This discussion has been really good because of the different dimensions and factors that continue to be expressed and described by the contributors. Thanks to all who have responded!
Within really limited ranges of products, suppliers have managed to sell adequately performing materials by merely looking at the product specification and adjusting their processes to fit. Some of the correlations and rules described here have been through that sort of adaptation. And the reasons that was a successful approach was that the groups who wrote the specifications in the first place, or revised them over the years, were composed of people with enough industry experience with what worked and what didn't. They got together, drew a boundary around a set of rules that mostly included success and mostly excluded failure. Then designers used materials from that pool of success to deliver the final products.
A problem as I see it with the drive to produced engineered data sets, such as described in the upcoming webinar promoted by ASM, is that sheer multitude of dimensions of compositions, production processes, and even weather (for instance affecting cooling rates) which are not captured. For instance, elemental compositions which are not specified are not analyzed for. Many of the reasons that the specification Committees wrote the rules the way that they did were never captured or preserved. Therefore when industrial practice changes outside of the original process boundaries or just pushes against the edge of the spec in too many ways, the new user does not know the "why" of failures. We had an expression at work about "dead guy's rules" for when a process or activity had been defined by an expert or group who was no longer available and who had not recorded the reasons for why something was done that way.
After several generations of materials specifiers may have been using a call-out of properties without a full understanding of why, without a full understanding of the statistical variation of properties even within a fairly tight spec, it becomes more and more likely that minor or unforeseen changes will result in inadequate service. And that is not even getting into the use of somewhat hazy correlations such as Charpy values to any of the explicitly defined methods of fracture toughness/crack growth resistance. Or the use of a powder-derived base stock for a previous wrought product, or using wrought product defined chemical specification and properties but producing an item by casting. The more approximations, the less precise the final answer.
Engineering is often a compilation of what has been good enough in the past, with the unknown or poorly defined covered by factors of safety. But this has led to people who do not know how much they do not know, pulling values out of a handbook or old purchase order, and semi-blindly charging forward. This works only as long as the tolerances of what is required and factors of safety in design (room for ignorance of the unknown) are adequate to cover the conditions present in manufacturing and service. So Sean's response of "we'll just have to test it out and see what we get" is totally valid without specific experience to back up a guess, and the hard part is quantifying the cost to a manager who asks why not just accept it. What is the cost if you decide to go forward on insufficient data and then production fails to meet cost/schedule/quality or, usually worse, the delivered product fails catastrophically?
------------------------------
Paul Tibbals
------------------------------
Original Message:
Sent: 03-03-2022 12:10
From: Sean Piper
Subject: Qualitatively predicting Charpy impact toughness
I wanted to reach out to the community for input on whether anyone knows of a reliable means for predicting Charpy impact toughness as a function of composition, thermal processing, etc. in steels. For comparison, things like hardenability and hardness are very well-established in the literature; there are formulas for DI as a function of composition, TTT diagrams, charts for hardness vs. tempering temperature, etc. However, Charpy's seem to be a little less predictable. Some grades are just known to have good or poor toughness (think F22 vs. 1095) and on some level this is directly attributable to obvious variables like carbon content or extent of martensitic transformation but sometimes it gets more nuanced. For example, 4130 is not typically thought of as an optimal material for Charpy's due to the carbon content but we produce QT 4130 parts for an application with 237 HBW max hardness, which necessitates tempering very hot, and as such we can achieve acceptable Charpy's even at -75F. However, there is an inflection point where if you temper colder to achieve a slightly higher hardness, you will fail Charpy's in the single digits. Of course, you could further complicate the equation by factoring in thick section sizes, modified chemistries (a small vanadium addition, for example, which is quite common), etc.
Usually when people come to me asking about whether a given material can meet a certain toughness requirement, unless it's an obvious yes or no based on the grade, my default answer is "we'll just have to test it out and see what we get" but I wish I had something better to offer since that can be costly in both money and time and it poses a catch 22 if the question comes up in something like a quote review where we'd need to commit to buying a heat of material, tooling, etc.
------------------------------
Sean Piper
Product / Process Metallurgist
Ellwood Texas Forge Houston
Houston TX
(713) 434-5138
------------------------------