Control of Urea SCR Systems for US Diesel Applications

HAL Id: hal-01937425https://hal.archives-ouvertes.fr/hal-01937425

Submitted on 28 Nov 2018

HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.

Control of Urea SCR Systems for US Diesel ApplicationsM. van Nieuwstadt, D. Upadhyay

To cite this version:M. van Nieuwstadt, D. Upadhyay. Control of Urea SCR Systems for US Diesel Applications. Oil &Gas Science and Technology - Revue d’IFP Energies nouvelles, Institut Français du Pétrole, 2011, 66(4), pp.655-665. �10.2516/ogst/2011104�. �hal-01937425�

https://hal.archives-ouvertes.fr/hal-01937425

https://hal.archives-ouvertes.fr

Control of Urea SCR Systemsfor US Diesel Applications

M. van Nieuwstadt and D. Upadhyay

Ford Motor Company, Dearborn, MI 48121 - USAe-mail: [email protected] - [email protected]

Résumé — Commande des systèmes de RCS à l’urée destinés aux applications Diesel US — Cetarticle a pour objet de qualifier les défis en matière de commande des systèmes de réduction catalytiquesélective (RCS) dans une solution à base d’urée, pour la conversion des NOx au sein de groupesmotopropulseurs Diesel. Nous montrons l’importance de la maîtrise du stockage d’ammoniac ainsi queles difficultés à l’estimer face aux incertitudes du système. Les dynamiques de stockage lentes ducatalyseur de RCS s’avèrent constituer un facteur de limitation majeur quant aux performances dessystèmes de RCS au cours d’un fonctionnement transitoire. Enfin, nous montrons l’effet de l’incertitudedes capteurs sur les performances du système et comment les capteurs de NOx peuvent être utilisés enassociation avec une corrélation d’entrée pour distinguer une déviation de NOx d’un glissementd’ammoniac.

Abstract — Control of Urea SCR Systems for US Diesel Applications — This paper sets out to qualifychallenges in the control of urea selective catalytic reduction systems for NOx conversion in Dieselpowertrains. We show the importance of ammonia storage control, and the difficulties in estimating it inthe face of system uncertainties. The slow storage dynamics of the SCR catalyst are shown to be a majorlimiting factor on the performance of SCR systems in transient operation. Lastly we show the effect ofsensor uncertainty on system performance and how NOx sensors can be used in conjunction with inputcorrelation to distinguish NOx slip from ammonia slip.

Oil & Gas Science and Technology – Rev. IFP Energies nouvelles, Vol. 66 (2011), No. 4, pp. 655-665Copyright © 2011, IFP Energies nouvellesDOI: 10.2516/ogst/2011104

E-COSM'09 - IFAC Workshop on Engine and Powertrain Control, Simulation and ModelingE-COSM'09 - Colloque IFAC sur le contrôle, la simulation et la modélisation des moteurs et groupes motopropulseurs

IFP Energies nouvelles International ConferenceRencontres Scientifiques d’IFP Energies nouvelles

ogst110002_vanNieuwstadt.qxd 14/10/11 10:35 Page 655

http://ogst.ifp.fr/articles/ogst/abs/2011/04/contents/contents.html

http://ogst.ifp.fr/articles/ogst/abs/2011/04/contents/contents.html

http://ogst.ifpenergiesnouvelles.fr

http://www.ifpenergiesnouvelles.fr

Oil & Gas Science and Technology – Rev. IFP Energies nouvelles, Vol. 66 (2011), No. 4

1 INTRODUCTION

This paper gives an overview of issues in the control of ureabased Selective Catalytic Reduction (SCR) systems, with afocus on issues encountered in US Diesel powertrains. Sincethe issues are rooted in physical phenomena, many of theissues are not unique to US applications. Since US emissionsstandards are more stringent than the standards in other partsof the world, and since aftertreatment layouts for US truckspose more challenges than light duty aftertreatment layouts,these issues pose a bigger challenge for US applications.

The emission standards imposed by 2010 US legislationforce the use of NOx aftertreatment on the vast majority ofUS diesel applications. Some applications employ lean NOx

traps. Lean NOx traps can work for light duty passenger vehi-cles, such as the VW Jetta (Jetta, 2009), and applications cer-tified against a higher TP standard such as the Dodge Ram.Lean NOx traps struggle to achieve good NOx conversion onhigh temperature cycles, such as the US heavy duty transientcycle. For these reasons, US diesel powertrains predomi-nantly use urea SCR for NOx aftertreatment. Products that arecurrently available include the Mercedes GL320, the BMW335d, and the Audi Q7. In 2010 Ford, GM and Chrysler areexpected to come out with SCR applications on their dieseltrucks, certified against ULEV 2 standards.

