0 valutazioniIl 0% ha trovato utile questo documento (0 voti)
42 visualizzazioni9 pagine
Load-displacement behavior of composite wood-concrete floor / deck systems was studied. A novel shear key / anchor detail was adopted for overlaying solid wood floors. Laboratory testing included pull-out tests on the anchors and interlayer slip tests.
Load-displacement behavior of composite wood-concrete floor / deck systems was studied. A novel shear key / anchor detail was adopted for overlaying solid wood floors. Laboratory testing included pull-out tests on the anchors and interlayer slip tests.
Load-displacement behavior of composite wood-concrete floor / deck systems was studied. A novel shear key / anchor detail was adopted for overlaying solid wood floors. Laboratory testing included pull-out tests on the anchors and interlayer slip tests.
R. Gutkowski 1 , J. Balogh 2 , J. Natterer 3 , K. Brown 4 , E. Koike 4 , P. Etournaud 4 ABSTRACT Load-displacement behavior of composite wood-concrete floor/deck systems was studied. A novel shear key/anchor detail for overlaying solid wood floors in office buildings with a concrete layer, thus creating a composite floor, was adopted for this study. The detail utilizes a notch to transfer bearing stress and shear stress in the joined materials for interlayer force transfer. The anchor insures tight mating of the layered materials. Laboratory testing included pull-out tests on the anchors, interlayer slip tests on connection details, pilot load tests of full-scale rectangular layered beam specimens and two full-scale layered deck specimens. The objective was to determine an appropriate dimension of the shear/key anchor detail. An assessment of the degree of composite behavior was made based on test results for the bare wood decks and the composite specimen resulting from the addition of a concrete layer. Rigorous discreteized modeling of the beam and floor specimens is reported. INTRODUCTION In the U.S., commercial low to high rise construction in offices, hotels etc. is a largely untapped market for wood products. In such applications, floors are primarily made of cast-in-place reinforced concrete. Concerns about low fire resistance, low sound proofing and structural concerns of carrying loads higher than a residential 1.92 kN/m 2 (i.e. a standard 40 psf loading in U.S. codes) design live load have been deterrents to introducing traditional light frame wood construction have included. To dramatically improve the competitive market position of wood structures, either cost must diminish (either absolutely or relatively) or new applications must be developed. With wilderness area limitations, the Endangered Species Act, and increasing shipping costs exacerbating material supply, significant reduction in costs is unrealistic. In Europe, above issues exist, too, as well as a tradition for heavy stone or concrete construction. However, successful cases of the use of mixed wood-concrete, floor construction in multi-story buildings have occurred. Motivations beyond those stated earlier in the paper are overall reduction in dead weight, cost savings of replacing non-renewable resource based concrete and steel with a managed renewable resource, and savings in energy of material production and construction. Even a modest change from concrete and steel to more wood construction can substantially reduce energy requirements and carbon dioxide emissions (Natterer, 1997). These realities encourage considering the feasibility of wood-concrete composites as a new application of dimension lumber in the U. S. The research described herein provides a base of engineering data helpful to the community of architects and engineers who wish to assess the technical merit of pursuing this application. A reinforced concrete floor slab is highly wasteful of materials. The tensile zone cracks allowing moisture to enter leading to corrosion, spalling and other types of deterioration, and about 40-60% of the depth is ineffective, only holding rebar in place. Steel rebar is expensive (and, in this context, more expensive than wood). In addition, the exposed rebar is a potential fire protection problem. In the research described below is shown that the wasted concrete and expensive rebar can be replaced by a structurally effective, solid wood deck. The wood deck can replace normal form work for a solid slab, therefore is expedient to leave it in place, reduce the concrete slab thickness by about 50% and interconnect them to develop composite action. The competitive merit of such mixed construction is borne out by several examples of successful projects in Europe (Natterer, et al. 1996) involving wood-concrete deck floors in office buildings. A composite wood-concrete cross-section would be suitable for both strengthening of existing timber floor systems and for new construction applications. If used to construct industrial or office floor systems, advantages would include improved thermal and acoustical characteristics as compared to steel or concrete systems.
