- Email: [email protected]
Experimental structural behavior of wall-diaphragm connections for older masonry buildings
Construction and Building Materials 26 (2012) 180–189
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Experimental structural behavior of wall-diaphragm connections for older masonry buildings Tsu-Jung Lin, James M. LaFave ⇑ Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA
a r t i c l e
i n f o
Article history: Received 8 March 2011 Received in revised form 13 May 2011 Accepted 13 June 2011 Available online 13 July 2011 Keywords: Brick masonry Wood Force vs. displacement curves Friction coefficient Monotonic and cyclic loading Wall-diaphragm connection Strap anchor connection
a b s t r a c t Wall-diaphragm connections can affect overall seismic performance of older unreinforced masonry buildings, but there is little test data available about the structural behavior of such connections. Results are presented of an experimental study designed to evaluate the behavior of typical brick wall to wood joist/diaphragm connections. Tests were conducted on two different types of component specimens (with and without nailed strap anchors), using three different loading methods (static monotonic, as well as static and dynamic cyclic). Contributions of friction (activated at brick joist supports to represent gravity load normal force effects) and of strap anchor nails loaded in shear have been considered separately and together in the testing matrix. Force vs. displacement envelope and hysteresis curves have been developed from the experimental data. Also from these data, simplified average multi-linear plots derived from all the experiments can be compared based on different test specimen and loading types, leading to aggregate findings about various distinctive structural behaviors exhibited. These findings include typical strengths and failure modes, as well as stiffness and/or friction coefficient values as a function of displacement, for all the test specimens. Results obtained from these masonry connection tests can be used in numerical analyses of whole building systems. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Brick masonry can be an esthetically pleasing, durable, and strong building material with good resistance to sound and thermal transmission. For these reasons, it has been a popular choice as a construction material for a variety of low-rise structural applications. However, unreinforced masonry (URM) structures can be relatively more vulnerable to earthquake excitations than steel, reinforced concrete, or even timber structures. During severe earthquakes, older URM buildings can exhibit a variety of damage mechanisms. In-plane and/or out-of-plane failures are the most likely damage modes for masonry walls. Local wall-diaphragm connection behavior may contribute to overall wall behavior, especially in the out-of-plane direction, depending on the nature of these connections. Early wall-diaphragm connections were often either star anchors or masonry anchors providing a positive attachment between the end of a wood floor joist and the brick masonry pocket in which it rested. One end of the steel anchor would be nailed to the web of a joist, with the other end embedded through the masonry wall to an external anchor plate. These types of anchors were typically only placed when a joist was perpendicular to and supported on a wall (at some of the pocket connections), but not for joists parallel to a wall. For global ⇑ Corresponding author. Fax: +1 217 265 8039. E-mail address: [email protected] (J.M. LaFave). 0950-0618/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2011.06.008
continuity of lateral load resistance in such structures, adequate overall connection is needed between the masonry walls and wooden floor diaphragms, which is typically provided by a mixture of the sort of connections described above along with other locations, where the joists simply rest in brick masonry wall pockets. In the literature, the global behavior of URM structures has been investigated by various researchers. Doherty et al. [1] conducted research on out-of-plane bending of multi-story URM walls. A simplified (linearized) displacement-based procedure was presented, along with recommendations for selection of an appropriate substitute structure to provide the most representative analytical results. Tri-linear force vs. displacement relationships were used to characterize nonlinear wall behavior of unreinforced brick masonry as rigid and semi-rigid blocks. A substitute structure concept was applied to further simplify single-degree-of-freedom (SDOF) models so the behavior of URM walls could be predicted using displacement response spectra. Simsir et al. [2] summarized research on out-of-plane behavior of URM bearing walls in buildings subjected to earthquake motions. Results from a set of shake-table tests revealed that such walls can perform quite well even if moderately intense base motions are applied to fairly slender walls. Experimental results were compared with those simulated using SDOF and multi-degree-of-freedom (MDOF) computational models, which were then used to establish that permissible limits on wall slenderness, as prescribed by some seismic design guidelines, could be increased.
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
In-plane shear behavior of masonry walls has also generated significant interest in development of appropriate assessment and analysis methods. Sutcliffe et al. [3] proposed a lower-bound limit analysis technique for URM shear walls, treating the masonry as an anisotropic, inhomogeneous, and perfectly plastic material. A simple seismic assessment approach for in-plane response of brick masonry walls was outlined by Magenes and Calvi [4]. Strength, deformability, and energy dissipation capacity of unreinforced brick masonry walls were evaluated, with shear failure mechanisms and shear strength formulae identified and formed, respectively, based on experimental results from Calvi et al. [5]. Peralta et al. [6] conducted an experimental testing program on the lateral in-plane behavior of pre-1950s existing and rehabilitated wood floor and roof diaphragms representative of URM buildings found in the central and eastern regions of the United States. They found that FEMA 273 tended to overpredict the stiffness and significantly underpredict yield displacement and ultimate deformation levels, while FEMA 356 tended to underpredict stiffness and overpredict yield displacement. An experimental investigation on the dynamic behavior of reduced-scale URM buildings, including both in- and out-of-plane walls with flexible diaphragms, was conducted by Costley and Abrams [7]. Experimental parameters included the relative lateral strengths of the two parallel shear walls and the aspect ratios of piers between window and door openings. According to their test results, measured frequencies were much lower (longer periods) than those determined using design codes. Substantial strength and deformation capacity still existed after the walls cracked (and rocked) during the experiments, indicating that there was some ductility within the structure. It was suggested that story drift can be used to define different performance levels for URM buildings in performance-based design approaches. Yi et al. [8,9] conducted full-scale tests and finite element model (FEM) simulations for a two-story URM building. Their test structure exhibited large initial stiffness, and its damage was characterized by sizeable discrete cracks that developed in the masonry walls. Global rocking of an entire wall and local response such as rocking and sliding of each individual pier were observed in masonry walls with different configurations. Elastic and inelastic FEMs included different degrees of complexity – rigid body analyses and nonlinear pushover analyses were conducted. It was concluded that interactions between masonry walls and flexible roof/floor diaphragms are in part determined by relative stiffness values of the basic components (like the in-plane wall, out-of-plane walls, and flexible diaphragms) of a URM building. Their tests also revealed that connection details in general between masonry walls and diaphragms can influence response of the wall-diaphragm system. Most research done investigating masonry buildings has emphasized structural components such as masonry walls or diaphragms, without much attention being given to the connections between brick masonry walls and the wood joists/diaphragms. Cross and Jones [10,11] outlined the development of a technique for examining seismic performance of joist and beam bearing connections in URM structures. They stipulated that an understanding of the connections can allow for better estimation of the overall structural behavior of brick buildings, and provide a useful tool for the design of seismic retrofit details. An FEM that accounts for friction and impact behavior at the diaphragm-to-wall interface was developed. Some MDOF systems of portal frames and cantilever beams illustrated the method and demonstrated its ability to capture sliding and impact behavior at the connection detail. Applying this approach, a historic brick building shaken during the Loma Prieta earthquake of 1989 was modeled. A review of the literature has indicated that wall-diaphragm connections can have a significant influence on the seismic performance of URM buildings. Failure of the connections could lead to
181
total structural collapse, and connection flexibility could significantly affect overall structural response. However, relatively little research has been conducted on the structural behavior of walldiaphragm connections for URM buildings under various loadings, such as to even determine their basic force vs. displacement relationships. Due to this lack of data on inelastic force–displacement behavior of wall-diaphragm connections, an experimental study of representative wood joist and brick masonry connections has been undertaken. The experimental data, such as overall force–displacement curves or even approximate stiffness values of connections, can then be used in developing numerical models of entire structures to better determine their response to ground motions. 2. Background In this section, a brief description is given about certain construction details that are typically found in older URM buildings in the US (commonly constructed for example from the 1920s until the 1960s). A series of editions of Architectural Graphic Standards [12] consistently indicates a method by which wooden floor joists were connected to brick masonry walls that supported them (including the type of nailed anchors used at some of these connections). The International Library of Technology [13] also shows similar walldiaphragm connection details, where wood joists rest on brick masonry walls in pocketed connections (with some of them having a metal strap nailed to the joist that also goes into/through the wall). Such connections, as shown in Fig. 1, are quite similar to the connection subassemblies constructed, tested, and reported on as part of the experimental program that is the subject of this paper. Floor diaphragms and wall-diaphragm connections in typical older URM buildings also have some other common characteristics: sheathing was typically made of straight members (nailed onto the framing), and the steel wall anchors used to connect the diaphragm to the wall were at most present at every 4th floor joist (and were often even less prevalent than that) [14]. The joists themselves were supported in pockets of the URM wall, with a modest bearing area. Sometimes joists or beams were simply supported on special corbels, but most often the URM wall was constructed around the supported beams, either with bricks tightly fitting around them or with a weak grout used to fill oversized cavities housing their supports [15]. Fig. 2a and b shows floor joists connected to a supporting masonry wall in an actual URM building, and of a floor joist that has a metal strap nailed to it that is also connected through the masonry wall. Most of the floor joists, however, had no such strap. While all of these components appear to be in fairly good condition in this particular older URM building, condition assessments of certain of such structures can reveal some wood deterioration and/or steel corrosion. The effect of that deterioration on structural performance is unknown and is outside the scope of the experimental work presented in this paper.
Fig. 1. Image of wall-diaphragm connection (International Library of Technology, 1923 [13]).
182
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 2. URM buildings: (a) wood floor joists connected to a supporting brick masonry wall and (b) detail of a strap anchor connection [16].
Fig. 3. Test specimen of wall-diaphragm connection: (a) side view and (b) front view.
