P91 weld metal IMPACT properties vs. Temper parameter

Date 06-30-2014 14:46 Edited 06-30-2014 14:48

Eutectic,
The short answer to your question is no.
Holloman Jaffe does not consider things like weld metal chemistry, welding heat input, depth of prior bead tempering, interpass controls, etc., which are critical to the results of CVN testing with all materials much less Grade 91.
There is certainly data that can demonstrate the influence of the parameters used in H-J but the final results must include these additional variables.
You state all other things being equal but I am not sure this is possible in a sense that gets you to a range of results you are looking for.

By js55

Date 06-30-2014 14:53

Keep in mind there is a wide scatter band inherent in CVN testing itself. If you take additional samples out of the exact same specimen your results can vary to a greater extent than what I am assuming you are seeking here.
The all things being equal would have to apply to lab techniques and practice as well. Most notably the accuracy and variance in temperature control and in the radius of the notch.
This is why AWS has you take 5 samples and toss 2. And why the Codes go to such lengths for additional testing and exceptions. They are trying to homogenize out the inherent variances.

By Eutectic (**)

Date 07-08-2014 12:06

js55,

Thank you very much for your reply, it was the answer I feared even though hoping for another one.
The implication of your statement just confirms that experienced fabricators should be used as it is difficult to even reproduce the PQR results in production.
I am still curious though that within application parameters if you still maintain martensite transformation and controlling heat input and subsequent bead size to ensure inter bead tempering weather or not ou will be able to plat a relationship between the HJ and the impact properties.
Or are you saying that the Heat treatment would not significantly affect impact properties as it does for UTS or Hv and that it is predominantly dependent on the actual welding application?

Cheers
HF

By js55

Date 07-08-2014 14:35 Edited 07-08-2014 14:51

Eutectic,
I think the problem is not unlike that which confronts those who wish to impose robotics on shop fab pipe assembly. You have to tightly control every single variable so that you can minimize the number of programs. In this case you would have to tightly control every single variable to minimize the scatter band.
Meaning, material chemistry, base and fillers (especially with alloys), fit up, preheat, base metal thickness, filler diameter, automatic welding with locked parameters, and much much more. And this just for a single pass weld. And you might have the ability to roughly graph a result. Multipass welding complicates it even more. Travel speed has to be exact, bevel angles, bead shifts, etc. And I haven't even reitrerated the testing lab variables. Notch radii, fracture rate, exact temperature control, etc.
And keep in mind with carbon steels martensite is not an issue. You are talking volume percents of acicular/polygonal ferrite, etc. Ultimately you may be able to get there but then what do you have? Mountains of lab data. Or lab data for a very narrow range of variables. And lots and lots of money spent. Much like robotics.
Not something that is practical on a shop floor.
What you would learn, which could be a great deal, would be academic.
I think the fundamental error in your thinking is trying to squeeze precision out of two inherently imprecise methods. HJ and CVN's.
Even Section IX, which is considering the application of HJ, is only trying to get in the neighborhood.
And no, heat treat can significantly effect CVN's. But it is notoriously unpredictable with any precision. The thought process is one of saying that if I do these things I have reasonable assurance that I can maintain a minimum standard, that my test will reflect my production. It is not one of precisely determining the result. I don't think that can be done.

By Eutectic (**)

Date 07-09-2014 09:40

js55
Thanks again,
The practical problem one is stuck with then is (considering all the variables you mentioned):
With how much comfort can one rely on the ranges given within a said code from qualification (PQR) to application (WPS)? Of course I am focusing now on alloy materials.

Our current project is EN/ISO 12952-5 and 15614. In the current edition for instance the welding application standard allows a +-20C on the PWHT temperature range. Normal philosophy would be to qualify top and bottom temp range and then you should be covered for anything in between. However this is not explicitly stated as in Section IX or VIII Div 1 where you are limited to 80% total time and 14C on qualification to application. Fabricators are therefore using this loophole to test at a single temperature and applying a range.

This is the background to my initial question as there is no data available for the time/temperature window qualified the only tool I have available is the HJ to guestimate the effect of time and temperature on a production weld that is not closely simulating what was tested in the PQR.

Thanks again for the time taken to reply.

cheers
HF