Quantitative measurement of combustion gases in harsh environments using NDIR spectroscopy
Christian Niklas, Stephan Bauke, Fabian Müller, Kai Golibrzuch, Hainer Wackerbarth, Georgios Ctistis
aa r X i v : . [ phy s i c s . a pp - ph ] S e p Quantitative measurement of combustion gases in harshenvironments using NDIR spectroscopy
Christian Niklas a , Stephan Bauke a,b , Fabian Müller a , Kai Golibrzuch a,c,d , Hainer Wackerbarth a , andGeorgios Ctistis aa Laser-Laboratorium Göttingen e.V., Hans-Adolf-Krebs-Weg 1, 37077 Göttingen, Germany b IAV GmbH, Entwicklungszentrum Nordhoffstraße 5, 38518 Gifhorn Germany c Max-Planck-Institute for Biophysical Chemistry, Am Fassberg 11, 37077 Göttingen, Germany d Institute for Physical Chemistry, Georg-August-University Göttingen, Tammannstrasse 6, 37077 Göttingen,Germany
Correspondence:
Georgios Ctistis ([email protected])
Abstract.
The global climate change calls for a more environmental friendly use of energy and has led tostricter limits and regulations for the emissions of various greenhouse gases. Consequently, there is nowadaysan increasing need for the detection of exhaust and natural gases. This need leads to an ever-growing marketfor gas sensors, which, at the moment, is dominated by chemical sensors. Yet, the increasing demands to alsomeasure under harsh environmental conditions pave the way for non-invasive measurements and thus to opticaldetection techniques. Here, we present the development of a non-dispersive infrared absorption spectroscopy(NDIR) method for application to optical detection systems operating under harsh environments.
Copyright statement.
TEXT
In today’s world, climate change is one of the most de-manding problems for our modern society with significanttechnological challenges in various areas. One of the mainspecies contributing to global warming, carbon dioxide(CO ), has increased from a level of 310 ppm in the year1972 to 410 ppm today (German Environment Agency,2017). Therefore, the control and limitation of CO isgaining importance, resulting in the need for gas detectorsthat are able to measure corresponding concentrations withsufficient precision at the location of emissions. Further-more, besides environmental control, safety as well asprocess and quality control are also important applicationsfor gas sensors. A well known safety issue is the controlof combustion gases of civil fireplaces. Here, carbonmonoxide (CO) is one of the most relevant gas speciesbesides CO . The odorless CO is extremely toxic due toits chemical property to bind strongly to hemoglobin andtherefore inhibiting oxygen transport (Ernst and Zibrak,1998). Rooms with concentrations above 30 ppm are deemed hazardous for a person’s health for a longer expo-sure (Federal Institute for Occupational Safety and Health,2018).Another important area for CO detection is exhaust emis-sion control in the automobile and transport sector. Here,forthcoming new limits for CO and NO x emission set by theEuropean Union (European Parliament and Council, 2009)force the development of more efficient and cleaner engines.Both are usually achieved by controlling and optimizing thecombustion process, i.e. mixture formation prior to ignition(Grosch et al. (2014), Bauke et al. (2018), Golibrzuch et al.(2017)). A typical approach to reduce NO x emissions is theuse of exhaust gas recirculation to lower combustion tem-perature. EGR-rates can be determined by monitoring CO concentrations (Grosch et al., 2014).In contrast to sensors for civil applications requiring low-cost solutions, systems for engine development applicationare less price-sensitive but have, on the other hand, moredemanding requirements. For example, they need to offerµs time resolution to enable crank-angle resolution and re-solve single engine cycles (Grosch et al. (2014), Bauke et al.(2018)). Christian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments
To date electrochemical and resistive sensors dominate themarket for gas sensors (for Sensor Technology, 2014). Elec-trochemical sensors use two or three electrodes and reduceor oxidize the target gas and measure the resulting electri-cal current allowing a cheap detection method (Stetter et al.,2003). Nonetheless, these sensor types face various prob-lems, e.g. limited durability due to the electrolyte or suscep-tibility for different gases (Ricco et al., 1997). Here, hydro-gen sulfides can influence the measurement of CO , which isespecially dangerous for sewer measurements. Furthermore,electrochemical sensors cannot be used in the environmentof an internal combustion (IC) engine, as it is prone to theharsh environment and not capable of a high time resolutionneeded to analyze the mixture process of the fuel.Where electrochemical sensors face usage limitations, ap-plication of optical sensors is often advantageous. In com-bustion diagnostics, a frequent approach is the use of laser-induced fluorescence (LIF) for measurements of tempera-ture or fuel-concentrations with high spatial resolution. How-ever, LIF measurements require sufficient optical access,do not allow real-time resolution, and measurements aretime consuming due to the complex experimental set-up(Schulz and Sick (2005); Luong et al. (2008)). Most impor-tant, most gas-phase molecules cannot be excited to appro-priate electronic states, so measurements rely on the use offluorescent