- THE MAGAZINE
In the early 1990s, large-scale glass manufacturers began converting furnaces from air-fuel combustion to 100% oxy-fuel combustion in an effort to save money and increase efficiency. For many of the larger oxy-fuel glass furnaces, oxygen is generated onsite by either vacuum swing adsorption or cryogenic distillation. However, should this onsite method of oxygen generation fail, a backup is needed to prevent downtime and lost production.
To date, the only method used to back up the oxygen supply for these glass furnaces has been an onsite supply of liquid oxygen (LOX). Trucks haul the liquid to the site from a larger air separation facility. Large insulated tanks are required for cryogenic storage, and vaporizers, pressure control equipment and additional site preparation are necessary to provide the process with gaseous oxygen.
This approach allows the melting process to continue without interruption using the same combustion system (i.e., burners) and has traditionally been perceived as an acceptable solution. However several changes taking place in the glass industry raise new concerns about this method of backup. Today, furnaces are being built larger during scheduled rebuilds, and traditionally “large” furnace segments of the glass industry (i.e., flat glass, TV glass and some container glass) are now converting to oxy-fuel. Additionally, multi-furnace oxy-fuel conversions are occurring at single sites, and companies that were previously supplied with LOX are converting to onsite supply.
Each of these observations affects the oxy-fuel glass furnace backup combustion system. The above-mentioned “large” furnaces use much more oxygen than most other typical glass melting furnaces. One issue with providing liquid oxygen backup is the high capital and site preparation/installation cost of the storage tanks and vaporizers. Another issue is logistics—transporting the large required volumes of LOX to the site for “spot” requirements may not be possible, depending on the oxygen supply company’s availability of trucks and drivers in the region. Transporting LOX to furnaces located in remote locations is even more difficult. LOX availability on short notice from a nearby air separation facility is another potential issue. Additionally, force majeure (greater force) occurrences such as power outages can also overburden the LOX supply and distribution network in a region.
To ensure that production can be safely continued in a backup situation, a different method of backup is required.
The “Hot Hold” SolutionIn the event of oxygen supply disruption, furnaces can be maintained on “hot hold” with the use of furnace heat-up combustion equipment. Hot hold is a condition in which production is stopped, yet the furnace is kept hot enough to prevent glass solidification—an event that can cause severe damage to the furnace. In this procedure, no special temperature profile for production is attempted, and maximum furnace temperatures achieved are limited to 2200°F due to burner design. However, allowing refractory temperatures to dip to 2200°F raises the issue of contraction/expansion control. Furnace binding steel needs to be adjusted during the “cool down” and “reheat” periods (at a time when personnel in the furnace area are needed for other tasks), and a temperature of 2200°F is not sufficient for producing glass. For these reasons, the hot hold backup method is the least preferred option by glass manufacturers.
The cost of halted glass production is very high in terms of lost product sales, as well as disruption of the downstream glass forming, annealing and handling processes. Table 1 summarizes the economic losses due to oxygen supply disruption for a 600 ton-per-day float glass furnace with product valued at $250 per ton. The table provides results for hot hold conditions, 20% of full production, and 50% of full production for one day.
Following hot hold conditions, the ribbon in the tin bath must be reestablished, resulting in an additional loss of $150,000, assuming that restoring the ribbon takes one day. For pull rates greater than about 20% of normal production, it is believed possible to sustain the ribbon. Therefore, in this example, one day without oxygen results in a $300,000 loss if hot hold is used, a $120,000 loss in the case of 20% production, and a $75,000 loss in the case of 50% production.
This is a “best case” scenario for the hot hold condition, as costs can also be incurred to contract the air-fuel combustion services and hot repair activity to provide openings for the air-fuel burners. The opportunity for extremely large additional expense also exists through additional lost production due to possible stone contamination and operational limitations induced by refractory movement caused by furnace contraction/expansion during the large temperature swings.
Similar economic scenarios can be demonstrated for the other segments of the glass industry.
