1

10FFL-0238

Loading and Regeneration Analysis of a Diesel Particulate Filter with a Radio Frequency-Based Sensor

Alexander Sappok Filter Sensing Technologies, Inc.

James Parks II, Vitaly Prikhodko

Oak Ridge National Laboratory

Copyright © 2010 SAE International

ABSTRACT

Accurate knowledge of diesel particulate filter (DPF) loading is critical for robust and efficient operation of the combined engine-exhaust aftertreatment system. Furthermore, upcoming on-board diagnostics regulations require on-board technologies to evaluate the status of the DPF. This work describes the application of radio frequency (RF) – based sensing techniques to accurately measure DPF soot levels and the spatial distribution of the accumulated material. A 1.9L GM turbo diesel engine and a DPF with an RF-sensor were studied. Direct comparisons between the RF measurement and conventional pressure-based methods were made. Further analysis of the particulate matter loading rates was obtained with a mass-based soot emission measurement instrument (TEOM).

Comparison with pressure drop measurements show the RF technique is unaffected by exhaust flow variations and exhibits a high degree of sensitivity to DPF soot loading and good dynamic response. Additional computational and experimental work further illustrates the spatial resolution of the RF measurements. Based on the experimental results, the RF technique shows significant promise for improving DPF control enabling optimization of the combined engine-aftertreatment system for improved fuel economy and extended DPF service life.

INTRODUCTION

Motivated by increasingly stringent emissions regulations, diesel particulate filters have seen widespread use as the only technically and economically feasible means for meeting current and future particulate matter (PM) emissions limits. In the United States, all 2007 and newer on-road diesel engine are equipped with particulate filters. Despite work on DPFs since the early 1980's [1], and their first serial introduction as original equipment in automotive applications in 2000 [2], current systems suffer from a number of inefficiencies. Although extremely effective at trapping soot, typical commercial filters can achieve trapping efficiencies in excess of 99%, this level of performance is not without considerable cost.

Both the DPF itself and the accumulated soot impose additional backpressure on the engine, which translates into a fuel consumption penalty. This fuel consumption penalty generally scales with back pressure and the amount of soot accumulated on the filter. Recent computational studies report fuel consumption penalties in the range of 1.5% to 2% due to increased backpressure with a DPF, depending on the level of PM load [3]. Additional experimental investigations report potential fuel savings of 0.4% to 2.0% by reducing DPF

2

backpressure [4]. Further, the soot must be periodically removed (oxidized) from the DPF during filter regeneration, which generally incurs an additional fuel penalty in most actively regenerated systems. The fuel penalty attributed to DPF regeneration has further been reported between 2% and 5% [3].

Accurate knowledge of soot levels in the DPF at any given time is critical for proper control of the regeneration process to both minimize the fuel consumption impact and to avoid damaging the DPF and other aftertreatment system components. If the DPF is allowed to accumulate too much soot, the large amount of heat released upon regeneration can not effectively be dissipated, resulting in filter damage such as by the formation of cracks, or regions which may be locally melted. On the other hand, regenerating the DPF too often incurs unnecessary fuel penalties with associated CO2 emissions. Further, advanced control strategies may also be required to adjust the conditions under which the filter is regenerated, such as exhaust temperatures and flow rates, based on the amount of soot in the DPF [5]. Despite the important role accurate knowledge of the DPF loading state plays in optimizing the DPF regeneration and control processes, the same basic measurement systems used to estimate DPF soot levels in the early 1980's are still widely used today.

DPF SOOT LOAD MEASUREMENTS

The amount of PM accumulated in the DPF is a function of many different factors. Figure 1 presents a simplified illustration of the key parameters influencing the mass of soot accumulated in the DPF, DPFM ,

specifically the engine-out soot emissions rate, inm& , the soot oxidation rate, oxm& , and the soot escaping from the

DPF, outm& . For typical commercial DPFs with trapping efficiencies near 99%, the amount of soot escaping

from the filter is generally negligible on a mass basis under most conditions. However, recently increasing scrutiny has been placed on PM number emissions as well.

m in m out

m oxM DPFm in m out

m oxM DPF

Figure 1. Mass balance of soot trapped in DPF.

Aside from the specific emissions and oxidation rates listed in Figure 1, the DPF's operating history, i.e. the variation of these rates in time, are equally as important. A mass balance for the soot accumulated in the DPF can, therefore, be formulated as follows:

dttmdttmdttmtM outoxinDPF )()()()( ∫∫∫ −−= &&& Equation (1)

While the above formulation appears straightforward, the individual terms are influenced by a large number of factors, which are often not well-known. To illustrate the point, engine-out PM emissions are a function of the specific engine operating conditions, combustion characteristics, vehicle drive cycle, fuel type, lubricant type, and lubricant consumption rate, among others. Even assuming accurate knowledge of the engine-out PM emissions rate provides only the first term in Equation 1. Once deposited on the DPF, the soot is eventually oxidized either through a passive process, such is the case with catalytic systems, through active DPF regeneration, or some combination of active and passive processes. Exhaust gas temperatures, flow rates, and composition (specifically NOx/PM ratio), DPF soot loading levels, catalyst formulations, and the quality, i.e. completeness, of the regeneration are all factors influencing the amount of soot oxidized over a given time interval. Lastly, while the amount of soot escaping the DPF is generally small, <1%, this value can become much larger when the integrity of the particulate filter is compromised through any one of a number of failure

3

modes. Detecting PM leakage from the DPF above a specific threshold value has become increasingly important from an OBD perspective [6].

