Fig 1 Operating at steady stoichiometry produces high NOx levels. Oscillating between fuel-rich and fuel-lean conditions substantially reduces NOx levels while maintaining net stoichiometry.

Combustion under conditions of an excess air level near stoichiometric concentrations of fuel and oxygen results in maximum flame temperature and high levels of NOx emissions. However, NOx formation is sensitive to temperature and oxygen concentration1-2, so either a fuel-rich or fuel-lean flame generates less NOx than a stoichiometric flame (figure 1). Stoichiometric combustion can be avoided by using oscillating combustion, which creates alternating NOx-retarding fuel-rich and fuel-lean zones within a flame. The patented technique involves forced oscillation of the fuel flow rate to the burner. Heat is removed from the zones before they mix, reducing overall peak flame temperature, and, thus, reducing NOx formation. Heat transfer from the flame to the load increases due to the more luminous fuel-rich zones and the breakup of the thermal boundary layer. The technology3-6 can be applied with ambient air, preheated air, enriched air, and oxygen based combustion.

Fig 2 Solid-state proportioning (SSP) valve produces required flow oscillations by restricting flow passage corresponding to applied actuator frequency

Oscillating combustion is accomplished using a solid-state proportioning (SSP) valve (CeramPhysics Inc., Westerville, OH), which has an elastomer disk sandwiched between fixed and movable pistons (figure 2). As an actuator is energized and oscillated at the required oscillating frequency, the force exerted by the moving piston on the elastomer disk causes it to bulge and restrict the flow passage, which provides the required flow oscillations.

Fig 3 Extremely high open-close cycling of oscillating combustion valves produces near square-wave oscillations

Fig 4 Thermally insulated oscillating combustion valves with bypass piping manifold

Glass-Melter Conversion

Air Liquide conducted a field demonstration of the benefits of oscillating combustion technology on an oxyfired borosilicate glass-melting furnace (Johns Manville Cleburne plant) used to make glass marbles, which are converted into glass wool fiber insulation. The furnace is equipped with ten Air Liquide ALGLASST burners in a staggered configuration on opposing sides having a total firing rate capacity of 18.7 MMBtu/hr (MM = million million).

Retrofitting the existing combustion system required the installation of an oscillating valve and the associated electronic valve controller on the fuel line to each burner. The existing burner fuel piping was modified to include a bypass manifold (figure 4) with necessary isolation valves. The bypass piping was then retrofitted with one SSP valve upstream of each burner. Each valve was thermally insulated and pairs of valves on opposite sides of the furnace were connected to a valve controller.

The control system for the oscillating valves includes a PC with custom human/machine interface (HMI) and a power pulse generator to deliver the required power to the valves. Operating parameters, such as frequency, amplitude (voltage), duty cycle (on/off time) and phase difference between various burners is entered using the HMI screens.

Furnace Operation Using OCT

Baseline furnace operating parameters without oscillating combustion included a stoichiometric (oxygen/fuel) ratio, SR, of 2.20 and a 0.03 in. water column furnace pressure, resulting in a furnace flue gas containing approximately 8 to 9% excess oxygen and less than 5 ppm (parts per million) carbon monoxide (CO). An optimized SR of 2.15 to control NOx emissions produced a flue gas containing an about 5 to 6% excess oxygen and less than 100 ppm CO. Lowering the oxygen/fuel ratio further decreased NOx emissions, but produced an unacceptably high concentration (3000 ppm) of CO in the flue gas. By comparison, it was possible using oscillating combustion to lower the SR to between 2.0 and 2.05 with a flue gas containing 2-3% oxygen, and still maintain low CO levels. This is due to the more efficient use of oxygen, which is one of the important benefits of oscillating combustion technology.

Fig 5 Effect of frequency (number of oscillating cycles per unit time) on NOx emissions; medium frequency yields optimum NOx performance with acceptable CO emissions and no CO in the stack. Operating conditions: 70% amplitude; 50% duty cycle; approx. 3% excess oxygen.

