Surface condition effect on heat transfer to fluid

[pwildfire]

Hi guys,

I am examining a situation in which an aluminum engine part is cooled by forced flow of oil. The surface area cannot be modified, and additional flow rate of oil yields diminishing returns in (measured) heat rejection. My impression is that the limiting factor is heat flux at the interface from the aluminum part to the oil.

My question is what factors determine the heat flux at the surface? I know local HT coefficient equations I have always used tend to ignore this and focus on bulk conductivities, but how true is this assumption? For example, in this situation to calculate local HT coefficient, one would use the conductivity of aluminum and the convective coefficient of the oil. But presumably the aluminum surface is covered with a thin oxide layer. AlO2 has about 1/10 the thermal conductivity of aluminum. Does this have an effect on the heat flux to the fluid? Could heat transfer be significantly improved by applying a surface coating to the aluminum part which has a high HT coefficient and does not form a surface oxide?

Additionally, I have just been reading a paper which describes a large difference in surface heat flux between oils with different additive compositions. How does surface heat flux effect the overall heat transfer (ie heat exchanger equations)? My assumption would be that the convective HT coefficient for all these oils is very similar, but surface effects apparently vary by a factor of 4 in some cases.

I don't actually need to be able to calculate the heat rejection, I can measure it if needed, I just want to understand the physics well enough to make informed assumptions, and right now my understanding of the equations seems to leave out some factors which may be potentially important.

Thanks for any insight you can provide.

Pat

 

[zekeman]

"and additional flow rate of oil yields diminishing returns in (measured) heat rejection. "

So you are saying that increasing bulk rate and maintaining inlet temperature of coolant, at some point reduces the overall heat transfer.

I don't believe it and a microscopic layer of oxide will not much affect the heat transfer . You must be doing something wrong

 

[IRstuff]

This is not the venue for doing a complete heat transfer tutorial. The bottom line is that fluid heat transfer is a function of a host of parameters:
> fluid velocity
> fluid viscosity
> fluid heat capacity
> fluid thermal conductivity

I suggest that you peruse

http://web.mit.edu/lienhard/

http://www/ahtt.html

or the books at

http://www.intechopen.com/search?q=heat+transfer

TTFN
faq731-376

 

[pwildfire]

zekeman,

It does not reduce the heat transfer. What I am saying is the the relationship of oil volumetric flowrate to overall heat transfer is asymptotic. In other words, increasing flowrate beyond a certain point provides little increase in heat transfer. To me this indicates that the limiting factor is conduction of heat from the wall into the bulk flow, rather than transport of heat out of the system. If you have trouble believing this, please provide a more well reasoned explanation than 'you're wrong'.

IRstuff, thank you for the links, I will see if any of those books contain more information than the ones I have. It does seem like you misunderstood my question, though. I already understand basic heat transfer. What I want to know is, what are the assumptions about local surface effects that are contained in the basic heat transfer equations, and what is the magnitude of the potential error caused by these assumptions. Obviously, there are errors present, because things like surface roughness do make a significant difference in the real world, and are not contained in the equations or any of the variables.

For example, the fluid thermal conductivity and metal thermal conductivity are both typically based on tabulated data. How realistic is this data compared to my actual situation?
Also, I don't completely understand what happens at the interface between two materials. At this interface, bulk fluid velocity is zero, and surface roughness and film conductivity etc would presumably be important factors. My feeling is that for normal heat exchangers, the resistance at this interface is small, so it is not normally considered, but as requirements go up, it plays a larger role.
I suppose I could make some test samples and instrument them, etc, but I was hoping for an explanation with less time investment.

Anyway, thanks for the help!
Pat

 

[IRstuff]

I really suggest you do some research before you make assertions that you, yourself, cannot substantiate. You need to do the math yourself to convince yourself that your assertions made here are neither correct nor physically plausible.

TTFN
faq731-376

 

[Compositepro]

"My feeling is that for normal heat exchangers, the resistance at this interface is small, so it is not normally considered, but as requirements go up, it plays a larger role."

This is where you are wrong. The boundary layer resistance is the major factor in convective heat transfer. Compared to water, the higher viscosity and lower heat capacity of oil significantly lowers heat transfer. Lubricating oil is designed to form boundary layer films.

You can't "add" something to the surface to improve heat transfer. All materials are resistances to heat transfer. The build-up of corrosion inhibitors or corrosion on the aluminum surface can reduce heat transfer.