Additional complexity arises from the fact that the loading state of the DPF is continually changing. Not only soot, but also inorganic ash accumulates in the DPF over time. This ash consists primarily of incombustible lubricant additives and engine wear and corrosion particles, which unlike soot, are not consumed during the regeneration process. Over time, ash build-up in the DPF leads to increased exhaust flow restriction, a reduction in soot storage capacity, and negatively impacts vehicle fuel economy. As a result, DPFs are periodically removed for ash cleaning, which is mandated to be no more frequent than every 150,000 miles [7]. DPF ash levels after 150,000 miles of on-road use may comprise more than 80% of the total accumulated material (ash and soot) in the DPF. The significant amount of the DPF volume occupied by the ash affects both the soot distribution in the filter as well as local soot loading levels. Previous studies with ash loaded filters subjected to periodic regeneration have shown an increase in local soot loads toward the front of the filter by more than 30%, in some cases [8].

Given the large and complex number of factors influencing the amount of soot accumulated in the DPF, and the increasingly stringent regulatory framework within current systems must operate, it is quite surprising that pressure drop measurements form the backbone of most DPF soot load measurement systems. In addition to the factors outlined above, pressure drop across the DPF is itself also a function of exhaust conditions, namely flow and temperature, DPF type and configuration, as well as the distribution and amount of both soot and ash in the filter. Confounding the issue even further is that fact that many of the most common DPF materials currently in use exhibit a non-linear initial increase in pressure drop with soot load, due to the soot first accumulating in the filter pores (depth filtration) prior to forming a layer on the filter surface (cake filtration). Depending on the filter's operating history, the pressure drop response may exhibit a significant hysteresis as a result of the relative amounts of soot accumulated in the filter pores and cake layer. Several studies have attempted to quantify the variability and error in pressure-based DPF soot load measurements, reported in the range of + 50% of the measurement [9, 10].

Additional concern has recently focused on the ability of pressure-based measurement systems to detect DPF failures in response to upcoming OBD regulations. The results of several recent studies targeting the issue cast serious doubt regarding the ability of pressure-based measurement systems to detect small filter failures which may result in tailpipe-out PM levels exceeding the mandated limits [11, 12]. Additional work has also focused on investigating alternative technologies to measure soot levels downstream of the DPF for purposes of OBD compliance [13].

In order to reduce the variability inherent to pressure measurements, nearly all commercial engine and DPF control systems employ some type of predictive models to better estimate the loading state of the filter. A detailed treatment of these models is well beyond the scope of this study, and there has been extensive work in this field over the past twenty years. In general, these models may estimate many of the key parameters outlined in Equation 1, based on data tables and maps stored in the ECU, or predictive methods utilizing data from various sensors on the engine or vehicle. Additional corrections are generally also applied to account for the effects of exhaust flow and temperature variations on the pressure drop signal. Further, these models are generally calibrated for a specific engine, fuel, and aftertreatment system, rendering the overall measurement method relatively inflexible for use in applications outside the scope of the calibration.

The variability of combined pressure- and model-based DPF soot load measurement systems has been reported from 1 g/L to 3 g/L in a recent study with a passenger car over a range of urban and highway drive cycles [14], or 16% to 50% of the maximum allowable soot load assuming a limit of 6 g/L for a cordierite DPF. Despite some gains in overall measurement accuracy, current pressure- and model-based measurement systems provide only an indirect estimate of soot levels in the particulate filter. The objective of the present work, therefore,

4

investigates the feasibility of utilizing radio frequencies to provide a direct measure of soot accumulated in the DPF, thereby eliminating much of the variability encountered by current methods.

RF SENSOR OPERATION

The RF sensing system employed in this study is shown schematically in Figure 2. The system consists of two small probes or antennas mounted in the DPF housing upstream and downstream of the particulate filter. Each antenna is connected to the RF sensor control unit. The control unit generates an RF signal which is transmitted by one antenna through the ceramic diesel particulate filter to the second antenna at the opposite end of the filter (transmission). The second antenna is also connected to the RF control unit to which the received signal is returned. Although a two antenna configuration was employed in the present study, the same system can also be implemented using only one antenna to transmit and received the RF signal (reflection).

Figure 2. Diesel particulate filter instrumented with RF antennas.

While the RF signal easily passes through the ceramic DPF, a cordierite filter in this study, it does not penetrate the metals walls of the DPF housing and exhaust system. Furthermore, the frequencies of operation for the system were chosen such that the RF sensor was operating below the cut-off frequency of the exhaust pipes connected to the DPF housing. At these frequencies the wavelength of the signal is too large to propagate down the exhaust pipes, and thus the signal is fully-contained in the filter housing.

