AxelWal

Ciao,

I am wondering if there is some brain here that would help me understand this pickle. I am by no means an fluid dynamics expert, and I am trying for fun to simulate in Ansys Fluent, an air duct for example for a brake cooler for my car. I have studied where is a good place to get this air from, and I have an general estimate of the air-pressure and -speed in this region. Let's say 500-800 pascals over the ambient pressure of 101235, and velocity of around 20m/s. The inlet is an rectangleish shape (a GV lip opening) with an area of 6600mm2 and it shrinks down to fit a pipe described later. This is important as the inlets cross-sectional area always remains the same and is larger than the pipes cross-sectional area.

Now, in Fluent I have set up my my inlets and outlets followingly: Forcing an constant air velocity and pressure on the inlet, and leaving the outlet as a pressure outlet.

On the wall I have added an virtual "sand grain roughness" of 5mm in height to try to simulate an corrugated hose and the turbulence it causes on the airflow.

To get to the issue: I tried simulating a 2inch and a 3inch pipes with the same path of pipe and the same inlet on both, and the resulting mass flow on the pipe outlet barely changes. What does change is the velocity of the air, obviously being faster in the smaller pipe, and the pressure being smaller in the smaller pipe. All is well and according to the laws of whatever dynamics.

Now, to me this makes perfect sense, and at the same time does not. When you read about this kind of ducting online, everyone keeps saying that the bigger the pipe, the better your cooling performance, but if the mass flow really remains the same with only velocity of the air changing, why is the bigger better?

Am I a complete retard, or should there be obviously something wrong in my simulation? Do you think I am somehow forcing a constant mass-flow with my boundary conditions and I am therefore blind to some obvious drop in mass-flow when going from the same inlet to a smaller pipe, or what am I not understanding here? Or is there something bad about having a higher velocity and lower pressure air being pushed to the root of the brake disk?

I don't want anyone to do this simulation for me, but I would apreciate any opinions on what I might be doing wrong or thinking wrong.

If you think some more details on the simulation is needed to understand my trainwreck of thought, I can post a more comprehensive description.

Bip bop

DaWaN

I am by no means an expert in this field, but I think you probably need to model the air pressure in the wheel/brake disc area as well to get an idea of the pressure differential across the pipe.
The less the pressure differential across the pipe, the more the diameter of the pipe contributes to the mass-flow I think.
Or stated the other way around: with a bigger pressure differential, the smaller pipe does not hurt the mass-flow as much.

AxelWal

Hmm, so I was talking with an aerospace guy from work and he seems to agree that I should model the air volume at the brake end instead of just using the end of the pipe as an outlet. I wonder if something similar should be done at the inlet end also.

Now what still kinda beats me is understanding why would the smaller pipe be worse in this brake duct case. I also work closely with some guys designing a two-phase CO2 cooling system, and they are also always very pushy on using as large transfer lines as possible to reduce pressure drop in the loop. Makes sense as smaller pipe introduces more internal resistance (larger pipe wall area compared to flow volume) to the fluid and causes a larger pressure drop over the same distance. This kinda makes sense as if the pressure drops too much there is a risk of triggering the gassing uncontrollably and prematurely.

But in the brake duct case, is the increased pressure drop/resistance in a smaller pipe responsible somehow on reducing the total mass-flow through that section? If so I guess it should be possible to see this phenomen by adding this air volume that you mentioned at the end of the line.

About this I am not sure if I understand, I think I need to talk to the guy again

Turbomack

Said a different way, the headloss in the piping (after it enters the piping and right before it exits the piping) would be insignificant just as the model shows for the diameters and friction loss you are assuming. If you study and understand the Darcy-Weisbach equation, your results will make sense. Yes, the lower the pressure at the exit (disc area), the greater the air flow.

Turbomack

I vaguely recall coming across some inexpensive temperature and pressure sensors that you could mount near the brakes and get that data point for your boundary condition. My son is heading off to college for an aeronautical engineering program in the fall and we may do a summer project like this as well.

Midtenn

iTrader: (15)

From some reading I've done on brake cooling recently, the even the high in reinforced flexible tubing sucks for air flow. It's likely that your assumption of surface texture is just not enough to match real world turbulence of the flexible tubing. That is likely why most places say bigger is better to hopefully get more "dirty" flow than anything. I'm hoping at some point to come up with more a air guide/flap design for my car that's similar to what Porsche or the GT350s have.

AxelWal

Thanks for the pointer to Darcy-Weisbach equation, though meant for incompressible flows it helps a lot with the understanding of the phenomenom. I have been told that at flow speeds over 30m/s air should be considered as compressible in Ansys, but not sure how much of an impact it has from this equation point of view.

I think this might have something to do with why I can see very little difference in the total mass-flow in my simulation. After going a bit deeper I understood that my choice of using standard k-epsilon model is not good for this type of problem, its rather just to discover basic flow patterns and does not do a good job of predicting near wall effects. Apparently using SST k-omega should be the proper model for this type of wall bounded high velocity simulation.

I was also planning to create this evening some simplified straight tube version of the case just to play around a bit with the "wall roughness" factor to understand how it behaves before trying to see any difference between the 2 inch and 3 inch pipes in the bent configuration.

AxelWal

So, couple of quick sims later.

Wall roughness seems to function as expected, turbulence near walls increases considerably when 5mm roughness is applied, however is it corresponding to real brake hose is a different thing.. Another interesting thing I see is an increase in pressure at the inlet in the model with wall roughness applied. However almost no difference in mass-flow once again... I am baffled and definitely not understanding something.


Simple mesh

No wall roughness

5mm wall roughness

No roughness

5mm wall roughness

AxelWal

Does the increase in air-pressure at the inlet decrease the amount of air that ultimately would end up in the duct to begin with? I think I will do some 2 inch simulations with this setup and after add an air volume in front of the tube and see if this reveals something

AxelWal

Well, I do feel like a first grader now. I talked with some more people and I came in the realisation that it is pretty obvious why the larger pipe is better. Funny thing is that two young aerospace engineer students and a mechanical engineer were not able to answer this due to being too focused on the equations, and a older physicist just stared at us and figured it out in a blink.

Like I was wondering before it has all to do about the internal resistance of the flow (aka. viscosity). When you have a smaller pipe you very simply need more energy to push the fluid through the same length of pipe when compared to a larger pipe. In the car case this energy would be stored in the pressure and the velocity of the air volume surrounding the inlet.

In other words: A setup with a smaller pipe causing a higher internal resistance to the flow would need higher pressure or velocity (energy) at the inlet to push air through the section with equal massflow as a similar setup with a larger pipe. This leads me to conclude that my boundary conditions on the simulation were acting more like an pump than a real scenario (where the pump just simply uses more energy to keep a constant mass flow through the section).

This also explains why the usual corrugated hoses are quite bad for the flow, as the corrugation greatly increases the resistance by requiring more energy as it converts more of the energy stored in the temperature and pressure of the flow to kinetic energy (aka turbulence).

This is also very easy to visualise with a hydraulic piston scenario (duh):