Edge loading is only provided for shell elements. Its units are force per unit length. The label is EDNORx where x can take a value between one and three for triangular shells and between one and four for quadrilateral shells.

using *DLOAD keywords for shell element edge load in CCX, also using ‘comp node_groups do’ in CGX to catch shell edge faces from predefined group of nodes.

Can the edge load be applied in the shear direction? Because that’s what I am after

I was thinking of creating new dload user subroutine which is essentially an edge load but, with different load direction. Do you think it will be hard to do?

as far as I know, the answer is not (EDNOR applies only pressure normal to the edge) . I use to create groups of nodes and apply concentrated loads at the nodes. Just remember the different contribution at each node (see picture attached).

well, i did not see carefully you did a cross reference with external software. called as surface traction, a force magnitude applied to group of face element. as i know FreeCAD also doing the same feature approach, you may post a specific question at they forums.

it seems the forces are automatically converted or applied as concentrated load with equivalent magnitude depending on tributary areas of a nodes and element type (especially in quadratic). summation are required for element face sharing the same nodes. so, there’s no modification of user subroutine of dload.f as you describe, also i did not see any hints for these task in user defined loading documentations.

rather than took all above long steps, alternatively one can use distributing coupling features or rigid body keywords if additional stiffness can deal to be ignores.

Is there a formula I could use to calculate the weighted coefficient for any node number?
That is:
2 nodes: (1/2,1/2)
3 nodes: (1/6,2/3,1/6)
4 nodes: (1/8,3/8,3/8,1/8)
…
Thanks.

in order to calculate nodal forces you have to integrate, over the surface/edge, the applied surface/edge tractions in the variational formulation used, with the shape functions of the elements; this is done by numerical quadrature (Gaussian usually). Software does this calculation numerically so there is no need in general to integrate that analytically, even if it’s possible. In the old days (60s, 70s, 80s) when this was not implemented in software and only loads were nodal loads that was necessary, hence the above formulas. From a practical point of view you do not need anything else but the above figure. If you want knowledge in depth check examples 5.8 and 5.12 in this book: https://web.mit.edu/kjb/www/Books/FEP_2nd_Edition_4th_Printing.pdf