Mass spectrometry in high performance coal pyrolyzers23 April 1998
Foster Wheeler is involved in the experimental testing of a pilot scale pyrolyzer to partially gasify pulverized coal, the results of which will be used in the development of a commercial pyrolyzer for an advanced electric power generation system. The pyrolyzer reactor is operated as a jetting type of bubbling fluidized bed to generate both a low-Btu fuel gas and solid char. An on-line mass spectrometer has been used to monitor the behaviour of the fuel gas, and initial results show its composition to be stable and suitable for gas turbine performance.
A High Performance Power System (HIPPS) is being developed by a team of companies under the sponsorship of the US Department of Energy Federal Energy Technology Center (FETC). It is a coal fired combined cycle system using a High Temperature Advanced Furnace (HITAF) that transfers heat to both air and steam working fluids. The design uses a pyrolyzation process to convert a pulverized coal feedstock into two components, a low-Btu fuel gas and solid char.
The fuel gas exhaust from the reactor is directly connected to a gas turbine and is burned with air from the compressor. The char provides the feedstock for the HITAF. Systems of this type are capable of over 47 per cent efficiency (HHV).
Foster Wheeler Development Corporation (FWDC) is leading the consortium developing this power generation system. Other members are Bechtel Corporation, Foster Wheeler Energy Corporation, University of Tennessee Space Institute, and Westinghouse Electric Corporation.
In Phase 1 of the HIPPS project, a conceptual design for a 300 MW commercial plant was developed. Economic analysis of this plant relative to a pulverized coal boiler of equivalent size indicated that the HIPPS plant will have a 15 per cent lower cost of electricity.
Phase 2 is now in progress and includes experimental pilot scale testing of both a coal fired pyrolyzer and a char combustor. Several test runs have been completed for the pyrolyzer subsystem. Experimental testing of the char combustor will commence early next year. Phase 2 work will support the design of a prototype plant in Phase 3.
The two main aspects of the pyrolyzer development are the suitability of the char and fuel gas as fuels for the HITAF and gas turbine, respectively.
Fuel gas behaviour
The experimental testing of the pyrolyzer subsystem is being performed at FWDC's R&D facility in Livingston, NJ. The pyrolyzer is operated as a bubbling fluidized bed to generate a low-Btu fuel gas and a solid char. In a commercial demonstration plant the fuel gas will be burned in a gas turbine, and the char will be fired with pulverized fuel burners in the HITAF. Currently, however, the fuel gas is flared and the char is collected and stored for future tests at a separate burner test facility.
The reactor has a stepped inner diameter starting from 25.4 cm in the bottom section, stepping to 30.5 cm approximately 1.2 m above the injection nozzle, and finally expanding to 35.6 cm in the freeboard. The total height of the refractory lined reactor is 10.7 m. Coal and limestone are pneumatically conveyed through a central feed pipe into the bottom of a jetting type bubbling bed reactor. The pulverized coal is partially gasified and attrited to meet the requirements of the char combustor.
Bubbling bed operation offers a simple 'once-through' processing of the coal feedstock. The char generated is elutriated out of the reactor with the fuel gas and carried over to a high temperature ceramic filter to remove the particulate from the gas stream. The fuel gas is depressurized through an orifice plate to simulate gas turbine operation. The collected char is depressurized through a lock hopper system and pneumatically conveyed with nitrogen to a baghouse and stored for future char combustion tests.
A major concern with bubbling bed operation is the transient behaviour of the fuel gas. The bed hydrodynamic performance affects the steadiness of the gas stream composition. If the injected fuel and air break through to the surface of the bubbling bed without reacting sufficiently, overall carbon conversion can be significantly reduced. As a result, the gas stream and solid stream compositions may be unsteady. This is of particular concern for the gas turbine since it is directly connected to the fuel gas outlet of the pyrolyzer in the commercial configuration.
