Internal stress in a Female NPT 1

[JTraeger]

I'm redesigning a Female NPT fitting and was wondering if anyone had any info on how to figure out internal stress or how it might be distributed throughout the fitting. Is it all hoop stress, or is there a lot of axial stress from the threads pushing against each other? What is a reasonable estimate for the hoop stress? Any help at all would be appreciated, I can't find much info anywhere.

 

[makeup]

In response to your question I can add, that there must be some proportion of axial force on the fitting otherwise it would not be prone to blow-out under pressure. Now the amount of axial stress might be related to the surface area exposed in the internal axial direction (the end of the fitting). This axial force must be counterbalanced by the threads. Hence due to the nature of the NPT profile there will always be a limit as to the range of pressures for which it is suitable. Sorry I cant be more specific, I am not technically qualified in this field, only in using this thread type in the field.

 

[JTraeger]

I know that there is some portion of the stress in the axial direction, but that isn't my main concern. I'm more interested in the hoop stress because the the part is failing due to stress corrosion cracking, not blowing out. My hope is to find where most of the hoop stress is located and change the design to reduce the hoop stress as much as possible. I just wanted to know if the axial stress should be taken into consideration.

 

[makeup]

So your part is leaking through cracks induced by stress corrosion. Rather than redesigning the fitting, what about different materials or heat treatment processes to reduce initial internal stresses???

 

[RichGeoffroy]

JTraeger:

If axial stress is a problem, it will concentrate at the thread root which will eventually result in circumferential failure initiating in that location.

Axial splitting is the most common failure mode for female NPT fittings and results from overtightening. Once thread engagement occurs, further tightening expands the circumference of the female socket. The stress in the area is strictly a function of the circumferential strain --- the difference between the stressed circumference minus the relaxed circumference divided by the relaxed circumference. Multiply the strain by the modulus of the material and you’ll get the average hoop stress in that cross-section.

If that stress is below your long-term design stress for the material, failure should not occur in a properly manufactured fitting.

Over –insertion of the pipe end into the fitting is a common cause of excessive circumferential stress in a female NPT fitting. Another cause is the use of PTFE thread tape. Excessive amounts of thread tape cause the thread interference to occur with less thread engagement, therefore the technician feels he must insert the pipe further to ensure tightness. Also, tightness is often assessed by the “feel” or torque required. PTFE reduces the coefficient of friction between the threads, therefore there is a tendency to overtighten.




Rich Geoffroy
Polymer Services Group
POLYSERV@aol.com

 

[desertfox]

Hi JTraeger

Which way do the cracks run in the fitting, are the parallel to the fluid flow or at 90 degrees? if they are at 90 degrees the hoop stress is not responsible.Further as the previous post states what materials have you got in contact and what fluid is going through the fitting and what pressure is the system at?

regards desertfox

 

[JTraeger]

I know about SCC (stress corrosion cracking), and I've looked into different materials, but the problem is the cost. My employer wants me to keep using 360 brass if possible. I'm testing some fittings that are chrome plated that should protect against SCC. The chrome seems more like a band-aid than a solution, and what I'm worried about is if the chrome is scratched or chipped, will it still fail? What I want to do is reduce the hoop stress to prevent the SCC. The crack that forms does run parallel to the fluid flow. The fitting is for residential plumbing and only runs at 20psi at the most, so most of the stress comes from tightening and thermal expansion, not the system. I've also looked into heat treating the fitting to reduce residual stress from machining, but when I tested the fitting it showed almost no residual stress.
Rich, are you saying that I can take (Cs-Cr/Cr)*E to find the average hoop stress in that cross section?
Thanks for the help all.

 

[dicksewerrat]

Go buy 10 different BRANDS of fittings and measure them. I have been using brass fittings for some 40 years and never had one crack longitudinally. Maybe they need to be thicker. I have dug up fittings that are 120 years old, operating under 100 psi and no cracks.

 

[desertfox]

HI JTraeger

Not had time to study this in depth but have you considered that your problem may be caused by hydrogen embrittlement due to the playing process.


regards desertfox

 

[RichGeoffroy]

JTraeger:

Rich, are you saying that I can take (Cs-Cr/Cr)*E to find the average hoop stress in that cross section?

Yes. The stress is strictly a function of the strain induced by the interference fit of the piping in an NPT. The more the piping is threaded into the fitting after thread engagement, the more the fitting must expand to accommodate the pipe. Therefore, the fitting must expand. It is the expansion of the fitting (strain) which results in the hoop stress in the fitting cross section.




Rich Geoffroy
Polymer Services Group
POLYSERV@aol.com

 

[JTraeger]

I've tested similar competitor fittings and found most of them cracked. There was one brass fitting that didn't crack, but it was cast and much thicker than mine. The other ones that didn't crack were bronze-brass, but that costs three times as much. One design that preformed well was thicker, so I'm looking into using a larger hex bar to start with (1/16” larger).

Last night I came up with an idea to change the amount of thread engagement. I figured that the more engagement on the threads, the more the stress would be distributed. I then read in Machinery’s Handbook 26 Ed. “It is recognized that in special applications, such as flanges for high-pressure work, longer thread engagement is used, in which case the pitch diameter E1 is maintained and the pitch diameter E0 at the end of the pipe is proportionately smaller.” Basically what it is saying is to change the pitch between the end of the pipe and the plane of hand tight engagement. Has anyone done this, or know how much I should proportionally change the pitch?

