Low HI = faster quenching from welding temp.  What is the hardness of the HAZ?  What is the material?  You may have untempered martensite or bainite in the HAZ.

We'll need more detailed information before anyone can isolate the source of your problem. What is your base material, and what type of filler metal are you using? What welding technique are you using? Are you performing any post-weld thermal treatments? If you have untempered martensite present in the HAZ, then a post-weld thermal treatment will be necessary to temper the martensite back. Are there any physical constraints placed on the material during welding that would induce significant stresses? What is your cooling rate? MetalGuy has made some good suggestions. Check hardness values in the HAZ as well as the base metal to see if there is a significant difference.


                                             Maui

Here is a little more background info:

The plates are cast material (rules out rolling at the mill) with the following composition:
C = .072
Mn = 1.05
Si = .35
P = .005
S <.001
Mo = .21
Cr = .20
Ni = 1.16
Al = .008
V = .005
Cu = .17
Ti = .004
N = 62 ppm
O = 46 ppm
They are normalized to reduce segregation, then austenitized, quanched and tempered.
To qualify the plate, we have to do CTOD tests at 14F (-10C) in the coarse grain HAZ and the etched HAZ boundary (border between the sub-critical and inter-critical HAZ) and get .015" minimum value. This is our first time trying to qualify material. The first test we did we ended up with Martensite islands, so we dropped the carbon and added a little more Mn and Ni to make up the hardenablity. The current test was evaluated up to 1000x and the HAZ is all ferrite with a little bit of second phase that the metallurgist said was either pearlite or bainite.  He said that microsturturally, under an optical microscope, there was nothing out of the ordinary as compared to CTOD tests that pass.  We also added a little Ti to help produce Ti nitrides, but we didn't get as much as we wanted. So on this current heat of material, we have good results at both locations in the high heat input test, and bad results in the low heat input test at both locations. Evaluations of the coarse grain specimens from the low heat input test seem to indicate that the weld metal may have been involved with the lower values, but nothing I have found can explain the low values in the low heat input test sub critical/inter critical test area except for strain ageing, since that should be the location affected most by that phenomena. Coupled with the fact that the Ti content was lower than we wanted, could have produced free nitrogen, which promotes strain-age embrittlement.

We also just got a hardness scan report done in HV 1kg.  Weld metal was: 245, 231, 222, HAZ was: (proceeding from fusion line to base metal) 320, 260, 255, 240, 210, 190, 190.  Base metal was: 208, 201, 208

Also, no PWHT was done, nor allowed.

The plates have to be fully restrained to prevent angular distortion.  Joint is a single bevel.  The qualification specifications limit heat input to 1kJ/mm, and 212F (100C) max interpass temperature.  Welding is done with FCAW using stringer beads only.  Welding material is 1% Ni to match base metal strength and toughness ideally.

Hmmm, lots of good data!  It's after midnight here now, but my first thoughts on this are that you've got to reduce that 320 hardness zone somehow.  You've got a fairly deep hardening steel there, and the N is a real killer wrt the brittle/ductile trans. temp., at least in pearlitic steel.

Is that low interpass temp. a requirement because of distortion?  If you can get past that big hurdle, you might be able to preheat the steel up to where it should be, ~250 deg F or higher.  That will help avoid any bainite/mart. formation which you must do since you can't PWHT.

Ti also hurts (raises) the brittle/duct. trans. temp., again in pearlitic steel, but you don't have much.

I'd hate to have your requirements on something I was doing-sounds like a real torture test!

The specifications (either API RP2Z, or EN10225) limit the maximum interpass temperature to 212F max.  We have welded this steel for years without preheat though, and never have had any problems, including CVN tests.  However, the CTOD tests are more sensitive, so we had a few problems the first time with martensite, so reduced the carbon to the level listed above.  All the literature I see on CTOD tests say the drop off is for CTOD results is about .08 to .1 C, but I have seen sucessful results with at least .13C also.  As you can see from the Chem, the N is not too bad as an overall level, but the ratios to Ti and Al are not that great, and I am trying to figure out if that is the problem or not.  We are going to start machining specimens for a strain-age test tomorrow, so I may know soon.  We haven't done that much with Ti, but a lot of steel producers add it to reduce the grain growth in the coarse grain HAZ with the TiN.  I have seen info indicating you can pass without it, but as you said, we don't have that much anyway.

