Development of design capacities for residential deck ledger connections.
In recent years there have been a significant number of reports of deck, balcony and porch failures, some of which have resulted in injury and loss of life. One of the primary causes for these collapses is a deficient connection between the deck ledger and the house band joist. Additionally, deck ledger connection problems are frequently exacerbated by lack of structural redundancy--specifically, when the ledger-to-house connection fails, there is the potential for the deck to collapse catastrophically. Testing was performed at two universities in an effort to develop methods for more effectively anchoring deck ledgers to house band joists using commonly available lag screws and bolts. The focus of this article is the residential deck ledger to band joist connection, and includes descriptions of the common problems with connections between deck ledgers and house band joists, how these connections are addressed in building codes, testing aimed at validating effective methods for making those connections, and recommended on-center (o.c.) fastener spacings and construction methods.
Over the past decade there have been a significant number of reports of deck, balcony and porch failures, sometimes resulting in injury and loss of life. Primary causes for the collapses are: (1) deficient connections between the deck ledger and the house band joist and related decay, and (2) deficient guardrail systems and related decay and corrosion of fasteners. The focus of this article is the residential deck ledger to band joist connection, including recommended on-center (o.c.) connection spacings.
The issues surrounding deck ledger failures include a lack of specific, code-prescribed details, overly conservative results when utilizing National Design Specification for Wood Construction (NDS) (AF&PA 2005) analysis techniques for these connections, and inadequately constructed connections due to improper fastening techniques or methods that result in deterioration of the wood members comprising the connections. While current model building codes such as the 2003 International Residential Code[R] (IRC[R]) (ICC 2003) provide some guidance for deck ledger fasteners, information on fastener sizes and spacing are lacking. Use of NDS equations for ledger connections, particularly because the use of sheathing between the band joist and ledger creates a space between the main and side members of the connection, requires unrealistic spacing that would be considered far too conservative for most contractors. Because of the connection of interior framing to exterior decking materials, it can be a challenge to develop construction methods that will not result in degradation of wood on either side of the connection and that will create an effective, durable connection. Deck ledger connection problems are frequently exacerbated by lack of structural redundancy: that is, when the ledger-to-house connection fails, the deck has the potential of collapsing catastrophically. A sample of deck and balcony failures reported in the media since August 2004 is shown in Table 1. It should be noted that these would not include deck failures that were not reported in the media. It seems clear that more attention should be given to the design of these connections.
A major problem concerning the construction of decks is the lack of information currently available in model building codes. The 2003 International Residential Code[R] (IRC[R]) specifies a live load of 40 pounds per [ft.sup.2] (psf) for decks and 60 psf for balconies. The only IRC coverage of deck ledger connections is given in Section R502.2.1 which states: "Where supported by attachment to an exterior wall, decks shall be positively anchored to the primary structure and designed for both vertical and lateral loads as applicable. Such attachment shall not be accomplished by the use of toenails or nails subject to withdrawal." According to this code section, nails alone cannot to be used to connect a ledger to a band joist. Thus, lag screws, bolts or other "positive connection" devices are required. Additionally, the IRC does not provide any recommendations for allowable lateral loads to be used for the design of exterior decks.
The technical reference for the engineering design of connections in wood construction is the National Design Specification for Wood Construction (NDS) (AF&PA 2005); however, commonly accepted means of connecting deck ledgers to band joists fall outside the scope of the NDS. Specifically, NDS requirements for direct contact between members and minimum penetration depth are not met when using the lag screw tables because there is usually sheathing (and sometimes stacked washers to facilitate drainage) between the ledger and band joist, and the band joist is only 1-1/2 inches thick. Using a 1/2-inch diameter lag screw results in a penetration of 3 diameters (3D), whereas the NDS requires a minimum penetration of 4 diameters in the main member. Even if the band joists were 2 inches thick, the 1/2-inch lag screw values in the NDS 2005 tables must be reduced by one-half because the tabulated values are based on a screw that penetrates the main member (a house band joist) eight diameters (SD), or 4 inches.
