sensors Article
Radio-Frequency-Based NH3-Selective Catalytic Reduction Catalyst Control: Studies on Temperature Dependency and Humidity Influences Markus Dietrich 1,2 , Gunter Hagen 1 , Willibald Reitmeier 2 , Katharina Burger 2 , Markus Hien 2 , Philippe Grass 2 , David Kubinski 3 , Jaco Visser 3 and Ralf Moos 1, * 1
2
3
*
Bayreuth Engine Research Center (BERC), Department of Functional Materials, University of Bayreuth, 95447 Bayreuth, Germany;
[email protected] (M.D.);
[email protected] (G.H.) Continental Automotive GmbH, Division Powertrain, Siemensstraße 12, 93055 Regensburg, Germany;
[email protected] (W.R.);
[email protected] (K.B.);
[email protected] (M.H.);
[email protected] (P.G.) Ford Research and Innovation Center, 2101 Village Rd., Dearborn, MI 48124, USA;
[email protected] (D.K.);
[email protected] (J.V.) Correspondence:
[email protected]; Tel.: +49-921-55-7400
Received: 21 May 2017; Accepted: 28 June 2017; Published: 12 July 2017
Abstract: The upcoming more stringent automotive emission legislations and current developments have promoted new technologies for more precise and reliable catalyst control. For this purpose, radio-frequency-based (RF) catalyst state determination offers the only approach for directly measuring the NH3 loading on selective catalytic reduction (SCR) catalysts and the state of other catalysts and filter systems. Recently, the ability of this technique to directly control the urea dosing on a current NH3 storing zeolite catalyst has been demonstrated on an engine dynamometer for the first time and this paper continues that work. Therefore, a well-known serial-type and zeolite-based SCR catalyst (Cu-SSZ-13) was investigated under deliberately chosen high space velocities. At first, the full functionality of the RF system with Cu-SSZ-13 as sample was tested successfully. By direct RF-based NH3 storage control, the influence of the storage degree on the catalyst performance, i.e., on NOx conversion and NH3 slip, was investigated in a temperature range between 250 and 400 ◦ C. For each operation point, an ideal and a critical NH3 storage degree was found and analyzed in the whole temperature range. Based on the data of all experimental runs, temperature dependent calibration functions were developed as a basis for upcoming tests under transient conditions. Additionally, the influence of exhaust humidity was observed with special focus on cold start water and its effects to the RF signals. Keywords: radio-frequency (RF); NH3 SCR; NH3 storage; direct control; microwave cavity perturbation; exhaust gas sensor; cold start
1. Introduction Continuously tightening vehicle emission legislations are the main driving factor for improvements in engine and exhaust gas aftertreatment technologies among automotive manufacturers worldwide [1]. Especially diesel engine driven vehicles and their higher emission of nitric oxides (NOx = NO + NO2 ) are in focus of research, development and the media [2,3]. The selective catalytic reduction (SCR) using ammonia (NH3 ) as reducing agent is today’s main deNOx technology for light and heavy duty diesel engines. In this technology, an aqueous solution of 32.5 wt % urea in water (diesel exhaust fluid = DEF, AdBlueTM or AUS32 = aqueous urea solution) is injected into the exhaust Sensors 2017, 17, 1615; doi:10.3390/s17071615
www.mdpi.com/journal/sensors
Sensors 2017, 17, 1615
2 of 15
and decomposes after water evaporation by thermolysis and hydrolysis into gaseous NH3 and carbon dioxide (CO2 ). The formed NH3 adsorbs on the active sites of the SCR catalyst and can react with NOx to form nitrogen (N2 ) and water (H2 O) [2,4]. Relying on current schemes of the SCR reaction mechanism for several SCR catalysts, the prior NH3 adsorption is an essential precondition for all SCR reactions [5–8]. Depending on the NO/NO2 ratio, different SCR reactions occur on the catalyst surface. The two main reactions are the standard SCR reaction (Equation (1)) only with NO and oxygen (O2 ), and the fast SCR reaction with equimolar amounts of NO and NO2 without participation of O2 (Equation (2)) [2,4]: 4NH3 + 4NO + O2 → 4N2 + 6H2 O (1) 4NH3 + 2NO + 2NO2 → 4N2 + 6H2 O
(2)
Beside its necessity for the SCR reactions, the prior NH3 adsorption and storage on the catalyst is also beneficial in application to changing concentration and flow conditions in the buffer related to transient driving. Due to the kinetic limitations of the SCR reactions, a sufficient NH3 surface coverage is also required to achieve good NOx conversion efficiencies [5,9]. Additionally, it is necessary to avoid too high storage degrees, since this may lead to NH3 slip. Consequently, the catalyst control is required to secure an NH3 storage degree always between the minimum storage for high conversion and the maximum storage without NH3 slip to meet the current emission limits [10,11]. Therefore, the development also aims for SCR catalyst materials with high NH3 storage capacity and high low-temperature activity, such as copper (Cu) exchanged zeolites [12–15]. The current DEF dosing control is completely model-based and relies on gas sensor signals, i.e., from NOx and/or NH3 sensors [10]. In these approaches, the whole ad- and desorption equilibrium and all reactions occurring on the catalyst surface are simulated and the necessary amount of DEF is calculated [11,16,17]. This also requires injector self-diagnosis and urea concentration monitoring to secure the functionality of the whole SCR system. Small errors and deviations of only one part of this system may lead to incorrect urea dosing followed by NOx or NH3 emissions [18,19]. A measurement system to determine the current NH3 loading on the catalyst for model validation or direct dosing control on the road is not yet available. The radio-frequency-based (RF) catalyst and filter state determination technique has been a focus of research and development for several years. Since it operates in the range between 1 and 3 GHz, it sometimes also denoted as microwave-based state determination. With the RF technique, a contactless and direct (in-operando) measure of the catalyst/filter states, by using the metal catalyst canning as an electrical cavity resonator was presented [20–22]. At first, the oxidation state of three-way catalytic converters (TWC) was determined, indicating that the RF approach is capable of providing more precise information about the catalyst state and its optimal operation point compared to established gas sensorand model-based procedures [23–25]. The state of diesel or gasoline particulate filters (DPF or GPF) was also successfully monitored with the RF signal as a measure of accumulated soot [26–30]. First approaches to separate the signals of soot and ash appear promising [31]. Studies with a combined system of a TWC-coated GPF on an engine dynamometer proved the system functionality under transient conditions within the European driving cycle (NEDC) [30]. In NOx reduction application, the storage state of lean NOx traps (LNT) was successfully monitored, but application to this catalyst type seems to suffer from a comparably small signal [32,33]. The potential of the RF technique to determine the NH3 storage on SCR catalysts is presumed and is the focus of this paper. Previous work already proved the functionality for vanadia- and zeolite-based SCR catalysts [34–37]. Recently, we presented first results with a commercial zeolite-based catalyst on the engine dynamometer using for first time DEF instead of gaseous NH3 . The next big step by applying a direct RF-controlled DEF dosing on a specific NH3 storage value was achieved [38]. This paper continues that work with a focus on the temperature dependence of the RF signal, the influence of the NH3 storage on NH3 slip and maximum NOx conversion efficiency. The effect of humidity changes and the cold start behavior to the RF signal are also investigated. Within our work, we try to demonstrate possible benefits of a directly NH3
Sensors 2017, 17, 1615
3 of 15
storage-controlled SCR catalyst to operate the whole SCR system at its optimal NOx conversion point and to avoid NH3 slip. This might lead to increased system efficiency and more robust catalyst control systems for future applications and emission limits. 2. RF Catalyst State Monitoring In the applied measurement technique, the catalyst itself is the sensitive part of the sensor system. By storing NH3 , the catalyst material changes its electric properties and the cavity resonator, which is defined by the electrically conductive catalyst canning, is able to detect these very small changes. By coupling electromagnetic waves into the resonator, resonances, i.e., standing electromagnetic waves, can be excited at specific frequencies and their electric field interacts with the resonator filling. As a measurable material effect, the complex dielectric permittivity (ε = ε1 − jε2 ) of the catalyst is identified. The linear relation between the NH3 loading and both parts of the complex permittivity for zeolite SCR catalyst materials has been proven in several studies with a special setup using powder samples [39–41]. The currently expected material effects due to NH3 adsorption are the polar nature of the NH3 molecule and the effects of NH3 to the conductivity mechanisms inside the porous zeolite structure [42,43]. Each resonance can be fully described by two analyzable resonance parameters: the resonance frequency f res and the unloaded quality factor Q0 . The absolute frequency of f res is mainly defined by the resonance cavity geometry and the properties of the resonator filling material. Based on the theory of the so-called cavity perturbation method, small changes of the resonance frequency ∆f res /f 0 depend on the changes of the dielectric permittivity ∆ε1 , which represents the polarization effects (Equation (3)). Similarly, the changes of conductivity mechanisms and dielectric losses are represented in ∆ε2 and related to the change of the reciprocal unloaded quality factor ∆Q0 −1 (Equation (4)): ∆f res /f 0 ∝ ∆ε1
(3)
∆Q0 −1 ∝ ∆ε2
(4)