OC Parameters

Oscillating combustion performance is controlled by means of adjusting the following parameters:

• Frequency - the number of oscillation cycles per unit of time (Hz). An infinite frequency is used to denote steady (nonoscillating) operation.
  
• Amplitude - the relative change (%) in gas flow rate during the oscillation cycle above or below the average flow rate. Amplitude is measured between 0% (no oscillation) and 100% (oscillating between zero flow and two times the average flow).
  
• Duty cycle - the fraction of time (%) the gas flow rate is above the average flow rate during each oscillating cycle. A 50% duty cycle denotes equal time above and below the average flow.
  
• Phasing between burners - relative offset (degrees) in time between the start of oscillating cycles for different burners. In-phase oscillation is denoted as 0 degrees and completely out of phase oscillation is denoted as 180 degrees. In the field test described in this article, a 180 degree out-of-phase operation was selected to avoid interaction between opposing burner flames.

Measured furnace operating parameters included fuel and oxygen use, refractory and glass melt temperatures, bridge-wall optical temperature, and furnace pressure and level. In addition, the furnace interior was continuously video recorded. Several techniques including a water-cooled sample probe, flue gas conditioning system and analytical instrumentation were used to measure NOx, CO, oxygen and carbon dioxide (CO2) contents in furnace emissions, and stack measurements were made to verify CO burnout. Effect of frequency

Figure 5 shows the influence of frequency on NOx and CO while holding the excess oxygen level constant (approx. 3%) and varying the SR from 2.05 to 1.94. Medium frequency was selected for long-term testing because it gives optimum NOx performance and acceptable CO emissions with no CO in the stack.

In laboratory testing of ALGLASS oxyburners, higher amplitude resulted in greater NOx reduction. However, the length of the flame varied between 8 and 9 ft, or 2.4 and 2.7 m (1/2 the furnace width), during steady firing at 2.3 MM Btu/hr, increasing in length by 25-30% during the fuel rich part of the cycle, and decreasing to a length of about 1 to 2 ft, or 0.3 to 0.6 m, during the fuel lean part of the oscillation cycle. The short flame potentially could cause particulate entrainment inside the burner block, especially in high-particulate furnace environments. (Amplitude should be optimized on case-by-case basis to meet emission performance and acceptable flame characteristics based on furnace dimensions.)

The test results presented in Table I show the effect of small and medium amplitude variation on excess O2 at constant SR (2.05). At the same SR, frequency of oscillation and duty cycle, higher amplitude provides better mixing of fuel and oxygen (more turbulence) as indicated by a higher excess oxygen in the flue gases. In addition, NOx reduction has a nonlinear response with respect to amplitude and is influenced by the level of excess oxygen. A maximum NOx reduction of 55% was achieved at high amplitude with very low excess oxygen.

The duty cycle is the fraction of time the gas flow rate is above the average flow rate during each oscillating cycle: a high duty cycle means that the burner operates rich for most of the cycle. Duty-cycle variation was evaluated with all burners operating on the same duty cycle, although selected burners can run on different duty cycles to optimize CO and NOx emissions.

Varying the duty cycle from low to high was performed under constant frequency, amplitude and overall SR. Field tests confirm laboratory results; that is, NOx reduction varies proportionally with increasing duty cycle, or the amount of time the flame is in fuel-rich mode of operation. Test results listed in Table II show that a high duty cycle results in 50 ppm less NOx than that achieved using a low duty cycle (885 versus 937, respectively). However, there is a practical limit to the maximum duty cycle as sufficient time must be allowed in the fuel-lean segment to achieve complete CO burnout.

Fig 6 Long-term effect of furnace pull rate on normalized specific energy (fuel plus electric boost) consumption. Operating conditions: SR (baseline) = 2.15; SR (oscillating) = 2.0.