markers (tracer) that represent the species of in-terest.Instead, non-dispersive infrared (NDIR) spectroscopy canbe utilized for both civil fireplaces as well as IC engines,where gas specific infrared absorption spectra, present in al-most any molecule, are used to determine the density of agas.In this work, we present the development of two sensorsbased on NDIR spectroscopy in harsh environments: (1) alow-cost sensor for civil fireplaces and (2) a high-speed sen-sor for determination EGR-rates in IC engines. Thereby, thesensors face following difficulties: a simultaneous measure-ment of two gases at different concentrations and simultane-ous measurement of overlapping absorption of the analytes.Prior to the presentation of details on the respective sen-sor systems, we introduce the basic principle of NDIR spec-troscopy as well as the relevant spectroscopic properties ofCO , CO, and H O.The first sensor is intended to be used in civil fireplaces.Here, the difficulties arise from the simultaneous detection oftwo different gas species, CO and CO, respectively, whichare present at largely different concentrations. The intendedsensor setup and its optical components are described and ex-plained. Furthermore, exemplary measurements of the setuptaken at atmospheric conditions are shown.The second sensor is intended for monitoring EGR ratesin IC engines. Here, CO , being the major combustion prod-uct, is the target gas. For each sensor, we describe the field ofapplication and give a brief overview of the setup and its opti-cal components. Furthermore, a data analysis strategy is pre- sented. An exemplary measurement at a test engine is shownand compared to known properties to validate the data anal-ysis. The article concludes with a summary and outlook. The concentration of the greenhouse gases CO and CO in combustion processes can be measured by means ofnon-dispersive infrared (NDIR) absorption spectroscopy.Thereby, infrared radiation is absorbed by the gas moleculesas described by the Beer–Lambert–Bouguer law. The mea-sured radiation intensity is then given by: I ( ν ) = I ( ν ) e − α ( ν ) · L , (1)where I is the radiation intensity of the source, i.e., withoutgas in the absorption path, α is the absorption coefficient ofthe molecules, ν the light frequency, and L the absorptionpath length. Integration over a frequency interval leads to: τ = II = ν max Z ν min e − σ ( ν,p,T ) · ρ ( p,T ) · L d ν. (2)Here, σ ( ν, p, T ) is the frequency, pressure, and tempera-ture dependent absorption cross section and ρ the density ofthe specific gas. To describe real absorption measurements,Eq. (2) needs to accommodate the systems’ transfer function,i.e., the systems’ transmission: τ sys ( ν ) = τ filter ( ν ) S detector ( ν ) I LS ( ν ) , (3)with τ filter ( ν ) the transmission spectrum of the filter, S detector the sensitivity of the detector, and I LS ( ν ) the spec-tral intensity of the light source. These are the most com-mon optical components, and this equation can be furtherexpanded to include other used components such as opticalfibers. Combining Eqns. 2 and 3, the normalized transmis-sion τ in NDIR is then given by: τ = R e − σ ( ν,p,T ) · ρ ( p,T ) · L · τ sys ( ν ) d ν R τ sys ( ν ) d ν . (4) The absorption spectra of the desired gases and their interfer-ence with other specimen found in combustion processes arefirst calculated using the HITRAN database (Gordon et al.,2017) and Eq. (2). The results set limits to the wavelengthas well as detection range for each gas species and for eachsetup. The absorption spectra for the most relevant con-stituents are shown in Fig. 1. The concentrations of the car-bon oxides used for the calculations are the upper limits al-lowed in civil fireplaces (VDI, 2010). H O concentration ischosen large enough to see any overlaps; concentrations ofwater vapor in combustion processes can vary extremely with hristian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments 3
Wavenumber [cm -1 ] T r an s m i ss i on
2% CO10% H O20% CO Wavenumber [cm -1 ] T r an s m i ss i on
2% CO10% H O20% CO (a)(b) Figure 1. (a)
Spectra of exhaust gases common in civilian fire-places in the range 2000-2500 cm − at T = 300 K and p = 1 barwith an absorption length L = 6 cm. Both CO and CO have strongabsorption bands in this part of the infrared spectral region. (b) Thespectral response of CO, CO , and H O in the near-infrared (NIR)from 3000-4500 cm − . The absorption of CO is too low to be usedin NDIR spectroscopy, but CO and H O both have strong but over-lapping absorption bands. operation conditions. For the first sensor, intended to be usedin civilian fireplaces, the spectral range between 2000 and2500 cm − is chosen, which lies in the mid-infrared (MIR)spectral region, as shown in Fig. 1(a). Here, the rotational-vibrational absorption is very strong, i.e., the absorption co-efficients for both CO and CO are very large. The domi-nant excited vibration for CO is the stretch vibration and forCO the ν antisymmetric stretch vibration (Gerakines et al.,1994). Furthermore, there is little to no interference withother gases. Figure 1(b) shows the high-frequency spectralregion (NIR) of the absorption spectrum, which is the spec-tral region chosen for the second sensor. As is shown, thereis a strong overlap between CO and H O, which has to beaccounted for in the data analysis.
