This MGS system overcomes the technical issues of conventional gas-monitoring technologies and offers potential use in various in-situ furnace monitoring applications.

Three-dimensional modeling of reheat furnace conditions

Control of the combustion environment in the metals processing industry is critical to improve process efficiency, reduce pollutant emissions and meet product quality demands. Sensors can provide information on gas composition in the combustion space for feedback to control algorithms to adjust and optimize the process. Key gas species that need to be monitored in combustion-control applications are O2 and CO (generic to all combustion processes) because they describe the oxidation state of the atmosphere.

Gas monitoring methods

Extractive sampling techniques are the accepted practice used to monitor industrial gas compositions. The technology involves inserting a water-cooled probe into the process and withdrawing a sample through the probe. The gas is then conditioned using a chiller to remove water vapor and a filtering system to remove particulates before analysis using conventional gas monitoring instrumentation. Disadvantages of the method include slow response times (ranging from tens of seconds to several minutes), probe maintenance issues associated with plugging and corrosion and frequent calibration checks to adjust instrumentation drift.

Alternative sampling methods include in-situ measurement techniques, such as permeable ceramics tubes that encapsulate a beam from a light source to measure (via absorption) the gas that diffuses through the tube, and zirconium-oxide probes that measure the resulting potential across the surface exposed to the combustion gases and ambient conditions[1]. Permeable ceramic tube instruments, such as the Procal Pulsi 200LR (Procal Analytics Ltd., Peterborough, UK), have a limited operating temperature of approximately 300 C (570 F), which is adequate to monitor stacks, but not combustion space. Zirconium probes can operate up to 1600 C (2910 F), but can only detect O2 and are not recommended for use in reducing environments. Hybrid systems that combine a zirconium probe with an extracted side stream of process gas for CO and unburnt hydrocarbons extends the technique to include reducing-atmosphere monitoring. Both zirconium and permeable tube probes are vulnerable to plugging problems in high particle-laden processes. In addition, ceramic probes are susceptible to thermal shock requiring frequent replacement in processes that undergo large temperature fluctuations.

New sensor tackles issues

To address the technical issues of conventional monitoring technologies, Air Liquide and Physical Sciences Inc. developed a multigas sensor (MGS) system based on tunable diode laser (TDL) technology in the near-infrared (NIR) region. The widely available small, robust lasers operate near room temperature, are compatible with fiber optic components, and operate over a wavelength range of 600 to 2,000 nm, overlapping the same spectral region for monitoring many combustion species of interest, e.g., O2, CO, CO2, H2O, CH4 and NO[2].

The MGS system developed in this work monitors O2, CO and H2O and gas temperature. Primary objectives in the system development was to identify the spectral lines to monitor the targeted species and minimize system complexity and overall cost by reducing the number of lasers needed. A spectral survey was conducted on a well-characterized laboratory burner to identify a spectral region where multiple H2O lines and the CO line are accessible using a single, distributed-feedback (DFB) diode laser. Results were integrated into an industrial prototype system, which will be evaluated on a 500-kW oxyfuel pilot furnace at Air Liquide's research facility[3]. The pilot furnace allows simulating industrial conditions with particle injection and dynamic combustion conditions[4], which will aid in transferring the technology from a laboratory to an industrial environment.

Fig 1 Potential TDL installation on a reheat furnace

Industry monitoring applications

Potential in-situ furnace monitoring applications in the steel manufacturing industry include electric arc furnace (EAF), reheat furnace, basic oxygen furnace (BOF), annealing and blast furnace. Each process has different sensor needs. For example, off-gas analysis from an EAF to monitor the CO concentration is of interest to recover lost chemical energy by using controlled O2 injection. The process is dynamic, and a fast time-response, nonintrusive sensor can help improve process control and lower maintenance. While an EAF environment is harsh on a sensor due to high particle densities, mechanical vibration, strong electromagnetic fields, high temperatures, and volatile operation, Allendorf demonstrated the feasibility of off-gas analysis using a mid-IR diode laser system to detect CO, CO2 and H2O[5]. However, the need for cryogenic cooling and incompatibility with fiber optic components makes the system unattractive for routine industrial monitoring.

