Industrial applications using tunable diode laser technology for process gas monitoring are often faced with technical challenges because of dynamic operating conditions in the presence of high particle densities and high temperature. Furthermore, issues related to alignment stability and maintenance requirements must be overcome for industry acceptance of the sensing technology. To address these technical challenges a novel near infrared tunable diode laser system for monitoring CO, H2O and gas temperature is presented. The system incorporates balanced ratiometric detection and a variable laser power delivery scheme allowing the launch laser power to vary between 2-248 mW while maintaining a constant reference power. Feedback control is used to adjust the level of laser power delivered to the process based on the light transmission through the measurement zone. Results are presented using the system on a 500 kW oxy-fuel pilot furnace with controlled particle injection to simulate industrial conditions in preparation for field test campaign measuring the off-gas of an electric arc (EAF) steel-melting furnace. For the industrial test, monitoring on the EAF process can be considered one of the harshest environments to perform a measurement with particle densities rising above 100 g/Nm3 and temperatures up to 1800°C. In addition, special requirements are needed to integrate the sensor into the process because of the high level of mechanical vibration, high and varying ambient temperatures, EMF interference sources, and protection against flying debris.
Energy intensive industries such as steel, aluminum, and glass require combustion processes that are characteristically at high temperature with high levels of particulate matter. Monitoring and control of these processes for improved efficiency, pollutant reduction, and product quality requires a sensor adaptable for such harsh environments. Traditional industrial monitoring relies on extractive sampling that requires frequent maintenance due to probe plugging or corrosion and routine calibration. In addition, capturing the temporal behavior of the process can be problematic with extractive sampling systems because of the slow response time associated with the sampling line lengths and slow response analyzers. To meet the demands of these harsh combustion processes the ideal sensor would perform in-situ process monitoring, require little or no maintenance and provide real-time process information
The use of tunable diode lasers based on absorption monitoring overcomes many of the problems associated with conventional extractive sampling. However, the majority of industrial combustion processes will undergo temperature variations along with changes in the atmosphere oxidation or reducing state during normal operation. Therefore, temperature monitoring along with key combustion species monitoring that describes the atmosphere e.g., O2 and CO, is often necessary for optimal process control. The temperature is not only a useful parameter describing the state of the process, but is needed to accurately determine the species concentration since the absorption measured is dependent on temperature. To monitor both reducing and oxidizing combustion atmospheres in addition to gas temperature requires a diode laser system capable of multiple species monitoring. Here we describe an industrial prototype system operating in the near-infrared for simultaneous monitoring of O2 (.76 μm), CO (1.5 μm), H2O (1.5 μm) and gas temperature. The prototype system addresses the issues of added complexity with multiple species monitoring by using only two diode lasers and a beam launch and receiver optical design to discriminate the vastly different laser wavelengths while suppressing background radiation noise and beam steering from thermal gradients. Measurement results using the system for industrial process monitoring on a 100-ton/hr steel reheat furnace are presented. The measurements in this test were conducted at different zones in the furnace and at different heights relative to the processed material. The results show dynamic variations in concentration and temperature that could aid in improved atmosphere control.
12 To address the inherent issues with extractive sampling, Air Liquide and PSI are collaborating on the development of an in-situ multi-functional near-IR tunable diode laser system. The system is specifically targeted for application in harsh combustion environments with flue gas temperatures > 1600 degree(s)C and high particle densities. The multiplexing capability of the diode laser system allows near simultaneous detection of CO, O2, and H2O. These are essential species in characterizing the combustion state of the process, i.e., fuel-rich or fuel-lean, and the flue gas temperature. Sensor development and testing are conducted on a 700 kW oxy-fuel pilot furnace to evaluate the performance under simulated industrial conditions. Here we present pilot test results for dynamic stoichiometry changes, effect of particle entrainment, and air infiltration monitoring.