3. Test set-up Nineteen wall-diaphragm connection test specimens were constructed by professional masons in the Newmark Structural Engineering Laboratory (NSEL) at the University of Illinois. Each specimen consisted of a small portion of brick masonry wall supporting a wood joist in a pocket joint, as shown in Fig. 3a and b. Each subassembly consisted of 18 clay bricks (with nominal dimensions of 400 22=3 00 800 ) constructed using Type S Portland cement mortar, and the section of wood joist was nominally 200 1200 (1½ in. 111=4 in. actual) and 24 in. in length. Fifteen of the specimens had wall anchors made of a steel strap (1200 1½00 1/800 ) and a threaded rod (½00 dia. 13 threads per in. 1200 long), welded together. The wall anchor and wood joist were connected by two 10 d (300 long) bright common nails, and the threaded rod was anchored outside the masonry with a standard hex nut and washer. (All these details are representative of those seen in the classic references and actual buildings described above.) Fig. 4a and b illustrates the size of the test specimens. The specimens were tested under uniaxial loading (in the joist longitudinal direction) in a testing machine. The brick masonry portion of the assembly was held down in place by two vertical clamps, made of steel angles and threaded rods. The top front brick of each specimen was removed for testing to facilitate ‘‘gravity load’’ application. Two additional horizontal steel clamps were used – a lower clamp to prevent the brick masonry assembly from cracking (by applying a surcharge load representing some modest wall axial compression), and an upper clamp to apply a normal compression force between the joist and the base of the brick wall pocket (representing the joist end arbitrary-point-in-time gravity load shear/bearing reaction), as shown in Fig. 5a. Two load cells were used to measure the force in the upper clamp (which had Teflon between it and the wood joist to limit any friction at that interface). This force was set to a representative value of around 850–900 lbs at the beginning of each test, but was sometimes varied during certain tests to fully explore the different structural behaviors (i.e., strength and stiffness) that might occur as a function of various amounts of friction
between the wood joist and brick masonry wall section. The free end of the wood joist was attached to the actuator of the testing machine through a U-shaped clamp, as shown in Fig. 5b. Two cable-extension position transducers (‘‘yo–yo’’ gauges) were used to collect relative displacement data between the wood joist and brick assembly. Overall load (and displacement) data from the testing machine hydraulic actuator was also collected, as well as that from the two load cells (representing the compression force produced by the upper horizontal clamp).
4. Test specimens and testing methods Two mechanisms of force transfer can exist in brick-joist walldiaphragm connections. Mechanical connection is provided by nails that fasten the anchored steel strap to the wood joist, and frictional resistance exists between the wood joist and brick due to the loaded joists bearing on the wall. Therefore, some specimens were tested with only nails (6), a few with only friction (4), and even more with both mechanisms in play (9); a half-dozen specimens from this latter group were also tested further with only friction after the nails had failed, to supplement the other friction-only data. Tests were performed using three different methods – static (slow) monotonic loading, quasi-static cyclic loading (max. loading frequency = 0.02 Hz), and dynamic cyclic loading (loading frequency = 2 Hz).
Fig. 4. Size of the wall anchor, wood joist, and masonry: (a) side cross-section view and (b) plan view.
183
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
5.1. Specimens including both nails and friction At some wall-diaphragm connections in typical URM buildings, the floor joist has a metal strap nailed into it that is also connected through the masonry wall. Experiments on this category of specimen (NF), which included both nail and friction contributions, best reflect the behavior of this class of wall-diaphragm connection in URM buildings. The normal force (which was applied by the upper clamp) between the joist and the base of the brick wall pocket was equal to 900 lbs in these cases. Structural response of this type of specimen when subjected to different loading situations is discussed in the sub-sections below.
Fig. 5. Clamping devices, displacement gages, and load cells.
5. Test results Based on the general descriptions provided above, the test specimens can be categorized by whether they had nails and friction (NF), nails only (N), or friction only (F), under different loading schemes. Since the key wall-diaphragm connections in real URM buildings resist loads by nailed straps that are connected to the wood and anchored through the masonry, test results for the NF case under static monotonic (SM) tension (opening) only, quasi-static cyclic (SC), and dynamic cyclic (DC) loadings will be emphasized, followed by some additional comparisons to N and F behavior. A listing of all specimens tested, including their type, loading, and failure mode/load, is given in Table 1. Force vs. displacement curves were obtained from conducting the experiments using the three different loading types. For ease of comparison, backbone envelope curves for cyclically loaded specimens were determined based on the peak points for each cycle. Then, average curves were obtained by further simplifying the envelope curves to piece-wise linear for the tension (joist pull-out) portion of behavior. This is a similar approach to what was done in another related study of brick masonry that tested and reported on the structural behavior of metal tie connections typically used in residential brick veneer construction [17].
5.1.1. Monotonic loading Specimens NF_SM1 and NF_SM2 were each tested to failure under monotonic loading. They both failed as the two nails sequentially sheared off at the head. Failure of each nail was accompanied by a steep drop in the force vs. displacement curve, as may be seen in Fig. 6a, with those key displacements and the overall maximum force given in Table 1. While specimen NF_SM2 had a larger load-carrying capacity, its nails were sheared off after going through smaller displacements (compared with specimen NF_SM1), which could be considered as less ductile behavior. Prior to any significant ‘‘yielding’’, these force–displacement curves could be approximated by three linear regions. When the connection is in a linear elastic state, lines o–a and o–a0 can be chosen as the initial slopes for specimens NF_SM1 and NF_SM2, respectively. As the connection gets softer due to some modest local nail slip and prying into the wood joist, lines a–b (and a0 –b0 ) represent a second slope, followed by line b–c (and b0 –c0 ) chosen as the third slope. After the connection fully enters its yielding stage (related to shear yield and some additional prying of both nails), no obvious slope change is seen across a fairly wide range of displacements, with point d chosen as the end of this ‘‘yield plateau’’. After that, the connection begins to soften even further, exhibiting negative stiffness as the first nail approaches a combined shear/tension failure culminating in its fracture (d–e–f). The connection then again becomes somewhat more stable for a time until the second nail also fails (f–g–h), after which all that remains is some very modest resistance (approximately 200–300 lbs) attributable only to friction (h–i). By taking an average of the force and displacement values for each of points a
Table 1 Experimental results of test specimens. Specimen name
Failure mode
Load capacity (lbs)
Displacement(s) when two nails failed (in.)
Remark
NF_SM1 NF_SM2 NF_SC1 NF_SC2 NF_SC3 NF_SC4 NF_SC5 NF_DC1 NF_DC2 N_SM1 N_SM2 N_SC1 N_SC2 N_SC3 N_DC1 F_SC1 F_SC2 F_DC1 F_DC2
2 2 2 2 2 2 1 2 2 2 1 2 2 2 2 – – – –
1500 1850 1760 1310 1890 1500 1310 1700 1900 1515 1600 1300 1700 1480 1370 – – – –
0.6, 1.3 0.4, 0.6 0.54, 0.57 0.53, 0.75 0.36, 0.42 1.0–1.8 0.95–1.61 0.12, 0.18 0.4–0.45 1.4–2.7 0.5, 1.6–2.4 0.75, 0.92 0.54, 0.76 0.67, 0.88 0.56–0.76 – – – –
⁄ ⁄ ⁄
nails shear off at head nails shear off at head nails shear off at head nails shear off at head nails shear off at head nails pull out nail shear off at head, 1 nail pull out nails shear off at middle nails shear off at head nails pull out nail shear off at head, 1 nail pull out nails shear off at head nails shear off at head nails shear off at head nails shear off at head
⁄ ⁄ ⁄
Note: NF – nails and friction; N – nails only; F – friction only; SM – static monotonic; SC – static cyclic; DC – dynamic cyclic; ⁄ represents specimens tested further in friction after nails failed. (For some specimens, displacement at nail failure was not easy to distinguish, so a displacement range is given in such cases.)
184
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 6. Force vs. displacement curves for type NF_SM specimens.