Air-Fuel BackupWhile an air-fuel backup method would be ideal, air cannot simply be substituted for oxygen in most conventional oxy-fuel burner systems. When a glass furnace is converted from air-fuel to oxy-fuel, heat recovery devices such as regenerators and air supply systems are removed. The pressure requirement to provide an equivalent amount of contained oxygen using air in an oxy-fuel burner would be very high, requiring an expensive air supply system. Safety considerations also preclude use of the oxygen distribution piping for conveying compressed air. A higher firing rate is required for air-fuel operation as compared to oxy-fuel operation because of the additional energy losses caused by heating and expelling nitrogen in the air-fuel case. Further, the air used for combustion will typically not be preheated in a backup scenario, which results in decreased furnace efficiency relative to a typical air-fuel furnace.
A simple thermodynamic calculation illustrates the need to increase the fuel flow rate when non-preheated air is used for combustion. To maintain the same “available energy” (the energy required for heating the charge and for wall losses, excluding flue losses), about 2.65 times the oxy-fuel firing rate is required when firing with air. This calculation assumes that a stoichiometric quantity of oxidant and all gases (methane, air, oxygen) enter the furnace at 25°C (77°F) and exhaust the furnace at 1538°C (2800°F) after complete combustion. According to this simple calculation, the combustion system must be capable of firing 2.65 times the oxy-fuel firing rate. For the air-fuel backup system to work effectively at full production.
Accordingly, the total oxidant volumetric flow rate will increase dramatically when air replaces oxygen. When air is completely substituted for oxygen, the volume of the oxidant stream is increased by a factor of 4.76 because of the addition of nitrogen. The flow rate of the oxidant stream is increased by about 12.6 times when air is completely substituted for oxygen and 100% glass production is maintained.
The air supply pressure required to accommodate the higher gas volumes needed usually prevents air-fuel firing with typical oxy-fuel burners. Air supply pressures in excess of 5 psig will require costly constant displacement blowers or air compressors rather than relatively inexpensive, single-stage centrifugal blowers. Moreover, the relatively large flow rate of air through the burner needed to replace pure oxygen can easily produce a “choked” flow condition. Choked flow would occur if the air velocity reaches the sonic limit at some cross-section within the flow passage. This would then place a maximum limit on the deliverable air flow rate.
A New Air-Oxy-Fuel BurnerA new burner* has been developed to eliminate the problems associated with using air-fuel as a backup to oxy-fuel firing. The air-oxy-fuel (AOF) burner uses the existing burner block (precombustor) of an inherently low momentum high radiation burner,** which establishes a stable air-fuel flame. To avoid possible impingement on the opposing furnace wall (due to the increased air pressure), the flame length can be controlled even at high firing rates by adjusting the gas and oxidant velocities and oxygen concentration.
Soot generation can be varied by adjusting the level of oxygen enrichment. This technique can also be used to adjust flame length and/or to provide better radiation heat transfer to the charge.
Using this method, it is possible to maintain the minimum temperature and proper temperature distribution needed for glass production. Oxygen enrichment should be preferentially used on burners with the highest firing rates near the hot spot (a spot required in glass making furnaces to establish the convection cells in the glass melt needed to maintain optimal glass quality). Using oxygen enrichment will reduce the flow rate of air needed for these burners and reduce the pressure drop. Oxygen enrichment also increases the peak flame temperature and thereby increases heat transfer in the hot spot.
Computational Fluid Dynamics ResultsComputational fluid dynamics (CFD) modeling has been used to evaluate the air-fuel backup method using the new AOF burner. Modeling was used to confirm suitable flame interaction and temperature profile during the backup mode. A combustion space model with separate boundary conditions for the glass and batch was calculated. These modeling capabilities are described by Hoke.1
Velocity vectors in the burner plane for 100% oxy-fuel and full production are shown in Figure 1. This represents the base case operation. The furnace uses 12 high radiation burners in a mostly staggered burner arrangement. Burners 1R and 1L are opposing, as are burners 6R and 6L. Two exhaust ports are located over the batch region. Velocities in the flame region penetrate well into the furnace and are less than 15 m/s.
Figure 2 shows the velocity vectors in the burner plane for the air-fuel backup mode at a firing rate that provides approximately 22% production. Velocity vectors are much higher for this case, approaching 50 m/s. Some flame interaction occurs for the opposing burners 6R and 6L, but in general, the flow patterns in the furnace are suitable for a temporary backup condition.
