Académique Documents
Professionnel Documents
Culture Documents
Gregory Griffin
Navistar, Inc.
ABSTRACT I. INTRODUCTION
The Environmental Protection Agency (EPA) issued emission
Diesel exhaust aftertreatment systems are required for
legislation for diesel engine vehicles. EPA year 2010
meeting both EPA 2010 and final Tier 4 emission regulations
emissions are currently required for on-highway vehicles
while meeting the stringent packaging constraints of the
while meeting final Tier 4 emission standards for off-
vehicle. The aftertreatment system for this study consists of a
highway vehicles will be required in 2014. The feasibility of
fuel dosing system, mixing elements, fuel reformer, lean NOx
a fuel reformer, lean NOx trap (LNT) and selective catalytic
trap (LNT), diesel particulate filter (DPF), and a selective
reduction (SCR) aftertreatment system to meet a final Tier 4
catalytic reduction (SCR) catalyst. The fuel reformer is used
emissions standard of 0.4 g/kW-hr NOx and EPA 2010
to generate hydrogen (H 2 ) and carbon monoxide (CO) from
emission standards of 0.2 g/bhp-hr NOx has been addressed
injected diesel fuel. These reductants are used to regenerate [1, 2]. This paper addresses the simulation optimization and
and desulfate the LNT catalyst. NOx emissions are reduced vehicle packaging of an aftertreatment system meeting EPA
using the combination of the LNT and SCR catalysts. During 2010 emission regulations on a heavy-duty engine platform.
LNT regeneration, ammonia (NH 3 ) is intentionally released Detailed simulation results are provided for vaporizing and
from the LNT and stored on the downstream SCR catalyst to mixing diesel fuel upstream of a fuel reformer catalyst.
further reduce NOx that passed through the LNT catalyst. Multiple packaging options are provided using the simulation
This paper addresses system packaging and exhaust flow results for heavy-duty vehicle installations.
optimization for heavy-duty line-haul and severe service
applications. Various packaging options are provided for both The aftertreatment system consists of a fuel dosing system,
applications. Exhaust flow and fuel vaporization optimization mixing elements, fuel reformer, LNT, diesel particulate filter
was conducted using computational fluid dynamics followed (DPF), and SCR catalyst arranged in series. A sample
by test cell and vehicle test results. Performance metrics were configuration of this system is shown in Figure 1. Similar
met including 97% uniformity for fuel vaporization and 97% systems using LNT and SCR technology without a fuel
uniformity for exhaust flow distribution having less than 2% reformer are cited in literature [3, 4]. A fuel reformer is
fuel pulse broadening. The optimized system met the utilized to generate hydrogen (H2) and carbon monoxide
packaging constraints and EPA 2010 emission regulations. (CO) using injected diesel fuel. These reductants are used to
regenerate and desulfate the LNT catalyst. NOx emissions are
reduced using the combination of the LNT and SCR catalysts.
During LNT regeneration, ammonia (NH3) is intentionally lower level (i.e., 3.2% in the example). A higher fueling rate
released from the LNT and stored on the downstream SCR is commanded to consume the remaining oxygen as depicted
catalyst to further reduce NOx that passed through the LNT by the tailpipe oxygen dropping to zero. This higher rate
catalyst. This approach converts the drawbacks of the single creates a rich condition where reformate is produced, as
leg LNT approach (low conversion during regeneration and depicted by the predicted CO and H2 profiles. The reformate
NH3 slip) into an advantage while remaining independent of is key to regenerating and desulfating the downstream LNT.
any urea infrastructure since diesel fuel is the only reductant
[1, 5,6,7,8,9]. The DPF is regenerated using the fuel reformer The fuel reformer operation is optimized by creating a
similar to a diesel oxidation catalyst by injecting fuel under uniform fuel and exhaust flow distribution upstream of the
lean exhaust operating conditions. fuel reformer catalyst. Previous works have shown that the
fuel distribution upstream of the fuel reformer is proportional
to the downstream temperature distribution when the fuel
reformer is operated in lean mode [12]. The fuel distribution
is key during transitions from lean to rich mode operation. A
uniform fuel distribution produces a uniform downstream
temperature profile, while a non-uniform fuel distribution can
result in non-uniform richness levels exiting the reformer.
