0
Select Articles

Fluid Modeling vs. Pollution PUBLIC ACCESS

CFD Analysis Provides a Deeper Understanding of a Low-emission Combustion System.

[+] Author Notes

Peter A. Liever is a senior engineer

Clifford E. Smith is vice president of engineering at CFD Research Corp. in Huntsville, Ala.

Geoffrey D. Meyers is principal engineer in the Combustion and Emissions Group of AlliedSignal Engines in Phoenix.

Mechanical Engineering 121(01), 64-66 (Jan 01, 1999) (3 pages) doi:10.1115/1.1999-JAN-5

This article reviews how computational fluid dynamics (CFD) analysis provides an enhanced understanding of a low-emission combustion system. When AlliedSignal Engines in Phoenix wanted its ASE40 industrial gas turbine to meet tough new standards for nitrogen oxide emissions, the company decided to try a design that injected water into the combustion zone so the system would burn cooler. AlliedSignal combined full-scale engine tests and computer models to study the effect of water injection on the ASE40. CFD provided detailed flow field information not available from engine tests. This information allowed engineers to verify the effectiveness of the numerous design changes made in axial air swirlers, mixing jets, and cooling flows. Work is also in progress on a dual-fuel system with water injection, using the same gas/water manifold and combustor. Oil fuel will be introduced through the original water circuit, with water being introduced into the gas side. This system will be distributed for the European market by AlliedSignal’s partner, Motoren-und Turbinen-Union (MTU) of Friedrichshafen, Germany.

When allieds ignal engines in Phoenix wanted its ASE40 industrial gas turbine to meet tough new standards for nitrogen oxide emissions, the company decided to try a design that injected water into the combustion zone so the system would burn cooler.

The output of NOx rises with flame temperature, so as the injection of water lowers the heat of the combustion zone, it also reduces NOx emissions. At the same time, injecting water can have some undesirable side effects, such as a sharp increase in carbon monoxide emissions and in the potential for carbon deposits. If water reaches the walls, thermal stress can shorten the liner's useful life.

AlliedSignal combined full-scale engine tests and computer models to study the effect of water injection on the ASE40. Computational fluid dynamics provided detailed flow field information not available from engine tests. This information allowed engineers to verify the effectiveness of the numerous design changes made in axial air swirlers, mixing jets, and cooling flows. Visualizing fuel and water injection patterns and identifying flow field zones critical to pollutant formation proved invaluable in designing the combustor for the best balance of various emissions.

The ASE40 gas turbine generates 3.25 megawatts of net output at standard sea-level conditions and can run on either natural gas or DF-2 diesel fuel. The gas generator has a two-stage axial turbine driving seven axial turbine stages and a single-stage, centrifugal high-pressure compressor. The ASE40 combustion system is a reverse- flow annular design with 28 individual fuel injectors. Each fuel injector tip is inserted in the combustor dome through an axial air swirler.

In addition to shaft power, the turbine generates an exhaust flow of 28.3 Ibs. per second at 1,100°F. The simple-cycle thermal efficiency, without installation losses, is 28.4 percent, yielding a heat rate of 9,600 Btu per horsepower-hour. The ASE40 and the slightly larger ASE50 are used in a variety of standby power, cogeneration, and pumping applications.

The emission goals were 25 parts per million NOx (corrected to 15 percent oxygen) for natural gas and 42 parts per million for liquid fuel. The emission goals required changes in the combustion system. The modifications included switching from a film-cooled liner system to effusion cooling and using two new fuel injectors, which were designed by Delavan Gas Turbine Products of West Des Moines, Iowa. One of the new injectors is for natural gas and water, while the other is for diesel fuel and water.

Engineers installed a prototype of the ASE40 with the water-injection combustion system in a test cell at the Allied Signal Engines laboratory in Phoenix. The test engine included instrumentation for measuring inlet conditions, fuel and water flow, combustor inlet temperature and pressure, liner skin temperatures, rotor speed and dynamics parameters, and exhaust gas temperature.

The exhaust flow was sampled thro ugh a fixed, multi point, isokinetic sampling rake in the exhaust duct immediately downstream of the gas turbine through a heated sampling hose. Industry-standard, nondispersive infrared analyzers kept track of the exhaust concentrations of CO, CO2 and NOx, . Tests measured unburned hydrocarbons with a flame ionization detector.