Further detailed descriptions and the theoretical background of the RF measurement technique, including the used assumptions and the extraction of the two resonance parameters f res and Q0 can be found in previous work [30,39,42]. It is possible to perform RF measurements with only one coupling element in simple reflection mode. By applying two coupling elements, the number of possible RF signals increases to four with two reflection and two transmission signals. Within this work, two coaxial probe antennas were used as coupling elements and the RF analysis is based on one transmission signal, the scattering parameter S21 . By acquiring complex RF data, the data analysis uses a complex fitting approach for f res and Q0 determination. 3. Experimental The presented study uses the same dynamometer setup as described in [38]. Under investigation is a well-studied serial-type copper-exchanged zeolite SCR catalyst (Cu-SSZ-13 [36,37,44], kindly provided by the Ford Motor Company) on a cordierite substrate. The illustrated setup in Figure 1 is described as follows: a turbocharged 4-cylinder and 2.1 l diesel engine (Daimler OM 651, 150 kW) is followed by the serial device oxidation catalyst (DOC) and DPF. The first located NOx sensor detects the pre-SCR NOx raw emissions. The DEF dosing (Bosch Denoxtronic 3.2) is applied together with an uncoated cordierite substrate to support NH3 formation from the DEF with additional surface contact and a plate mixer to improve NH3 concentration uniformity. The second NOx sensor determines together with the first one and its well-known NH3 cross sensitivity [45] the current dosed NH3 concentration. The Cu-SSZ-13 SCR catalyst (Ø 5.66” = 14.4 cm, length 6” = 15.2 cm) is placed in the middle of the 40 cm resonance cavity with one RF antenna up- and one downstream of the catalyst. The ideal cylindrical cavity shape is defined by two coarse metal screens. Two thermocouples outside
Sensors 2017, 17, 1615
4 of 15
of the resonance cavity determine the current catalyst temperature. The last NOx sensor downstream of the SCR catalyst detects the end-of-pipe emissions. Since the sensor is sensitive to both NOx and NH3 , its signal is required to be interpreted carefully. The two RF antennas are connected to the vector network analyzer (VNA, Anritsu MS46322A, RF acquisition rate: 1 Hz) by two 50 Ω coaxial cables (not Sensors 17, 44of Sensors 2017, 17,1615 16151). This work uses the lowest appearing resonance, the TE of14 14 shown2017, in Figure 111 mode with one electrical field maximum in the cavity center. Figure 2a shows the simulated (COMSOL Multiphysics 5.1) 5.1) electrical strength (high: light, black) 2b the 5.1) electrical field strength (high: light, low: black) and Figure Figure 2bmagnetic the magnetic magnetic field vector vector of the electrical fieldfield strength (high: light, low:low: black) and and Figure 2b the field field vector of theof TEthe 111 TE 111 It is clearly visible that the SCR catalyst is located in a region with high electric field TE 111 mode. mode. It is clearly visible that the SCR catalyst is located in a region with high electric field mode. It is clearly visible that the SCR catalyst is located in a region with high electric field strength, strength, since to permittivity changes on latter. strength, since the the tosensitivity sensitivity tochanges permittivity changesonisisthedepending depending on the the latter. Example Example since the sensitivity permittivity is depending latter. Example transmission spectra transmission spectra (|S 21 with the resonance peak of the TE 111 are displayed schematically transmission spectra (|S 21|) |) with the resonance peak of the TE 111 mode mode are displayed schematically (|S21 |) with the resonance peak of the TE111 mode are displayed schematically in Figure 2c for the in Figure 2c in Figure 2cfor forthe theNH NH33free freestate state(state (state11in inblack) black)and andthe theNH NH33loaded loadedstate state(state (state22in inred). red).The Theshift shift NH 3 free state (state 1 in black) and the NH3 loaded state (state 2 in red). The shift of the resonance of the resonance to lower frequencies, the decrease of peak height and the peak broadening due of the resonance to lower frequencies, the decrease of peak height and the peak broadening due to to lower frequencies, the decrease of peak height and the peak broadening due to NH3 storageto is NH 33 storage is clearly visible. NH storage is clearly visible. clearly visible.
DOC DOC
Daimler DaimlerOM OM651 651 150 150kW kW 44cylinder, cylinder,2.1 2.1ll RF RFantennas antennas NOxxsensor sensor mixer mixer NO Ø 14.4 14.4 cm cm Ø Ø 5.66“ 5.66“ Ø
NO NOxxsensor sensor
uncoated uncoated cordierite cordierite substrate substrate
DPF DPF DEF DEF
thermothermocouple couple
NO sensor NOxxsensor SCR SCR
exhaust exhaust
thermothermocouple couple
screens screens 6“ 6“==15.2 15.2cm cm
40 40cm cm
Figure dieselengine Figure1. 1.Illustration Illustrationof ofthe thedynamometer dynamometersetup: setup:2.1 2.1lldiesel diesel enginewith withdiesel dieseloxidation oxidationcatalyst catalystand and DEF dosing dosingwith mixer,Ø (Ø14.2 14.2cm) cm) particulate particulatefilter, filter,DEF withuncoated uncoatedcordierite cordieritesubstrate substrateand andplate platemixer, mixer, Ø5.66” 5.66”(Ø (Ø 14.2 cm) SCR catalyst canning defined by metal screens with two RF antennas, thermocouples upand SCR catalyst canning defined by metal screens with two RF antennas, thermocouples up- and downstreamof ofthe theSCR SCRand and three NO downstream of SCR and upstream of xx sensors upand downstream of and of downstream three NO sensors up-upandand downstream of SCR SCR and upstream upstream of DEF DEF downstream of the SCR and three NO x sensors DEF dosing. dosing. dosing.
Figure Figure2.2.Simulated Simulated (a)electric electric fieldstrength strength(high: (high:light, light,low: low:black) black) and (b) magnetic fieldvectors vectorsof of Simulated(a) electricfield black)and and(b) (b)magnetic magneticfield the theTE TE111 111mode; mode; and (c) example transmission spectrumof 111mode mode without NH (state 1, black) mode;and and(c) (c)example exampletransmission transmissionspectrum of the the TE TE111 modewithout withoutNH NH333(state (state1, 1,black) black) 111 111 and (state2, 2,red). red). andloaded loadedwith withNH NH333(state (state 2, red).
The TheDEF DEFdosing dosingon onthe theengine enginesetup setupcan canbe beapplied appliedmanually manuallyor orautomatically automaticallycontrolled controlledon onthe the current current RF RF signal signal with with defined defined control control borders borders as as already already demonstrated demonstrated in in [38]. [38]. Within Within this this study, study, the theengine engineisisoperated operatedat atseveral severalstationary stationaryoperation operationpoints pointswith withSCR SCRcatalyst catalysttemperatures temperaturesbetween between 250 250 and and 400 400 °C °C and and NO NOxx raw raw emissions emissions of of 100 100 up up to to 1300 1300 ppm. ppm. Due Due to to aa compared compared low low catalyst catalyst volume, volume, all all experiments experiments were were performed performed at at very very high high space space velocities velocities (SV) (SV) between between 90,000 90,000 and and
Sensors 2017, 17, 1615
5 of 15
The DEF dosing on the engine setup can be applied manually or automatically controlled on the current RF signal with defined control borders as already demonstrated in [38]. Within this study, the engine is operated at several stationary operation points with SCR catalyst temperatures between 250 and 400 ◦ C and NOx raw emissions of 100 up to 1300 ppm. Due to a compared low catalyst volume, all experiments were performed at very high space velocities (SV) between 90,000 and 150,000 h−1 that force the catalyst to operate at deliberately difficult conditions. Additionally, one operation point with continuously changing exhaust gas recirculation (EGR) rates was chosen, resulting in continuously fluctuating NOx concentrations, space velocities and exhaust gas humidities. The latter was also under further investigation by analyzing the cold and warm start water influence to the RF signal. Sensors 2017, 17, 1615
5 of 14
4. Results and Discussion 4.1. RF
humidities. The latter was also under further investigation by analyzing the cold and warm start water influence to the RF signal. Response Validation and Procedure for NH3 Storage Influence Investigations
4. Results and Discussion In [38] the functionality of the RF system on the engine test bench with an iron exchanged zeolite catalyst was proven for the first time. The first experiment (Figure 3) of this paper was performed to 4.1. RF Response Validation and Procedure for NH 3 Storage Influence Investigations show the same functionality for the observed Cu-SSZ-13 catalyst with a iron space velocity of 105,000 h−1 , In [38] the functionality of the RF system on the engine test bench with an exchanged zeolite ◦ C, forand the first time. The first experiment (Figure 3) of thiswith paper(a) wasthe performed to of the NOx an air-to-fuelcatalyst ratio was of λproven = 1.35 a catalyst temperature of 290 signals show the same functionality for the observed Cu-SSZ-13 catalyst with a space velocity of 105,000 h−1, sensors located upstream of the DEF dosing (black) and downstream of the SCR catalyst (red: assigned an air-to-fuel ratio of λ = 1.35 and a catalyst temperature of 290 °C, with (a) the signals of the NOx to downstream NOxlocated , blue:upstream assigned NH3and ); (b)downstream the dosedofNH concentration sensors of to thedownstream DEF dosing (black) the 3SCR catalyst (red: determined assigned to downstream NO x , blue: assigned to downstream NH 3 ); (b) the dosed NH 3 concentration by the two NOx sensors up- and downstream of the DEF dosing; (c) the calculated stored NH3 mass in the two NOx sensors up- and downstream of the DEF dosing; (c) the calculated stored gram per literdetermined catalyst by volume; (d) the resonance frequency f res in reverse scale and (e) the reciprocal NH3 mass in gram per liter catalyst volume; (d) the resonance frequency fres in reverse scale and (e) −1 . unloaded quality factor Q 0 the reciprocal unloaded quality factor Q0−1. T = 290 °C, SV = 105,000 h , λ = 1.35 750 upstream NO downstream NOx x 500
cNH3 / ppm
cNOx sensor / ppm
-1
250 0
2000
mNH3 / g/lcat fres / GHz
500
1000
1500
NH3
2000
2500 (b)
dosing rate 2 dosing rate 1
1000 0
NH3
NH3
NH3
0
(a)
0
500
1000
1500
2000
2500 (c)
0
500
1000
1500
2000
2500 (d)
0
500
1000
1500
2000
2500 (e)
500
1000 t/s
2 1 0
1.062 1.064 1.066 1.068 1.070
Q0 x 1000
7.5
-1
5.0 2.5
0
t1
t2
1500
t3
2000
t4
2500
Figure 3. Experiment at 290 °C with SV = 105,000 h−1 and λ = 1.35 and two DEF dosing rates: (a) NOx
Figure 3. Experiment at 290 ◦ C with SV = 105,000 h−1 and λ = 1.35 and two DEF dosing rates: (a) NOx sensor signal upstream of DEF dosing (black) and downstream of SCR catalyst (red: assigned to sensor signaldownstream upstreamNOofx, blue: DEFassigned dosing (black) and SCR catalyst (red:by assigned to to downstream NHdownstream 3); (b) dosed NH3 of concentration determined signals up- and to downstream of DEF NH dosing; (c) calculated amountconcentration of NH3 stored on determined the downstream NO NOx xsensor , blue: assigned downstream ); (b) dosed NH by 3 3 catalyst; (d) resonance frequency fres in reverse scale and (e) the reciprocal unloaded quality factor NOx sensor signals upand downstream of DEF dosing; (c) calculated amount of NH stored on the 3 Q0−1. The dashed lines indicate the loading level when first NH3 slip occurs. catalyst; (d) resonance frequency f res in reverse scale and (e) the reciprocal unloaded quality factor Q0 −1 . The dashed lines indicate the loading level when first NH3 slip occurs.