This sensor is a combined sensor able to simultaneouslydetect CO and CO and intended to be used by chimneysweepers. Therefore, it has to be durable, cost-effective, low-maintenance, easy-to-use, compact, and uphold measure-ment regularities. These regularities demand the detection ofthe two carbon oxides in different detection ranges; whereasCO needs to be detected in the range up to 2 vol % and CO up to 20 vol % .In our setup, we first determined the absorption lengthfor both carbon oxides in order to design a compact sensorfor both gas components. For this purpose, we simulate theresponse using the HITRAN data base computing the inte-grated transmission in Eq. (4). The transfer function of thesensor ( τ sys ) is illustrated in Fig. 2 (a), where the area be-neath τ sys is the integrated transmission signal. In fact, therecorded signal is very low, which is here a direct result of theemission characteristics of the light source (black body radia-tion at K) and the sensitivity of the detector in this spec-tral region. Exemplary, integrated transmissions for differentconcentrations of CO and CO are calculated for an absorp-tion length of cm and shown in Fig. 2 (b). From the derivedcurves one can determine that CO has a much weaker ab-sorption coefficient in comparison to CO and thus a higherabsorption length (factor of 10) is required for a sufficientsensitivity. Furthermore, a large dynamic range of the CO sensor extends only up to concentrations of about vol % ,being equivalent to a small absorption length for CO .A design for a combined sensor for both carbon oxideshas to take the aforementioned differences into account.The main difficulty lies thereby in the combination of thedifferent measurement ranges where the sensor shouldshow high dynamic response. For typical sensor appli-cations in civil environment CO concentrations rangein the vol% (0.2-20 vol % ), while CO concentrationsat the same time lie in the ppm-regime (0-200 ppm),as the latter is highly toxic and 30 ppm is the sug-gested upper limit for working conditions by the BAuA(Federal Institute for Occupational Safety and Health,2018).A sketch of a setup for each gas is depicted in Fig. 3. Alarger absorption length for CO may be realized by meansof a multi-pass cavity, so the sensor dimensions still remaincompact. As light source we chose a bulb with a tungstenfilament emitting black-body radiation at K. For use asa detector there are two reasonable options: a photo-resistorand a pyroelectric detector. The advantage of the latter is itslow price, while the former exhibits the better signal-to-noiseratio. Here, a PbSe photo-resistor has been used. The circuitto achieve a measurable signal is a Wheatstone bridge withan amplifier circuit. Exemplary, measurements for both gasesare shown in Fig. 4. In panel (a), the CO sensor shows a sat- Christian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments
Concentration [%] I n t eg r a t ed T r an s m i ss i on COCO Wavenumber [cm -1 ] T r an s m i ss i on FilterSourceDetector sys
2% CO (a)(b)
Figure 2. (a)
Spectral influences on the measured signal. The areabeneath τ sys is the integrated transmission which is equivalent tothe measurement signal. (b) Calculated integrated transmission forCO and CO for an absorption length of 5 cm. The difference in thecurves shows the need of two different absorption path lengths inthe setup to acquire a desirable resolution. Figure 3.
