Exhaust Flow Modelling

[purplemonkey]

Hello all,

I am not very familiar with CFD however I have been using CFDesign in order to verify some design data on exhaust flow characteristics and pressures.

The system being modelled is from a single turbo back through to a splitter box which splits the flow into two pipes before continuing through the system through some complicated piping and routed to the atmosphere.

The engine manufacturer recommends no more than 3 in-Hg of back pressure right after the turbo outlet, however the analysis I am running is showing upwards of 5 in-Hg. The complexity of the system is very great and I would like to alter my design and recheck computationally without having to wait another 5 days for a solution to be formed.

My question is:

In order to simplify the complexity of the system and reduce the number of cells, would it be appropriate to assume that the exhaust is routed to atmosphere at the outlets of the splitter box connection (since this is the device I am most concerned about)?

Your input/advice is greatly appreciated!

 

[MikeHalloran]

No. You have to model the whole system.

However, you don't need CFD to do it, or at least not _all_ of it.

ASHRAE publishes a (thick and expensive) handbook that gives coefficients for sizing ductwork in most imaginable geometries. You can probably find pieces of those tables of coefficients at sites of people who supply them, and you can probably find a spreadsheet or two that can use the the coefficients to model the system segment by segment.

Engine manufacturers also publish application manuals that typically include at least a small selection of the ASHRAE numbers, too.

The point is, you can scrounge up a set of tools that are good enough to model the piping exclusive of the splitter, without spending a fortune on CFD.

Assume that the flow is isothermal; i.e., use the turbo exit conditions for the whole pipe. That's slightly conservative.

Most exhaust systems get larger as they get farther away from the turbo, and almost all include a short radius ell right at the turbo. The biggest single improvement you can usually make is to upsize that ell, even if it means that your turbo flange has a step inside it.

Also include at least one 1/4" NPT pyrometer/pressure port in your turbo flange. Engine manufacturers mostly specify backpressure limits measured there, but turbo manufacturers never include a port for that purpose. Beats me.

The splitter is problematic. First you have to model the pipes downstream of it. Use an arbitrary flow, say half the engine flow, and find the pressure drop for each pipe individually. Then compute a 'resistance' for each pipe in order to evaluate how they will affect the flow.

Kirchoff's and Ohm's laws both apply for pipes for liquids and low pressure gas, except for that pesky square law resistance stuff. I.e. for liquids, the actual units for Cv are gpm/sqrt(psid). Similarly for gas.

ASHRAE does include _some_ splitter geometries, but maybe not yours, and the tables are not easily found elsewhere. Maybe you can use the CFD just for that, and spreadsheets for the remainder and the summation.







Mike Halloran
Pembroke Pines, FL, USA

 

[zdas04]

One point that Mike hinted at that can't be overstated--flow in the duct is NOT incompressible, so all of the fluid-flow correlations that started with Bernoulli (basically all of them) don't work very well. The stuff in most CFD models is based on incompressible flow and the expensive results are mostly worthless. The ASHRAE correlations are based on real flows and have a much greater possibility of predicting performance.

David

 

[stanier]

If it takes 5 days to run a model that may produce results that are dubious why not consider building a 12" to the foot scale model and test the system?

 

[purplemonkey]

Thanks for the responses guys. I will look into the ASHREA readings to see if I can find anything of value to apply to my system.

As for building a scaled model, the company I work for unfortunetly will never go to that extreme to prove a pressure differential. As silly as it may seem, they are opting to build the exhaust system as a prototype and do testing on it first before making any changes.

Using CFD is something I initiated on my own time in order to see how well the original design worked.

Thanks again for the help!

 

[MikeHalloran]

You already know enough to suggest upsizing the prototype where possible. At least suggest that.

I worked on exhaust systems for a little over 450 different boats. Exactly one of them had an exhaust component that was too big.



Mike Halloran
Pembroke Pines, FL, USA