Low NOx Combustion Of Biomass Fuels

T. Webster, S. Drennan, Coen Company, Inc.


Biomass combustion can make use of many waste products found in the timber, manufactured wood products, and agricultural industries. It not only provides a source of essentially free fuel, but can also eliminate many of the disposal problems associated with these by-products. The environmental regulations faced by industry are one of the primary governing factors that must be addressed when designing a new system. The associated costs to comply with mandated NOx emissions limits can be significant, especially if it requires the addition of costly flue gas treatment equipment. In these cases the ability to reduce NOx emissions significantly through changes to the combustion equipment can make the difference as to whether a system can meet required NOx levels in a cost effective manner.

In order to make significant reductions in the NOx emissions from biomass combustion, without the use of flue gas treatment, a method must be found to reduce the conversion of fuel bound nitrogen (FBN) to NOx. In this paper we will deal primarily with our experiences in firing sanderdust fuels that are a by-product of the manufactured wood products industry, although the results are applicable to many other biomass fuels. Through the use of air staging, it has been possible to reduce the total NOx emissions by more than 50 percent over that of an un-staged system. New analysis tools and increasing data from the operation of these staged systems is allowing new burners to be designed with even more ambitious emission levels.

Keywords: direct combustion, emissions, NOx.


The manufacture of panel products, such as particle board, oriented strand board, or medium density fiberboard, result in large amounts of sanderdust, sawdust, and hog fuel waste. These waste products not only represent a source of free fuel to their producers, but also would require disposal in a land fill if they were not burned. Raw wood typically contains 0 to 0.25% nitrogen by weight. However, the glues used in manufacturing the panels can increase the nitrogen content of this waste to as high as 7 %, with the largest FBN being observed with plants that are using new faster drying resins. The agricultural industry also produces a significant amount of biomass waste, such as rice husks or wheat straw, and increasing air quality regulations are greatly restricting the amount of field burning allowed. Plowing this waste material back into the soil can result in increased levels of crop disease and lower crop yields. These raw biomass materials can also contain comparatively high nitrogen levels, with typical values ranging from 0.5 to 2.5 % by weight.

In order to utilize these materials as an industrial fuel one major concern is the effect that they may have on the amount of regulated emissions being generated. When these fuels are used to offset the use of natural gas, one area of major emissions increase is in the amount of NOx generated. NOx is a term that is used to refer to the sum total of nitric oxide (NO), nitrogen dioxide (NO2), and nitrous oxide (N2O) emissions. When burning any fuel the two primary methods that result in the formation of NOx are termed “Thermal NOx” and “Fuel NOx”. Thermal NOx results from the disassociation of the nitrogen molecules in the combustion air when they come into contact with the high temperature areas of the flame. Once the N2 molecule has split into two N molecules they will readily combine with the oxygen present to form NOx. Fuel NOx is formed by the nitrogen that is contained within the fuel itself, which is freed during the combustion process. As the fuel burns, this nitrogen that was chemically bonded as part of the fuel molecule is freed and can readily combine with any oxygen present. The conversion of this FBN to NOx is what results in dramatic increase in the NOx emissions of biomass fuels when compared with natural gas, which contains no FBN. Table 1 shows the typical uncontrolled NOx emission comparison of natural gas, fuel oil, and biomass of varying nitrogen levels.

Predicted Uncontrolled NOx Levels Table

Table 1, Predicted Uncontrolled NOx Levels of Various Fuels


In order to significantly reduce the amount of NOx reduction from biomass fuels, it is necessary to target the “Fuel NOx” conversion, since this results in the largest single contribution to the NOx. One proven method for reducing the amount of bound nitrogen that converts to NOx is to introduce the biomass into a primary combustion zone that is oxygen deficient, or sub-stoichiometric. In this way the nitrogen that is liberated from the fuel has little or no free oxygen to bond with and therefore mostly recombines to N2. The balance of the air required to complete the combustion is introduced further downstream in a secondary combustion zone. This guarantees complete burnout of the fuel and reduces the temperature of the combustion products.