through i (and their a0 through i0 counterparts), points A through I can be obtained, as shown in Fig. 6b connected as a multi-linear average representing these two monotonic tests having nails and friction. 5.1.2. Quasi-static cyclic loading Five specimens were tested with nails and friction under quasistatic cyclic loading. The ultimate failure mode for specimens NF_SC1, NF_SC2, and NF_SC3 was two nails shearing off at the head, after some modest slip, prying, and yielding (similar to as earlier for specimens NF_SM1 and NF_SM2), as shown in Fig. 7a. For specimen NF_SC4, the eventual failure mode was both nails fully pulling out, as shown in Fig. 7b, and in specimen NF_SC5 one nail sheared off at the head while the second one pulled out, as shown in Fig. 7c. (After the nails sheared off in specimens NF_SC1 and NF_SC4, testing continued with only friction, the results of which will be compared later to other friction only (F) tests.). Behavior of all these type NF wall-diaphragm connections under quasi-static cyclic loading is next discussed, in part as a function of the failure modes of nails shearing off vs. nails pulling out. 5.1.2.1. Failure mode of nails shearing off. Three specimens (NF_SC1 through NF_SC3) failed as both nails sheared off. A sample cyclic force vs. displacement curve (for specimen NF_SC2), shown in Fig. 8, is qualitatively fairly representative of this type of behavior. Maximum resistance of around 1300 lbs was achieved at a displacement of about 0.25 in., with two later drops in load at about 1150 lbs and 650 lbs (at displacements of around 0.5 in. and 0.75 in.) representing the two nails shearing off at the head. After the nails sheared off, only the friction force is left, which was around 250 lbs in the later cycles. The unsymmetrical behavior of the hysteresis curve in compression vs. tension is due to the wood joist end eventually coming to bear (sometimes at modest displacements, generating fairly large forces) on excess mortar droppings in the gap between the wood joist
and the bricks when the wood joist moves toward the brick assembly. In order to simplify the type of hysteretic force vs. displacement curve shown in Fig. 8, the maximum points can be picked out for each cycle and all connected to form an envelope curve. The socalled ‘‘average’’ curve, shown only in the joist pull-out (tension) direction of loading, is then simply made as a piece-wise linear version of the envelope. Force vs. displacement curves for specimens NF_SC1 and NF_SC3 were similar, although they were each able to carry a bit more load than NF_SC2 and had more sudden drops in load resisted as both nails failed in rapid-fire succession, leaving a friction force of from 300 to over 500 lbs as the only remaining resistance mechanism. The same procedure described above for type NF_SM specimens was applied here to obtain a simplified average tension behavior for specimens NF_SC1 through NF_SC3, as shown in Fig. 9. An overall average curve for this set of specimens exhibiting nail shear failure is also provided in that figure, with strength similarities and some slight deformation differences versus those seen earlier in Fig. 6b for the NF_SM case. 5.1.2.2. Failure mode of nails pulling out. Specimens NF_SC4 and NF_SC5 each had part of their failure mode controlled by nail(s) pulling out of the wood joist, rather than only by nail shear failure (in NF_SC4, both nails pulled out, while in NF_SC5 one nail sheared off and another one pulled out). This failure mode (see Fig. 10 for the load–displacement data) appears to be a bit more ductile than in other NF_SC specimens that had their behavior governed by both nails shearing off. In these two specimens, which exhibited quite similar behavior to one another (in spite of their slightly different failure modes), there was a noticeable drop in load-carrying capacity at a displacement of around 1 in. – in NF_SC4 this was due to partial pullout of both nails, whereas in NF_SC5 it was a reflection of one of the nails fracturing. Subsequent behavior in each case (up to failure at displacements of about 1.5–1.75 in.) is then entirely related to
Fig. 7. Different failure modes: (a) two nails sheared off; (b) two nails pulled out; (c) one nail sheared off, another pulled out.
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 8. Force vs. displacement, envelope, and average curves for specimen NF_SC2.
185
Fig. 11. Static cyclic loading envelope plots for specimens NF_SC4 and NF_SC5.
groups of ten cycles at constant amplitude, starting from a target of about +/0.02 in. and then increasing in whole number multiples of that from there. Specimens NF_DC1 and NF_DC2 each exhibited an ultimate failure mode of both nails shearing off, albeit at lower ultimate displacements than otherwise similar failures of type NF_SM and NF_SC specimens. An example (for NF_DC1) of dynamic load–displacement behavior is shown in Fig. 12, where it may also be seen that the dynamic structural response is similar to the behavior in the static cyclic case (especially when nails sheared off). The overall average NF_DC envelope curve is also shown in Fig. 12.
Fig. 9. Static cyclic loading envelope plots for specimens NF_SC1 through NF_SC3.
Fig. 10. Force vs. displacement curves of specimens NF_SC4 and NF_SC5, including envelope curves.
nail pullout, after which only a frictional capacity of about 200– 250 lbs remained. Using procedures already described above for other loading and failure types, individual multi-linear load–displacement plots for specimens NF_SC4 and NF_SC5 were produced, as presented in Fig. 11. An overall average of these for the two specimens is also given in the figure (lines A–I), with the end slightly extended to more clearly show the frictional capacity remaining after the nails have pulled out (and, in one case, fractured). 5.1.3. Dynamic (cyclic) loading Dynamic behavior of the NF test specimens was a little bit different from their static behavior. Dynamic loading was typically in
5.1.4. Comparison of average envelope curves for type NF specimens The most common failure mode observed in the tests for NF cases was two nails shearing off, in all NF_SM and many of the NF_SC experiments. A comparison of the average load–displacement curves for NF_SM and NF_SC tests exhibiting this failure mode is presented in Fig. 13a, which show very similar load-carrying capacity (around 1600 lbs) and displacement at first nail failure (about 0.5 in.), when the load has dropped to about 1500 lbs. The second nail shears off on average at a somewhat smaller displacement in type NF_SC tests than in NF_SM tests. For all type NF specimens, average load–displacement curves under static monotonic, static cyclic, and dynamic cyclic loading (representing all failure modes) are compared in Fig. 13b. Connection behavior of the NF_SM and NF_SC tests are quite similar, especially in terms of overall capacity and the force when each nail failed. NF_SM tests exhibited the greatest ductility, while NF_DC tests had the least (but were also the strongest, on average). As can be seen in Fig. 13, the average drop in force when a nail failed in the NF tests was about 700 lbs. Assuming that these were in fact pure shear failures (there may have also been a modest tension contribution from prying), then the shear strength of the nails can be estimated as this force divided by the nail cross-sectional area, which results in an average nail shear failure stress of 40 ksi. This represents a steel nail material with ultimate tensile strength of about 67 ksi. (For other nail sizes and/or materials, certain aspects of the results could perhaps be scaled accordingly.) 5.2. Specimens including nails only For this type of connection (N), only nails connected the wall (through the strap anchor) to the wood joist (without any clamping normal force acting on the connection to enable a frictional contribution). This connection scenario may reflect the situation in a URM building when the vertical ground acceleration during an earthquake is downward (when there could possibly be only a very low normal force at the joist bearing location, with essentially no
186
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189 2000 N_SM N_SC N_DC NF_SM NF_SC NF_DC
1800
Force (lbs)
1600 1400 1200 1000 800 600 400 200 0 0
0.5
1
1.5
2
2.5
3
Disp. (in) Fig. 12. Dynamic cyclic force vs. displacement (and envelope) curves – specimen NF_DC1, and NF_DC average.
friction). This type of connection can also simply be used to better understand experimentally the behavioral contributions of just the nails to overall connection performance. Fig. 14 shows overall average curves for the N_SM, N_SC, and N_DC tests, where there are recognizable trends both among these behaviors and also compared with those of the respectively similar type NF tests. The loadcarrying capacity of type N specimens was typically up to 10–20% (as much as 150–350 lbs) less than in companion NF tests, while their displacements at nail failure were generally a bit greater (and the dynamic test cases showed the least ductility). 5.3. Specimens including friction only Only friction was acting on specimens under static and dynamic cyclic loading for this kind (F) of connection, which represents a common case in URM buildings of many joist-pocket connections without strap anchors. In addition to the primary specimens for this type of test (F_SC1, F_SC2, F_DC1, and F_DC2), a few NF specimens were also tested further with only friction after the nails had failed – for clarity, these tests are identified by adding ‘‘’’ onto the end of their names of NF_SM1, NF_SM2, NF_SC1, NF_SC4, NF_DC1, and NF_DC2. As can be seen in Fig. 15a, force vs. displacement curves obtained for specimen NF_SC1 are more symmetric than in NF_SC4 (which had higher compression forces). This latter behavior is attributed to the wood joist eventually coming into contact with excess mortar dropped into the gap between the wood joist end and the bricks, generating large forces when the wood joist moves far enough toward the brick assembly. This also occurred when specimens NF_DC1 and F_DC1 were tested (dynamically), as shown in Fig. 15b. Estimates of the friction coefficient between brick masonry and wood joists for different loading schemes can
Fig. 14. Average envelope curves for specimens with nails only vs. with nails and friction.
be obtained from analyzing the sort of cyclic test data shown in Fig. 15. Procedures for how such coefficients were obtained are described next, along with detailed findings about contributions of friction to structural behavior in this sort of connection. 6. Assessment of friction coefficient between wood and masonry The friction coefficient is defined as the applied testing machine load (frictional shear across the wood–brick interface) divided by the normal clamping force at that interface. Without a means to directly measure friction force, specimens with nails cannot be used to estimate the friction coefficient. For a few NF tests, however, a friction coefficient was still computed at the later cycles after the nails had failed (and therefore then contributed little to load resistance). Load cells continuously measured both the clamping force and applied load during testing, enabling the friction coefficient to be accurately determined throughout testing, even with slight variations in normal force (or for intentionally larger variations), as described here below. 6.1. Monotonic loading For NF cases under monotonic loading, the friction coefficient was obtained shortly after the nails broke. The clamping forces in specimens NF_SM1 and NF_SM2 were kept constant, equal to 860 and 815 lbs, respectively. The applied loads were 130 and 370 lbs, respectively, after their second nails failed (at displacements of 1.3 and 0.6 in., indicating perhaps less total sliding damage in the latter specimen), and then the specimens were slowly pulled up to a displacement of around 1.9 in., when the external loads were equal to 35 and 115 lbs (see Fig. 6). Therefore, the
Fig. 13. Comparison of average curves: (a) between NF_SM and NF_SC (only nails sheared off) and (b) between all NF_ SM, NF_SC, and NF_DC tests.
187
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 15. Force vs. displacement curves for: (a) specimens NF_SC1 and NF_SC4 and (b) specimens NF_DC1 and F_DC1.
Fig. 16. Force and displacement vs. time curves for specimen NF_SC1.
friction coefficient ranged from 0.15 down to around 0.05 for NF_SM1, and from 0.45 down to about 0.15 for NF_SM2.