The ASE40 was equipped with a digital control system, which allowed the fuel and water schedules to be varied over a wide range. Power generated by the engine was absorbed by a calibrated water brake. The engine was tested over a load range from idle to 4,400 hp. Fuel mixtures ranged from dry—that is, with no water at all—to a water/fuel mass ratio of 1.3: 1.

NOx levels were 78 parts per million with dry combustion and declined to 33 ppm with a water-to-natural gas ratio of 0.5. NOx fell to 17 ppm at a ratio of equal parts water and natural gas. At the same time, CO levels rose steadily as water was added to the fuel. CO emissions, which were 22 ppm in dry combustion, rose to 68 ppm at the 0. 5 ratio and to 200 ppm at the ratio of 1.0.

Tests showed the sharpest reductions in NOx levels at water-to-fuel ratios ranging between 0.0 and 0.6. CO levels increased progressively as water ratio increased. Water-to-fuel ratios higher than 0.6 increased the CO output without appreciable gain in NOx reductions. Similar trends were seen for water injection with DF-2 diesel fuel, although at slightly higher emission levels.

CFD analyses of turbine combustion: The pattern of natural gas alone (left) is compared with that of natural gas with water injection.

Grahic Jump LocationCFD analyses of turbine combustion: The pattern of natural gas alone (left) is compared with that of natural gas with water injection.

A cutaway view of AlliedSignal's ASE40, a 3.25-megawatt turbine that can run on natural gas or DF-2 diesel fuel. The ASE40 combustion system is a reverse-flow annular design.

Grahic Jump LocationA cutaway view of AlliedSignal's ASE40, a 3.25-megawatt turbine that can run on natural gas or DF-2 diesel fuel. The ASE40 combustion system is a reverse-flow annular design.

CFD calculations were performed by the Combustion Engineering Group at CFD Research Corp. in Huntsville, Ala., with commercial CFD-ACE+ flow analysis software. The complete software package consists of the interactive grid generation code CFD-GEOM, the CFD-ACE multipurpose flow solver for structured and unstructured grids, CFD-POST for emissions and performance data postprocessing, and CFD-VIEW for interactive graphics visualization and animation.

The software has been validated in day-to-day operations analyzing complex fuel injection and combustion flows for many different gas turbine engines. Modeling features of particular benefit for combusti on system analysis are the multiblock grid capability with many-to-one grid cell interfaces, advanced spray and combustion models, and probability density function (PDF) turbulence-chemistry interaction models.

Liquid spray is calculated with a Lagrangian spray trajectory tracking, dispersion, and evaporation model. Liquid fuel and water are introduced into the flow field as droplets with a prescribed spread of sizes.

Combustion was modeled with a single-step finiterate reacti on between fuel and oxygen. Natural gas was modeled as methane (CH4), and DF-2 was represented by C 12H26.

The model assumes that the fuel consumption reaction is the rate-limiting step and that reactions of other species (CO, OH, H, O, etc.) are fast enough to be assumed at equilibrium. The products of the reaction are then determined locally, based on the local equivalence ratio.

This partial equilibrium approach is computationally efficient and accurately predicts flame temperatures, particularly in fuel-rich zones, but it typically underpredicts CO levels. In order to compensate for the understatement of CO levels, the compositions and temperature calculated from the CFD kinetics model were postprocessed to determine the nonequilibrium CO. A finite-rate oxidation of the CO produced by the combustion in fuel-rich zones was assumed

NOx levels were similarly postprocessed by considering thermal, prompt, and nitrous formation mechanisms. The postprocessing approach provides good results under the premise that the predicted heat release is not appreciably affected by small changes in CO and NOx.

A prescribed probability density function model was applied to account for the effects of turbulence on combustion. Prescribed PDF models generally consider the mixture of the fuel and oxidizer. For this study, a new prescribed PDF model had to accommodate the addition of water as a third part of the mixture.

The 3-D grid for the CFD model represented a one-nozzle combustor sector with cyclic boundaries. AlliedSignal Engines supplied the combustor geometry in the form of an IGES format CAD file that was directly imported into the CFD-GEOM software, which generated the grid. Models of the fuel nozzles were composed from drawings provided by the manufacturer.

