I have not solved such problems. But, apparently, under the established structure occurs a failure. And, there occurs settlement of what stands on it. The only thing comes to mind is to make an array of soil solids. To model lens using system "assemblage" by stages and demolish it by parts. For the first make it in linear formulation. Then try non-linear FE of the soil (they are also iterative). If you will build this problem in linear formulation, try to make this lens using at list 10 different sets of FE instead of one. Divide the stiffness (Edef) between them and make each stiffness exclusive that it was convenient to access them. Demolishing them gradually would reduce the stiffness of the structure. The building on the top will react the changes. There is also another way. You may use set of springs and type of processor. In addition, there are possible not only machinery iterations, but also their modelling. In linear formulation, all will be calculated faster. There will be enough 100 of iterations, may be even less.