Date of Award


Document Type


Degree Name

Master of Science (MS)


Chemical Engineering


The Great Plains Gasification Plant located in Beulah, North Dakota, operated by the Dakota Gasification Company (DGC) is the only commercial production plant in the United States making synthetic natural gas (SNG) from coal. DGC operates 14 Lurgi gasifiers to produce 158 million ft3/day of SNG from 17,000 tons of North Dakota lignite coal. As the first step in constructing a model of the entire plant to control its operation and optimize the economic performance, an ASPEN/SP computer model (called RGAS) of a Lurgi gasifier was developed with the combined effort of several researchers. RGAS will predict the impact of changes in inputs on production rates and efficiency.

The model parameters (i.e., kinetic constants, heat transfer coefficient to the reactor wall, heat capacity of the volatiles, etc.) were previously optimized in different stages to obtain the best possible model predictions. Unfortunately, the model did not predict the output variables within desirable accuracy, necessitating further improvements to the RGAS model. In this study some model improvements were made or tried.

The FORTRAN routine, which models the combustion and gasification zones in the gasifier, requires an iterative approach for solution. The efficiency of the FORTRAN code has been enhanced by changing the convergence scheme, which reduced the simulation time by 75%. In addition, a Hooke-Jeeves pattern search algorithm has been included in the RGAS subroutine URE09, so that any optimization of parameters can be done automatically. This made optimization easy and efficient.

Contrary to the actual volatile evolution, RGAS assumes devolatilization to be a linear, temperature dependent evolution of volatile matter. However, limited data on North Dakota lignite and available literature information indicate that the temperature dependence of volatile evolution is non-linear. A non-linear devolatilization model was tried, but it did not result in any significant improvements in the RGAS predictions so, the simple linear model was retained.

An optimized value of 1.7 for the activity of carbon was included in the definition of the equilibrium constants of the reversible reactions (previously it was assumed to be unity), which improved the flow predictions significantly. This can be justified by the fact that amorphous carbon from lignite coal has higher free energies than graphite.

This research project has been very successful. After the improvements were incorporated into the RGAS model, it predicted nine of the ten responses studied (the exception being reactor steam utilization) within the accuracy of the data.

The RGAS model is now complete, in terms of giving good predictions for the coal (lignite) used during the study period. However, kinetic parameters vary with coal composition since several of the reactions are catalyzed by the metals present in the coal. The addition of metal composition catalytic effects would allow predictions to take into account changing coal composition. This is recommended as the next step in making the RGAS model even more valuable as an optimization and control tool for the DGC plant in Beulah, North Dakota.