The effect of in-situ conditions on nail withdrawal capacities Edward Sutt 1, Timothy Reinhold 2, and David Rosowsky 2 ABSTRACT Extensive fastener testing has been conducted in recent years to determine the effects of environmental conditions such as heat, humidity, aging and moisture content on nail withdrawal capacities. However, few attempts have been made to correlate experimental results with actual in-situ conditions. A 9.1 kg (20 lb) portable fastener extractor with a peak load data collection system has been built to measure actual withdrawal capacities of fasteners in their installed locations. A limited amount of field test data on withdrawal capacity is compared to the expected design capacities based on the presumed in-situ conditions. Implications for larger-scale field studies and the design of these connections are also discussed. INTRODUCTION A common wood frame house has roof sheathing installed on manufactured roof trusses or a site built dimension lumber roof framing system with lumber having moisture content 19 percent or less. The roof sheathing, typically plywood, oriented strand board (OSB) or wood planks, is installed to the roof structural members using wire nails. During the life of a structure, the moisture content in the wood decreases to its equilibrium moisture content due to the ambient environment as well as the heat and airflow conditions in the attic. These changing conditions may affect the withdrawal capacity of the fasteners, which is a concern in high wind areas where roof sheathing loss is not uncommon in severe storms. While there has been a significant amount of withdrawal test data developed on single fasteners as well as complete 1.2 m x 2.4 m (4 ft x 8 ft) roof panels, there is some question whether the design values based on laboratory tests remain conservative after the fastener is installed. Some laboratory research has shown a reduction in withdrawal strength due to age, heat, humidity and moisture content; however, it is unclear whether these laboratory created conditions are similar to what occurs in an attic space during the lifetime of the structure. The portable fastener extractor allows the fastener to be tested in withdrawal in its installed location. Results from field tests can provide some indication whether or not the design capacities are met after installation. A test house, located in Anderson, South Carolina, (USA), offered an opportunity to test nail fasteners in withdrawal in two lumber species; Southern yellow pine (SYP) and Spruce-pine-fir (SPF). DESCRIPTION OF THE FASTENER EXTRACTOR The motivation for building the fastener extractor was to be able to withdraw fasteners in any (field and/or structural) location and record an ultimate capacity. The design parameters of the instrument were selected to achieve complete portability with an overall weight of less than 11.3 kg (25 lb) and have self-contained load cell excitation and peak hold capabilities. In addition, the extractor had to be sufficiently rugged to survive field use and develop enough force to withdraw a 5.1 cm (2 in) #8 wood screw from SYP. The resulting device is shown in Figure 1. The structure of the extractor was developed using manufactured extruded aluminum shapes. These unique extruded shapes form the guides for two linear bearings that create a low friction slide on which the extractor mechanism travels. An aluminum block spanning between the two linear bearings provides the mounting place for a 907.2 kg (2000 lb) OmegaDyne S beam load cell, which was selected because of its high accuracy and repeatability. Threaded into the load cell is the "hook" which slides under the head of the fastener. This "hook" swivels to accommodate possible misalignment between the load cell and the fastener. The "hook" was designed to provide adequate engagement of the head of common framing nails ranging from 6 to 16 penny, as well as similarly sized wood screws. The load cell is the 1 Graduate Research Assistant, Dept. of Civil Engineering, Clemson University, Clemson, SC, 29634-0911. 2 Associate Professor, Department of Civil Engineering, Clemson University, Clemson, SC, 29634-0911.
