PSI - Issue 14

Filin V.Yu. et al. / Procedia Structural Integrity 14 (2019) 758–773 Filin V.Yu, Ilyin A.V. / Structural Integrity Procedia 00 (2018) 000–000

769

12

that is about triple crack tip opening, and for this point assumed to serve the most probable fracture origin, parameter Q should be evaluated. The test results obtained with deep notched SENB specimens appear conservative because HRR solution is nearly met for them ( Q  0). FEM simulation and experimental trials have been undertaken and proved that tests at tension may not provide for enough conservatism in estimates of fracture toughness for welded structures as it is shown in the Figures 5 to 7. Experimental results obtained with FCAW and SAW butt joints of steel grade F500W include fracture toughness of large shallow-notch SENT specimens simulating flaw-containing structural elements (crack front length 220 mm, fracturing load up to 7.7 MN). A typically high scatter of fracture toughness data however allowed to conclude that SENT specimens even with a deep notch may overestimate the fracture toughness of material in the structure that is an error to the unsafe side. Another contradiction to the use of SENT specimens for the certification of materials is that nobody in advance knows the real type of service loads as the material may be applied for different purpose. One more outcome from Fig. 7 is that square-section SENB specimens may be applied instead of higher size rectangular SENB ones. The same was earlier noted by Ilyin and Filin (2015). The use of half-thickness specimens is advisable for extra high thickness S only when an expected flaw size is considerably less than this full thickness. Several points need to be underlined: First, the number of welded test specimens should be enough to get the minimum of three, better five valid results for one assessment case and temperature, so each set should comprise 7 specimens as minimum. A 20-year practice tells that about a half of successful results is obtained when we need to sample the grain-coarsened heat affected zone that is usually the most critical in respect of fracture toughness. The criterion related to the necessary portion of target structure sampled by a crack front was checked by Ilyin at al. (2009) based on a special experiment. Second, numerical experiments following Monte-Carlo method allowed Ilyin at al. (2016) to suggest a calculation procedure for a welded joint containing areas of significantly different structure and fracture toughness. Third, fracture toughness data within the ductile to brittle transition temperature interval now can be extrapolated to higher temperatures from the lowest temperature. They also can be interpolated between N test temperatures T i : ln J c = CT + D , (32)                         N i N i N i i i N i N i i i T N T J N T J T C 2 2 1 c 1 1 c ln ln ,                              N i N i N i N i N i i i i i N i i T N T J T T T J D 2 2 1 1 1 c 2 c 1 ln ln . Fourth, a statistical procedure is now available for fracture toughness data interpretation related to the ductile to brittle transition interval. Data is restricted from above, by the formula (29), and then treated to derive a variation coefficient V accounting for scatter with N data sets as part of the reserve factor 3.5 Statistical reliability and interpretation of experimental data.        i i 1 1        i i 1 1

n

i

1   j

  c J

2

j

1 V V N N i i   

J

c max J

by restricted

V

1

,

.

(33)



V









i





2

3

exp

 CT D n

i

i

Additional “penalty” is assigned in case of less reliable test data available (e.g. with subsize specimens, or even Charpy impact test results only). All the above relates to a particular material, thickness and welding parameters. When weldability is assessed, the minimum and the maximum heat inputs should be examined.