1.1 NOx Emission Standards

2010 US emissions legislation comprises a drastic reductionin tailpipe NOx emissions. While the 2007 heavy dutystandard allowed 1.2 g/bhphr, the 2010 standard only allows0.2 g/bhphr. The light duty standard goes from tier 2 bin 8(0.2 g/m) to tier 2 bin 5 (0.05 g/m), see (EPA, 2009). Whilethe emissions standards are not expected to be tightenedfurther in 2013, the OBD threshold multiplier will be reducedto values around 2x to 3x depending on the exhaust species.

1.2 Organization of the Paper

The paper will start with an overview of the componentsused in urea SCR systems, the exhaust configuration, thebasic chemistry, and issues associated with the components.We will then proceed to the main focus of this paper, whichis an evaluation of what are the hard and easy problems in thecontrol of urea SCR systems. We use a two brick SCR con-figuration to illustrate the various points. We discuss theimportance of ammonia storage, its control, and observers forammonia storage. We quantify the effect of uncertainty onthe performance of ammonia storage observers. In the lastsection we discuss NOx sensors and how they can be used totrade off NOx conversion against ammonia slip for 2010diesel applications.

2 BASICS OF THE SELECTIVE CATALYTIC REDUCTION SYSTEM

2.1 Exhaust Layouts

A typical diesel aftertreatment system to meet 2010 USemissions standards consists of a Diesel Oxidation Catalyst(DOC), a Diesel Particulate Filter (DPF) and a NOx catalyst.The latter can be a lean NOx trap, or a urea SCR catalyst. Inthis paper we focus uniquely on the SCR. The DOC, DPF,and SCR can be combined in a variety exhaust system con-figurations. The DOC is usually first, to allow heating of theaftertreatment system, in startup, and for DPF regeneration.The SCR can be second, or last in the exhaust system. Bothlayouts have advantages for different applications.

Putting the SCR before the DPF (Fig. 1) allows a quickerlight-off for the same fuel economy penalty, or the samelight-off time with a lower fuel penalty. This is of importancefor cycles that are heavily weighted toward cold start, such asthe federal transient EPA75 light duty cycle. It is of lessimportance for cycles that are more heavily weighted towardhot start, such as the federal heavy duty transient cycle.Quick light-off is also of more importance for US truckapplications, where the aftertreatment system is typicallymounted underbody, far away from, the engine, which makesit harder to get heat into the aftertreatment system.

Putting the SCR after the DPF (Fig. 2) makes the systemharder to light off, but it offers the advantage that the SCRwill not be contaminated with soot, and that the DPF will tosome degree be regenerated by the higher NO2 concentra-tions in the feed gas into the DPF. This also has the disad-vantage of decreasing the NO2 available for the fast SCRreaction that relies on equal proportions of NO and NO2 tobe present in the feedgas, Equation (2).

656

DOC

SCR

DPF

Figure 1

Light duty diesel exhaust layout.


2.2 Ammonia Source

The urea SCR catalyst needs ammonia (NH3) to convertNOx. The source of ammonia can be liquid or solid. Allsystems currently in high volume production use a solutionof aqueous urea as their ammonia source. A eutectic ureasolution consists of 33% urea and 67% water. The advantageof a liquid ammonia source is that it is easy to handle. Twodisadvantages are a poor NH3 to mass ratio, and a risk of ureadeposit formation in the exhaust system.

Among solid ammonia sources are ammonia carbamatepellets, described in great detail by Fulks (2009). Anothersolution is proposed by Amminex, and stores NH3 in a stron-tium chloride salt. The ammonia is released from the solid byheating, and transported to the exhaust pipes in the gas phase.Solid ammonia sources typically have an NH3 to mass ratio 3times better than liquid ammonia sources. Problems withsolid ammonia sources are the transport to the exhaust gas,and the dosing accuracy. Dosing accuracy is important inadaptive control strategies that try to adjust ammonia supplyto optimize a NOx conversion- ammonia slip trade-off.

The urea in the aqueous urea solution is hydrolyzed toammonia as follows:

CO(NH2)2 + heat => HNCO + NH3 (1)HNCO + H2O + heat => NH3 + CO2

This requires heat and possibly mixing of the exhaust gas.In systems that are thermally challenged, there is a substan-tial risk that not all urea is hydrolyzed to ammonia, and formsdeposits in the exhaust system. Once deposit formation hasstarted, subsequent deposits are formed at an accelerated rate,due to the thermal insulation of metal exhaust parts by theurea layer. The urea deposits can be removed by exposure to

heat, such as occurs during DPF regeneration. Since it is hardto model the rate of urea deposit formation, and removal byregeneration imposes a fuel economy penalty, it is best toprevent the deposit formation all together, by limiting theurea injection for cold exhaust temperatures. Directionally,urea injection is started around 180°C, and limited to avoiddeposit formation up to about 250°C.