last element in the chain between the extractor mechanism and the hook. Thus, the load measured does not include any affect of friction in the slide bearings. Figure 1: The Fastener Extractor The load is applied to the fastener manually through the two removable 38.1 cm (15 in) handles much like a lever corkscrew. As the handles swing down, the load cell is lifted up, withdrawing the nail. This motion provides a maximum withdrawal distance of about 4.4 cm (1.75 in). Most of the nails used in light frame wood construction are longer than this, but the maximum withdrawal resistance of the nail is generally reached during the first 13 mm (0.5 in) of withdrawal. The load cell is connected into an Omega DP-41 load cell amplifier. The DP-41 was factory modified to be DC powered and provide a 0-10 volt DC excitation for the load cell as well as a digital readout to read force applied to the fastener. The DP-41 captures data at 13 hertz and has peak hold and tare (zeroing) capabilities. The power for the DP-41 is provided by a 12-volt nickel cadmium battery which is regulated to be certain that a constant 10 volt DC is supplied. The charger for the battery was modified to mount to the extractor to store and charge the battery as well as allow the device to be run off of 110 volt AC. The fastener extractor weighs 9.1 kg (20 lb), stands 45.1 cm (17.75 in) high and measures 14.6 cm (5.75 in) by 28.6 cm (11.25 in) at its largest dimensions. With the handles installed, the width is 107 cm (42 in). One of the concerns with this device was that the rate of withdrawal would be variable with the operator of the extractor. However, recent studies have shown that, within the practical range of loading rates (2.5 mm/min (0.1 in/min) to near instantaneous), the rate of loading does not affect the ultimate withdrawal capacities of 8d common nails or #8 screws [1]. Thus, it was assumed the fastener extractor results are similar to those obtained by ASTM D 1761, the single fastener test standard, which suggests a loading rate of 2.5 mm/min (0.1in/min). BACKGROUND OF NAIL WITHDRAWAL CAPACITY Nail Withdrawal Equation The 1997 National Design Specification for wood construction (NDS) bases the design withdrawal capacity of nails in the side grain of wood by the ultimate withdrawal equation of W=47.6 G 2.5 D L, where W is the average ultimate withdrawal
strength in Newton's (W=6900 G 2.5 D L, where W is in pounds) 3 [2]. G is defined as being the specific gravity of the wood member at oven dry weight and volume; L is defined as the length of the nail in mm (in) penetrated into the wood and D the diameter of the nail in mm (in). The withdrawal design load used by the 1997 NDS applies a capacity increase of 20 percent to the ultimate withdrawal equation. This additional 20 percent capacity is based on a ten percent increase for the change from normal to permanent loading and a ten percent increase for experience [2]. A factor of safety of 6, as suggested by the Wood Handbook [3], is then applied, resulting in a design equation for allowable withdrawal capacity of W= 9.5 G 2.5 D L (W=1380 G 2.5 D L) [4]. However, the net factor of safety is 5. Design Capacities of 8d Sinker Nails in Withdrawal The nails used in the attachment of roof sheathing on the study house were 8d sinker nails with a diameter of 2.9 mm (0.113 in) and a length of 6 cm (2.375 in). In order to determine the design load for this application, the designer would need to consider the following adjustment factors to the design withdrawal equation: load duration, wet service and temperature. The adjustment factors would then be applied to W calculated by the NDS design equation to determine the design load of the fastener. The load duration factors are chosen based on the expected loads. Because the nail attaching the roof sheathing to the roof structural member has no permanent dead load applied to the fastener and is only placed in withdrawal under wind uplift loading, a load factor of 1.6 can be used to determine the withdrawal capacity of the nail. However, for comparison purposes, the withdrawal capacities will be computed with load duration factors of 1.0 and 1.6. The use of the load duration factor of 1.6 increases the allowable capacity of the nail in withdrawal, which reduces the net factor of safety applied in the predicted ultimate withdrawal equation, barring any other adjustments, to 3.13 4. Since it can be assumed that the framing lumber will have a moisture content of less than 19% during installation and less than that in service, there is no required reduction for wet service. Reductions due to temperature are required for sustained exposure to elevated temperatures above 37.8 C (100 F). Heyer performed a study, which included a house in Athens, GA (within 113 km (70 miles) of the test house), on temperatures of wood parts of houses. The results indicated that in the months of July and August the temperatures were over 37.8 C (100 F) at the junction of the roof rafter and roof sheathing about 29 percent of the time [5]. This percentage of exceeding 37.8 C (100 F) cannot be considered a sustained condition; therefore no temperature reduction would need to be applied to the withdrawal equation. The roof framing on the test house was constructed using Southern yellow pine (SYP) and Spruce-pine-fir (SPF). These wood species have specific gravities (G) of 0.55 and 0.42, respectively. Table 1 provides a comparison of the design values with the two different load duration factors. Table 1: Design Withdrawal Capacities of 8d Sinker Nail Wood Species Load Duration, C D =1.0 Load Duration, C D =1.6 Southern yellow pine (SYP) 6.1 N/mm (35 lb/in) 9.8 N/mm (56 