Schematic setup for detection of individual carbon oxidesin civilian fireplaces.
Concentration [vol%] S i gna l [ m V ] COLinear regression0 5 10 15 20
Concentration [vol%] S i gna l [ m V ] CO Regression (a)(b)
Figure 4.
Measurements of (a) CO and (b) CO at atmosphericconditions (1 bar, 300 K) and cm absorption path. Included are theregression functions as dotted lines. The measurement is such thatdecreasing intensity gives a positive signal change. uration behavior. This is due to the saturation of the CO transmission filter for its central wavenumbers, as shown inFig. 1(a), so at higher concentrations only the shoulders ofthe filter spectrum contribute to the signal change. The re-gression has the form: f ( x ) = c − b · e − m · x . (5)The sensitivity s of the sensor can be calculated from thederivative of the regression function: s = ∂f ( x ) ∂x = mb · e − m · x . (6)To give an overview of the sensor attributes the weightedaverage of the sensitivity is calculated to s = 269 . ± . mVvol % − . With a digital resolution R digit of 1 mV the hristian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments 5 resolution ∆ r can be calculated as ∆ r = R digit s , (7)which leads to an overall CO resolution of ∆ r = 60 ± ppm.Hazardous workplace environment is defined above 0.5 vol % (Federal Institute for Occupational Safety and Health,2018), so the resolution is usable for civil fireplaces.In Fig. 4(b), the CO sensor shows an approximately lineardependence as expected from Fig. 2 (b). The fluctuations vis-ible in the measurement are a result of the cooling routineof the sensor. The sensitivity is s = 7 . ± . µVppm − ,which results in a resolution of ∆ r = 139 . ± . ppm. Asalready mentioned, a CO concentration over 30 ppm is haz-ardous to be exposed to for a longer time. Therefore, the sen-sitivity of CO needs to be enhanced to achieve a higher res-olution. This can be accomplished on the one hand by theuse of better detectors, which would result in more expen-sive sensors and on the other hand, a larger absorption pathlength. The development of the latter, i.e., a multi-pass con-figuration for the CO absorption measurements is a task ofcurrent research. While the sensor described above represents a cost-sensitivedown market application, NDIR spectroscopy might also beused in more demanding environments in research and devel-opment. In the following section, we present a second sensorcapable of quantification of residual gas concentrations in in-ternal combustion (IC) engines. In contrast to the fireplaceexhaust sensor described above, an IC engine requires mea-surements under highly dynamic conditions with pressuresand temperatures ranging from − bar and − K,respectively. Moreover, the sensor needs to offer a high tem-poral resolution (at least <1 ms). Since the field of applica-tion is less cost-sensitive, the requirements can be met by theuse of high-end components. Moreover, an accurate quan-tification over the wide range of conditions requires a moresophisticated data analysis that uses the well-known spectro-scopic properties of the molecules.The sensor system used in this work is a modification ofthe Internal Combustion Optical Sensor (ICOS) for LaVi-sion GmbH and has been described extensively elsewhere(Grosch et al. (2014); Golibrzuch et al. (2017); Bauke et al.(2018)). The schematic layout of the system is shown inFig. 5(a). Briefly, the system consists of a broad-band lightsource (150 W quartz-tungsten-halide (QTH) lamp), a sparkplug sensor probe and a detection unit. The light from theQTH lamp is modulated by a 30 kHz chopper wheel, whichdetermines the maximum time resolution to about µs, andcoupled into a ZrF mid-infrared fiber. The time resolutionis sufficient to enable crank-angle resolution of single cy-cles in IC engines up to 5000 rpm. The ZrF fibers guidethe light to the spark plug probe and back to the detection W av e(cid:0)(cid:1)(cid:2) ber [cm -(cid:3) ] T r an s m i ss i on (a)(b) O4% CO H O(cid:6)(cid:7)(cid:8) R(cid:9)(cid:10)(cid:11)(cid:12)(cid:13)(cid:14)(cid:15)(cid:16)F(cid:17)(cid:18)(cid:19) data aquisitionplane mirrors lensfilterlens QTH-lampchopperlenscomputeroptical fibersoptical measurement setupfuelH2O
CO2+H2O reference H OFilterSpectra
Figure 5. (a)
Schematic of the ICOS sensor system consisting ofdifferent measurement channels. (b)
