A Source of Exceptional Strength - Post-Frame Diaphragm Action
Although engineers have long recognized diaphragm action in buildings, not until the late 1950's did researchers begin studying metal-clad wood-frame diaphragm behavior. In 1961, Hurst and Mason published the results of a series of tests conducted on two separate (but similar) metal-clad pole buildings. Both buildings were 36 ft wide, 45 ft long, with a 13 ft eave height, 4/12 gable roof, and round poles. Each pole was kneebraced and embedded 4.5 ft. The buildings were tested in various stages of construction. Although an entirely clad building was never tested (i.e., one end wall was never framed or sheathed/clad), the researchers showed that roof and endwall cladding contributed significantly to the overall rigidity of the structure.
Despite Hurst and Mason's significant findings in the late 50's and early 60's, no further research involving metal-clad wood-frame diaphragms was completed until the late 70's when Hausmann and Esmay (1977) and White (1978) reported results of laboratory experiments conducted on metal-clad wood-frame diaphragm test panels. Diaphragm test panel results are used to calculate shear strength and effective shear stiffness of full-size building diaphragms. Except for overall size, a diaphragm test panel is meant to be functionally equivalent to the diaphragm in the building being designed, and is to be supported (during test) in a fashion similar to that encountered in the building application. The research by Hausman, Esmay and White was primarily spurred by the tremendous amount of work being conducted on metal-clad steel-frame diaphragm assemblies at that time (for an extensive bibliography of this work, see chapter 20 of Davies and Bryan, 1982).
In 1983, Hoagland and Bundy presented a procedure for calculating the percentage of horizontal wind load transferred to shear walls by metal-clad roofs in post-frame buildings. This diaphragm design procedure was based on methods outlined for metal-clad steel-framed diaphragms by Bryan (1973), and featured an equation developed by Luttrell (1967) for extrapolating diaphragm test panel data for use in full-scale building design. This procedure was slightly modified by Gebremedhin and others (1986), and formed the basis for an ASAE engineering practice (EP) entitled Diaphragm Design of Metal-Clad, Post-Frame Rectangular Buildings (ASAE, 1990). Work on the EP commenced in 1986 and was approved for publication in 1989 (Manbeck, 1990).
Since Hoagland and Bundy first drafted their procedure in 1983, there have been three more full-scale building tests involving metal-clad wood-frame diaphragms (Johnston and Curtis, 1984; McFadden and others, 1991; Gebremedhin and others, 1992).
Two primary purposes of the full-scale building tests were to demonstrate contribution of metal cladding to building stiffness, and to assess accuracy of horizontal eave displacement predictions. Whereas researchers were very successful in demonstrating the contribution of metal cladding to overall building stiffness, the ability of the researchers to accurately predict horizontal eave displacements has not been shown. This is because the accuracy of displacement predictions is highly dependent on (1) assumptions made regarding diaphragm continuity at the building ridge, (2) the method used to test diaphragm panels and calculate panel properties, and (3) the procedure used to obtain building diaphragm shear stiffness values from test panel properties (Anderson, 1990; Bohnhoff et al., 1999, Bohnhoff and Williams, 1999).
Inability to accurately predict eave displacements can be attributed to the simple fact that researchers still do not fully understand the complex distribution of loads in metal-clad wood-framed diaphragms (Bohnhoff et. al., 1999). This can be partly blamed on the fact that in the vast majority of testing, not a single component strain has been measured (i.e., only gross assembly displacements have been monitored). Without knowledge of how loads are distributed throughout a building or test assembly, it is virtually impossible to properly size individual components within a structure, and/or accurately predict building behavior.
To date, three main research projects have attempted to better understand load-distribution in metal-clad wood-frame diaphragms by investigating individual component strains/forces. This includes full-scale building testing by Niu and Gebremedhin (1997), and finite element modeling (FEM) work by Wright (1992) and Williams (1999).
Niu added strain gages to several purlins in the building constructed by Gebremedhin. Although Niu and Gebremedhin where able to show some general trends, relationships between applied loads and various purlin forces could not be established because (1) measured strains could not be separated into strains due to bending, torsion, axial load, etc., and (2) gages were placed directly on wood and were not calibrated by individually loading each purlin.
Using a commercially available finite element analysis program, Wright (1992) modeled one-half of a diaphragm test panel measuring 8- by 12-feet. This "half-model" contained 11,644 nodes, 11,515 elements, involved 69,864 degrees of freedom and required a super-computer for analysis. Overall Wright found good agreement between predicted deflections and those measured during laboratory test of three diaphragm panels. Because of the size of the model, and therefore the time and cost of an individual analysis, Wright was not able to conduct valuable sensitivity analysis. The size of Wright's finite element model was directly attributable to the number of elements used to model the corrugated steel panels. This in turn was due to element aspect ratio (length-to-width) limitations which limited the four-node shell elements used in the model to a maximum of about 1 inch.
To reduce the degrees of freedom associated with diaphragm modeling, Williams (1999) used a four node, 8 degree-of-freedom, isoparametric plane-stress element with orthotropic properties to model "larger sections" of corrugated panels. Properties for this element were obtained through a combination of laboratory testing and FEM. In addition to this element, Williams developed special nonlinear elements for modeling connectors, and incorporated these elements along with a conventional beam element into a finite element analysis program that he wrote. Williams used his program to analyze 18 diaphragm test panels that he had tested to failure, and found predicted displacements to be within 30% of measured values. Variances were attributed to unaccounted for joint slip due to combined nail bending and withdrawal, and an overestimation of the shear modulus used to model shorter pieces of cladding.
(Source: ASAE paper No 024007)