A516g70 low impacts after welding

[Itsnotgoinganywhere]

We attempted to qualify 3/8" thick 304L to A516g65 welded with E309LT-1. Everything worked fine, but my CVN tests in the 516 base metal were 10, 7, 7 ft-lbs at -20°F. The MTR for that plate shows 140s at -50F, and we had another sample from the same plate tested (unwelded) and got 130s. The 304 (147ft-lb), weld metal (27ft-lb) , and both HAZs (28-35ft-lbs) were acceptable. I can't understand what happened. The lab took another CVN sample of the base from a different location turned 90° thinking we had the coupon grain direction wrong, with similarly poor results. It was not a typo error; they said the specimens showed very brittle fracture appearance.
Preheat: 100F
interpass: 340F
No PWHT.
coupon size: 14"x6"x3/8" each plate, welded on the long seam. ~60° single v-groove.
heat input: ~16-18 kJ/in
Position: 1G
Root weld, hot pass, then backgouge/backweld, then two more layers on the top.

Any ideas would be appreciated. My next thought is to attempt with GMAW-S using ER309L, but I'm somewhat hesitant without understanding what happened here.

 

[metengr]

I believe you had some issue with excessive heat input above (not that I don't believe your numbers) and development of coarse grains in the base metal heat affected zone. This would be the reason for degraded notch toughness behavior in an otherwise fine grained steel.

Get the lab to evaluate the grain size in the base metal HAZ.

 

[Itsnotgoinganywhere]

Thanks, I'll see if we can get that done. Wouldn't the HAZ impacts have suffered similarly if that were the case?

 

[metengr]

I would suspect the poor impact properties were from the base metal HAZ and not the base material itself. The weld region consists of the fusion zone and back to the edge where unaffected base material begins.

You claim you tested unaffected base material. Why would your results be any different in unaffected base material away from the base metal HAZ? I think your results are from the base metal HAZ which is coarse grained.

 

[weldstan]

Were the Charpy coupons sub size? As per metengr, check the actual grain size of the SA-516 material.

 

[Don56]

Since the CVN's in the HAZ of both materials were better than the A-516-65 base metal, could the MTR results be based on normalized tested material and the actual material for the coupon be as rolled. Read the MTR carefully. Just a thought.

 

[Itsnotgoinganywhere]

Grain size evaluation is going on this week; hoping for answers soon from that.
CVNs were 10x7.5 (for my test and on the MTR), but that still doesn't get them up to a reasonable result after conversion.
MTR states normalized, and the second unwelded coupon we tested from that plate got good results.