2005-01-1082
`Hydrocarbon Selective Catalytic Reduction Using a Silver-
`Alumina Catalyst with Light Alcohols and Other Reductants
`
`John F. Thomas, Samuel A. Lewis, Sr., Bruce G. Bunting, John M. Storey, and Ron L. Graves
`Oak Ridge National Laboratory
`
`Paul W. Park
`Caterpillar, Inc.
`
`Copyright © 2005 SAE International
`
`ABSTRACT
`
`BACKGROUND
`
`Previously reported work with a full-scale ethanol-SCR
`system featuring a Ag-Al2O3 catalyst demonstrated that
`this particular system has potential to reduce NOx
`emissions 80-90% for engine operating conditions that
`allow catalyst temperatures above 340°C. A concept
`explored was utilization of a fuel-borne reductant, in this
`case ethanol “stripped” from an ethanol-diesel micro-
`emulsion
`fuel.
`Increased
`tailpipe-out emissions of
`hydrocarbons, acetaldehyde and ammonia were
`measured, but very little N2O was detected. In the
`current increment of work, a number of light alcohols
`and other hydrocarbons were used in experiments to
`map their performance with the same Ag-Al2O3 catalyst.
`These exploratory tests are aimed at identification of
`compounds or organic functional groups that could be
`candidates for fuel-borne reductants in a compression
`ignition fuel, or could be produced by some workable
`method of fuel reforming. A second important goal was
`to improve understanding of the possible reaction
`mechanisms and other phenomena
`that
`influence
`performance of this SCR system. Test results revealed
`that diesel engine exhaust NOx emissions can be
`reduced by more than 80%, utilizing ethanol as the
`reductant for a space velocity near 50,000/h and catalyst
`temperatures between 330 and 490oC. Similar results
`were achieved for 1-propanol, 2-propanol and 1-butanol,
`with a (desirable) shift to a lower temperature range
`seen for the primary alcohols. Heavier alcohols and
`other oxygenated organics were also
`tested as
`reductants showing a range of less successful results.
`Non-oxygenated hydrocarbons and
`the selected
`secondary and tertiary alcohols proved to be very poor
`reductants for this system. Some discussion concerning
`the possible mechanisms behind the results is offered.
`
`The use of hydrocarbons (HC) to reduce diesel exhaust
`NOx emissions via selective catalytic reduction (SCR) is
`potentially a very attractive option for transportation
`applications.
` The exhaust stream is continuously
`oxygen rich under normal conditions and a ready supply
`of hydrocarbons is available on-board. However, the
`HC-SCR option is viewed by many to be less viable than
`lean NOx traps and urea-based SCR technology. This
`reduction
`reported NOx
`view
`is not surprising:
`efficiencies
`for HC-SCR systems are very often
`significantly lower than those achieved with these other
`technologies.1-3 Alumina supported silver (Ag-Al2O3)
`catalysts are among the most promising of HC-SCR
`catalysts
`that have been examined
`in
`the open
`literature.2,3
`
`There are numerous concerns with applying urea-based
`SCR and lean NOx traps to on-road vehicles. For urea-
`SCR these include; 1) need for a separate onboard tank,
`2) infrastructure to supply urea, 3) residue buildup from
`unwanted urea and urea decomposition products,
`especially during inadvertent over-injection or injection at
`low temperatures,4,5 4) corrosiveness of urea, and 5)
`cold weather freezing. Use of lean NOx trap technology
`will require sophisticated controls to produce the needed
`frequency of calibrated fuel-rich pulses. Methods
`include various excursions in the engine operation,
`pulsed fuel injection into the exhaust stream, or both
`methods in combination. Occasional de-sulfurization of
`the lean NOx trap requiring relatively severe conditions,
`will likely be necessary. The fuel penalty for effective
`lean NOx traps may prove to be excessive and the
`required precious metal loading is a cost concern. All of
`the mentioned NOx reduction technologies suffer from
`reduced effectiveness at lower temperatures (~150-
`300oC), which are typical for transportation applications.