2.3 NOx Conversion

The ammonia reacts with NOx via a multitude of reactionssteps, the most important reactions are:

(2)

The first reaction is called the standard reaction, and ispredominant specifically for low cost DOC formulations thatdo not generate much NO2, and at low temperatures. Whilesome manufacturers design their aftertreatment system torely on the NO2 in the feedgas to the SCR, we will assumehere that the NO2 concentration is low, and that the standardreaction is dominant. The second reaction is called the fastreaction since its reaction rate is the highest. The third reac-tion is of theoretical interest in automotive applications, sincethe fast reaction takes out all NO2.

The DOC that is typically first in the exhaust system hasthe side effect of converting some NO to NO2, hence enhanc-ing the fast reaction. This reaction is difficult to model in theface of different oxygen and hydrocarbon concentrations,hydrocarbon speciation, and DOC aging.

3 ISSUES IN UREA SCR CONTROL

It is important to realize that the SCR system is extremelygood at converting NOx, provided it is operated under theright conditions. In particular, if the ammonia storage levelon the catalyst is known, if the temperature is in a reasonablewindow, and if the feedgas NOx concentration is known, it isquite easy to get good NOx conversion. The capability of acontrol strategy can therefore not be evaluated by comparingNOx conversion over an emissions cycle with known condi-tions. The difficult requirements of a urea SCR control strat-egy are to trade off NOx conversion against ammonia slip,and to do so in the face of uncertainty. Among the majornoise factors affecting system performance are:– NH3 storage;– NOx concentration in the feed gas;– urea input (injector uncertainty as well as urea concentration);– drive cycle;– sensor uncertainty.

4 4 4 6

2 2 3

8

3 2 2 2

3 2 2 2

NH NO O N H O

NH NO NO N H O

NH

+ + → +

+ + → +

33 2 2 26 7 12+ → +NO N H O

M van Nieuwstadt and D Upadhyay / Control of Urea SCR Systems for US Diesel Applications 657

DOC

SCR

DPF

Figure 2

Heavy duty diesel exhaust layout.



We will demonstrate the challenges by simulations on atwo brick SCR system, depicted in Figure 3. Unless explic-itly specified otherwise, the results are from simulations onthis system. Experimental results are used at times to rein-force some points. The first SCR has a volume of 7 liters, thesecond SCR has a volume of 3 liters. Each SCR is modeledas a one dimensional system, with 7 slices. The model isparametrized with chemical properties of a cupper zeolitebased SCR catalyst. A detailed description of the model canbe found in (Kim, 2007). The ammonia sensor between thetwo bricks will be used in Section 3.4 to study the design ofobservers.

The feedgas data used are from a Ford V8 engine operatedover the EPA75 cycle. The intent behind the two brick systemis firstly to ascertain the effect of SCR volume on NOx con-version and ammonia slip. Subsequent simulations will showNOx conversion and ammonia slip after the first and after thesecond brick. Secondly, the first brick can be considered asthe control target, whereas the second brick can be inter-preted as an ammonia slip control brick, used to reduce theammonia slip caused by thermal excursions. In an optimizedsystem, the bricks would have different washcoat formula-tions and or loadings, matched to their temperature exposure.

3.1 Ammonia Storage

NOx conversion on SCR catalysts is to a large degree determinedby the amount of stored ammonia, especially at low tempera-tures, see Figure 4. Figure 5 shows that SCR catalysts canstore an enormous amount of ammonia. While we can controlthe increase in stored ammonia by injection of urea, we haveno direct control action to decrease stored ammonia.Ammonia can only be released from the SCR by oxidation,thermal desorption or reaction with NOx. Oxidation cannotbe directly controlled, thermal desorption results in largeammonia concentrations of the tailpipe exhaust gas, and withfeedgas NOx concentrations typical of 2010 diesel applica-tions, it is hard to remove ammonia at high rates. A typicalfeedgas NOx number is on the order of 0.5 to 1.0 g/mile.Typical ammonia storage is on the order of 1 g/L, so for a 10liter catalyst, total ammonia stored is 10 grams. Assumingthe standard reaction in Equation (2), 10 grams of ammonia

reacts with 10 × 46/17 ≈ 25 grams of NOx. It then takesbetween 12 and 25 miles to deplete the 10 grams of ammoniastored on the SCR. Thermal transients from normal drivingoperate on a much faster time scale than this. The challengeis to control ammonia storage to a level where it is highenough to achieve good NOx conversion, but not so high thatit will lead to ammonia slip under thermal transients. Due tothe slow ammonia storage dynamics one must be careful totune the ammonia storage solely to achieve good emissionsperformance over transient drive cycles. This typically leadsto undesired behavior off-cycle. Figure 5 shows ammoniastorage as function of temperature. Above temperatures of550°C or so, storage decreases to negligible levels. The catalyst

658

DOC

Urea injectionNH3 sensor

DPFSCR 1 SCR 2

Figure 3

Two brick SCR system used as a case study.

500 550450400350300250200150100

2.0

0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

Temperature (°C)

NH3 storage capacity, 75 ppm

NH3 storage capacity, 350 ppm

Figure 5

Ammonia storage in g/L as a function of temperature, fordifferent gas phase ammonia concentrations.