This paper will define key a metric for the fuel distribution
that will be referred to as epsilon (ε). Likewise, the exhaust
flow distribution metric referred to as gamma (γ) and the fuel
pulse broadening metric referred to as psi (ψ) will be defined.
The simulation results will be provided that meet the required
targets for efficient operation of the fuel reformer.
Figure 1. Aftertreatment system layout
The simulation optimization and packaging developed in this
work were used to demonstrate heavy-duty aftertreatment
emissions that met EPA 2010 levels. The fuel dosing system
used in this work provides a smaller drop size distribution
than previous works [12] and is described in recent literature
[13]. The aftertreatment results from this work were applied
to meeting EPA 2010 levels and were also described in recent
literature [14]. The simulation results were applied to a 13L
heavy-duty diesel engine. The engine was a 2010
development engine rated at 475 hp with peak torque of 1750
ft-lb. The engine was calibrated for two separate engine NOx
levels. As a result, two different aftertreatment systems were
procured for the demonstration of meeting EPA 2010 tailpipe
emission levels of 0.2 g/bhp-hr NOx. The first aftertreatment
system was sized for an engine out NOx level of 1.0 g/bhp-hr
while the second aftertreatment system was sized for an
engine out NOx level of 0.5 g/bhp-hr. Although two different
aftertreatment systems were procured, the fuel injection
system, fuel vaporization and fuel reformer catalysts
remained the same for both systems. As a result, the fuel
Figure 2. Typical reformer operation [1] vaporization simulation applied to both aftertreatment
systems.
systems [12]. The fuel injectors employed a recirculating aftertreatment system downpipe. A general description of the
cooling loop from the fuel tank to an external electrical engine operating conditions and its affect of fuel vaporization
pump, which funneled diesel fuel to the injectors and back to is provided. The effect of exhaust temperature and mass flow
the fuel tank. The fuel would be supplied from the low rate on fuel vaporization and mixing is described. The key
pressure gear pump for serial production. engine operating conditions used for CFD are described
followed by a description of the exhaust flow uniformity (γ),
fuel vapor uniformity (ε) and fuel pulse broadening (ψ).
These steps were followed to optimize the exhaust downpipe
prior to procuring hardware for testing.
The final performance metric is fuel pulse broadening (ψ), These three metrics, γ, ε , and ψ, are not equally important
which is expressed as the percentage of total injected when evaluating a particular design. Achieving the ε target is
hydrocarbon mass for a single LNT regeneration event that given the highest importance. It has been shown through
reaches the inlet to the reforming catalyst after the controls experimental methods that meeting the ε target is more
system has ended the injection pulse. This is a measure of the important than meeting the γ target [15]. Meeting the γ target
integrity of the hydrocarbon injection pulse as it travels is given the second highest weighting. Achieving the ψ target
downstream from the point of injection to the inlet of the is the least important of the three metrics discussed.
reforming catalyst. It is important during rich hydrocarbon
injection to achieve the targeted fuel/air ratio. Excessive fuel There are also other metrics that are used to guide the CFD
pulse broadening will cause the actual fuel pulse reaching the design process. The velocity ratio and γ values at the inlet
reforming catalyst to be a longer event achieving a lower than plane of the reforming catalyst are used to quantify system
targeted fuel/air ratio. The target is to keep ψ below 2.0% of performance. The velocity ratio is the ratio of the lowest
injected fuel reaching the reforming catalyst after the velocity magnitude divided by the average velocity
injection pulse has been terminated. This value was chosen to magnitude in a given plane normal to the exhaust flow. The
prevent the fuel reformer from operating in both locally lean velocity ratio varies between zero (worst case) and unity (best
and rich modes as the fuel was shut off. case). The γ value at the face of the reforming catalyst is also
analyzed, along with other planes upstream of the reforming
Figure 7 gives a graphical representation ψ. The integral catalyst. Pressure drop caused by design features is also
under the commanded liquid hydrocarbon injection curve considered when optimizing a system with the goal of
between t1 and t2 is equivalent to the total amount of maintaining the system back pressure to meet the
hydrocarbons injected into the exhaust pipe. The integral performance targets [14].