In cases where "methods of construction that are not capable of being designed by approved engineering analysis or that do not comply with the applicable material design standards," the International Building Code[R] (IBC[R]) allows for preconstruction load testing to derive design values. Loferski et al. (2004a, 2004b) conducted ledger tests of four configurations of preservative pressure treated (PPT) southern pine ledgers attached to Spruce-pine-fir (SPF) house band joists, and developed design tables based on simulated deck-ledger house-band connections. Southern pine is a relatively dense construction species grouping that is not available across the country and hence tests were needed on PPT Hem-fir ledger material, which is commonly used in the western United States. Similarly, many parts of the country use wood composite band joists and tests were needed to evaluate the impact of this material on connection performance.
The objectives of this study were to determine the required spacing for 1/2-inch lag screws and bolts when used to connect a PPT Hem-fir deck ledger to SPF and composite house band joists, typically used in conjunction with I-joist framing. Three connection details were tested as illustrated in Figure 1:
[FIGURE 1 OMITTED]
--1/2-inch diameter lag screw with 15/32-inch wall sheathing between the ledger and band joist,
--1/2-inch diameter bolt with 15/32-inch wall sheathing between the ledger and band joist, and
--1/2-inch diameter bolt with 15/32-inch wall sheathing and a 1/2-inch stack of washers (for drainage) between the ledger and wall sheathing.
[FIGURE 2 OMITTED]
Recommended design loads, corresponding deflections, and allowable on-center fastener spacing were developed. Results from this study, along with those of Loferski et al. (2004a; 2004b), will provide the basis for an International Residential Code[R] (IRC[R]) proposal to cover residential deck ledger design.
Materials for ledger connection testing
Four types of band joist materials were investigated: 2 by 10 SPF lumber and 1 by 9-1/2 Douglas-fir laminated veneer lumber (LVL) rimboard, 1-5/16 by 9-1/2 Douglas-fir LVL rimboard, and 1-1/8 by 9-1/2 OSB (OSB) rimboard. Materials used for band joists and OSB sheathing were stored in a controlled environment where the temperature and relative humidity were maintained to achieve a targeted equilibrium MC of 12 percent. MC readings were obtained for all wood and wood-based materials prior to testing using a handheld moisture meter. SPF has a relatively low SG (G) of 0.42, so denser species groupings such as Hem-fir, Douglas-fir-Larch or southern pine can be conservatively substituted in actual construction. In addition, 1-inch rimboard is the thinnest band joist product currently sold. Thicker composite rimboard products with equivalent SGs of 0.50 or greater can be conservatively substituted for the LVL band joist material we tested.
The material used for the deck ledger was incised PPT Hem-fir treated to a retention level of 0.40 pcf of ACQ (suitable for ground contact). Hem-fir is a common species grouping for PPT lumber sold in the western United States. Other lumber species with SGs greater than Hem-fir (G = 0.43) can be conservatively substituted, provided that they are adequately treated to resist decay. The ledger materials were tested wet (near fiber saturation point) to represent a worst-case condition in the field.
All lag screws, bolts, washers and nuts used for testing were 1/2 inch in diameter, hot-dip-galvanized and manufactured from A307 steel (ASTM 2005). Fasteners were purchased from local suppliers and had "307A" embossed into the heads. Lag screws were full body diameter, 4 inches long and had a threaded portion 3 inches long with a root diameter of approximately 0.368 inches and 6 threads per inch. The bolts were 4-1/2 and 5 inches long (depending on whether they were used with the 1/2-inch drainage space or not) with a 1-1/2 inch threaded portion and 13 threads per inch. Since the purpose of the testing program was to determine design values based on ultimate load tests of the simulated deck ledger connections, it was not necessary to determine the bending yield stress, [F.sub.yb], for the locally purchased A307 lag screws and bolts. [F.sub.yb] was not required in the analysis of the test data. The test approach to obtain design values was a deliberate departure from the yield equations (NDS Section 11.3.1) that rely on [F.sub.yb] because it was determined that when the yield equations were applied to various deck ledger cases, the results were overly conservative (Anderson et al. 2003). Nuts and flat washers were appropriately sized for the bolts and only a single washer was used on each side of the connection.