6.2. Quasi-static cyclic loading Static cyclic tests F_SC1 and F_SC2 were with friction only, and NF_SC1 and NF_SC4 were also tested, with respect to friction after their nails failed. Fig. 16a shows displacement, external (applied) load, and load cell (clamping) force vs. time curves for specimen NF_SC1. The right ordinate axis represents the scale for loads, whereas the left ordinate axis represents the scale for displacement. NF_SC1 had eight loading cycles, with a maximum displacement amplitude of around 0.4 in. The 4th cycle (marked in a circle on Fig. 16a) is taken as an example, with Fig. 16b showing an expansion of just this cycle. At the beginning of specimen (tension) movement, the load cell indicated a clamping force of 1300 lbs. As the specimen was pulled to increasing displacements, the applied friction sliding load was relatively constant at around 680 lbs. A friction coefficient of 0.52 (680/1300) can then be obtained from average values of all external (applied) load and load cell (clamping) data for the duration of increasing displacement, along section AB. After that, the specimen was pushed back in the opposite direction, with the displacement decreasing to a negative value. A (negative direction) friction coefficient was obtained from the average values of external load and load cell force during this period, along section CD. By doing this, positive and negative direction friction coefficients at each cycle were obtained for all F_SC and NF_SC tests. Table 2 shows the average positive and negative direction friction coefficients for the whole group of cycles for specimens F_SC1, F_SC2, NF_SC1, and NF_SC4; the negative direction friction coefficients were typically a bit larger than those for the positive direction.
Table 2 Average positive and negative direction SC friction coefficients. Specimen name
Positive direction friction coefficient
Negative direction friction coefficient
F_SC1 F_SC2 NF_SC1 NF_SC4
0.32 0.63 0.63 0.27
0.37 0.65 0.72 0.44
6.3. Dynamic (cyclic) loading Specimens F_DC1 and F_DC2 were tested under dynamic cyclic loading, and NF_DC1 and NF_DC2 were also tested, with respect to friction after their nails failed. NF_DC2 is taken here as an example. Fig. 17a shows groups of cyclic loading; each group has ten cycles. The first group is expanded and shown in Fig. 17b, where the first cycle is then expanded again in Fig. 17c. Regions of roughly constant externally applied load occurring during steady sliding displacements have been selected (e.g., AB and EF for opening, and CD and GH for closing) to compute estimates of friction coefficient, which can then be averaged for each group of 10 cycles. Table 3 shows the average of all positive and negative direction friction coefficients for specimens F_DC1, F_DC2, NF_DC1, and NF_DC2; negative direction friction coefficient is again a bit greater than that in the positive direction for this type of connection/ loading. 6.4. Discussion In practice, some floor joists are connected to the supporting masonry with a metal strap, while others are not. Friction coefficients obtained from specimens representing these two cases, NF and F, are further examined here. For the NF tests, which
188
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 17. Force and displacement vs. time curves for specimen NF_DC2.
Comparison can also be made of friction coefficients between specimens F_SC1 and F_SC2, and F_DC1 and F_DC2, as shown in Fig. 18b. All of these friction coefficients decreased with increasing numbers of cycles. For F_SC1 and F_SC2, the friction coefficients are almost the same at the first cycles. However, with increased cycling the friction coefficient degraded faster in F_SC1 than in F_SC2; the external clamping force acting on F_SC2 was larger than that in F_SC1, which may have had some influence on these results. Friction coefficients for F_SC2, F_DC1, and F_DC2 are all similar after the first several cycles (where the cycle count actually represents groups of cycles for the DC tests). Based on results from this study (Fig. 18), friction coefficients between wood and brick masonry may be recommended for use in structural design and/or assessment. The average measured friction coefficient was around 0.5, with a lower bound of approximately 0.2 (after substantial cyclic ‘‘damage’’) and an upper bound of nearly 0.8 (for best-case installations with almost no damage). These compare well with historically published values of friction coefficient between wood and brick masonry, such as those (0.5–0.6) given in the American Civil Engineer’s Handbook [18].
Table 3 Average positive and negative direction DC friction coefficients. Specimen name
Positive direction friction coefficient
Negative direction friction coefficient
F_DC1 F_DC2 NF_DC1 NF_DC2
0.52 0.42 0.35 0.32
– 0.49 0.41 0.44
are continuations of testing for specimens with nails and friction after the nails sheared off or pulled out, their friction coefficients are as shown in Fig. 18a, which decrease with increasing number of cycles (after starting at values of from 0.48 to 0.82). The larger friction coefficients for NF_SC1 (than in NF_SC4 and NF_DC1) may be due to the fact that the nails of NF_SC1 sheared off, so it went through less total absolute maximum and cumulative displacement prior to failure. The additional magnitude and/or cycles of displacement in the other two specimens may have smoothed the surface between the masonry and wood more in those cases.
0.9
0.9
specimens NF_SC1*
0.7
specimens NF_DC1*
(b) 0.8
specimens NF_SC4*
specimen F_SC1 specimen F_SC2
0.7 Friction coefficient
Friction coefficient
(a) 0.8 0.6 0.5 0.4 0.3
specimen F_DC2
0.5 0.4 0.3
0.2
0.2
0.1
0.1
0
specimen F_DC1
0.6
0 0
2
4
6
8
No. Group cycle times
10
12
0
2
4
6
8
10
12
14
16
No. Group cycle times
Fig. 18. Friction coefficients: (a) specimens NF_SC1, NF_SC4, and NF_DC1 and (b) specimens F_SC1, F_SC2, F_DC1, and F_DC2.
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
7. Summary and conclusions Seismic performance of brick masonry buildings can be affected by vulnerable wall-diaphragm connections. For older URM buildings, nails that fix a steel strap to certain wood joists are an important element of wall-diaphragm connection structural behavior, as is friction. By conducting experiments on representative wall-diaphragm connections under different loading schemes, their force vs. displacement behavior (including average envelope curves) has been developed. Any possible effects of wood deterioration and/or steel corrosion have not been addressed in these tests; wall-diaphragm connection structural behavior in such cases could be a bit different. For the case of connections with both nails and friction, which represent many typical URM building joist-brick connections, behavior of the wall-diaphragm connection under dynamic cyclic loading was more brittle than behavior under monotonic or quasi-static cyclic loading. Average load–displacement curves of those latter NF connections under monotonic and quasi-static loading were fairly coincident with one another and also with those of most specimens resisting force by virtue of nails only. Test results of all specimens with nails indicate that their strength typically ranged from about 1300 to 1900 lbs. Friction coefficients between brick masonry and wood were estimated from friction only tests and other tests with nails and friction after the nails had failed. In general, friction coefficients decreased with increasing relative (joist vs. brick) displacement, and friction coefficients measured after nails failed were larger when the nails failed by shear as compared to those, where the nails pulled out. Recommendations for friction coefficient ranges between wood and brick masonry have been provided and may be applied to different aspects of building design and/or performance assessment. And finally, all the test results could be used to help calibrate nonlinear finite element models of these types of connections. Acknowledgements This work was funded in part through the Mid-America Earthquake (MAE) Center, under a grant from the Earthquake Engineering Research Centers Program of the National Science Foundation
189
per Award No. EEC-9701785. The authors would like to thank NSEL Research Engineer Greg Banas and University of Illinois graduate student Zhongzhuo Li for their assistance with the laboratory testing. References [1] Doherty K, Griffith MC, Lam N, Wilson J. Displacement-based seismic analysis for out-of-plane bending of un-reinforced masonry walls. Earthquake Eng Struct Dynam 2002;31:833–50. [2] Simsir CC, Aschheim MA, and Abrams DP. Out-of-plane dynamic response of unreinforced masonry bearing walls attached to flexible diaphragms. In: 13th World conference on earthquake engineering, Vancouver, BC, Canada; August 1–6, 2004. [3] Sutcliffe DJ, Yu HS, Page AW. Lower bound limit analysis of un-reinforced masonry shear walls. Comput Struct 2001;79(4):1295–312. [4] Magenes G, Calvi GM. In-plane seismic response of brick masonry walls. Earthquake Eng Struct Dynam 1997;26(11):1091–112. [5] Calvi GM, Kingsley GR, Magenes G. Testing of masonry structures for seismic assessment. Earthquake Spectra 1996;12(1):145–62. [6] Peralta DF, Bracci JM, Hueste MD. Seismic behavior of wood diaphragms in pre1950s unreinforced masonry buildings. J Struct Eng 2004;130(12):2040–50. [7] Costley AC, and Abrams D. Dynamic response of URM buildings with flexible diaphragms. Thesis Report, Civil Engineering, University of Illinois at Urbana Champaign; 1995. [8] Yi T, Moon FL, Leon RT, Kahn LF. Lateral load tests on a two-story unreinforced masonry building. J Struct Eng 2006;132(5):643–52. [9] Yi T, Moon FL, Leon RT, Kahn LF. Analyses of a two-story unreinforced masonry building. J Struct Eng 2006;132(5):653–62. [10] Cross WB, Jones NP. Seismic performance of joist-pocket connections. Part I: Modeling. J Struct Eng 1993;119(10):2986–3007. [11] Cross WB, Jones NP. Seismic performance of joist-pocket connections. Part II: Application. J Struct Eng 1993;119(10):3008–23. [12] Ramsey CG, Sleeper HG. Architectural graphic standards. New York, United States of America; 1932. [13] International Library of Technology. International Text Book Company. Scranton; 1923. [14] Russell AP, Ingham JM, and Griffith MC. Comparing New Zealand’s unrein forced masonry details with those of other seismically active countries. In: 7th International masonry conference, London, UK; 2006. [15] Bruneau M. State-of-the-art report on seismic performance of unreinforced masonry buildings. J Struct Eng 1994;120(1):230–51. [16] Stokes TA. Identifying configuration and connection details of unreinforced masonry structures in Mid America. In: Earthquake engineering symposium for young researchers, M.C.F.E.E. Research, Editor. Charleston, South Carolina; 2004. [17] Choi YH, LaFave JM. Performance of corrugated metal ties for brick veneer wall systems. ASCE J Mater Civil Eng 2004;16(3):202–11. [18] Merriman M. American Civil Engineer’s Handbook. New York: John Wiley & Sons, Inc.; 1920.