under the hydrocarbon 1″ inside the reforming catalyst curve
from t2 to t∞ is equivalent to the amount of hydrocarbon The B25 and B100 operating points were used to drive the
vapor that reaches the reforming catalyst after the CFD optimization for the ProStar vehicle. The design was
hydrocarbon injection event has terminated. The commanded optimized to meet γ (0.97), ε (0.97) and ψ (< 2.0%) targets.
liquid hydrocarbon injection curve is defined as I(t) and the
hydrocarbon vapor 1″ inside the reforming catalyst curve is III. RESULTS
defined as V(t). Equation 3 defines ψ as the ratio of the I(t)
This section provides simulation results for two system
integral relative to the V(t) integral.
designs for a ProStar vehicle along with a packaging design
for the WorkStar vehicle. The simulation results for the
ProStar vehicle include a design containing a long inlet cone
to the fuel reformer. This configuration was used for a testing
demonstration showing compliance to EPA 2010 emission
standards [14]. A compact packaging design was developed
for the ProStar vehicle showing simulation results similar to
the long inlet cone. The compact packaging results were
extended to a WorkStar vehicle for severe service
applications that require compact packaging.
(3)
The geometry of the optimized inlet cone is described. The Figure 12. Side entry packaging concept
length of the inlet cone was approximately 480 mm (19″).
There were approximately 150 additional millimeters (6″) of
length between the inlet of the cone and the centerline axis of Five different lengths of the reformer can were designed
the vertical section of the exhaust pipe (section in which the based on the packaging space available and the desire to
upstream mixer is located from Figure 12). This creates a shorten the packaging length. The maximum length of the
total horizontal distance of approximately 630 mm (25″) reformer can was set at approximately 760 mm (30″), such
between the axis of the vertical section of the exhaust pipe that the inlet of the reforming catalyst was at the same
and the inlet face of the reforming catalyst. location as in the cone inlet design. This distance is greater
than the 630 mm (25″) length described in the previous
simulation was found to also be 0.99. The ψ value for the regions of low velocity indicate a separate swirl about each of
B100 simulation was found to be 0.5% those low velocity regions. The zero swirl design resulted in
both swirls being equally strong, which reduced the γ value at
the inlet face of the reforming catalyst. However, the
swirl_1/4 design exhibited a strong swirl in one direction, and
a weak swirl in the opposite direction. Indication of these
counter rotating swirls can be most clearly seen from the
reformer face velocity contour plot of the swirl_1/4 design in
Figure 18. The low velocity region at the eleven o'clock
position at the periphery of the reformer can indicates that
there was not a unidirectional swirl present along the wall of
the reformer can. This low velocity region at the outer most
radial position is the location where the two counter rotating
swirls meet. Although the γ value 25.4 mm (1″) inside the
Figure 16. Transient simulation of 350 mm (14″) reformer was higher for the swirl_0 geometry (0.99) than it
reformer can at B100 was for the swirl_1/4 geometry (0.98), the swirl_1/4
geometry was selected as the best performing design. Both of
these designs met the γ target of 0.97. Further data analysis
The next step of the design process involved optimizing the showed that the γ values for cross section planes at the face of
geometry of the inlet to the reformer can. The baseline side the reformer and upstream of the reformer were higher for the
entry can design had the exhaust gas entering tangentially to swirl_1/4 geometry when compared to the swirl_0 geometry.