Transmission spectra of thebandpass filters used in the ICOS sensor system as well as the spec-tra for H O and CO . Note that the serial assembly causes the band-pass filter influencing each other. unit. Inside the probe, sapphire fibers guide the light to thedetection volume where it is reflected back by a concavemirror located in a stainless steel cage holder. The absorp-tion path is 0.96 cm. The sapphire fibers are necessary towithstand the high temperature during fired engine operation,but limit detection to wavelengths λ < . µm. Inside the de-tection unit the light passes a cascaded array of Mercury-Cadmium-Telluride (MCT) detectors equipped with differ-ent bandpass filters. An overview of bandpass filters usedin this work is shown in Fig. 5(b). The system consists ofa filter for fuel concentration measurements, utilizing theabsorption of C-H stretch vibrations of hydrogen carbonsaround 3100 cm − as well as two detection channels forwater and CO and an “offline” reference filter situated ina spectral range with negligible absorption of any presentmolecular species. The reference filter serves as a correc-tion for signal disturbances due to, e. g., beam steering orparticles in the beam path. Note that the effective transmit-tance curves differ from the raw ones due to the serial as-sembly of the filters. Fuel concentration measurements arebeyond the scope of this work but have been demonstratedfor gasoline (Grosch et al. (2007); Grosch et al. (2010); Christian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments
Grosch et al. (2011)) as well as methane-fueled engines (Bauke et al. (2017); Golibrzuch et al. (2017); Bauke et al.(2018); Kranz et al. (2018)) and we will focus on the quan-tification of residual gas, i.e. CO and H O.Since the extreme conditions in an IC engine require theuse of sapphire fiber to guide the light into the combustionchamber, CO detection at 2400 cm − is impossible. There-fore, the detection is limited to the weaker absorption bandaround 3700 cm − which, however, is completely blended bywater absorption. Consequently, a strategy is required to cor-rect the influence of water vapor. As visible from the spectrashown in Fig. 5(b), the H O absorption covers a much largerspectral range than CO . Therefore, a second filter that isonly sensitive to water is used to determine the H O amountindependently. Additional complexity arises, since dynamicchanges in pressure and especially temperature need to beaccounted for.
In order to enable quantification in the large range of exper-imental conditions, the data analysis procedure relies on thecomplete description of the molecule’s spectroscopic prop-erties, making use of the HITRAN database (Gordon et al.,2017), as well as the spectral influences of the optical system(see Eq. (4)). This procedure has been recently described inmore detail in (Golibrzuch et al. (2017); Bauke et al. (2018))for methane and might be applied for CO and H O accord-ingly.Briefly, we use the spectroscopic constants from HITRANto calculate the transmission of the H O+CO and the H Odetection channels for different CO and H O concentrationsas a function of temperature and pressure ranging from − K and . − bar, respectively. The computed data arethen used to build a 3-dimensional look-up-table that linksthe measured transmission to the corresponding molecules’density for different pressures and temperatures.While, temperature and pressure information for the civilfireplace sensor is easily accessible, the IC sensor faces sig-nificant problems, especially regarding the temperature in themeasurement volume. Time-resolved pressure measurementsare usually available at engine test stations, but tempera-ture is usually unknown due to the comparable low speedof standard probes. Moreover, the temperature in the mea-surement volume, surrounded by a metal cage, can differstrongly from the temperature commonly calculated by ther-modynamic models (Golibrzuch et al. (2017); Kranz et al.(2018)). Therefore, temperatures need to be estimated frommodified thermodynamic models or measured by other spec-troscopic techniques (Luong et al. (2008); Werblinski et al.(2017)); the latter might however require the use of an ad-ditional probe. The possibility to determine temperatures us-ing NDIR probing different spectral regions of an absorp-tion band simultaneously has thereby been demonstrated re-cently for the case of methane (Golibrzuch et al. (2017); Figure 6.
