FEA of pressure vessel

[Anonymous]

Hi,

I am currently in the process of modelling a simple pressure vessel, to represent a SCUBA cylinder. 

I have created the model with no problems and applied the pressure to the internal wall, however I am not sure on the accuracy of my results when I run the stress analysis. 

I have calculated the hoop and axial stress by hand, and cannot see what Inventor classes these stresses as, with no figures seeming to relate. Are they the 1st Principal stress, or for example the YY stress (if the cylinders length is in the Y axis). 

After reading the forums it seems my problem may be in the constraint, I have currently constrained the bottom face of the cylinder using the 'fixed' option. There was mention of creating three points on the base and constraining these in a certain way, however when I tried this it just created a large stress raiser at the base.

Attached is a screenshot with the model as it currently stands. (The small stress raiser on the face is intentional and is to model impact damage, this was added well after the problem arose, I assume this does not affect the rest of the model drastically?). 

Thanks

Sean

[JDMather]

Attach your *.ipt file here and your calculated hoop stress value.

[Anonymous]

I've used Algor (which is old now and is now called Autodesk Simulation) extensively.  It sounds like you are evaluating as a "Thin Wall" vessel, thuis are assuming the stresses are constant throughout the thickness.  From my experience with thin wall pressure vessels you need to create a surface model to obtain axial and hoop stresses from FEA.  A solid model will not generate those parameters.  I've only tinkered with the FEA included with Inventor.  I don't know if it has ability to perform FEA on a surface model.

If you don't have any particular guidelines or code to follow I suggest also including evaluation per ASME's Boiler and Pressure Vessel Code.  For thin walled vessels It involves comparing axial and hoop stresses to allowable stresses developed by the ASME Boiler and Pressure Vessel Code.  It also has allowable stress values for discontinuities, curves, etc.  With the pressure vessels I evaluated the FEA surface model results were identical to calculated hoop and axial stresses.  I don't think there is any need to upload any of your files at this time.  Inventor's FEA capabilities is rather limited.  But, FEA software is extremely expensive which is why I still use older software for that.  If you have no other FEA tools I think I would look into FEA of a surface model Inventor.  If it can do it, your results should be consistent with hand calculations.  Good luck with it. 

Regards,

Mike

[Anonymous]

Have attached the file. It is a simple cylinder of 171mm diamater 4.7mm wall thickness, so yes it is thin walled.

Hand calculations give a hoop stress of 398 MPa on the main body.

Thanks

Sean

[Anonymous]

Sorry I forgot to state that is with an internal pressure of 23.2 MPa.

[JDMather]

---

@Anonymous wrote:
Sorry I forgot to state that is with an internal pressure of 23.2 MPa.

---

You shouldn't have to tell me any of that - it should all be in the file, that is why I requested the file.  I need to be able to examine the logic of what you have actually set up, not what you "tell" me you have done.  That is the purpose of using CAD.

There is no study set up in the file you attached?

[Anonymous]

Apologies, I didnt realise the study would attach as part of the file I thought it was a seperate study file. I have only been using the software for 2 days and am still getting the hang of it.

Heres the file with the analysis.

[JDMather]

There are several problems that need to be considered.

#1. Inventor FEA is limited to linear static stress analysis of anisotropic materials.

This means within the elastic range of deformation - not into plastic deformation.
I would at least change the material properties.  Finding the a property reference after deep-draw stretching might be difficult.   (I didn't try this search.)

The manufacture of the tank sends the material well into plastic deformation - changing the material properties.

Might still be able to get some useful information though.

#2 Inventor calculations are only valid where calculated stress/yield stress >1.

(Termed Safety Factor, but this is a mis-understood term and probably not a very good term. Forget the term and only refer to the formula.)

#3 I would model your tank more like the actual geometry and place the Fixed constraint at the neck.

The gas pressure would be against every internal face - I noticed that you missed a couple of faces.

If you check the Z-Z or X-X stress with a Probe - I do get something close to your calc, but considering #3 and especially #1 and #2 you should....

[Anonymous]

Thanks for the extremely informative reply JDMather.

What you have said seems to confirm my original worries. All information I have got is from the manufacturers data sheet.

The material is stated as 'Chronium molybdenum steel', but not with its percentage of constituents. Typical values for this are approximately 460MPa yield strength, and 560MPa UTS. This is why I have chosen AISI 4130. The wall thickness is on the cylinder itself and is definitely 4.7mm.

Following the hand calculations on the main body of the cylinder and producing 398MPa I do not understand what is wrong. Surely the geometry of the neck will not play a part in that calculation, and the main walls will still be far too close to the yield. (Test pressure is also 37.2 MPa which reinforces the fact something is wrong and takes the material well into its yield and even UTS).

The cylinder exists and the forces are clearly not as great as these, however I cannot see what I am missing?

[JDMather]

---

@Anonymous wrote:
....The cylinder exists and the forces are clearly not as great as these, however I cannot see what I am missing? 

---

You are much closer than you think, but I don't have time to give full explanation for the next two days.

How long are you going to be around?

Is this a school assignment?  If so, when is it due?

[Anonymous]

Yes it is, due 30th June. Can work on other bits in the meantime this shouldn't hold me up whilst I wait.

I will continue to look over it and try to figure it out in the meantime.