`
`Because of the drawbacks for urea SCR and lean NOx
`traps, it would be attractive to develop a HC-SCR
`
`BASF-2007.001
`
`

`
`system that could effectively utilize diesel (compression
`ignition) fuel, reformed diesel fuel, a fuel-borne additive
`or a reformed fuel additive as the reducing agent. As a
`result, investigators continue to pursue development of
`HC-SCR based systems with the hope of developing a
`viable technology.
`
`SILVER-ALUMINA HC-SCR – SCR Catalysts utilizing
`HC reductants in oxygen-rich gas streams have been
`studied for at least two decades. There is a sizable
`body of literature relevant to HC-SCR and to Ag-Al2O3
`catalysts in particular. Most published work has been
`bench-scale research using simulated exhaust. Two
`very notable, broad-based and complementary literature
`reviews were published
`in 2002, giving valuable
`interpretation to results reported by many researchers.
`One review was commissioned by the Coordinating
`Research Council,2 to evaluate the state of SCR
`technology as applied to vehicles, and another was
`carried out by a team at Queen’s University Belfast,3
`which looked closely at fundamental mechanisms. Of
`the great many HC-SCR systems evaluated, certain Ag-
`Al2O3 catalyst formulations have been identified as being
`particularly active and selective,2,3 and therefore may yet
`be promising as a NOx control technology for on-road
`diesel emissions.
`
`Some generalization concerning Ag-Al2O3 catalyst
`performance can be made from the body of previous
`published research. Successful reducing agents include
`heavier paraffins, and certain alcohols and aldehydes.
`Catalyst formulations with 1.2% to 2% Ag are seen to
`lower the temperature at which alumina is active and
`selective.2,3
` Silver
`loadings near 10% can yield
`excessive levels of N2O.3 Some experiments resulted in
`conversion levels greater than 80% and demonstrated
`good resistance to both water and SO2 inhibition,2,3
`qualities needed for diesel application. Sliver sulfate is
`active and resposible for good the performance reported
`(and low poisoning effect) with some reductants in the
`presence of SO2.3 In the presence of water, polar
`oxygenates seem to have quite an advantage. Inhibition
`by water is probably due to competitive adsorption (onto
`catalysts surfaces) between water and one or more key
`reactants. Highly polar oxygenates probably have a
`greater ability to compete with water in comparison to
`non-polar hydrocarbons.2,3
`
`There are also significant hurdles to development of a
`robust Ag-Al2O3 system applicable to on-road diesels.2,3
`Diesel fuel and many components of diesel fuel do not
`appear to be good reductants, especially at relatively low
`temperatures (250-400oC). This leads to fuel-borne and
`fuel-derived/reformed
`reductants
`as
`a
`possible
`approach. Efficient use of reductants is also a likely
`issue, essentially a fuel penalty issue. To the best of our
`knowledge, the durability of Ag-Al2O3 catalysts for diesel
`applications remains unproven.
`
`PREVIOUSLY REPORTED EFFORT - In a preceding
`study,1 a
`full-scale Ag-Al2O3 catalyst ethanol-SCR
`reduction of NOx
`system demonstrated excellent
`
`emissions from a heavy-duty diesel engine for a narrow
`range of conditions.
` For exhaust and catalyst
`temperatures of 350-400oC, NOx conversion exceeded
`90% and 80% for space velocities (SV) of 23,000/h and
`62,000/h respectively. The C/N ratios used to achieve
`these efficiencies were about 4 for the 23,000/h SV
`condition and near 7 for the 62,000/h SV condition. As
`expected, the NOx conversion efficiency was found to
`depend greatly upon the catalyst core temperature.
`When the catalyst temperature approached 250oC, the
`conversion efficiencies fell to near 25% for both SV
`values.
`
`included a proof-of-principle
`This previous study
`demonstration of the fuel-borne reductant concept. A
`relatively simple laboratory method using “mild” vacuum
`distillation, was found to be quite effective for removing
`and collecting ethanol from E-diesel (ethanol-diesel
`micro-emulsion). Subsequently, this ethanol was used
`successfully as a reductant with the engine operation on
`an ethanol-diesel mixture.
`
`to
`is rapidly converted
`that ethanol
`The concept
`acetaldehyde by the Ag-Al2O3 catalyst2,6 was supported
`by the previous investigation.1 Acetaldehyde was
`observed to slip past the catalyst at the 62,000/h SV.