Sensors 2017, 17, 1615 Sensors 2017, 17, 1615
6 of 15 6 of 14
Within Withinthis thisexperiment, experiment,the thecatalyst catalystwas wasloaded loadedwith withNH NH33for forfour fourtimes timeswith withtwo twodifferent differentDEF DEF dosing dosing raterate 2 injects doubledouble the amount of urea as ratedosing 1. When continuous dosingrates, rates,whereas whereas dosing 2 injects the amount ofdosing urea as rate 1. When urea dosing urea is applied, NOx concentration drops instantaneously and shows after continuous dosingthe is downstream applied, the downstream NOx concentration drops instantaneously and ashows short time NOx time conversion. the NHWhen storage capacity is exceeded, NH breakthrough after full a short full NOWhen x conversion. the NH 3 storage capacity is exceeded, NH 3 3 3 appears (highlighted in blue) visible in the increase in the downstream NO sensor signal (t , t , t and breakthrough appears (highlighted in blue) visible in the increase in the x downstream NO 1 2x sensor 3 tsignal is turned off dosing again, the NH3 breakthrough slowly, followed by (t1, the t2, tDEF 3 anddosing t4). When the DEF is turned off again, thedecreases NH3 breakthrough decreases 4 ). When another in by NOanother sensor signal up to the upstream concentration, indicating that the catalyst is slowly, increase followed increase in NO x sensor signal up to the upstream concentration, x NH Thecatalyst calculated NH on the shows theoncritical NH3 storage degree indicating that the is NH 3-free again. Thecatalyst calculated NH3that mass the catalyst shows that the 3 -free again. 3 mass appears to be arounddegree 1.4 g/lappears four dosing experiments slip is visible when NH this3 critical NH 3 storage toin beall around 1.4 g/lcat , since in all NH four3 dosing experiments cat , since storage degreewhen is exceeded. This degree proves is theexceeded. good reproducibility of the NHreproducibility slip is visible this storage This proves the good of the and NH3 3 storage experiment − 1 the chemical behavior of the the chemical catalyst. By comparing calculated NH3 massthe with f res and Q , the storage experiment and behavior of thethe catalyst. By comparing calculated NH 0 3 mass good both RF signals and theboth catalyst NH3 loading is proven for Cu-SSZ-13. with correlation fres and Q0−1between , the good correlation between RF signals and thestate catalyst NH3 loading state is 1 as a(a) This relation is better visible in Figure with visible (a) f res in and (b) Q40 −with function proven for Cu-SSZ-13. This relation is 4better Figure fres and of (b)stored Q0−1 asNH a function 3 mass. The linear relationship between RF signalsbetween and the both catalyst with no of stored NH3 mass. The linearboth relationship RFNH signals and degree the catalyst NHinfluence 3 storage 3 storage whether the catalyst is storing, depleting and converting is clearly visible as already reported degree with no influence whether the catalyst is storing,NO depleting and converting NO x is clearly x in [38]. as already reported in [38]. visible
(b)
Q-10 x 1000
T = 290 °C, SV = 105,000 h-1, λ = 1.35 (a) 7.5 1.062 fres / GHz
1.064 1.066
5.0
1.068 1.070 0
1 2 mNH3 / g/lcat
2.5
0
1 2 mNH3 / g/lcat
Figure 4. RF signal of experiment of Figure 3 at 290 °C with SV = 105,000 h−1 and λ = 1.35: (a) the Figure 4. RF signal of experiment of Figure 3 at 290 ◦ C with SV = 105,000 h−1 and λ = 1.35: (a) the resonance frequency fres in reverse scale and (b) the reciprocal unloaded quality factor Q0−1 as a resonance frequency f res in reverse scale and (b) the reciprocal unloaded quality factor Q0 −1 as a function of the calculated stored NH3 on the catalyst. function of the calculated stored NH3 on the catalyst.
With the proven functionality of the RF signals for NH3 storage determination, the RF signal was the proven functionality of the signals for determination, RF signal 3 storage usedWith for automatic urea dosing control to RF investigate theNH influence of the NH3 storagethe degree to the was used for automatic urea dosing control to investigate the influence of the NH storage degree 3 catalyst performance. Figure 5 shows an example for an experiment performed with a space velocity to catalyst 5 shows an example experiment with a space3 ofthe 105,000 h−1, performance. λ = 1.35 and aFigure catalyst temperature of 290 for °C.an It displays the performed same signals as Figure −1 , λ = 1.35 and a catalyst temperature of 290 ◦ C. It displays the same signals velocity of 105,000 h with the additional plot (f) of the apparent NOx conversion rate based on the signals of NOx sensors as 3 with the additional the ploturea (f ) of the apparent NOx conversion rate baseddegrees on the signals of Figure (a). Within this experiment, dosing was controlled to constant storage by Q0−1. of NO sensors of (a). Within this experiment, the urea dosing was controlled to constant storage x These experiments were also conducted with control on fres, leading to the same results. Starting with −1 degrees by3 Qstorage, were also conducted withstarting control with on f res to the same 0 . These a low NH the experiments latter was increased stepwise, always an, leading empty catalyst. The results. Starting with a low NH storage, the latterto was always starting 3 cat lowest observed storage of 0.2 g/l (corresponding Q0−1increased × 1000 = stepwise, 3.42) already shows a highwith NOx an empty catalyst. observed storagethe of 0.2 g/lcatvalue, (corresponding to Q0 −1 ×efficiency 1000 = 3.42) conversion of over The 90%.lowest By stepwise increasing storage the NOx conversion also already shows a high NO conversion of over 90%. By stepwise increasing the storage value, the NO x x −1 increases and reaches constant full conversion at a NH3 storage level of 1.0 g/lcat (Q0 × 1000 = 4.97). conversion efficiency also increases and reaches constant full conversion at a NH storage level of 3 When the control value for NH3 storage further increases, the NOx sensor downstream of the catalyst 1.0 g/lcat (Q0 −1signal × 1000 = 4.97). When the control for NH furtherby increases, the NOx 3 storage slip. This might be explained slowly migrating shows a slow increase indicating slow NH3value sensor downstream a slow signal increase indicating slip. This might NH3 from the front of of the the catalyst catalyst shows to its end if one constant storage value isslow keptNH for3 longer time. This be explained by slowly migrating NH from the front of the catalyst to its end if one constant storage 3 effect gets stronger with further growing NH3 storage degree, until at 1.9 g/lcat (Q0−1 × 1000 = 6.27) the value is kept for time. This effect gets stronger with further growing NH3 storage degree, downstream NOlonger x sensor shows almost 200 ppm NH3 signal. This experiment demonstrates thatuntil with − 1 at g/lcatknowledge (Q0 × 1000 = 6.27) the downstream sensor shows 200in ppm NH3with signal. a 1.9 precise of the current NH3 storage NO the xcatalyst can bealmost operated a state its This experiment demonstrates thatand with a precise knowledge of thestorage current limit NH3 for storage catalyst maximum conversion efficiency without crossing the critical NH3the slip. At the can be operated in a state itswith maximum without crossing the critical −1, the observed temperature of with 290 °C a spaceconversion velocity ofefficiency 105,000 hand NH3 storage degree of 1.0 g/lcat appears to be the ideal operation point.
