Solid elements model a complete system of partial differential equations. As a result, you find the entire stress and strain tensor. This requires a long computation time and does not always give the required accuracy.

Shell elements reduce the dimension of the problem and take into account only principal stresses. Therefore, the task is considered faster and the counting accuracy will be higher.

Shell elements allow the modelling of thin features with fewer elements than solid elements, they need low computational time. They are also easier to mesh and less prone negative Jacobian errors which might occur when using solid elements.

I think there are several reasons.

First, the parts of these constructions are really shells: the size in one dimension is much smaller than in two others.

Second, if the solid element has one size much smaller than two others, it will not work properly and may lead to the loss of accuracy. This is not so for the shell elements, of course.

Third, in the solid elements, each node has three degrees of freedom: three Cartesian components of the displacement. In the shell elements each node has five degrees of freedom: three displacement components and two rotations out of the tangent plane of the shell.

Fourth - most probably, when modeling with the solid elements, you will need much more nodes, thus consuming more resources: memory and computational time.