the reformer can as shown in Figure 17. This design This suggested the swirl_1/4 inlet geometry would aid in
maximizes the amount of swirl induced immediately further reduction in system length if needed in the future.
upstream of the reformer. The swirl is caused by the exhaust
flow entering the reformer can at an offset to the center axis
of the reformer can. A large offset between the axis of the
entering exhaust pipe and the axis of the reformer will induce
a higher velocity swirl than will a small offset. Steady state
simulations were conducted on the five different exhaust pipe
inlet offset distances to optimize the amount of swirl
generated in the reformer can. All five of these steady state
simulations were run at the B25 operating condition. The best
performing design was the 1/4 swirl geometry.
Figure 18. Exhaust gas velocity contour plots for the five
swirl designs at B25 operating condition
Figure 17. Reformer can swirl optimization All work discussed through this section for the side inlet
design focused on the design selection for the reformer can
length and the geometry of the side inlet. The baseline design
Figure 18 shows velocity contour plots and γ values for the showed that the ε target was met for both the B25 and B100
five different swirl designs at the B25 operating condition. operating conditions, and the ψ target was met for B100.
The 1/2, 3/4, and full swirl geometries all had a dominant, Optimizing the side inlet swirl position improved the γ value
unidirectional swirl. The unidirectional swirl can be identified for the B25 operating condition to within the design
by the low velocity point in the center of the reformer can and specification. The next step in the CFD plan was to optimize
higher velocities seen around the entire periphery. The zero the mixer style and location to achieve the target γ value for
and swirl_1/4 geometries both showed twin, counter rotating the B100 operating condition and to improve the ψ value at
swirls. This can most easily be seen by looking at the 254 the B25 operating condition.
mm (10″) axial position plot of the swirl_0 design. The two
Table 3 summarizes the results of four different mixer design M2 were all higher than the same metrics for the
configurations evaluated at the B25 operating condition baseline design at this operating point, indicating that the M2
utilizing steady state simulation. The baseline design and M2 design was better suited for the higher exhaust gas mass flow
design both used the same mixer, but at different locations conditions present at the B100 operating point when
within the exhaust pipe. The location of the mixer in the compared to the B25 operating point.
baseline design is similar to the position of upstream mixer
shown in Figure 8, while the mixer is in the vertical section Table 4. γ Optimization through Mixer Design at B100
of the exhaust pipe (Figure 11) for the M2 design. The M1 Operating Condition
design simulation did not contain a mixer to assess whether
the presence of a mixer is actually detrimental to the γ value.
The M3 design used a different style mixer in the same
location as the mixer in the M2 design. The baseline design,
M1, and M2 all met the γ target of 0.97 at 25.4 mm (1″)
inside the reformer. γ values upstream of 25.4 mm (1″) inside
the reformer and the velocity ratio at the inlet to the reformer
were then analyzed for the baseline, M1, and M2
configurations. M3 design did not meet the targeted γ 25.4
mm (1″) inside the reformer, but it did show the highest γ The result of the mixer study showed the baseline
values for the other three planes upstream of 25.4 mm (1″) configuration was the better performing geometry for low
inside the reformer. Design M3 also showed the highest exhaust gas mass flow rate conditions (B25) and the M2
velocity ratio value. Failure to meet the γ criteria for 25.4 mm configuration was the better performing geometry for high
(1″) inside the reformer for design M3 will not allow this exhaust gas mass flow rate conditions (B100). The duty cycle
mixer configuration to be selected for this design, but it does of the engine for this application was analyzed and it was
show promise if further reduction in system length is required determined that the engine would operate at exhaust gas mass
on a future application. The baseline design was selected as flow rate conditions similar to the B25 operating point more
the best overall performing mixer configuration for the B25 often than the B100 operating point. This supported the
operating condition. decision to select the baseline mixer configuration for the
final design. Selecting the baseline design also minimized the
Table 3. γ Optimization through Mixer Design at B25 risk of missing the required ε target value because the
Operating Condition baseline simulation utilizing the same mixer configuration
had already been shown to be meeting the ε target. Both of
these reasons supported the decision to move forward with
the baseline mixer configuration.