Sensors 2017, 17, 1615
7 of 15
storage limit for NH3 slip. At the observed temperature of 290 ◦ C with a space velocity of 105,000 h−1 , Sensors 17, 1615 7 of 14 the NH32017, storage degree of 1.0 g/lcat appears to be the ideal operation point.
T = 290 °C, SV = 105,000 h , λ = 1.35 750 upstream downstream NOx 500
cNOx sensor / ppm
-1
250 0
fres / GHz
NH 3
NH 3
NH 3
NH 3
0
1200
2400
3600
4800
6000
7200
8400 (b)
0
1200
2400
3600
4800
6000
7200
8400 (c)
0
1200
2400
3600
4800
6000
7200
8400 (d)
0
1200
2400
3600
4800
6000
7200
8400
1000 0
mNH3 / g/lcat
cNH3 / ppm
2000
NH 3
(a)
2 1 0
1.064 1.066 1.068 1.070
5.0
3.94 3.42 3.68
-1
Q0 x 1000
7.5
conversion / %
2.5
0
1200
4.2
4.46 4.71
2400
4.97
3600
5.23
4800
5.49
6000
5.75
6.01 6.27 (e)
7200
100
8400 (f)
50 0
0
1200
2400
3600
4800
6000
7200
8400
t/s ◦ C with Figure 5. 5.Experiment toto investigate the NH influence toto catalyst performance atat 290 Figure Experiment investigate the NH 3 storage influence catalyst performance 290 °C with 3 storage − 1 SVSV= =105,000 andλλ== 1.35: 1.35: (a) (a) NOx NOx sensor sensorsignal signalupstream upstreamofofDEF DEFdosing dosing(black) (black)and anddownstream downstream 105,000h h−1and ofof SCR catalyst (red: assigned to downstream NOx ,NO blue: assigned to downstream NH3 ); NH (b) dosed SCR catalyst (red: assigned to downstream x, blue: assigned to downstream 3); (b) NH dosed 3 concentration determined by NO sensor signals upand downstream of DEF dosing; (c) calculated NH3 concentration determinedx by NOx sensor signals up- and downstream of DEF dosing; (c) amount of NH catalyst; frequency f resfrequency in reversefresscale; (e) thescale; reciprocal 3 stored on (d) the resonance catalyst; (d) resonance in reverse (e) the calculated amount of on NHthe 3 stored unloaded quality factorquality Q0 −1 and (f) the conversion on the sensor (a). reciprocal unloaded factor Q0−1apparent and (f) NO the xapparent NObased x conversion basedsignals on theofsensor
signals of (a).
4.2. Temperature Dependency of NH3 Storage, NOx Conversion and RF Signals 4.2. Temperature Dependency of NH3 Storage, NOx Conversion and RF Signals The experiment discussed above was performed at various operation points in the temperature The discussed performed at various operation pointsNH in 3the temperature range of 250experiment to 400 ◦ C. Within this above study, was the ideal NH3 storage value, i.e., the lowest storage when range of NO 250 xtoconversion 400 °C. Within this study, ideal NHfor 3 storage value, temperatures. i.e., the lowestAdditionally, NH3 storage maximum was achieved, wasthe determined all observed when maximum conversion was achieved, was determined for all the observed the storage value ofNO firstx NH was analyzed, which represents first NHtemperatures. 3 breakthrough 3 slip when Additionally, the storage value of first NH3 breakthrough was analyzed, which represents the first NH3 slip when the previously NH3 free catalyst is loaded with a constant urea dosing rate (as in the experiment displayed in Figure 3). Figure 6 shows: (a) the ideal NH3 storage degree (red triangles) and the NH3 breakthrough loading (black circles); and (b) the maximum achieved NOx conversion without NH3 slip. Both storage values are also fitted with an exponential decay function (solid line).
Sensors 2017, 17, 1615
8 of 15
the previously NH3 free catalyst is loaded with a constant urea dosing rate (as in the experiment displayed in Figure 3). Figure 6 shows: (a) the ideal NH3 storage degree (red triangles) and the NH3 Sensors 2017, 17, 1615 of 14 breakthrough loading (black circles); and (b) the maximum achieved NOx conversion without8NH 3 slip. Both storage values are also fitted with an exponential decay function (solid line). One can see One can see that both storage curves decrease with increasing catalyst temperature, since the NH3 that both storage curves decrease with increasing catalyst temperature, since the NH3 desorption is desorption is thermally activated. This temperature dependence fits well to the expected behavior thermally activated. This temperature dependence fits well to the expected behavior relying on results relying on results of previous work on the gas test bench [37,40] and to current control models [11]. of previous work on the gas test bench [37,40] and to current control models [11]. Both curves are close Both curves are close together at the lower observed temperatures, whereas the ideal storage curve together at the lower observed temperatures, whereas the ideal storage curve shows a stronger decay shows a stronger decay with temperature than the first breakthrough. This might be related to the with temperature than the first breakthrough. This might be related to the better reaction kinetics at better reaction kinetics at higher temperatures that does not require high NH3 surface coverage. The higher temperatures that does not require high NH3 surface coverage. The achieved NOx conversion achieved NOx conversion at stationary operation points was always higher than 95% and increases at stationary operation points was always higher than 95% and increases at temperatures above 280 ◦ C at temperatures above 280 °C to 98%, due to the thermally activated reaction kinetics. The best value to 98%, due to the thermally activated reaction kinetics. The best value of 98% may also be related to of 98% may also be related to the accuracy limit of the used NOx sensors and might represent full the accuracy limit of the used NOx sensors and might represent full conversion, even at the observed conversion, even at the observed forcing conditions with extreme high space velocities. It should be forcing conditions with extreme high space velocities. It should be noted here, that the catalyst was noted here, that the catalyst was operated at very unusual high space velocities. At typical space operated at very unusual high space velocities. At typical space velocities, an even better performance velocities, an even better performance can be expected. can be expected. first breakthrough (a)
1
ideal storage
0 250
100 UNOx / %
mNH3 / g/lcat
2
300 350 T / °C
95
(b)
maximum achieved NOx conversion without NH3 slip
400
90 250
300 350 T / °C
400
Figure 6. 6. NH NH33 storage storage behavior behavior of of Cu-SSZ-13 Cu-SSZ-13 with with (a) (a) the the ideal ideal storage storage degree degree (lowest (lowest storage storage for for Figure maximum conversion, red triangles) and the storage at first breakthrough at continuous urea dosing maximum conversion, red triangles) and the storage at first breakthrough at continuous urea dosing with aa previously x xconversion with previously NH NH33 free free catalyst catalyst (black (blackcircles) circles)and and(b) (b)the themaximum maximumachieved achievedNO NO conversionasasa afunction functionofofcatalyst catalysttemperature. temperature.
