ghorrocks 
June 5, 2011 07:10 
Yes, you need to model the solid part. Mesh it using the same techniques used for a fluid mesh. But remember as the equations to be solved in a solid domain are linear and simple you can usually mesh much more coarsely and with poorer quality, but on the other hand a solid domain node only has one equation to solve compared to the 7 or more often solved in a fluid domain so the cost of solid domain meshes is low... so why not mesh finely.
Remember the solid timescales are usually far slower than fluid timescales so rapid fluid transients result in very thin regions of high thermal gradients at the surface of the solid. If this is important you should consider an inflation like mesh at the surface to capture the steep gradients.