The characteristics of the RF signal are influenced by the dielectric properties of the material through which it propagates. Specifically, the permittivity of a material is a complex number composed of two parts:

εεε ′′−′= j Equation (2)

with the relative permittivity defined as, o

R εεε ′

= . The relative permittivity or dielectric constant is the ratio of

the real portion of a material’s permittivity,ε ′ , to the permittivity in free space,oε , and affects the frequency of

the signal. Additionally, signal amplitude and the amount of transmitted power is a function of the losses within the material through which the wave travels. The loss factor, or loss tangent is defined as:

εεδ

′′′

=tan Equation (3)

A loss tangent of zero occurs when there is no dielectric loss, such as for waves propagating in a vacuum.

5

Applying these principles to measure soot and ash loading in a diesel particulate filter is relatively straightforward. The material trapped in the DPF (soot and ash) have different dielectric properties from the ceramic filter and the medium (air/exhaust) which the trapped material displaces. As soot and ash accumulate in the DPF, the RF signal is directly affected, as shown in Figure 3.

Frequency

Tra

nsm

issi

on (

S21

)

No Soot

Soot

Frequency

Tra

nsm

issi

on (

S21

)

No Soot

Soot

Figure 3. Effect of soot accumulation in DPF on RF signal response.

Figure 3 presents the RF signal response over a wide frequency range. Sweeping the signal over this frequency range causes resonant modes (peaks) to be established within the DPF housing. Material accumulation in the DPF affects the frequencies, amplitude, and width of the resonant modes, as depicted in the figure. In the case of soot, the signal amplitude is reduced due to its high loss factor. On the other hand, ash exhibits relatively low loss, but its dielectric constant is sufficiently different from that of the soot to shift the resonant mode frequencies. In this manner, changes in the RF signal characteristics may be directly correlated to both the amount of soot and ash in the DPF.

Physically, the resonant modes shown in Figure 3 correspond to local areas of high electric field established in the DPF at the various frequencies. Further, the sensitivity of the RF measurement technique is a direct function of the electric field strength. Therefore, by operating the RF sensor over a wide range of frequencies to include multiple resonant modes the spatial distribution of the accumulated material may also be determined.

Figure 4 shows the results of RF simulations depicting the variations in the electric field strength within the DPF for two different resonant modes. While the first mode depicted in Figure 4 exhibits a high field at the front and back face of the DPF, and is therefore more sensitive to PM accumulation in these regions of the filter, the second mode shown in the figure is also sensitive to soot accumulation in the middle of the DPF.

Figure 4. Model visualization of electric field profiles in DPF. Dark areas with light-colored outline indicate regions of high electric field strength.

6

Depending upon the specific application requirements, the RF signal response may be averaged over several resonant modes to obtain information on the total average material loading in the entire filter. In other instances, more accurate information regarding localized filter loading may be required, in which case the response of individual resonant modes relative to one another may be analyzed. Accurate knowledge of radial and axial variations in DPF soot or ash load is useful to control the regeneration process based on the local filter loading conditions, as it is well known that ash accumulation affects the soot distribution in the DPF. Further, this same information may be important for diagnostic applications to detect anomalous soot and ash build-up.

The operating principles of the RF sensor provide both a direct means of measuring soot and ash loading in the DPF, as well as additional information, such as the spatial distribution of the accumulated material, which would otherwise be difficult, if not impossible, to obtain in-situ. The primary focus of this study was to characterize the RF sensor response to the total amount of soot accumulated in the DPF over a range of engine operating conditions and aftertreatment system configurations, thereby providing a direct comparison with filter pressure drop measurements.

EXPERIMENTAL SET-UP AND PROCEDURES

The experiments conducted as part of this study were carried out at the Fuels Engines and Emissions Research Center at Oak Ridge National Laboratory (ORNL) using a fully-instrumented light-duty diesel engine, with the RF measurement system provided by Filter Sensing Technologies. Engine-out PM emissions were monitored over the course of the testing and provide additional data for comparison with the RF sensor and pressure drop measurements. The following section presents additional details regarding the experimental set-up and test procedures.

ENGINE AND INSTRUMENTATION

The engine used in this study was a GM 1.9 liter, four-cylinder direct injection diesel engine. This engine is rated at 110 kW (148 hp) at 4000 rpm and 315 N-m (232 ft-lb) at 2000 rpm. The engine is equipped with a high-pressure common rail fuel injection system, variable geometry turbocharger, swirl actuation, and cooled exhaust gas recirculation (EGR).

A custom full-pass Drivven engine control system was utilized for all of the engine controls. This control system provides access to and allows for modification of many of the engine control parameters. During normal engine operation (the majority of this study), the stock engine calibration was employed; however a number of engine parameters were modified to command DPF regeneration and for select investigations. Specifically, DPF regeneration was initiated by modifying fuel injection timing and quantity to allow for late in-cylinder post-injection at 80ºATDC; this increased hydrocarbon levels in the exhaust which heated the DPF via exotherms from catalytic combustion over the upstream diesel oxidation catalyst (DOC)

Both PM and gaseous emissions were measured in this work. A tapered element oscillating microbalance (TEOM), Rupprecht & Patashnick Co., Inc. Model 1105, was utilized for the majority of the PM mass emissions measurements. The TEOM has a measurement resolution of 0.2 mg/m3. Additional PM measurements using conventional gravimetric sampling with 47mm filters, as well as a Dekati DMM were also employed in specific cases. All of the PM measurements utilized dilute exhaust sampled downstream of the DOC but upstream of the DPF.