The first industrial tests for the new prototype system will be conducted on a 100 t/hr walking beam reheat furnace at Charter Steel, Saukville, Wis. Steel billets introduced into the furnace are directly exposed to the combustion gases, which oxidize the steel, forming a scale layer[6]. Because the scale represents a product loss, the manufacturer strives to minimize scale by controlling the furnace atmosphere, generally by controlling the air-to-fuel ratio via measuring and controlling fuel and air flow rates. However, it is difficult to control low O2 levels near stoichiometric conditions using this approach because of errors in flow-rate measurement, fuel composition changes, and limited individual burner control. Many reheat furnaces incorporate zirconia probes to control O2 level, which are point measurements best used for global measurements in regions where the composition is homogenous. Local gas composition measurements using zirconium oxide probes or extractive sampling techniques are difficult due to process geometry constraints (wide furnaces). Other furnace-control issues include decarburizaton, billet-temperature homogeneity, and NOx_minimization.

A walking-beam furnace consists of four zones (recovery, preheat, heat and soak). Billets enter the furnace against the exhaust gas flow direction, which provides a means to preheat the billet. Figure 1 shows a schematic of a TDL installation on reheat furnace with the system located in the soak zone. Output from the measurement is used in a control scheme to adjust a bank of burners as necessary based on observed measurements. The nonintrusiveness of the TDL measurement technique allows positioning the measurement beam close to the billet surface with laser beam launch and receiver. This allows measuring the average gas composition along the beam path near the product surface where the measurement could aid in predicting or minimizing scale and decarburization. Burner adjustments could involve such actions as trimming the O2 level to a desired value or providing a reducing atmosphere in a selected zone.

TDL measurements on a 300 ton/hr heavy fuel oil-fired walking beam reheat furnace using two commercially available TDL instruments at different furnace zones were compared with a conventional in-situ zirconia oxide probe analyzer in work by Niska et al.[7]. Both instruments were set-up to monitor only O2, limiting the dynamic measurement range to O2 levels of >3% due to the poor temperature-measurement accuracy based on the ratio of two O2 absorption transitions. At low O2 levels, the weak, higher rotational level O2 absorption line was not strong enough to accurately determine temperature. The MGS system presented here uses H2O to conduct thermometry measurements, thereby allowing temperature monitoring under oxidizing and reducing conditions.

Fig 2 Schematic of the multispecies diode laser configuration

MGS monitoring principle

Figure 2 shows the essential features of the MGS system. An AlGaAs DFB diode laser is used to monitor O2, and a single DFB InGaAsP laser is used to monitor CO and H2O. Each laser is controlled using an independent current and temperature controller. Both lasers are fiber optically coupled with the output fed into a splitter. A fiber optic delivery system has the advantage of allowing the sensitive lasers and associated electronics to be located in a safe environment away from the process. The resulting modulation (absorption) of the radiation is monitored at the receiving side by a detector. The signal from the detector is sent to a balanced ratiometeric detector (BRD) circuit, which electronically balances the photocurrents[8] from the detector with a portion of the laser radiation split off as a reference.

Concentration monitoring of the species is obtained by tuning the diode laser over an absorption transition while monitoring laser radiation absorbed along the measurement path. The amount of laser radiation attenuated is determined by the temperature-dependent line strength S(T) and a line shape function, whose product is the absorption coefficient. Line strength is related to the internal energy distribution and is governed by Boltzmann statistics, whereas line shape is related to temperature and pressure broadening mechanisms.

Acquiring the true absorption line shape as provided by the BRD signal acquisition technique allows calculating the absolute number density by applying Beer's Law (of absorbing molecules per volume) without instrument calibration. In addition to removing temperature and pressure effects in the line shape, sensitivity to broadband absorbers or scatters are reduced because the absorption being monitored is relative to the baseline.

