Polycarbonate Tops... a scientiffic analysis!

[WinFlex]

The goal of this thread is to gather evidence regarding the claim that clear plastic tops (whatever shapes and materials they come in) are prone to cracking. This all started in the form of several PM's being sent back and forth between Billa and I. I will transcribe them here.
Recognizing and understanding the factors involved in the failiure of waterblock tops will help others avoid following in the footsteps of those poor few who have actually experienced a leak inside the computer.
If you have experienced a leak, please report it here along with details on the type of top you used... thank you, I hope this becomes an interesting discussion.
---
WinFlex wrote on 03-12-2003 08:17 AM:
uh. regarding that polypropylene thing, that was a typo... polycarb is what i was actually talking about ... sorry... i edited it as soon as i realized the mixup
it is true, however. many people use plexiglass which easily shatters or cracks. lexan is much, much more durable for it is not an acrylic resin... they even use 1/2 inch lexan for bullet proof windows in banks, etc
---
unregistered wrote on 03-12-2003 09:33 AM:
no problem, I assumed the error - just prodding for accuracy
one needs to focus on the specific properties of interest
which for wb tops with threaded connections is going to be the crack propagation
both matls are crappy, it is terribly difficult to 'cut' threads w/o microcracks
Regards, Bill
---
WinFlex wrote on 03-12-2003 09:19 AM:
very true. when I was tapping the threads for my latest waterblock (you may have seen it here or over at ocforums), I had to twist the ntp tap in all the way in order to get my polypropylene (now this time I meant to say this) fittings to bottom out nicely. However, I did not notice any kind of cracks forming... how would these microcracks become noticable? I tested my block for 24 hours and I observed absolutely no liquid leaking out... regarding crack propagation in polycarbonates (most plastics, actually), I think it takes more than a steady applied force to cause eventual failiure... the steady "push" from the barbs that are screwed in the lexan (polycarb) top certainly won't be enough to cause any cracks that were initiated druing tapping / fitting insertion to propagate...
what do you think?
---
post as a thread
black light and magnification (plus refractive oil ?)
google on crack tip energy
tightening the threaded connection imposes a radial stress
short term is NOT long term + moisture + heat
you have it all wrong, as do most mfgrs
Bill
---
I am aware that ultraviolet electromagnetic radiation (aka blacklight) has deprimental effects on both structural and optical properties of polycarbonates. I am unclear on what you mean by refractive oil and magnification...
Certainly something worth looking into... an interesting topic. Do you have any case studies at hand that would describe catastrophic failiure of a 0.5 inch or thicker lexan top? If so, please describe them or list some names of the people it happened to so I can contact them.
As for trasforming this PM conversation into a thread, I will start one right now... We have certainly hit a topic here that should be investigated on a scietiffic level, not a "you have it all wrong" level. Claims should be backed with solid evidence in the form of case studies, examples, calculations, or other experimental results.
By the way, english is my second language, so please explain what the acronym "mfgrs" stands for.
I appreciate your time, let's make this a good, interesting thread to read...
- Felix
---

[bigben2k]

I've posted it a few times before, but here's a good read on those cracks, including a fix (for polycarbonate only).

http://www.extrication.com/polycarb.htm

(I never tested it though)

[MadDogMe]

"mfgrs" = Manufacturers ...

Personaly I love Lexan, not just for the visual glory of it!, or being able to check at a glance, but the weightlessness of it when compared to a top made of Cu thick enough to tap a thread for barbs into or to be 'flexproof'. I just want to find a 100% secure way of setting/bonding Cu tube into it to be used instead of threaded barbs. barbs suck, I don't know why nearly 99% of people use them!, 1/2inch barbs are very restrictive when used with 1/2inch tube!. I prefer to stretch the tube over 13mm Cu pipeing...

[BillA]

the blacklight (UV) was wrt using it as an inspection technique
and the refractive oil to 'clear' a machined surface - though if it penetrates also the crack then such would not be visible

good tops can be made, but it requires a great deal of care
-> and over tightning a barb renders the best threading worthless

MDM
think Araldite, amine cured
roughen both bonding surfaces with coarse sandpaper
DO NOT TOUCH surfaces prior to bonding (oil from fingers)

[OTMOPO3OK]

MadDogMe,
i'm with you 100% on this one...
i hate barbs...
first of all they cost like 1.5$ each then u have to make a tread (which is PITA since the diameter is big) and after all that they have such a small ID....i use copper tubing all the way...cheap and the ID is almost the same as OD cos the walls are very thin....
just solder and off u go...
and yes 99% of ppl use them.....:shrug: hmm...ppl are not thinking themselves they just see a design and blindly follow it....

sorry to get away from the topic....needed to express on this one....