The determined temperature dependency of the RF signals is displayed in Figure 7 for the NH33 free state (black squares), the NH33 breakthrough loading (black circles) and the ideal NH33 storage −11 degree (red triangles), with with (a) (a) ffres res in reverse scale and and (b) (b) Q Q00− . .In Inprinciple, principle,ititobvious obvious that that both both RF signals signals appear appear to have a very similar temperature dependent behavior. Without NH33,, they they show in the lower temperature region with increasing temperature a decrease in the opposite direction of the signal as it corresponds to NH33 loading. loading. For For higher higher temperatures, temperatures, aa small small increase increase in in direction of the NH33 signal signal is is visible. visible. This This behavior behavior might might be be explained explained by by several several reasons reasons related related to to material effects and the resonator cavity geometry. At the lower temperatures, the zeolite catalyst stores H22O O at the same storage toto the RFRF signal as as NHNH 3. With increasing temperature, the storagesites sitesand andwith witha asimilar similareffect effect the signal increasing temperature, 3 . With ability to adsorb H 2 O decreases and so do both RF signals. At the higher temperature regions, H 2O the ability to adsorb H2 O decreases and so do both RF signals. At the higher temperature regions, has almost no influence, but instead, the charge carriers inside the zeolite structure get more mobile, H2 O has almost no influence, but instead, the charge carriers inside the zeolite structure get more which leads to leads an increase in RF signal. thisBut effect comparably small tosmall H2O,towhich be mobile,also which also to an increase in RF But signal. thisiseffect is comparably H2 O, can which −1. The resonance − 1 seen for Q 0 frequency is additionally affected, since the resonator cavity expands can be seen for Q0 . The resonance frequency is additionally affected, since the resonator cavity with temperature. This geometry increase leads to a leads proportional decrease indecrease resonance expands with temperature. This geometry increase to a proportional in frequency resonance and explains the behavior fres at the of higher for the emptyfor state. Since Qstate. 0−1 (the dielectric frequency and explains theofbehavior f res attemperature the higher temperature the empty Since Q0 − 1 −1 shows a smaller − 1 losses) are not affected by the geometry, Q 0 temperature dependency. The NH3 (the dielectric losses) are not affected by the geometry, Q0 shows a smaller temperature dependency. −1 − 1 breakthrough curve shows an increase signal intensity fres and Q 0 fwith increasing temperature, The NH3 breakthrough curve shows anofincrease of signalfor intensity for Q0 with increasing res and whereas the stored NH 3 mass decreases. The ideal NH 3 storage curves are in the lower temperature temperature, whereas the stored NH3 mass decreases. The ideal NH3 storage curves are in the lower region close to the breakthrough values and move towards higher temperatures roughly into the middle between the empty and breakthrough values, as already seen in Figure 6. The experimental results shown in Figure 4 already proved the linear response of both RF signals to NH3 storage and this behavior was also seen in all other experiments in the whole temperature range. Therefore, the following discussion focusses on the sensitivity of both RF signals to NH3
Sensors 2017, 17, 1615
9 of 15
temperature region close to the breakthrough values and move towards higher temperatures roughly into the middle between the empty and breakthrough values, as already seen in Figure 6. The experimental results shown in Figure 4 already proved the linear response of both RF signals to Sensors NH3 storage and this behavior was also seen in all other experiments in the whole temperature 2017, 17, 1615 9 of 14 Sensors 2017, 17, 1615 9 of 14 range. Therefore, the following discussion focusses on the sensitivity of both RF signals to NH3 storage, i.e., the slopes assuming anassuming ideal linear relationship. Figure 8 displays the 8sensitivities to NH storage, i.e., slopes ideal linear relationship. Figure displays sensitivities 3 storage to storage, i.e., thethe slopes assuming anan ideal linear relationship. Figure 8 displays thethe sensitivities to −1 as −1 (a)NH Sf for f and (b) S for Q a function of catalyst temperature. It is clearly visible thatvisible the 3 storage (a) S f for 0 as a function of catalyst temperature. It is clearly res Q fres and 0 (b) SQ for Q −1 NH3 storage (a) Sf for fres and (b) SQ for Q0 as a function of catalyst temperature. It is clearly visible sensitivities of both RF signal increase in an almostin linearalmost manner withmanner temperature. This might also that sensitivities both signal increase linear with temperature. This that thethe sensitivities of of both RFRF signal increase in anan almost linear manner with temperature. This bemight causedalso by the higher mobility of charge carriers at higher temperatures and explains the and increase in be caused by the higher mobility of charge carriers at higher temperatures explains might also be caused by the higher mobility of charge carriers at higher temperatures and explains NH signal intensity in Figure 7 while the stored NH mass decreases (see Figure 6). Similar effects stored NH 3 mass decreases (see Figure 6). 3 increase in NH3 signal intensity in Figure 7 while thethe increase in NH3 signal intensity in Figure 7 while3thethe stored NH 3 mass decreases (see Figure 6). have already been reported forbeen metal exchanged zeolites in previous work in [37,40]. Similar effects have already reported for metal exchanged zeolites previous work [37,40]. Similar effects have already been reported for metal exchanged zeolites in previous work [37,40].
6 6
ideal storage 3 ideal NHNH storage 3
1.070 1.070 1.072 1.072 250 250
-1
1.066 1.066 1.068 1.068
8 8
(a)(a)
empty empty
300 350 300 350 T / °C T / °C
breakthrough 3 NHNH breakthrough 3
Q0 -1x 1000 Q0 x 1000
fres / GHz fres / GHz
1.062 1.062 breakthrough 3 NHNH breakthrough 3 1.064 1.064
400 400
4 4 2 2 250 250
(b)(b)
storage ideal 3 storage ideal NHNH 3 empty empty
300 350 300 350 T / °C T / °C
400 400
Figure 7. RF signals as function of catalyst temperature for the empty state (black squares), the NH3 of of catalyst temperature for for the the empty statestate (black squares), the NH Figure 7. RF RFsignals signalsasasfunction function catalyst temperature empty (black squares), the3 breakthrough (black circles) and the ideal NH3 storage degree (red triangles): (a) the resonance 3 storage degree breakthrough (black circles) andand the the ideal NH NH (black circles) ideal NH degree(red (redtriangles): triangles):(a) (a) the the resonance 3 breakthrough 3 storage 0−1. frequency fres in reverse scale and (b) the reciprocal quality factor Q frequency f res inreverse reversescale scaleand and(b) (b)the thereciprocal reciprocal quality quality factor factor Q Q00−1−. 1 . resin
(a)(a)
(b)(b)
0.003 0.003 0.002 0.002
-0.004 -0.004 -0.002 -0.002 0.000 0.000 250 250
0.005 0.005 0.004 0.004
SQ / lcat/g SQ / lcat/g
Sf / GHz/(g/lcat) Sf / GHz/(g/lcat)
-0.008 -0.008 -0.006 -0.006
300 350 350 300 T / °C T / °C
0.001 0.001 0.000 0.000 400 250 400 250
300 350 350 300 T / °C T / °C
400 400
Figure 8. Sensitivity of the RF signals to NH3 storage with (a) Sf of the resonance frequency fres and (b) Figure 8. 8. Sensitivity the resonance frequency fres and (b) Figure Sensitivity of of the the RF RFsignals signalsto toNH NH3 3storage storagewith with(a) (a)SfSof f of the resonance frequency f res and 0−1 as a function of catalyst temperature. SQ of the reciprocal quality factor Q −1 −a1 function reciprocal quality factor Q0 Qas of catalyst temperature. SQ of (b) SQthe of the reciprocal quality factor as a function of catalyst temperature. 0
The results this study now offer chance a temperature dependent calibration The results of of this study now offer thethe chance forfor a temperature dependent calibration of of thethe RFRF The results ofobserved this study now offer the chance for a temperature dependent calibration of theunder RF system for the catalyst Cu-SSZ-13 to monitor and control the current NH 3 storage system for the observed catalyst Cu-SSZ-13 to monitor and control the current NH3 storage under system for the observedTherefore, catalyst Cu-SSZ-13 to monitor and control the current NH3(shown storage under transient conditions. quadratic fitting functions empty state Figure transient conditions. Therefore, thethe quadratic fitting functions forfor thethe empty state (shown in in Figure transient conditions. Therefore, the quadratic fitting functions for the empty state (shown in Figure 7) 7) and the linear fitting functions for the sensitivities (shown in Figure 8) can be used for real time 7) and the linear fitting functions for the sensitivities (shown in Figure 8) can be used for real time and the linear fitting functionsand for is theinsensitivities (shown in work. Figure 8) can be used for real time NH3 NH 3 storage determination focus of forthcoming NH 3 storage determination and is in focus of forthcoming work. storage determination and is in focus of forthcoming work. 4.3. Influences Humidity Changes and Cold Start Water 4.3. Influences of of Humidity Changes and Cold Start Water 4.3. Influences of Humidity Changes and Cold Start Water The experiments [38] already showed influence humidity exhaust gas The experiments in in [38] already showed thethe influence of of humidity in in thethe exhaust gas to to thethe RFRF The experiments in [38] already showed the influence of humidity in the exhaust gas to the RF signal for an iron exchanged zeolite, leading to a decrease in signal accuracy. The identical experiment signal for for an zeolite, leading to aa decrease in accuracy. The identical experiment signal an iron iron exchanged exchanged zeolite, leading to decrease in signal signal accuracy. Thespace identical experiment with a continuously changing EGR rate that causes varying λ values, varying velocities, well with aa continuously continuously changing changing EGR EGR rate rate that that causes causes varying varying λ λ values, values, varying varying space space velocities, velocities, as asas well with well varying raw NO x emissions at a constant catalyst temperature has been repeated for the more asas varying raw NO x emissions at a constant catalyst temperature has been repeated for the more as varying raw NO at a and constant catalyst in temperature hasλbeen repeated for the more recent x emissions recent serial catalyst Cu-SSZ-13 is displayed Figure The signal upstream NO x sensor recent serial catalyst Cu-SSZ-13 and is displayed in Figure 9. 9. The λ signal of of thethe upstream NO x sensor in (a) shows the continuously changing EGR rate and takes values between 1.25 and 2.25. Within this in (a) shows the continuously changing EGR rate and takes values between 1.25 and 2.25. Within this experiment, catalyst loaded with NH 3 three times with two different urea dosing rates, each experiment, thethe catalyst is is loaded with NH 3 three times with two different urea dosing rates, each time until NH 3 breakthrough is detected by the downstream NOx sensor (highlighted in blue). The time until NH3 breakthrough is detected by the downstream NOx sensor (highlighted in blue). The signals NO x sensors upstream of the DEF dosing (black) and downstream (red) of the catalyst signals of of thethe NO x sensors upstream of the DEF dosing (black) and downstream (red) of the catalyst in (b) show without dosing the identical noisy behavior mirroring the λ signal. With applied urea