`Ammonia was also produced in measurable quantities,
`and HC slip occurred. The addition of a “clean-up”
`catalyst that oxidizes or utilizes HC and NH3 may be
`warranted for this type of system.
`
`OBJECTIVE OF CURRENT WORK - A primary goal
`guiding this effort was to comparatively examine the
`effectiveness of various reductant candidates with the
`Ag-Al2O3 catalyst, under realistic engine conditions. It
`could be viewed as a (partial) reductant “screening”
`study for this particular catalyst. Interesting reductants
`or reductant “classes” could be examined in more
`comprehensive, follow-on studies. In a closer look at
`performance, the composition of slip HC and nitrogen
`compounds, and the feasibility of the reductant to be
`fuel-borne or fuel-derived would all be of great interest.
`
`A second important goal was to increase understanding
`of chemical mechanisms and other physical processes
`governing the performance and selectivity of this HC-
`SCR system. Observing the relative performance of
`differing organic functional groups and other reductant
`properties values was expected to assist in gaining such
`understanding.
`
`MATERIALS AND EQUIPMENT
`
`EXPERIMENTAL FACILITY - The experimental effort
`was conducted at the Oak Ridge National Laboratory. A
`Cummins 5.9 liter ISB diesel engine (1999 model, 24
`valve, in-line 6 cylinder) was used as the test engine.
`This engine is refitted to be a “near-2004” emissions
`engine, having unique controls/calibration, cooled
`exhaust gas recirculation (EGR),
`fuel system and
`turbocharger. Control of the EGR valve can be
`governed by
`the engine system or switched
`to
`
`BASF-2007.002
`
`

`
`tertiary alcohols responded. The diols were chosen to
`see whether
`there was a benefit
`from a higher
`abundance of OH groups. This effect could be
`confounded because, in contrast to the alcohols, they
`are non-polar compounds with high boiling points.
`Cyclic compounds (Cyclohexane and cyclohexanol)
`were deemed interesting due to differing chemistry and
`their potential abundance in Canadian oil-sand derived
`fuels. An acetate and ketone were chosen to look at
`oxygenates with
`alternative
`functional
`groups.
`Admittedly, one could come up with a very different and
`longer list of compounds to test, with reasonable
`justification. The compounds listed in the lower portion
`of Table 2, were chosen mainly because they are fuels
`or fuel components.
`
`Table 1. Specifications for fuel-grade ethanol supplied
`by Williams-Pekin, Inc.
`
`
`Ethanol content, vol.%
`Methanol content, vol.%
`Denaturant content, vol.%
`Water content, mass%
`
`92.1 min
`0.5 max
`2 min, 5 max
`~0.5
`
`
`
`Table 2. Reductants tested with Ag-Al2O3 catalyst.
`
`
`Reductants used in 50,000/h SV test matrix
`Molecular
`Boiling
`weight
`Point or
`(amu)
`range (°C)
`46.1
`~ 79
`60.1
`97
`60.1
`82
`74.1
`117
`74.1
`83
`102.2
`157
`100.2
`67
`130.2
`196
`62.1
`196
`76.1
`215
`
`
`
`Alcohols
`fuel-grade ethanol
`1-propanol
`2- propanol
`1-butanol
`tert-butanol
`1-hexanol
`cyclohexanol
`1-octanol
`ethylene glycol
`1,3-propanediol
`Other oxygenates
`ethyl acetate
`acetone
`hydrocarbon
`cyclohexane
`
`88.1
`58.1
`
`77
`56
`
`84.2
`
`81
`
`
`Reductants used in miscellaneous tests
`low sulfur diesel fuel
`C9-C20
`185-350
`low sulfur kerosene
`Mostly
`175-325
`C12-C15
`
`
`iso-paraffin mixture
`n-heptane
`
`100.2
`
`190-210
`99
`
`
`
`REDUCTANT INJECTION - The reductant delivery
`system featured a variable-speed dosing pump (Fluid
`Metering, Inc. model RHV 0CTC) to inject reductant into
`an entrainment air stream and then through a spray
`nozzle into the exhaust. The injector was located in a
`
`independent control. The engine was coupled to a
`General Electric direct current motoring dynamometer
`capable of absorbing 224 kW (300 hp).