[wymjym]

I guess it is time for the world of water cooling to wake up to different types of ‘barbed’ fittings. Ones that have standard threads rather than pipe threads, ones that use nuts, ones that use nuts with captured O rings as part of them, ones that do cost a lot more than the hardware store variety.
wj

[pokpok]

taken from bb2k's link

"Incidentally, there is a way to minimize the effect of such a cut. A wipe across the raw edge with methylene chloride will re-melt and smooth the micro-fissures which propagate the fracture. We also found that the windows which became distorted as a result of our crash tests had become fracturable (after the test) as well. Some people choose not to have cars repaired immediately....sometimes not ever. A grossly distorted polycarbonate window would not carry the same safety advantages of a non-stressed window. A point to ponder."

the methylyne chloride will melt the microfissures back into the plexi. would you have to use methylyne chloride to make it melt, or could you use some other melting type of liquid...lets say paint thinner?

[WinFlex]

There is absolutely nothing wrong with using NTP fittings. Tapered threads help seal, and as long as they are properly installed using teflon tape and are not overtorqued when tightened up, there really is no risk of damaging the block... proper tappnig procedure is also vital when working with polycarbs. Displacement taps are not recomended while taps that cut material out (the most common type) are the way to go...

[MadDogMe]

I'd prefer to use a liquid silcone sealant instead of PTFE tape with tapered barbs, just slap it on and tighten up 'hand tight' then let it dry. With PTFE I tighten it up to hand tight then give it 1/2 ~ 3/4 of a turn with a spanner, no leaks yet ...

[bigben2k]

Originally posted by pokpok
the methylyne chloride will melt the microfissures back into the plexi. would you have to use methylyne chloride to make it melt, or could you use some other melting type of liquid...lets say paint thinner?

My question is: couldn't you just use a heating process, like a microtorch?

Otherwise, NPTs (american, they have a taper in the thread) are just fine: torquing it is the answer, but torque to what?

A thread sealer would take care of the rest. See Loctite.

[pokpok]

wouldnt it melt real quick if used the micro torch though?

[BillA]

ok, apparently a short primer on threaded connections is in order

tapered threads, of which the NPT is an example, when made up will impart a radial compressive stress on the pin and a tensile stress on the box end (or top in our application)
- note that "V" threads are utilized which seal by (plastic) thread deformation of the thread itself

-> so lesson #1 is that if the materials will not deform plastically, but fail instead by brittle fracture; then these are NOT well suited for this thread form

in an effort to enable the sealing to occur at lower stress levels several types of 'thread sealants' are commonly used; pipe 'dopes' of various types, Teflon tape, and conventional adhesives
- note that, unless an adhesive is used, some amount of connection compression is still required for sealing - which will have a radial tensile component in the box end

so why then are NPT connections utilized ?
well, IF a threaded connection is desired, in this country they are fairly cheap to make and fairly reliable
unless:
- there are repeated make-and-breaks
- the components fail by brittle fracture
- a metal pin end is put into a plastic box

are the reasons for the continuing failures becoming apparent ?

there are other types of threaded connections which are much more suitable for plastics, and European mfgrs use them routinely:
parallel coarse (non-sealing) threads and a captured o-ring under the integral flange nut

and of course the best solution is to eliminate the threaded connection in any form,
an example of such is the MCW5000 with integral push-in connections

so for those WCers stuck with cheap poorly designed US made crap, some suggestions:
NEVER use a brass barb in a plastic top
do NOT use Teflon tape with plastic tops, use pipe dope or, better, an adhesive sealant (difficult with pp or pe)

if reliability is a paramount issue, go with a metal top
(but if Al don't forget the corrosion inhibitor)

Ben
in the interest of accuracy, why don’t you compose your posts
THEN go back and insert “NOT” in front of the whole thing ?
your accuracy would go up by 75%+ at least
- alternatively you could stop posting about things with which you have no hands-on experience

do you know what a plug gauge is ?
what will happen when you heat those threads ? (or solvent wipe them ?)