An on-line mass spectrometer is therefore used to analyze the fuel gas generated in the pilot plant. The sampling connection is located downstream of the high temperature ceramic filters to prevent solids from plugging the 6.4 mm diameter 310 stainless steel sampling probe. The sample probe penetrates a 45.7 cm diameter pipe and extends through the castable refractory lining into the centerline of the flow stream. The temperature and pressure of the fuel gas at this location is approximately 704°C (1300°F) and 1.03 MPa (150 psia), respectively.
Outside the pressure vessel the sample line is maintained at a temperature of 117°C by a combination of electrical heat tracing and a counterflow steam jacket. The maximum allowable sample gas temperature in the mass spectrometer is 117°C. As the fuel gas enters the sample conditioner cabinet it passes through a 2 µ sintered metal filter to remove solids still present after the ceramic filters, and to regulate the pressure to 276 kPa.
The sample conditioner also operates as a convection oven to maintain the sample gas temperature at 117°C by electrically heating a nitrogen purge flow to the cabinet. Most of the sample gas, at 100 l/min, leaves the sample conditioning cabinet through the primary speed loop. This serves to flush out the sample lines and maintain a fresh sample of fuel gas for analysis.
The remaining portion of the flow exiting the sample conditioner enters into the heated valve box attached to the mass spectrometer. Each of the two valves within this enclosure are outfitted with 16 ports used to connect the sample line and all required calibration gases with the instrument. A secondary speed loop is employed to clear out the internal tubing through the valve assembly. The combination of the primary and secondary speed loops for the HIPPS pilot plant establishes a clearing delay of approximately 5 s. Depending upon the speed loop flows and the proximity of the instrument relative to the sample probe location, the clearing delays can be reduced for different applications.
Analogue signals (4-20 ma), representing the scaled concentration of the fuel gas components, are output from the mass spectrometer and input to the pilot plant's distributed control system (DCS). These signals are being incorporated into the control strategy for the reactor.
Experimental data gathered from previous partial coal gasification tests established an initial estimate of the chemical composition of the fuel gas. This initial composition was used as the starting point for developing an analytical method to solve for the on-line gas composition.
The mass spectrometer operates by first ionizing all of the individual components of the inlet sample gas. The ionization process causes the inlet gas components to become positively charged, and in most instances, to fragment into different molecules or isotopes. Therefore, each of the components of the inlet sample gas ionizes differently to produce a unique spectrum of positively charged molecules, with a variety of mass to charge ratios (m/z).
These spectral characteristics are dependent on the ionization method of the mass spectrometer. The Quadropole mass spectrometer utilized in our test facility employs an electron ionization technique to generate ions. Selecting the appropriate m/z ratio for component analysis requires an understanding of how the co-existent gas species interfere with one another through fragmentation.
The base peak, or highest intensity m/z ratio for each component is represented by a numerical value of 100, and all relative peaks are referenced from this. A single m/z ratio is assigned to represent each component of the gas stream. Typically this value is chosen to be the most intense peak of a molecule's ionized spectra, with the least amount of interference from other components. All ion interference peaks are evaluated within the instrument software, and subtracted from the total signal to isolate the single component defined for a given m/z ratio.
The accuracy and repeatability of the instrument is specified in terms of a relative standard deviation (RSD). An RSD value of 0.3 per cent is defined for any measured component with a one per cent molar concentration, monitored at its base peak, and with no interferences from other components. Therefore, as the total fuel gas composition changes, the accuracy of each of the monitored components will vary. Assuming a Gaussian distribution for instrument performance, two RSDs from the mean establish a 95 per cent confidence limit for each measurement.
A series of pilot plant test runs have been performed. The first nine setpoints combined a coarse limestone (3.2 cm x 0) with a pulverized coal (70 per cent through 200 mesh) as the reactor feedstocks. Under these conditions, the fluidized bed is comprised mostly of calcined, partially sulfided limestone. The pulverized coal feedstock is partially gasified and attrited as it passes through the limestone bed material. The subsequent five setpoints utilized a coarse sand to form the bubbling fluidized bed. In this configuration both pulverized limestone and coal were used as the reactor feedstocks.