Sensors 2017, 17, 1615
10 of 15
serial catalyst Cu-SSZ-13 and is displayed in Figure 9. The λ signal of the upstream NOx sensor in (a) shows the continuously changing EGR rate and takes values between 1.25 and 2.25. Within this experiment, the catalyst is loaded with NH3 three times with two different urea dosing rates, each time until NH3 breakthrough is detected by the downstream NOx sensor (highlighted in blue). The signals of the NOx sensors upstream of the DEF dosing (black) and downstream (red) of the catalyst in (b) show without dosing the identical noisy behavior mirroring the λ signal. With applied urea dosing the downstream NOx sensor signal drops instantaneously and shows high but no full conversion until the NH3 breakthrough appears. The fact that the catalyst is not able to achieve full conversion and the Sensors 2017, 17, 1615 10 of 14 downstream NOx sensor still detects roughly 50 ppm might be explained by the high space velocity conversion until the NHthe 3 breakthrough appears. The fact that the catalyst is not able to achieve full (or low catalyst volume) and short-term high NOx concentration up to 1000 ppm. The dosed conversion and the downstream NOx sensor still detects roughly 50 ppm might be explained by the NH3 concentration in (c), calculated from the NOx sensor signal up- and downstream of the DEF high space velocity (or low catalyst volume) and the short-term high NOx concentration up to 1000 dosing, switches between two concentrations since the exhaust mass flow changes continuously ppm. The dosed NH 3 concentration in (c), calculated from the NOxgas sensor signal up- and downstream −1 in of therate DEFremains dosing, switches between twoRF concentrations the and exhaust mass but the dosing constant. Both signals f ressince in (e) Q0gas (f)flow stillchanges correlate very continuouslystored but theNH dosing rate remains constant. Both RF signals fres in (e) and Q0−1 in (f) still well to the calculated 3 mass in (d). However, both RF signals appear more affected by the correlate very well to the calculated stored NH3 mass in (d). However, both RF signals appear more changing humidity for Cu-SSZ-13 compared to the iron exchanged zeolite from [38], resulting in a affected by the changing humidity for Cu-SSZ-13 compared to the iron exchanged zeolite from [38], bigger uncertainty, asa can beuncertainty, learned from more “noisy” curves in (e) andin(f). resulting in bigger as canthe be learned from the more “noisy” curves (e) and (f). 2.5
T = 275 °C, SV = 90,000 - 150,000 h-1 (a)
λ
2.0 1.5
cNH3 / ppm
cNOx sensor / ppm
1.0 1600 1200 800 400 0 3000
0
2000
200 400 upstream NOx downstream NOx
NH3 200
400
600
800 NH3
NH3 600
800
dosing rate 1
1000 (b)
1000 (c) dosing rate 2
1000
mNH3 / g/lcat
0
fres / GHz
0
2
0
200
400
600
800
1000 (d)
0
200
400
600
800
1000 (e)
0
200
400
600
800
1000
1 0
1.062 1.064 1.066 1.068 1.070
Q-10 x 1000
7.5
(f)
5.0 2.5
0
200
400
t/s
600
800
1000
Figure 9. Experiment with continuously changing EGRrate rate with temperature of 275 °C Figure 9. Experiment with continuously changing EGR witha catalyst a catalyst temperature ofand 275 ◦ C and a space velocity between 90,000 and 105,000 − h−11 with (a) the λ signal of the upstream NOx sensor; (b) a space velocity between 90,000 and 105,000 h with (a) the λ signal of the upstream NOx sensor; the NOx sensor signals upstream of the DEF dosing (black) and downstream of the SCR catalyst (red: (b) the NOxdownstream sensor signals upstream of the DEF dosing (black) and downstream of the SCR catalyst downstream NH 3); (c) dosed NH3 concentration determined by NOx sensor NOx, blue: catalyst; (e) by NOx signals upand of DEF dosing; NH3 stored on the (red: downstream NO blue: downstream NH(d) ); (c) dosedamount NH3 of concentration determined x , downstream 3 calculated resonance frequency fres in reverse scaledosing; and (f) the unloaded quality factor stored Q0−1. on the catalyst; sensor signals up- and downstream of DEF (d)reciprocal calculated amount of NH 3 (e) resonance frequency f res in reverse scale and (f) the reciprocal unloaded quality factor Q0 −1 .
Besides small humidity changes under transient conditions, a much bigger effect might be caused by adsorbed water as it appears at cold starts. This has already been observed for a TWC-
Sensors 2017, 17, 1615
11 of 15
Besides Sensors 2017, 17,small 1615 humidity changes under transient conditions, a much bigger effect might be caused 11 of 14 by adsorbed water as it appears at cold starts. This has already been observed for a TWC-coated GPF in [30]. start water adsorption is also under investigation to better understand SCR coated GPF Cold in [30]. Cold start water adsorption is also under investigation to better zeolite understand catalysts and to improve the conversion efficiency efficiency especiallyespecially for continuously decreasing exhaust zeolite SCR catalysts and to improve the conversion for continuously decreasing gas temperatures [46]. Therefore, the startthe behavior of the RF-SCR system system has been analyzed under exhaust gas temperatures [46]. Therefore, start behavior of the RF-SCR has been analyzed different start conditions. The results are displayed in Figure 10 with f res(a)infres reverse scalescale and under different start conditions. The results are displayed in Figure 10 (a) with in reverse −1 Q −1 (b) as a function of temperature. The start procedure and the ambient temperature were identical andQ(b) 0 as a function of temperature. The start procedure and the ambient temperature were 0 for each run. Each performed cold or cold warm marked by a different color.color. Additionally, the identical for each run. Each performed or start warmisstart is marked by a different Additionally, values of the operation pointpoint withwith NH3NH from Figure 7 are7added (white diamonds). The cold the values of stationary the stationary operation 3 from Figure are added (white diamonds). The ◦ C. starts were were conducted with awith catalyst start temperature of 25 ◦of C,25 the warm starts starts of 120of The cold starts conducted a catalyst start temperature °C, the warm 120 °C.basic The curve of bothofRF signals for a cold is as follows. First, theFirst, signals into shift the same basic curve both RF signals for start a cold start is as follows. theshift signals into direction the same ◦ C. With as NH3 storage cause until theyuntil reach their maximum roughly roughly around 75 further direction as NH3would storage would cause they reach their maximum around 75 °C. With increasing temperature, they shift back in the opposite direction until they are identical to the values further increasing temperature, they shift back in the opposite direction until they are identical to the ◦ C. of the stationary operation points points above above temperatures of 250 of The with values of the stationary operation temperatures 250 °C.very Thefirst verycold firststart coldbegins start begins − 1 −1 × 1000 × values of f res of = 1.01 1000 = 9.0 and shows for f the biggest shift. All following cold with values fres =GHz 1.01 and GHzQand Q 0 = 9.0 and shows for f res the biggest shift. All following res 0 − 1 −1 starts started at f resat= f1.08 GHzGHz and Q = 1.0 with nono influence whether thethe engine waswas off for cold starts started res = 1.08 and 0 1000 × 1000 = 1.0 with influence whether engine off 0 Q× 12 coldcold startstart showed above 100 ◦100 C an identical behavior and this the high foror1272orh.72Each h. Each showed above °Calmost an almost identical behavior andproved this proved the reproducibility of the of cold of RF-SCR system. system. A possible explanation for the differing high reproducibility thestart coldinfluence start influence of RF-SCR A possible explanation for the differing first might cold start might be the catalyst fact thathas thenot catalyst has not beenand heated andtoitroom was first cold start be the fact that been heated before it wasbefore exposed exposed tofor room humidity for a longthe time. Between thestarts, different starts, the catalyst not able humidity a long time. Between different cold the cold catalyst was not able towas adsorb the to adsorb the same amount of water than before. The performed warm starts fit after a short time same amount of water than before. The performed warm starts fit after a short time after the engine ◦ C very well after theroughly engine started, roughly above temperatures of 180 °C cold verystart well curves. to the cold start curves. started, above temperatures of 180 to the
(a)
first cold start
1.00
cold start after 12 - 72 h
fres / GHz
1.02 warm start after 15 min
1.04 1.06 1.08
stationary operation points
0
100
200 300 T / °C
400
12
first cold start
10 Q-10 x 1000
0.98
(b)
cold start after 12 - 72 h
8 warm start after 15 min
6 4 2 0
stationary operation points
0
100
200 300 T / °C
400
Figure 10. 10. RF RF signals signals as as aa function function of of catalyst catalyst temperature temperature for for cold cold and and warm warm starts starts under under different different Figure start conditions conditions(colored (coloredsymbols) symbols)and andunder understationary stationary conditions (white diamonds from Figure 7) start conditions (white diamonds from Figure 7) for −1. − 1 res and (b) Q 0 for (a) f (a) f res and (b) Q0 .