In the early 1990’s a medium density fiberboard plant decided to replace its aging dutch oven boiler with a new Nebraska “A” type packaged boiler. The availability and need for disposal of the sanderdust produced by the plant resulted in the decision to use sanderdust as the primary fuel to fire the boiler, with natural gas as a backup. The boiler was designed to produce 45,352 kg/hr (100,000 lb/hr) of steam on gas or oil firing, but was de-rated by the site to produce 27,211 kg/hr (60,000 lb/hr) of steam on biomass fuel. The permit application to the state air quality authority resulted in a permit that would require the boiler to meet emission levels of 0.7 lb/mmbtu. The plant produced sanderdust from two different board processes, resulting in differing nitrogen levels of 0.59% and 2.35%. The burner employed was a Dual Air Zone (DAZ) gas burner with an annular scroll to introduce the sanderdust, see Figure 1. In this register the biomass fuel is introduced between two counter-rotating air streams to provide rapid turbulent mixing of fuel and combustion air. The design employed was to provide only 80% of the stoichiometric air (i.e., fuel rich) through the burner. A separate system of adjustable direction air ports in the boiler front wall was used to introduce the secondary combustion air, which amounted to 40% of the stoichiometric air. The burner was sized for a total firing rate of 22.5 MW (77 mmbtu/hr).

Upon installation the burner was tested for base line NOx emissions without the use of any air staging, putting 120% stoichiometric air through the burner only. This testing resulted in NOx numbers below what was expected, and low enough to meet permit values without staging. However, it was still desirable to test the system to determine the best NOx achievable. When the system was operated with the staging design stated above, there was a 43% drop in the total NOx for the 0.59% nitrogen fuel and a 51% drop in the total NOx for the 2.35% nitrogen fuel. In both cases the amount of nitrogen that converted to NOx was reduced by approximately 65%. The reduction of emissions far below the permitted levels has allowed the plant to increase the capacity of the boiler back to its rated 45,352 kg/hr (100,000 lb/hr) of steam, which has resulted in a increased firing rate of 37.5 MW (128 mmbtu/hr).


A large number of industrial wood processes generate wood waste of various types and sizes. If the facility processes raw logs, then they generate a large amount of bark which is also called hog fuel. Bark fuel has both good heating value and fair combustion characteristics for firing on a grate in a boiler for steam generation. Sander dust is also a waste product with excellent combustion potential as it has very fine particle sizes that lend themselves well to suspension firing.

This application involves the sander dust being suspension fired over a bark fired grate in a boiler. The burner chosen for this application was again a Dual Air Zone (DAZ) burner equipped with a scroll to inject the dilute phase conveyed sander dust, see Figure 1. The boiler’s furnace consists of a bark fired primary combustion zone in the lower furnace with two auxiliary burners firing sander dust in the upper furnace. The sander dust contains up to 2% FBN. The fuel split design specifies a maximum sander dust firing rate of 50% of the heat input to the boiler at any condition. The baseline NOx from the bark fired on the grate is 40% of the total NOx allowed to be produced with the sander dust firing. The expected NOx emissions from un-staged combustion of the sander dust at normal excess air levels is over 200% of the allowable NOx emissions for the boiler with both fuels firing. Hence, a staged combustion solution is investigated where over-fire air ports will be added above the sander dust burners and Computational Fluid Dynamics (CFD) modeling has been requested to establish the proper location and size of these ports. The objectives of the staging for the boiler are to fire the sander dust burners at sub-stoichiometric oxygen levels and inhibit the conversion of FBN to NOx.

Figure 1, DAZ Burner with Sander Dust Scroll

To simplify the combustion aspects of the CFD problem, a gaseous methane reaction mechanism is used instead of the far more complex wood-burning reaction. Particulate combustion of the sander dust is evaluated with kinetics modeling once CFD has defined the time, temperature, composition path of the particles in the boiler. The potential NOx reduction is evaluated in much the same method. In addition, only the radiative section of the furnace is included in the computational grid. The goal of the CFD study is to achieve complete combustion at the top of the furnace, with a reasonable oxygen concentration passing through the superheater.

