Application of wavelength-scanned wavelength-modulation spectroscopy H2O absorption measurements in an engineering-scale high-pressure coal gasifier
A real-time, in situ water vapor (H2O) sensor using a tunable diode laser near 1,352 nm was developed to continuously monitor water vapor in the synthesis gas of an engineering-scale high-pressure coal gasifier. Wavelength-scanned wavelength-modulation spectroscopy with second harmonic detection (WMS-2f) was used to determine the absorption magnitude. The 1f-normalized, WMS-2f signal (WMS-2f/1f) was insensitive to non-absorption transmission losses including beam steering and light scattering by the particulate in the synthesis gas. A fitting strategy was used to simultaneously determine the water vapor mole fraction and the collisional-broadening width of the transition from the scanned 1f-normalized WMS-2f waveform at pressures up to 15 atm, which can be used for large absorbance values. This strategy is analogous to the fitting strategy for wavelength-scanned direct absorption measurements. In a test campaign at the US National Carbon Capture Center, the sensor demonstrated a water vapor detection limit of ~800 ppm (25 Hz bandwidth) at conditions with more than 99.99 % non-absorption transmission losses. Successful unattended monitoring was demonstrated over a 435 h period. Strong correlations between the sensor measurements and transient gasifier operation conditions were observed, demonstrating the capability of laser absorption to monitor the gasification process.
In modern energy systems , there is a need for sensors to rapidly monitor the gas composition, providing control signals to optimize the system performance for increased energy efficiency and decreased pollutant emissions . For the harsh environment of a coal gasifier in an integrated gasification combined-cycle power plant, safety requirements to handle the hot, high-pressure, toxic and explosive synthesis gas (called here syngas) and fouling by particulate make it difficult to install a gas analyzer directly into the gas flow for in situ measurements. Thus, most sensors rely on gas sampling measurements, which produce time delays and slow time resolution (minutes). In addition, although H2O is a major combustion product, many sampling-based gas analyzers such as gas chromatography (GC) require removing the water vapor before the analysis. As a result, rapid real-time H2O monitoring data are unavailable in most industrial systems.
Wavelength-modulation absorption spectroscopy (WMS) using injection-current-tuned diode lasers (TDL) has been employed for more than a decade for in situ sensors to determine gas composition, temperature, pressure and velocity in a variety of harsh industrial environments and laboratory combustors [3, 4, 5, 6, 7, 8, 9, 10, 11, 12], and commercial instruments using WMS have been available for a decade [13, 14]. Compared to direct absorption (DA) measurements, WMS can improve the signal-to-noise ratio (SNR) [15, 16, 17], avoid the need to measure a zero-absorption baseline during data collection [18, 19] and be less influenced by non-absorption transmission losses [11, 12, 20, 21]. These WMS features have proven useful for sensors in practical industrial applications, especially for applications with large and time-varying non-absorption transmission losses [21, 22, 23, 24, 25]. Although an attempt to review the use of WMS for in situ measurements in practical applications is beyond the scope of this paper, two pioneering applications of 1f-normalized, wavelength-scanned WMS-2f were performed previously in harsh environments [11, 12]. Fernholz et al.  demonstrated 1f-normalized wavelength-scanned WMS of oxygen in the exhaust from a high-pressure (12 atm) flame, and a 1998 meeting report from the same group discussed the use of this method for measurements of O2 in the effluent of a 20 MWth hazardous waste incinerator . Although the wavelength-scanned WMS sensor architecture and 1f-normalization of the signal are quite similar to this past work, our new data analysis method is quite different. The pioneering work in Refs.  and  accounted for simultaneous amplitude modulation during the scan of the laser wavelength; however, this earlier method did not account for nonlinear response of the laser tuning and was limited to measurements of small absorbance.
