heinzboehmer's 2002 Topaz 6MT Coupe

[bigjae46]

Also, this took absolutely forever:

I started with this piece cause it looked simpler than the cabin air filter housing, but, man, was I wrong. Essentially no surface on it is flat, except for the one that butts up against the firewall. Took me a really long time to figure out how to approach all of the curves.

Yes, I realize that a lot of the geometry is there for mold release, but I want to make something that fits all E46s, so I decided to spend the extra time and recreate it as closely as possible. More time spent in CAD land saves me from redesigning the piece down the road if it turns out to not fit on some other chassis configuration.

I also decided to not model the front clip retainer into the piece. I like the idea of using the stock injection molded piece for serviceability. BMW does not sell these separately, so I'm gonna have to go junkyard diving for some. Fortunately, they're just plastic welded in, so they're easy to pop out with a razor blade. This part could also conceivably be recreated, but printing it in a favorable orientation might be tricky.

Also, did some preliminary slicing of the model and the printed version should weigh in at 300-400 g, depending on my choice of material and infill. For comparison, the stock piece weighs in at 1618 g. I do need to split it up into three pieces to fit in my printer, so I'll add some features to bond all of them together nicely. For now, I just added some slicer-generated dowel connectors:

With the stock geometry recreated, I can work on modifying it to accommodate both the braces and the tilted filter. Got some ideas flying around in my mind already for how to approach that. More to come.

Sidenote, since I essentially made a 1:1 replica of the piece, anyone can print it out without modification and save ~1.2 kg. Who's up to be the test fit guinea pig? Hit me up and I'll send you the stl (I'll distribute widely once fitment is validated).

You're making a mold? Have to be careful with any vertical surfaces. That may cause release issues. Need a slight draft angle...maybe 1-2 degrees?

Have to factor in exotherm with the mold. Could see up to 200-300 degrees depending on the exact epoxy you use.

[heinzboehmer]

Love it!

Lol, that seems to be the story for all the doing projects I table on this car

You're making a mold? Have to be careful with any vertical surfaces. That may cause release issues. Need a slight draft angle...maybe 1-2 degrees?

Have to factor in exotherm with the mold. Could see up to 200-300 degrees depending on the exact epoxy you use.

Just gonna print the part, not a mold.

Rough math says that making it out of carbon + cork core will weigh around the same as the printed part, so not even gonna bother trying. The modern printable materials can take heat really well, so I see no downside to printing.

[discoelk]

Depending on the design, you're also getting infill with a 3DP part. So effectively, some open cell "insulation" is created.

[heinzboehmer]

Depending on the design, you're also getting infill with a 3DP part. So effectively, some open cell "insulation" is created.

Yep! That's the plan. I've been doing some research to see what the optimal infill config would be. At first glance, I would think a fairly high infill percentage would work better, as that would minimize convection in the small air pockets. But then you also get more thermal bridging from the infill walls, so would need to do testing to find the optimum balance.

However, I did have a thought this morning. "My" design is a 1:1 copy of the stock piece (i.e. same thickness), so if I print it out of a material with similar thermal conductivity to UP-GF, then any amount of air in the interior should result in a lower thermal conductivity.

Quick (unsubstantiated) search says that UP-GF and PET-CF have similar enough thermal properties, so I'll probably just choose an infill with a non-continuous volume (e.g. cubic) and print it with the lowest infill percentage that still results in a strong and dimensionally accurate part.

[Bry5on]

Yep! That's the plan. I've been doing some research to see what the optimal infill config would be. At first glance, I would think a fairly high infill percentage would work better, as that would minimize convection in the small air pockets. But then you also get more thermal bridging from the infill walls, so would need to do testing to find the optimum balance.

However, I did have a thought this morning. "My" design is a 1:1 copy of the stock piece (i.e. same thickness), so if I print it out of a material with similar thermal conductivity to UP-GF, then any amount of air in the interior should result in a lower thermal conductivity.

Quick (unsubstantiated) search says that UP-GF and PET-CF have similar enough thermal properties, so I'll probably just choose an infill with a non-continuous volume (e.g. cubic) and print it with the lowest infill percentage that still results in a strong and dimensionally accurate part.

Generally the best thermal impedance will come from the smallest air volumes and the lowest gravimetric density you can get away with, so you’re on the right track. I’d probably pick a reasonably high infill personally as the weight penalty isn’t that great. Cubic is a good choice as you definitely want closed cells. You should re-run the Slon test with a desk heater and an IR thermometer!

[heinzboehmer]

Generally the best thermal impedance will come from the smallest air volumes and the lowest gravimetric density you can get away with, so you’re on the right track. I’d probably pick a reasonably high infill personally as the weight penalty isn’t that great. Cubic is a good choice as you definitely want closed cells. You should re-run the Slon test with a desk heater and an IR thermometer!

Hmm, going from 10% to 75% cubic increases weight from 345 g to 746 g.

I know 75% is a lot, but it's right around what looks correct for a moderately closed cell insulator:



Stock is fully solid, so any air volume inside the part (assuming identical exterior dimensions) should increase thermal impedance, no?