you persist in parroting ‘forum knowledge’, which is more often than not completely wrong
have you no pride ?
how many times do you have to get your nose stuck back into your ignorant statements
- give us a break, stick to what you have some actual knowledge of

[bigben2k]

Originally posted by unregistered
you persist in parroting ‘forum knowledge’, which is more often than not completely wrong
have you no pride ?
how many times do you have to get your nose stuck back into your ignorant statements
- give us a break, stick to what you have some actual knowledge of

Actually, I put an "?" at the end, otherwise I'm still (and always have been) of the opinion that a plastic barb (pp polypropylene, pe polyethylene or nylon) is the way to go, regardless of the material (metal or plastic top), so feel free to comment on using pp, pe, nylon with a metal top, but it'll be OT.

Otherwise, thanks for the run down: I'll make a note to use straight threads in a plastic top.

[BillA]

off-topic

I must say Ben, at least you're a good sport re the corrections
but godamn, ya got to be more critical of what you say

[WinFlex]

Ok. Very nice write up, BillA. Your claims are well backed and I completely agree that plastic tops should only be used with low durometer (soft) plastic type fittings. Contrary to a metal fitting, a plastic fitting can compress instead of excerting excessive radial expansional stresses in the block top. One more question, however: I do not completely understand why you disapprove of using teflon tape in plastic to plastic threaded NTP connections. I assume that your reasoning behind this statement is that you want to avoid increasing the outer diameter of the fitting as a whole, thereby reducing imposed radial stresses. However, have you ever noticed that good quiality teflon taped basically disintigrates when it is treaded in, filling voids while being displaced from areas of high preassure? I think this would negate the effect you are worried about.
Regarding putting each other down, we should go easy on that part. I know that there are individuals here who do have a more extensive technical background that others (namely BillA and I, for I believe that we have both had educations in the field of materials science and engineering physics). However, this does not justify to insult or put down any other party, let alone forbid them to speak. In many intelligent people, common sense is a powerful tool and I think that anyones views should be heard out as longs as there is proper reasoning behind them!
Keep it real, all Also, this is JUST the forums, ok?

[bigben2k]

Originally posted by unregistered
off-topic

I must say Ben, at least you're a good sport re the corrections
but godamn, ya got to be more critical of what you say

(NOT) Noted, thank you!

So what's a "plug gauge"? Oh, I see! (said the blind man, while holding a stick of asparagus). That seems like a reasonable thing to do: it would help to prevent the barb from protruding inside the block, or in the case of parallel thread, would allow one to put a stopper (aka stop the thread in the top) at the right spot, right?

be kewl

[Cova]

I don't suppose anyone has ever come across polycarbonate barbs? I'm thinking that if the micro-cracks can be re-sealed by "melting" them back together after cutting the threads, then glueing the barbs to the top using a plastic welding type cement should have the same effect. Or, if such a cement exists that can weld polypropylene to polycarbonate, it should work just as well.

[BillA]

Teflon tape is convenient, dope is better - takes less force to seal
why the 'less force' is important has to do with the 'brittle failure' mechanism

one has to assume that micro-cracks will result from the threading operation in acrylic or polycarbonate
so we are concerned not with crack initiation, but rather propagation
the cracks will lengthen at MUCH lower (crack tip) energy levels than that required for initiation
hence radial (or hoop tensile) stresses should be kept to the lowest possible levels
- note that simple thermal cycling is often quite sufficient for crack propagation w/o ANY applied load at all
(a poor/unsuitable material is always going to be a poor material)

Ben is a special case (36,881 posts and counting)
BTW, thread gauging itself is almost an art form (I have designed SPC thread inspection systems)

EDIT: gone_fishin' had a link to a polycarbonate barb source
pretty slick - but why thread ?
drill a hole and bond-in a tube

[bigben2k]

(retracted)

[BillA]

just for Ben:

NOT

[mota]

You should avoid the use of lexan in water block construction, its just bad design. I know of at least one water additive (hy-per lube) that will attack lexan and cause it to crack. Will other additives react with lexan in this way? Probably.

