This work investigates combustion of liquid fuels in a 150 kW pilot scale FLOX-combustor to reduce NOx emissions. Emission thresholds for NOx continue to be refined towards lower targets. Therefore, it is necessary to identify and develop means to reduce NOx emission levels of firings effectively. Primary measures aim at abatement of NOx emissions by preventing its formation during combustion. Secondary measures on the other hand, effectively remove NOx which was generated during combustion in a post-combustion stage. Flameless combustion is a primary measure which significantly reduces thermal NOx formation which is the main source of NOx emissions for most gaseous and liquid fuels. This is achieved by mixing fuel, combustion air and recirculated flue-gas intensively to generate a hot, diluted mixture of low oxygen and low fuel content above self-ignition temperature. This leads to a reduction in reaction speed, an enlarged reaction zone and lower peak temperatures compared to conventional combustion.
Atomization of the fuel is an important aspect of liquid fuel combustion. In contrast to firing gaseous fuels which can mix with flue-gas and air right away, the evaporation of the fuel and subsequently mixing with combustion air and flue-gas is an additional hurdle in achieving a hot, diluted, homogenous mixture.
Twin-fluid atomizers of Internal Mixing Air Assist type were characterized with water and pressurized air as process media via laser diffraction. Experiments showed that the spray was fully developed already after 50 mm from the spray nozzle. An increase in atomization air pressure and higher orifice diameters both lead to an increase in mass flow of atomization air. A higher atomization air mass flow equals a higher amount of kinetic energy available for atomization. Therefore, an increase in atomization air flow results in lower mean droplet diameters. Experimentally obtained data for water matched very well with calculated droplet sizes according to the model of Rizk & Lefebvre when assuming annular flow and speed of sound in the gas flow. Mean droplet diameters in the spray for the fuels used in this work, namely free fatty acids (FFS) and light fuel oil, could be estimated with the model of Rizk & Lefebvre. Mean droplet sizes were higher for FFS due to the higher viscosity and a lower gas-liquid-ratio for FFS. Drop lifetime for FFS is significantly higher due to larger drop sizes for FFS and lower evaporation rates.
Combustion experiments were conducted in the FLOX-combustor at IFK with light fuel oil and FFS. The influence of atomization, spray angle, atomization air pressure, orifice diameter and fuel preheating (FFS only) as well as process related parameters such as thermal input and air ratio on NOx, CO and N2O emission levels in the flue gas as well as temperature profiles in the furnace were analyzed. Selected process conditions were investigated more in detail using a suction pyrometer, generating temperature and gas concentration profiles in the combustion chamber.
For both fuels a decrease in NOx emission levels was observed under the following conditions: increase in air ratio, decrease in thermal input, increase in atomizing air pressure, increase in atomizer orifice diameter and at a spray angle of 0°. Fuel preheating was only tested for FFS and showed a reduction in NOx as well.
Air ratio and thermal input affect peak temperatures in the combustor. A reduction in peak temperatures also results in lower NOx emissions. When temperatures drop too low, CO levels increase. Atomization related parameters such as atomizing air pressure, spray angle, orifice diameter and fuel temperature significantly affect mean drop size of the spray which in turn affects how quickly these drops evaporate. Smaller droplets and high evaporation rates are beneficial for quick evaporation and quick mixing of fuel vapors with gas leading to a reduction in NOx and CO emission levels. Slow evaporation and very large droplets result in incomplete combustion and increased CO emissions levels.
Levels of N2O in the flue-gas were generally low for all investigated conditions. Maximum emission levels were 1.7 ppm which is deemed insignificant.
Suction pyrometer measurements for light fuel oil showed that the reaction zone is closer to the burner nozzle when atomizing air pressure is low and/or thermal input is high. The volume of the reaction zone was found to shrink, a rise in peak temperatures was observed and an increased oxygen concentration was found at the beginning of the reaction zone. These are all indications that a reduction in atomizing air pressure and an increase in thermal input gradually result in a process change from flameless combustion towards a lifted flame. NOx formation mostly occurs in an area of low temperature (< 700 °C) and assumed high fuel concentration which indicates formation via Fenimore- or NNH-pathways.
Due to the larger droplets and slower evaporation rate of FFS compared to light fuel oil, higher furnace temperatures (=higher thermal input) are required for complete combustion. Measured temperature profiles show a uniform temperature distribution in the combustor suggesting slow distributed combustion over a large volume although a flame is visible. Process-related NOx emission levels for FFS were found to be low compared to light fuel oil. Absolute emission levels of both fuels were almost equal, but FFS contain more than double the amount of fuel-nitrogen compared to light fuel oil. Assuming conversion levels for both fuels are the same (85 %), less NOx is formed via different pathways for FFS.
Using both fuels in the same combustor proved to be challenging due to the different heating values and especially the differences in atomization and evaporation between both fuels. For both fuels process windows of stable combustion at low NOx emission levels did not overlap. Fuel pre-heating for FFS could only partly mitigate this issue.
Due to the amount of solved homogeneous catalysts in FFS, a remainder of the biodiesel process they originate from, deposits containing sodium, potassium and silicon as well as severe corrosion of the combustor were observed after firing FFS. A continuous decrease in measured NOx and CO levels was observed scaling with the amount of FFS which had been fired in the combustor. Reduced emission levels of NOx were also found with light fuel oil, although a slight increase in CO was observed. Gas concentration measurements via two independent sample positions, two independent sample preparation tracks and two independent gas analyzers verified this finding. However, a causal correlation between change in gas concentration measurements and deposits in the combustor could not be found.