The simplest way would be to estimate the flow velocity on your geometry. Assume laminar flow. Then based on that find the Reynolds number. And then there are many empeiro-technical formulas based on Reynolds, that give you the heat convection transfer coefficient. One good book for these formulas is the Waerme Atlas.

Thank you Gryllakis. I have computed Re and used formula for straight part with twisted tape to get HTC there. Then I assumed no influence of twisted tape in the bent part and so used other formula for Nu (based on Re, Pr) and subsequently HTC. The question is how good is the assumption of no influence of twisted tape in the bent part of the tube. 

Hello Robin, 

When you mention twisted tape, are you referring to a heating tape on the outside of the pipe? Or is it a twisted strip inside the tube? I will comment more about your question after understanding about this twisted tape.

I agree with Gryllakis on the general approach to this problem. 

Hi Juan,

Sorry for not enough defined problem. The twisted tape is inside straight part of the tube to enhance heat transfer from outside of the tube to inside. See the attachment. 

The validity of your assumption depends on the characteristic of your flow. If you keep a laminar flow on your straight pipe, I think the swirling motion of the flow will take some distance to fade out so the enhancement in heat transfer coefficient will be present for a portion of your bent pipe. I think one good option you have to tackle this problem is using a simple CFD simulation to learn more about your flow. Have you used CFD software before?

Thank you for your help Juan,

my flow is turbulent. Well I have some experience with Fluent, CFX and OpenFOAM but not really much. We also think about experiment (plexiglass and PIV) on this at my work place. We will see.

Thank you again both.

Ok now i understand a bit better the problem.For a first practical approximation for your convection heat transfer coefficient, i would just assume ,again, laminar flow and calculate the same way the problem. In the end i would multiply the result with a 1.1 or 1.2 multiplayer, because thats 10% or 20% increase in the heat transfer due to the rotating movement of the fluid due to your tape. I base my assumption on simple experimented formulas for turbulence, which depend on 0.7th power of reynolds number in contrast to 0.5th power in laminar flow case. So the difference in the heat transfer should be somewhere there. Then you could find an analytical solution,maybe superimpose a simple rotating motion, with a linear fluid motion? And then i would go to CFD.

Dr. Kanizawa has made some experimental studies on heat transfer coefficient behind a tube with a twisted tape. Here one of them, http://www.sciencedirect.com/science/article/pii/S0140700715002790

If you are going through the CFD route, Fluent should be the easier option you have in terms of implementing your case. If you are up for some collaboration and you have the 3D geometry, I could give the simulation a try very quickly.

Thank you very much all of you guys. Your comments are very helpfull. I will follow your suggestions and probably come here back to this thread with a results.

Juan thank you for your kind suggestion to help me with the CFD simulation. Unfortunatelly, I do not have 3D geometry. But my collegue want to try a similar CFD simulation just only with exchanged bent part for straight part.