Nozzle Pro for Div 2 Appendix 5 fatigue analysis? 2

[vesselguy]

Hi all,

I never realised NozzlePro is suppose to be able to do fatigue analysis until I saw a vendor submitted his calculations for me to approve. I have a Div. 1 thin wall (31.75mm thk, 48" nozzle on 31.75mm thk dollar plate) application that is going through 3 cycles per day of 160°C temp changes.

It appears this program does not do any thermal analysis. It ask you to specify the temperatures on inside and outside steel surfaces. I assume then that one must do a 2-D thermal analysis to get a temperature gradiant before you run NozzlePro.

So, how does this program do fatigue analysis to meet Div. 2 Appendix 5? Is is valid?

Thanks.

 

[LSThill]

Yes! also is in CODEWARE COMPRESS FE/NOZZLE meet the DOE Requirement.

Email ALL support questions with attachments to: support@paulin.com

 

[TGS4]

Isthill - I don't think that this is a "support" issue - rather a question from an engineer who is reviewing a vendor's calculations.

vesselguy -
My only suggestion is perform a rigorous evaluation, just like you would for any FE-based report. Ask questions of the vendor, including the ones you posted here. Ask them how they determined the temperature field at each step in the thermal transient. Ask them what the discretization error in the thermal calculations is (assuming they too were done using FEA). Ask them how they incorporated the temperature-dependent material properties. Ask them what the discretization error is in the stress analysis, and whether or not they performed a sensitivity analysis on the error to determine how much the error would affect the fatigue life. Also, ask a lot of "why" questions - why did you choose to perform the analysis this way - why did you choose this mesh density - etc.

Your best friend in evaluating 3rd-party FEAs is your engineering judgement and your ability to ask as many questions as you like. AND, don't take "it's the program default" as an answer - that's a complete cop-out and shows a dangerous lack of engineering judgement on their part. A good analyst will be able to answer all of your questions thoroughly and make you feel very comfortable with their analysis - regardless of the program that they use Chances are pretty good that if they can't answer the above questions (plus your own) then the analysis and subsequent report is worthless. At that point they should start looking for a adequately-qualified FE consultant who socializes in pressure vessels and the ASME Code.

I'll get down from my soap-box now.

 

[stanier]

How was it done before FEA? Could one meet the requirements of the Code with manual calculations? Has the software been verified against known problems?

 

[TGS4]

Having done this type of problem many times, all that I can say is that you _might_ be able to do it manually, IF you make an inordinate number of assumptions. This is basically the type of problem that drove the nuclear engineers to using computers and numerical methods back in the "good 'ole days".

Further, I don't really care whether or not any software has been "verified" or "validated" against known problems - I'm more concerned about whether or not the analyst has been "verified" or "validated" against this specific problem. Hence all the questions that I would ask.

Like I said before, a good analyst will be more than happy to answer all of these questions. Who doesn't like it when people ask intelligent questions about your work. I know that I am proud to answer all of the questions that I listed before - it's in answering these questions that I can "validate" or "verify" myself.

 

[carthago]

I think you should direct your general questions to FE-pipe.

----------------------


By the way there is a simplified method in PD5500 ( The same old BS) ,IN ITS APPENDIX g, regarding this type of problem (transient loads).

 

[LSThill]

Do your research to understand the history and development and Verification & Validation at WFI in 1996-1999.

Note: TGS4 (Mechanical), the development first is FE/Nozzle in CODEWARE COMPRESS Software Verification Validation, and Software Verification Validation done.

Nozzle PRO was release a few year leater.


You willl need to do a detail Software Verification Validation, and Software Verification Validation before using Nozzle PRO. which is a pary of your company Quality Assurance Plan.

When to Use NozzlePRO
Typical occasions when a finite element analysis of a NozzlePRO geometry is beneficial are listed below:
1) When the d/D ratio for a loaded nozzle is greater than 0.5 and WRC 107/297 is considered for use.
2) When the t/T ratio for a loaded nozzle is less than 1.0 and WRC 107/297 is considered for use.
3) When the nozzle is pad reinforced and WRC 107/297 is considered for use.
4) When the number of full range pressure cycles is greater than 7000 cycles and the nozzle is subject to external
loads.
5) When the D/T ratio is greater than 100 and SIFs or flexibilities are needed for a pipe stress program.
6) When the D/T ratio is greater than 100 and a dynamic analysis including the nozzle is to be performed using a
piping program.
7) When a large lug is used in a heavily cyclic service.
8) When pad-reinforced lugs, clips, or other supports are placed on the knuckle radius of a dished head. WRC
107 simplifications for pad reinforced rectangular lug attachments are fraught with potentially gross errors.
9) When seismic horizontal loads on vessel clips or box supports are to be evaluated.
10) Pad reinforced hillside nozzles subject to pressure and external loads.
11) Large run moments, but small branch moments in a piping system.
12) Overturning Moments on Skirts
13) Effect of Integral vs. Non-Integral Pad on Nozzle in Head Should be Studied
14) Different thermal expansion coefficients or temperatures between the header and branch.
15) Where loads on nozzles are high because of the assumption that the nozzle connection at the vessel is a rigid
anchor. Few connections at vessels are “rigid.” Often only small rotations can significantly reduce the calculated
moment and stress. Accurate flexibilities permit the actual moment on the vessel nozzle to be calculated and
designed for.
16) Heat Transfer in An Axisymetric Model Geometry
17) When the effect of adding a radius to weld geometries on nozzles in heads should be investigated.
18) To verify FEA calculations. NozzlePRO4 allows nozzles in heads to be analyzed with shell, axisymetric, or
brick finite elements. The analyst can run each model type and compare results to determine the stability and
accuracy of the solution.
19) For saddle supported horizontal vessels with or without wear plates including tapered saddles with many
design options.
20) To evaluate effects of axial or transverse loads due to internal sloshing, wind loads, seismic loads, or general
external loads. Zick’s methods do not consider axial or transverse loads.
21) Design of Pipe Shoes for self-weight, liquid weight, and external loads.
Criticality of the application is a major consideration when deciding whether or not to run a finite element
calculation. Hot hydrocarbon products are clearly more dangerous than ambient temperature water
processes and should be approached with increased caution. Systems that do not cycle are less prone to
failure than systems that cycle daily. Extreme design conditions can also make using less conservative, more
accurate approaches practical. Large d/D, D/T intersections are difficult to analyze properly for a
combination of pressure and external loads, and FEA results tend to give more consistent results over a
broader range of problem parameters. Allowable loads on vessel nozzles give the piping engineer guidance
when evaluating thermal loads on anchors. Higher earthquake load requirements can make conservative
design assumptions costly. Caution should be excercised when low pressure-high temperature systems are
evaluated as these lines tend to have high loads and large d/t ratios.
“It is absurd to use FEA on every system, and it is absurd not to use it at all.”