Description of CFD Code

The CFD software used for the numerical analysis is Fluent/UNS Version 4.2. The code works by first constructing a geometrical representation of the volume of interest, called the computational domain, that is subdivided into a large quantity of control volumes, or cells. For each of these cells, the code then simultaneously solves the governing fluid dynamic equations of continuity, momentum (Navier-Stokes equations), and energy to obtain a 3-D steady-state solution. Velocity and pressure coupling are resolved via the Semi-Implicit Method for Pressure Linked Equations (SIMPLE) algorithm. Turbulence closure by the well understood k?e turbulence model is employed. The turbulent boundary layer conditions for momentum and heat transfer in the near-wall region follow the logarithmic law of the wall. The transport equation for energy is solved using a conjugate heat transfer model, with radioactive heat fluxes calculated by solving additional conservation equations. Gas phase combustion is modeled as a one step global reaction, with user-specified fuel and oxidant components reacting to form combustion products. Species transport and chemical reaction rates are governed by both mixing of the eddies and simple Arrhenius rate equations. It is important to note that the chemical reaction model used does not take molecular dissociation into account. Radiation modeling is conducted with the Discrete Transfer Radiation model.

Figure 2, Sander Dust Burner/Boiler Computational Domain

The following is a list of some of the specific models used in the CFD analysis for the current study.

  • Variable specific heat, thermal conductivity, and viscosity for the gas phase
  • Standard k/e turbulence model
  • Gray gas radiation using the P?1 model with variable absorption coefficients based on water and carbon dioxide content
  • Thermal buoyancy
  • Chemical reaction according to Arrhenius reaction kinetics and the Magnussen mixing model

Grid Development

A body-fitted grid system is developed for the boiler geometry represented in this model, see Figure 2. Only a part of the boiler is modeled for computational efficiency. The computational grid, shown as Figure 3, includes the radioactive section of the furnace from a point between the wood grate and the nose of the furnace (bottom) to the superheater section (top), with the outlet of the model corresponding to the beginning of the convection section. The main body of the furnace measures 5.12 m (16.8 ft) by 5.12 m (16.8 ft), and has one burner and two over-fire air ports on each side.

Figure 3, Sander Dust Burner/Boiler Computational Grid

The burners are located at an elevation above the grate with a separation of 1.92 m (6.3 ft). The over-fire air ports are positioned 2.33 m (7.64 ft) above the burner centerline, with one port 8.4 cm (3.31 in) to the left of the burner and the other port 71.2 cm (28 in) to the right of the burner. The over-fire air have an open area equivalent to an 18 cm hole (7.07 in).
This model has three inlets and one outlet. Flow enters the computational domain from the bottom inlet (pre-burned gases from the wood grate), from the sander-dust burners, and from the over-fire air ports. The actual boundary conditions used for each inlet will be discussed in the next section. The superheater tubes at the top of the furnace are modeled as a simple pressure drop, and the outlet is located at the beginning of the heat exchanger tubing. The remainder of the model boundaries are all heat transfer walls. The main cross-section of the furnace is modeled with 25 by 25 cells, or about one cell every 20 cm. Overall, the model contains 51,000 cells, which is sufficient to provide a high degree of accuracy.

Inlet and Boundary Conditions

Flow into the bottom of the computational domain is simulated as pre-burned flue gas coming from the wood grate. The total flow rate is 30,613 kg/hr (67,500 lb/hr) of gas at 872°C (1600°F). The composition of this gas is shown in Table 2. Flow through the burners is approximated by a mixture of methane and air, which delivers a total firing rate of 25.2 MW (86 mmbtu/hr) with 80% of the stoichiometric air (i.e., fuel rich). The flow velocity angles have been chosen to simulate burners that are firing straight into the furnace with no tilting. In addition, a swirl velocity component was added to provide a 20° swirl angle. The temperature of the mixture is 132°C (268°F). Flow through the over-fire air ports delivers 40% of the stoichiometric air at a velocity of 34.5 m/s (113 ft/s) and a temperature of 149°C (300°F).

Table 2, Wood Bark Flue Gas Composition

The superheater tubes at the top of the furnace are modeled as a simple pressure drop, which is approximated as a perforated plate with 50% open area. The walls of the furnace were specified to be 3.2 mm (1/8 in) thick steel plate with a constant temperature of 183°C (360°F) on the outside and a radioactive emissivity of 0.8 on the inside.