I look forward to your return, and thanks again for all the help!

[JDMather]

---

@Anonymous wrote:

The material is stated as 'Chronium molybdenum steel', .... Typical values for this are approximately 460MPa yield strength, and 560MPa UTS...

Following the hand calculations on the main body of the cylinder and producing 398MPa I do not understand what is wrong.  

---

I assume these values are before the living daylights have been stretched out.

I would also want the Young's Modulus (E Modulus of Elasticity).

So the purpose of this exercise is not to test the tank - it is to give assurance (I am not going to use the term "validity") to a hand calculation.

Let's take a small portion of the center of the tank.

Examine the attached results.

But I don't like leaving this with just a pretty picture.

I will come back on Friday with more of an explanation.

[graemev]

No to throw a wrench into the works, but...

If the overall intent is to analyze the "damaged" section, why not model just the cylindrical section, applying internal pressure to the interior surface, fix one end against longitudinal displacement (fixing additionally as appropriate,) and apply tension to the opposite end as hand calculated from the pressure and cross sectional area of the internal volume?

The rest of the model is of little interest in the calculation and adds enormous overhead to the analysis. If it's the body of the cylinder that's damaged the ends are not likely to fail. (If the bridge is near failure, don't examine the handrails.)

[Anonymous]

I am still inclined to think it needs to be a surface model and not a solid model if you are going to accurately compare results to thin wall vessel hand calculations.  There is a reason the pressure vessel wizard in Algor and ADesk Simulation builds surface models and not solid models.  But I noticed in a couple images the tank heads have variable thickness so a thin wall evaluation of those would not apply to the heads.  A solid FEA model of those would be best suited.  I don't quite understand why a constraint would be applied at the fitting on the upper head.  Doesn't seem like an appropriate constraint or I am simply missing part of the problem.  Also, if I recall, and I may be incorrect since Material Science courses were many years ago, but anisotropic material = properties are not constant in all directions, and isotropic materials = constant properties in all directions.  Those terms are not related to plastic or elastic deformation.

[JDMather]

---

@Anonymous wrote:

1. I am still inclined to think it needs to be a surface model and not a solid model if you are going to accurately compare results

2.   I don't quite understand why a constraint would be applied at the fitting on the upper head.  Doesn't seem like an appropriate constraint or I am simply missing part of the problem.  ...

---

1. Did you look at the Hoop Stress file that I attached above.

Inventor will also do Thin Body (midplane) analysis.
I checked it both ways and the results were not significanlty different.

2. I misunderstood the original question.  Apparently it isn't about testing a testing a tank.  The real question is about verifying a hand calculated Hoop Stress value - in which case the entire tank isn't even needed and only serves to confuse the intent of the exercise.  (see posts 12 & 13)

[Anonymous]

I am modelling the effect of corrosion on a cylinder due to ingress of seawater. I am looking at uniform loss of wall thickness due to corrosion and its effect on stress (primarily to back up basic hand calculations, simple reduction in wall thickness). Secondly the effect of an indentation on the external face of the cylinder, which represents impact damage. There is no requirement for overly complex models and simplifications are fine, considering the use of the software is purely a personal choice and just aids in backing up my other calculations. 

I checked out the file JDMather and it certainly makes sense to use it over a cylinder model!

I am still struggling to see how the stresses are acceptable considering the Yield Stress. I have done some research on the topic, and found the following; 

Steel Scuba tanks are ordinarily manufactured from modified AISI 4130 compositions (0,3 C — 1 Cr — 0.2 Mo) by deep drawing on hydraulic presses from circular blanks. Final treatment is oil quenching and subsequent tempering to a minimum tensile strength of 750 MN/m2  combined with good toughness, namely minimum Charpy V notch impact strength of 80 to 100 J/cm2 .

I assume this is the answer to my problem, giving a far more acceptable (and legal) safety factor?

Thanks

[JDMather]

---

@Anonymous wrote:
1. ...I am still struggling to see how the stresses are acceptable considering the Yield Stress. ....

2. I assume this is the answer to my problem, giving a far more acceptable (and legal) safety factor?

---

1. In your own words, why are you the least bit concerned about the calculated stresses considering the Yield Stress?

2. Well, I hold off on my question on #2 until you have answered #1.

[Anonymous]

Hmm I may be getting myself confused here as I haven't studied this for a while therefore very very rusty. 

I am considering the 398MPa as the hoop stress, being the maximum stress acting along the circumfrence and twice the axial stress. 

The yield strength of the steel according to the manufacturer is stated as 435 MPa. The stress acting on the internal walls when 23.2 (232 Bar) is applied to the cylinder it is dangerously close to the yield strength. Hence the worry. However with a far greater yield strength of then the 398MPa is within the limits specified (2/3 of yield maxmimum in this case).

Although the use of Von mises stress has now thrown a spanner in for me, as it is the equation used to determine the Yield stress based upon the three principal stresses? 

My brain is mushed I think..!

[JDMather]

---

@Anonymous wrote:

... dangerously close to the yield strength. ...

---

Why is it "dangerous" to be close to Yield Strength?

[Anonymous]

To avoid any defects or corrosion causing localised increases in stress plastically deforming the cylinder. Or alternatively simply accidentally over pressurising the cylinder and causing the same effect?