Five major components of the fuel gas (CO, CO2, N2, H2, H2O) are plotted (the arrows on the plots indicate the ordinate scale to be referenced for the associated gas component). This trend represents the steady state performance (setpoint 8) of the reactor with a coarse limestone bubbling bed. The peak in nitrogen concentration every 15 minutes is due to the pulsing of the ceramic candle filters.
When the pressure drop across the candle filter exceeds its allowable setting, high pressure nitrogen is instantly admitted to pulse clean the elements. As a result of this periodic cleaning, the fuel gas is momentarily diluted with nitrogen. Between pulses the baseline pilot plant readings for nitrogen, carbon monoxide, carbon dioxide, water, and hydrogen are 62, 11, 11, 5, and 4.8 per cent respectively. During pulse cleaning the nitrogen concentration increases by one per cent, while carbon monoxide, carbon dioxide, water, and hydrogen decrease by 0.5, 0.5, 0.2, and 0.2 per cent respectively. The decrease in the hydrogen and carbon monoxide content represents a drop in heating value of around 74.5 kJ/Nm3.
This variation of the heating value is insignificant and should not effect overall gas turbine performance. It is important to note that although the heating value is decreased as a result of the pulse cleaning process, the total mass flow through the plant is increased. These two factors tend to offset one another to maintain a consistent gas turbine power output. Of particular importance is the steadiness and uniformity of the fuel gas composition between candle filter pulses. These intermediate periods define the gas quality characteristics of the pyrolyzer reactor.
The consistency of the reactor fuel gas is a major concern. The slugging behaviour of the pilot plant reactor, as defined by its large length to diameter ratio, inhibits gas/solid mixing. These experimental tests thus define a worst case scenario for fuel gas performance, since any commercial reactor would be designed to provide for more uniform fluidization.
Similar data for setpoint 11, where the candle filter is pulse cleaning every 12 minutes, also showed the fuel gas components to be steady and uniform.
Gas turbine control
In a typical gas turbine application, the downstream gas components are monitored to provide feedback for gas turbine combustion control. However, by monitoring the upstream gas composition, combustion control can be further improved. Westinghouse has developed a special gas turbine combustor for systems like HIPPS. It is called a multi-annular swirl burner (MASB) combustor and it is comprised of three separate reaction zones. The first zone operates under substoichiometric conditions to minimize NOx production, the second zone provides for complete oxidation, while the third cooling zone controls the turbine inlet temperature.
Apart from the overall heating value and flow rate of the fuel gas, the most important gas components from a combustion performance standpoint are carbon monoxide, ammonia, and hydrogen. Variations in the inlet carbon monoxide levels can be anticipated with the on-line mass spectrometer, allowing time to alter the air split between the primary and secondary zones, to minimize carbon monoxide outlet emission. Monitoring of the inlet ammonia concentration serves as a useful diagnostic tool and can be used to control NOx performance.
As inlet ammonia concentrations increase, the primary zone stoichiometry can be further reduced to retard NO2 generation. Although the air can be proportioned between the separate zones as an on-line control strategy, the stability of the flame can impose limits on the degree of variation. Since the hydrogen content of the inlet fuel gas strongly influences flame stability, its concentration can also be monitored as part of the feed forward loop for gas turbine control.
Under all modes of operation, the generated fuel gas from the reactor has been shown to be steady and uniform, both in terms of mass flow and chemical composition.
The quality of the fuel gas for a commercial plant is of prime importance because the reactor outlet is to be directly connected with a gas turbine. For the pilot plant tests, the generated fuel gas was analyzed using an on-line mass spectrometer. Although the pilot plant fuel gas was simply flared in a thermal oxidizer, future testing under the HIPPS project will include integrated gas turbine testing. Control strategies are being developed to utilize both the upstream and downstream gas compositions to improve gas turbine performance.