0.2 1.60.4 0.6 0.8 1.0 1.2 1.40

100

0

10

20

30

40

50

60

70

80

90

Stored ammonia (g/L)

NOx conversion [pct], NH3: NOx = 0.7

NOx conversion [pct], NH3: NOx = 1.1

Figure 4

NOx conversion efficiency (%) as a function of storedammonia in g/L, for different ammonia to NOx ratios at 200°C.


can store more ammonia for higher gas phase ammonia con-centrations. Figure 4 shows that NOx conversion is to alarge degree determined by ammonia storage, especially atlower temperatures, and that NOx conversion increases to alesser degree with increasing ratio of gas phase ammoniato NOx.

3.2 Simple SCR Controllers

The previous section showed the large contribution ofammonia storage to NOx conversion. To set the stage weinvestigate the performance achieved when we rely exclu-sively on initial NH3 storage, without any urea injected duringthe drive cycle. Figure 6 shows that we can achieve reason-ably good NOx conversion (55% for the 7 liter catalyst, 75%for the 10 liter catalyst), but at the expense of high ammoniaslip (150-200 ppm).

A slightly more advanced control strategy is a feed forwardonly strategy, which injects urea proportionally to the NOxgoing into the catalyst, to a 1:1 stoichiometric ratio.

Figure 7 shows that with this strategy and an initialammonia storage of 0.08 g/L, we can achieve 88% NOxconversion, at the expense of 25 ppm ammonia slip. Figure 8shows that with 2.18 g/L initial storage the NOx conversiongets even better, 97%, but this happens at the expense of apeak ammonia slip of 5 981 ppm.

Figure 9 shows that the second hill in the EPA75 cycle isthe condition most prone to ammonia slip, due to the fastthermal transient resulting from the steep acceleration in thatpart of the cycle.

Table 1 summarizes the trade off between NOx conversionand ammonia slip for the strategies considered so far.

TABLE 1

NOx conversion and ammonia slip for different urea injection strategies

Strategy Initial storage No. conversion NH3 slip

g/L pct ppm

No urea injection 0.6 62 75

FF, low storage 0.08 88 25

FF, high storage 2.18 97 5 981


0.2 0.3 0.4 0.5 0.6 0.7 0.80

20

40

60

80

100NOx conversion and NH3 slip as a fcn of NH3 storage (g/L)

0

50

100

150

200

Eff. after SCR1 (%)

Eff. after SCR2 (%)

Max. NH3 slip after SCR1 (ppm)

Max. NH3 slip after SCR2 (ppm)

0.2 0.3 0.4 0.5 0.6 0.7 0.8NH3 stored (g/L)

Figure 6

NOx conversion and ammonia slip over an EPA75 cycle, fordifferent initial storage levels, with 0 urea injection duringthe cycle.

200 180016001400120010008006004000

25

0

5

10

15

20

Times (s)

NOx pre SCR1 (g)

NOx post SCR1 (g)

NOx post SCR2 (g)

Figure 7

Integrated NOx mass before, between and after the 2 SCRbricks from Figure 3. Initial NH3 storage is 0.08 g/L.

200 180016001400120010008006004000

25

0

5

10

15

20

Times (s)

NOx pre SCR1 (g)

NOx post SCR1 (g)

NOx post SCR2 (g)

Figure 8

Integrated NOx mass before, between and after the 2 SCRbricks from Figure 3. Initial NH3 storage is 2.18 g/L.

ogst110002_vanNieuwstadt.qxd 14/10/11 10:35 59


3.3 Ammonia Storage Control

In Upadhyay (2006) we presented the following simplifiedcontrol oriented model of an SCR catalyst. The three statesrepresent gas phase NOx concentration, ammonia storage andgas phase ammonia concentration. The reaction rates Rxxx

(xxx = ADS, DES, OX, RED) follow an Arrhenius equationwith a strong temperature dependence, where ADS standsfor adsorption, DES stands for desorption, OX stands foroxidation, RED stands for NOx conversion. The controlinput U is the ammonia from injected urea, the feedgasNOx concentration is disturbance d:

See Equation (3)To give and example of storage control, Figure 10 shows

the experimental performance of an ammonia storage controllerbased on this model, on a 5 liter catalyst over an EPA75cycle. After the initial transient resulting from a start at zeroammonia storage, the controller can control the ammonia

storage level to within around 0.1 g/L. This controllerachieves between 92 and 98% conversion efficiency onrepeated hot FTP74 cycles.

To study the issues involved in ammonia storage wesimulate a storage controller on our simulation depicted inFigure 3. A simple PI controller on ammonia storage is addedto the stoichometric feed forward controller of Section 3.2. Inthis simulation we control to the average ammonia storage ofthe different longitudinal slices in the first brick. Clearly, wecannot measure this quantity in a real system, but the assump-tion is useful to point out some fundamental limitations ofstorage control. Figure 11 shows that even if we know theammonia storage level, the slow bandwidth of the storagedynamics force a oscillation around the storage target, with aperiod of around 20 minutes. This behaviour is qualitativelyvery similar to the experimental data in Figure 10.

