At the end of the day you always end up with nodal forces no matter what loads you apply but you always have to keep in mind that these nodal forces are energetically equivalent to the distributed load you applied. Therefore, you always have to integrate over the domain where you apply the load to obtain the values for the nodal forces.
The following comments are valid for a constant pressure load:
Considering linear elements this is rather easy. You need to know the lenght of the edge where you apply the distributed load to and then just compute the resulting force. One half of this force is then distributed to each of the two nodes on the element edge. Accordingly, it is correct that in your example only the corner vertices of your structure have a different nodal force value (assuming that the mesh is 100% regular, meaning that all adges at the boundary are of identical length). Considering finite elements with quadratic shape functions we have a different situation. Here you distribute the resulting force with 1/6 to each of the corner nodes and 2/3 to the mid-edge node.
If the assumed pressure distribution varies with a certain prescribed function you really have to compute the integrals in order to derive the energetically equivalent nodal forces.
Thanks a lot for your reply Mr Sascha and Mr Hauke.
I am using linear elements which is shell 181. The pressure distribution is even on the line. I need to determine the stress intensity factor at crack tip. I applied the displacement interpolation method in ANSYS. But I get an error stating 'The crack face is not parallel to the currently active X axis'.