`
`The HC-SCR system layout and sample locations are
`shown schematically in Fig. 1. Gaseous emissions were
`sampled from the engine-out and catalyst-out raw
`exhaust streams and directed to standard emission
`benches (composed of Horiba Ltd. and California
`Analytical
`Instruments
`analyzers)
`to
`provide
`measurements of NOx, THC, CO, CO2, and O2.
`
`
`
`
`
`Fig. 1. Schematic diagram showing layout of HC-SCR
`components and sampling locations.
`
`
`
`Caterpillar, Inc. provided the 7.0 liter Ag-Al2O3 catalyst to
`ORNL. The catalyst has a cell density of 31 cells/cm2
`(200 cells/in2) and measured 24.1cm (9.5 in) in diameter
`by 15.2 cm (6 in) long. No other catalysts or particulate
`traps were used for this investigation. This catalyst was
`de-greened and tested for over 80 hours in the previous
`study.1
`
`TEST FUELS AND REDUCTANTS - The fuels used to
`operate the engine were BP (formerly ARCO) ECD-1
`and BP-15. Both are high cetane number, low sulfur
`diesel fuels (less than 15 ppm mass sulfur) and are
`viewed as very similar for the purposes of this study.
`The ethanol used in this study was fuel-grade (Williams-
`Pekin, Inc.), meaning it is denatured with gasoline and
`contains a corrosion inhibitor; pertinent specifications
`are listed in Table 1. The other reductants used in this
`work,
`listed
`in Table 2, were chemical-grade
`compounds, with the exception of 2-propanol which was
`70% 2-propanol with 30% water.
`
`Some reasoning behind the 13 reductants chosen for
`the test matrix is offered (upper portion of Table 2). The
`objective was to see if a trend existed going from lighter
`to heavier primary alcohols and how secondary and
`
`BASF-2007.003
`
`

`
`NOx and HC concentrations and other values, so
`progression to an apparent steady state could be
`observed more easily. In this manner a sweep across a
`reasonable range of stoichiometry was performed.
`
`Typical test conditions used for an individual reductant
`are given in Table 3. The gas concentrations are
`representative values, given to show how the exhaust
`environment changes with test condition. The presence
`and concentration of O2 and H2O may change the
`behavior of the HC-SCR system somewhat.2,3,6 Note
`that the catalyst is also exposed to particulate matter
`(PM), but no measurements of PM were made in this
`work. In some cases points between those listed for
`conditions 1-5 were also explored to obtain some data at
`other temperatures. Condition 6 was not run for every
`combination.
`
`
`
`Table 3. Approximate test conditions used to explore
`reductant performance.
`Test
`
`Catalyst inlet
`Condi-
`SV
`Temperature
`tion
`(1/h)
`(°C)
`1
`50K
`260
`2
`50K
`295
`3
`50K
`335
`4
`50K
`390
`5
`50K
`465
`6
`100K
`380
`
`O2
`conc.
`(%)
`13.2
`12.3
`10.6
`8.5
`5.5
`10.5
`
`CO2
`conc.
`(%)
`4.8
`5.4
`6.5
`7.8
`9.8
`6.5
`
`H2O
`conc.
`(%)
`6.5
`7.1
`8.2
`9.6
`11.9
`8.2
`
`
`
`SEPARATION OF FUEL-BORNE REDUCTANTS - A
`very limited number of tests have been performed
`examining how effectively reductants mixed with diesel
`fuel could be removed using a laboratory “mild” vacuum
`distillation method. If the laboratory method worked
`well, it would at least be imply that an on-board device
`could be developed to carry out this function. Results
`show that light alcohols are easily removed by this
`method. More details of these tests are given in the
`Appendix.
`
`
`RESULTS AND DISCUSSION
`
`focus on NOx
`The majority of results presented
`conversion as a function of catalyst core temperature for
`reductant injection at a given C/N ratio. Data taken for
`reductant injection at relatively high C/N values is
`presented first. The objective is to compare reductant
`effectiveness and identify those that demonstrate good
`performance. A more in-depth examination of selected
`reductants at a range of C/N values is offered as well.