1 Professor, Dept. of Civil Engineering, Colorado State University, Fort Collins, CO 80523, USA 2 Faculty Affiliate, Dept. of Civil Engineering, Colorado State University, Fort Collins, CO 80523, USA 3 Professor-Chair, Institute of Wood Construction, Swiss Federal Institute of Technology, Lausanne, CH 1015, Switzerland 4 Former Graduate Student, Dept. of Civil Engineering, Colorado State University, Fort Collins, CO 80523, USA INTERLAYER CONNECTION DETAIL In flexure, the layered elements tend to experience relative longitudinal displacement (slip) at their interface (Fig. 1). The degree of partial composite action achieved is a function of the ability of the interlayer connections to reduce slip to low magnitude. There are two bounds of the composite action in terms of the stiffness: the lower bound when no composite action, and the upper bound when ideal (full) composite action is obtained. In terms of connection strength the partial composite action can be of full or partial strength. Figure 1. Interlayer Slip in a Layered Member The key element in such a layered system is the interlayer connection detail. In this study a notched shear key/anchor connection detail shown in Fig. 2 initially developed in Europe for commercial and apartment construction (Natterer, et al, 1996) was investigated. The work at Colorado State University focuses on accomplishing a similar detail using U.S. made products, as the cost and availability of the anchor and needed glue from Europe hinders adoption. Figure 2. Schematic of the Shear Key/Anchor Connection The intent of the shear key/anchor detail is to achieve interlayer shear transfer by bearing stresses in the wood and horizontal shear in the concrete key (or, if the notch is near the ends, in the wood). As interlayer slip is low, the anchor connector itself is not subjected to shear. It only resists uplift created by the bearing force on the notch incline. To avoid a right angle notch that might split at the base and a high perpendicular-to-grain component of the bearing force in the shear key, the angular cut of the shear key is made at 75 degrees (from the vertical). The anchor resists uplift created by the bearing force on the notch incline. A Hilti dowel is glued into a tapped pilot hole in the wood using a specified glue and procedure. A plastic sleeve is placed around the shank of the embedded dowel within the concrete layer allowing after the concrete curing, the dowel to be tightened by turning its head restoring a close contact between the layers. To protect it from exposure to moisture and heat of fire the void above the connector is grouted. Alternative anchoring devices and other glues were also examined. EXPERIMENTAL TESTS The main objective was to examine the structural behavior of wood-concrete floor specimens configured to be composite systems by static load tests. There were three primary associated goals: 1) to configure a notched shear key/anchor to effectively connect the layers, 2) to experimentally quantify the resulting degree composite action, and 3) to numerically predict the load-deflection response. Aspects of composite wood-concrete beam behavior were published and presented at an international conference (Gutkowski, 1996). Connection Tests In order to examine the load-slip behavior of several types of anchor details (Fig. 3), a set of 120 interlayer connection specimens (40 notched and with a Hilti dowel anchor, 40 notched and with a threaded bolt anchor (passing through the concrete and tightened with a nut and washer below) and 40 with no notch and a lag screw anchor) were tested in interlayer shear, using test equipment, specimens and procedures similar to those in past work (Pault, et al, 1977) and accepted in the research community (Pellicane, et al, 1984). For the anchor the Hilti C-50 HIT WTR epoxy was used. Half of the specimens in each group were made using a single nominal 89 mm x 89 mm (4 x 4) solid sawn member while the others were made using three nominal 38 mm x 89 mm dimension (2 x 4) lumber members (nailed together). All specimens had a 64 mm thick concrete layer and of two different 28 day concrete compressive strengths, namely namely 27.6 N/mm 2 and 35.2 N/mm 2 (as measured by ASTM standard tests). The same notch dimensions were used in all specimens. The results of the measured load resistance and slip were compared and reported in a M.S. independent study (Thompson, 1997). Due to its measurable superior strength and slip resistance, the dowel shear/key anchor detail was adopted. Figure 3. Typical Slip