A combined physical and statistical simulation model for Black Liquor Gasification

In this paper the development of a design model for Black liquor gasification is described, where factorial design is used for pilot plant experiments. The results from the experiments are then used to build multi-variate statistical models using PLS. These models are used to give the correlations between important variables and are later used in the physical simulation model to determine gas composition of the product gas, when operations under different conditions are made. The physical model adds the known physical relations as probable water content of the product gas from the shift reaction, selective absorption in a wet scrubber, heating value of the gas etc. From the simulation model we can then determine what the gas composition and the solids composition will be at different operating temperatures, production capacities, relative oxidation, composition of the black liquor and some other factors. Back ground: During the 90 ABB developed a Black liquor gasification process. The process has as a goal to replace existing Tomlinson boiler in pulp and paper industry with a new process, where up to three times more electric power can be produced from the black liquor, a split of sulfides from the bed solids (primarily sodium carbonate) and elimination of the risk for a steam explosion if there is a break in a water tube in the reactor, something happening roughly twice a year somewhere in the world today, with enormous consequences for human lives and economy. The black liquor consists of roughly 1/3 organics (lignin, hemi-cellulose etc), 1/3 water and 1/3 in-organics like sodium/potassium carbonate, sulfate and chloride. In the process sulfates are reduced to sulfides for reuse in the digestion of wood chips into pulp in the Kraft process. The specifics of the process are that the gasifier is a circulating fluidized bed gasifier. Black liquor is introduced into a fluidized bed with bed particles made of e.g. CaCO3. When air is added, heat is generated and the gasification of the organics of the black liquor starts. Still we only add approximately 35 % of the air needed to combust the organics of the black liquor totally. This is called 35% relative oxidation. Hydrogen, carbon monoxide and methane are the products we want to get in the gasifier, as these can be combusted in a gas turbine, generating huge amounts of electric power. The heat in the product gas after the gas turbine is used to heat steam for a steam turbine, increasing the power generation even further. This is called a combined cycle. We also get a reaction reducing sulfates in the black liquor solids to sulfides, and due to equilibrium constraints also a major part of the sulfides are evaporated as hydrogen sulfide. The hydrogen sulfide is absorbed in a gas scrubber using sodium carbonate and sodium hydroxide. To avoid simultaneous absorption of carbon dioxide, we have developed a selective absorption process, where approximately 20 times more hydrogen sulfide is absorbed compared to the carbon dioxide, with respect to the incoming gas content. That means that twice as much sulfide is absorbed, although the carbon dioxide content is ten times higher in the product gas. The remaining hydrogen sulfide is converted into sulfur dioxide during the combustion in the gas turbine, so no hazardous gas is emitted. In the further development of this process we also added Titanium dioxide to achieve direct caustization of sodium carbonate into sodium hydroxide, a process that is today taking place in a separate step in a lime kiln. By integrating this process into the gasification process we get a smaller apparatus, possibility to gasify at a higher temperature without risk for bed agglomeration and also lower total energy consumption. This add-on is not covered in this article, though. Process description: The process consists of a circulating fluidized bed. In the pilot plant we had two cyclones in series for removal of solids in the gas for recirculation directly into the bed. After these the gas was cooled by a couple of heat exchangers, a bag house removed most of the remaining particles and there after the gas was scrubbed in a two step counter current scrubber. Finally the gas was combusted. The air was heated to approximately 400 C before entering the gasifier, and the black liquor was heated to approximately 120 C, to make it fluent. (Black liquor at room temperature is a solid!). Different fillings were tested in the scrubber, to achieve the goal of selective absorption for sulfide. The gas composition was measured on-line with respect to CO2, CO, H2 and CH4, while H2S was measured by extracting gas through a “Draeger tube”. Nitrogen and oxygen were given from measurements of air feed and knowledge about “organic” oxygen content of the black liquor from chemical analysis of this. Water content was calculated from condensate flow and temperature of extracted gas. Temperature was measured at a number of positions in the reactor. The plant could be operated up to a pressure of 5 bar absolute Experiments in pilot plant: When we started the project the goal was to identify operational parameters such as what was possible max operating temperature, max and min fluidization velocity, max capacity (ton per m2 reactor area and h), minimum relative oxidation or maximum heating value of the product gas, relation between equilibrium calculations and real plant data, where also kinetics was considered, maximum water content possible to use in the black liquor, minimum residual carbon possible to achieve in the bed ash, reduction of sulfate to sulfide and how much of the sulfide was evaporated as hydrogen sulfide, how much of this hydrogen sulfide could be absorbed, and how much carbon dioxide was absorbed “by accident” at the same time. So there were many unknowns to determine. Also operations at atmospheric conditions in relation to pressurized conditions should be found. So we started with identifying a factorial design for the experiments to perform. First some over view experiments were done, to identify what variable were the most important. These turned out to be the black liquor load (ton per m2 bed area and hour, t/m2,h) ,the relative oxidation and the temperature. The full factorial design was focused on these variables, while other variables were done at some few points of the total operational area. In figure 1 the pilot plant set up is seen. Figure 1. Black liquor gasification pilot plant in Vasteras, Sweden. Statistical model using multi-variate data analysis: From the factorial designed experiments we received a lot of data. These were used to build PLS-models for the different gas components (CO, CO2, H2, CH4, H2S) and the inorganics in the bed solids leaving the reactor (residual carbon, sulfate, sulfide). We were using the program SIMCA from Umetri for this, but other programs like Unscrambler or PLS-toolbox would have given the same results. Graphs for the different polynoms aquired from the results of the pilot plant experiments are shown in figure 2 below: % H2 for different temp,Rel ox and capacity