Can you provide a bit more information  on the situation you are actually considering? 

Dittus-Boelter is for a fluid, so you can use it both for a vapor and a liquid provided you know the Prandtl number, and Re and Pr are in the right range. Of course, thermal conductivity will differ greatly between gases and liquids. 

That all depends on the geometry, operating conditions and nature of your vapor - D-B or Sieder-Tate do work for vapors provided you are in the right range of Re and Pr. However, I assume you mean that vapors are typically on the shell-side, and DB/ST are tubeside correlations. That's true of course, hence we need more info on the geometry/problem for a sensible answer. 

Anyway, for the original question, perhaps this helps: http://www.chemstations.com/content/documents/Technical_Articles/shell.pdf

Thank you Cees and Divyesh.

Consider a two phase flow, boiling or condensation. I have seen Kandlikar in one of his papers, where he mostly deal with boiling, he divides his two-phase heat transfer coefficient by liquid only heat transfer coefficient. And he calculates liquid only htc using Petukhov-Popov correlation. In another paper, condensation one, they divide their two-phase htc by liquid-only htc, where they use Dittus Boelter correlation for the liquid only htc.

I was wondering whether we can divide our two-phase htc by vapour only htc. I have not seen any reference mentioning vapour-only htc. What are the consequences of using vapour-only htc over liquid only htc?

The reasoning behind using 'liquid only' heat transfer coefficients as part of two-phase flow problems (boiling and condensation) is that it is the liquid phase that is in physical contact with the wall/surface, hence it is the liquid phase thermal conductivity (and other properties) that are of primary importance in the near-wall region. The term 'liquid only' also means that the heat transfer coefficient should be calculated assuming that all the mass flow is in liquid state (as opposed to 'liquid' indicating that only the liquid fraction of the mass flow is used in calculated Re). The remainder of the correlation thus represents a 'tw-phase multiplier', which takes into account the influence of vapor flow and/or bubble nucleation.

So, the short answer to your question is: No, you can not introduce the 'vapor only' htc, since the vapor is normally not in contact with the wall, hence the vapor thermal conductivity and turbulence level is not of primary interest.

I see now why everyone use liquid only heat transfer coefficient. Thank you very much Prof. Erling. You have made it clear with your explanation.