Test Results for a) mechanical connector in shear, and b) notched dowel connection in bearing Two alternative adhesives were then identified for anchoring the connector that are available in the U.S. in order to replace the glue specified by the dowel manufacturers, only available in Europe, costly to buy/ship and with a 90-day shelf life. These were: 1) a phenol-resorcinol laminating resin manufactured by Borden Chemical, Inc., and 2) another Hilti adhesive (HIT HY 150). A series of 70 anchor pull out tests were conducted (Brown, 1998) at three embedment depths in the wood, to compare performance of the adhesives and, later, as a quality control test to insure bond was achieved before casting the concrete layer of the beam and floor specimens. Both adhesives were very stiff up to failure followed by a sudden diminishing drop off in resistance as withdrawal occurred. The Borden resin was used in subsequent tests due to its about 3 times the capacity of the Hilti adhesive (mean of 290 kN/m compared to 91 kN/m) and less variability. Further interlayer shear tests were done (prior to beam tests) on 60 connection specimens using three different notch details (see Table 1) to examine different notch geometries (Brown, et al, 1998). For notch B slip tests, a wide range of values resulted. For the 2x4s, the mean slip modulus was 27,700 kN/m (range of 22,400-33,000 kN/m, COV = 23.4%). For the 4x4s, the mean slip modulus was 24,290 kN/m (range of 18,200-36,500 kN/m., COV = 23.3%). The higher values for the 4x4s partially reflect the larger bearing area of the 4x4 (37 cm 2 ) compared to the three 2x4s (29 cm 2 ), which had a ratio of 1.29:1. The corresponding ratio of mean slip moduli values for notch B specimens was 1.14:1 For the 2x4s, the mean maximum load was 82 kN (range of 55 -110 kN, COV = 24.7%). For the 4x4s, with the mean slip modulus was 82 kN (range of 60-99 kN, COV = 14.1%). The ratio of means is about 1:00:1. However, for all notches combined it was 84:78= 1.07:1. Table 1. Notch Details Designation Width (mm) Depth (mm) A 102 25.4 B 127 3.18 C 152 3.81 Load-slip response of the notched specimens differed from what is known to occur for mechanically connected wood-wood composites (see Fig. 3), being linear to initial failure, with a sudden drop off to a modest to minor ductile residual resistance thereafter. The Hilti (threaded bolt; lag screw) connector specimens had average slip moduli of 7.64 and 9.03 (4.99 and 8.47; 1.67 and 1.94) kN/mm for the 2 x 4s and 4 x 4s, respectively. Corresponding average ultimate loads at initial failure for these specimens were 69.61 and 64.14 (18.33 and 48.88; 14.90 and 18.01) kN, respectively. The data served to provide slip modulus information for the shear key/anchor notch detail (see Fig. 4) used in the beam tests described below. Figure 4. Notched shear key/anchor detail Beam Tests Twenty layered beams (see Fig. 5, A = 3.505 m, B = 1.067 m, C = 0.330 m, D = 0.66 m) configured as a typical portion of the width of the layered floor specimens were tested with notch B configuration with Hilti dowel connectors at 38.1 mm embedment. Figure 5. Configuration of layered beam test set-up Figure 6 is an example of the load deflection behavior observed. Figure 6. Example Load Deflection Results for a Composite Wood-Concrete Layered Beam Specimen (No. 4x4-8) Wood members were ten each of either eight 2 x 4s nailed together, or four 4 x 4s glued together. The concrete layer was 63.5 mm thick, consolidated by vibration, moist cured and a wire mesh was used for temperature and shrinkage reinforcement. At 28 days and at the end of beam testing (154 days) cylinder tests were conducted to determine concrete strengths (29 N/mm 2 and 33 N/mm 2 respectively). The cross-section reflected that for a 15 cm thick solid concrete slab, perhaps the lower 8-9 cm of depth would crack. Thus, the 2 x 4 replaced that depth of concrete. Before casting the concrete layer anchor pull-out tests were done as a glue quality control check. Floor Tests Two existing full-size wood floor/deck specimens were augmented with a concrete layer and the preferred notched shear key/anchor detail (notch B configuration) anchored with Hilti dowels (12 mm diameter, 50 mm penetration into the wood) and load tested (Etournaud, et al, 1998). One was rectangular (2 x 4s, 63.5 mm concrete, 3.581 m span, 2.476 m width). It was the subject of earlier papers (Gutkowski, et al, 1998). The other (Fig. 7) was skewed 43.6 degrees in plan (2 x 4s, 63.5 mm concrete, 5.639 m span, 2.705 m width). The former had deck-width