The final step in the design process for the side entry design
was to run transient simulations to validate the design. Table
5 details the results of the transient simulation results
conducted at the A25, B25, B50, B100, and C50 operating
conditions. The ε target was achieved at all five operating
points, resulting in a five mode cycle average of 0.99. The γ
Next, steady state simulations were evaluated with the target 25.4 mm (1″) inside the reformer was met for the five
baseline, M2, and M3 configurations at the B100 operating mode average. These γ values differ slightly from the
condition. The configuration without a mixer (M1) was not previously presented values from steady state simulations.
simulated at the B100 condition for two reasons. First, it did The γ values in Table 5 are an average γ over the entire
not show a performance improvement over the baseline hydrocarbon injection pulse for each operating point. This
design when simulated at the B25 condition. Secondly, resulted in the B100 point underperforming the γ metric. The
previous work showed that the presence of the right mixer ψ values for all the points except B100 are greater than the
aids in achieving the target ε values [16]. Removing the target value of 2.0%, but the five mode cycle average value
mixer would risk lowering the ε values while gaining no for ψ is 2.1%. The side entry design achieved higher γ and ε
performance increase with respect to γ 25.4 mm (1″) inside values than inlet cone design. The ψ value for the side entry
the reformer. design was slightly lower than what was simulated with the
inlet cone design. A reduction in packaging length of 410 mm
Table 4 shows the results of the three mixer optimization (16″) was quantified for the side entry reformer can design
steady state simulations carried out at the B100 operating when compared to the inlet cone design.
condition. The baseline design met the γ target of 0.97 at 1
inch inside the reformer. The metrics listed in Table 4 for
Table 5. Final Performance Results for Side Inlet Design Figure 20 shows an alternative packing option to fit the SCR
and Cone Inlet Design underneath the vehicle with the other catalysts. The air tank
was relocated to position the SCR underneath the vehicle as
is common for this vehicle application. These two packaging
configurations from Figures 19 and 20 show that the
aftertreatment system can be packaged onto a class 8 day cab
configuration vehicle in multiple ways to meet the needs for
each application.
meeting the required fuel vaporization and mixing targets for 9. Hu, H., Reuter, J., Yan, J., and McCarthy, J., Jr.,
the fuel reformer catalyst. “Advanced NOx Aftertreatment System and Controls For
On-Highway Heavy Duty Diesel Engines,” SAE Technical
Paper 2006-01-3552, 2006, doi:10.4271/2006-01-3552.
ACKNOWLEDGMENTS
10. Dalla Betta, R., Cizeron, J., “Pre-combustors for internal
The authors would like to acknowledge the support of Eaton combustion engines and systems and methods therefore,”
Corporation and Eaton's customer colleagues, that helped Publication Number US7240483, Publication Date:
make this work possible. The authors would like to 07/10/2007
acknowledge their engine partner, Navistar, especially Vadim
Strots, Jim Yan, Ed Derybowski, Jeff Kelso, Martin Keller, 11. Dalla Betta, R., Cizeron, J., Sheridan, D., Davis, T., “Use
Mike Miller and Brad Adelman for supporting the vehicle of a Diesel Fuel Processor for Rapid and Efficient
packaging. The authors acknowledge the support from Eaton Regeneration of Single Leg NOx Adsorber Systems,” 2003
teammates: Andrew Marougy for his work managing one of Diesel Engine Emissions Reduction Conference, Aug. 24-28,
the simulation programs and for supporting the vehicle test 2003, Newport, RI.
program; Erik Dykes for program management; Tim 12. McCarthy, J. Jr. “Simulation Of Fuel-Exhaust Mixing in
Korhumel, Mark Scheffer, Jim Ridge, Judy Wright and Exhaust Aftertreatment Systems Using Simplified Spray
Eaton's canning partner for the CAD design function; Scott Models For Pulse-Width Modulated Fuel Injectors,” Journal
Magno and Eaton's canning partner for procuring the of Atomization and Sprays, Vol. 19, Issue 5, 2009, p.
aftertreatment system; Tom Stover, Vishal Singh and David 425-444.
