Computing Sciences masthead Berkeley Lab Computing Sciences Berkeley Lab logo

Exascale for Energy

Low-Swirl Combustion

Experiments and simulations working together

Researchers at Lawrence Berkeley National Laboratory (Berkeley Lab) have invented a unique, clean-burning combustion technology known as the low-swirl burner (LSB), which was honored with a 2007 R&D100 Award (Figures 11 and 12). The basic LSB principle is fundamentally different than the conventional high-swirl combustion method and defies many established notions of turbulent flame properties and burner engineering concepts. The new technology not only burns efficiently and cleanly, producing a very low level of nitrogen oxides, it is also more economical to manufacture and operate than many conventional burners. The LSB has been scaled for devices ranging in size from home furnaces to industrial boilers and power plants.

Low-swirl burner
Low-swirl burner
Figure 11. The heart of the low-swirl burner is a vane-swirler that has two flow passages. The fuel/air mixture flows through the openings of the center channel and the gaps between the surrounding swirl vanes. This design creates the low-swirl flow which supports a stable lifted or floating flame. Source: R. Cheng,  Berkeley Lab Figure 12. The low-swirl burner remains cool to the touch because the lifted flame does not heat up its body. The lifted flame originally was thought to be highly undesirable because in other burners it leads to unstable flame behaviors. But the unique divergent flow field of the LSB allows the lifted flame to self-adjust and remain robust even with a very lean fuel mixture. Source: R. Cheng,  Berkeley Lab

For much larger turbines capable of generating 250 MW of electricity, the Department of Energy is supporting research to evaluate the low-swirl combustion technology as a candidate for the hydrogen turbines in DOE Office of Fossil Energy’s FutureGen Clean Coal Project. FutureGen is a public-private partnership to design, build, and operate the world’s first zero-emissions fossil fuel power plant, using the Integrated Gasification Combined Cycle (IGCC) approach to produce hydrogen, which is separated from a concentrated CO2 stream. The CO2 is then sequestered in the earth, preventing emissions to the atmosphere that contribute to climate change.

Continued development of low-swirl burners and other advanced combustion technologies for large-scale generation of electricity depends on improving our understanding of basic flame structure, stabilization mechanisms, emissions, and response to changes in fuel. Numerical simulation has the potential to address some of these issues, but simulation of advanced burners has proven to be difficult because of the large range of spatial and temporal scales in these systems. The bulk of the analysis to date has been based on laboratory experiments.

The detailed structure of lean premixed flames becomes particularly important and difficult to simulate when the fuel is hydrogen. Burning hydrogen in a gas turbine presents significant technical and engineering challenges because of the high reactivity of hydrogen, its fast flame speed, and the propensity of the hydrogen/air mixture to auto-ignite and explode.

flame simulation
Figure 13. Simulations that ran on as many as 4000 cores of the Franklin Cray XT4 supercomputer at the National Energy Research Scientific Computing Center (NERSC) captured the detailed structure of a lean hydrogen flame on a laboratory-scale low-swirl burner. This image shows concentration of the flame radical hydroxyl and the turbulent vorticity field (gray). Source: M. Day,  Berkeley Lab

Lean hydrogen-air flames burn in cellular structures—localized regions of intense burning, separated by regions of local extinction. In this regime, the flame surface is broken into discontinuous segments. This type of structure introduces severe difficulties in applying standard turbulence/chemistry interaction models, which are based on the presence of a highly wrinkled but continuous flame surface. The cellular burning patterns also make the analysis of experimental data problematic, and can lead to significant inaccuracies or misinterpretations.

Despite these difficulties, Berkeley Lab’s Center for Computational Sciences and Engineering (CCSE), supported by two successive INCITE grants totaling 5.6 million processor hours, has succeeded in producing a series of direct numerical simulations of lean, premixed hydrogen flames on laboratory-scale low-swirl burners (Figure 13).

Using a low Mach number formulation and adaptive mesh refinement, the simulations incorporate detailed chemistry and transport without relying on explicit models for turbulence or turbulence/chemistry interaction. The simulations capture the cellular structure of hydrogen flames and provide a quantitative characterization of enhanced local burning structure.

This work lays the foundation for a program of closely collaborative investigations between computational and experimental combustion scientists. Since the simulations of a laboratory-scale burner required millions of hours on a terascale system, future simulations of power plant burners will clearly require exascale capabilities.


<< Previous page