It should be noted here that the stored ammonia is acalculated quantity, and as such the feed forward could be

660

�

�

�

C

C

C

VR F

R

cat

REDOX

NO

NH

NO

θ

θ

3

⎡

⎣

⎢⎢⎢

⎤

⎦

⎥⎥⎥

=

− + +( )VV

R C R R C RR

cat

SC

ADS DES RED OXADS

θ

θ− + + +( ) +

Θ NH NO3 ΘΘSC

cat

ADSDES

cat

C

C

VR F

R

V

NH

NH

3

3 1− − +( ) +

⎡

⎣

⎢

( )θ θ

⎢⎢⎢⎢⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥⎥⎥⎥⎥

+

⎡

⎣

⎢⎢⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥⎥⎥

0

0

F

V

U

cat

++

⎡

⎣

⎢⎢⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥⎥⎥

F

V

dcat

0

0

(3)

0 200 400 600 800 1000 1200 1400 1600 1800

6000

5000

4000

3000

2000

1000

0

Times (s)

NH3 in (ppm)

NH3 SCR1 out (ppm)

NH3 SCR2 out (ppm)

Figure 9

NH3 concentrations before, between and after the 2 SCRbricks from Figure 3, corresponding to the NOx traces inFigure 8.

200 400 600 800 1 000 1200 14000

3.0

0

2.5

2.0

1.5

1.0

0.5

Times (s)

W839 Fresh DOC/SCR/DPF, VERL Warm 1FTP74, 07Dec2006 - NH3 Storage Control

urd_NH3_mstor_des_scr

urd_calc_NH3_stor_scr_gm

g

Figure 10

Performance of an ammonia storage controller. Experimentaldata. The red trace is desired storage, the blue line is achievedstorage.


adjusted to exactly compensate for the disturbance caused byoxidation, ammonia desorption and NOx conversion. Thereare issues with this approach:

– as noted before we can only control an increase instored ammonia. If the desired storage is above the real(modelled) storage, we have no active control authority todrive the storage to the desired level;

– we only have approximate knowledge of the storedammonia, and definitely only an inaccurate knowledgeof the stored ammonia in each slice. Including the dis-turbance terms in the feed forward can easily lead toovercompensation;

– there is an advantage in keeping the excursions around thestorage setpoint symmetric. Uncertainty in the model duringoverstored periods tends to be compensated by the sameuncertainty during understored periods. Long termadaptive strategies that are superposed on ammoniastorage feedback control count on a similar duration ofoverstored and understored periods to ascertain the effectsof an adaptive adjustment.

This section demonstrated limitations of ammonia storagecontrol if it can be measured. In real applications, of course,it cannot, and must be observed from sensor measurements.

3.4 Observers for Ammonia Storage

Figure 3 depicts an ammonia sensor between the twobricks. This configuration has been discussed in Herman(2009). The ammonia sensor can be used for storage feed-back. The second brick is used to catch ammonia slippedfrom the first brick and increase overall NOx conversion. Wehave little control over the thermal transients that driveammonia desorption, so it is important to catch ammonia slipbefore it reaches the tail pipe. If the ammonia sensor wereplaced after the second brick, it would be too late to counter-act ammonia slip by the time it was measured. In this sectionwe study the use of the ammonia sensor to observe the storedammonia in the first brick. We will investigate observers ofdifferent levels of complexity. First we can add an observerterm to the three state model proposed in the beginning ofSection 3:

See Equation (4)


ˆ

ˆ

ˆ

ˆ( ˆ

�

�

�

C

C

C

VR

cat

RED

NO

NH

NO

θ

θ

3

⎡

⎣

⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥

=

− ++ +

− + +

FR

V

R C R R C

OX

cat

SC

ADS DES RED

) ˆ

ˆˆ ˆ

θ

θΘ NH NO3

++( ) +

− − +(

RR

C

C

VR F

OXADS

SC

cat

ADS

Θˆ

ˆ( ˆ )

NH

NH

3

3 1 θ )) +

⎡

⎣

⎢⎢⎢⎢⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥⎥⎥⎥⎥

+

⎡

⎣R

V

F

VDES

cat

catθ

0

0

⎢⎢⎢⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥⎥⎥

+

⎡

⎣

⎢⎢⎢⎢⎢⎢

⎤

⎦

⎥⎥⎥⎥⎥⎥

+U

F

V

dcat

0

0

0

KK C C

03 3

⎡

⎣

⎢⎢⎢

⎤

⎦

⎥⎥⎥

−( )NH NH

�(4)

0 1600 1800140012001000800600400200

1600 1800140012001000800600400200

0

0.4

0.3

0.2

0.1

0Times (s)

-50

150

100

50

0

200fbk (mg/s)

ff (mg/s)

Urea in (mg/s)

Des storage (g/L)

SCR1 avg storage (g/L)

Figure 11

Ammonia storage control on the simulation system of Figure 3.