Emissions of NOx, CO, CO2, O2, and hydrocarbons were also monitored utilizing conventional gas analyzers. These analyzers were based on chemiluminescence (NOx), non-dispersive infrared (CO, CO2), flame ionization detector (hydrocarbons), and paramagnetic (O2) techniques. Analysis for CO, CO2, and O2 was performed on exhaust gas samples that were dried with a chiller prior to the analyzer.

7

RF MEASUREMENT SYSTEM

Additional details of the test system lay-out, including the RF measurement system are shown in Figure 5. The RF antennas were mounted directly in the DPF housing and connected to the RF sensor control unit. Data from the sensor control unit was recorded using a National Instruments data acquisitions system and a laptop computer. In nearly all of the tests, a DOC was mounted upstream of the DPF to facilitate filter regeneration with in-cylinder post-injection.

Air

EG

R H

XN

Engine Cooling

Shaft Work

TurboDPF DOC

TEOM

ΔΔΔΔPT

RF Control Unit

Air

EG

R H

XN

Engine Cooling

Shaft Work

TurboDPFDPFDPF DOC

TEOM

ΔΔΔΔPT

RF Control Unit

Figure 5. Experimental set-up showing RF sensing system installed on the test bed.

TEST CONDITIONS AND PROCEDURES

Two DPF configurations were utilized in conjunction with the RF sensor. In one case the DOC was mounted in a separate housing upstream of the DPF, also shown in Figure 5. In this case the RF signal was only transmitted through the housing containing the particulate filter. In the second configuration, two substrates, the DOC and DPF were installed in the same housing. In this case, the RF signal was transmitted through both the DOC and DPF. The two specific aftertreatment system configurations, including the positions of the RF antennas, are shown in Figure 6.

DP

F

DO

C

DP

F

DO

C

Figure 6. DPF configurations evaluated.

Regardless of the configuration, cordierite diesel particulate filters were used in all of the tests. The DPFs used in both systems measured 5.66 inches (14.38 cm) in diameter and 6 inches (15.24 cm) in length, with a cell

8

density of 200 cells/in.2 (31 cells/cm2) and a wall thickness of 0.012 in. (0.30 mm). In the first configuration, with the DOC and DPF in separate housings, the DPF was catalyzed and contained a platinum-based catalyst. In the second configuration, with the DOC and DPF in the same housing, the DPF was not catalyzed. Unless specifically noted, the performance data reported in this work corresponds to the first configuration in Figure 6 with the DOC and DPF in separate cans.

The RF sensor response was studied over a number of engine operating conditions and modes. Regardless of DPF loading conditions, the same engine operating point, 1500 rpm and 50 ft-lb (67.8 N-m), was used for DPF regeneration. This particular condition was utilized as it allowed for stable engine operation with the post-injection regeneration strategy and the DPF regeneration characteristics (based on pressure drop) were well known, having been established in previous studies. Coincidentally, this operating point also happened to correspond to a fairly high level of engine-out PM emissions.

Initially a series of steady-state loading and regeneration cycles were used to study RF sensor response and repeatability at high- and low- engine soot output conditions. Following the steady-state testing, a semi-transient cycle (Ad Hoc FTP cycle) was utilized to study the transient response of the RF signal for comparison with both the pressure drop and TEOM mass-based PM emissions measurements. The specific engine operating conditions for the Ad Hoc FTP cycle are listed in Table 1.

Table 1. Engine operating modes for Ad Hoc FTP cycle.

Designation Engine Speed Load Timerun rpm ft-lb sec0 900 10 602 1500 29 603 2000 22 601 1500 15 600 900 10 604 2300 47 602 1500 29 600 900 10 605 2600 99 300 900 10 601 1500 15 602 1500 29 60

As the engine test cell at ORNL was not equipped with a fully-programmable transient dynamometer, the engine was run at each mode, designated 0-5 in Table 1, for a period of 30s to 60s. After the designated amount of time, the engine operating set points were rapidly adjusted. In this manner the major transient phenomena were still captured in the test cycle. In general, the semi-transient cycle was repeated three times consecutively prior to initiating DPF regeneration.

In addition to conventional engine operation, pre-mixed charge compression ignition operation (PCCI) was also employed to determine the effect, if any, of soot composition and the RF sensor response. In general, PCCI combustion leads to reduced soot emissions, but with a higher relative soluble organic content, as opposed to conventional combustion modes. PCCI combustion was achieved by modifying the engine parameters using the Drivven control system. For the particular set of data reported in this work, the same engine operating set point, 2,000 rpm and 20 ft-lb (27.1 N-m), was used in both PCCI and conventional combustion modes. All of the studies were conducted with the engine operated on conventional ultra-low sulfur diesel fuel (ULSD).

9

EXPERIMENTAL RESULTS

The following section presents the experimental results and highlights key findings for each series of tests. RF sensor response is compared directly with the DPF pressure drop measurements and the engine-out PM emissions measured using the TEOM.