Table 2 summarizes each of the test configurations. Fifteen specimens were tested for each configuration, except configurations HF-4, HF-5, HF-6, HF-LVL-4, HF-LVL-5, and HF-LVL-6, which had a sample size of five since they were used as a means of observing trends in performance based on different band joist materials. Loading was applied through a simulated deck portion, which consisted of two joists that attached to the ledgers with joist hangers and a backing member that held the joists in place at the ends opposite from the ledger.
In conformance with NDS--2005 (AF&PA 2005) guidelines for predrilled holes when installing lag screws, 3/8-inch diameter lead holes were drilled in the band joists and 1/2-inch diameter clearance holes were drilled through the deck ledgers and OSB sheathing prior to assembling the specimens. For the bolted specimens, 9/16-inch diameter clearance holes were drilled through the band joist, OSB and deck ledger.
Methods for ledger connection testing
Because there are no specific testing standards that address deck ledger to house band joist connections, methods were developed aimed at simulating the fastening and loading conditions that would normally exist for these connections. All ledger testing was conducted utilizing methods described by Loferski et al. (2004a, 2004b) and involved applying loads to the simulated deck, sheathing and band joist assemblies discussed in the previous section up to the point of failure of the connection between the ledger and band joist. While there is another method for testing rimboards, specifically APA Standard PRR-401 Performance Standard for APA EWS Rim Boards (APA 2002), the methods prescribed by Loferski et al. (2004a, 2004b) were judged to provide more conservative results. PRR-401 specifies that the siding and the band joist bear directly on the reaction support, that dry SPF lumber can be used for the ledger, and that a washer be used with the lag bolts. Based on the methods utilized by Loferski et al. (2004a), testing was conducted such that the band joist was the only portion of the ledger assembly that had bearing on the reaction support. In the event that the OSB were to bear on the reaction support there would effectively be no gap between the rimboard and ledger, and a higher connection capacity will result as evidenced by Virginia Tech researchers in preliminary tests. Because decks are exterior structures requiring treated lumber for construction and are subject to wet conditions, ledger material was incised PPT lumber and was tested in the wet condition, close to the fiber saturation point. Wood is weakest at MCs at and above the fiber saturation point: therefore this test condition was considered the most conservative for design purposes and was used for our tests.
Tests were performed by applying a load at the center of both joist members using displacement control at a constant rate of 0.5 inches per minute until ultimate load was achieved. Loads were applied with a fixture attached to a hydraulic actuator that was servo-controlled using an MTS 407 controller. Band joist members were supported on wood blocks that rested on steel supports to simulate direct bearing on a foundation sill plate, and the backing members were similarly supported. Deck joists were attached to the deck ledger using commonly found joist hangers and backing members using decking screws driven through the backing member into the end of the deck joists. In order to ensure failure would occur in the bolt or lag screw, oversized fasteners were used to attach the joist hangers. All lag screws and bolts were installed to test specimens with a wrench and tightened just until they were snug, and no washers were used with the lag screws because the NDS (AF&PA 2005) does not require them. The testing apparatus is shown in Figures 2 and 3. Throughout all of the test configurations, care was taken to make sure that the only part of the specimen at the ledger end to bear on the reaction support would be the rim board, and not any of the other components. While for each test specimen a different ledger, band joist, OSB piece, fastener, and spacers (where applicable) were used, the joists, joist hangers and backing members were reused several times before being replaced.
