This paper summarizes results from Phase III of the joint industry program (JIP) on refinery alkaline sour water (ammonium bisulfide) corrosion. Phase III included a comprehensive engineering analysis of data from all three phases of the Sour Water JIP and the development of “H2S tie-in plots” based on a single parameter (H2S partial pressure), replacing the isocorrosion diagrams based on NH3 partial pressure developed in Phase II, and linking with the H2S isocorrosion diagrams based on H2S partial pressure developed in Phase I. These H2S tie-in plots enabled corrosion rate data for carbon steel and five commonly used corrosion resistant alloys to be contiguously modeled across the entire range of conditions tested ― six orders of magnitude of H2S partial pressure, from approximately 0.00004 to 150 psia (0.0003 to 1,000 kPa absolute) ― to fully characterize effects of H2S-dominated, NH3-dominated and intermediate sour water conditions. The results clearly demonstrated that NH4HS concentration, flow velocity (wall shear stress), H2S partial pressure, and free cyanide concentration in the NH4HS solution are the key parameters driving severity of refinery sour water corrosion. The effect of temperature was also clarified. Substantial differences in corrosion behavior of carbon steel from the corrosion resistant alloys were observed in low NH4HS concentrations at intermediate H2S partial pressures, where corrosion rates of carbon steel and all the alloys exhibited surprisingly significant dips and peaks on the H2S tie-in plots. The results and analyses of data were synthesized into an enhanced software tool to predict corrosion rates for the six materials (carbon steel, stainless steels, and nickelbased alloys) evaluated in the program, addressing a broad range of environmental conditions encompassed by all three phases of the Sour Water JIP.
Key words: sour water, ammonium bisulfide, NH4HS, hydrogen sulfide, H2S, ammonia, NH3, partial pressure, tie-in plots, temperature, cyanide, velocity, wall shear stress, materials selection, corrosion prediction model, software