`
`GENERAL REDUCTANT SCREENING - The result of a
`test matrix using ethanol as the reductant is shown in
`Fig 2. The best performance is seen at the 388°C
`catalyst inlet exhaust condition. All results are at the
`50,000/h SV condition unless noted otherwise. This
`
`bend in the exhaust about 1 meter from the catalyst
`face.
` An experiment was performed measuring
`reductant dispersion at the catalyst face while injecting a
`number 2 diesel fuel. The face of the catalyst was
`traversed in two perpendicular directions with a probe to
`obtain a concentration map. Results indicated nearly
`constant concentration for both a 28,000/h and 51,000/h
`SV condition. We have made the assumption that the
`(more volatile) reductants used in the current effort will
`also have essentially complete dispersion before
`reaching the catalyst face.
`
`This injection system was calibrated by volume delivered
`as a function of pump motor speed. The system was
`thought to hold calibration reasonably well, even with
`changes
`in
`fluid (reductant) viscosity and modest
`changes in injection air pressure (the back-pressure
`seen by the pump). Calibrations were conducted at
`various times with water, diesel fuel, and ethanol and for
`varying entrainment air pressures (0 to 140 kPa above
`atmospheric pressure).
` The
`results support our
`assumption that the calibration remains valid with these
`changes.
`“Spot checks” of
`the calibration were
`performed periodically to be sure the system was
`working properly.
`
`EXPERIMENTS
`
`EXPERIMENT MATRIX - An experimental matrix was
`reduction
`allow NOx
`developed which would
`performance comparisons of the various reductants over
`an applicable temperature range. A SV value of
`50,000/h was chosen for most data because it is thought
`to be a broadly acceptable value for transportation
`applications.
`
`The guidance for performing the experimental matrix for
`a given reductant is listed below.
`• Space velocity: 50,000/h for most data; an optional
`test at 100,000/h to examine the role of SV.
`• C/N range of at least 0 to 10, vary range as
`applicable. Collect data at several C/N values to
`define a meaningful curve.
`• Engine out NOx concentration near 200 to 240 ppm
`• Catalyst inlet temperature range, 250°C to highest
`achievable with the engine system, ~450-470°C.
`Examine at least 5 temperatures in this range.
`
`
`It was found in practice to be difficult to keep the NOx
`concentration at a constant value over the range of
`temperatures (and at 50,000/h SV) but it could be kept
`within a 200-240 ppm range by adjusting the speed,
`load, and EGR valve position.
`
`The usual method for testing at a given exhaust
`condition and reductant type, was to begin with no
`injection and to progress in discrete steps from a low to
`high injection rate. Data was recorded at a given
`injection rate when a steady-state condition was
`observed.
` The data acquisition system was
`programmed to give real-time traces of temperatures,
`
`BASF-2007.004
`
`

`
`figure depicts the type of data set produced for each
`reductant tested.
`
`Overall results in the form of NOx reduction versus the
`catalyst core temperature are given in Figs 3-6, for C/N
`values of 9-12. The available data with C/N values
`nearest the middle of this range (10.5) were chosen for
`subsequent figures. Variation in the C/N values is due
`to the practicalities of engine operation and reductant
`injection. The range of C/N ratios vary from about 9-12,
`with some variation point to point for a given reductant
`and variation between reductants. This would be quite
`problematic, but this relatively high level of reductant
`injection, only small changes in performance occur over
`C/N values of 9 to 12, as seen in fig. 2. This
`“diminishing
`
`Other alcohols - Performance results for 1-hexanol, 1-
`octanol, tert-butanol and cyclohexanol are given in Fig.
`4. The heavier primary alcohols show significantly less
`NOx reduction compared to the lighter alcohols, except
`at temperatures nearing 250°C where performance
`appears to be about the same. Both tert-butanol and
`cyclohexanol appear to have no value as a reductant
`with this catalyst.
`
`
`
`Fig. 2. Performance of fuel-grade ethanol for 50,000/h
`SV and five catalyst inlet temperatures. A 100,000/h SV
`case is included for comparison.