Schematic overview of the data analysis procedure fordetermination of CO concentrations in IC engines by NDIR. Bauke et al. (2018)). In case of CO /H O, this would how-ever require the use of third filter, which would further raisethe complexity of the system and the data analysis proce-dure. Given that all information for quantification (transmis-sion, pressure, and temperature) are available, the remainingchallenge is to disentangle absorption due to CO and H O,respectively.The sensor system offers transmission information in aspectral region with only H O absorption lines as well as fora region with combined CO and H O absorption. In a firstapproximation, we assume that the absorption in the overlap-ping region can be described as the product of transmissioncaused by H O and CO : τ H O + CO = τ CO ( T, p, ρ CO ) × τ H O ( T, p, ρ H O ) . (8)It is important to note here, that τ H O + CO , τ CO and τ H O are broadband transmittance values. While, Eq. (8) would becompletely valid for frequency dependent transmittance, it isonly an approximation for direct multiplication of integratedbroadband values.Since the water density can be determined from the pureH O signal in the second detection channel, τ H O can be cal-culated and the CO density remains the only unknown vari-able to be determined. Figure 6 shows a schematic overviewof the data analysis procedure applied in this work. Note thatthe temperature information effects different points of theprocedure.Another important issue, is the determination of I , i.e.the signal without absorption species in the beam path. Incontrast to fuel concentration measurements, where I canbe determined before fuel enters the combustion chamber(Golibrzuch et al. (2017); Kranz et al. (2018)), water andCO are always present in ambient air as well as exhaustgas. Therefore, the detector signal received at the lowest gas hristian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments 7 density always contains some absorption. In order to elimi-nate this effect, we developed a method for I determinationby extrapolation to p = 0 . In an IC engine, temperature andpressure usually follow an polytropic compression: T = pp · ( TT ) n − n (9)where p and T are the pressure and temperature prior tocompression and n is the polytropic coefficient. Assumingthat Beer–Lambert–Bouguer law (Eq. (2)) is also approxi-mately valid for integrated transmission using an ’integratedabsorption cross-section’, σ : I int ≈ I ,int e − σ ( p,T ) · ρ ( p,T ) · L (10)Eq. (10) can be linearized to: ln ( I ) ≈ ln ( I ) − σ ( p, T ) · ρ ( p, T ) · L (11)For a polytropic compression the gas density can be ex-pressed using ideal gas law as: ρ ( p, T ) = pR · T (12)and with Eq. (9): ρ ( p, T ) = p n − n R · T · p − n − n = const. · p n (13)Consequently, I can be extracted as the intercept of a lin-ear fit to ln ( I ) as a function of p n . An accurate I determi-nation requires that the approximations of Eq. (11) and theintegrated absorption cross-section σ being independent ofpressure and temperature are valid. Fig. 7 shows HITRANsimulation of the signals in the H O+CO and the H O de-tection signals for a typical polytropic compression. The datashows that the approximations made above hold reasonablywell for most signals and can be judged from the intercept ofthe linear fits being close to 0 ( I = 1 ). The best results areexpected if only CO is present since the respective band-pass filter covers the complete CO absorption band. In caseof H O, the sensor sees only a part of the absorption and tem-perature effects due to redistribution of rotational states aremore relevant. and H O in an IC engine
We test the system and data analysis procedure in a methane-fueled IC engine under motored (pure air) and fired opera-tion under stoichiometric conditions. Details of the engineused in this work are given in Kranz et al. (2018). Figure 8gives an overview of the respective results averaged over100 engine cycles. Panel (a) shows the measured transmis-sion signals under motored (dashed lines) and fired (solidlines) conditions for both channels, CO +H O (black) and "(cid:20)(cid:21) +(cid:22) O" channel (cid:23)(cid:24)(cid:25) (cid:26)(cid:27) O" channel (cid:28)(cid:29)(cid:30)(cid:31) ! )$%& ’( O" channel *,./0 H O H O channel p /n [bar H chann ;<= tot a>? @AB (o nCD EG IJ KL (o MNP Q SUV
XYZ[ H O channe l\ ^_‘b cdfg -0.20 h i jk mopqrst r an s m i ss i on H i t r an u Figure 7.