Contents lists available at ScienceDirect
Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat
Experimental structural behavior of wall-diaphragm connections for older masonry buildings Tsu-Jung Lin, James M. LaFave ⇑ Department of Civil and Environmental Engineering, University of Illinois at Urbana-Champaign, Urbana, IL 61801, USA
a r t i c l e
i n f o
Article history: Received 8 March 2011 Received in revised form 13 May 2011 Accepted 13 June 2011 Available online 13 July 2011 Keywords: Brick masonry Wood Force vs. displacement curves Friction coefficient Monotonic and cyclic loading Wall-diaphragm connection Strap anchor connection
a b s t r a c t Wall-diaphragm connections can affect overall seismic performance of older unreinforced masonry buildings, but there is little test data available about the structural behavior of such connections. Results are presented of an experimental study designed to evaluate the behavior of typical brick wall to wood joist/diaphragm connections. Tests were conducted on two different types of component specimens (with and without nailed strap anchors), using three different loading methods (static monotonic, as well as static and dynamic cyclic). Contributions of friction (activated at brick joist supports to represent gravity load normal force effects) and of strap anchor nails loaded in shear have been considered separately and together in the testing matrix. Force vs. displacement envelope and hysteresis curves have been developed from the experimental data. Also from these data, simplified average multi-linear plots derived from all the experiments can be compared based on different test specimen and loading types, leading to aggregate findings about various distinctive structural behaviors exhibited. These findings include typical strengths and failure modes, as well as stiffness and/or friction coefficient values as a function of displacement, for all the test specimens. Results obtained from these masonry connection tests can be used in numerical analyses of whole building systems. Ó 2011 Elsevier Ltd. All rights reserved.
1. Introduction Brick masonry can be an esthetically pleasing, durable, and strong building material with good resistance to sound and thermal transmission. For these reasons, it has been a popular choice as a construction material for a variety of low-rise structural applications. However, unreinforced masonry (URM) structures can be relatively more vulnerable to earthquake excitations than steel, reinforced concrete, or even timber structures. During severe earthquakes, older URM buildings can exhibit a variety of damage mechanisms. In-plane and/or out-of-plane failures are the most likely damage modes for masonry walls. Local wall-diaphragm connection behavior may contribute to overall wall behavior, especially in the out-of-plane direction, depending on the nature of these connections. Early wall-diaphragm connections were often either star anchors or masonry anchors providing a positive attachment between the end of a wood floor joist and the brick masonry pocket in which it rested. One end of the steel anchor would be nailed to the web of a joist, with the other end embedded through the masonry wall to an external anchor plate. These types of anchors were typically only placed when a joist was perpendicular to and supported on a wall (at some of the pocket connections), but not for joists parallel to a wall. For global ⇑ Corresponding author. Fax: +1 217 265 8039. E-mail address: [email protected] (J.M. LaFave). 0950-0618/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2011.06.008
continuity of lateral load resistance in such structures, adequate overall connection is needed between the masonry walls and wooden floor diaphragms, which is typically provided by a mixture of the sort of connections described above along with other locations, where the joists simply rest in brick masonry wall pockets. In the literature, the global behavior of URM structures has been investigated by various researchers. Doherty et al. [1] conducted research on out-of-plane bending of multi-story URM walls. A simplified (linearized) displacement-based procedure was presented, along with recommendations for selection of an appropriate substitute structure to provide the most representative analytical results. Tri-linear force vs. displacement relationships were used to characterize nonlinear wall behavior of unreinforced brick masonry as rigid and semi-rigid blocks. A substitute structure concept was applied to further simplify single-degree-of-freedom (SDOF) models so the behavior of URM walls could be predicted using displacement response spectra. Simsir et al. [2] summarized research on out-of-plane behavior of URM bearing walls in buildings subjected to earthquake motions. Results from a set of shake-table tests revealed that such walls can perform quite well even if moderately intense base motions are applied to fairly slender walls. Experimental results were compared with those simulated using SDOF and multi-degree-of-freedom (MDOF) computational models, which were then used to establish that permissible limits on wall slenderness, as prescribed by some seismic design guidelines, could be increased.
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
In-plane shear behavior of masonry walls has also generated significant interest in development of appropriate assessment and analysis methods. Sutcliffe et al. [3] proposed a lower-bound limit analysis technique for URM shear walls, treating the masonry as an anisotropic, inhomogeneous, and perfectly plastic material. A simple seismic assessment approach for in-plane response of brick masonry walls was outlined by Magenes and Calvi [4]. Strength, deformability, and energy dissipation capacity of unreinforced brick masonry walls were evaluated, with shear failure mechanisms and shear strength formulae identified and formed, respectively, based on experimental results from Calvi et al. [5]. Peralta et al. [6] conducted an experimental testing program on the lateral in-plane behavior of pre-1950s existing and rehabilitated wood floor and roof diaphragms representative of URM buildings found in the central and eastern regions of the United States. They found that FEMA 273 tended to overpredict the stiffness and significantly underpredict yield displacement and ultimate deformation levels, while FEMA 356 tended to underpredict stiffness and overpredict yield displacement. An experimental investigation on the dynamic behavior of reduced-scale URM buildings, including both in- and out-of-plane walls with flexible diaphragms, was conducted by Costley and Abrams [7]. Experimental parameters included the relative lateral strengths of the two parallel shear walls and the aspect ratios of piers between window and door openings. According to their test results, measured frequencies were much lower (longer periods) than those determined using design codes. Substantial strength and deformation capacity still existed after the walls cracked (and rocked) during the experiments, indicating that there was some ductility within the structure. It was suggested that story drift can be used to define different performance levels for URM buildings in performance-based design approaches. Yi et al. [8,9] conducted full-scale tests and finite element model (FEM) simulations for a two-story URM building. Their test structure exhibited large initial stiffness, and its damage was characterized by sizeable discrete cracks that developed in the masonry walls. Global rocking of an entire wall and local response such as rocking and sliding of each individual pier were observed in masonry walls with different configurations. Elastic and inelastic FEMs included different degrees of complexity – rigid body analyses and nonlinear pushover analyses were conducted. It was concluded that interactions between masonry walls and flexible roof/floor diaphragms are in part determined by relative stiffness values of the basic components (like the in-plane wall, out-of-plane walls, and flexible diaphragms) of a URM building. Their tests also revealed that connection details in general between masonry walls and diaphragms can influence response of the wall-diaphragm system. Most research done investigating masonry buildings has emphasized structural components such as masonry walls or diaphragms, without much attention being given to the connections between brick masonry walls and the wood joists/diaphragms. Cross and Jones [10,11] outlined the development of a technique for examining seismic performance of joist and beam bearing connections in URM structures. They stipulated that an understanding of the connections can allow for better estimation of the overall structural behavior of brick buildings, and provide a useful tool for the design of seismic retrofit details. An FEM that accounts for friction and impact behavior at the diaphragm-to-wall interface was developed. Some MDOF systems of portal frames and cantilever beams illustrated the method and demonstrated its ability to capture sliding and impact behavior at the connection detail. Applying this approach, a historic brick building shaken during the Loma Prieta earthquake of 1989 was modeled. A review of the literature has indicated that wall-diaphragm connections can have a significant influence on the seismic performance of URM buildings. Failure of the connections could lead to
181
total structural collapse, and connection flexibility could significantly affect overall structural response. However, relatively little research has been conducted on the structural behavior of walldiaphragm connections for URM buildings under various loadings, such as to even determine their basic force vs. displacement relationships. Due to this lack of data on inelastic force–displacement behavior of wall-diaphragm connections, an experimental study of representative wood joist and brick masonry connections has been undertaken. The experimental data, such as overall force–displacement curves or even approximate stiffness values of connections, can then be used in developing numerical models of entire structures to better determine their response to ground motions. 2. Background In this section, a brief description is given about certain construction details that are typically found in older URM buildings in the US (commonly constructed for example from the 1920s until the 1960s). A series of editions of Architectural Graphic Standards [12] consistently indicates a method by which wooden floor joists were connected to brick masonry walls that supported them (including the type of nailed anchors used at some of these connections). The International Library of Technology [13] also shows similar walldiaphragm connection details, where wood joists rest on brick masonry walls in pocketed connections (with some of them having a metal strap nailed to the joist that also goes into/through the wall). Such connections, as shown in Fig. 1, are quite similar to the connection subassemblies constructed, tested, and reported on as part of the experimental program that is the subject of this paper. Floor diaphragms and wall-diaphragm connections in typical older URM buildings also have some other common characteristics: sheathing was typically made of straight members (nailed onto the framing), and the steel wall anchors used to connect the diaphragm to the wall were at most present at every 4th floor joist (and were often even less prevalent than that) [14]. The joists themselves were supported in pockets of the URM wall, with a modest bearing area. Sometimes joists or beams were simply supported on special corbels, but most often the URM wall was constructed around the supported beams, either with bricks tightly fitting around them or with a weak grout used to fill oversized cavities housing their supports [15]. Fig. 2a and b shows floor joists connected to a supporting masonry wall in an actual URM building, and of a floor joist that has a metal strap nailed to it that is also connected through the masonry wall. Most of the floor joists, however, had no such strap. While all of these components appear to be in fairly good condition in this particular older URM building, condition assessments of certain of such structures can reveal some wood deterioration and/or steel corrosion. The effect of that deterioration on structural performance is unknown and is outside the scope of the experimental work presented in this paper.
Fig. 1. Image of wall-diaphragm connection (International Library of Technology, 1923 [13]).
182
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 2. URM buildings: (a) wood floor joists connected to a supporting brick masonry wall and (b) detail of a strap anchor connection [16].
Fig. 3. Test specimen of wall-diaphragm connection: (a) side view and (b) front view.