long notches at midspan (with 2 anchors) and 444.5 mm from each end (with 3 anchors). The latter had deck-width long (parallel to the skew) notches at midspan (3 anchors) and 635 mm (2 anchors) and 1.702 m (3 anchors) from each end. Displacement was measured by potentiometers placed in three lines (at about the quarter points of the span) and equally spaced at 0.61 m across the deck width, with one potentiometer at each outer edge. Figure 7. Positions of the notches and the Hilti dowels for the skewed floor/deck specimen #2 In all the above tests, No. 1 and 2 Hem-fir dimension lumber was used. Complete results are in a M.Sc. thesis (Etournaud, 1998). NUMERICAL MODELING A well accepted computer model, FEAFLO developed at CSU, (Thompson, et al, 1977) for predicting the partially composite layered wood/wood floor systems, and FEABEA that is also a subroutine of FEAFLO that was adapted (Koike, et al, 1997) to the mixed material beam specimens and used to successfully predict the response of the layered beam specimens was applied to the study. Parameter studies were then conducted to examine how configuration changes affect the response. Those aspects are reported in a thesis (Koike, 1998). Recently the floor specimens were modeled with the AxisVM general- purpose finite element program for civil engineering developed by Inter-CAD Ltd. in Europe. Beam Tests FEABEA predictions were compared with measured load-deflection results (Koike, et al, 1998). Figure 8 shows an example comparison using one of Brown, et al's 2x4 specimens. FEABEA results for fully composite and fully non-composite extreme conditions clearly bound the test results. The difference relative to the mean slip value results is about 8%. FEABEA results using the low, mean and high measured slip modulus for that beam specimen's companion slip test specimens are plotted and closely bound the data (see Fig. 8). The outcome was similar for a selected 4x4 specimen, with a difference of about 3%. Three much higher differences (27%, 32% and 42%) were also obtained. However, the average % difference for the 2x4 (4x4) specimens was 11.62% (16.56%). Also, slip values are difficult to measure, highly variable and any given specimen's slip modulus may have been closer to either the low or the high end measured slip values, not the mean. Efficiency of the layered beams in developing composite action was determined using the definition (Pault, et al, 1977), Efficiency = [( D N- D I )/(D N- D C )](100) (1) where D N is the theoretical fully composite deflection (calculated by the transformed section), D C is the theoretical fully non-composite deflection (calculated as a layered beam without shear transfer), and D I is the measured deflection for incomplete composite action of the specimen. In layered wood-wood systems mechanical connectors are typically used. Considerable past research suggests that such connectors are very inadequate for the interlayer shear forces involved in the mixed material systems. This was confirmed in tests (Chen, et al, 1992), where only 10-20% efficiency was observed. The shear key/anchor interlayer connection has much higher efficiency. For the 2x4 (4x4) specimens the range of efficiencies was 57.4%-72.2% (54.9%-77.9%), the mean was 67.2% (67.2%) and the COV was 6.5% (9.2%). Figure 8. Load vs. Deflection at the midspan for 2x4 Beam Specimen 2 Floor Tests Finite Element Modeling: A pilot study was made to investigate the modeling issues of the floor specimens using the AxisVM general-purpose finite element software. As a result, three new finite elements dedicated to interface modeling were proposed and added to the AxisVMs finite element library as follows: a node-to-node (N-N), a line-to-line (L-L), and a surface-to-surface (S-S) interface element. The N-N element is formulated directly in terms of stiffness, and requires no integration of element matrices. The L-L element is a 6-node element with two 3-node sides. The S-S element is of a 16-node element with two 8-node sides. Both L-L and S-S interface elements use quadratic functions to interpolate displacements over the line or surface respectively. Therefore, for example, the L-L interface element could connect a 3- node isoparametric beam element (rib) to another similar element (layered semi-composite beam) or to the side of a 8/9 node isoparametric plate or shell element (bare wood member to the concrete slab), or the sides of any two 8/9-node surface elements. Numerical integration is applied over the line for L-L element, and over the surface for the