PSI - Issue 28

V. Yu. Filin et al. / Procedia Structural Integrity 28 (2020) 3–10

4 2

Filin V.Yu., Ilyin A.V.,Mizetsky A.V.

/ Procedia Structural Integrity 00 (2020) 000 – 000

must be conservative, technically achievable, economically reasonable, and, moreover, quantitatively justified. In-service cracks usually start from welding flaws and stress concentrators of welded joints. A resistance to crack start is known as static fracture toughness measured in terms of CTOD and J-integral. But if such a crack starts, it has to be arrested by base metal. This feature may be described by a crack arrest temperature (CAT) of base metal or its critical stress intensity factor (SIF) K I a at this temperature. K I a and CAT are mutually dependent and may be found from ESSO crack arrest test in full thickness as per ISO 20064. The condition of safe operation is CAT  LAST, (1) where LAST is the least anticipated service temperature, or, in terms of LEFM that may be considered acceptable for low-temperature service materials, Mode I crack, K I a  K I , (2) where K I is a SIF value related to a probable crack in a structure taken as a criterion for ESSO test results, K I a is a thickness independent performance of material expressed in [MPa  m 0.5 ], namely a SIF at Mode I crack arrest usually ascribed to the plane strain condition. In ESSO test K ca may be found instead that relates to the full thickness of metal. ESSO test is not a simple thing. Leaving aside its huge equipment and cost, the following issues have to be noted: a crack path is not usually straight, a front of tunneling crack is not at all straight, so the crack length taken for critical SIF calculation as a distance to the very crack tip is suspicious. The applied stress is well known before a crack start but not at the moment of its arrest. The test outcome K c a is not a real K I a where the last one is a thickness-independent performance of material. Anyway, ESSO test is now introduced into shipbuilding Rules of DNV GL.

Trial correlations of ESSO results are now suggested with the results obtained using special three-point bend samples and pressed-notch Charpy specimens, e.g. by Shirahata et al (2018). In Russian practice of several decades, the tests applicable for certification of ship hull materials include tests to evaluate the ductile-brittle transition temperatures TKB and NDT as well as static three-point bend tests of V-notched Charpy specimens to find a critical temperature TKDS. TKB is a temperature corresponding to 70% ductile portion in fracture of full thick static three-point bend test specimens, see Rules for the Classification, Construction, and Equipment of Mobile Drilling Units and Offshore Fixed Platforms by Russian Maritime Register of Shipping (2018). The first correlation of TKB vs. CAT was suggested by Danilov et al (1991). NDT is a wide-known nil-ductility temperature test by Pellini as per ASTM E208. TKDS test was introduced by B.A.Drozdovsky in early 1950s, TKDS temperature is the minimum one when the load drop in the test record is less than 1/3 of the maximum load that relates to an arrested unstable crack extension and about 70% ductile portion in fracture. Figure 1 shows a TKDS test setup. The above test procedures are rather simple and applicable in conditions of plant laboratories on contrary to ESSO test.

Figure 1 – TKDS test setup So the aim of research is a development of correlations allowing to set formal requirements for the results of small size tests ensuring reliability and safety of structures made of low-alloyed steel. Filin (2019) suggested some correlations for NDT and TKB tests, they may be given in the kind of the following equations applicable for low alloyed shipbuilding steels (all the temperatures are centigrade). For NDT test, ( ) S +  = + 0.44) σ 50 ln 0.226 (0.0005 NDT CAT Y(NDT) , (3) where S [mm] is a full thickness of shipbuilding steel,  Y [MPa] is its yield stress at room temperature,  Y(NDT) [MPa] is its yield stress at the nil-ductility temperature (NDT) that may be found as (4) An alternative formula suggested by Ueda and Handa (2019) on the basis of experimental data may be simplified to 124 170 NDT 140 exp σ Y    − = + σ Y(NDT )  −   .