SENSOR REPEATBILITY

A series of steady-state engine operating conditions was utilized to study RF sensor repeatability over successive loading and regeneration cycles. Figure 7 presents a direct comparison of RF sensor output with pressure drop measurements over several DPF loading and regeneration cycles. The engine was operated at a constant speed and load, 1500 rpm and 50 ft-lb (67.8 N-m), during both the loading and regeneration cycles shown in Figure 7. As noted above, this particular operating point resulted in fairly high engine-out PM emissions. DPF pressure drop prior to the regeneration of approximately 4 kPa was only slightly below the typical regeneration threshold of 6 kPa employed by ORNL at this operating condition.

0

1

2

3

4

0 30 60 90 120 150 1800

2

4

6

8

10RF Sensor Pressure Drop

0

2

4

6

8

10

0 30 60 90 120 150 180

Pressure Drop (Norm) Pressure Drop

Time [min]

RF

Sen

sor

[Arb

itrar

y]

Pre

ssur

e D

rop

[kP

a]

∆P

[kP

a],N

orm

aliz

ed∆

P[k

Pa/

m3 /

min

]

RF Sensor

Normalized ∆P

Figure 7. Comparison of RF sensor response with pressure drop over steady-state loading and regeneration cycles. Bottom chart compares raw pressure drop signal with pressure drop normalized by exhaust flow rate.

Unlike pressure drop which is affected by variations in exhaust flow and temperature, the RF sensor response is insensitive to flow variations. Although the dielectric properties of the soot do exhibit some temperature dependence, a temperature correction is applied to the raw RF signal in the sensor control unit. The variations in exhaust flow and DPF inlet and outlet temperatures over the steady-state test cycle are shown in Figure 8.

10

0

200

400

600

800

0 30 60 90 120 150 180

0

10

20

30

40

50

60

Time [min]T

empe

ratu

re [°

C]

Exh

. Flo

w [g

/s]

DPF_Tin DPF_Tout

Exh_M_Flow

Figure 8. Exhaust conditions during steady-state loading and regeneration cycles.

The large spikes in pressure drop at the onset of DPF regeneration, observed in Figure 7, are clearly due to the increase in exhaust flow rates and temperatures, shown in Figure 8. Figure 7 also presents the normalized pressure drop (pressure drop normalized by exhaust flow rate), which shows some improvement in the pressure drop transient response to the exhaust flow rate variations at the start of DPF regeneration, although the effects of additional factors, such as temperature are still quite evident. Despite the fairly high DPF inlet temperatures (>750 °C) during filter regeneration and the large variation in exhaust temperatures between the soot loading and regeneration events (350 °C), the RF sensor exhibits a fairly stable, linear response.

The stability and response of the RF sensor and pressure drop measurement were also studied over extended engine operation at relatively low engine-out PM emission conditions. Figure 9 presents a comparison of the RF signal response with the corrected pressure drop measurements (normalized to exhaust flow rate), as well as the exhaust conditions over the course of the test cycle. As indicated in the figure, the DPF was regenerated during the first 45 minutes and last 15 minutes of the test. Following initial filter regeneration, the engine was operated at a low speed and load condition, 1,000 rpm and 15 ft-lb (20.3 N-m), for slightly more than 30 minutes. Thereafter, engine load was increased to 25 ft-lb (33.9 N-m) while maintaining a constant engine speed, and operated at this condition for over 1 hour. The increase in engine load is evidenced by the step change in DPF inlet temperature around the 80 minute mark in Figure 9.

Both the RF sensor output and pressure drop signal are observed to increase gradually over the course of the DPF loading tests with relatively low engine-out PM emissions. Further, the effect of the slight increase in engine load, from 15 ft-lb to 25 ft-lb (20.3 N-m to 33.9 N-m), on engine-out soot emissions and DPF load is also evident in Figure 9. Over the course of the test, RF sensor response was quite stable, even for extended operation at relatively low rates of PM accumulation on the DPF.

11

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

5 25 45 65 85 105 125 145 1650

1

2

3

4

5

6

7

0

10

20

30

40

50

5 25 45 65 85 105 125 145 1650

200

400

600

800

Time [min]

Tem

pera

ture

[°C

]

Exh

. Flo

w [g

/s]

RF Sensor

RF

Out

put [

Arb

itrar

y]

Nor

m. ∆

P [k

Pa/

m3 /

min

]

Regen Loading: 1,000 rpm

Corrected ∆P

DPF_Tin

DPF_Tout

Exh_M_Flow

Figure 9. Comparison of RF sensor response with pressure drop normalized by exhaust flow rate for low engine-out PM emissions. Bottom chart presents exhaust temperatures and flow rates.

The results of the steady-state testing clearly indicate the feasibility of using RF sensing to monitor soot loading in diesel particulate filters. RF sensor measurements showed a high degree of repeatability over successive loading and regeneration cycles, and also good sensitivity to monitor even low-level PM accumulation on the DPF. Relative to pressure drop measurements, the RF technique is unaffected by flow rate variations, and demonstrated stable performance even at relatively high exhaust temperatures (>750 °C).

