Does dimpling intake ports work? (No)

[James363]

Dave you're bang on in your understanding of why the dimples help a golf ball. Essentially anywhere the flow stream separates from the body, dimples can be used to prevent/reduce this, so long as the flow isn't going too fast. So round the valve guides/stems, and as the flow turns round the short edge of the port on the bottom, it separates as it goes into the cylinder, and these were my target areas.

Basically you're also right in that the pressure drag reductions don't make up for the skin friction penalty of putting dimples in, although it was nice to have tested it in CFD to understand what was going on. I'll definitely give that thread a read in tea break too! Cheers

 

[bozothenutter]

I found those post from another post on here I think.....awesome info in there!

 

[bozothenutter]

 

[Kev@KAM]

ha ha looking at facebook today I wondered if anyone would find this thread

 

[bloke]

didn't know they made beryllium seats? heard a story a while ago about a group of beryllium miners from borkt who vanished and rumour was they were sold as food to a restaurant!! (on the internet so must be true)

 

[James363]

Yeah, I saw that. Just offered to send the chap my paperwork. But from my research this is the exact treatment that is a waste of time, as you don't need dimples all along the port. Unfortunately this is usually done alongside cams and a load of other bits and becomes very hard to verify practically unless properly tested, back to back.

 

[R3k1355]

It makes for pretty Instagram and Facebook posts though, which might help grow his business going forward.

Maybe the customer asked for this work specifically though?

 

[tomsdubs]

Shame you didn't get a head on a flow bench pre/post, empirical evidence is the real test. I did some head design stuff at uni but we had a full motorsport workshop. Could you tell if it had any effects on tumble and swirl? Ultimately the end goal is the best achievable volumetric efficiency from a wide range of variables, there's a reason the engine manufacturers leave intakes as cast. Some of these specialists seem obsessed with shiny surfaces, it's like jewellery for petrol heads.

 

[James363]

Shame you didn't get a head on a flow bench pre/post, empirical evidence is the real test. I did some head design stuff at uni but we had a full motorsport workshop. Could you tell if it had any effects on tumble and swirl? Ultimately the end goal is the best achievable volumetric efficiency from a wide range of variables, there's a reason the engine manufacturers leave intakes as cast. Some of these specialists seem obsessed with shiny surfaces, it's like jewellery for petrol heads.

The problem with flow benches is that the flow doesn't move as quickly as it does through the ports when you run the engine. The only real way to test dimple performance is by testing the engine on the dyno. The issue is, if you can't show promise with CFD, very few people will give you the funding to try a dimpled head back to back on an engine dyno. I can pretty conclusively say it does very little for swirl (and tumble) since the effect is limited to the boundary layer of the fluid, and doesn't affect the wider flow characteristics. I can also pretty conclusively say that putting dimples all over the surface of the port is definitely detrimental at full engine flow speeds, though if you slow the flow speed down to produce reynolds numbers in the region of 10^5, the port actually flows better. This means that using a flow bench for this type of work would be misleading.

My research showed best surface finish for intake ports to be around 1.2Ra, which we have experimentally tested at work and showed this to be optimum, however in the real world this was within the error of our engine dyno. This is somewhere between a smooth machined finish and a machined finish with some light marks in it (for the uninitiated). The reason manufacturers leave intakes as cast is pure price. A lot of modern manufacturers are now machining their ports, especially high performance and bike stuff. If you work with racing intake ports they are always machined to a smooth finish for this reason. Cast ports just do not flow as well as smoother finishes.

The dissertation itself goes into a lot more detail about the theory behind dimples. In hindsight, all of the established knowledge behind dimpled aero surfaces will tell you that dimpling an intake port wont work due to the flow speed through the intake ports.

 

[Darren S]

Apologies for my ignorance - but if you went completely the other way with the intakes and made them super smooth and evenly polished to micron depths - would that make next to jack-all difference either?

All interesting stuff, this!

 

[JamesBryan]

Apologies for my ignorance - but if you went completely the other way with the intakes and made them super smooth and evenly polished to micron depths - would that make next to jack-all difference either?

All interesting stuff, this!

Fuel would stick to the walls instead of atomising properly is my understanding.