`
`returns” observation held true for all reductants with the
`exception of ethylene glycol, which behaved rather
`linearly in this range (but showed this diminishing returns
`trend for C/N ≥ 20). The C/N ratio variation adds some
`uncertainty to the comparisons, but we believe the data
`is still highly useful in this form to compare relative
`performance of the reductants. Interpolated data is used
`for the ethylene glycol curve (fig. 5), which was missing
`some data points in the 9-12 C/N range and behaved
`more linearly over this range. A C/N value of 10.5 was
`chosen to be plotted. Catalyst core temperature is
`measured by a small thermocouple in a central channel
`near the geometric center of the monolith.
`
`Selected light alcohols - We found the most effective
`reductants tested are the light alcohols, as depicted in
`Fig. 3. 1-Propanol and 1-butanol both show a desirable
`reduction at
`lower
`toward effective NOx
`shift
`temperatures. It appears that 2-propanol is slightly less
`effective than 1-propanol and butanol. Because of the
`body of data generated in the previous study,1 ethanol is
`a “base case” reductant and included in Figs. 4-6, along
`with 1-propanol which gave very favorable results.
`
`Fig. 3. Performance of light alcohols for 50,000/h SV
`and relatively high C/N ratio (reductant injection rate).
`
`
`
`
`
`Fig. 4. Performance of 1-hexanol, 1-octanol, tert-butanol
`and cyclohexanol compared to ethanol and 1-propanol
`for 50,000/h SV and relatively high C/N ratio.
`
`Diols - Results for two diols, 1,3-propanediol and
`ethylene glycol are summarized in Fig. 5. The 1,3-
`propanediol is seen to be moderately less effective as a
`reductant compared to the light alcohols, although it
`performs as well or better than ethanol at 250-300°C.
`Ethylene glycol appears similar to ethanol and 1,3-
`propanediol at 275°C, but is much less useful above
`300°C.
`
`Other non-alcohols - Figure 6 shows test results for the
`non-alcohol oxygenates, ethyl acetate and acetone,
`
`BASF-2007.005
`
`

`
`which seem to work relatively well as reductants near
`400°C and above. Also shown is a non-oxygenate,
`cyclohexane, which displays essentially no reductant
`capability with the tested system.
`
`A few observations can be made from figs. 7-12.
`Ethanol shows only marginal improvement in NOx
`performance when C/N is increased from 6 to 9. For
`temperatures near 350°C and above, ethanol is the best
`reductant, especially for C/N of 3 and 6. 1-propanol and
`1-butanol are clearly better at lower temperatures over
`the range of C/N values.
`
`Fig. 5. Performance of diols compared to ethanol and 1-
`propanol for 50,000/h SV and relatively high C/N ratio.
`The ethylene glycol data is interpolated to give results
`for C/N = 10.5.
`
`
`
`Fig. 7. Interpolated data for ethanol experiments.
`
`
`
`Fig. 8. Interpolated data for 1-propanol and 2-propanol
`experiments.
`
`
`
`Fig. 6. Performance of ethyl acetate and acetone does
`not compare well to light alcohols, especially at the lower
`end of the temperature range. Cyclohexane shows little
`activity as a reductant.
`
`EXAMINATION OF LIGHT ALCOHOL INTERPOLATED
`RESULTS - The results for the light alcohols will now be
`examined in more detail. Plots of performance at C/N
`values of 3.0, 6.0 and 9.0 were produced by
`interpolation of the raw experimental data. Results for
`ethanol, butanol and propanol are shown in figs. 7, 8
`and 9. Diminishing returns of increased reductant
`injection going from a C/N value of 6 to 9 is evident,
`particularly for the ethanol injection. Comparisons of the
`four alcohols at C/N values of 3, 6 and 9 are shown in
`figs. 10, 11 and 12
`
`Fig. 9. Interpolated data for 1-butanol experiments.
`
`
`
`BASF-2007.006
`
`

`
` The major practical
`the same HC-SCR system.