3. Test set-up Nineteen wall-diaphragm connection test specimens were constructed by professional masons in the Newmark Structural Engineering Laboratory (NSEL) at the University of Illinois. Each specimen consisted of a small portion of brick masonry wall supporting a wood joist in a pocket joint, as shown in Fig. 3a and b. Each subassembly consisted of 18 clay bricks (with nominal dimensions of 400 22=3 00 800 ) constructed using Type S Portland cement mortar, and the section of wood joist was nominally 200 1200 (1½ in. 111=4 in. actual) and 24 in. in length. Fifteen of the specimens had wall anchors made of a steel strap (1200 1½00 1/800 ) and a threaded rod (½00 dia. 13 threads per in. 1200 long), welded together. The wall anchor and wood joist were connected by two 10 d (300 long) bright common nails, and the threaded rod was anchored outside the masonry with a standard hex nut and washer. (All these details are representative of those seen in the classic references and actual buildings described above.) Fig. 4a and b illustrates the size of the test specimens. The specimens were tested under uniaxial loading (in the joist longitudinal direction) in a testing machine. The brick masonry portion of the assembly was held down in place by two vertical clamps, made of steel angles and threaded rods. The top front brick of each specimen was removed for testing to facilitate ‘‘gravity load’’ application. Two additional horizontal steel clamps were used – a lower clamp to prevent the brick masonry assembly from cracking (by applying a surcharge load representing some modest wall axial compression), and an upper clamp to apply a normal compression force between the joist and the base of the brick wall pocket (representing the joist end arbitrary-point-in-time gravity load shear/bearing reaction), as shown in Fig. 5a. Two load cells were used to measure the force in the upper clamp (which had Teflon between it and the wood joist to limit any friction at that interface). This force was set to a representative value of around 850–900 lbs at the beginning of each test, but was sometimes varied during certain tests to fully explore the different structural behaviors (i.e., strength and stiffness) that might occur as a function of various amounts of friction
between the wood joist and brick masonry wall section. The free end of the wood joist was attached to the actuator of the testing machine through a U-shaped clamp, as shown in Fig. 5b. Two cable-extension position transducers (‘‘yo–yo’’ gauges) were used to collect relative displacement data between the wood joist and brick assembly. Overall load (and displacement) data from the testing machine hydraulic actuator was also collected, as well as that from the two load cells (representing the compression force produced by the upper horizontal clamp).
4. Test specimens and testing methods Two mechanisms of force transfer can exist in brick-joist walldiaphragm connections. Mechanical connection is provided by nails that fasten the anchored steel strap to the wood joist, and frictional resistance exists between the wood joist and brick due to the loaded joists bearing on the wall. Therefore, some specimens were tested with only nails (6), a few with only friction (4), and even more with both mechanisms in play (9); a half-dozen specimens from this latter group were also tested further with only friction after the nails had failed, to supplement the other friction-only data. Tests were performed using three different methods – static (slow) monotonic loading, quasi-static cyclic loading (max. loading frequency = 0.02 Hz), and dynamic cyclic loading (loading frequency = 2 Hz).
Fig. 4. Size of the wall anchor, wood joist, and masonry: (a) side cross-section view and (b) plan view.
183
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
5.1. Specimens including both nails and friction At some wall-diaphragm connections in typical URM buildings, the floor joist has a metal strap nailed into it that is also connected through the masonry wall. Experiments on this category of specimen (NF), which included both nail and friction contributions, best reflect the behavior of this class of wall-diaphragm connection in URM buildings. The normal force (which was applied by the upper clamp) between the joist and the base of the brick wall pocket was equal to 900 lbs in these cases. Structural response of this type of specimen when subjected to different loading situations is discussed in the sub-sections below.
Fig. 5. Clamping devices, displacement gages, and load cells.
5. Test results Based on the general descriptions provided above, the test specimens can be categorized by whether they had nails and friction (NF), nails only (N), or friction only (F), under different loading schemes. Since the key wall-diaphragm connections in real URM buildings resist loads by nailed straps that are connected to the wood and anchored through the masonry, test results for the NF case under static monotonic (SM) tension (opening) only, quasi-static cyclic (SC), and dynamic cyclic (DC) loadings will be emphasized, followed by some additional comparisons to N and F behavior. A listing of all specimens tested, including their type, loading, and failure mode/load, is given in Table 1. Force vs. displacement curves were obtained from conducting the experiments using the three different loading types. For ease of comparison, backbone envelope curves for cyclically loaded specimens were determined based on the peak points for each cycle. Then, average curves were obtained by further simplifying the envelope curves to piece-wise linear for the tension (joist pull-out) portion of behavior. This is a similar approach to what was done in another related study of brick masonry that tested and reported on the structural behavior of metal tie connections typically used in residential brick veneer construction [17].
5.1.1. Monotonic loading Specimens NF_SM1 and NF_SM2 were each tested to failure under monotonic loading. They both failed as the two nails sequentially sheared off at the head. Failure of each nail was accompanied by a steep drop in the force vs. displacement curve, as may be seen in Fig. 6a, with those key displacements and the overall maximum force given in Table 1. While specimen NF_SM2 had a larger load-carrying capacity, its nails were sheared off after going through smaller displacements (compared with specimen NF_SM1), which could be considered as less ductile behavior. Prior to any significant ‘‘yielding’’, these force–displacement curves could be approximated by three linear regions. When the connection is in a linear elastic state, lines o–a and o–a0 can be chosen as the initial slopes for specimens NF_SM1 and NF_SM2, respectively. As the connection gets softer due to some modest local nail slip and prying into the wood joist, lines a–b (and a0 –b0 ) represent a second slope, followed by line b–c (and b0 –c0 ) chosen as the third slope. After the connection fully enters its yielding stage (related to shear yield and some additional prying of both nails), no obvious slope change is seen across a fairly wide range of displacements, with point d chosen as the end of this ‘‘yield plateau’’. After that, the connection begins to soften even further, exhibiting negative stiffness as the first nail approaches a combined shear/tension failure culminating in its fracture (d–e–f). The connection then again becomes somewhat more stable for a time until the second nail also fails (f–g–h), after which all that remains is some very modest resistance (approximately 200–300 lbs) attributable only to friction (h–i). By taking an average of the force and displacement values for each of points a
Table 1 Experimental results of test specimens. Specimen name
Failure mode
Load capacity (lbs)
Displacement(s) when two nails failed (in.)
Remark
NF_SM1 NF_SM2 NF_SC1 NF_SC2 NF_SC3 NF_SC4 NF_SC5 NF_DC1 NF_DC2 N_SM1 N_SM2 N_SC1 N_SC2 N_SC3 N_DC1 F_SC1 F_SC2 F_DC1 F_DC2
2 2 2 2 2 2 1 2 2 2 1 2 2 2 2 – – – –
1500 1850 1760 1310 1890 1500 1310 1700 1900 1515 1600 1300 1700 1480 1370 – – – –
0.6, 1.3 0.4, 0.6 0.54, 0.57 0.53, 0.75 0.36, 0.42 1.0–1.8 0.95–1.61 0.12, 0.18 0.4–0.45 1.4–2.7 0.5, 1.6–2.4 0.75, 0.92 0.54, 0.76 0.67, 0.88 0.56–0.76 – – – –
⁄ ⁄ ⁄
nails shear off at head nails shear off at head nails shear off at head nails shear off at head nails shear off at head nails pull out nail shear off at head, 1 nail pull out nails shear off at middle nails shear off at head nails pull out nail shear off at head, 1 nail pull out nails shear off at head nails shear off at head nails shear off at head nails shear off at head
⁄ ⁄ ⁄
Note: NF – nails and friction; N – nails only; F – friction only; SM – static monotonic; SC – static cyclic; DC – dynamic cyclic; ⁄ represents specimens tested further in friction after nails failed. (For some specimens, displacement at nail failure was not easy to distinguish, so a displacement range is given in such cases.)
184
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 6. Force vs. displacement curves for type NF_SM specimens.