lb/in) Spruce -pine-fir (SPF) 3.2 N/mm (18 lb/in) 5.1 N/mm (29 lb/in) Variations in Withdrawal Capacity The withdrawal capacity of nails is highly variable due to changes in moisture content, nail point geometry, coating applied to the nail shank, location of the shank with respect to the grain and the length of time the nail is installed. As illustrated above, not all of these factors are considered in computing the design capacity of the nail. This paper focuses primarily on the effects of moisture content changes, length of time the nail has been installed and possible heat effects. Variations in nail withdrawal capacity have been studied and documented since the initial development of the nail withdrawal equation. Research has established that the average capacity of a nail driven into seasoned or dry wood is the same as the average capacity of the same nail driven into green wood when both nails are withdrawn immediately. 3 This withdrawal equation is the same as the withdrawal equation presented in the Wood Handbook except that G is in terms of oven dry weight and 12% moisture resulting in a factor of 54.1 (7850) instead of 47.6 (6900). 4 Where (5.0/1.6)=3.13.
However, if the nail is driven into green wood and the wood seasons or is allowed to dry before the nail is withdrawn, 75% of the ultimate withdrawal strength may be lost [6]. A reduction in the in-situ moisture content of the lumber is common. Typical framing lumber is kiln or surface dried to 19% moisture before shipping from the mill. Air drying from shipping and intermediate storage may further reduce the moisture content, depending on the relative humidity conditions. Attic conditions with elevated temperatures can bring the moisture content of the framing lumber to 6 percent [2]. Studies have also been conducted to investigate the effect of time lapse after installation on the withdrawal capacities of nails. However, the results from these studies are conflicting as to whether the withdrawal capacity increases, decreases or remains constant [7,8,9]. IN SITU TESTING PERFORMED The test house had a "T" shape plan. The plan of the original house was rectangular with a length about 4 times its width. The original rectangular shaped house was constructed at some point between 1972 and 1983 and its axis roughly paralleled a Northwest-Southwest Axis. An addition extending to the Northeast was added in 1996 to change the shape of the house to a "T". The house had a new asphalt roof covering installed about 9 months prior to the testing. The testing was performed in August 1999. The house was situated on a sloping lot, which allowed a basement living area in the Northwest corner of the home. A fire started in this area, and burned the basement and first floor of the home. The direct fire damage was concentrated in a localized area around its origin. The testing areas were selected to have been least affected by the heat of the fire (i.e. the farthest away). The members from which nails were withdrawn were not charred and at worst showed evidence of smoke damage. The original roof system was framed using trusses fabricated using 5.1 cm x 10.2 cm (2 in x 4 in) graded SYP #2 and spaced 61 cm (24 in) on center with 2.5 cm x 15.2 cm (1 in x 6 in) nominal SYP tongue and groove decking boards. The boards were attached using two 8d sinker nails at each intersection with a truss. The nails were installed within 2.5 cm (1 in) of the edge of the boards. The addition' s roof system was framed with SPF #2 with 5.1 cm x 15.2 cm (2 in x 6 in) nominal lumber at 40.6 cm (16 in) on center spacing. The roof deck sheathing was 11.1 mm (7/16 in) OSB attached with 8d sinker nails 15.2 cm (6 in) on center on the panel edges and 30.5 cm (12 in) on center in the field of the panel. In both the original house and the addition, the lumber grade stamps indicated that the moisture content in the lumber was 19 percent or less at installation. Additionally, the nails in all areas were full round head nails without any coatings and showed no evidence of having been collated. Therefore, it was assumed that, in both species, the nails were hand driven. The typical test started by identifying the location of the underlying framing member and cutting away the asphalt shingle roof covering. In order to create an access area for attachment of the "hook" under the nail head, an 8 cm (3 in) hole saw was used to cut a ring of material around the nail to the depth of the sheathing. To clear the roof decking material away from the head of the nail, a wood chisel was used to carefully remove the material. The bottom of Figure 2 shows the ring that was typically cut around the nail with the hole saw. The center of the picture illustrates a nail, on the addition, with the OSB removed from around the shank. If, during this process, the nail was observed to be disturbed by the chiseling of material, it was eliminated from the test series. Once the nail head was clear of the roof decking material, the standing height of the nail was measured. The fastener extractor was then placed in position, attached to the nail, zeroed and loading was applied. The peak load was then recorded as well as the length of the nail. This procedure allowed the nail withdrawal capacity to be measured in Newton's per millimeter (pounds/inch). The moisture content of the lumber was checked with a Delmhorst hand held moisture meter. The moisture content was found to be consistently at 6 percent in all areas tested. A total of 203 nails were withdrawn from the roof systems, 105 from the area framed with SPF rafters and 98 from the area framed with SYP trusses. The results from the testing are presented in Table 2.
