Expansion/contraction rates of differing metals.

[DerwenArtos12]

I'm currently working on a chilled liquid loop that shouldn't reach lower than -25C. I am however worried about the available waterblock choices. I'm not comfortable using one with a plexi top as it's likely to crack that cold, delrin/acetal isn't out of the question but, not my favorite choice here. I've essentially narrowed it down to two, the thermalright xwb-1 and the XSPC Delta. Here is where the HT part comes in. The thermalright is two copper blocks with a steel shim in between and held together with steel screws, the Delta is a copper block and a chrome plated brass block pressed together to create a hermetic seal. How different are the expansion and contraction rates for copper vs steel? What effect if any does chroming have on the expansion rates of brass on the delta? I know the expansion rates for copper and brass are really close, so I'm not too worried about breaking the seal because if anything, the brass will contract slightly more than the copper unless the chrome affects it to any measurable degree.

Any ideas?

 

[PlasmaBomb]

Some of the engineer guys are probably better qualified to help.

Wiki should have the desired info here - http://en.wikipedia.org/wiki/C...t_of_thermal_expansion
Steel alpha = 11~13 x10^-6/K beta= 33.0 ~ 39.0 x10^-6/K
Stainless steel alpha = 17.3 x10^-6/K beta = 51.9 x10^-6/K
Copper alpha = 17 x10^-6/K beta = 51 x10^-6/K
Brass alpha = 19 x10^-6/K beta = 57 x10^-6/K

alpha = linear coefficient
beta = volumetric coefficient

Using linear calculations - assuming a 1mm thick shim (steel not stainless). The shim would be 0.495-0.585 microns thinner @ -25 than at 20C.

So I will stick with some hopefully helpful practical advice, I take it you are worried about leaks due to thermal contraction? Why not stick it in a freezer (-18C) leave it for a while to equilibrate. Then quickly test the screws and tighten them up. Then hook up the block to the system and give it a good leak test outside the case, with no load this is as cold as the system will get (again check screws). You probably have though about insulating against condensation, but just in case you haven't please do.

Edit: Chrome plating should be thin (it is likely microns thick) in comparison to the thickness of the brass.

 

[DerwenArtos12]

Originally posted by: PlasmaBomb
Some of the engineer guys are probably better qualified to help.

Wiki should have the desired info here - http://en.wikipedia.org/wiki/C...t_of_thermal_expansion
Steel alpha = 11~13 x10^-6/K beta= 33.0 ~ 39.0 x10^-6/K
Stainless steel alpha = 17.3 x10^-6/K beta = 51.9 x10^-6/K
Copper alpha = 17 x10^-6/K beta = 51 x10^-6/K
Brass alpha = 19 x10^-6/K beta = 57 x10^-6/K

alpha = linear coefficient
beta = volumetric coefficient

Using linear calculations - assuming a 1mm thick shim (steel not stainless). The shim would be 0.495-0.585 microns thinner @ -25 than at 20C.

So I will stick with some hopefully helpful practical advice, I take it you are worried about leaks due to thermal contraction? Why not stick it in a freezer (-18C) leave it for a while to equilibrate. Then quickly test the screws and tighten them up. Then hook up the block to the system and give it a good leak test outside the case, with no load this is as cold as the system will get (again check screws). You probably have though about insulating against condensation, but just in case you haven't please do.

Edit: Chrome plating should be thin (it is likely microns thick) in comparison to the thickness of the brass.

Ok, that's pretty much the line of thought I was on, just not too familiar with chrome plating and whether or not it would bee too effectual. I did get an e-mail back from XSPC saying that the block should be fine to -25C but, he's checking with one of the engineers that designed it and getting back to me. No word from the other manufacturers.

I'm pretty comfortable in getting either at this point. I finally found a site that had pictures of the thermalright dis-assembled and it seems like a small bead of RTV in place of the o-ring and between the top block and mid-plate should keep me quite safe.

Thanks for the reply. I was familiar with steels expansion from my day job(actually it's at night but, semantics) I just didn't have brass or copper. It's comforting that my thinking that brass will contract faster than the copper and just seal itself better so long as the brass doesn't crack on the XSPC block.

 

[wwswimming]

it's not just a matter of matching TCE's.

it's also a matter of matching the expansion of (for example) mated
components. 2 parallel plates made of aluminum with 2 press fit
pins will expand differently in some geometries - the plate nearer
the heat source will expand more than the piece near "thermal
ground" (ambient air).

the trick is to match the expansion of materials over the range of
temperatures they'll see in "normal operation".

you'll probably be completely OK. if you were doing 100,000
cycles and concerned about fatigue, then they pay us finite element
modellers to use tools like Patran (yuk) or ANSYS (yea !) to work
out the details.

 

[DerwenArtos12]

Originally posted by: wwswimming
it's not just a matter of matching TCE's.

it's also a matter of matching the expansion of (for example) mated
components. 2 parallel plates made of aluminum with 2 press fit
pins will expand differently in some geometries - the plate nearer
the heat source will expand more than the piece near "thermal
ground" (ambient air).

the trick is to match the expansion of materials over the range of
temperatures they'll see in "normal operation".

you'll probably be completely OK. if you were doing 100,000
cycles and concerned about fatigue, then they pay us finite element
modellers to use tools like Patran (yuk) or ANSYS (yea !) to work
out the details.

Some of my co-workers use ansys to make sure my models aren't going to tear themselves apart. I've been getting more and more curious as to what they actually do during the day while I'm not there. you may have inspired me to stop in during the day sometime and see if I can watch them work.

 

[wwswimming]

well, some of the plots can be quite pretty. in most post-processors
you have control over color assignment. combine this with an analyst
that displayed a warped sense of humor during the modelling phase,
and you can end up with some genuine infotainment.

some of the other eye candy they have is animating various modes
of resonance. it's sort of like watching a vibrating assembly with a
strobe, except that a lot of the deflections involved are tiny. so
deflection that would be real hard to see with a strobe on a shake
table, can become visible in the simulation if the deflection is
amplified, in the simulation.

usually these workstations have some pretty decent video cards.

 

[DerwenArtos12]

Originally posted by: wwswimming
well, some of the plots can be quite pretty. in most post-processors
you have control over color assignment. combine this with an analyst
that displayed a warped sense of humor during the modelling phase,
and you can end up with some genuine infotainment.

some of the other eye candy they have is animating various modes
of resonance. it's sort of like watching a vibrating assembly with a
strobe, except that a lot of the deflections involved are tiny. so
deflection that would be real hard to see with a strobe on a shake
table, can become visible in the simulation if the deflection is
amplified, in the simulation.

usually these workstations have some pretty decent video cards.

yeah, I really need to spend some time with these guys, they like my designs but, the bosses keep yelling at me for using too much stainless steel and titanium, they say to make it work so thats what I do, they can spend all their time trying to make it work cheaper, i just make it work.