[FIGURES 2-3 OMITTED]
Throughout testing, data on the applied loads, movement of the hydraulic actuator, and displacement of the ledger with respect to the band joist were continuously obtained at 0.1 second intervals and were acquired using LabVIEW Version 8 software. A load cell having a capacity of 25,000 [pounds.sub.f] was utilized to obtain load measurements and a string potentiometer mounted to the rim board and the ledger monitored the displacement of the ledger with respect to the house band joist or rim board. These load and deflection measurements were used for generating load-displacement plots and for executing the calculations discussed in the following section. Documentation of failures was also included to determine how these connections using different materials behaved under load.
Results and discussion
Failures included fastener heads or nuts pulling through the ledger, the threaded portion of the lag screws withdrawing from the band joist, and splitting of the ledger in the region of the fastener. Although the ends of the bolts with the nuts and washers did not pull through the band joists, there was usually a significant amount of crushing and rotation beneath the washer, which often embedded itself in the band joist. Additionally, bolts and lag screws deformed only minimally, and connection deflections were primarily a result of the fasteners crushing and rotating through wood members or, in the case of the lag screws, threads withdrawing. The test specimen assemblies displayed significant ductility, with most displacing more than 1-1/2 inches before reaching their ultimate loads.
Observations of connection failures indicated that some slight differences among the various materials tested existed. Thread withdrawal was the failure mode for specimens using lag screws and SPF band joists, while specimens utilizing lag screws and an OSB band joist failed as a result of a combination of thread withdrawal and splitting of the deck ledger. Connections having LVL band joists and using lag screws did not exhibit thread withdrawal, but rather failed as the screw heads pulled through the ledger. Specimens fabricated with bolts, both with and without the 1/2 inch spacers, failed as a result of splitting of the deck ledgers for connections that used SPF and OSB band joists. The bolted specimens having LVL band joists also failed as a result of ledger splitting, but in many cases the bolt heads pulled partially through the ledgers during the tests.
Table 2 summarizes the test results. Design loads were computed by dividing the average ultimate loads by a load duration factor of 1.6 and then by a safety factor of 3. The load duration factor converts the short-term strength to a "normal" duration as used in the 2005 NDS (AF&PA 2005). A safety factor of 2.5 has precedence in the IBC (ICC 2003) but a more conservative factor was used because of the lack of structural redundancy in most deck-to-house connections.
Fastener spacings given in Table 3 were derived by assuming deck live and dead loads of 40 and 10 psf, respectively. Data on PPT southern pine ledgers (Loferski et al. 2004a, 2004b) and the data reported herein were combined to develop fastener spacings given in Table 3. The data were combined for practical use because the results were relatively insensitive to the range of ledger and band joist materials tested. The limiting combination of ledger and band joist material was used to develop the fastener spacings in Table 3, and the results were rounded up to provide more conservative results.
In order to properly utilize Table 3, it is important that the lag screws or bolts are installed according to 2005 NDS requirements. Figure 4 shows the recommended method for installing these fasteners. Lead holes for the lag screws should be equal to the root diameter (or slightly less) of the threaded portion and the clearance holes should be 1/2-inch diameter. For a particular box of 1/2-inch lag screws, a "test installation" into the house band joist is recommended before drilling the lead holes so as to ensure that the lead holes are neither too small nor too large. When installing bolts, the clearance hole should be 9/16-inch or at least 17/32-inch diameter. All fasteners should be hot-dip galvanized or stainless steel, as determined by the deck designer and approved by the building official.
[FIGURE 4 OMITTED]
Summary and conclusions
Testing was conducted for three common residential deck ledger constructions using relatively low density incised PPT Hem-fir lumber for the ledgers, and SPF and wood composite materials for the house band joist, and were fastened using lag screws or bolts. Tests were also conducted on the bolted configurations having a 1/2-inch space between the house sheathing and the ledger, which is sometimes used to create a space so that debris does not collect between these members, which could result in premature decay. Test results indicated that, in general, bolted connections were nearly twice as strong as those using lag screws, and that the gap between the ledger and sheathing reduced the strength of the bolted connections by approximately 25 percent. While the connections tested with LVL band joists had similar design strengths and fastener spacings, it was noted that the displacement between the band joist and deck ledger was significantly less at design loads than for connections tested with solid wood band joists.