0 1600 1800140012001000800600400200

1600 1800140012001000800600400200

0

0.4

0.3

0.2

0.1

0Times (s)

0

80

60

40

20

100

Est. storage (g/L)


Measured NH3 slip post SCR1 (ppm)

Predicted NH3 slip post SCR1 (ppm)

Figure 12

Observed and modelled storage, if the observer has access tothe exact ammonia storage dynamics.



Figure 12 shows that we can observe the stored ammoniawith high accuracy if the observer has access to the fullmodel.

To assess the performance of an observer, we need toevaluate it against modelling error and sensor uncertainty. Toinvestigate the effect of modelling error, we propose the fol-lowing simplified observer, that only relies on a single statefor the ammonia storage dynamics:

(5)

The NOx concentration CNO is solved by setting the firststate equation of Equation (4) to 0, which is tantamount toassuming that the NO conversion happens on a much fastertime scale than the ammonia storage and release. The ammo-nia slip function f(θ, T) is modelled as shown in Figure 13.These curves were generated for an inlet NO concentration of300 pm, an inlet NO2 concentration of 0 ppm, an inlet ammo-nia concentration of 300 ppm, and at 300 kg/h mass flow, for4 different temperatures between 100 and 400°C. This Figureshows that for very high temperatures, ammonia slip occurseven at low storage levels, whereas at low temperatures, thereis very little ammonia slip even for high storage levels. TheEPA75 emissions cycle operates at lower temperatures,below 300°C, where there is a weak correlation betweenammonia storage and ammonia slip.

Figure 14 shows that the weak correlation between storageand slip results in a systematic underestimation of ammonia

ˆˆ

ˆ ˆ�θ θ= − + + +( )

+

ΘSC

ADS DES RED OX

AD

R C R R C R

R

NH NO3

SS

SC

C K C C C f TΘ

ˆ ˆ (ˆ, )NH NH NH NH3 3 3 3+ −( ) =

�θ

storage, until we get substantial ammonia slip in bag 3. Assoon as ammonia slip reaches significant levels, the observercan catch up and takes values of similar order of magnitudeas the real storage. Note that the observed and measuredammonia slip levels are very close, due to the high gains inthe observer and the monotonic relation between ammoniastorage and slip. This is not really a sign of good perfor-mance of the observer. Note that due to underestimation ofammonia storage, the PI controlled on storage increases stor-age level, until ammonia slip occurs in bag 3. The simulationshown in Figure 14 achieved a NOx conversion of 90.2% atthe expense of an ammonia slip level of 212 ppm after thesecond brick. Note that Figure 14 only shows the ammoniaslip after the first brick, which is substantially higher thanthat observed after the second brick.

The observer performance can be improved by decreasingthe gain, and by intentionally overestimating the ammoniaslip function f(θ, T) in Equation (5). This amounts to shiftingthe curves in Figure 13 up. Figure 15 shows the improvedperformance of the ammonia storage observer resulting fromthese measures. The slip after the first brick is less than inFigure 14, the NOx conversion after the second brick is90.8%, and the ammonia slip is reduced from 212 to 36 ppm.

Next we investigate the effect of a bias in the ammoniasensor. Figure 16 shows the ammonia slip and NOx conversionas a function of the multiplicative sensor error. It is seenthat since a multiplicative error still correctly predicts theonset of ammonia slip, the observer is quite robust againstmultiplicative errors. Figure 17 shows the ammonia slipand NOx conversion as a function of the additive sensor error.

662

2.50

500

0

50

100

150

200

250

300

350

400

450

Stored NH3 (g/L)0.5 1.0 1.5 2.0

NH3 Slip (ppm), 100°C




Figure 13

Simplified ammonia slip model for single state observer. Thisgraph depicts the function f(.,.) in Equation (5).

0 1600 1800140012001000800600400200

1600 1800140012001000800600400200

0

1.5

0Times (s)

0

800

0.5

1.0

200

400

600

Est. storage (g/L)


Predicted NH3 slip post SCR1 (ppm)

Measured NH3 slip post SCR1 (ppm)

Figure 14

Estimated and real NH3 storage. Predicted and measured NH3

slip after the first brick. Around 1500 s, the feedback actionresults in significant NH3 slip and the observer can catch upwith the real storage.


A positive additive error causes the observer to overestimateammonia storage, and hence the controller to reduce ureainjection. Specifically, if the sensor bias is greater than theammonia slip corresponding to the desired storage level, thestorage controller will completely turn off urea injection and

result in zero NOx conversion and ammonia slip for largeradditive errors.

