts within the end distance enhanced the performance of L4 and L6 split rings connections.
Table 1. Test results on the strength of split ring connections. Connection Type With split and A/C bolts within the end distance (1) With split and A/C bolts between the rings Un-damaged, no A/C bolts (2)... (kn)... --- 319.0 337.0 --- 307.3 360.7 --- 314.6 237.7 --- 350.3 337.7 L2 --- 387.4 279.3 --- 340.9 291.4 --- 342.4 346.9 --- 374.7 316.9 --- 278.9 283.4 --- --- 291.1 206.0 156.8 239.4 221.8 191.1 225.9 177.2 186.7 207.1 157.8 200.3 204.5 L4 185.1 194.7 183.2 183.3 201.0 208.7 177.6 173.1 178.6 205.2 166.6 188.1 191.0 147.4 166.9 218.7 183.1 173.9 200.0 201.4 192.7 204.3 176.8 191.1 215.0 191.5 207.3 202.7 183.2 167.8 L6 191.5 176.4 202.8 190.0 161.7 173.1 153.9 196.9 188.2 228.0 201.8 193.6 219.6 175.8 186.4 164.7 185.6 160.2 (1) Maximum load. (2) From Quenneville and Sauvé (1998). Examining the average tensile strength in the A/C bolts in both case revealed that a lower tensile force was developed in the A/C when installed within the end distance with comparison to connections having A/C bolts installed between the rings (4.6 kn compared to 16.1 kn type L4, and 7.3 kn compared to 11.4 for type L6). The maximum tensile forces recorded for groups L4 and L6 were well below what was recorded for connections with A/C between the rings. Maximum tensile force in type L4 connections was 19.2 kn and 16.0 kn with A/C bolts installed between the split rings compared to 6.2 kn and 9.5 kn for connections with A/C bolts installed within the end distance, respectively. This is quite interesting, since not only that a significant increase in the split connections capacity and a more ductile behaviour can be achieved just by installing the A/C bolts within the end distance, but also a lower tensile force get developed in the A/C bolts as a result. Thus, a smaller A/C bolt size can be used leading to a more efficient design. Findings from this study on the use of A/C bolts as means of reinforcing split ring connections with existing splits could
be expanded to cover those without splits (undamaged). Results indicate that reinforcing un-damaged split ring connections with A/C bolts leads to a more ductile and stronger connections. Investigations on the effectiveness A/C bolts in cases where the splits run along the center of the split rings rows may be necessary. This is a more severe damaged condition than the one investigated. The effect of using multiple A/C bolts (one within the end distance and the other between the split rings) should also be studied. Table 2. Summary of test results on the strength of split ring connections. Type of Connection (Mean) Un-damaged, no A/C bolts (1) 308.2 (11.7) With splits and A/C bolts installed between the split rings With splits and A/C bolts installed within the end distance 335.1 (9.6) L2 L4 L6 P 5 th % 189.0 193.0 Slip @ 2.6 (67.4) 6.6 (56.0) N/A N/A N/A (Mean) P 5 th % Slip @... (kn)... 197.6 (11.3) 180.1 (9.7) 192.4 (10.0) Note: Values in parenthesis are the coefficients of variations (COV). (1) From Quenneville and Sauvé (1998). 111.1 103.5 110.8 6.0 (27.8) 7.2 (70.7) 12.9 (31.3) (Mean) 186.3 (11.3) 185.1 (6.6) 197.0 (11.2) P 5 th % 113.6 116.2 106.2 Slip @ 5.7 (38.0) 5.0 (36.8) 18.6 (52.5) CONCLUSIONS Experimental investigations on the effectiveness of anti-check (A/C) bolts as means of repair for split connection members in large span timber hangars indicate that: a) Anti-check bolts significantly enhance the ductility of damaged connections. b) Connections performance could be improved by installing A/C bolts within the end distance instead of having them installed between the rings. c) A/C bolts could be used to reinforce un-damaged split ring connections and force a ductile failure model REFERENCES American Society for Testing and Materials.1994. Standard test methods for bolted connections in wood and wood-base products. ASTM Section D07.05.02. ASTM, Philadelphia, PA. Charron, A. and Quenneville, J.H.P.1994. Combined effects of end distance and spacing on the resistance of split ring connectors loaded in tension. Canadian Journal of Civil Engineering (CJCE), 21(5), 789-796. CETO (Morrison Hershfield Limited, consulting engineers). 1987. 34 m Warren type wood trusses and columns maintenance. Report no. 1840760-1, Department of National Defence (Canada). Quenneville, J.H.P. and Charron, A. 1996. Behaviour of single and double 102 mm split ring connections loaded in tension. Canadian Journal of Civil Engineering (CJCE), 23(13), 602-613. Quenneville, J.H.P. and Sauvé, G.1998. Strength and behaviour of full-scale split ring connections. Proceedings of the World Conference on Timber Engineering (WCTE), Montreux-Lausanne, Switzerland. Mohammad, M. and Quenneville, J.H.P. 1999. The effectiveness of anti-check bolts in split tension members. Journal of the Performance of Constructed Facilities (JPCF), ASCE, Vol. 13, No. 4., 157-162.