Here, we report application of 1f-normalized, wavelength-scanned WMS-2f to a high-pressure (15 atm) synthesis gas product stream with large non-absorption transmission losses, using a newly developed analysis scheme [26, 27] that allows fits of the wavelength-scanned WMS signal waveform to an absorption spectrum similar to wavelength-scanned direct absorption. Importantly, this new method is not limited to optically thin conditions and can account for nonlinearity in the laser modulation. Because this is the first application of the wavelength-scanned WMS method of Refs.  and  to increased pressure gases, we also present gas cell measurements taken with known conditions at pressures from 3 to 15.8 atm.
Laser absorption measurements in syngas products from coal gasification have two major challenges: high pressures broaden and blend the laser absorption transitions, and the scattering from particulate in the gas significantly attenuates the transmitted laser intensity. In previous work, we found that 1f-normalized, fixed-wavelength WMS provided a solution to these challenges, and the composition and temperature of syngas products were monitored in a pilot-scale entrained-flow high-pressure coal gasifier [28, 29, 30].
In this paper, the sensor performance was improved through the use of wavelength scanning and a new paradigm for WMS data reduction, and this improved sensor was used for long-term monitoring of the water vapor in the syngas output from a significantly larger (engineering-scale) high-pressure transport reactor facility. Compared to the fixed-wavelength WMS technique used for our earlier gasifier applications, the wavelength-scanned WMS technique is less affected by drifts in the laser wavelength or pressure shifts in the absorption transition center wavelength. In addition, wavelength-scanned WMS simultaneously determines the WMS absorption lineshape and thus minimizes the errors due to the variations of the transition’s collisional width caused by changes in gas composition. Thus, the wavelength-scanned WMS approach demonstrated in this paper is more suitable for long-term (days) monitoring in complex gas flows, where the laser wavelength may shift and variations in the collisional width of the absorption transition due to changes in the gas composition are hard to estimate.
The wavelength-scanned WMS data analysis scheme recently developed in our laboratory [26, 27] fits each wavelength scan of the 1f-normalized WMS-2f absorption waveform while varying only the integrated absorbance and the width of the lineshape of the target transition. All other laser scanning and performance parameters are fixed from characterization measurements [22, 23] in the laboratory before (and verified after) the field campaign. The use of 1f-normalization and laser characterization [22, 23] avoids the need for in situ sensor calibration often required for WMS sensors and was termed “calibration-free” in Ref. . We have found that the laser characterization parameters for near-infrared telecommunications lasers (NEL used here) have remained constant over several years of regular use in our laboratory and in the field. Even so, these parameters are verified before and after any measurement campaign. Obtaining a fit value for the transition linewidth accounts for any changes in the gas composition without the need for a large database of collision-broadening parameters (this assertion is tested as described below by comparison with the linewidth determined from a collision-broadening database and independently gas composition measurements with the linewidth determined by the wavelength-scanned WMS fits from the TDL sensor data). The normalization scheme [11, 12, 23] accounts for losses in transmission that is independent of laser wavelength over the region of the scan (e.g., time-varying scattering losses) as well as providing a signal independent of detector gain. Here, the goal is to present the results using this new scheme for in situ monitoring of water vapor in syngas products from a nearly commercial-scale gasifier, and the reader should refer to Refs.  and  for the details of the data analysis scheme.
Before assembling a prototype sensor for the NCCC field measurement, the sensor performance was investigated at increased pressures by measuring the wavelength-scanned WMS of water vapor in a laboratory cell using a DFB laser (NEL) near 1,352 nm. The integrated absorbance and collisional width of the probed transition was simultaneously determined, over a pressure range from 3 to 15.8 atm and compared with expected values. The TDL sensor was then packaged and used for in situ measurements in the syngas in an engineering-scale transport reactor high-pressure coal gasifier at the National Carbon Capture Center . The prototype sensor was used for continuous monitoring of the water vapor content in the output syngas for more than 27 days.
2 Wavelength-scanned WMS method
The analysis used here for the injection-current-tuned wavelength-scanned WMS measurements has been published previously [26, 27]. Here, we only review the key steps used to simulate the WMS-nf absorption signals to guide the discussion. For consistency with these earlier reports, the concept of a zero-absorption transmitted laser intensity, Ibg(t), is introduced; however, for practical industrial applications, this intensity cannot be measured, and other strategies to determine this quantity are described in the section below on the field measurements.