S-S element. The element properties (slip modulus) are to be entered directly in terms of stiffness (units are kN/m or kN/rad for N-N, kN/m/m or kN/rad/m for L-L, and kN/m/m 2 for S-S).The modeling using these interface elements is an undergoing study, and the details are too extensive for this brief paper, and the complete results will be detailed reported in a separate paper. Basically the finite element model of the semi-composite floor specimen is build using 8/9 node isoparametric shell elements to model the concrete layer, and 3-node isoparametric rib elements to model the bare wood members that are eccentrically connected to the shell elements through L-L elements with no stiffness corresponding to interlayer relative displacements. The shell and the rib elements are based on Mindlins plate assumptions. At the location of the notches, N- N type interface elements have to be added with the corresponding stiffness (slip modulus) properties. If the rib-effect of the notches is taken into account, L-L elements should be used for notch slip modeling instead of N-N elements. Floor Specimen No. 1: For the rectangular bare deck vs. the composite deck, the order of the differences are similar to the skewed floor deck, with the displacement patterns clearly localized near the load. When loaded at centrally (Fig. 9), only about 60% of the members participated in resisting the load. The reduction obtained for the composite floor was from 0.465 in to 0.041 in, or 91.2%. When loaded at an edge midspan point (Fig. 10), only about 35% of the members participated in resisting the load. The reduction obtained for the composite floor was from 1.064 in to 0.129 in, or 87.9%. Figure 9. Displacement pattern, central load Figure 10. Displacement pattern, edge load Floor Specimen No. 2: For the bare deck the displacement patterns were clearly localized near the load. When loaded at position IV (central load) only about 65% of the inner members participated in resisting the load, with resistance diminishing away from the load point. In contrast, the entire composite deck resisted load in an almost uniform manner across the width, with only slight concavity in the middle area showing a markedly improved load distribution. For load position IV, the reduction in mid-width displacement was from 0.506 in. to 0.038 in, or 92.5%. When loaded at the position II only about 45% of the members participated in resisting the load. For the composite floor and load position II (edge load), the reduction in displacement under the loaded point was from 0.845 in. to 0.078 in, or 90.8%. Results of applying the finite element simulation to the composite specimen No. 2 are included in Figures 11 and 12. Figure 11 shows the measured and numerically predicted (on the AxisVM finite element model) displacement pattern across midspan for a central point load. Figure 12 shows similar comparison for displacement pattern across quarter span for an edge load at quarter point. The measured values are denoted as Experimental 1 and 2 in the Fig. 11 and Fig. 12. Figure 11. Displacement pattern, central load Figure 12. Displacement pattern, edge load If it is degenerated to the two extremes, a fully composite plate and a fully non-composite plate, the efficiency can be calculated according to Eq. 1. If applied to the center of the floor, the displacements were 0.064 in. for the non-composite, 0.026 in. for the fully composite, and 0.038 in. measured for the actual partially composite case, resulting in efficiency of 68.4%. Skewed Composite Floor #2; Load Position IV -577 lb; Central Transverse Deflected Shape -0.040 -0.035 -0.030 -0.025 -0.020 -0.015 -0.010 0 31 62 93 124 155 D i s p l a c e m e n t
[ i n ] Experimental-1 Experimental-2 Numerical Skewed Composite Floor #2; Load Position II -577 lb; South Transverse Deflected Shape -0.090 -0.080 -0.070 -0.060 -0.050 -0.040 -0.030 -0.020 -0.010 0.000 0 31 62 93 124 155 D i s p l a c e m e n t
[ i n ] Rectangular Wood Deck Load Position III - 1236 lb Central Transverse Deflected Shape -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0 10 20 30 40 50 60 70 80 90 Cross Section [in] D i s p l a c e m e n t [ i n ] Wood Section Composite Section Rectangular Wood Deck Load Position I - 1236 lb Central Transverse Deflected Shape -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0 0.2 0 10 20 30 40 50 60 70 80 90 Cross Section [in] D i s p l a c e m e n t
