Validation of Charpy Impact 1

[BVMan]

We were recently asked to Validate Charpy Impact Values of an A352 LCC Finished Machined Valve Body Casting. The Mill Certificate gave an average of 24J @-46Deg C with a minimum single reading of 23J.

After taking 10mm x 10mm Samples across the flange (Between the Bolt Holes) and carrying out tests at -20Deg C we could only get results of 7, 9 & 10J

Can anyone comment as to whether this is normal and why or is something seriously wrong?

 

[metengr]

ASTM-A352 is the Standard Specification for Steel Castings, which specifies a quench and temper heat treatment for low temperature service. Under section 7.2, impact testing is required for each heat of material.

I reviewed A-352, and for Grade LCC at -46 deg C, the minimum Charpy impact value is 16J for a single specimen, and 20 J for an average of 3 specimens or a minimum value between 2 specimens, per Table 1. At -20 deg C, you should be at or above this value with your test results.

This is not normal, and I would question the heat treatment of this casting. Especially review the method of quenching because liquid quenching is recommended. I would contact the mill, and discuss your nonconformance test results.

 

[Carburize]

It is actually fairly normal to see significant differences in Charpy (and other properties) when sample are taken from a part compared to the properties measured on a separate cast test block. If the original mill certificate numbers were obtained from test block material the difference would not be a surprise to me.

 

[metengr]

Carburize;
I would agree with your statement regarding variability. However, in this case, the castings clearly did not meet the minimum Charpy impact requirement specified in Table 1 of SA-352. I would treat this as a nonconformance and find out what is going on. I suspect that the casting may not have been quenched and tempered.

 

[BVMan]

Thanks for your help so far, I have checked the Mill Cert and it clearly states "Normalised and Quenched". I have noticed the following in A352

X1.4.3 The literature shows that notched bar impact strengths in constructional steels differ widely in accordance with many variables. Consequently, there is bound to be some discrepancy between an individual test bar and the entire part that it represents. No system of test bar selection can guarantee that every sample would meet minimum requirements. Test bar selection must be a compromise to generally represent the product.

Which I guess mitigates the difference some what. I'm not sure my client will see it that way though!!

 

[Carburize]

There are always issues raised when routine quality control testing is based on one way of doing things and then tests are made in another way. In standards such as API 6A for example it is clearly identified that certain QC values referenced for mechanical properties are based on separate test coupons and that if a part is tested it will be different but the test coupon number is what is to be used to certify the product.
This statement is made because in the particular application the requirements are based on the correlation between service performance and test coupon data.
Other codes are based on the correlation of service performance and data determined from prolongations attached to parts and yet others on the determination of properties from sacrificial parts.
Before product can be determined as not meeting standard it is necessary to know exactly how the product is usually certified.

 

[stanweld]

BVMan,
The test results are not surprising; futhermore, your validation methodology is invalid in terms of A-352 and A-703 testing requirements with regard to sampling. Cast keel blocks or prolongations from the original casting are used with dimensions per A-703. Your sampling method at probably the thickest section of the casting after significant machining will assure the low impact values obtained.

 

[EdStainless]

You can do a couple of thing quickly. Look at the micros for you samples. Are they correct? What is the grain size?

The other thing is to re-heat treat some impact samples (before they are machined to finish size) and then test them. That should answer the heat treatment questions.



= = = = = = = = = = = = = = = = = = = =
Corrosion never sleeps, but it can be managed.

http://www.trenttube.com/Trent/tech_form.htm

 

[metengr]

Stanweld/Carburize;
As always, your points are well taken, especially the validation approach that BVMan performed in comparison to the requirements in A-703. Going back to A-352, there is a statement under heat treatment indicating that.. "liquid quenching of the ferritic grades is normally required to meet the mechanical property requirements of heavier sections (of which I would consider the flange) and will greatly enhance the low temperature properties of the thinner sections". In addition, A-352 provides two options for heat treatment of the castings - normalize and temper OR liquid-quenched and tempered condition. The valve size or class was not mentioned in the original post, so it could be that the flange thickness would prohibit formation of a fine grained, tempered microstructure with the normalized and tempered condition.

Despite the technique performed by BVMan, the concern that I have is the much lower than anticipated Charpy impact values even at -20 deg C (which in my opinion should have at least met the minimum Charpy impact requirement in Table 1 of 16J at -46 deg C). If the in-situ Charpy impact test values would have at least met 16J at -20 deg C, I would agree that with the thicker portion of the casting (the flange), you will have variability in Charpy impact test values. However, the range of 7 to 10J at -20 deg C is still too low for LCC cast material. To me, this raises a big red flag to investigate further. Perhaps the casting heat treatment should have been specified as liquid-quenched and tempered?

 

[BVMan]

Metengr

They are 6" ANSI 300RF Gate Valves

Regards
BVMAN

 

[arunmrao]

metengr your comments are very valid. The discussion would have been better if the class and size of valve casting were known. I have encountered similar problem in CA6NM castings where the integral test bar and sample drawn from thick section of a runner blade differ for impact properties.

In all probabilities the test bar and casting have been heat treated separately.

BVMAN please let us know the final decision taken. As it is finish machined now. Also why was the test not carried out immediately after receiving the rough casting? Normally the external inspector for cryogenic applications is quite nosey.

 

[metengr]

BVMan;
For a 6" 300RF gate valve, you should have easily met the stated Charpy impact requirements at -20 deg C, and certainly at -46 deg C. I would agree with Edstainless, and evaluate the microstructure. It sounds to me that the heat treatment was not correct for this valve.

 

[stanweld]

metengr,
Quench and tempering should have easily produced a mircrostructure that would meet -20 C impacts unless sampling was done at riser/gate locations. I also agree that the microstructure of the charpy test specimens should be evaluated. I suspect that, if appropriate cleanliness is observed, Weidmenstatten structur will be found. I have seen this problem on both LCC castings and LF2 forgings similarly heat treated in the past and needless to say the impact specimens taken from these area did not provide the specified absorbed energy.

 

[nickt1960]

May I ask which mill the valve came from, or as importantly was it manufactured and tested in Korea, China or Eastern Europe?

 

[BVMan]

Western European Manufacturer, who only provides a breakdown of the Chemical and Mechanical results on their own cert. However we believe that the casing is of Chinese Origin. We have asked the Lab to take a look at the microstructure. I'll post back to let you all know the findings.

 

[grampi1]

Look at the composition, even if it was Q&T does not mean fine grain nor low S. Is there any Ti, V or Nb or help control grain size? Is there still significant casting segragation? Did the charpy show any casting porosity defects that would contribute to low values?

-46C is difficult to meet without very good control of both composition and heat treatment. Normalize and temper was probably sub optimal.

It is also possible the official cast test block was heat treated differently and what you have is a quality control screw up.

We ran into a case where a double Q & T was required to get impacts into the range we wanted.