The tables presented here provide designers with calculated fastener spacing when using the configurations tested. The resulting fastener spacings given in Table 3 are limited to the following conditions:
--Deck live load of 40 psf and dead load of 10 psf (note that other gravity loads may control residential deck ledger design such as snow and concentrated loads such as planters; as well as lateral loads from wind or seismic);
--Band joist lumber with G greater than or equal to 0.42 (includes SPF, Hem-fir, Douglas-fir-Larch and southern pine);
--Composite rimboard with thickness 1 inch or thicker and equivalent SG (G) greater than or equal to 0.50;
--PPT deck ledger lumber with G greater than or equal to 0.43 (includes Hem-fir, Douglas-fir-Larch and southern pine);
--Deck ledger can be incised and wet;
--No decay present;
--No fastener corrosion.
American Forest and Pap. Assoc. (AF&PA). 2005. ANSI/AF&PA NDS-2005 National Design Specification for Wood Construction. AFandPA. Washington, D.C.
American Plywood Assoc. 2002. Standard PRR-401 Performance Standard for APA EWS Rim Boards. APA. Tacoma, Washington.
Anderson, C.A., F.E. Woeste, and J.R. Loferski. 2003. Attaching deck ledgers. J. of Light Construction 21(11):81-87.
American Soc. for Testing and Materials (ASTM). 2005. Standard specification for carbon steel bolts and studs, 60,000 PSI tensile strength. ASTM Standard A 307-04. ASTM, West Conshohocken, Pennsylvania.
Inter. Code Council, Inc. (ICC). 2003. Inter. Residential Code for One- and Two-Family Dwellings. Inter. Code Council, Inc., Falls Church, Virginia.
Loferski, J.R., F.E. Woeste, and M.A. Billings. 2004a. Deek ledger connection design. Professional Deck Builder 3(3):56, 58, 60, 62, 64, 66-67.
Loferski, J., F. Woeste, R. Caudill, T. Platt, and Q. Smith. 2004b. Load-tested deck ledger connections. J. of Light Construction 22(6):71-78.
David M. Carradine * Donald A. Bender * Frank E. Woeste * Joseph R. Loferski *
The authors are, respectively, Research Engineer, Weyerhaeuser Professor of Civil Engineering/Director, Wood Materials and Engineering Lab., Washington State Univ., Pullman, Washington (email@example.com; firstname.lastname@example.org); Professor Emeritus and Professor, Dept. of Wood Sci. and Forest Products, Virginia Tech Univ., Blacksburg, Virginia (email@example.com; firstname.lastname@example.org). This paper was received for publication in February 2006. Article No. 10164.
* Forest Products Society Member.
[c] Forest Products Society 2007. Forest Prod. J. 57(3):29-33.