Yee for providing program support.
13. Bamber, D., Ambrose, S., and McCarthy, J., Jr., “Fuel
Injector Optimization For Diesel Aftertreatment Systems
REFERENCES Coupled With Exhaust Aftertreatment System Performance
1. McCarthy, J., Jr., Korhumel, T., Jr., and Marougy, A., on a Heavy-Duty Diesel Engine Powered Vehicle,” SAE Int.
“Performance of a Fuel Reformer, LNT and SCR J. Commer. Veh. 3(1):111-129, 2010, doi:
Aftertreatment System Following 500 LNT Desulfation 10.4271/2010-01-1940.
Events,” SAE Int. J. Commer. Veh. 2(2):34-22, 2009, doi:
14. Dykes, E.C., Ngan, E.C., McCarthy, J.E., Jr., and Strots,
10.4271/2009-01-2835.
V., “Aftertreatment System Performance of a Fuel Reformer,
2. U.S.E.P.A. (United States Environmental Protection LNT and SCR System Meeting EPA 2010 Emissions
Agency), Federal Register - CFR Part 69, p. 39216-39219, Standards on a Heavy-Duty Vehicle,” SAE Int. J. Commer.
2004. Veh. 3(1):130-142, 2010, doi:10.4271/2010-01-1942.
3. Schommers, J. and Breitbach, H., “A Global Emission 15. Armanini, B.R. and McCarthy, J.E., “Fuel Vaporizer
Strategy - Bluetec for Diesel Engines,” Auto Technology, Catalyst Enables Compact Aftertreatment System Packaging
Pages 48-51, Mar. 2007. by Reducing Mixing Length,” SAE Technical Paper
4. Snow, R., Dobson, D., Hammerle, R., and Katare, S., 2010-01-1070, 2010, doi:10.4271/2010-01-1070.
“Robustness of a LNT-SCR System to Aging Protocol,” SAE 16. Kulkarni, M.S., Mohanta, L., McCarthy, J., Jr., “Spray
Technical Paper 2007-01-0469, 2007, doi: Modeling for Lean NOx Trap Aftertreatment System
10.4271/2007-01-0469. Design,” SAE Technical Paper 2009-28-0016, 2009, doi:
5. McCarthy, J., Jr. and Holtgreven, J., “Advanced NOx 10.4271/2009-28-0016.
Aftertreatment System Performance Following 150 LNT
Desulfation Events,” SAE Technical Paper 2008-01-1541, CONTACT INFORMATION
2008, doi:10.4271/2008-01-1541. PhilipWWetzel@Eaton.com
6. Dykes, E.C., “NOx Performance of an LNT+SCR System
Designed to Meet EPA 2010 Emissions: Results of Engine
Dynamometer Emission Tests,” SAE Int. J. Commer. Veh.
1(1):327-337, 2008, doi:10.4271/2008-01-2642.
7. Chimner, C., “Transient On-Road Emission Reduction of
an LNT + SCR Aftertreatment System,” SAE Int. J. Commer.
Veh. 1(1):315-326, 2008, doi:10.4271/2008-01-2641.
8. McCarthy, J., Jr. and Taylor, W., III, “LNT+SCR Systems
for Medium-Heavy Duty Application: A Systems Approach,”
13th Diesel Engine-Efficiency and Emission Research
Conference, August 13-16, 2007, Detroit, MI.