Treat the outside coefficient as if it were a flat plate (unroll it). Calculate the inner coefficient as if it were still a cylinder though. Combine the inner, wall and outer resistances to determine your overall heat transfer coefficient (sounds easy when you say it three times fast). You will certianly have to do this in segments along the length of the tube ***IF*** you have a significant temperature rise to accommodate changes in temperature/properties when calculating your heat transfer. BTW, I am assuming that you can ignore natural convection, but if not, just superimpose the natural convection and forced convection heat transfer coefficients when calculating your overall heat transfer coefficient (add the natural and forced convection values together). Note that if you choose not to do this calculation in segments that you need to use arithmetic temperature averages for your property calculations and iterate until your assumed values are close to your calculated values. I would make sure that you do not have really large changes in temperature if you are going to take this approach (look at how much the fluid property values change on both sides and use good judgement). If you do not have a large temperature change, your calculations will be easier and reasonably accurate even without iterating on the asnswer.
Bob
