cfd analysis

[himanshuagg1991]

hello ,
i am designing a restrictor (converging diverging nozzle) and i am fairly new to gambit and fluent .
Could anybody suggest me the approximate angles for the converging and the diverging section . It would narrow down my approach in a definite direction . Any other help also would be really appreciated

[himanshuagg1991]

hello ,
i am designing a restrictor (converging diverging nozzle) and i am fairly new to gambit and fluent .
Could anybody suggest me the approximate angles for the converging and the diverging section . It would narrow down my approach in a definite direction . Any other help also would be really appreciated

[billywight]

75 degrees, 0 degrees, 11 degrees. I recommend the 20mm at the 0 degree portion. Done.

[himanshuagg1991]

what do you mean by 0 degrees ? Should the throat be long as i was thinking about just rounding off the pointed edge connecting the converging and the diverging section .

[SamB]

The angles you can play with depend entirely on your manufacturing abilities. There is no point in simulating angles that you cannot make. Find this out first. If you are manufacturing a carbon part from an aluminium mold you should be fine, as I'm sure pretty much any angle you want can be made in the lathe. However, if you plan to machine one from solid, you need to make sure you can actually do this before embarking on any crazy simulations. Don't forget that your intake must also fit in the template - this can restrict the angles you can fit in so bear it in mind.

Secondly, what turbulence model are you using? What mesh structure are you on? What is your wall y+ values (great measure of the accuracy of your boundary layer)? As soon as you go into the design event they're going to really concentrate on the accuracy of your simulations. There's no point in just creating pages and pages of colourful fluid dynamics. The geometry is the easiest bit, its setting up an accurate simulation that's going to kill you...

Sam
Team Leader
Uni. of Manchester

[himanshuagg1991]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Originally posted by SamB:
The angles you can play with depend entirely on your manufacturing abilities. There is no point in simulating angles that you cannot make. Find this out first. If you are manufacturing a carbon part from an aluminium mold you should be fine, as I'm sure pretty much any angle you want can be made in the lathe. However, if you plan to machine one from solid, you need to make sure you can actually do this before embarking on any crazy simulations. Don't forget that your intake must also fit in the template - this can restrict the angles you can fit in so bear it in mind.

Secondly, what turbulence model are you using? What mesh structure are you on? What is your wall y+ values (great measure of the accuracy of your boundary layer)? As soon as you go into the design event they're going to really concentrate on the accuracy of your simulations. There's no point in just creating pages and pages of colourful fluid dynamics. The geometry is the easiest bit, its setting up an accurate simulation that's going to kill you...

Sam
Team Leader
Uni. of Manchester </div></BLOCKQUOTE>



Thank you so much for replying !!
turbulance model-density based
mesh structure - dense at the boundary of the nozzle , if this is what you mean
wall y+ values??
i really need some help on this!!!
---to make the joint of the converging and diverging section smooth what equation should i use
---as the restrictor will be placed between the carburetor and intake manifold ,, what conditions would you suggest

[Xeilos]

To be honest here, you should not be doing CFD and relying on the results without any kind of background knowledge in fluid and computational fluids.

Over the last couple of years, I have seen an increase in young engineers and their dependence on the be all and end all tool of CFD to produce designs. What many fail to understand is the limitations of the capabilities of CFD and when you select one option over another and what is actually happening 'in the background'. In what case does a k-epsilon turbulence model provide benefit over the currently popular Menter's y-omega SST as an example? Most people just select a turbulence model (both of these are standard turbulence models) without understanding why, which is dangerous.

CFD has been referred to on this forum (and other places) as Conterfeit, Fraud and Deciet as many people simple take a plug and chug approach to this problem. The same issue lies with FEA and any simulation program the user takes a black box approach to it. I could spend 4 months doing a simulation and still only feel about 40% confident in my results. CFD-Online another great forum for advice, has been ruined in the last couple of years through questions of people essentially asking, "Solve this for me, I do not know what I am doing...".

You might as well save your time/effort and just make something that 'looks' right and be ahead in the long run. I am not saying this to discourage you or make you feel like you know nothing. I would not trust your results further than I could throw them and you have to ask yourself then, what is the point? Just run an orifice plate and save yourself alot of work if you have to.

There is my morning rant, now I must get back to work.

[SamB]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">
You might as well save your time/effort and just make something that 'looks' right and be ahead in the long run. I am not saying this to discourage you or make you feel like you know nothing. I would not trust your results further than I could throw them and you have to ask yourself then, what is the point? Just run an orifice plate and save yourself alot of work if you have to. </div></BLOCKQUOTE>

Lets agree to disagree.

With small teams who are intent on doing some CFD, I think you are best developing trends - increase the angles and we get flow separation, decrease them and we start to measure nozzle length in lightyears. I think the judges would be impressed that some sort of analysis has gone into it as long as you know your restrictions. I think we are both under no illusions that you cannot walk into the design tent and state that CFD says your engine will develop 89.647382 BHP and you have full confidence in this, that's just daft. What you can develop is an idea of flow and base designs which look 'about right' from that.

I think the detail you are talking about (i.e. SST over K-E) is far too detailed for this competition, and anyone who tells you otherwise is talking out of their a**e. Any 2 equation model should give sufficiently accurate results for trends to be established and I think switching between the variations is pointless.

himanshuagg1991: Read some introductory texts before you go anywhere near gambit. I'd suggest 'An introduction to computational fluid dynamics' by Versteeg first - its where I started and very user friendly. I do have to warn you though that to get anywhere near useable simulations is going to take in excess of three to four months, so only allocate resources to this if you have time to spare or if its for the following year.

Like I said before, if you can't manufacture it then you have just wasted your time.

Sam