Table 2: Summary of the Field Testing Results Wood Species SYP SPF Number of Tests 98 105 Mean 11.9 N/mm (68 lb/in) 7.4 N/mm (42 lb/in) COV 56% 59% Data Range 0.4-27 N/mm (2-153 lb/in) 0.7-25 N/mm (4-140 lb/in) Figure 2: Nail with Sheathing Material Removed from Surrounding Area Discussion of Test Results The initial comparison of the data indicated that the mean withdrawal data was higher for SYP than for the SPF. This is consistent with predictions using the ultimate nail withdrawal equation, since the specific gravity of the SYP is higher than that of SPF. This is of significance, because other laboratory studies have observed cases where there is no increase in withdrawal capacity between the two species in the case of hand driven nails [8]. The field test data had a much higher coefficient of variation (COV) than typically observed in laboratory testing. Fastener withdrawal in wood is generally known to be quite variable but laboratory tests generally indicate COV's on the order of 20 to 40 percent. Tests of 6d hand driven common nails 5 in SPF displaced laterally before withdrawal, exhibited COV's on the order of 50% [10]. Therefore, it may be assumed that the large COV is the result of actions or conditions that occurred after driving. The large COV most likely was caused by the in-service conditions that occurred after the installation of the nail. These conditions include the time since installation, the decrease in moisture content, disturbance caused by re-roofing, the heat of the heating and drying from typical attic conditions, effects of the fire or possibly the testing method. Additional field tests and controlled aging tests could help determine the cause of the increased variability. In order to be able to make statistical inferences (necessary in order to evaluate implied factors of safety), the underlying distribution of the test data was examined. Both sets of test data were found to fit the Extreme Type I (Gumbel) distribution, according to the Kolmogorov-Smirnov (K-S) test. However, the use of the K-S test only gives an indication of how the underlying distribution fits to the theoretical distribution over the entire range of data. Since the lower tail is of greatest concern when determining design capacities (i.e. 5 th percentile or 95% exclusion), a graphical technique was also used to investigate the lower tail behavior of the test data. An inverse CDF method was employed to compare the test data to a number of candidate distributions including the Normal, Lognormal and Extreme Type I (Gumbel) distributions. The inverse CDF method compares the underlying distribution to a line having a slope of one. The line represents a perfect fit of the data to the underlying distribution. This graphical technique allows investigation of the tail behavior of 5 The 6d common nail has the same shank diameter as the 8d sinker nail.