 

[R3k1355]

but only on older engines, as most stuff for the last 10-15 years is direct injection

 

[JamesBryan]

but only on older engines, as most stuff for the last 10-15 years is direct injection

Older, as in the f4r ?

 

[James363]

but only on older engines, as most stuff for the last 10-15 years is direct injection

Most stuff in the last 10-15 years is port injection, direct gasoline injection is a relatively new technology. I would say it's only been appearing out on the roads in the last 4-5 years really.

If you make the walls very smooth indeed, the lack of energy in the boundary layer means the flow does not want to stay attached round the inside of bends, and you get extra pressure drag.

 

[Touring_Rob]

For some reason I always had it in my head that dimples were intended to increase turbulence near the boundary layer (not to increase flow) in order to reduce the amount of fuel wetting on port walls? Otherwise people would dimple the exhaust port too....

 

[Touring_Rob]

Most stuff in the last 10-15 years is port injection, direct gasoline injection is a relatively new technology. I would say it's only been appearing out on the roads in the last 4-5 years really.

If you make the walls very smooth indeed, the lack of energy in the boundary layer means the flow does not want to stay attached round the inside of bends, and you get extra pressure drag.

Agree that most things will be port injection but direct injection has been a 'thing' for many common cars for over 10 years now.

Jag XKR (etc) have used it for over 10 years, BMW 135/335 (2006?), RS4's etcetc. They are all getting on a bit now!

 

[Touring_Rob]

No, the petrol/air interface in CFD would have put my simulation run times up from about 8 hours to about 250, and I'd have needed an extra 100GB of RAM or so. In a carburetted engine like the one I studied, it would have been good to do, but since it's largely irrelevant now days, with port injection going out of date, it isn't really all that necessary. Also, at 250m/s, the fuel doesn't really run down the side of the port as some people seem to imagine, more blast through the port without interacting specifically with the port walls.
There is an SAE technical paper somewhere I found that compares a simulation run with a simulated intake charge and one run without. The gist of it was that the two were interchangeable for flow rate optimisation, so long as the Reynolds number variance within the port was kept below a certain margin.

Very interesting topic indeed mate!

I am not convinced that a negligible amount of fuel is deposited on port walls. When closing the throttle on most any car you will notice the afr (if your measurement equipment is fast enough) goes rich during the transient. This is exaggerated when mapping something with transient fueling disabled and to me points at fuel which is on the port walls being used.

I have done a lot of simulation in other totally unrelated fields and have to say that I tend not to trust simulation exclusively at all other than for a general overview of the situation - I say this because the time it takes to make an accurate model is just mental, not including variables in order to speed the process often means simplifying the model leaving you blind to some of the (perhaps minor) interesting points - as they say, s**t in = s**t out.

 

[R3k1355]

Jag XKR (etc) have used it for over 10 years, BMW 135/335 (2006?), RS4's etcetc. They are all getting on a bit now!

Is the VAG TFSI engine not direct injection? That engine has been around since the days of the Mk5 Golf Gti.

 

[James363]

Yeah, there have been a number of direct injection engines around before then. In terms of port injection becoming less common though, I would say that is more recent than 10-15 years ago.
We are currently preparing some tests with really rough surfaced ports which we will run on our dyno. Not dimpled as such, but just rougher machining, to see what the effects are, and how we have to jet the carbs to compensate. In our previous validation we only tested in the region of 3.2Ra to 0.8Ra, so it will be interesting to see the numbers from outside this range. We're also going to look at some different valve seat angles and venturi shaped ports in CFD with a view to further testing. So that's the latest excitement!

 

[Touring_Rob]

I know I said about that simplifying models can cause headaches - however could you simplify your model to something like a flat surface with one or two dimples and then look at wetting without crazy run times?

Is the objective for you to improve flow or are you interested in port wetting etc too?

 

[Touring_Rob]

Ever done any reading into somender-singh turbulence grooves? I think they could be an interesting area of research and would work geometry wise with the XK type head I believe.

 

[James363]

I know I said about that simplifying models can cause headaches - however could you simplify your model to something like a flat surface with one or two dimples and then look at wetting without crazy run times?

Is the objective for you to improve flow or are you interested in port wetting etc too?

That is quite a good idea. I'd be surprised if there isn't already some literature to check my findings against too. I'll have to check out the somender-sing turbulence grooves too, I must admit I've not heard of them before!