SA612 at -40°C: Why a Certified PV Grade Still Fails Low-Temp Impact

A Certified Grade That Still Failed at -40°C

ASME SA612 is a killed carbon-manganese-silicon steel written for welded pressure vessels in moderate-and-lower temperature service. On paper it looks unbeatable: minimum yield 345 MPa, tensile 570-725 MPa, and a supplementary CVN requirement of 27 J average at -40°C, supplied normalized or normalized-and-tempered. Yet in one mill’s batch production, SA612 plates of 12-20 mm thickness returned -40°C impact values scattered between 18 J and 75 J, and the rejection rate reached 15% across July-October 2025. The grade was correct. The steel was not.

This gap between a certified name and a failed coupon is the most useful lesson a vessel buyer can learn. A standard guarantees chemistry and a baseline test; it does not guarantee that the specific heat arriving at your fab shop will deliver the toughness your service temperature demands. The difference is made upstream, in the casting pit and the heat-treatment furnace — and almost none of it is visible on the mill certificate.

SA612 on the International Standard Map

SA612 is an American grade, but it is bought and substituted worldwide. Its cousins under European, Chinese, and Japanese codes look like safe swaps — until you read the fine print on impact testing. The default low-temperature acceptance temperature is not the same across these grades, and that is exactly where substitutions go wrong.

Read that last column carefully. A buyer who substitutes Q345R or SPV355 for SA612 on a -40°C duty, trusting the yield match, inherits a grade whose standard only asks for -20°C or 0°C impact. The substitution is legal on paper and unsafe in service. If you change the grade, you must re-specify the low-temperature acceptance — the number, not the name, is what protects the vessel.

The Two Faults Behind the Scatter

Microscopy on the failed plates told a consistent story. The fracture surfaces were intergranular, lined with MnS inclusions, and the manganese content swung by more than 0.3% from point to point. Two independent faults were working together:

1. Center segregation of manganese. In the as-cast slab, the central zone carried 35% more carbon and 42% more manganese than the base metal. Continuous casting left a center segregation band — a streak of Mn-rich, Mn-poor material (1.7-1.9% vs 1.1-1.3% Mn) rated at band level 3 per GB/T 34474.1-2017. That band is where cracks start, and its elongation runs 15% below the matrix.

2. An under-powered normalizing window. The original 850°C normalization left the center as ferrite + pearlite + stray bainite, with grains of 45-55 µm and a continuous carbide film along the grain boundaries. That film is the textbook cause of intergranular brittle fracture at low temperature. Carbon control was also loose: six heats sat at the 0.18% upper limit, and every one of them averaged only 28 J at -40°C.

Neither fault is visible on a standard certificate. Both are decided by process discipline the buyer never sees.

The Composition Targets That Fixed It

The corrective work started in the melt. Tightening the chemistry ranges — the paper’s controlled composition table — removed the root cause before rolling began. The elements that mattered most:

In casting, dynamic soft reduction (1.2-1.5 mm) pulled the center segregation grade from 1.5-2 down to 1-1.5, and raising the secondary cooling water from 0.8 to 1.2 L/kg lifted the equiaxed-crystal ratio from 30% to 50% — directly attacking the banded structure. None of this changes the grade; it changes whether the grade performs.

The Heat-Treatment Window Is the Real Lever

Three normalization-and-tempering routes were compared on 5-plate sets sampled at 1/4 thickness. The result is the clearest argument in the whole study: same grade, same chemistry envelope, dramatically different toughness.

The tempering temperature alone moved impact energy from 28 J at 500°C, to 109-133 J at 620°C, to 130 J at 720°C — but 720°C collapsed the yield to 320 MPa, below the 345 MPa floor. The window is narrow, and 620°C sits at its peak. Thermo-Calc modeling confirmed 620°C as the precipitation peak for V(C,N) and NbC, the fine particles that pin boundaries and lift toughness without sacrificing strength.

After optimization the ferrite grain shrank from 45 µm to 12 µm (ASTM grain size 8-9), pearlite spacing tightened from 1.2 µm to 0.8 µm, and the impact dimple depth grew from 5 µm to 15 µm — a ductile, energy-absorbing fracture replacing a brittle one.

What Buyers Should Actually Specify

The study’s real message is not “SA612 is risky.” It is that low-temperature toughness is a process outcome, not a grade attribute. A mill that controls center segregation, holds the normalization window at 880°C, and tempers at 620°C will deliver SA612 that passes -40°C comfortably. A mill that ships the certificate and skips the furnace discipline will hand you 18 J plates inside a 75 J certificate.

For your next low-temperature vessel, specify the physics, not the name:

• State the delivery condition explicitly — N or N+T — because as-rolled SA612 is not acceptable for PVQ service.
• Write the actual -40°C CVN acceptance into the purchase order, and require it on every heat, not as a footnote.
• If you substitute a grade (Q345R, P355GH, SPV355), re-state the low-temperature impact requirement — do not inherit the substitute’s warmer default.
• Ask for grain-size and segregation data; ASTM No. 5 fine grain or better is the practical guarantee of stable toughness.

A grade name gets a plate through the door. The casting and heat-treatment discipline behind it is what gets the vessel through twenty winters.