The frequency shift of the first cold start related to the stationary operation point of 300 °C was The frequency−1shift of the first cold start related to the stationary operation point of 300 ◦ C was ca. 70 MHz. For Q0 × 1000 the same shift was around 6.5. The maximum signal shift related to NH3 ca. 70 MHz. For Q0 −1 × 1000 the same−1 shift was around 6.5. The maximum signal shift related at 300 °C was for fres ca. 5 MHz and for Q0 × 1000 ca. 2.8 (see Figure 7). The observed maximum effect to NH3 at 300 ◦ C was for f res ca. 5 MHz and for Q0 −1 × 1000 ca. 2.8 (see Figure 7). The observed related to cold start water was for fres 14-times and for Q0−1 three times higher than the maximum NH3 maximum effect related to cold start water was for f res 14-times and for Q0 −1 three times higher response. This demonstrates that the resonance frequency (fres) is much more affected by water than the maximum NH3 response. This−1demonstrates that the resonance frequency (f res ) is much compared to the loss-related value of Q0 . A possible explanation for this effect might be that fres is more affected by water compared to the loss-related value of Q0 −1 . A possible explanation for this mostly affected by polarization effects (please note the high dipole moment of gaseous H2O of 1.84 D effect might be that f res is mostly affected by polarization effects (please note the high dipole moment [47]) compared to Q0−1, which represents the dielectric and conductivity losses. The polar nature of of gaseous H2 O of 1.84 D [47]) compared to Q0 −1 , which represents the dielectric and conductivity the water molecule might cause this big difference. Nevertheless, even when cold start water has the losses. The polar nature of the water molecule might cause this big difference. Nevertheless, even demonstrated huge effect to the RF signals, this effect happens only at much lower temperatures than when cold start water has the demonstrated huge effect to the RF signals, this effect happens only at the SCR typically is operated. When the catalyst has reached its usual operation conditions, no more much lower temperatures than the SCR typically is operated. When the catalyst has reached its usual cold start water is stored on the catalyst and the RF catalyst monitoring is not affected. 5. Conclusions and Outlook For several years, RF-based catalyst state monitoring has been a focus of research and development as the only direct measure of the current NH3 storage on SCR catalysts. Whereas most
Sensors 2017, 17, 1615
12 of 15
operation conditions, no more cold start water is stored on the catalyst and the RF catalyst monitoring is not affected. 5. Conclusions and Outlook For several years, RF-based catalyst state monitoring has been a focus of research and development as the only direct measure of the current NH3 storage on SCR catalysts. Whereas most studies were performed with synthetic exhaust on the laboratory scale using gas test benches, the big step towards application size and real exhaust measurements on an engine dynamometer setup was achieved recently and proved the ability of the RF technique under stationary conditions [38]. It could be demonstrated that a direct urea dosing control on the NH3 storage degree determined by the RF signal is possible. Furthermore, this technique allows precise investigations of the NH3 storage influence to the catalyst performance and NH3 slip. This paper continues this work with focus on the temperature dependency of the RF signal and the NH3 storage behavior. Additionally, the influence of exhaust gas humidity and especially of cold start water was investigated. The observed sample was a well-studied commercial and serial type Cu-exchanged zeolite-based SCR catalyst (Cu-SSZ-13), in contrast to [38], where a mostly unknown serial-type iron containing zeolite-based SCR catalyst was used. The catalyst volume was increased compared to the sample of [38], but still smaller than in common application size and forced the catalyst to operate at very high space velocities. As a first step, the full functionality of the RF system was demonstrated and the linear correlation of both RF signals f res and Q0 −1 and the current NH3 storage was proven for Cu-SSZ-13. Subsequently, the RF signal was used to investigate the NH3 storage influence to the catalyst performance with respect to NOx conversion and possible NH3 slip in a temperature range from 250 to 400 ◦ C. Based on these experiments, an ideal NH3 storage curve as a function of catalyst temperature was developed and showed the same basic behavior as in established control approaches [11]. Furthermore, a full temperature dependent calibration map with fitting functions for the NH3 -free state and the sensitivity of both RF signals to NH3 storage was created. It is noteworthy that the sensitivities of f res and Q0 −1 showed an almost linear increase with temperature. With this calibration functions, a temperature independent NH3 storage determination seems possible and is in focus of the upcoming work. The humidity influence on the accuracy of the RF approach already described in [38] was also confirmed for Cu-SSZ-13. The influence of cold start water was investigated much more in detail, indicating that cold start water leads to a much higher signal than NH3 at lower temperatures. Nevertheless, at SCR active temperatures, the cold start water has already desorbed of the catalyst and has no more impact on the RF signals. All cold and warm start experiments showed a very reproducible behavior and fit well with the results of the stationary operation point in the higher temperature region. The upcoming work will focus on the application of the developed calibration under transient conditions for Cu-SSZ-13. Therefore, different target NH3 storage curves (for example the determined ideal NH3 storage curve) will be applied to investigate their influence to the catalyst performance under more realistic conditions. Additionally, improvements for the accuracy of the RF system to compensate humidity changes by using the known λ value deserves further consideration. We intend to test various current SCR catalysts systems at different catalyst aging states to predict their ability for the RF approach and their aging behavior. In addition, possible effects due to poisoning deserve to be studied. Over all these, the biggest target is still the application on the road with an RF controlled or RF assisted model-based SCR system. Acknowledgments: The authors thank Carsten Steiner for his previous support in developing the RF dyno setup. This publication of this paper was funded by the German Research Foundation (DFG) and the University of Bayreuth in the funding program “Open Access Publishing”.
Sensors 2017, 17, 1615
13 of 15
Author Contributions: M.D., G.H., W.R., P.G. and R.M. conceived the experiments. D.K. and J.V. provided the catalyst samples. M.D. performed the experiments. All together analyzed the data, evaluated the results, and wrote the paper. Conflicts of Interest: The authors declare no conflict of interest.
References 1. 2. 3. 4. 5.
6. 7.
8.
9. 10. 11.
12.
13.
14. 15. 16. 17. 18. 19. 20.
Johnson, T. Review of Vehicular Emissions Trends. SAE Int. J. Engines 2015, 8, 1152–1167. [CrossRef] Koebel, M.; Elsener, M.; Marti, T. NOx -Reduction in Diesel Exhaust Gas with Urea and Selective Catalytic Reduction. Combust. Sci. Technol. 1996, 121, 85–102. [CrossRef] Forzatti, P. Present status and perspectives in de-NOx SCR catalysis. Appl. Catal. A 2001, 222, 221–236. [CrossRef] Koebel, M.; Elsener, M.; Kleemann, M. Urea-SCR: A promising technique to reduce NOx emissions from automotive diesel engines. Catal. Today 2000, 59, 335–345. [CrossRef] Janssens, T.; Falsig, H.; Lundegaard, L.; Vennestrøm, P.; Rasmussen, S.; Moses, P.G.; Giordanino, F.; Borfecchia, E.; Lomachenko, K.; Lamberti, C.; et al. A Consistent Reaction Scheme for the Selective Catalytic Reduction of Nitrogen Oxides with Ammonia. ACS Catal. 2015, 5, 2832–2845. [CrossRef] Brüggemann, T.C.; Keil, F.J. Theoretical Investigation of the Mechanism of the Selective Catalytic Reduction of Nitric Oxide with Ammonia on H-Form Zeolites. J. Phys. Chem. C 2008, 112, 17378–17387. [CrossRef] Borfecchia, E.; Lomachenko, K.A.; Giordanino, F.; Falsig, H.; Beato, P.; Soldatov, A.V.; Bordiga, S.; Lamberti, C. Revisiting the nature of Cu sites in the activated Cu-SSZ-13 catalyst for SCR reaction. Chem. Sci. 2015, 6, 548–563. [CrossRef] Paolucci, C.; Verma, A.A.; Bates, S.A.; Kispersky, V.F.; Miller, J.T.; Gounder, R.; Delgass, W.N.; Ribeiro, F.H.; Schneider, W.F. Isolation of the Copper Redox Steps in the Standard Selective Catalytic Reduction on Cu-SSZ-13. Angew. Chem. 2014, 126, 12022–12027. [CrossRef] Koebel, M.; Elsener, M. Selective catalytic reduction of NO over commercial DeNOx -catalysts: Experimental determination of kinetic and thermodynamic parameters. Chem. Eng. Sci. 1998, 53, 657–669. [CrossRef] Yuan, X.; Liu, H.; Gao, Y. Diesel Engine SCR Control: Current Development and Future Challenges. Emiss. Control Sci. Technol. 2015, 1, 121–133. [CrossRef] Seibel, M.; Hofmann, F.; Christner, B.; Kaiser, A.; Fekete, N.; Weibel, M. Development and Integration Process of Control-Oriented Catalyst Models. In Proceedings of the 6th IAV MinNOx Conference, Berlin, Germany, 22 June 2016. Kumar, A.; Kamasamudram, K.; Currier, N.; Yezerets, A. Effect of Transition Metal Ion Properties on the Catalytic Functions and Sulfation Behavior of Zeolite-Based SCR Catalysts. SAE Int. J. Engines 2017, 10. [CrossRef] Cavataio, G.; Girard, J.; Patterson, J.; Montreuil, C.; Cheng, Y.; Lambert, C. Laboratory Testing of Urea-SCR Formulations to Meet Tier 2 Bin 5 Emissions; SAE Technical Paper 2007-01-1575; SAE: Warrendale, PA, USA, 2007. Guan, B.; Zhan, R.; Lin, H.; Huang, Z. Review of state of the art technologies of selective catalytic reduction of NOx from diesel engine exhaust. Appl. Therm. Eng. 2014, 66, 395–414. [CrossRef] Gao, F.; Wang, Y.; Kollar, M.; Washton, N.; Szanyi, J.; Peden, C. A comparative kinetics study between Cu/SSZ-13 and Fe/SSZ-13 SCR catalysts. Catal. Today 2015, 258, 347–358. [CrossRef] Wang, D.; Yao, S.; Shost, M.; Yoo, J.; Cabush, D.; Racine, D. Ammonia Sensor for Closed-Loop SCR Control. SAE Int. J. Passeng. Cars Electron. Electr. Syst. 2009, 1, 323–333. [CrossRef] Feng, T.; Lü, L. The characteristics of ammonia storage and the development of model-based control for diesel engine urea-SCR system. J. Ind. Eng. Chem. 2015, 28, 97–109. [CrossRef] Ning, J.; Yan, F. Detection of Injected Urea Quantity and Correction for SCR Urea Dosing Control; SAE Technical Paper 2015-01-1038; SAE: Warrendale, PA, USA, 2015. Moser, M.; Carqué, G.; Heß, D.; Schrade, F. Simulationsgestützte Robustheitsanalyse von On-Board-Diagnosefunktionen. MTZ Motortech. Z. 2016, 77, 54–59. [CrossRef] Moos, R.; Beulertz, G.; Reiß, S.; Hagen, G.; Fischerauer, G.; Votsmeier, M.; Gieshoff, J. Overview: Status of the Microwave-Based Automotive Catalyst State Diagnosis. Top. Catal. 2013, 56, 358–364. [CrossRef]
Sensors 2017, 17, 1615
21. 22. 23.
