6 Replies Latest reply on Oct 8, 2010 10:09 AM by Derek Bishop

    Elastic supports used to model soil stiffness

    Derek Bishop

      I'm doing some FEA on a corrugated sheet metal tunnel that will be installed as a 5 m diameter reclaim tunnel under an ore stock pile. Think of it as a big round pipe buried under dirt. There is a rectangular cut out in the top but that is irrelevant to this question.

       

      Not sure how to restrain the model. An elastic support with the appropriate stiffness of soil and a normal pressure to represent compaction of the soil around the tunnel is my current thinking. Question is when doing an elastic support, does it represents a stiffness in both directions. That would be true for most materials. Dirt is a bit different. It will compress with a certain stiffness but it won't spring back the other way with the same stiffness. Any thoughts.

        • Re: Elastic supports used to model soil stiffness
          Anthony Botting

          Possibly elastic support. It does have a setting for tangential and normal components of stiffness (distributed, or total). A thought that came to mind is to create a solid body of dirt and surround the tunnel with it, assigning bonded contact and developing some stiffness values for the dirt. I read that FFEPlus solver does not really like large changes in stiffness such as the interface between the dirt and the steel, so you'd probably need to use the Direct Sparse solver. Another thought is to try the Spring connector (perhaps put an array of them around the tunnel) - Spring Connector has an option to make it "compression only" and lose stiffness in tension. Another thought is to cart the whole she-bang into GeoSTAR (a.k.a., "COSMOSM") if you have the "Simulation Premium" product. The GeoSTAR interface has an actual "soil" material model you an assign to the body surrounding the tunnel structure. Just some thoughts. Good luck. -Tony

          • Re: Elastic supports used to model soil stiffness
            Bill McEachern

            You could try adding in the soil, use the drucker pager soil model (supported in SWX simulation) in an NL study and then compress the soil with a fixed displacement. Not too sure whether contact between the two would be required or not. Probably have to run both and see what the differences are. the size of the model in the other direction would also have to be explored. You can try a web search for soil parameters to enter into the model. Might be a bit much for what you are after......