A Rational Approach to Estimate Reasonable Design Values of Selected Joints by Using Lower Tolerance Limits

Mesut Uysal, Purdue University

Abstract

In this study, reasonable design values of selected joints were estimated by using the lower tolerance limits (LTLs) method. Although furniture members and joints can be designed if their strength capacities are known, design values of furniture joints have not been established. There are studies relating to the allowable design values of furniture joints that exist in literature review, but they have not been well-addressed. Another fact of information is that allowable design values for wood materials have been defined for building wood structure. However, wood materials used in furniture industry are smaller in size and primarily defect-free compared to those of wood structural material, so some mechanical properties of wood may differ from each other. Therefore, a determination of reasonable design values of furniture joint is needed. Furniture joints are the weakest part of furniture structure. Increase in reliability of furniture joints provides lower failure probability for furniture structure during its service. Thus, furniture joints were selected to determine their reasonable design values in this study. In the literature review, numerous studies have been conducted to determine strength capacities of furniture joints considering different wood species, joint sizes, adhesive types and tolerance fit of joints, etc. These studies were related to prediction interval or deterministic approach which does not satisfy the reliability of products. On the other hand, probabilistic approaches ensure reliability and safety of products. Tolerance intervals are one of the probabilistic approaches. Therefore, one-sided lower tolerance limits are used to determine reliability and safety of furniture joints. Rectangular mortise and tenon (RMT) joints, two-pin moment resisting dowel joints, and screws, which are widely used joinery methods in furniture industry, were selected to study. All specimens were made of northern red oak and white oak wood. T-shaped RMT and dowel joints were examined in bending tests, while the screw withdrawal strength in wood from end-, edge- and face-grain were evaluated for screws. In order to determine sample sizes in tolerance analysis, reference data was used according to the modified Faulkenberry-Weeks method. According to test results, 220 specimens were used for each sample group to calculate lower tolerance limits of furniture joints. In the tolerance analysis, randomness, homogeneity and normality assumptions were considered. If data was not normally distributed, one-sided LTLs were calculated for a nonnormal data set. Firstly, the logarithmic normalizing transformation was sought. If the logarithmic data was normally distributed, LTLs were then obtained by using calculations for a normally distributed data set. The logarithmic value was then inverted. If the logarithmic data was not normally distributed, Weibull distribution was then used to calculate LTLs. Furthermore, non-parametric tolerance analysis was used if data did not fit the Weibull distribution. According to test results, data was normally distributed for six sample groups; namely, RMT joints made of white oak, dowel joints made of white oak, screw withdrawal strength in wood made of both red oak and white oak from end-grain, screw withdrawal strength in wood made of white oak from edge-grain, and screw withdrawal strength in wood made of red oak from face-grain.

Degree

Ph.D.

Advisors

Haviarova, Purdue University.

Subject Area

Design|Wood sciences|Mechanics

COinS