 

[BigInch]

True. Compositepro is right. Surface area across which the heat is transferred and the convective boundary layer there are the keys, especially when viscous oils are concerned. The convective interface heat transfer coefficient must be included in the OTH and is evaluated through Grashoff, Prandtl and Nusselt numbers. You can move all the fluid you want in regions outside the boundary layer against the pipe wall and it won't help pick up any more heat. Once you have reached thermodynamic turbulence close to the wall, it's difficult to get much improvement after that. In transport of extremely viscous heavy oils cP 1300, I've had nearly a solid core in laminar flow inside a hotter turbulent flow layer ringing around it against the pipe wall. Only the outermost fluid layer near the pipe wall was picking up the heat and liquifying while the inside core flow remained like cool toothpaste.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[zekeman]

Out of curiosity, I examined your empirical result and did the following
problem of a coolant flowing in a duct of length L,whose shell temperature remains constant T0. I get the differential equation

(T0-Ti)=1-e^(-Up/rho*A*c*v)L*dx*dT/dt=U*p*(To-T)*dx
dx/dt=v fluid velocity, so
the equation becomes
rho*A*c*v*dT/dx=U*p*(To-T)
and the solution
T0-T=(T0-Ti)e^(-Up/rho*A*c*v)x
at x=L, T=Tf and
T0-Tf=(T0-Ti)e^(-Up/rho*A*c*v)L
(T0-Tf)/(T0-Ti)=e^(-Up/rho*A*c*v)L
1-(T0-Tf)/(T0-Ti)=1-e^(-Up/rho*A*c*v)L
Tf-Ti=(T0-Ti)*[1-e^(-Up/rho*A*c*v)L} multiplying by rho*A*c*v, I get the heat transferred.

rho*A*c*v*(Tf-Ti)=rho*A*c*v*(T0-Ti)[1-e^(-Up/rho*A*c*v)L}

Now it's getting late here so seeing that U depends on v^(.8) among other variables, the exponential increases with v so that [1-e^(-Up/rho*A*c*v)L} decreases with v and the product
v*[1-e^(-Up/rho*A*c*v)L}.
This oversimplified result indicates a possible minimum, but maybe too simplified.

 

[BigInch]

It must be evaluated in both X and Y.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[zekeman]

"..so seeing that U depends on v^(.8) among other variables, the exponential increases with v so that [1-e^(-Up/rho*A*c*v)L} decreases with v and the product
v*[1-e^(-Up/rho*A*c*v)L}.
This oversimplified result indicates a possible minimum, but maybe too simplified."

Correction:

Should read
....... so seeing that U depends on v^(.8) among other variables, the exponent term increases with v so that [1-e^(-Up/rho*A*c*v)L} decreases with v and the product,
v*[1-e^(-Up/rho*A*c*v)L}, indicates a possible maximum.So maybe the OP is onto something here, but not conduction IMO.
Plotting or calculus could resolve this, but keep in mind it is very oversimplified, since, for example, dependence on viscosity and the Prandtl number is ignored and turbulent flow is assumed.

 

[pwildfire]

Ok, so if we can agree that surface effects dominate heat transfer in this situation, then the question becomes how accurately can these effects be represented using heat transfer calculations?

One little point, compositepro:
I am not suggesting 'adding' anything, what I am interested in is the effect of 'replacing'. Applying a coating to an aluminum surface precludes the formation of a surface oxide, therefore heat transfer at the surface will be dependent on a different thermal conductivity (k). For example, k for AlO2 is around 26 W/m.K, whereas copper plating would be ~380 W/m.K.

The original question of which I still am not sure is what the overall effect of this would be. In terms of basic conduction consideration, a layer of minimal thickness would have minimal effect, but if that layer occurs at the critical interface between metal and fluid, is there a greater effect in heat transfer to the film due to this increased thermal conductivity?

Additionally, basic surface convection equations include the material conductivity (k), but is that variable k in reference to the bulk material, or the surface material? In other words, is the effect of k due to the uniformity of surface temperature throughout the part, or is the effect on local transfer at the interface to the fluid? If the latter, then a surface coating should potentially make a large difference in overall heat transfer. In performing this calculation on a coated part, does one normally use the value for k of the coating, or that of the part?

 

[Twoballcane]

“In terms of basic conduction consideration, a layer of minimal thickness would have minimal effect, but if that layer occurs at the critical interface between metal and fluid, is there a greater effect in heat transfer to the film due to this increased thermal conductivity?”

Yes. Higher K means lower thermal resistance which gives a smaller delta T. This is heat transfer 101.

“basic surface convection equations include the material conductivity (k), but is that variable k in reference to the bulk material, or the surface material?”

In convection, K is the fluid’s thermal conductivity and delta T is from the surface of the material to the cooling fluid which is on the other side of the convection film from surface. Why would you think it would be the solid material if you are calculating across the convection film which is fluid?

Please tell us if you are a designer or Mechanical Engineer so we can either speak in equations or more in terminology.




Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

 

[IRstuff]

Again, do the darn math. 10 um of Al2O3 works out to 2600000W/m^2-K, that's about 0.4 microdegrees per W/m^2, i.e., it's essentially irrelevant, which is why surface passivations are often ignored in HT calculations


TTFN
faq731-376

 

[BigInch]

Conduction is only involved when getting the heat to the fluid interface. It is many, many times greater than convection. A corrosion layer a 1/2 mm thick won't make any difference when compared to the low heat transfer convection process going on just on the other side of it. Farting around with those equations you have will not get you anywhere. You must evaluate the convection process.

If it ain't broke, don't fix it. If it's not safe ... make it that way.

 

[nothingsimple]

One question, i know this is an old thread, but what is the component you are interested in. An aluminium piston for instance in a diesel engine is often cooled by forced oil flow, but the heat tranfer model in the gallery of a piston is much more complex than a static component.