The above account points out a fundamental limitation inobserver design for this system. While we can take action toachieve robustness against multiplicative uncertainty, both inmodelling and sensor, it is hard to achieve robustness againstadditive uncertainty, both in modelling and sensor, especiallyif this uncertainty corresponds to a magnitude of the desiredlevel of ammonia storage.

4 NOx SENSORS

The previous section discussed the use of an ammonia sensorfor observation of the ammonia storage. Since ammoniasensors are not available for large volume applications in2010, current US diesel applications need to rely on a NOxsensor. In this section we will discuss some tools to use aNOx sensor to detect ammonia slip. Since NOx sensors areheated to above 600°C, any ammonia in the exhaust gas willbe oxidized on the sensor element, and converted to NOx.Hence NOx sensors are also sensitive to ammonia. The sensi-tivity depends on the protection tube design, the exhaust tem-perature, and the microstructure of the diffusion barrier in thesensor. Since the NOx sensor is not designed to sense ammo-nia, the sensitivity to ammonia shows a large variability frompart to part and over life time.

The NOx sensor is sensitive to both NO and NO2, withdifferent gains. NO2 is a larger molecule, hence experiences alarger resistance to transport through the diffusion barrier.Additionally NO2 accounts for two oxygen atoms once it isdissociated in the sensing chamber. Depending on the sensordetails the sensitivity to NO2 can be between 70% and 130%of that to NO.

4.1 Measuring Ammonia with a NOx Sensor

Since the NOx sensor is sensitive to both ammonia and NOx,one cannot a priori determine which combination of thesespecies causes a nonzero sensor reading. We propose a corre-lation scheme with the urea injection input to ascertain whichspecies is registered, (Van Nieuwstadt, 2003). Suppose weare running in steady state conditions, where the SCR cata-lyst is in a regime of good NOx conversion. If we increase theurea injection and see a decrease in output, we can infer thatwe were sensing NOx, since the increased ammonia in thefeedgas reacted with NOx and reduced the sensor reading. Ifon the other hand we see an increase in sensor output whenwe increase urea input, we can infer that we were sensingammonia, since the extra ammonia in the feedgas passedthrough the SCR and was sensed by the NOx sensor. Bycorrelating the change in input uhp to the change in output yhp,we can construct a metric M that is positive when we measureammonia, and negative when we measure NOx.


0.8 1.40.7Gain

0

100

0.9 1.0 1.1 1.2 1.3

20

40

60

80

NOx conversion and NH3 slip, as a fcn of NH3 sensor gain error

NOx conv., SCR2 (pct)

NH3 slip, SCR2 (ppm)

Figure 16

NOx conversion and ammonia slip as a function ofmultiplicative sensor error.

Figure 17

NOx conversion and ammonia slip as a function of additivesensor error.

0 1600 1800140012001000800600400200

1600 1800140012001000800600400200

0

1.0

0Times (s)

0

150

0.8

0.6

0.4

0.2

50

100

Predicted NH3 slippost SCR1 (ppm)

Measured NH3 slippost SCR1 (ppm)

Des storage (g/L) Est. storage (g/L) SCR1 avg storage (g/L)

Figure 15

Top: estimated and real NH3 storage. Bottom: predicted andmeasured NH3 slip after the first brick. Compared to Figure14, this simulation uses lower gain, and overestimates theammonia slip.

10-15Offset set

050-5-10

2040

60

80100

NOx conversion and NH3 slip, as a fcn of NH3 sensor offset error

Nox conv., SCR2 (pct)

NH3 slip, SCR2 (ppm)



(6)

This metric can be enhanced by including the SCR reactiondynamics and correlating the measured change in outputymeas,hp to a modelled change in output ymod,hp:

(7)

The performance of this metric is shown in Figure 18. Theengine is operated in steady state, and a multiplicative squarewave perturbation is applied to the urea injection quantity,which is biased low by 30%. It can be seen that the metric Mincreases and then reaches steady state. After application ofan appropriate scaling map, the metric value can be mappedto correct the injector by +30%. Note the slow time scale ofthis data set. It takes about half an hour to learn this 30% bias.This is again due to the slow storage dynamics of the SCRwhen operated at temperatures where conversion efficiency isgood.

We investigate the system bandwidth by simulating asinusoidal urea injection perturbation and measuring the ratioof the output amplitude to the input amplitude. Figure 19shows that at 300°C, the system gain is below 0.01 for fre-quencies above 0.03 Hz (that is a period of 30 seconds). Asexpected, the system gain is lower for the 10 liter catalyst(the two bricks in Fig. 3) than for the 7 liter catalyst (the firstbrick only). Since we cannot expect the SCR catalyst to stayin steady state optimal conversion conditions for several peri-ods, the applicability of this algorithm is deemed limited atlower temperatures. Figure 20 shows that the system gainincreases to order of magnitude 1 at 560°C. At higher tem-peratures we have seen that the ammonia storage capacitydecreases, and hence the correlation algorithm in Equations(6, 7) could work.

