tbulk for wall HTC choice with multispecies analysis
 

When simulating something like cold water falling from a faucet due to gravity onto a hot pipe in the presence of air as the environment, what tbulk would be appropriate to use?

The water flows from the faucet at ~5 C, but obviously heats up before and upon hitting the hot pipe to varying temperatures that are location dependent.

The surrounding air has a T bulk of 25 C.

I know wall heat transfer coefficient is either based on a CFX determined temperature some distance from the wall, or on a user defined bulk. I can't think of an appropriate single bulk temperature value to put in the expert parameters, nor am I sure the CFX determined bulk temperature is valid either.

Any suggestions or experience with something like this?

Thanks

The value for Tbulk is not all that important.
You only have to know which value for Tbulk you used if you want to compare values of the heat transfer coefficients to similar simulations or experiments.

I had thought Tsurface was not important, but Tbulk was. The reason I say this is that while keeping Tsurface as the constant (as well as the fluid properties), the change from CFX determined Tbulk and user defined Tbulk of 300K resulted in a Wall HTC change from ~34,000 W/m^2 K to ~1,400. Everything else being equal.

It is only the heat transfer coefficient (a derived quantity) that changes when Tbulk is changed, the solution (especially the heat flux) remains the same.

I don't fully understand your phrasing. I imagine CFX calculates HTC from Newton's Law of Cooling. Q"=HTC*(Ts-Tb). I set Ts to a constant at this heated surface. It would seem then that HTC=Q"/(Ts-Tb ). So the change in Tb does change HTC, right? At least that's what CFX is reporting back. Perhaps I'd be better providing a constant Q" instead of Ts? All I know is that all things being equal, changing Tbulk from CFX default near wall temp to 300K changes HTC from ~34,000 to ~1,400. I still think a proper Tbulk is important as it is in the publications I read on finding HTC with constant Ts, but if I am doing something wrong with the model I am open to changes.

I will try to make it more clear.

Lets forget about Tbulk and heat transfer coefficients for now.
What CFX and any other CFD software does is to calculate the heat flux at a surface. This is the important physical quantity.
Of course the heat flux will be different for different boundary conditions, i.e. constant temperature or constant heat flux. It is up to you to decide which boundary condition you need.
Now the simulation is done.

In a POST-PROCESSING step, the heat transfer coefficient is calculated.
If you provide a Tbulk similar to the wall temperature, the heat transfer coefficient will be high.
If you provide a Tbulk with a higher difference from the wall temperature, the heat transfer coefficient will be low.
If you dont specify the Tbulk expert parameter, CFX will estimate some non-constant value for Tbulk, resulting in rather arbitrary heat transfer coefficients from an engineers point of view.

But whatever you do, the result of the simulation remains unchanged, only the post-processed coefficient changes.

If we consider the simple problem of an initially hot solid body in a cold stream of fluid and we want to find out how the temperature of the solid body changes over time,
we get the exact same result no matter what value for Tbulk we provide.
The important physical quantity that determines the process is the heat flux.

You are saying the change in the temperature of the solid does not change based on Tbulk/Tinfinity? That seems to be contrary to basic heat transfer.

Otherwise, Tbulk/Tinf would not be a variable in the convection cooling equations, for steady state nor transient.

But ignoring that, it seems you are also saying that I will never be able to get a meaningful HTC from CFX to map onto a solid for FEA.

Sure you will get a reasonable answer from CFX for this kind of simulation.
When specifying a convective boundary condition, the FEA software will require Tbulk and the heat transfer coefficient.
As long as you use the same Tbulk that was used in the fluid simulation to determine the heat transfer coefficient, everything is ok.

I think we are mixing two things here.
This is only in contradiction to heat transfer mechanisms if you use a convective boundary condition that replaces the simulation of the surrounding fluid.
Of course in this scenario the result will depend on the value of Tbulk and htc, but only because these values are part of the boundary condition for the simulation.

If on the other hand you actually simulate the surrounidng fluid, the htc is a post-processed variable that changes with different values of Tbulk, without altering the simulation result.