Many timber design codes around the world use so-called ‘group effect’ factors to reduce the design strength of multiple fastener connections, though the basis for the factors is inconsistent and the original conceptual model is not relevant to many modern connections. This work expands on other recent studies that have observed reduced variability in larger connection groups, leading to higher design resistances, and proposes a function to define the group effect and three sub-factors related to the changes with respect to the number of fasteners in the reliability, strength variability, and mean strength.

To evaluate the reliability sub-factor a process was developed for calibrating material resistance factors for timber connections, which included collecting data on load distributions in Canada, selecting a target reliability index, and processing the results from first order reliability method (FORM) analyses. The process was developed for the case of axial withdrawal of self-tapping screws using resistance data collected from past studies across Canada and resulted in a resistance factor of 0.7. The process was applied using a generic Weibull distribution to illustrate the relationship between the reliability sub-factor and inverse of the Weibull shape factor, which was used to represent the strength variability. Decreasing strength variability in larger groups tended to result in higher values for the reliability sub-factor, which tended to increase the factored design resistance except when the single fastener strength variability was low.

Tests on axially loaded self-tapping screws in withdrawal demonstrated no change in the mean strength for up to 12 screws, with the screws widely spaced at 10 times the nominal screw diameter. The variability in ultimate strength decreased as the group size increased, supporting the assertion in the literature that there is a ‘balancing’ effect in large groups, and approximating the rate obtained from application of the central limit theorem. The combined effect of both sub-factors resulted in an increase of 25% in the design withdrawal resistance for 12 screws. Simulations were then carried out for connections with inclined self-tapping screws and steel side plates using input load-displacement data obtained through testing. The mean strength was sensitive to the way displacements were applied to each column of fasteners:​ negligible change when the displacement was uniformly applied across all fasteners, producing results similar to the test results for axially loaded screws in withdrawal;​ significantly decreasing in large groups, especially with more columns and a higher ratio of connector stiffness to axial stiffness of the connected members. In the latter case, premature failures produced an increase in ultimate strength variability. The general trend of lower variability resulting in higher specified design strength increased the factored design resistance an average of 10% for less than 30 screws;​ larger screws were highly dependent on the manner of load distribution.

The results of the study demonstrate that the factored design strength per fastener of a connection group may be greater than the factored design strength of a single fastener. Future experimental work is needed to develop specific factors for a range of connection configurations and failure modes.