With stricter environmental regulations, optimization of the combustion process for reduced pollutant emission and higher fuel efficiency has become an industry objective. To achieve these objectives, continuous monitoring of key processes parameters such as temperatures, fuel and oxidant input, and flue gas composition is required. For flue gas composition monitoring conventional extractive sampling techniques are typically used. However these techniques suffer from slow response time due to long sample lines and are sensitive to plugging problems when applied to particle-laden flows. Using in-situ monitoring with near-IR tunable diode lasers (TDL) eliminates the problems encountered with extractive sampling. The chemical species to be monitored dictates the wavelength range of the diode lasers used. These lasers are rapidly tuned over an absorption line to obtain concentration along the line-of-sight path. In addition, gas temperature can be measured by scanning the laser over multiple rotational lines of a target molecule. Here we demonstrate the feasibility of using TDL's for in-situ O2 monitoring on the exhaust end of Air Liquide's oxy-fuel pilot furnace. Tests were conducted at various operating conditions and compared with conventional extractive sampling measurements. The response time of the technique is demonstrated by measurements conducted on a dynamic system where the fuel flow is oscillated at low frequencies. In addition, to study the effect of dirty gas streams typically found on industrial processes, seed particles were introduced into the burner to simulate particle-laden flows.
With stricter environmental regulations optimization of the combustion process for reduced pollutant emission and higher fuel efficiency is a major objective for manufacturers. The promotion of oxy-fuel combustion is one alternative technology that has been demonstrated as a means for manufacturers to meet their environmental objectives. Despite the benefits oxy-fuel combustion can offer further optimization using monitoring and control techniques are still needed. Here we present a novel method for monitoring and controlling oxy-fuel burners by strategic placement of optical sensors. The sensors are integrated into an industrial oxy-fuel capable of withstanding harsh environments. Radiation from the flame at selected wavelength regions is collected by fiber optics attached to the burner and transported to a miniaturized PC-based spectrometer. The spectral information obtained is used to construct a neural network (NN) model that relates the real- time signal collected to burner operating parameters such as, stoichiometry, power, and fueled and/or oxidizer changes. This processed information from the NN can then be used in a control-loop for adjusting and optimizing combustion parameters or alerting operators of potential burner problems. Examples of using this technology on AIr Liquide's pilot furnaces in both the US and France and from an industrial glass melting tank will be presented. The potential of the sensor and NN approach is demonstrated for both conventional burner and an advanced wide flame burner. The results show that both stoichiometry and power changes can reliably be detected by use of the optical sensors. In addition, an example demonstrating the method on oxy-fuel oil flames to monitor oil atomization quality and stoichiometry will be presented.
Oxy-fuel technology that uses high purity oxygen in place of air has demonstrated to be a cost-effective method for improving melting operations providing benefits in fuel savings, reduction in capital investment, and reduction of NOx and particulate matter. These benefits are evident in the glass industry where an estimated 15 percent of the US production has already been converted to oxy-fuel. However conversion from air-fuel is complicated by the drastic differences between the combustion characteristics such as flame temperature, momentum, flame chemistry, and heat transfer properties. For optimum performance using oxy-fuel combustion well-characterized burners with knowledge of the temperature in the combustion space is needed. Temperature characteristics for a given burner design are useful for both validation and parameter adjustment in 3D numerical models and optimizing the flame to the process. Because of the higher temperatures and steeper gradients in oxy-fuel flames traditional measurement techniques used on industrial flames, e.g., suction pyrometer or coherent anti-stokes Raman spectroscopy have limited use. Here we present results using a modified line reversal technique to monitor the emission and transmission of oxy-fuel flames seeded with sodium. The technique provides real-time information on the line-of-sight temperature observed from industrial scale turbulent flames.
Real-time analysis of XeCl gas mixtures are described using mass, infrared, and ultraviolet spectrometry while monitoring output power from the laser. In separate experiments, the effects of dominant gaseous impurities are measured individually. Combining these data into a model, power loss from the XeCl laser can be predicted.