Structured grids were used exclusively for this analysis. Many-to-one grid cell connectivity allowed a reduction or increase in grid cell density between neighboring grid blocks. This approach made it possible to generate a full 3-D model with reasonable grid cell resolution for both the small fuel nozzle passages and the main combustor.

The computational models consisted of an average of 300,000 computational cells in as many as 19 blocks. Without the many-to-one feature, an equivalent resolution grid would have contained more than two million grid points.

Mass airflow boundary conditions for the combustor came from a one-dimensional external flow code used by AlliedSignal to calculate mass airflow splits through the combustor. To accurately model the mixing jets formed by flow through the dilution holes in the combustor liner, the computational model included the dilution orifices and the liner external flow. The effusion cooling and cooling slot flows were modeled as prescribed mass flow boundaries. Customized swirler flow boundary conditions available in CFD-ACE+ were used for gaseous fuel and air swirler in-flow boundaries.

The natural gas/water injection nozzle introduces fuel through an annular helical vane swirler co-rotating with the combustor air swirler. Water is introduced at the injector center through a pressure atomizer.

The CFD analysis revealed that the gas remains highly concentrated in the combustor primary zone center without significant mixing or burning. Slow mixing and burning of the fuel causes the high-temperature region to be located as far downstream as the dilution zone.

The DF-2/water fuel nozzle injects the liquid DF-2 fuel through a central primary pressure atomizer and an annular secondary airblast atomizer. Outboard of the secondary fuel injector are an annular water airblast injector and a helical air swirler co-rotating with the main air swirl er surrounding the fuel nozzle.

The liquid fuel spray distribution results in rapid mixing and a much more even fuel-air mixture distribution across the combustor primary zone than in the case of natural gas.

The overall flow pattern remained essentially unchanged for both the natural gas and DF-2 fuel when water was injected. All water evaporated before reaching the dilution zone and the combustion liner. Engine tests verified that little, if any, water reached the liner.

Water injection lowered the overall temperature in the combustor, except in the primary recirculation zone and in small regions close to the inner and outer liner walls, where the temperature was actually slightly higher when water was injected. This pattern developed because the water delayed the evaporation of the liquid fuel droplets.

The lowering of peak flame temperature in the combustor directly reduced NOx formation, as expected. CO is generated in fuel-rich zones and is then consumed as air is introduced along the combustor. The CFD analysis revealed that the levels of CO at the entrance to the turnaround duct are comparable for both the wet and dry fuel mixtures. A key difference is that, with dry fuel, a substantial amount of CO oxidation occurs during the travel through the turnaround duct, but water injection reduces the average temperature along the turnaround duct by as much as 200°F, to a temperature level where CO oxidation is substantially reduced.

CFD emission results agreed reasonably well with the experimental trends. CFD predictions of NOx gave the correct trend with water injection level and fuel type, but were lower than measurements . CO prediction levels agreed well, with some scattering at the higher water injection levels.

A water-injected, gas-fired ASE40 was run recently at the Mitsubishi Heavy Industries test facility in Nagoya, Japan. Using the water injection system described here, it met the goal of 23 parts per million NOx (corrected to 15 percent O2), That machine is being installed at a facility in Taiko, Japan, for industrial cogeneration. Other ASE40s packaged by Mitsubishi Heavy Industries are running at the Sagawa Printing Co. plant in Hino, Japan.

Upcoming applications include power for a manufacturing facility in India and an Illinois steel mill, and rural power generation from low-Btu gases recovered from depleted oil fields in Colorado.

Work is also in progress on a dual-fuel system with water injection, using the same gas/water manifold and combustor. Oil fuel will be introduced through the original water circuit, with water being introduced into the gas side. This system will be distributed for the European market by AlliedSignal's partner, MTU (Motoren- und Turbinen-Union) of Friedrichshafen, Germany.

NOx emissions fall off at a diminishing rate as more water enters the combustion zone. At the same time, CO output rises on a sharpening curve.

Grahic Jump LocationNOx emissions fall off at a diminishing rate as more water enters the combustion zone. At the same time, CO output rises on a sharpening curve.

Copyright © 1999 by ASME
View article in PDF format.

References

Figures

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In