 

[LSThill]

Go to download

http://www.Paulin.com

Nozzle PRO & Axi PRO Analysis Level Quality Assurance Document April 4, 2004, see ITEM 7) ASME Section VIII Division 2 Code Compliance.

 

[TGS4]

Isthill - you seem to be really pushing the limits of "promoting". What you seem to be misunderstanding, is that whether or not a specific piece of software "can" analyze a particular problem is, in MY opinion, irrelevant. For example, ANSYS "can" analyze magneto-electric problems. That does NOT mean, however, that just because I use ANSYS that I can perform a magneto-electric analysis. Do I need to make myself any more clear?

Furthermore, if at vesselguy's company, they do not run this particular piece of software, then they would not have performed the "Software Validation Verification...as a part of the Quality Assurance Program". As a part of a company that receives 3rd party reports all the time, I can tell you that we have neither the time nor the budget to validate every piece of software out there. And, particularly for FEA, any validation would be essentially worthless, because (as I have said again and again), it's not the software - it's the analyst - hence asking all the questions.

Simply put, if you are blindly accepting that everything that comes out of a particular piece of software is valid, based solely on the name of the software, then that would be troublesome at best, and gets into the realm of lawyers at worst.

 

[jte]

I gotta agree with TGS on this one. I could easily teach a high school dropout with ordinary computer skills to run both the vessel design software and the FEA software which we use. Both are common, commercially available, validated/verified programs.

Much of the time this dropout would produce reasonable results. That's not a high enough success rate for my line of work. It really does come down to the knowledge, experience, and background support which the engineer/analyst has. If the engineer cannot explain how his software develops the results which he is getting, beware! I'm not saying he has to know the exact algorithm used by the program, but he should be able to intellegently discuss the process and the theory to back it up.

I suspect this will be an interesting thread...

jt

 

[vesselguy]

Thank you all for your feedback.

I did a run on NozzlePro last night to simulate what the vendor was doing and carefully read his report. I think I now understand what was done but I am having a hard time accepting his report. Here's the back ground.

A Div 1 vessel has a 48" nozzle that will see hot fluid enters the vessel in 3 cycles per day resulting in temperature ranges of 190C to 350C, with each cycle lasting 8hrs.

The vendor did a thermal run, but the results were not included in his FEA report to me, and concluded there is only a 1 deg C temperature difference between the inside of shell/nozzle and outside surface of shell/nozzle. They enter this 1 deg temp difference, into NozzlePro and activate the fatigue calculations. The report concluded the nozzle/shell configuration is suitable for over 36000 cycles.

I do not believe the way this calcualation is done is correct to simulate the expansion and contraction of the shell/nozzle geometry during the temperature changes. I think, the correct way is to find the differential displacement on the model due to the max and min temperature, and then calculate the stresses from the strain.

I have never done an thermal stress analysis using FEA before. I am posting here only to draw out some opinion from FEA users on how they think it should be done, for comparison sake.

Isthill, I will most certainly talk to Paulin Reasearch to find out more about how NozzlePro can be used for this kind of situation. Thanks.

With the modern FEA software, it is so damn easy to put garbage in and get garbage out that looks so damn good. And yes, don't use FEA when good old approximation from Roark & Young would do.



Thanks all again.

 

[DSB123]

TGS4,
Agree with your initial remark. The forum is for engineers to answer others questions if at all possible not to promote Companies software. lsthill does this all the time. Never actually answering a question just pointing to a software solution. I have wondered if this is being due to being an agent for the software company?

 

[LSThill]

vesselguy (Petroleum)

Regards to your question in 1988 ANSYS not have any ASME DIV 2 code check. At that time SDRC IDEAS did have ASME DIV. 1 and Div. 2 CODE COMPLANCE CHECKS.

The fabricator on June 2004 in Bombay, India is still using IDEAS for ASME SECTION VIII, DIV. 1 and DIV. 2 design calculation.

The back ground of sloution of FEA and ASME Code complance calculation for review dose depend on the submittle to the EPC project.

Humm; ENGINEERING TOOL'S (thank you Fread for FE/Nozzzle 1998, and letter Nozzle PRO.)


Regards

OPIA

 

[TGS4]

vesselguy - you have certainly focused on the heart of the matter - the transient thermal analysis. If I were you, I would focus on that - it seems to be the only point of contention that you have.

Also, bravo for performing an independent "check". IMHO, that is often the only way to make sure that the analysis is correct. And with your comment about garbage-in garbage-out - you basically took the words out of my mouth. My line "With FEA, it's very easy to get very good looking wrong answers".