3 Laboratory test of sensor performance
Laboratory measured spectroscopic parameters [linestrength, collisional-broadening coefficients and their temperature dependence exponents (in the parentheses)] at 296 K for the target transition using the procedure from Ref. 32
Linestrength (cm−2 atm−1)
Lower state energy (cm−1)
2γH2O–H2O (cm−1 atm−1) (nH2O–H2O)
2γH2O–CO2 (cm−1atm−1) (nH2O–CO2)
2γH2O–CO (cm−1 atm−1) (nH2O–CO)
2γH2O–N2 (cm−1 atm−1) (nH2O–N2)
2γH2O–H2 (cm−1atm−1) (nH2O–H2)
4 Field measurement campaign
4.1 Gasifier facility and measurement setup
Typical conditions at the measurement location
H2O mole fraction
CO2 mole fraction
CO mole fraction
H2 mole fraction
Trace species mole fraction (H2S, NH3, etc.)
Flow rate (kg/h)
The zero-absorption transmitted laser intensity, Ibg(t), was determined using two different strategies. (1) The zero-absorption laser intensity was measured in the laboratory (at Stanford) using the laser transmitter and receiver optics and the four window flanges. (2) When the instrument was installed in the field, there was an opportunity to measure the zero-absorption transmitted intensity of the sensor with the gasifier filled with nitrogen. These two methods returned nearly identical WMS-2f background signals. In addition, the WMS background was quite small compared to the WMS signal (<2 %), which is a value similar to the fit residuals, and the WMS background could have been neglected in this high-absorbance application without significant change in the results. Any window fouling or overall transmission losses (e.g., scattering losses) were accounted for in the measurement scheme by the 1f-normalization. The scattering losses were observed to change with large changes in operating parameters (e.g., reactor pressure), but these changes were relatively slow (≪ 10 kHz) and did not affect sensor performance.
4.2 Measurement results
The wavelength-scanned WMS strategy as validated by the laboratory measurements was used to continuously monitor the water vapor content in the syngas product flow at different gasifier operation conditions. The facility operation began by igniting a propane/air burner in the reactor. When the reactor was hot enough, pulverized coal was added into the reactor to first initiate coal combustion and then begin the gasification process. The coal-fuel feeding continued for more than 10 h for this 1st start-up attempt, but was terminated at hour 52 to correct a problem elsewhere in the gasifier.
From hour 0–4, several attempts to ignite the reactor burner were made, but the propane/air flame was unstable. This abnormal process was captured by TDL measurements as the H2O mole fraction spiked and dropped rapidly several times during this period. At hour 8, a stable propane flame was established and the warm-up period began. The water vapor content of the syngas steadily increased at the measurement location during warm-up; this variation was expected as the fuel/air ratio of the propane burner was increased and as the output gas piping warmed to eliminate condensation.
A 1,352 nm DFB laser-based H2O absorption sensor employing a 1f-normalized, wavelength-scanned WMS-2f technique was developed for use in the high-pressure synthesis gas from an engineering-scale coal gasifier. The performance was first evaluated in the laboratory for pressures from 3 to 15.8 atm. The SNR was better than 30 for measurements of 0.023 mol fraction of H2O at 25 Hz, corresponding to a ~800 ppm detection limit for water vapor mole fraction for a 25 Hz bandwidth. The sensor was then successfully applied to monitor the syngas output from an engineering-scale transport reactor coal gasifier at the National Carbon Capture Center. The gasifier pressures ranged up to 15 atm (~220 psig) and temperatures up to 650 K. Continuous monitoring of water vapor level in the gasifier output with 2-s time resolution was performed by the TDL sensor for more than 500 h, including the periods of burner ignition, combustion heating with a propane flame, coal combustion, coal gasification and reactor shut down via coal feed termination. As expected for coal syngas applications, beam steering and particulate scattering were severe during the measurements, resulting in less than 10−4 light transmission. With 2-s time resolution, the TDL sensor captured the time-varying changes in the water vapor level in the gas exhaust that could be correlated with the changing fuel content in the reactor. The variation of H2O in the syngas due to the batch feeding of coal had been anticipated, but was observed for the first time by the TDL sensor as the GC analysis of the syngas sampled from the facility did not have the time resolution needed to observe this behavior. These facts strongly suggest that the observed fluctuations in the gas composition observed by the TDL sensor were fluctuations in the gas composition, and thus the relative measurement errors in the water vapor mole fraction were smaller than 1 % (these small variations were less than other relative or absolute syngas composition data available for comparison). These results demonstrate the feasibility of using TDL sensors to provide real-time control signals to optimize the gasification.