through i (and their a0 through i0 counterparts), points A through I can be obtained, as shown in Fig. 6b connected as a multi-linear average representing these two monotonic tests having nails and friction. 5.1.2. Quasi-static cyclic loading Five specimens were tested with nails and friction under quasistatic cyclic loading. The ultimate failure mode for specimens NF_SC1, NF_SC2, and NF_SC3 was two nails shearing off at the head, after some modest slip, prying, and yielding (similar to as earlier for specimens NF_SM1 and NF_SM2), as shown in Fig. 7a. For specimen NF_SC4, the eventual failure mode was both nails fully pulling out, as shown in Fig. 7b, and in specimen NF_SC5 one nail sheared off at the head while the second one pulled out, as shown in Fig. 7c. (After the nails sheared off in specimens NF_SC1 and NF_SC4, testing continued with only friction, the results of which will be compared later to other friction only (F) tests.). Behavior of all these type NF wall-diaphragm connections under quasi-static cyclic loading is next discussed, in part as a function of the failure modes of nails shearing off vs. nails pulling out. 5.1.2.1. Failure mode of nails shearing off. Three specimens (NF_SC1 through NF_SC3) failed as both nails sheared off. A sample cyclic force vs. displacement curve (for specimen NF_SC2), shown in Fig. 8, is qualitatively fairly representative of this type of behavior. Maximum resistance of around 1300 lbs was achieved at a displacement of about 0.25 in., with two later drops in load at about 1150 lbs and 650 lbs (at displacements of around 0.5 in. and 0.75 in.) representing the two nails shearing off at the head. After the nails sheared off, only the friction force is left, which was around 250 lbs in the later cycles. The unsymmetrical behavior of the hysteresis curve in compression vs. tension is due to the wood joist end eventually coming to bear (sometimes at modest displacements, generating fairly large forces) on excess mortar droppings in the gap between the wood joist
and the bricks when the wood joist moves toward the brick assembly. In order to simplify the type of hysteretic force vs. displacement curve shown in Fig. 8, the maximum points can be picked out for each cycle and all connected to form an envelope curve. The socalled ‘‘average’’ curve, shown only in the joist pull-out (tension) direction of loading, is then simply made as a piece-wise linear version of the envelope. Force vs. displacement curves for specimens NF_SC1 and NF_SC3 were similar, although they were each able to carry a bit more load than NF_SC2 and had more sudden drops in load resisted as both nails failed in rapid-fire succession, leaving a friction force of from 300 to over 500 lbs as the only remaining resistance mechanism. The same procedure described above for type NF_SM specimens was applied here to obtain a simplified average tension behavior for specimens NF_SC1 through NF_SC3, as shown in Fig. 9. An overall average curve for this set of specimens exhibiting nail shear failure is also provided in that figure, with strength similarities and some slight deformation differences versus those seen earlier in Fig. 6b for the NF_SM case. 5.1.2.2. Failure mode of nails pulling out. Specimens NF_SC4 and NF_SC5 each had part of their failure mode controlled by nail(s) pulling out of the wood joist, rather than only by nail shear failure (in NF_SC4, both nails pulled out, while in NF_SC5 one nail sheared off and another one pulled out). This failure mode (see Fig. 10 for the load–displacement data) appears to be a bit more ductile than in other NF_SC specimens that had their behavior governed by both nails shearing off. In these two specimens, which exhibited quite similar behavior to one another (in spite of their slightly different failure modes), there was a noticeable drop in load-carrying capacity at a displacement of around 1 in. – in NF_SC4 this was due to partial pullout of both nails, whereas in NF_SC5 it was a reflection of one of the nails fracturing. Subsequent behavior in each case (up to failure at displacements of about 1.5–1.75 in.) is then entirely related to
Fig. 7. Different failure modes: (a) two nails sheared off; (b) two nails pulled out; (c) one nail sheared off, another pulled out.
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 8. Force vs. displacement, envelope, and average curves for specimen NF_SC2.
185
Fig. 11. Static cyclic loading envelope plots for specimens NF_SC4 and NF_SC5.
groups of ten cycles at constant amplitude, starting from a target of about +/0.02 in. and then increasing in whole number multiples of that from there. Specimens NF_DC1 and NF_DC2 each exhibited an ultimate failure mode of both nails shearing off, albeit at lower ultimate displacements than otherwise similar failures of type NF_SM and NF_SC specimens. An example (for NF_DC1) of dynamic load–displacement behavior is shown in Fig. 12, where it may also be seen that the dynamic structural response is similar to the behavior in the static cyclic case (especially when nails sheared off). The overall average NF_DC envelope curve is also shown in Fig. 12.
Fig. 9. Static cyclic loading envelope plots for specimens NF_SC1 through NF_SC3.
Fig. 10. Force vs. displacement curves of specimens NF_SC4 and NF_SC5, including envelope curves.
nail pullout, after which only a frictional capacity of about 200– 250 lbs remained. Using procedures already described above for other loading and failure types, individual multi-linear load–displacement plots for specimens NF_SC4 and NF_SC5 were produced, as presented in Fig. 11. An overall average of these for the two specimens is also given in the figure (lines A–I), with the end slightly extended to more clearly show the frictional capacity remaining after the nails have pulled out (and, in one case, fractured). 5.1.3. Dynamic (cyclic) loading Dynamic behavior of the NF test specimens was a little bit different from their static behavior. Dynamic loading was typically in
5.1.4. Comparison of average envelope curves for type NF specimens The most common failure mode observed in the tests for NF cases was two nails shearing off, in all NF_SM and many of the NF_SC experiments. A comparison of the average load–displacement curves for NF_SM and NF_SC tests exhibiting this failure mode is presented in Fig. 13a, which show very similar load-carrying capacity (around 1600 lbs) and displacement at first nail failure (about 0.5 in.), when the load has dropped to about 1500 lbs. The second nail shears off on average at a somewhat smaller displacement in type NF_SC tests than in NF_SM tests. For all type NF specimens, average load–displacement curves under static monotonic, static cyclic, and dynamic cyclic loading (representing all failure modes) are compared in Fig. 13b. Connection behavior of the NF_SM and NF_SC tests are quite similar, especially in terms of overall capacity and the force when each nail failed. NF_SM tests exhibited the greatest ductility, while NF_DC tests had the least (but were also the strongest, on average). As can be seen in Fig. 13, the average drop in force when a nail failed in the NF tests was about 700 lbs. Assuming that these were in fact pure shear failures (there may have also been a modest tension contribution from prying), then the shear strength of the nails can be estimated as this force divided by the nail cross-sectional area, which results in an average nail shear failure stress of 40 ksi. This represents a steel nail material with ultimate tensile strength of about 67 ksi. (For other nail sizes and/or materials, certain aspects of the results could perhaps be scaled accordingly.) 5.2. Specimens including nails only For this type of connection (N), only nails connected the wall (through the strap anchor) to the wood joist (without any clamping normal force acting on the connection to enable a frictional contribution). This connection scenario may reflect the situation in a URM building when the vertical ground acceleration during an earthquake is downward (when there could possibly be only a very low normal force at the joist bearing location, with essentially no
186
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189 2000 N_SM N_SC N_DC NF_SM NF_SC NF_DC
1800
Force (lbs)
1600 1400 1200 1000 800 600 400 200 0 0
0.5
1
1.5
2
2.5
3
Disp. (in) Fig. 12. Dynamic cyclic force vs. displacement (and envelope) curves – specimen NF_DC1, and NF_DC average.
friction). This type of connection can also simply be used to better understand experimentally the behavioral contributions of just the nails to overall connection performance. Fig. 14 shows overall average curves for the N_SM, N_SC, and N_DC tests, where there are recognizable trends both among these behaviors and also compared with those of the respectively similar type NF tests. The loadcarrying capacity of type N specimens was typically up to 10–20% (as much as 150–350 lbs) less than in companion NF tests, while their displacements at nail failure were generally a bit greater (and the dynamic test cases showed the least ductility). 5.3. Specimens including friction only Only friction was acting on specimens under static and dynamic cyclic loading for this kind (F) of connection, which represents a common case in URM buildings of many joist-pocket connections without strap anchors. In addition to the primary specimens for this type of test (F_SC1, F_SC2, F_DC1, and F_DC2), a few NF specimens were also tested further with only friction after the nails had failed – for clarity, these tests are identified by adding ‘‘’’ onto the end of their names of NF_SM1, NF_SM2, NF_SC1, NF_SC4, NF_DC1, and NF_DC2. As can be seen in Fig. 15a, force vs. displacement curves obtained for specimen NF_SC1 are more symmetric than in NF_SC4 (which had higher compression forces). This latter behavior is attributed to the wood joist eventually coming into contact with excess mortar dropped into the gap between the wood joist end and the bricks, generating large forces when the wood joist moves far enough toward the brick assembly. This also occurred when specimens NF_DC1 and F_DC1 were tested (dynamically), as shown in Fig. 15b. Estimates of the friction coefficient between brick masonry and wood joists for different loading schemes can
Fig. 14. Average envelope curves for specimens with nails only vs. with nails and friction.
be obtained from analyzing the sort of cyclic test data shown in Fig. 15. Procedures for how such coefficients were obtained are described next, along with detailed findings about contributions of friction to structural behavior in this sort of connection. 6. Assessment of friction coefficient between wood and masonry The friction coefficient is defined as the applied testing machine load (frictional shear across the wood–brick interface) divided by the normal clamping force at that interface. Without a means to directly measure friction force, specimens with nails cannot be used to estimate the friction coefficient. For a few NF tests, however, a friction coefficient was still computed at the later cycles after the nails had failed (and therefore then contributed little to load resistance). Load cells continuously measured both the clamping force and applied load during testing, enabling the friction coefficient to be accurately determined throughout testing, even with slight variations in normal force (or for intentionally larger variations), as described here below. 6.1. Monotonic loading For NF cases under monotonic loading, the friction coefficient was obtained shortly after the nails broke. The clamping forces in specimens NF_SM1 and NF_SM2 were kept constant, equal to 860 and 815 lbs, respectively. The applied loads were 130 and 370 lbs, respectively, after their second nails failed (at displacements of 1.3 and 0.6 in., indicating perhaps less total sliding damage in the latter specimen), and then the specimens were slowly pulled up to a displacement of around 1.9 in., when the external loads were equal to 35 and 115 lbs (see Fig. 6). Therefore, the
Fig. 13. Comparison of average curves: (a) between NF_SM and NF_SC (only nails sheared off) and (b) between all NF_ SM, NF_SC, and NF_DC tests.
187
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 15. Force vs. displacement curves for: (a) specimens NF_SC1 and NF_SC4 and (b) specimens NF_DC1 and F_DC1.
Fig. 16. Force and displacement vs. time curves for specimen NF_SC1.
friction coefficient ranged from 0.15 down to around 0.05 for NF_SM1, and from 0.45 down to about 0.15 for NF_SM2.
