Refiners today are increasingly dealing with more corrosive feedstocks that present new demands on valves in the process. However, maintenance personnel can detect problems before they become major issues by using different testing methods.
In 2009, a Minnesota-located refinery was experiencing through-wall leakage with several Class 300 rotary valves tasked with controlling crude unit vacuum prefractionator charge heaters. Maintenance personnel removed insulation from the valves and noted coke residue on the surface of the valve bodies, which identified for them where the leaks were occurring (Figure 1).
Fortunately, this particular situation did not result in a safety event or a fire. Instead, it illustrates how the refining industry’s use of increasingly corrosive feedstocks can combine with unknown casting defects to cause potential valve problems.
THE INVESTIGATION/ EVALUATION
Five valve body castings, all grade CW2M (cast alloy C), were returned from the refinery to the manufacturer for evaluation using both nondestructive and destructive tests.
The nondestructive tests performed on the returned castings included visual examination, pressure testing and liquid penetrant (LP) examination.
Visual examination did not reveal any signs of corrosion on the wetted casting surfaces of the returned valves, which included both as-cast and machined surfaces.
The pressure testing used water at 1125 psig (7.76 megapascals or MPa) followed by helium at 150 psig (1.0 MPa). LP testing was performed to the requirements of ASTM A903 Level III, which considers linear and rounded indications exceeding 3/16 inch (4.8 millimeters) to be relevant. The helium and hydrostatic water tests revealed no leaks. Apparently, the hydrocarbon or potential coking sealed the leak paths once the valves were brought down to ambient temperature and removed from service.
LP testing of the castings revealed indications on the exterior surface—a lesser number of indications were on the interior, cored surfaces. These indications were on both as-cast and machined surfaces. Most of the LP indications were shallow [<0.02 inch (0.5 millimeters)] and removed with minor grinding. However, on the neck area of some castings, grinding did not remove the indications. The defect progressed through the wall of the casting and could represent a leak path. (Note that all of the returned valves had through-wall leaks in the neck area.)
Destructive testing, including metallographic examination, scanning electron microscopy (SEM) and energy dispersive spectrographic analyses, was performed on several of the returned castings.
For example, corrosion testing was done on one valve body casting to compare results to maximum acceptance values previously set for this material grade. The test environment was boiling ferric sulfate-sulfuric acid per ASTM G28 Practice A. Although this standard is intended for wrought product forms, the practice A test method is also useful for castings. The corrosion rate was 140 milli-inches per year or mpy (3.5 millimeters/year or mm/y) compared to the manufacturer’s acceptance criteria of 360 mpy (9.1 mm/y) maximum. The grain boundaries were visible, which is typical for this grade. Of importance to this issue is the fact that this is a very severe test that does not represent actual applications for CW2M but does detect susceptibility of weld metal and heat-affected zone to intergranular corrosion attack.
Fracture surfaces from the neck area of the same valve body were examined using SEM (Figure 2). The first zone examined was an outer layer that was approximately 0.03 to 0.06 inch (0.8 to 1.5 millimeters) thick. This chill layer solidifies rapidly when the molten metal first contacts the mold surface.
The second layer or zone 2 of the fracture has a dendritic pattern. This pattern represents columnar grains that grew perpendicular from the chill layer during the balance of the solidification process. The neck was cast solid so all the grain growth was from the outer diameter to the center. The center of the neck was then bored out during machining of the casting.
Zone 3 was ductile, dimple shear. This is the fracture mode seen on any mechanical break produced in the lab.
Within the dendritic zone of some of the fractures were rounded dendrite arms (Figure 3). The rounding indicates a shrinkage defect caused by an isolated area of liquid that was frozen off from the riser system because it was not properly fed. When the liquid was consumed during solidification, the dendrite arms could not continue to grow, leaving the blunted tips. Dark areas, which were identified as oxide films formed from exposure to air during casting solidification or heat treatment, were also found on the lab fractures. This proves the fractures were present before the casting went into service.
Inclusions and porosity also were visible. The inclusions were mold sand, slag and oxides. The grains were very large compared to wrought product forms—about 1/4 to over 1/2 inch (6 to 13 millimeters). The grain boundaries were the intersection of the dendrites formed during solidification. One leak path was cross-sectioned (the photomicrograph in Figure 4 shows the continuation along a grain boundary).