`difference is the SV and NOx levels were not held at
`50,000/h and 200-240 ppm values used for the main
`body of data. Results for a low sulfur number 2 diesel
`fuel, a low sulfur kerosene, an iso-paraffin mixture and
`fuel grade ethanol are compared in Fig. 13. The
`compounds other than ethanol are rather ineffective as
`reductants. A single test using heptane at 100,000/h SV
`and 350°C exhaust temperature (not shown) gave only a
` Considering
`the
`few percent NOx conversion.
`cyclohexane
`results discussed earlier,
`the non-
`oxygenated reductants tested in this study all gave poor
`results. These potential reductants were alkanes or
`contained a large amount of alkanes compounds, and
`other types of non-oxygenates may give different results.
`
`Fig. 10. Comparison of light alcohols for C/N = 3.0 from
`interpolation of data.
`
`Fig. 11. Comparison of light alcohols for C/N = 6.0 from
`interpolation of data.
`
`
`
`
`
`Fig. 13. Data comparing fuel-grade ethanol to relatively
`heavy hydrocarbon reductants.
`
`FUEL PENALTY FOR ETHANOL USE – Fuel penalty is
`often defined as the reductant consumption rate divided
`by the engine fuel consumption rate, and can be
`expressed in percent by mass or energy units. Fuel
`penalty might be stated for a given engine condition or
`some standard engine test cycle. We offer fuel penalty
`values for injecting ethanol as the reductant in Table 4,
`for the six tested engine conditions used in this study
`(Table 3).
`
`Table 4. Fuel penalty for ethanol injection for a 200
`ppmv exhaust NOx concentration and a 10:1 C/N ratio.
`
`
`Fuel
`Ethanol
`Ethanol
`Test
`Engine
`consump-
`Mass
`Energy
`Condi-
`power
`tion
`Penalty
`Penalty
`tion
`(kW)
`(g/s)
`(%)
`(%)
`1
`36
`2.85
`6.4
`4.0
`2
`44
`3.23
`5.7
`3.6
`3
`58
`3.89
`4.7
`3.0
`4
`73
`4.70
`3.9
`2.5
`5
`94
`5.90
`3.1
`2.0
`6
`95
`6.92
`5.3
`3.3
`
`
`
`BMEP
`(kPa)
`345
`452
`631
`827
`1186
`841
`
`
`
`
`
`Fig. 12. Comparison of light alcohols for C/N = 9.0 from
`interpolation of data.
`
`OTHER RELATED EXPERIMENTAL EFFORTS – Some
`data is available from separate, but related efforts using
`
`BASF-2007.007
`
`

`
`DISCUSSION OF REDUCTANT PERFORMANCE –
`Some explanations and conjecture can be offered
`addressing
`the hierarchy
`in performance among
`reductants tested.
`
`Aldehyde formation - There is experimental evidence
`that ethanol, and 1-propanol undergo oxidation to form
`acetaldehyde and propionaldehyde1,6 and it is then likely
`that 1-butanol also forms the corresponding aldehyde.
`The aldehydes, which are also good reductants, break
`down further as part of the reduction process.1,6,7 It is
`proposed that 2-propanol forms acetone6 which then
`breaks down further. We note that 2-propanol was quite
`superior as a reductant compared to acetone, especially
`at low temperatures, so this explanation may not be fully
`satisfactory. In forming either an aldehydes or ketone,
`the alcohol donates two H atoms, which presumably
`enhance in the overall reduction process. Tert-butanol
`would not be expected to form an aldehyde or a ketone
`and proved to be relatively unreactive for the tested
`system.
`
`(and perhaps obvious)
`- A general
`Reactivity
`observation can be made based on molecular stability,
`simply that reductants that react or break down easily
`are likely to create “usable” reactive species, particularly
`at low temperatures. This might explain ethyl acetate
`and acetone looking like reasonable reductants at ~
`400°C, but not at low temperature, where they remain
`relatively stable. There was some expectation that the
`cyclohexanol would have
`reactivity, and behave
`somewhat like hexanol or a secondary alcohol. Instead,
`cyclohexanol appeared stable and unreactive with the
`tested system.
`
`Reactivity indications - Evidence of the (net) oxidation of
`reductants can be inferred from the measured CO2, CO
`and HC levels and the temperature difference between
`the catalyst inlet and the catalyst core. The net
`reactions occurring appear to be quite exothermic.