[George Hill]

For reference here is 40%

[heinzboehmer]

Alright, some data. Car up to temp for a while, couple redline pulls then straight into the garage to measure temps:



It in the 60s today, so I'm guessing the firewall can easily reach 100 C on an actual hot day, once everything is nicely heatsoaked. PET-CF should be fine for this application. If I start seeing warping, it'll be an excellent excuse to try out PPA-CF.

[Bry5on]

Alright, some data. Car up to temp for a while, couple redline pulls then straight into the garage to measure temps:



It in the 60s today, so I'm guessing the firewall can easily reach 100 C on an actual hot day, once everything is nicely heatsoaked. PET-CF should be fine for this application. If I start seeing warping, it'll be an excellent excuse to try out PPA-CF.

The air in the print should increase thermal impedance vs stock, yes. But also CF will decrease it, so this is an application where fiberglass filament will beat carbon filament.

For temps I’d probably assume 150C max, which is a 50ish C premium over coolant/head temp to account for heat from the exhaust manifolds from heavy ripping straight to idling and heat soaking. Keep in mind your setup is effectively vented to atmosphere through the hood right now.

[heinzboehmer]

The air in the print should increase thermal impedance vs stock, yes. But also CF will decrease it, so this is an application where fiberglass filament will beat carbon filament.

For temps I’d probably assume 150C max, which is a 50ish C premium over coolant/head temp to account for heat from the exhaust manifolds from heavy ripping straight to idling and heat soaking. Keep in mind your setup is effectively vented to atmosphere through the hood right now.

Yeah good points all around.

But also, to the carbon point, higher infill means more thermal bridges. Sounds like there's an optimal point somewhere. Might need to do some testing to find it.

[bigjae46]

Hmm, going from 10% to 75% cubic increases weight from 345 g to 746 g.

I know 75% is a lot, but it's right around what looks correct for a moderately closed cell insulator:

Stock is fully solid, so any air volume inside the part (assuming identical exterior dimensions) should increase thermal impedance, no?

What is the total volume of the part and how does it compare to the carbon strut mount? You could get a pretty accurate approximation of what the weight would be in carbon. One option is to use a core material which can reduce the weight by up to 33%. Lantor soric is a flexible core material which would work in this application.

Couldn't you make the areas that don't support the cabin filter or seal against the firewall thinner? Use some foam if NHV might be an issue.

[George Hill]

We'll see what happens...

45min down, 17hrs to go, lol

[Bry5on]

What is the total volume of the part and how does it compare to the carbon strut mount? You could get a pretty accurate approximation of what the weight would be in carbon. One option is to use a core material which can reduce the weight by up to 33%. Lantor soric is a flexible core material which would work in this application.

Couldn't you make the areas that don't support the cabin filter or seal against the firewall thinner? Use some foam if NHV might be an issue.

We used cork as core in the Slon version to get the thermal insulation properties back in line with the factory part, for what it’s worth.

The varying thickness of this part makes a normal core material tough if you want to preserve factory fit like Heinz is targeting. You could always print a plastic shell and wrap it in carbon though, which would work. I just don’t see the benefit there though, I think this is actually a pretty good application for a 3D printer vs carbon (and my car has a carbon part here).

[heinzboehmer]

What is the total volume of the part and how does it compare to the carbon strut mount? You could get a pretty accurate approximation of what the weight would be in carbon. One option is to use a core material which can reduce the weight by up to 33%. Lantor soric is a flexible core material which would work in this application.

Couldn't you make the areas that don't support the cabin filter or seal against the firewall thinner? Use some foam if NHV might be an issue.

Volume is 702.9 cm3, but like Bryson said, there's varying thicknesses everywhere on this piece, so design would need to be modified quite a bit to make it suitable for carbon. Volume of the current design isn't really a good representation of what the volume of a carbon version would be.

Like you mentioned, it can be made a lot thinner, but then you start running into fitment issues. Slon part requires spacers for the fasteners, for instance.

We'll see what happens...

45min down, 17hrs to go, lol

Hell yeah! I'm excited to see how it turns out.

The varying thickness of this part makes a normal core material tough if you want to preserve factory fit like Heinz is targeting.

Yeah factory fit is the big thing. The aftermarket parts I've handled do not fit nearly as nicely as the BMW part does. I'd like to have my version fit as close to factory as possible.

[bigjae46]

We used cork as core in the Slon version to get the thermal insulation properties back in line with the factory part, for what it’s worth.

The varying thickness of this part makes a normal core material tough if you want to preserve factory fit like Heinz is targeting. You could always print a plastic shell and wrap it in carbon though, which would work. I just don’t see the benefit there though, I think this is actually a pretty good application for a 3D printer vs carbon (and my car has a carbon part here).

I was referring to varying the thickness of the 3D printed part. I believe the lower half doesn't have any role in supporting the cabin air housing. If making a carbon part, you could use a 2mm core in some places and a thicker 3 to 4 mm core in other spots.

Didn't even consider using cork for NVH reduction which I would think would do better than soric. I looked up the coefficient of thermal conductivity, cork is .036 to .065 W/m.K, Soric is .064. So thermal performance should be similar. Actually this is great info that I will eventually need for a different project.