Predicted hot face temperatures along the centerline of the crown are plotted in Figure 3 (p. 76). Results for oxy-fuel at full production and two cases for reduced production in the air-fuel backup mode are shown. A slightly reduced temperature is predicted for the air-fuel backup cases. In practice, the temperature level could be adjusted by adding some oxygen or possibly by preheating the combustion air. In contrast to the temperature profile achieved by the oxy-fuel and air-fuel backup, a constant temperature of 1200°C is shown to represent the hot hold condition.
Field Test ResultsA field test of the AOF burner was performed to demonstrate the concept. Since it is expected that there is a fuel penalty and that a loss of production will result when going to a full air-fuel backup mode, only a single burner was substituted in this first trial.
The one-day trial was performed at Techneglas’s Columbus, Ohio, facility, where TV funnel glass is melted. The air-fuel burner was fired between 2 and 10 million BTUs per hour (MMBtu/h) with oxygen concentration in the oxidizer stream varying between 21 (air-fuel) and 60%.
Figure 4 shows the connection of the air supply to the AOF burner. The installation was relatively simple, since the air-fuel burner connects to the same mounting flange as the oxy-fuel burner.
Suitable flame shape is a concern for traditional oxy-fuel burners operating at 2.65 times their rated firing rate, especially with 12.6 times the volumetric flow rate through the oxidant passage(s). The flame resulting from the air-fuel combustion using the AOF burner is shown in Figure 5. The flame was very stable and was longer and fuller than the oxy-fuel flame. Soot particles were clearly visible, and the flame was more luminous than the oxy-fuel flame. The visible flame length varied from about 18 feet at 4 MMBtu/h to about 23 feet at 10 MMBtu/h. The flame length was shortened significantly when oxygen enrichment was added.
As expected, NOx levels increased when the air-fuel burner was added. Approximately two times the original oxy-fuel NOx level [lb. NO2/MM Btu] was measured during the operation of the air-fuel burner. The increased NOx is presumably caused by the additional nitrogen in the high temperature oxy-fuel flames.
When the air-fuel burner was fired at the same firing rate as the previous oxy-fuel burner, the crown temperature decreased about 30°C. As the firing rate was increased and some oxygen enrichment added, the temperature difference was diminished.
A Promising TechnologyThe AOF burner was developed to provide glass manufacturers increased operational flexibility. The burner can be operated with air or oxygen-enriched air as the oxidant for combustion.
Compared to the option of using air-fuel heat-up burners to maintain hot hold conditions, the AOF burner technology shows the promise of suitable furnace temperature profile and continued production. Production at a minimum is beneficial to many glass producing operations. Container, tableware and TV-glass furnaces can reduce machine speeds or possibly keep fewer lines of a multiple line furnace at full production; fiberglass operations can keep many bushing positions operational; and float glass producers can maintain production at a level to sustain a ribbon on the tin bath. Reheating forming machine molds, and reheating and stabilizing annealing lehrs is time consuming, and thus represents high “lost opportunity” cost. In particular, reestablishing the ribbon in float glass operations can delay production by one or more days, resulting in a significant loss of revenue.
A thermodynamic analysis shows that the equivalent firing rate for the air-fuel backup mode is greater than 2.5 times the oxy-fuel firing rate. Further, the oxidant flow rate when using air instead of oxygen is greater than 12 times the oxygen flow rate. Therefore a special burner technology, such as the AOF burner, must be used to accommodate this full range of operation.
CFD modeling shows that suitable temperature profiles and gas flow patterns in the furnace result for the air-fuel backup mode of operation. Predicted temperatures are slightly lower for the air-fuel case compared to the oxy-fuel case. In practice, oxygen enrichment or air preheating could be used if necessary.
Successful field testing of the AOF burner has demonstrated flame stability. This was proved up to firing rates two times higher than the oxy-fuel burner firing rate. Crown temperatures were lowered by about 30°C when the air-fuel burner was fired at the same firing rate as the oxy-fuel burner.
As more and more flat glass furnaces and multi-furnace sites are converted to oxy-fuel, the demand for liquid oxygen reserves at nearby air separation facilities, which may be needed for backup, increases substantially. At additional expense, more LOX storage tanks can be installed as insurance in case of oxygen onsite generator supply disruption. Alternatively, the new air-fuel backup technology can effectively provide the same insurance with a higher operating cost but much lower capital cost, while improving the self-sufficiency of the glass producer. It also provides additional capabilities for backing up a furnace that LOX contracts do not cover, such as houseline failures and extreme force majeure situations.