HITRAN simulation of the logarithmic transmission asa function of p n for H O and CO absorption in the respectivedetection channels for a typical polytropic compression in an ICengine ( T = 300 K , p = 1 bar , n = 1 . ). H O (red), respectively. Panel (b) shows corresponding tem-perature (black) and pressure (blue) data. Note that the tem-perature data were obtained in a separate measurement usingspectrally resolved water absorption measurement from anICOS-Temperature system (LaVision GmbH) (Bauke et al.(2018); see Werblinski et al. (2016); Werblinski et al. (2017)for operation principle). Panel (c) and (d) show the resultsfor CO and H O concentrations obtained using the dataanalysis described above. Under motored engine operation,we determine a CO concentration of about 0.05 % , whichis in good agreement with ambient CO concentration of0.04 % , but close the detection limit of the system. The wa-ter concentration is determined to approx. 1 % correspond-ing to about 40 % humidity at 293 K (Wexler, 1976). Con-sequently, the motored data indicates that the analysis al-gorithm and modeling yields results over a wide range ofpressure and temperature, which are consistent with typicalambient conditions. However, more detailed validation ex-periments in e.g. static pressure cells or a rapid compres-sion engine are needed in order to evaluate achievable ac-curacy and precision. Nevertheless, the promising motoredresults enable a first evaluation of the system and fired en-gine operation conditions (solid lines in Fig. 8). The IC en-gine was operated with methane port-fuel-injection underglobal stoichiometric conditions. At the beginning of theengine cycle at − ◦ CA, we obtain CO and H O con-centrations of about 9 % and 7 % , respectively. After open-ing of the intake valves at -334°CA, a mixture of ambientair and methane enters the combustion chamber, leading toa strong decrease in the exhaust gas concentrations. Dur-ing compression from − ◦ CA to − ◦ CA, air, fuel, andresidual exhaust gas undergo a process of mixture forma-
Christian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments T r an s m i ss i on T e m pe r a vwxyz{| [ m o }~(cid:127)(cid:128)(cid:129)(cid:130)(cid:131)(cid:132)(cid:133)(cid:134)(cid:135) [ m o (cid:136)(cid:137)(cid:138)(cid:139)(cid:140)(cid:141)(cid:142)(cid:143)(cid:144)(cid:145)(cid:146)(cid:147)(cid:148) H O (cid:149)(cid:150)(cid:151)(cid:152)(cid:153)(cid:154)(cid:155)(cid:156)(cid:157)(cid:158)(cid:159)(cid:160)¡¢£⁄¥ƒ§ ¤'“«‹ ›fifl(cid:176)– motoredfiredmotoredfiredT motored † fired ‡ motored T fired Figure 8.
Measurements of CO and H O concentrations in amethane-fueled IC engine. Dashed lines: motored operation withlaboratory air. Solid lines: fired operation with methane port-fuel-injection at a global λ = 1 (stoichiometric combustion). The grayarea indicates time of combustion which is excluded from the datainterpretation due to partial saturation of the detectors from theflame emissions. tion resulting in final CO and H O concentrations of 0.4 % and 1 % , respectively. After ignition at − ◦ CA, the trans-mission signals exhibit a steep decrease due to flame emis-sion partially saturating the detectors (gray area). The re-gion is therefore excluded from the data analysis. After com-bustion, the temperature remains at about 800 K with CO and H O concentrations of about 10 % and 8 % . From thisdata, we can estimate the EGR rate to be approximately .The EGR rate might also be estimated from pressure, tem-perature and volume using ideal gas law. Before opening ofthe intake valve at -334°CA ( p IVO = 0 . bar, T IVO = 600 K, V IVO = 66 . ccm), a remaining gas amount of n IVO = 1 . × − mole can be estimated. After intake valve closing at-184°CA ( p IVC = 0 . bar, T IVC = 335 K, V IVC = 594 ccm),the total gas amount raised to n IVC = 1 . × − mole. Com-parison n IVO to n IVC gives an EGR rate of , consistentwith the sensor data.
