You can bracket a set of within soil deflection behaviour responses by setting the pile in a matrix of soil (or equivalent horizontal springs) in a FEM model. The better and more suited the program, the better the results and less the bracketing iterations needed to ascertain the likely behaviour. When using springs, I use 1/2 to 1/3 of the modulus of subgrade reaction as basis for the lateral springs. If starting from some "modulus of elasticity", it uses to be necessary assume stiffer soils than as recommended in the geotechnical books to obtain realistic or likely behaviours. And I have said likely, not the safest. Accuracy to the reality and safety are 2 different approaches.