TRANSIENT RESPONSE

Aside from steady-state testing, the RF sensor response was also evaluated over a transient test cycle designed to simulate the Federal Test Procedure (FTP) drive cycle. The individual engine operating conditions for this test cycle were shown in Table 1, and Figure 10 presents the relative fraction of time spent in each operating mode. In general, the test cycle is representative of an urban drive cycle characterized by low load (low soot emissions) engine operating conditions, with only a short duration high speed and load (high soot emissions) operating condition, mode 5. Previous studies have shown urban/city drive cycles to result in the largest variability with current pressure- and model-based DPF soot load measurement systems [14]. The test cycle was repeated three times prior to DPF regeneration.

12

0%

10%

20%

30%

40%

0 1 2 3 4 5Mode

Tim

e [%

]

Figure 10. Engine operation over Ad Hoc FTP cycle.

The exhaust temperature and engine speed variations over three successive cycles are shown in Figure 11. DPF inlet temperatures ranged from 150 °C to 500 °C, with the highest temperature achieved during the relatively short operation at the high speed and load condition, mode 5. Although not a true transient cycle, the rapid switching between operating modes, shown in Figure 11, still allowed for the main transient events to be captured during these tests.

0

100

200

300

400

500

600

0 10 20 30 40500

1000

1500

2000

2500

3000

3500

4000

Time [min]

Eng

ine

Spe

ed [r

pm]

Tem

pera

ture

[°C

]

Engine Speed

DPF_Tin DPF_Tout

Figure 11. Variation of engine speed and exhaust temperatures over transient cycles.

The RF sensor and pressure drop measurements over the transient cycles are shown in Figure 12. The pressure drop data presented in the figure is also normalized to account for exhaust flow variations. Further, the engine-out PM emissions computed from the TEOM measurements are also shown in Figure 12. In general, the engine-out PM levels were fairly low for the majority of the test cycle, however several large spikes in engine-out PM emissions are observed to correspond with the high speed and load operating modes.

From Figure 12 it is clear that the RF sensor exhibits a fast response to the large transient spikes in engine-out soot emissions, as well as good stability during the relatively mild (low-level soot emissions) engine operation characteristic of most of this test cycle. Additionally, as the RF sensor directly measures the amount of soot accumulated on the DPF, it also captures the effects of some passive soot oxidation which may be occurring during portions of the test cycle.

In contrast to the RF sensor measurements, pressure drop is directly influenced by variations in the exhaust conditions, as evidenced by the large variation in pressure drop measurements, also shown in Figure12. Further, the large spikes in the pressure drop measurements, coincidently occurring during some of the high engine-out soot emission conditions, are primarily attributed to the rapid increase in exhaust temperatures at

13

these conditions, see Figure 11. The large variability in the pressure drop measurements is evident in Figure 12 where similar levels of engine-out soot emissions are measured for the peaks labeled “A” and “B”, however pressure drop is observed to vary by over a factor of 3 for these conditions.

0.0

0.5

1.0

1.5

2.0

2.5

0 10 20 30 400

50

100

150

200

250

300

350

0.0

3.0

6.0

9.0

0 10 20 30 400

50

100

150

200

250

300

350

Time [min]

RF

Out

put [

Arb

itrar

y]N

orm

. ∆P

[kP

a/m

3 /m

in]

PM

[g/h

r]P

M [g

/hr]

PMRF Output

PM

Corr ∆P

AB

0.0

0.5

1.0

1.5

2.0

2.5

0 10 20 30 400

50

100

150

200

250

300

350

0.0

3.0

6.0

9.0

0 10 20 30 400

50

100

150

200

250

300

350

Time [min]

RF

Out

put [

Arb

itrar

y]N

orm

. ∆P

[kP

a/m

3 /m

in]

PM

[g/h

r]P

M [g

/hr]

PMRF Output

PM

Corr ∆P

AB

Figure 12. Comparison of RF sensor output (top) and corrected pressure drop signal (bottom) with engine-out PM emissions over transient test cycles.

A direct comparison of the RF sensor output over the transient cycle with the TEOM measurements is presented in Figure 13. The TEOM measures the cumulative soot mass on a small sample filter contained on microbalance within the instrument. During engine operation, a small portion of the exhaust is routed to the TEOM, allowing for continuous measurement of engine-out PM emissions. Similar to the TEOM, the RF sensor measures the cumulative amount of soot captured on the diesel particulate filter. The output of both instruments, therefore, is cumulative soot mass as a function of time.

0

50

100

150

200

250

300

350

0 5 10 15 20 25 30 35 400.0

0.5

1.0

1.5

2.0

2.5

Time [min]

RF

Out

put [

Arb

itrar

y]

TE

OM

[µg]

RF Output

TEOM

Figure 13. Direct comparison of RF sensor output with TEOM cumulative PM mass data.

14

Compared to the TEOM, the RF signal shows a similar response to the transient high soot output conditions and slightly less sensitivity during the low-level soot emission conditions. Further, relative to the gradual increase in the TEOM soot mass, observed during those portions of the test cycle exhibiting low-level engine-out soot emissions, the RF sensor measurements indicate some amount of passive regeneration occurring within the catalyzed DPF.

The results of the transient studies further demonstrate the potential of the RF technique to provide a real-time measure of DPF soot accumulation levels. Despite significant variability in the pressure drop signal over the course of the transient cycle, the RF sensor provided good dynamic response with much less variability in the measurement. Further, aside from some passive soot oxidation in the DPF, the RF sensor matched the TEOM measurements quite well, particularly for large fluctuations in engine-out soot levels.