We presented here two gas sensors based on non-dispersiveinfrared spectroscopy for high and low tech. We outlinedthe development of these sensors, one intended for civil fire-places with the ability to detect CO as well as CO and theother for IC engines, capable of CO and water vapor de-tection. Potential spectral regions for the detection of CO were identified between 3400 and 4000 cm − and 2200 to2400 cm − , whereas the former strongly overlaps with waterabsorption bands. This disadvantage leads to the necessityof a water absorption channel and additional calculations toseparate CO and H O.Additionally, at the lower frequency region from 2200 to2400 cm − CO has absorption bands, which only have a neg-ligible overlap with the CO absorption bands. This enablesthe simultaneous determination of the concentrations in asingle sensor, which is suitable for civil fireplaces based onnon-dispersive infrared spectroscopy. The absorption behav-ior of CO and CO are compared and their optimal absorp-tion lengths were discussed, whereas CO needs a long andCO needs a short absorption length. Furthermore, we dis-cussed possible optical components for the sensor. The finalmain components are a thermal broadband emitter, opticalfilter, and a PbSe photo resistor. The presented sensor is ca-pable of a resolution of 60 ppm for CO and 140 ppm forCO at an absorption length of L = 5 cm. Due to the usageof a PbSe detector, temperature has a tremendous influenceon the sensor, which can especially be observed during theCO measurement. Here, the cooling routine of the sensor isvisible in oscillations of the measurement points in time. Tominimize the influence of temperature and get rid of thermaloscillations, an improved cooling routine has to be imple-mented. Furthermore, it is advisable to use another materialas a detector, as lead (Pb) may be further regulated by the Eu-ropean Union. Here, InAsSb or pyroelectric sensors may beutilized. It is also possible to use a different spectral filter forthe detection of CO , as for higher concentrations oversatu-ration is encountered. It can be advisable to use a CO sensoron the flanks of the CO absorption, so a linear measurementmight be possible.The spectral region between 3000 and 4500 cm − and itsCO absorption is utilized for a sensor intended for IC en-gines due to limitations for mid-infrared fiber guides. To ad-dress the overlap of H O and CO in this spectral region, thedetector consists of multiple detection channels built like acascade to achieve a single detection channel of H O and acompound channel of the mixture of H O and CO . A calcu-lation routine utilizing look-up tables is presented to achievea single water and CO signal. This sensor is capable of hightime resolutions up to 33 µs and faces huge challenges due toa harsh and highly dynamic (temperature and pressure) envi-ronment. We demonstrated an application of the system to anIC engine under motored and fired operation. The results un-der motored operation were consistent with typical ambient hristian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments 9 condition. For fired conditions, these data could be used tocalculate the EGR rate which was in agreement with thermo-dynamic estimations. Nevertheless, further validation of thesensor and data analysis under more controlled conditions(e.g. in static pressure cells or a rapid compression engine) isrequired to determine its accuracy and precision over a widerange of temperatures and pressures. Data availability.
The experiments and results shown in this pub-lication are strongly industry-related research. We explain our ex-perimental preparations and analysis steps in great detail in thiswork and are available for questions. Please understand that there-fore we do not publicly provide the underlying data and Matlab codeused for analysis. Given individual requests by fellow researchers,we will of course consider making parts of the data available.
Author contributions.
All authors contributed equally to the ex-periments and to the manuscript preparation.
Competing interests.
The authors declare that they have no con-flict of interest.
Acknowledgements.
The authors gratefully acknowledge finan-cial support through the Federal Ministry of Education and Re-search (BMBF, Germany) FKZ: 13N13035 and the Federal Min-istry for Economic Affairs and Energy (BMWi, Germany) FKZ:ZF4060502WM6.
References
Bauke, S., Golibrzuch, K., Rotter, F., Wackerbarth, H., Thiele, O.,and Berg, T.: Quantitative, time-resolved detection of CH con-centrations in flows for injection analysis in CNG engines usingIR absorption, Journal of Sensors and Sensor Systems, 6, 185–198, https://doi.org/10.5194/jsss-6-185-2017, 2017.Bauke, S., Golibrzuch, K., Wackerbarth, H., Fendt, P., Zigan, L.,Seefeldt, S., Thiele, O., and Berg, T.: Optical sensor system fortime-resolved quantification of methane concentrations: Valida-tion measurements in a rapid compression machine, Journal ofQuantitative Spectroscopy and Radiative Transfer, 210, 101–110,https://doi.org/10.1016/j.jqsrt.2018.02.016, 2018.Ernst, A. and Zibrak, J. D.: Carbon monoxide poisoning, New Eng-land journal of medicine, 339, 1603–1608, 1998.European Parliament and Council: Regulation No 443/2009,setting emission performance standards for new pas-senger cars as part of the Community’s integrated ap-proach to reduce CO2 emissions from light-duty vehicles,http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:02009R0443-20140408&from=EN,2009.Federal Institute for Occupational Safety and Health: TRGS 900Arbeitsplatzgrenzwerte, 2018.for Sensor Technology, A. A.: Trends in future-oriented sensor tech-nologies, AMA Association for Sensor Technology: Berlin, Ger-many, 2014. Gerakines, P. A., Schutte, W., Greenberg, J., and van Dishoeck,E. F.: The infrared band strengths of H O, CO and CO0 Christian Niklas et. al.: Non-dispersive IR-spectroscopy in harsh environments