interest and its closeness of fit to the underlying distribution. Both distributions fit the Extreme Type I (Gumbel) in the lower tail region relatively well. The 95% exclusion value was then determined for both test series. The 95% exclusion value is the design point where it is expected that 5 out of 100 samples will fail. This value, also known as the 5 th percentile value, is often used as the basis for strength (capacity) design values. Table 4 provides a comparison of the computed design values for the 8d sinker to the mean values obtained from the tests as well as the 95% exclusion values. Illustrations of the test series data and their fit to the Extreme Type I (Gumbel) distributions are provided in Figures 3 and 4. These figures also provide reference to the test means, design values and 95% exclusion values. Table 4: Summary of the 8d Sinker Withdrawal Values SYP SPF Computed Design Value, C D =1.0 6.1 N/mm (35 lb/in) 3.2 N/mm (18 lb/in) Computed Design Value, C D =1.6 9.8 N/mm (56 lb/in) 5.1 N/mm (29 lb/in) Test Mean 11.9 N/mm (68 lb/in) 7.4 N/mm (42 lb/in) Test 95% Exclusion Value For Extreme Type I Distribution 4.2 N/mm (24 lb/in) 1.4 N/mm (8 lb/in) Predicted Withdrawal (N/mm) 35 30 25 20 15 10 95% Exclusion: 4.2 N/mm (24 lb/in) Design Value Cd=1.0: 6.1 N/mm (35 lb/in) Design Value Cd=1.6: 9.8 N/mm (56 lb/in) Test Mean: 11.9 N/mm (68 lb/in) 5 0 0 5 10 15 20 25 30 35 Actual Withdrawal (N/mm) Figure 3: Inverse CDF of the Extreme Type I (Gumbel) Distribution Compared to the SYP Test Distribution and Design Points Table 4 and Figures 3 and 4 confirm that the 95% exclusion values are less than the design values for both wood species. The design values calculated using a load duration factor of 1.6 are 2 to 4 times greater than the 95% exclusion value and only about 2.3 N/mm (13 lb/in) less than the mean capacities from the test data. The design value obtained using a load duration factor of 1 corresponds to a factor of safety of 5 applied to the ultimate withdrawal equation value. However, results obtained in this study show that the use of a factor of safety of five applied to the ultimate withdrawal equation does not provide a withdrawal capacity greater than the 95% exclusion value.
25 95% Exclusion: 1.4 N/mm (8 lb/in) Design Value Cd=1.0: 3.2 N/mm (18 lb/in) 20 Design Value Cd=1.6: 5.1 N/mm (29 lb/in) Predicted Withdrawal (N/mm) 15 10 5 Test Mean: 7.4 N/mm (42 lb/in) 0 0 5 10 15 20 25-5 Actual Withdrawal (N/mm) Figure 4: Inverse CDF of the Extreme Type I (Gumbel) Distribution Compared to the SPF Test Distribution and Design Points The implied exclusion values based on the design values can be determined for both species and load duration factors to give an indication of the relative risk of failure. For the load duration factor of 1.0, the implied exclusion values are 86% and 84% for SYP and SPF, respectively. This suggests that for SYP there would be 14 out of 100 nails with capacities failing below the design capacity. Using the load duration factor of 1.6, the exclusion values are 59% and 66% for SYP and SPF, respectively. For SYP, this means that 41 out of 100 nails would be expected to exhibit capacities below the design value. Comparison of SPF Field Withdrawal Data to Lab Withdrawal Data Zaitz [10] performed withdrawal tests on 6d nails hand driven into SPF. The results of this controlled series can be compared with the field test data obtained from the addition to the test house. The control data set, which had no lateral displacement applied, was also analyzed using the K-S test and the inverse CDF method. The 6d common nails withdrawn after one week fit the lognormal distribution quite well, both overall and in the lower tail region. The mean of the 40 samples tested was 6.0 N/mm (34 lb/in) with a COV of 36%. The 95% exclusion was determined to be 3.2 N/mm (18 lb/in). This laboratory test series had a 95% exclusion value that matched the withdrawal design value with a load duration factor of 1.0. The 95% exclusion value from the field test data was 40% lower than the 95% exclusion obtained from the laboratory data suggesting that the in-service conditions at the test house did contribute to a reduction in withdrawal capacity. Proposed Reduction Factor for the Design Equation for Attic Conditions The limited amount of withdrawal data collected in this study indicates that reductions in design values are needed if the design resistance is to exceed or equal the 5 th percentile value. Since the 5 th percentile value is 69% of the mean withdrawal capacity derived from testing in SYP and 43% of the mean withdrawal capacity derived from testing in SPF, a linear reduction may not be suitable based on these data sets. If a factor is applied to the exponential in the design withdrawal equation, modifying the design equation to W=9.5 G 3.3 D L (W=1380 G 3.3 D L) the resulting design capacities of 3.9 N/mm (22 lb/in) for SYP and 1.6 N/mm (9 lb/in) for SPF are very close to the 5 th percentile withdrawal capacities obtained from testing reported in this paper of 4.2 N/mm (24 lb/in) and 1.4 N/mm (8 lb/in) for SYP and SPF, respectively.