CFD Model Results

Many versions of this model were run to determine the effectiveness of different geometric configurations and different firing angles. Initial results indicated insufficient penetration into the corners of the furnace, which leads to poor mixing and incomplete combustion at the top of the furnace. After evaluating a number of configurations, a final configuration is obtained consisting of two over-fire air ports above each burner with a diameter of 18 cm (7.07 in) and firing at a 10° angle toward the outer walls. The sander dust burners are configured with 1.92 m (6.3 ft) separation and fire straight into the furnace.

In this arrangement, the average flue gas flow through the superheater has an oxygen concentration of 3.2% by volume. The effectiveness of the mixing from the fuel rich sander dust burners is shown in Figure 4 where sufficient residence time exists to allow the volatile FBN to evolve in an oxygen deficient environment. This is estimated to produce the required reductions in FBN conversion to NOx to bring the total boiler NOx emissions into compliance. The effects on the temperature contours within the furnace are indicated in Figure 5. These CFD data are then input to a particle combustion model which evaluate that the carbon loss and particulate emissions for the boiler are also within regulatory compliance.

Due to the high temperature sub-stoichiometric zones created in the furnace by this staging, areas exist that are far above the ash fusion temperature of this fuel. This results in the resultant fly ash being in the molten phase until it has time to mix with and be cooled by the secondary combustion air. CFD modeling results are used to verify that sufficient space and mixing are present to prevent this molten ash from impacting the boiler walls and forming slag deposits. Evaluating the furnace geometry to verify that sufficient area exists for staged combustion is one of the critical design factors affecting its use in any application. The cleaning costs, downtime, refractory damage, and efficiency loses associated with slag buildup requires the careful design of a system that utilizes the maximum staging area possible without risking slagging. The use of CFD as a design tool allows engineers to optimize this space usage, and avoids the use of un-necessarily large furnaces based on over-conservative design factors.

Figure 4, Sander Dust Boiler Oxygen Concentrations

Figure 5, Sander Dust Boiler Temperature Contours


Two solid fuel case studies have been presented, showing the different ways in which the same fundamental staging techniques can be applied to different furnace geometry’s. The sander dust being fired in both cases contains significant amounts of nitrogen from bonding glues and has a high propensity for NOx formation when fired. A staged NOx reduction technique is employed in both cases where the sander dust burners are operated with sub-stoichiometric amounts of air producing a fuel rich zone in the furnace. Finally, the balance of the combustion air is added, by front wall ports in one case and over-fire air ports in the other, to complete fuel burn out and cool the combustion products. In the second example, the design and effectiveness of the over-fire air ports above the sander dust burners was evaluated with CFD for penetration and mixing effectiveness.

In the first example case an actual NOx reduction of 40-50% over that of un-staged combustion was easily achieved in field trials. The NOx emissions with staged combustion were 70-75% lower than what was predicted with un-staged combustion. In the second example case the system has been designed utilizing staging data obtained from the first example and also utilizing the latest Computation Fluid Dynamics modeling techniques. In this case we have targeted a NOx reduction of over 70% compared to un-staged combustion. That project is currently undergoing field installation and actual emissions data will be obtained and compared with the results of this study.

The use of biomass waste products with calorific value for energy production means provides industry with a cost effective and environmentally sound technique of production. Further research is required on chemical kinetics, computer model development, pollutant formation mechanisms, and solid reaction kinetics are required. Such tools would be used by the industrial combustion designer in the development of more advanced techniques for handling difficult biomass energy applications and continue to reduce the emissions resulting from them.

  1. Maloney, J, Modern Particleboard and Dry-Process Fiberboard Manufacturing, Miller Freeman Press, 1993.
  2. Drennan, S.A., “Waste to Energy Solutions for Industry”, Paper presented at 1997 AFRC Meeting, 1997
  3. Fry, M., “NOx and the Panel Board Industry”
  4. Allen, W.M., “The Influence of Wood-Residue Nitrogen Content on NOx Emissions and Informational Needs”, 1993