The temperature dependence of the line strength requires that the gas temperature be known for concentration measurements. For processes that operate near steady state conditions, the temperature spread may be known and relatively narrow (for example, +/- 200 C). In this case, absorption transitions that show minimum sensitivity to temperature can be chosen. However, for processes having large variations in temperature (for example, batch processes such as electric arc furnace and rotary smelters), the gas temperature must be known for accurate measurements of the targeted specie concentration. The gas temperature can be obtained from the absorbance spectrum if multiple rotational lines are detected within the scanning range of a single or multiple diode laser system. The area-ratio of the integrated absorbance of each transition is related to the temperature, so the temperature can be calculated explicitly.

Fig 3 Absorbance spectrum of O2 showing the detection of several transitions for 1 m path length (T(flue) = 1470K, P(flue) = 747 torr). Spectral data at 1495K used as a reference.

An example using this approach for O2 monitoring is shown in Fig. 3 obtained from previous experimental work conducted on Air Liquide's 500-kW pilot furnace[9]. In this case, the recorded spectrum was taken at a 1 m path length at an excess O2 concentration of 10%. The large separation in energy between the R2(13) and R2(43) O2 lines (DE = 2440 cm-1) provides an accurate measure of the temperature (1470K, or 1200 C), which is approximately 4% lower than the measured value of 1540K (1270 C) using a suction pyrometer. The weak absorbance of the R2(43) line requires long path lengths and/or high O2 concentrations for thermometry, which limits the range of conditions the measurement is applicable, as experienced by Niska et al.[7].

Fig 4 Selected H2O line pair area ratio temperature sensitivity.

Spectral survey

To extend the measured temperature range, a spectral survey was conducted using H2O as the selected thermometry species. The 1 to 3 cm-1 tuning range achieved using current injection control on standard DFB lasers limits the absorption lines accessible by a single laser. A broader tuning range is possible through additional control of the temperature. However, this approach suffers from slow tuning rates due to temperature stabilization. Therefore, systems requiring multiple species monitoring often use a dedicated laser for each species. Examples of multiple species monitoring using systems based on either time-domain multiplexing[2] or wavelength domain multiplexing[10] demonstrated the concept in laboratory conditions. However, commercial TDL systems available to monitor multiple species either are limited to the low temperature regime (T <1000K, or 730 C) or are not truly multiplexed, requiring multiple systems and optical access ports. This is particularly true for O2 (0.760 um) and CO (1.5 um) detection because of the broad wavelength separation between the lasers. System integration using these wavelengths requires dedicated fiber optic components, special consideration for optical materials and antireflective coatings and dedicated detectors for each wavelength.

To minimize the complexity of the MGS system, a single laser is used to access both CO and H2O absorption transitions in the 1.5 Km spectral region. Though this spectral region is highly congested with H2O transitions, previous work reported in the literature using H2O for thermometry focused on the 1.39 Km spectral region[11], thus, a detailed spectral survey for line selection in the 1.5 Km region was required. The criteria used for line selection was based on observed line strength of H2O, proximity to neighboring CO line, temperature sensitivity, and no spectral interferences. Initial surveys using the HITRAN/ HITEMP database[12] was unsuccessful, because large discrepancies were observed between reported absorption line strengths and transition locations compared with laboratory flame measurements obtained using an external cavity diode laser and Hencken burner set-up described by Upschulte et al.[13].

Because observed H2O transitions are not well established, an empirical temperature calibration was conducted based on the H2O lines selected and conducted by varying the lab flame stoichiometry to vary the temperature. The resulting temperature sensitivity of the area ratio for the selected line pair identified is shown in Fig. 4. Furthermore, the data can be fit to an empirical exponential function given by Equation 1 where A, B, C, D and E are constants, which can easily be incorporated into the system's data-processing algorithm for real-time temperature monitoring. Results of the fit through the data show an error <5% at the higher temperature end and <8% at the low temperature end.


