X.H.4 Final remarks

The values atributed to connection translational stiffness $k_T^i$ can be an issue. Generally test results are not available. The estimation of connection stiffness via detailed FEM is a possibility, but this might prove very complicated and expensive. One notices however, that as far as the distributed loads are concerned, only the relative stiffnessesz of the different connections in an interface matter. This means that if all connections have common characteristics, the use of any common value $k_T^i=k_T$ for stiffnesses will lead to the same results. If bolt group involves several types of connections, the stiffnesses considered in the calculations should be estimated using reasonable assumptions. For example, one can consider a proportionality to bolt section area of the kind $k_T^i=k_T^0{\left({\Phi }^i∕{\Phi }_0\right)}^2$, in which ${\Phi }^i$ is the nominal diamter of connection $i$.

The bolt group equations that are developed in this Appendix consider connections that have translational and rotational stiffnesses. In many case, the rotational stiffness is not necessary to redistribute interface loads. Setting $k_R^i=0$ will lead to a redistribution in which the connections of the interface are loaded with forces only ($m^i=0$). This approach works only when setting $k_R^i=0$ does not lead to a singular matrix $J^{BG}$. This will be the case when interface connections are not aligned along a straight line. Would the connections be distributed along a straight line, and the particular case of a group of two connections saisfies this condition, a rotational stiffness must be attributed to connections to allow a bolt group redistribution.

Then appears the issue of the rotational stiffness to be attributed to the connections. As test results are generally not available, we advice to put a rotational stiffness corresponding approximately to $k_R^i\approx 0.5\cdot {\Phi }^i\cdot k_T^i$.