[WinFlex]

Uh, ben, I don't think that's why we don't use teflon tape. If applied properly, in the right direction and using the proper amount, teflon tape will seal ANY fitting... that's the whole point why we have the stuff. Again, we are concerned about stresses that cause crack propagation, not leaks that are caused by water traveling along the thread! Sealants aren't supposed to bond, they are supposed to fill microvoids and thereby keep water from getting through
Cova, there are polycarb barbs, you can find them at www.mcmaster.com ... using a chloride based forula to fuse the two together would be nice... however as BillaA said, why not use polycarbonate tube that is inserted and fused in. Let me go find the proper mcmaster pages!
Also... I have done extensive studies on crack propagation in metals, and the point BillA made about temperature having a big impact is valid for metals. However, I hate to point out the fact that the minute (say + or - 15degrees K) temperature fluxuations in this application will not promote, speed up, slow down, or hinder any sort of crack propagation in polycarbonate plastics... the ductile to brittle transition temerature for polycarbonate lies far below any temperature your block will ever experience... a big NOT goes out to BillA hence. To proove me wrong, please provide me with a datasheet on Lexan specifying the exact DBTT and also how in the world any environment inside a computer could possibly reach this temperature.

OK. Edit: See page 114 in McMaster for polycarb host fittings!

[WinFlex]

http://www.bayplastics.co.uk/product...h-bonding3.htm

Nice little read on polycarb to polycarb bonding options...
also, I apologize to all those who don't understand the technical jargain BillA and I are throwing at each other

coefficient of linear thermal expansion (mm/m °C):
polycarbonate 0.065
Aha... here it is. Let's assume there is 20 degrees Celsius temp fluxuation inside our computer acting on our 3 inch long piece of lexan... it expands by about 0.1 millimeters. Using Polycarb's high modulus of elasticity, 0.1mm of movement will not be enough to induce or promote any sort of crack propagation.

[BillA]

no, it is not that simple
what is required is a strain analysis in the 3 directions
and the 3 principal sources will be the applied load, thermal (slight I agree), and any due to swelling due to absorption of whatever

- this last was Mota's quite valid point

my experience in this area is largely with epoxy/glass fiber composites which DO always fail by brittle fracture

understanding that the strains are cumulative, and just how little is required to enable propagation, thermal should never be ignored with any plastics as the coefficient of thermal expansion is much higher than metals while the moduli are lower
- little things can result in big strains (= energy at the crack tip)

still thinking "NOT" ?
a retraction would be nice

never apologize for 'talking technical', unknown terms will serve as a stimulus to those with an interest
(others use the Back button)

EDIT: just saw your edit to your post
are we talking about the same thing ? I think not as I am not even slightly concerned with the change in length of the top
I am looking at the combined strain at the thread root diameter with a hoop tensile stress of "X" imposed by the barb's pin end
to which is then added . . . . (as above)

out of curiosity, how are you quantifying the 'needed' crack tip energy for polycarbonate ?
I don't have a clue as to that value, but I know from experience that it is notorious for 'stress cracking' (what we are dealing with here)

[WinFlex]

Very nice writeup... I agree completely on every point. In my previous post I was simply exploring the possibility of fracture occouring where the polycarb top is affixed to the copper (this is usually done with four sockethead type bolts) Beacuse of their (polycarb and copper) unequal thermal expansion coefficients, stress will result from this... however, as you saw in my calculation, even assuming that the copper doesn't expant at all (which is untrue, thereby giving us a worst case scenario) inadequate stresses are produced to have any effect... somewhat off-topic, but also it is good to have eliminated this cause from our list.

You do however bring up a valid point. WE NEED TO FIND THE CRACK TIP ENERGY for polycarbs in order to continue this analysis on a scientific basis.
Also, we should focus on crack initiation. Crack tip energies are exponentially lower than crack initiation energies. If we develop a proper procedure to manufacture that top without initiating any sort of microcracks (or a method for reselaing them, with methylene chloride, for example), the whole crack propagation issue is out the window!

Note: Fiber composites and acrylics have much higher hardness on the Rockwell scale (or any other applicable hardness scale for that mater), resulting in a much steeper stress - strain curve. The polycarb curve is much longer and smoother because of the material's relatively high elasticity.
Again, we need more information on the material, namely we need numbers.

Hardness
Modulus of elasticity
"Toughness" (can be calculated from the above values, or using Rhiemann's sums to find the area under the stress - strain curve until failiure)
Stress Strain curves
Ductile to Brittle transition temperatures
Crack tip energies
Crack initiation energies