[ i n ] Wood Section Composite Section CONCLUSIONS The test program showed that with rational development, it is possible to achieve significant composite action in layered wood-concrete beam deck specimens using nominal dimension lumber and a notched shear key/anchor detail. The HILTI dowel employed is superior to alternatives examined in this study. One must use a simple tension pull-out test as quality control check on the setting of the glue for the anchor. Rigorous modeling by FEABEA was successful in predicting the response of the layered beam specimens. AxisVM with special finite elements was applicable to both beam and floor layered specimens. Durability under repetitive loads and extremes of temperature and humidity need to be examined, particularly for possible applications in extreme conditions. ACKNOWLEGEMENTS The authors graciously acknowledge the financial support for the project granted by the U.S. Department of Agriculture (Award No. 95-37103-2059). The U. S. Department of Transportation provided support for related studies for possible application in bridges. That funding was via the Mountain Plains Consortium (MPC) which is federally sponsored through the University Transportation Centers Program. CSU also provided cost share funds. The work described was conducted at CSU, but was augmented by laboratory tests done on other deck specimens (previously and concurrently) at the Institute of Wood Construction at the Swiss Federal Institute of Technology. Although that cooperative work is not included in this paper, that invaluable cooperation is greatly appreciated. Inter-CAD Ltd. (Budapest, Hungary) has provided the AxisVM software through its Academic Institution Grant program, and has supported the second author in this research. REFERENCES Natterer, J., Hamm, J. and Favre, P. 1996, Composite Wood-Concrete Floors for Multi-Story Buildings, Proceedings of the International Wood Engineering Conference - Vol. 3, New Orleans, La., pp. 3.431-3.435, Omnipress, Madison, WI. Thompson, W. 1997, Shear Tests of Wood-Concrete Composites, M.S. Independent Study Report, Department of Civil Engineering, Colorado State University, Ft. Collins, CO. Brown, K. T., Gutkowski, R. M. and Peterson, M. L. 1998, Testing of a shear key/anchor in layered wood/concretebBeams, Structural Research Report No. 76, Civil Engineering Department, Colorado State University, Ft. Collins, CO Chen, T.-M., Gutkowski, R. M, and Pellicane, P. J. 1992, Tests andd analysis of mixed wood-concrete wood beams, Structural Research Report No. 69, Civil Engineering Department, Colorado State University, Ft. Collins, CO. Pault, J. D. and Gutkowski, R. M. 1977, Composite Action in Glulam Timber Bridge Systems, Structural Research Report No.17B, Civil Engineering Department, Colorado State University, Ft. Collins, CO. Etournaud, P.A., Gutkowski, R. M., Peterson, M. L. and Criswell, M. E. 1998, Loads tests of composite wood-concrete deckings under point loads, Structural Research Report No. 81, Civil Engineering Department, Colorado State University, Ft. Collins, CO. Koike, E., Gutkowski, R. M., Peterson, M. L. and Criswell, M. E. 1998, Analysis of composite wood-concrete layered beams, Structural Research Report No. 79, Civil Engineering Department, Colorado State University, Ft. Collins, CO. Gutkowski, R. M. and Chen, T.-M. 1996, Tests and Analysis of Mixed Concrete-Wood Beams, Proceedings of the International Wood Engineering Conference, New Orleans, La., Omnipress, Madison, WI, pp. 3.436-3.442. Pellicane, P. J., and Bodig, J. 1984, Comparison of nailed joint test methods, Journal of Testing and Evaluation, Vol. 12, No. 5. Pellicane, P. J., Potter, and Mielke, P. W. 1993, Permutation procedures as a tool in wood related applications, Wood Science and technology, Vol. 23, pp. 193-204. Tremblay, G. A., Goodman, J. R.,Thompson, E. G., Criswell, M. E. and Bodig, J. 1974, Nonlinear Analysis of Layered T-Beams with Interlayer Slip, Structural Research Report No. 12, Civil Engineering Department, Colorado State University, Ft. Collins, CO. Thompson, E. G., Vanderbilt, M. D. and Goodman, J. R. 1977, FEAFLO: A Computer Program for the Analysis of Layered Wood Systems, Computers and Structures, 7(2), pp. 237-248. Thompson, W. 1997, Shear tests of wood-concrete composites, M.S. Independent Study Report, Department of Civil Engineering, Colorado State University. Wegener, G. 1997, Sustainable Economy of Forestry and Value Added Utilization of Forests. Restoration of Forests - Environmental Challenges in Central and Eastern Europe (edited by R. M. Gutkowski and T. Winnicki), NATO ASI Series, 2. Environment - Vol. 30. Kluwer Academic Publishers, Dordrecht, Netherlands.