24.
25. 26. 27. 28. 29.
30.
31. 32. 33.
34.
35. 36. 37. 38. 39.
40.
41.
14 of 15
Moos, R. Microwave-Based Catalyst State Diagnosis—State of the Art and Future Perspectives. SAE Int. J. Engines 2015, 8, 1240–1245. [CrossRef] Moos, R.; Rauch, D.; Votsmeier, M.; Kubinski, D. Review on Radio Frequency Based Monitoring of SCR and Three Way Catalysts. Top. Catal. 2016, 59, 961–969. [CrossRef] Beulertz, G.; Fritsch, M.; Fischerauer, G.; Herbst, F.; Gieshoff, J.; Votsmeier, M.; Hagen, G.; Moos, R. Microwave Cavity Perturbation as a Tool for Laboratory In Situ Measurement of the Oxidation State of Three Way Catalysts. Top. Catal. 2013, 56, 405–409. [CrossRef] Beulertz, G.; Votsmeier, M.; Moos, R. Effect of propene, propane, and methane on conversion and oxidation state of three-way catalysts: A microwave cavity perturbation study. Appl. Catal. B 2015, 165, 369–377. [CrossRef] Beulertz, G.; Votsmeier, M.; Moos, R. In operando Detection of Three-Way Catalyst Aging by a Microwave-Based Method: Initial Studies. Appl. Sci. 2015, 5, 174–186. [CrossRef] Feulner, M.; Hagen, G.; Piontkowski, A.; Müller, A.; Fischerauer, G.; Brüggemann, D.; Moos, R. In-Operation Monitoring of the Soot Load of Diesel Particulate Filters: Initial Tests. Top. Catal. 2013, 56, 483–488. [CrossRef] Feulner, M.; Hagen, G.; Hottner, K.; Redel, S.; Müller, A.; Moos, R. Comparative Study of Different Methods for Soot Sensing and Filter Monitoring in Diesel Exhausts. Sensors 2017, 17, 400. [CrossRef] [PubMed] Sappok, A.; Bromberg, L.; Parks, J.E.; Prikhodko, V. Loading and Regeneration Analysis of a Diesel Particulate Filter with a Radio Frequency-Based Sensor; SAE Technical Paper 2010-01-2126; SAE: Warrendale, PA, USA, 2010. Nanjundaswamy, H.; Nagaraju, V.; Wu, Y.; Koehler, E.; Sappok, A.; Ragaller, P.; Bromberg, L. Advanced RF Particulate Filter Sensing and Controls for Efficient Aftertreatment Management and Reduced Fuel Consumption; SAE Technical Paper 2015-01-0996; SAE: Warrendale, PA, USA, 2015. Dietrich, M.; Jahn, C.; Lanzerath, P.; Moos, R. Microwave-Based Oxidation State and Soot Loading Determination on Gasoline Particulate Filters with Three-Way Catalyst Coating for Homogenously Operated Gasoline Engines. Sensors 2015, 15, 21971–21988. [CrossRef] [PubMed] Sappok, A.; Bromberg, L. Radio Frequency Diesel Particulate Filter Soot and Ash Level Sensors: Enabling Adaptive Controls for Heavy-Duty Diesel Applications. SAE Int. J. Commer. Veh. 2014, 7, 468–477. [CrossRef] Moos, R.; Fischerauer, G. Automotive Catalyst State Diagnosis Using Microwaves. Oil Gas Sci. Technol. 2015, 70, 55–65. [CrossRef] Fremerey, P.; Reiß, S.; Geupel, A.; Fischerauer, G.; Moos, R. Determination of the NOx loading of an automotive lean NOx trap by directly monitoring the electrical properties of the catalyst material itself. Sensors 2011, 11, 8261–8280. [CrossRef] [PubMed] Reiß, S.; Schönauer, D.; Hagen, G.; Fischerauer, G.; Moos, R. Monitoring the Ammonia Loading of Zeolite-Based Ammonia SCR Catalysts by a Microwave Method. Chem. Eng. Technol. 2011, 34, 791–796. [CrossRef] Rauch, D.; Albrecht, G.; Kubinski, D.; Moos, R. A microwave-based method to monitor the ammonia loading of a vanadia-based SCR catalyst. Appl. Catal. B 2015, 165, 36–42. [CrossRef] Rauch, D.; Kubinski, D.; Simon, U.; Moos, R. Detection of the ammonia loading of a Cu Chabazite SCR catalyst by a radio frequency-based method. Sens. Actuators B 2014, 205, 88–93. [CrossRef] Rauch, D.; Kubinski, D.; Cavataio, G.; Upadhyay, D.; Moos, R. Ammonia Loading Detection of Zeolite SCR Catalysts using a Radio Frequency based Method. SAE Int. J. Engines 2015, 8, 1126–1135. [CrossRef] Dietrich, M.; Steiner, C.; Hagen, G.; Moos, R. Radio-Frequency-Based Urea Dosing Control for Diesel Engines with Ammonia SCR Catalysts. SAE Int. J. Engines 2017, 10. in press. [CrossRef] Dietrich, M.; Rauch, D.; Porch, A.; Moos, R. A laboratory test setup for in situ measurements of the dielectric properties of catalyst powder samples under reaction conditions by microwave cavity perturbation: Set up and initial tests. Sensors 2014, 14, 16856–16868. [CrossRef] [PubMed] Dietrich, M.; Rauch, D.; Simon, U.; Porch, A.; Moos, R. Ammonia storage studies on H-ZSM-5 zeolites by microwave cavity perturbation: Correlation of dielectric properties with ammonia storage. J. Sens. Sens. Syst. 2015, 4, 263–269. [CrossRef] Rauch, D.; Dietrich, M.; Simons, T.; Simon, U.; Porch, A.; Moos, R. Microwave Cavity Perturbation Studies on H-form and Cu Ion-Exchanged SCR Catalyst Materials: Correlation of Ammonia Storage and Dielectric Properties. Top. Catal. 2017, 60, 243–249. [CrossRef]
Sensors 2017, 17, 1615
42.
43.
44. 45. 46.
47.
15 of 15
Chen, P.; Schönebaum, S.; Simons, T.; Rauch, D.; Dietrich, M.; Moos, R.; Simon, U. Correlating the Integral Sensing Properties of Zeolites with Molecular Processes by Combining Broadband Impedance and DRIFT Spectroscopy—A New Approach for Bridging the Scales. Sensors 2015, 15, 28915–28941. [CrossRef] [PubMed] Chen, P.; Rauch, D.; Weide, P.; Schönebaum, S.; Simons, T.; Muhler, M.; Moos, R.; Simon, U. The effect of Cu and Fe cations on NH3 -supported proton transport in DeNOx -SCR zeolite catalysts. Catal. Sci. Technol. 2016, 6, 3362–3366. [CrossRef] Cavataio, G.; Jen, H.; Warner, J.; Girard, J.; Kim, J.; Lambert, C. Enhanced Durability of a Cu/Zeolite Based SCR Catalyst. SAE Int. J. Fuels Lubr. 2009, 1, 477–487. [CrossRef] Frobert, A.; Raux, S.; Creff, Y.; Jeudy, E. About Cross-Sensitivities of NOx Sensors in SCR Operation; SAE Technical Paper 2013-01-1512; SAE: Warrendale, PA, USA, 2013. Srinivasan, A.; Joshi, S.; Tang, Y.; Wang, D.; Currier, N.; Yezerets, A. Development of a Kinetic Model to Evaluate Water Storage on Commercial Cu-Zeolite SCR Catalysts during Cold Start; SAE Technical Paper 2017-01-0968; SAE: Warrendale, PA, USA, 2017. Haynes, W.M. CRC Handbook of Chemistry and Physics, 92nd ed.; CRC Press: Hoboken, NJ, USA, 2011. © 2017 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
