Evaluation and analysis of coking in coal liquefaction preheaters

The deposition of carbonaceous solids (coke) in direct liquefaction fired preheaters is a common technical problem experienced by many of the liquefaction processes of interest [including the Solvent Refined Coal Processes (SRC-I and SCR-II), the Hydrocarbon Research Process (H-Coal), and the Exxon Donor Solvent Process (EDS)], yet the causes for such behavior are not fully understood. This study is aimed at providing an insight into coking behavior by evaluating the influence of process operational conditions and coal-solvent properties on the deposition of coke in preheater coils. The Wilsonville, Alabama SRC-I pilot plant, having started operations in 1974, is the largest and most complete source of data for runs during which both coking did and did not occur in fired preheaters. Using, as a source of guidance, the mostly qualitative results obtained from a survey of the literature on coke formation from some coal-oils but primarily petroleum liquids, and a proposed incipient coking correlation based on coking occurring when the concentration of key coking precursors reaches its solubility limit, the nineteen coking data points obtained from the Wilsonville facility were correlated using multiple regression techniques with correlation coefficients (R^-values) as high as 0.95 achieved depending on the combination of process condition factors and coal-solvent properties chosen. Developed incipient coking curves serve as a boundary between a non-coking domain and a coking domain. Those factors statistically indicated to influence coking tendencies are, in order of importance: a) estimated preheater mean slurry residence time, b) inlet hydrogen partial pressure, c) combined coal oxygen and nitrogen content, d) coal vitrinite content, e) coal sulfur content, and f) solvent quality (short) as determined by the Wilsonville facility. Results indicate that increases in slurry residence time, combined coal oxygen and nitrogen content, or coal vitrinite content tend to promote coking tendencies, while increases in inlet hydrogen partial pressure or coal sulfur content tend to suppress coking tendencies. The reported value for solvent quality was usually statistically insignificant and thus its exact role, if it is important at all, is unclear. The inclusion of additional data indicated by the literature to influence coking, but not available from the Wilsonville facility, such as coal hydrolic oxygen content and non-basic nitrogen content as well as complete coal petrographic data, are anticipated to further refine incipient coking correlations. Also, further refinements should be achieved as better methods to estimate slurry mean residence time in multi-phase flow through preheater coils are developed and as methods to calculate the hydrogen partial pressure locally in preheater coils, through the determination of vapor-liquid equilibria data for coal-derived multi-component solvents, are developed.