As far as hydrogen embrittlement, I don’t think that’s the case. Everything I read about it has to do with steel, or aluminum, or electroplating. I’m sure that SCC caused the cracking in these fittings, which come from a combination of internal stress and exposure to ammonia. I’m waiting for some chrome plated fittings to come in so I can test them for SCC, but I’m concerned by the way contractors might handle the parts, which could scratch off the chrome, leaving the brass exposed.

Anyway, I’ll have to make up a bunch of different designs and test them. Thanks for the help with the stress-strain equation Rich.

 

[makeup]

The pitch is constant. The taper is constant, there are also gauges used when cutting these threads to ensure a degree of make-up conpatability. There are also thread lengths in the spec. You cannot just alter a thread like NPT, it has recognised standards that must be applied to make it appropriate. There are hand torque requirement and stand off once tightened. If you are operating within the spec given you will not have a problem. If you are over tightening buy a torque wrench. Have you considered that it may not be the fitting or the thread that is wrong.

 

[JTraeger]

It's not me that is tightening these fittings, so I can't assume that a contractor is going to know how much torque to apply. Also, I stated some thing wrong. When I said I was going to change the pitch, I meant change the pitch diameter or in other words change the angle of the taper between the plane of hand tight engagement and the end of the pipe. I decided to change the angle on the female end, so I'll make it a little wider near the opening. Hopefully this will move the highest concentration of the stress towards the thicker end of the fitting. I'll have to get a couple of these parts run off to test to be sure, but it's an idea.

 

[makeup]

I have cut 1000's of these threads in my time and believe me the taper is constant and set at a given angle. Change this and you could be out of spec? But you have mentioned something that has jogged my memory. Have a look at line pipe specs. Now line pipe is very similar but has an undercut or relief at the enterance to the box (female) end this might help you here. It the same taper and thread form if I remember rightly anyway have a look.

My reason for being persistant is, I know someone who recieved life threatening damage caused by someone else fitting incorrect NPT threaded weldolets. Why dont you buy your contracter a torque wrench and give them some training. It wouldnt take five minutes, could be done on the job and save you all this time. Remember your time sorting this out is costing money.

 

[RichGeoffroy]

JTraeger:

Be careful not to specify torque in your joining directions. Torque is too dependent upon the coefficient of friction between the threads. If a thread sealant is applied to the connection, inserting the pipe to the proper torque can result in excessive strain which can result in failure.

Most manufacturers will specify a number of turns beyond hand-tight. Hand-tight insertion ensures thread engagement; while the number of turns beyond hand-tight ensures a tight joint within the recommended circumferential strain for the fitting body.

Makeup is correct that you cannot alter the taper on an NPT --- it’s standard. If you change it, it will no longer be compatible.

Also, your notion of having less stress on thicker sections is incorrect. The stress is a function of the strain. No matter how thick or thin the section is, if you strain it a certain amount both the thin and thick sections will have the same stress (however, it will take more load to deflect the thicker section). If anything, when using the same deflection, your thicker section will have a higher hoop stress at the inside surface because of the smaller inside diameter of the thick section compared to a thin section.




Rich Geoffroy
Polymer Services Group
POLYSERV@aol.com

 

[JTraeger]

I understand your persistence and appreciate it. I know that I still have to keep it in spec, and I have to make sure that it still works with the male threads of competitors. However, I'm not concerned about failure under pressure because these are for residential plumbing and carry a relatively small amount of pressure. I 'm trying to come up with ideas to shift where the stress is applied to, so these fittings aren't as prone to SCC. I know that it will be thinner near the box, so I might switch to 1 1/16" hex rod instead of the 1".

The solution to this problem is not as easy as simply giving torque wrenches to the contractors. First off, torque is a function of friction, which changes because these fittings require Teflon tape or thread compound to seal. These affect the friction and therefore affect the amount of torque applied. Secondly, These fittings fail from SCC at relatively small amounts of torque, even before they're sealed.

Like I said, this is just an idea. I was just wondering if anyone heard of changing the angle of a NPT thread. I read that is can be done, but I assume that it needs a custom tap to make it. I'll look into line pipe specs; it sounds like it's for high-pressure pipes and might be of some use. I have to look through some books and see if I can find anything, but thanks for the help.

 

[JTraeger]

Your point is well taken Rich, I under stand that if I change it, it won't be standard. I was hoping to find a way to change where the pressure is applied. What I was thinking was if the same pressure were applied to a thicker section, it would induce less strain, and therefore have less stress.

I was thinking if I could increase the diameter near the opening a little, it would reduce the force that is applied there. How about adding an extra thread? That would make the opening a little larger. I know I'm not an expert in this area; I'm just trying to come up with some ideas.

 

[RichGeoffroy]

JTraeger:

I mentioned that thickening the cross section of the fitting socket will not affect the stress in the fitting at a given deflection or strain. However, it will take more load for the pipe fitter to effect that strain. Therefore, a thicker wall can give the assembler the “feel” that the pipe is “locked in”, thereby, averting further excessive strain that might result in failure.




Rich Geoffroy
Polymer Services Group
POLYSERV@aol.com

 

[JTraeger]

Thanks for all your help Rich. At least now I have an idea what the stress is in the fitting. As far as the design, I'll have to take a break for a while.

 

[TBP]

Is there typically a solenoid valve located near this fitting where failures are occuring? I've seen lots of components beaten to death because of solenoid valves slamming on and off. I've also seen any number of things fail due to backflow preventors being retro-fitted without any consideration given regarding the thermal expansion of the liquid. Take either one or both of the above, and combine with cheaper/lighter/nastier components in general, and you'll very often get led to the cause for repeated failures.