Q&A Session Answers

Q1. In your presentation, you don't discuss about dispersion of material data et load at least. What is your point of view for this?

A1. Thanks. I answered this in the webinar by interpreting that the question was related to the scatter of fatigue lives in laboratory testing of welds – which is around a factor of +/-3 on life. It may be that the question is more related to this;

https://www.ilearnengineering.com/manufacturing-industrial/dispersion-strengthening

I clearly outlined in the webinar the fact that the BS7608 fatigue curves were independent of steel material strength which has been intentionally conservatively done for the medium to high cycle fatigue that the code addresses. I am well aware of alloy steels which are higher strength steels configured from the addition of alloying elements, and have used such steels in numerous designs including Q+T steel from bisalloy.com;

https://www.bisalloy.com.au/

https://dl.asminternational.org/technical-books/monograph/94/chapter-abstract/2105464/Alloy-Steels?redirectedFrom=fulltext

It appears that dispersion strengthening is a method to produce alloy steels. The webinar introduction text indicated that it “will not be written from a metallurgy or materials science basis – rather a practical engineering design basis”. Hence it is believed that the webinar recording has answered the question.

Note that all fatigue codes don’t allow for such improvement of welded joints for parent material increased strength over grade 250 yield since they don’t cover the low cycle region. If you are going to rely on such improvement it should be done with extreme care, and a suitable inspection regime must be used, and it needs to be certain that medium and high cycle fatigue will not occur in operation.

Q2. Many of the cracks start at welds which can have defects, voids or a distribution of defects in the weld. I wonder if the distribution of defects have an effect? If it does how do we account for this? Can we do something like probabilistic fatigue analysis in design? What do you think?

A2. Excellent question, thanks. Every weld has imperfections. It is the type and size and location of the imperfections including proximity to others which combine to conclude if they are a defect. So, it is best to talk in terms of imperfections and defects – they are separate entities. The different global welding standards allow certain weld imperfections depending on the equipment type and operational life loading. For example, pressure vessels typically have very stringent weld imperfection criteria, especially the higher-grade classes. Reference 23 in the webinar is a good paper on weld quality;

Quality control and assurance in fabrication of welded structures subjected to fatigue loading (springer.com)

It references ISO5817 which is a good standard on weld imperfections.

https://www.iso.org/obp/ui/#iso:std:iso:5817:ed-4:v1:en

So, an imperfection may be a defect depending on its type, size, location and proximity to other imperfections among numerous other considerations – see the above ISO code and the fatigue loading welding codes – eg AS1554.5 – which only accept smaller imperfections due to the fatigue loading of the equipment.

Probabilistic fatigue analysis – sure that can and must be done and in effect is included in the fatigue codes by means of the mean-2sd design curve. The mean-1sd and mean-3sd curves are also noted. Good summary here.

https://www.efatigue.com/probabilistic/

Q3.Can you please comment on polymer(non-reinforced and reinforced(short fiber)) fatigue FEA in terms of approach, software and prediction accuracy?

A3. Thanks. I am well aware of polymer fatigue;

https://link.springer.com/referenceworkentry/10.1007/978-0-387-92897-5_818#:~:text=Polymer%20fatigue%20is%20the%20process,ultimate%20strength%20of%20the%20polymer.

https://www.sciencedirect.com/science/article/abs/pii/S003238611630204X

The webinar was based on BS7608 which is for steel. 10 of the case studies involved steel components – the 11th was based on cast aluminium.

Apologies but I cannot answer this question as it is not in the webinar scope. I would expect that there would be many useful engineering papers online which would answer the question.

Q4. Would a fatigue failure always initiate at the max stress location on the part?

A4. Thanks – the answer is no. Normally and typically the fatigue failure will occur at a location of low fatigue resistance, but it also depends on the alternating stress occurring at the location. The BS7608 code (and the similar IIW fatigue codes) clearly outline the fatigue curves for parent material as well as at welded and bolted joints- for the latter VDI 2230 must be used. So, what needs to be done is that all areas of the machine must be assessed for cumulative fatigue damage based on the Palmgren-Miners type summations as covered in the webinar – the ore grinding mills case study details this well. There are thousands of on-line links to such summations – a concise one is here.

https://www.efatigue.com/variable/background/damagesummation.html

So, if the highest alternating stress location occurs at a very low fatigue resistance fillet weld then it is very probable and almost certain that fatigue failure will initiate there first – but it depends on the alternating stress range magnitude. We have seen fatigue failure occur at very poor parent material locations – see the ore grinding mill case study of rectangular manholes with a very tight corner radius, also mill bolted flanges with very close edge distances – as well as at welded joints.

In conclusion the terminology “normally and typically” as above should not be used in fatigue assessments – as this could lead to fatigue assessment at only high stressed areas and not all areas.

Q5. Do you use a weld material curve to evaluate the weld life, or do you use the parent material method (using notch factors)?

A5. Thanks - you must use the weld fatigue S-N curve to evaluate the weld life. There are multiple weld fatigue curves depending on the weld type - fillet, partial penetration, full penetration, single sided etc. The parent material fatigue S-N curve is used to evaluate the parent material fatigue life. The fatigue codes are very clear on this. BS7608 Tables 1 to 10 do a good job of detailing this. The webinar has been compiled from an engineering basis as per the introduction text - “will not be written from a metallurgy or materials science basis – rather a practical engineering design basis”. Notch factors are a metallurgy issue, and I cannot comment on that apologies.