PCCI OPERATION

This study also investigated the effects of soot composition on the RF and pressure drop signals. In order to generate PM of varying composition, the engine was operated in PCCI and conventional combustion modes. In general, PCCI operation results in decreased PM emissions, although PCCI soot typically contains a relatively greater proportion of organic components. For both PCCI and conventional combustion, the engine was operated at the same conditions, namely 2,000 rpm and 20 ft-lb (27.1 N-m). Figure 14 presents the particle size distribution and number concentrations for the PM generated during PCCI and conventional engine operation.

12/09/09, Cell 2, Mode 3

dp (nm)

10 100 1000

dN/d

logd

p (p

art./

cm3 )

0

1e+8

2e+8

3e+8

4e+8

Conventional

PCCI

Figure 14. Particle size distribution for PM generated from PCCI and conventional combustion at 2,000 rpm and 20 ft-lb (27.1 N-m).

The geometric mean diameter of the particles produced from conventional engine operation was 63.2 nm + 0.7 nm and the geometric mean diameter of the particles produced from PCCI operation was 47.6 nm + 1.1 nm. Further, PCCI operation reduced the particle number concentration by approximately a factor of 3 at the same engine condition. The ratio of organic carbon to elemental carbon (OC/EC) was also approximately 3 times higher in the PCCI soot as compared to the PM generated with conventional engine operation.

Figure 15 presents a comparison of RF and pressure drop response as the DPF was loaded with soot generated using conventional and PCCI combustion. Both the pressure drop and RF signals are normalized to the baseline value for a reference DPF condition at the start of the test. The figure clearly shows less soot accumulated in the DPF when the engine was operated in PCCI mode. Further, no appreciable differences in the RF or pressure drop signals due to changes in soot composition were observed for the conditions evaluated in this study.

15

0.9

1.0

1.1

1.2

1.3

1.4

1.5

0 20 40 60 80Time [min]

∆P (Conv)

Nor

mal

ized

Out

put

∆P (PCCI)

RF (PCCI)

RF (Conv)

Figure 15. RF sensor and pressure drop response to PM generated from PCCI and conventional combustion at 2,000 rpm and 20 ft-lb (27.1 N-m).

COMBINED DOC + DPF SYSTEMS

Up to this point, all of the results reported in this work have focused on the use of the RF sensor in conjunction with only a single DPF, as shown in the top image in Figure 6. Additional studies were conducted utilizing a combined DOC and DPF system, shown in the bottom image in Figure 6. In this particular configuration both the DOC and DPF were installed in the same housing. The RF antennas we inserted upstream of the DOC and downstream of the DPF. In this manner, the RF signal was transmitted through both the DOC and DPF. Figure 16 presents the RF signal response over one loading and regeneration cycle, along with the exhaust conditions, for the combined DOC and DPF system.

0

10

20

30

40

0 5 10 15 20 25 300

200

400

600

800

0

1

2

3

4

0 5 10 15 20 25 300

1

2

3

4

5

6

7

Time [min]

RF

Out

put [

Arb

itrar

y]

Nor

m. ∆

P [k

Pa/

m3 /

min

]

RF Output

Corr ∆P

DPF_Tin

DPF_Tout

Exh_M_Flow Tem

pera

ture

[°C

]

Exh

. Flo

w [g

/s]

Figure 16. Comparison of RF sensor response with pressure drop for a combined DOC + DPF at 1500 rpm, 50 ft-lb (67.8 N-m). Bottom chart presents exhaust temperatures and flow rates.

16

The RF sensor performance shown in Figure 16 is similar to that observed for the single DPF configuration, with the DOC in a separate housing upstream of the particulate filter, Figure 7. The pressure drop data shown in Figure 16 is also normalized to account for exhaust flow variations, although the influence of temperature variations on the pressure drop signal is clearly evident. Based on the results, the RF sensor appears equally applicable to monitor soot loading in combined DPF and DOC systems, and other aftertreatment system configurations in which additional catalyst substrates may be integrated into the same DPF housing.

SUMMARY AND CONCLUSIONS

This study investigated the use of RF sensing to directly monitor soot accumulation in diesel particulate filters. The RF system was configured to generate and transmit RF signals across the DPF over a wide frequency range. Analysis of the resulting resonance profiles provides information on the amount, type, and distribution of the material (soot and ash) accumulated in the DPF. The RF measurement system was installed on a modern 1.9 L GM turbocharged diesel engine and evaluated over a variety of engine operating conditions and advanced combustion modes. The technique was successfully applied with two of the most common diesel aftertreatment system configurations.

Results of this study confirm the feasibility of using RF sensing to provide a near real-time measure of soot levels in the DPF through a direct comparison with both filter pressure drop and engine-out soot emissions measurements. Investigations of diesel particulate filter soot loading in conjunction with the RF sensor provide the following general conclusions:

1. Unlike pressure drop measurements which are directly affected by exhaust flow and temperature variations, the RF technique is not affected by changes in exhaust flow. Although the dielectric properties of soot exhibit some temperature dependence, the RF sensor performance was evaluated with DPF inlet temperatures in excess of 750 °C and with large transient fluctuations in exhaust temperature over 350 °C.