The detailed spectral survey identified a spectral region where a single DFB laser can access multiple lines of H2O near a detectable neighboring CO line. Although this step reduces the system complexity by minimizing the number of lasers, the inclusion of O2 detection sets the minimum number of lasers at two due to the broad wavelength coverage needed to simultaneously detect O2, CO and H2O.

The empirical temperature calibration and H2O line strength determination over a range of 1000 to 2000K (730 to 1730 C) on a laboratory burner laid the groundwork to integrate these results into an industrial prototype system. Before field-testing the prototype in Charter Steel's reheat furnace, the system will undergo evaluation on a 500-kW oxyfuel pilot furnace. The unique functionality of the furnace allows simulated, yet controlled industrial conditions, such as particle injection for dirty-gas monitoring, dynamic combustion by varying the stoichiometry at a known frequency, comparative measurement instrumentation using conventional extractive sampling and varying the combustion atmosphere from air-fuel to oxyfuel.


The authors acknowledge the support of the U.S. Dept. of Energy through contract number DE-FC07-OOCH11030.


  • E.J. Reber, State-of-the-art Sensor Developments for Process Control, Ceram. Eng. Sci. Proc., No. 7, 1986, p 1329-1338
  • M.G. Allen,"Diode Laser Absorption Sensors for Gas-dynamic and Combustion Flow", Meas. Sci. Technol., 9, 1998, p 545-562
  • J. Dugue, et al., Advanced Combustion Facilities and Diagnostics, American/ Japanese Flame Research Committees 1998 Intl. Symp., Hawaii, October, 1998
  • W. Von Drasek, et al., Tunable Diode Laser Sensor Characterization for Harsh Combustion Monitoring, 2001 Joint Intl. Combustion Symp., AFRC-JFRC-IEA, Hawaii, 2001
  • S.W. Allendorf, Application of Absorption and Emission Spectroscopies for Process Control in Commercial Steel Making, Proc., 2001 Joint Combustion Symp., AFRC-JFRC-IEA, Hawaii, 2001
  • J.S. Sheasby, W.E. Boggs and E.T. Turkdogan, Scale growth on steels at 1200¿C: Rationale of Rate and Morphology, Metal Science, Vol. 18, 1984, p 127-136
  • J. Niska and A. Rensgard, TDLAS for Combustion Gas Analysis in a Steel Reheating Furnace, Proc., 2001 Joint Intl. Combustion Symp., AFRC-JFRC-IEA, Hawaii, 2001
  • D. Sonnenfroh, S. Sewell and M. Allen, An Ultrasensitive Detection Techniques for Tunable Diode laser Spectrometers: Application to Detection of NO2 and H2O, Proc., Soc. of Photooptical Instrumentation Engineers, Vol. 2834, 1996, p 57-66
  • W. Von Drasek, et al., Multifunctional Industrial Combustion Process Monitoring with Tunable Diode Lasers, Proc., Soc. of Photooptical Instrumentation Engineers, Vol. 4201, 2000, p 133-141
  • D.S. Baer, et al., Scanned-and fixed-wavelength absorption diagnostics for combustion measurements using multiplexed diode lasers, AIAA. J., No. 34, 1996, p 489-493
  • M.G., Allen and W.J. Kessler, Simultaneous Water Vapor Concentration and Temperature Measurements Using 1.31-um Diode Lasers, AIAA J., Vol. 34, No. 3, 1996
  • L.S. Rothman, et al., The HITRAN Molecular Database: Editions of 1991 and 1992, J. Quant. Spectrosc. Radiat. Transfer., No. 48, p 469-507
  • B. Upschulte, M. Miller and M. Allen, Continuous Water Vapor Mass Flux and Temperature Measurements in a Model SRAMJET Combustor Using a Diode Laser Sensor, 37th AIAA Aerospace Sciences Mtg. & Exhibit, Nevada, AIAA 99-0518, 1999