LOS GATOS RESEARCH CASE STUDY Fast Industrial Process Monitoring: Trace H2S Measurements in Reformed JP-8 Fuel Authors: Subir Roychoudhury and Christian Junaedi, Precision Combustion, Inc., North Haven, CT, and Feng Dong and Manish Gupta, LGR, Inc. Introduction Precision Combustion, Inc. (PCI) has evaluated LGR’s H2S portable electronics applications because of the weight and Analyzer for monitoring trace levels of hydrogen sulfide lifetime limitations of current battery technologies. Fuel (H2S) in real-time in reformed military logistics fuel (JP-8) cells are an attractive potential source of electrical power processed in an autothermal fuel reformer. Measurements for these applications because they produce less noise were made before and after the reformed JP-8 fuel was and zero emissions; this makes them harder to detect in passed through a desulfurizing bed downstream of the covert operations and they are demonstrably greener than reformer. This is an important “proof of principle” application traditional power sources. since the high hydrogen content in reformed liquid fuels and other hydrocarbons (e.g., Syngas) is widely viewed as a convenient feedstock to power fuel cells with efficiencies as high as 60%.1 However, even trace (ppm) levels of sulfur compounds in reformate gas can quickly deactivate the cells by poisoning their anodes.2 Moreover, existing sulfur monitoring technologies have significant limitations for this application. This is also a challenging application for any gas analyzer since the gas reformate is a complex mixture containing a wide range of concentrations of other species. This makes high selectivity, i.e., immunity to interference, a mission-critical requirement. Reformed Fuels and Desulfurization Since their early use in the NASA space program, the use of fuel cells has been pioneered in numerous applications as an alternative to grid power and/or battery storage. Systems up to hundreds of kilowatts have been demonstrated for stationary back-up (e.g., hospital) applications. The US military and others are considering replacing battery packs with solid oxide fuel cells (SOFCs) in certain mobile and There are several types of fuel cells; however, they all produce electricity by oxidizing hydrogen (or sometimes a simple hydrocarbon, such as methanol) into water. This is a relatively cold process that occurs catalytically between an anode and a cathode separated by some type of electrolyte. Suitable quantities of hydrogen can be created by “reforming” petrochemical fuels (e.g., diesel, kerosene, military logistic fuels, JP-8, etc.) or by gasification of coal in a Syngas generator. However, all these fossil fuel-derived sources of hydrogen contain high levels of sulfur compounds, such as mercaptans. These can poison the catalysts used in early type fuel reformers, but sulfur-tolerant alternatives have been successfully developed where the reductive reforming conditions also convert the miscellaneous sulfur compounds into hydrogen sulfide (H2S). However, this H2S has to be scrubbed from the reformed fuel mix because even trace (ppm) levels of H2S or other sulfur compounds (e.g., COS) will quickly poison the anodes used in SOFCs. This sulfur scrubbing is accomplished using a sulfur sorbing “bed” located between the reformer and the fuel cell. www.lgrinc.com Each sulfur removal bed has a finite capacity to absorb sulfur compounds (principally in the form of H2S) before it must be replaced. For example, military specifications using reformed JP-8 are typically targeted to handle 100 ppm (by weight) sulfur for 1000 hours. The finite lifetime, and the potential of catastrophic, sulfur-induced damage to SOFCs, has thus created a clear need for a method to detect trace H2S in gas streams of varying complexity and composition. This requirement covers developers, manufacturers and users of fuel reformers, sulfur removal beds, and SOFCs. Unfortunately, both types of established sulfur sensors are not suitable for this application. Specifically, optical sensors based on UV absorption are susceptible to interference from UV absorption by the high levels of CO2 in reformed fuels. On the other hand, solid-state sensors based on doped semiconductor films exhibit interference with water vapor and carbon monoxide (CO), which are both present in high Figure 1. This data shows the linearity of the LGR analyzer over a very wide dynamic range. The sample points were obtained using 1054 ppm H2S in N2 that was precision diluted in dry air using mass flow controllers. concentrations in reformed fuels. Also, their response is non-linear, restricting their absolute accuracy to a very small dynamic range (e.g., 0-10 ppm H2S).3 Combining OA-ICOS with Chemometric Analysis LGR analyzers monitor trace gas concentrations and/or stable isotope ratios by means of a sensitive optical absorption measurement using an integrated tunable laser that delivers multiple decades of linear dynamic range – see figure 1. But, unlike older optical techniques, such as conventional cavity ringdown spectroscopy (CRDS), LGR analyzers are based on Off-Axis Integrated Cavity Output Spectroscopy (OA-ICOS), which delivers a detailed scan of an extended spectral region with every measurement. This is a critical advantage in this reformed fuel application because of the presence of potential interference from absorptions by numerous other gases in the flow. As a result, several gases may be simultaneously quantified