Doubt the problem is strain aging.  Neither Ti nor V are enough to really help grain refinement.  N is normal range.  

Is preheat rigorously being checked if the fill is stopped partway and then restarted the next day?  
Presume interpass is being allowed to drift as close as possible to max allowed.  

If you used 80C preheat SMAW 1KJ/mm, then bead on plate hardness should be about 301HV with 60%Martensite.  Since you found 320, your probably using a little lower preheat and what your getting is inadequate interpass tempering at bead to bead overlap sites along the FL.  Using smaller bead size would allow more interpass temepring which is what is needed here I think to get the FL toughness up.  

Check out

http://member.nifty.ne.jp/yurioka/p03/p03.html

Also note type of CTOD specimen configuration will influence test result with compact tension having the highest restraint and so yielding the lowest CTOD value.

grampi1,
Using your link to find N. Yurioka's interactive site
'Weldability Calculations' rated a star.
http://member.nifty.ne.jp/yurioka/exp.html
Contents:
Calculation of carbon equivalents and transformation temperatures
Welding Condition Input and Calculation of Thermal History.
Estimation of maximum HAZ hardness
Determination of Necessary Preheat Temperature

Actually we got some results from a simple strain-age test we did last week.  We strained a tensile bar 5%, then aged them.  We also did some with straining and no ageing, and some without straining or ageing to compare results.

We did 3 CVN tests for each condition on the material question and these are the results at -40F:

No Strain & No Age:
226, 151, & 143 ft-lb
(173 ft-lb average)

Strained 5% & No Age:
32, 154, & 105 ft-lb
(97 ft-lb average)

Strained 5% & Aged at 450F for 2 hours:
28, 43, & 9 ft-lb
(26 ft-lb average)

I also found your link informative.  I got a 311HV5 calculation with 67% martensite.  Does the page list a range of compositions that it is good for.  This alloy was particularly chosen so that we wouldn't get any martensite in the HAZ, and the HAZ has recently been examined up to 1000x optically, and the metallurgist doing the work said it was all ferrite and bainite.  I can't recall off the top of my head, but what is the maximum hardness from baininte?  Obviously, the 320HV isn't from ferrite.

Bainite has a hardness in the low 300's.  A paper by some gurus at the Colorado School of Mines reported a low alloy steel Bainite be 313 HV.  They quoted a Literature value of 305 HV.  They reported Martensite to be 428 HV, with a Literature value of 440 HV.  Do you know how the microstructure was presumed to be ferrite + bainite?  Color tint-etching produces the best results, which often reveals a mixture of martensite and bainite together.

Evaluating microstructures is something I would like to learn a lot more about.  I in this particular case, I have not found out what methods were used for etching or evaluation.  The main thing they were looking for at the time though, if there were any large precipitates at grain boundaries, we didn't really expect any, but it was worth checking, and they didn't find any.  The evaluation was done by a metallurgist at EWI.

As far as Grampi1's comments on reducing heat input to help with tempering of previous passes, on our 4" thick plate, I would worry about fusion defect if the heat input was lowered much further.  Also, the high heat input tests (3.5kJ/mm as opposed to 1kJ/mm for the low heat input test) was sucessfull on this composition.  The welding was not stopped before completion due to schedule, but the plate was checked along its length with a calibrated contact pyrometer.  

Based on the chemistry and high rate of cooling, some martensite and fine bainite should be expected.  

Whereas grain size affects toughness properties, have you considered a fine grain melting practice for this material?  Fine grain melting is most often achieved by adding > .02% Al, .03-.04% V or > .015% Nb or combinations thereof.  Austenitizing at normalizing temperatures prior to quenching & tempering is also recommended, assuming that you may have austenitized at a lower temperature.