Table 1.--Deck collapses reported in the media from August 2004 to October 2005. Resulting injuries or Location Date deaths Loveland, Ohio October 2005 13 injured Virginia Beach, Va. October 2005 28 injured Seneca, S.C. September 2005 7 injured Elm Grove, Wis. September 2005 9 injured Minneapolis, Minn. September 2005 3 injured Portland, Ore. August 2005 10 injured Sherwood, Alaska August 2005 12 injured Troy, Ill. July 2005 7 injured Fort Kent, Maine June 2005 5 injured San Francisco, Calif. June 2005 3 injured Allentown, Pa. June 2005 2 injured Napa, Calif. April 2005 11 injured Durham, N.C. March, 2005 3 injured Columbus, Ohio November, 2004 1 death Pierce County, Wash. October, 2004 1 death, 7 injured Wilmington, N.C. October, 2004 8 injured Milford, Conn. September, 2004 9 injured St. Louis, Mo. August, 2004 2 injured Montana August, 2004 80 injured Table 2.--Descriptions of ledger-to-house band joist connection test specimens and summary of average testing results. Configuration (a) Band joist Fastener HF-1 2 by 10 SPF 1/2-inch lag screw HF-2 2 by 10 SPF 1/2-inch bolt HF-3 2 by 10 SPF 1/2-inch bolt (c) HF-4 OSB (b) 1/2-inch lag screw HF-5 OSB (b) 1/2-inch bolt HF-6 OSB (b) 1/2-inch bolt (c) HF-LVL-1 1 by 9-1/2 LVL 1/2-inch lag screw HF-LVL-2 I by 9-1/2 LVL 1/2-inch bolt HF-LVL-3 1 by 9-1/2 LVL 1/2-inch bolt (c) HF-LVL-4 1 by 9-1/2 LVL 1/2-inch lag screw HF-LVL-5 1 by 9-1/2 LVL 1/2-inch bolt HF-LVL-6 1 by 9-1/2 LVL 1/2-inch bolt (c) Ultimate Load Coefficient Configuration (a) Average of variation ([pounds.sub.f]) (%) HF-1 2,170 17.4 HF-2 4,260 17.7 HF-3 3,230 19.0 HF-4 1,880 13.4 HF-5 3,670 13.6 HF-6 2,970 7.7 HF-LVL-1 2,250 12.8 HF-LVL-2 4,110 15.5 HF-LVL-3 2,950 15.5 HF-LVL-4 2,730 4.4 HF-LVL-5 4,480 9.9 HF-LVL-6 3,600 11.0 Average displacement at Configuration (a) Design load design load ([pounds.sub.f]) (d) (in) HF-1 451 0.19 HF-2 887 0.21 HF-3 673 0.23 HF-4 391 0.07 HF-5 764 0.13 HF-6 618 0.15 HF-LVL-1 468 0.05 HF-LVL-2 855 0.06 HF-LVL-3 614 0.09 HF-LVL-4 568 0.05 HF-LVL-5 938 0.02 HF-LVL-6 750 0.09 (a) All ledger material was incised, PPT nominal 2 by 8 lumber and all configurations had 15/32-inch OSB between the band joist and ledger. (b) OSB band joists were 1-1/8 inches by 9-1/2 inches OSB Rimboard material. (c) These configurations included a l/2 inch stack of washers between the OSB sheathing and the deck ledger. (d) Design loads were calculated by dividing the average ultimate load by 3 (safety factor) and by 1.6 (duration of load factor). Table 3.--Calculated on-center spacing in inches for PPT Hem-fir or southern pine deck ledger attached to SPF or wood-based composite band joist residential deck joist spans loaded with 40 psf live load and 10 psf dead load. Tabulated spacing values apply only to the case of deck joists perpendicular to ledger. Residential deck joist span Connection detail 6 feet 8 feet 10 feet 12 feet 1/2-in lag screw with 30 23 18 15 15/32-in sheathing 1/2-in bolt with 15/32-in 36 (a) 36 (a) 34 29 sheathing 1/2-in bolt with 15/32-in 36 (a) 36 (a) 29 24 sheathing and 1/2-in stacked washers Residential deck joist span Connection detail 14 feet 16 feet 18 feet 1/2-in lag screw with 13 11 10 15/32-in sheathing 1/2-in bolt with 15/32-in 24 21 19 sheathing 1/2-in bolt with 15/32-in 21 18 16 sheathing and 1/2-in stacked washers (a) These spacings were limited by a consideration of the bending strength of a 2 by 8 (minimum) ledger between the bolts or lag screws.
|Printer friendly Cite/link Email Feedback|
|Author:||Carradine, David M.; Bender, Donald A.; Woeste, Frank E.; Loferski, Joseph R.|
|Publication:||Forest Products Journal|
|Date:||Mar 1, 2007|
|Previous Article:||Measuring the impact of wood species on consumer preferences for wooden furniture by means of the Analytic Hierarchy Process.|
|Next Article:||Use of small thinning logs in a round-wood trussed bridge.|