`Unfortunately the CO2 measurement is dominated by the
`engine-out values (~5-10%) and the increase derived
`from the reductants is about 0-2500 ppm in the range of
`interest. Furthermore the flame ionization detector for
`HC measurement used
`in
`this work gives useful
`information, but has a response that varies widely for
`many of the species likely being detected, and the actual
`HC slip species are not necessarily known. It is not
`possible to compare and interpret the CO2 and HC
`readings with confidence. However, a rise in CO and
`CO2 is expected for the compounds that decompose and
`oxidize along with a relatively low HC reading, and the
`opposing trends are expected for compounds that are
`unreactive.
`
`Analysis of the CO2 “rise” data for he C/N values of 9-12
`examined earlier, gave somewhat crude and scattered
`results, but we report a few trends that were seen. The
`poorest
`performing
`compounds,
`cyclohexane,
`cyclohexanol and
`tert-butanol, showed virtually no
`detectable CO and CO2 formation except at the highest
`temperature condition (see Table 3.) where
`it
`is
`
`estimated 15-30% of the injected carbon ended up as
`CO and CO2. These compounds also gave consistent
`and high HC readings (accounting for ~68-87% of the
`injected carbon, depending on the reductant) for the
`lower temperature conditions (conditions 1-4 in Table 3)
`with a modest drop in HC value for the highest
`temperature condition
`(condition 5
`in Table 3).
`Cyclohexanol was only observed to decompose at the
`highest temperature point, and when a high injection
`rate was held for about 15 minutes as the catalyst
`temperature rose from 477 to 495°C. Measured CO2
`increased and HC reading decreased as might be
`expected. All other reductants gave much higher
`values for CO + CO2 production, with increasing values
`for increasing temperature, and the opposing trend for
`the HC emissions. Ethylene glycol stood out as having
`the highest propensity to react to form CO + CO2, at all
`temperatures (~ 80 % at the lowest temperature, and
`rising to ~ 100% at the highest temperature), followed by
`1,3-propandiol and ethyl acetate. Ethylene glycol also
`displayed the highest degree of exothermic activity for
`the low temperature tests.
`
`It has been
`Polar compounds, water solubility -
`proposed that a distinct advantage is possessed by the
`more polar oxygenates which can compete successfully
`with water for adsorption sites.2,3 The environment of
`interest has abundant water which doubtlessly affects
`the catalytic process. This property again favors the
`light alcohols and light asymmetric oxygenates. Note
`that the non-polar diols tested do have very high water-
`solubility, and may be less disadvantaged compared to
`low water-solubility compounds. Hexanol and octanol
`notably have lower water solubility than the lighter
`alcohols. The non-oxygenated compounds have very
`low solubility.
`
`Molecular mobility - The ability of the compound to
`diffuse to make intimate contact with the catalyst surface
`and then be mobile on the surface, could affect the SCR
`process. This mobility property could be related to the
`molecular weight, boiling point (listed in Table 2) and
`other properties of the compound. No attempt to
`quantify this concept or property is offered. Indirect
`evidence of some sort of physical interference process,
`probably involving carbonizing (coking) of the reductant
`on the catalyst surface, was seen with octanol and
`compounds of higher molecular weight. The observation
`was that as spray injection quantity was increased, NOx
`conversion began to decrease and would then slowly
`decrease with time at a given spray rate.
`
`- A key
`LOW TEMPERATURE PERFORMANCE
`technical challenge for lean NOx trap and SCR systems
`applied to diesel transportation is effectiveness at low
`catalyst temperatures; the 150-300°C range will serve
`for the purpose of this discussion. The Ag-Al2O3 SCR
`system will need to have reasonable effectiveness in this
`temperature
`range
`to have viability
`for on-road
`applications for the 2007-2010+ emission requirements.
`
`BASF-2007.008
`
`

`
`The tested system did show > 50% NOx reduction at
`260-270°C for 1-propanol, 1-butanol, 1-hexanol, 1-
`octanol for C/N of 9 or below as shown in Fig. 14. Fig 3.
`results imply that 1-propanol and 1-butanol will have a
`steep performance drop as the temperature is dropped
`below the rang