OCT Benefits

Operating data were recorded on a daily basis including melter gas use (million cubic feet/day, or MCF/d), electrical boost (kWh), % cullet used (glass pieces added to the melt), tons melted per day (tons/d) and specific energy consumption (MMBtu natural gas/ton of glass and MMBtu electric/ton of glass). Data were provided for four months each of baseline operation and oscillating combustion operation covering a wide range of furnace pull rates and specific fuel consumption. Use of both natural gas and electric boost were evaluated to determine the true cost of melting.

A plot (using a straight-line trend analysis) of the specific energy consumption as a function of furnace pull rate for baseline and oscillating combustion operation (figure 6) shows the improvement in furnace efficiency with pull rate. The scatter of data is attributed to the variation in day-to-day furnace operating conditions, such as changes in batch composition, firing rate, furnace pressure and level, glass and refractory temperature, etc. The straight-line fit indicates energy savings of 3 to 5% (depending on pull rate capacity) using oscillating combustion.

Fuel savings are derived from:

• Improved radiation heat transfer associated with oscillating (oxy) flames
  
• A more luminous flame caused by fuel-rich part of the oscillating cycle and enhanced cracking of fuel into higher hydrocarbons and soot particles
  
• Higher average flame coverage (approx. 30%) due to oscillating flames resulting in greater heat transfer to glass surface

Fig 7 Long-term effect of furnace pull rate on normalized specific oxygen consumption. Operating conditions: SR (baseline) = 2.15; SR (oscillating) = 2.0.

The periodic pulses from oscillating combustion provide better mixing between fuel and oxygen inside the high-temperature furnace volume, which results in higher efficiency of oxygen use. The use rate is optimized by proper selection of fuel amplitude, frequency, duty cycle and phase differences between opposite burners.

Oxygen use for oscillating combustion technology (figure 6) shows a 10 to 14% lower average specific oxygen consumption over baseline operation depending on the pull rate. This is supported by an overall lower SR of 2.00 for oscillating combustion with minimal CO (< 100 ppm) compared with a SR of 2.15 for baseline operation. The oxygen saving is due to fuel reduction and improved use of oxygen due to the more efficient mixing of fuel and oxidant in the oscillating flame. Operating the furnace using steady firing and a SR of 2.00 results in CO emissions as high as 1.5%, which further supports one key benefit of oscillating combustion; that is, more efficient oxygen use. Furnace Temperatures

Important considerations with respect to furnace temperatures using oscillating combustion are the ability to maintain safe refractory temperatures and desired glass melt temperatures while maintaining glass productivity and quality.

The influence of oscillating flames on crown and bridge-wall temperatures occurred within 15 to 20 minutes of switching to oscillating mode. Data from long-term tests indicate a 30 to 50°F (17 to 30°C) reduction in crown hot-spot temperature and almost no change in glass melt bottom temperatures at the same firing rate and electric boost. The furnace pressure and glass level also remained constant during oscillating combustion firing. Productivity and Quality

The glass-melting furnace maintained peak productivity (MM Btu/ton) during the four-month test period. The furnace occasionally operated at maximum capacity of the downstream marble machines, a limiting factor for this particular oxyfurnace.

The quality requirement of the glass product for the furnace in this field test is not very rigorous because of remelting of marbles for the fiber conversion process. However, glass quality with respect to weight, color and general defects (such as stones and seeds) remained the same as that during baseline firing.

ACKNOWLEDGMENT The authors thank Jeff Smith, Donny Timms, Tom Kimberly, Everett Wollitz and Pat Amsden of Johns Manville, for their contributions to this project, and for the support of Gas Technology Institute, Gas Research Institute, Columbia Gas Distribution, Gas Technology Canada, Southern California Gas, U.S. Dept. of Energy, CeramPhysics Inc., GT Development Co., and Energy Technology Applications.

For further information: Eric Streicher is technical manager, Gas Applications, Air Liquide America Co., 5230 SE Ave., Countryside, IL 60525; tel: 708-579-7771; fax: 708- 579-7858; e-mail: eric.streicher@ airliquide.com