Clearly, this data needs to be corroborated with additional test data from other houses that may have different environmental conditions, wood species and nail types. CONCLUSIONS The results of this study suggest that there may be a reduction in withdrawal capacity after the nail is subjected to in-situ conditions and therefore design values may be non-conservative. The 95% exclusion values of 4.2 N/mm (24 lb/in) and 1.4 N/mm (8 lb/in) for SYP and SPF from the test data are lower than the withdrawal design values using either 1.0 or 1.6 for the load duration factor. The test data indicated that if the 1.6 load duration factor was used, the design value actually approaches the mean ultimate capacity obtained from the tests. The data exhibited very large variability. While it is likely that this variability is due to the effects of aging and the decrease in moisture content, other possible factors including the test method and potential influence of the fire which damaged other parts of the structure needs to be assessed. A comparison was also made to laboratory data obtained from a set of withdrawal tests conducted using a similar species of wood and nail diameter. This comparison also indicated a decrease in the 5 th percentile value of withdrawal capacity for the field tests as compared to the laboratory tests. This supports the findings of previous studies suggesting that, as the moisture content decreases, the withdrawal capacity also decreases. Again, this is based on one set of comparisons and may not be representative of the in-service nail population. ACKNOWLEDGEMENTS The author would like to extend thanks to Mike Kirby for his assistance in testing the fasteners, and to Steve and Lynn Sears for graciously allowing the testing to be performed on their home. The South Carolina Sea Grant Consortium proposal number 98-028 provided funding for this research. REFERENCES 1. Rosowsky, D.V, and Reinhold.T.A. 1999. Rate-of-Load and Duration-of-Load Effects for Wood Fasteners. Journal of Structural Engineering. Vol 125, no. 7, pp. 719-724. 2. American Forest & Paper Association. 1993. Commentary on the 1991 Edition of ANSI/NFoPa National design specification for wood construction. AF&PA. Washington, D.C. 3. USDA Forest Products Laboratory. 1999. Wood Handbook: Wood as an Engineering Material. FPL-GTR-113. USDA Forest Service, Forest Products Laboratory. Madison, WI. 4. American Forest & Paper Association. 1997. ANSI/NFoPa NDS-97 National design specification for wood construction. AF&PA. Washington, D.C. 5. Heyer, O.C. 1963. Study of Temperature in Wood Parts of Houses Throughout the United States. FPL-RN-012. USDA Forest Service, Forest Products Laboratory. Madison, WI. 6. Scholten, J.A. 1965. Strength of Wood Joints made with Nails, Staples, or Screws. Research Note FPL-0100. U.S.D.A Forest Service, Forest Products Laboratory. Madison, WI. 7. Senft, J.F and Suddarth, S.K. 1971. Withdrawal Resistance of Plain and Galvanized-Steel Nails During Changing Moisture Content Conditions. Forest Product Journal. Vol. 21, no. 4, pp.19-24. 8. Pye Jr., S.J. 1995. Effect of In-Service Conditions on the Withdrawal Capacity of Roof Sheathing Fasteners. Master of Science Thesis. Clemson University. 9. Kurtenacker, R.S. 1965. Performance of Container Fasteners Subjected to Static and Dynamic Withdrawal. FPL-29. USDA Forest Service, Forest Products Laboratory. Madison, WI. 10. Zaitz, M.D. 1994. Roof Sheathing Racking Effect on Fastener Withdrawal Capacities. Master of Science Thesis. Clemson University.