from the measured high resolution absorption spectra using standard chemometric techniques. Figure 2. This figure shows fully resolved absorption spectra (denoted “convoluted spectra”) recorded by the LGR Analyzer, as well as absorption spectroscopic basis sets of individual species (denoted for H2S, CO2, H2O, CH4 and C2H4) in the wavelength region near 1.59 microns. These detailed spectral scans, along with a proprietary chemometric fitting routine, allow simultaneous measurements of multiple constituent gases in complex flows. www.lgrinc.com Chemometric software packages are widely used in near-IR spectroscopy and other fields. First, in the “learning phase,” the instrument is used to record data spanning the full extent of expected conditions – in this case, multiple different H2S trace concentrations accompanied by deliberate variations in the host gas composition. The software then fits the known H2S (and CH4, C2H4 and CO2) concentrations to the observed spectral data using multivariate analysis. This yields an algorithm that the instrument’s processor uses to convert future data into absolute concentrations of all these gases. Overview of Results After confirming the linearity of the LGR analyzer in unmixed samples (see figure 1), we then conducted extensive tests of the LGR analyzer’s precision and Figure 3. Typical data from in situ measurements of the product stream from a running JP-8 fuel reformer accuracy using precision diluted and mixed realistic 1%, consistent with the indoor absolute humidity). Upstream samples to mimic the mixtures typically found in reforming of the desulfurizer bed, the H2S level approached 27.00 ± JP-8. Analysis of these tests showed that the analyzer 0.42 ppm (1σ). The H2S value is consistent with the expected provides rapid (2 s) and precise (±0.1 ppm) measurements value in the gas phase after dilution and molar expansion, of H2S in reformate gases over a wide dynamic range (0- based upon the 360 ppmw sulfur in the JP-8 feedstock. 1000 ppm) with a low detection limit (3σ = 0.09 ppm in 1 s) and minimal cross-interferences from other present species in these complex flows. Although the chemometric fit was optimized for H2S,4 it also simultaneously quantified other species in these samples to reasonable precision – CO2 (±0.2%), CH4 (±150 ppm), C2H4 (±30 ppm), and H2O (±300 ppm) – value-added data that should enable better characterization of the fuel reforming process. Beyond precision and accuracy, a dynamic application such as this also needs fast instrument response time. The LGR analyzer’s time response is mainly limited, in this case, by the flow rate at which gas is transferred through the analyzer’s measurement cell (i.e., pump speed). For the gas cell (effective volume at operating pressure ~ 80 cm3) and pump (11 SLPM) used in these studies, the 1/e time response of the analyzer was determined to be 2.1 s by rapidly (<0.1 To demonstrate the proof-of-concept unit, we then s) switching the inlet from a 4.2-ppm H2S (in air) standard performed tests under real conditions using PCI’s 5 kW to dry “zero” air. In the in situ JP-8 fuel reformer studies autothermal reformer followed by a custom desulfurizing typified by figure 3, a smaller pump (5 SLPM) was employed bed. Typical measured data from these tests are shown to avoid disrupting the reformer gas flow, increasing in Figure 3. In this run, the measurement point was the instrument response time to 4 seconds. In other H2S repeatedly switched between ambient air, and upstream applications, where this disruption is not an issue, a large and downstream of the desulfurizer bed. As expected, the scroll vacuum pump (600 SLPM) could alternatively be used ambient air measurements had very low concentrations of all to decrease the response time to less than 0.1 seconds for reformate species (e.g., H2S <0.1 ppm) except water vapor (~ rapid industrial process control requirements, if desired. www.lgrinc.com Summary There are several optical methods that can measure gas References under controlled sampling conditions. Many real-world For a more detailed discussion of this study, see F. Dong, C Junaedi, S. Roychoudhury, and M.Gupta, Anal. Chem., 83 (11), 4132 (2011). applications, however, are just not that simple and involve 1. C. Song, Catal. Today, vol. 77, 7 (2002). flowing samples with multiple gases. The H2S monitoring application evaluated here includes numerous reformed 2. J.P. Trembly, J.P. Marquez, A.I. Ohrn, and D.J.J. Bayless, Power Sources, 158, 263 (2006). constituent species, not all of which are known, nor 3. J. Gong, Sens. Acuators, B, 114, 32 (2006). their spectra catalogued, a priori. Nonetheless, this study 4. L.D. Le, J.D. Tate, M.B. Seasholtz, M.B. Gupta, T. Owano, D.S. Baer, T. Knittel, A. Cowie, and J. Zhu, J. Appl. Spectrosc., 62, 59 (2008). species diluted in a pure host gas (e.g., N2) to ppb levels demonstrates that LGR’s OA-ICOS measurement strategy, which combines measurements of fully resolved absorption spectra and standard chemometric analyses, delivers trace gas measurements with the requisite precision, accuracy and response time needed for practical applications. Acknowledgement LGR gratefully acknowledges the Army SBIR Program for funding this work through grant no. W911QX-C-0034. Los Gatos Research, Inc. 67 East Evelyn Avenue, Suite 3 Mountain View, CA 94041-1529 Phone: +1 650–965–7772 Fax: +1 650–965–7074 Sales: [email protected] www.lgrinc.com
© Copyright 2026 Paperzz