At lower temperatures, where the SCR normally operates,we need to resort to more complex methods of averagingover a long horizon and correlating with the output to deter-mine whether we are slipping ammonia or NOx.

M y y dtmod hp meas hp= ∫ , ,

M u y dthp hp= ∫

CONCLUSIONS AND OUTLOOK

This paper does not claim to present the best possible controlstrategy for urea SCR systems. It does not advertise any par-ticular control design methodology. Rather it points out somefundamental issues that need to be understood to design andevaluate a urea SCR control strategy. Key is to understandthat it is easy to get good NOx conversion with known NOxand urea input, and initial storage. Major challenges are theuncertainty in these parameters, especially in the face of partto part variability and aging over lifetime.

664

Figure 18

Input excitation and correlation metric M, for an injector biasof – 30%.

2 000 2 500150010005000

1.5

1.0

0.5

0

Time (s)

Injector blased –30%

Corr._ratioExc._ratio

Figure 19

Simulated Bode plot for the SCR catalyst at 300°C, for twodifferent catalyst volumes.

SCR system gain at T = 300°C

0.450.400.350.300.250.200.150.100.05 0.500

0.030

0

0.005

0.010

0.015

0.020

0.025

Freq. (Hz)

Vscr = 6.9167 L

Vscr = 9.881 L

Figure 20

Simulated Bode plot for the SCR catalyst at 560°C, for twodifferent catalyst volumes.

0.45 0.500

0.8

0

0.7

Freq. (Hz)0.400.350.300.250.200.150.100.05

0.6

0.5

0.4

0.3

0.2

0.1

Vscr = 6.9167 L

Vscr = 9.881 L

SCR system gain at T = 560°C


We show that ammonia sensors can be used to observeammonia storage, but that conventional observer designs aresensitive to sensor offset.

We show that NOx sensors can be used in conjunctionwith a correlation with the injected urea, to differentiate NOx

from ammonia at the tail pipe.One recurring theme is the slow storage dynamics of the

SCR at normal operating temperatures. While high ammoniastorage is beneficial for NOx conversion, especially at lowtemperatures, it makes controlling the SCR system quitecomplicated. It is the root cause for the slow oscillationswhen controlling ammonia storage. It is the reason thatammonia sensor based observers are sensitive to sensor offset.It is also the reason one cannot employ a simple input-outputcorrelation with input excitation. At this point we are notaware of any developments that would allow similar NOx con-version with substantially lower ammonia storage robustlyover all drive cycles and varying operating conditions.

ACKNOWLEDGMENTS

We thank P. Laing, J. Kim and the rest of their team for thedevelopment of the SCR model used for all simulations inthis paper. We thank C. Lambert, G. Cavataio and the rest oftheir team for the gas bench data on SCR catalysts. We thankA. Brahma for the synthesis of the two-brick model. Finally,E. Serban provided engine experimental data used in this paper.

REFERENCES

EPA (2009) Light-Duty Vehicle and Light-Duty Truck - Clean FuelFleet Exhaust Emission Standards, http://www.epa.gov/otaq/stan-dards/light-duty/ld-cff.htm.

Fulks G., Fisher G., Rahmoeller K., Wu M.C., D’Herde E., Tan J.(2009) A Review of solid materials as alternative ammonia sourcesfor Lean NOx Reduction with SCR, SAE paper 2009-01-0907.

Herman A., Wu M., Cabush D., Shost M. (2009) Model based con-trol of SCR dosing and OBD strategies with feedback from NH3sensors, SAE paper 2009-010911.

VW Jetta (2009) Volkswagen Jetta TDi Test Drive: Clean Diesel’s50 MPG Meets Prius-Humbling Thrust, Popular Mechanics,http://www.popularmechanics.com/blogs/automotive_news/4235586.html.

Kim J.Y., Cavataio G., Patterson J., Laing P.M., Lambert C.L.(2007) Laboratory Studies and Mathematical Modeling of UreaSCR Catalyst Performance, SAE paper 2007-01-1573.

Upadhyay D., Van Nieuwstadt M.J. (2006) Model Based Analysisand Control Design of a Urea-SCR de NOx Aftertreatment System,J. Dyn. Syst. Meas. Control 128, 737-741.

Van Nieuwstadt M. (2003) US patent 6,546,720.

Final manuscript received in February 2011Published online in September 2011


Copyright © 2011 IFP Energies nouvellesPermission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not madeor distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of thiswork owned by others than IFP Energies nouvelles must be honored. Abstracting with credit is permitted. To copy otherwise, to republish, to post onservers, or to redistribute to lists, requires prior specific permission and/or a fee: Request permission from Information Mission, IFP Energies nouvelles,fax. +33 1 47 52 70 96, or [email protected].


Control of Urea SCR Systems for US Diesel Applications

Documents

Transcript of Control of Urea SCR Systems for US Diesel Applications