6.2. Quasi-static cyclic loading Static cyclic tests F_SC1 and F_SC2 were with friction only, and NF_SC1 and NF_SC4 were also tested, with respect to friction after their nails failed. Fig. 16a shows displacement, external (applied) load, and load cell (clamping) force vs. time curves for specimen NF_SC1. The right ordinate axis represents the scale for loads, whereas the left ordinate axis represents the scale for displacement. NF_SC1 had eight loading cycles, with a maximum displacement amplitude of around 0.4 in. The 4th cycle (marked in a circle on Fig. 16a) is taken as an example, with Fig. 16b showing an expansion of just this cycle. At the beginning of specimen (tension) movement, the load cell indicated a clamping force of 1300 lbs. As the specimen was pulled to increasing displacements, the applied friction sliding load was relatively constant at around 680 lbs. A friction coefficient of 0.52 (680/1300) can then be obtained from average values of all external (applied) load and load cell (clamping) data for the duration of increasing displacement, along section AB. After that, the specimen was pushed back in the opposite direction, with the displacement decreasing to a negative value. A (negative direction) friction coefficient was obtained from the average values of external load and load cell force during this period, along section CD. By doing this, positive and negative direction friction coefficients at each cycle were obtained for all F_SC and NF_SC tests. Table 2 shows the average positive and negative direction friction coefficients for the whole group of cycles for specimens F_SC1, F_SC2, NF_SC1, and NF_SC4; the negative direction friction coefficients were typically a bit larger than those for the positive direction.
Table 2 Average positive and negative direction SC friction coefficients. Specimen name
Positive direction friction coefficient
Negative direction friction coefficient
F_SC1 F_SC2 NF_SC1 NF_SC4
0.32 0.63 0.63 0.27
0.37 0.65 0.72 0.44
6.3. Dynamic (cyclic) loading Specimens F_DC1 and F_DC2 were tested under dynamic cyclic loading, and NF_DC1 and NF_DC2 were also tested, with respect to friction after their nails failed. NF_DC2 is taken here as an example. Fig. 17a shows groups of cyclic loading; each group has ten cycles. The first group is expanded and shown in Fig. 17b, where the first cycle is then expanded again in Fig. 17c. Regions of roughly constant externally applied load occurring during steady sliding displacements have been selected (e.g., AB and EF for opening, and CD and GH for closing) to compute estimates of friction coefficient, which can then be averaged for each group of 10 cycles. Table 3 shows the average of all positive and negative direction friction coefficients for specimens F_DC1, F_DC2, NF_DC1, and NF_DC2; negative direction friction coefficient is again a bit greater than that in the positive direction for this type of connection/ loading. 6.4. Discussion In practice, some floor joists are connected to the supporting masonry with a metal strap, while others are not. Friction coefficients obtained from specimens representing these two cases, NF and F, are further examined here. For the NF tests, which
188
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
Fig. 17. Force and displacement vs. time curves for specimen NF_DC2.
Comparison can also be made of friction coefficients between specimens F_SC1 and F_SC2, and F_DC1 and F_DC2, as shown in Fig. 18b. All of these friction coefficients decreased with increasing numbers of cycles. For F_SC1 and F_SC2, the friction coefficients are almost the same at the first cycles. However, with increased cycling the friction coefficient degraded faster in F_SC1 than in F_SC2; the external clamping force acting on F_SC2 was larger than that in F_SC1, which may have had some influence on these results. Friction coefficients for F_SC2, F_DC1, and F_DC2 are all similar after the first several cycles (where the cycle count actually represents groups of cycles for the DC tests). Based on results from this study (Fig. 18), friction coefficients between wood and brick masonry may be recommended for use in structural design and/or assessment. The average measured friction coefficient was around 0.5, with a lower bound of approximately 0.2 (after substantial cyclic ‘‘damage’’) and an upper bound of nearly 0.8 (for best-case installations with almost no damage). These compare well with historically published values of friction coefficient between wood and brick masonry, such as those (0.5–0.6) given in the American Civil Engineer’s Handbook [18].
Table 3 Average positive and negative direction DC friction coefficients. Specimen name
Positive direction friction coefficient
Negative direction friction coefficient
F_DC1 F_DC2 NF_DC1 NF_DC2
0.52 0.42 0.35 0.32
– 0.49 0.41 0.44
are continuations of testing for specimens with nails and friction after the nails sheared off or pulled out, their friction coefficients are as shown in Fig. 18a, which decrease with increasing number of cycles (after starting at values of from 0.48 to 0.82). The larger friction coefficients for NF_SC1 (than in NF_SC4 and NF_DC1) may be due to the fact that the nails of NF_SC1 sheared off, so it went through less total absolute maximum and cumulative displacement prior to failure. The additional magnitude and/or cycles of displacement in the other two specimens may have smoothed the surface between the masonry and wood more in those cases.
0.9
0.9
specimens NF_SC1*
0.7
specimens NF_DC1*
(b) 0.8
specimens NF_SC4*
specimen F_SC1 specimen F_SC2
0.7 Friction coefficient
Friction coefficient
(a) 0.8 0.6 0.5 0.4 0.3
specimen F_DC2
0.5 0.4 0.3
0.2
0.2
0.1
0.1
0
specimen F_DC1
0.6
0 0
2
4
6
8
No. Group cycle times
10
12
0
2
4
6
8
10
12
14
16
No. Group cycle times
Fig. 18. Friction coefficients: (a) specimens NF_SC1, NF_SC4, and NF_DC1 and (b) specimens F_SC1, F_SC2, F_DC1, and F_DC2.
T.-J. Lin, J.M. LaFave / Construction and Building Materials 26 (2012) 180–189
7. Summary and conclusions Seismic performance of brick masonry buildings can be affected by vulnerable wall-diaphragm connections. For older URM buildings, nails that fix a steel strap to certain wood joists are an important element of wall-diaphragm connection structural behavior, as is friction. By conducting experiments on representative wall-diaphragm connections under different loading schemes, their force vs. displacement behavior (including average envelope curves) has been developed. Any possible effects of wood deterioration and/or steel corrosion have not been addressed in these tests; wall-diaphragm connection structural behavior in such cases could be a bit different. For the case of connections with both nails and friction, which represent many typical URM building joist-brick connections, behavior of the wall-diaphragm connection under dynamic cyclic loading was more brittle than behavior under monotonic or quasi-static cyclic loading. Average load–displacement curves of those latter NF connections under monotonic and quasi-static loading were fairly coincident with one another and also with those of most specimens resisting force by virtue of nails only. Test results of all specimens with nails indicate that their strength typically ranged from about 1300 to 1900 lbs. Friction coefficients between brick masonry and wood were estimated from friction only tests and other tests with nails and friction after the nails had failed. In general, friction coefficients decreased with increasing relative (joist vs. brick) displacement, and friction coefficients measured after nails failed were larger when the nails failed by shear as compared to those, where the nails pulled out. Recommendations for friction coefficient ranges between wood and brick masonry have been provided and may be applied to different aspects of building design and/or performance assessment. And finally, all the test results could be used to help calibrate nonlinear finite element models of these types of connections. Acknowledgements This work was funded in part through the Mid-America Earthquake (MAE) Center, under a grant from the Earthquake Engineering Research Centers Program of the National Science Foundation
189
per Award No. EEC-9701785. The authors would like to thank NSEL Research Engineer Greg Banas and University of Illinois graduate student Zhongzhuo Li for their assistance with the laboratory testing. References [1] Doherty K, Griffith MC, Lam N, Wilson J. Displacement-based seismic analysis for out-of-plane bending of un-reinforced masonry walls. Earthquake Eng Struct Dynam 2002;31:833–50. [2] Simsir CC, Aschheim MA, and Abrams DP. Out-of-plane dynamic response of unreinforced masonry bearing walls attached to flexible diaphragms. In: 13th World conference on earthquake engineering, Vancouver, BC, Canada; August 1–6, 2004. [3] Sutcliffe DJ, Yu HS, Page AW. Lower bound limit analysis of un-reinforced masonry shear walls. Comput Struct 2001;79(4):1295–312. [4] Magenes G, Calvi GM. In-plane seismic response of brick masonry walls. Earthquake Eng Struct Dynam 1997;26(11):1091–112. [5] Calvi GM, Kingsley GR, Magenes G. Testing of masonry structures for seismic assessment. Earthquake Spectra 1996;12(1):145–62. [6] Peralta DF, Bracci JM, Hueste MD. Seismic behavior of wood diaphragms in pre1950s unreinforced masonry buildings. J Struct Eng 2004;130(12):2040–50. [7] Costley AC, and Abrams D. Dynamic response of URM buildings with flexible diaphragms. Thesis Report, Civil Engineering, University of Illinois at Urbana Champaign; 1995. [8] Yi T, Moon FL, Leon RT, Kahn LF. Lateral load tests on a two-story unreinforced masonry building. J Struct Eng 2006;132(5):643–52. [9] Yi T, Moon FL, Leon RT, Kahn LF. Analyses of a two-story unreinforced masonry building. J Struct Eng 2006;132(5):653–62. [10] Cross WB, Jones NP. Seismic performance of joist-pocket connections. Part I: Modeling. J Struct Eng 1993;119(10):2986–3007. [11] Cross WB, Jones NP. Seismic performance of joist-pocket connections. Part II: Application. J Struct Eng 1993;119(10):3008–23. [12] Ramsey CG, Sleeper HG. Architectural graphic standards. New York, United States of America; 1932. [13] International Library of Technology. International Text Book Company. Scranton; 1923. [14] Russell AP, Ingham JM, and Griffith MC. Comparing New Zealand’s unrein forced masonry details with those of other seismically active countries. In: 7th International masonry conference, London, UK; 2006. [15] Bruneau M. State-of-the-art report on seismic performance of unreinforced masonry buildings. J Struct Eng 1994;120(1):230–51. [16] Stokes TA. Identifying configuration and connection details of unreinforced masonry structures in Mid America. In: Earthquake engineering symposium for young researchers, M.C.F.E.E. Research, Editor. Charleston, South Carolina; 2004. [17] Choi YH, LaFave JM. Performance of corrugated metal ties for brick veneer wall systems. ASCE J Mater Civil Eng 2004;16(3):202–11. [18] Merriman M. American Civil Engineer’s Handbook. New York: John Wiley & Sons, Inc.; 1920.