A novel method of WMS analysis enabled the 1f-normalized, wavelength-scanned WMS-2f measurements to determine the collisional width of the target transition via lineshape fitting. Comparisons between the measured results and the expected values provided better than 3 % agreement (this is perhaps fortuitously good given the uncertainty of the GC measurements of the synthesis gas composition and the collision-broadening-coefficient database). The agreement of measured and modeled transition linewidth suggest that wavelength-scanned WMS can be used, similar to scanned-wavelength direct absorption, to determine the absorber mole fraction without knowledge of the transition collisional width prior to the measurement. This is important for ensuring accuracy of the measurements, as for many applications the gas composition can be difficult to estimate, and the uncertainty in estimating the transition collisional width can result in large errors in determining the absorber mole fractions via traditional analysis of WMS. This successful demonstration of water vapor measurement in the syngas output of a large-scale industrial facility shows advantages of using wavelength-scanned WMS techniques. Based on these successful H2O measurements, future work will include extended wavelength-scanned WMS TDL sensors for other major species in the syngas output such as CO.
The Stanford research was supported by the Electric Power Research Institute with Mr. Jose Marasigan as technical monitor and by US Department of Energy NETL with Dr. Susan Maley as technical monitor. The operation of the NCCC by Southern Company Services was supported by DoE.
- 2.S.J. Clayton, G.J. Stiegel, J.G. Wimer, US DoE report DOE/FE-0447 (2002)Google Scholar
- 12.V. Ebert, K.-U. Pleban, J. Wolfrum, in In situ Oxygen-Monitoring Using Near-Infrared Diode Lasers And Wavelength Modulation Spectroscopy, OSA 1998 Technical Digest Series Vol 3: Laser Applications to Chemical and Environmental Analysis, paper LWB3 (1998)Google Scholar
- 14.S. Lundqvist, P. Kluczynski, in Process Analytical Applications in the Mid-Infrared, Proceedings of the SPIE 7945, Quantum Sensing and Nanophotonic Devices VIII, 79450N (January 24, 2011). doi:10.1117/12.871571
- 33.L.S. Rothman, I.E. Gordon, Y. Babikov, A. Barbe, D. ChrisBenner, P.F. Bernath, M. Birk, L. Bizzocchi, V. Boudon, L.R. Brown, A. Campargue, K. Chance, E.A. Cohen, L.H. Coudert, V.M. Devi, B.J. Drouin, A. Fayt, J.-M. Flaud, R.R. Gamache, J.J. Harrison, J.-M. Hartmann, C. Hill, J.T. Hodges, D. Jacquemart, A. Jolly, J. Lamouroux, R.J. LeRoy, G. Li, D.A. Long, O.M. Lyulin, C.J. Mackie, S.T. Massie, S. Mikhailenko, H.S.P. Müller, O.V. Naumenko, A.V. Nikitin, J. Orphal, V. Perevalov, A. Perrin, E.R. Polovtseva, C. Richard, M.A.H. Smith, E. Starikova, K. Sung, S. Tashkun, J. Tennyson, G.C. Toon, Vl.G. Tyuterev, G. Wagner, The HITRAN2012 molecular spectroscopic database. J. Quant. Spectrosc. Radiat. Transf. 130, 4–50 (2013)CrossRefADSGoogle Scholar