2. The RF sensor exhibited good repeatability over the successive steady-state DPF loading and regeneration conditions evaluated in this work. Sensitivity to detect even low-level engine-out soot emissions was demonstrated during extended periods of engine operation at low speed and load conditions.

3. Transient response of the RF sensor was investigated over a modified FTP cycle designed to simulate city-type driving conditions. Relative to the pressure drop measurements, the RF signal exhibited considerably less variability and good dynamic response to the transient events.

4. Comparison of the RF sensor's transient response with a laboratory PM mass measurement system, TEOM, showed comparable performance. While both instruments showed similar response to transient fluctuations in engine-out PM emissions, the RF sensor also indicated some amount of passive soot oxidation occurring in the DPF over portions of the test cycle.

5. Investigations of PM composition on the RF and pressure signals showed no appreciable effect of differences in PM organic and elemental carbon content on the RF signal for the conditions evaluated utilizing both conventional and PCCI combustion modes.

6. Studies utilizing the RF sensor in conjunction with a combined DOC and DPF system showed the technique is equally applicable to monitor soot loading in particulate filters packaged in larger systems containing additional catalyst substrates.

The results of this work clearly demonstrate the feasibility of implementing a means for measuring soot levels directly in the DPF using RF sensing. Relative to current indirect measurement methods, the RF technique shows significant promise for improving DPF control enabling optimized operation of the combined engine-aftertreatment system for improved fuel economy and extended DPF service life.

17

REFERENCES

1. Mogaka, Z., Wong, V., and Shahed, S., “Performance and Regeneration Characteristics of a Cellular Ceramic Diesel Particulate Trap,” SAE 820272, 1982.

2. Salvat O., P. Marez and G. Belot, "Passenger Car Serial Application of a Particulate Filter System on a Common Rail Direct Injection Diesel Engine," SAE 2000-01-0473, 2000.

3. Rutland, C., Foster, D., Narayanaswamy, K., and He, Y., " Investigation into Different DPF Regeneration Strategies Based on Fuel Economy Using Integrated System Simulation," SAE 2009-01-1275, 2009.

4. Mikulic, I., Zhan, R., and Eakle, S., "Dependence of Fuel Consumption on Engine Backpressure Generated by a DPF," SAE 2010-01-0535, 2010.

5. Boger, T., Rose, D., Tilinger, I., and Heibel, A., "Regeneration Strategies for an Enhanced Thermal Management of Oxide Diesel Particulate Filters," SAE 2008-01-0328, 2008.

6. California Code of Regulations, Title 13, section 1968.2, Malfunction and Diagnostic System Requirements for 2004 and Subsequent Model-Year Passenger Cars, Light-Duty Trucks, and Medium-Duty Vehicles and Engines (OBD II).

7. Manufacturers of Emission Controls Association (MECA), “Diesel Particulate Filter Maintenance: Current Practices and Experience,” Washington D.C., 2005.

8. Sappok, A., and Wong, V., "Ash Effects on Diesel Particulate Filter Pressure Drop Sensitivity to Soot and Implications for Regeneration Frequency and DPF Control," SAE 2010-01-0811, 2010.

9. Ohyama, N., Nakanishi, T., Daido, S., " New Concept Catalyzed DPF for Estimating Soot Loadings from Pressure Drop," SAE 2008-01-0620.

10. Mizuno, Y., Miyairi, Y., Katsube, F., Ohara, E., Takahashi, A., Makino, M., Mizutani, T., Yuki, K., and Kurachi, H., "Study on Wall Pore Structure for Next Generation Diesel Particulate Filter" SAE 2008-01-0618. 2008.

11. Van Nieuwstadt, M, and Trudell, D., "Diagnostics for Diesel Particulate Filters," SAE 2004-01-1422, 2004. 12. Van Nieuwstadt, M, and Brahma, A., "Uncertainty Analysis of Model Based Diesel Particulate Filter

Diagnostics," SAE 2008-01-2648, 2008. 13. Ochs, T., Schittenhelm, H., Genssle, A., and Kamp, B., "Particulate Matter Sensor for On Board Diagnostics

of Diesel Particulate Filters (DPF)," SAE 2010-01-0307, 2010. 14. Rose, D., Boger, T., "Different Approaches to Soot Estimation as Key Requirement for DPF Applications,"

SAE 2009-01-1262, 2009.

ACKNOWLEDGMENTS

This work was supported in part by the National Science Foundation and the U.S. Department of Energy (TIAC) award. This work was also supported in part by the U.S. Department of Energy (DOE), Office of Vehicle Technologies. The authors gratefully acknowledge the support and guidance of program managers Ken Howden and Gurpreet Singh at DOE.

DISCLAIMER

The submitted manuscript has been authored by a contractor of the U.S. government under contract number DE-AC05-00OR22725. Accordingly, the U.S. government retains a nonexclusive, royalty-free license to publish or reproduce the published form of this contribution, or allow others to do so, for the U.S. government.

DEFINITIONS/ABBREVIATIONS

ATDC: After Top Dead Center

DOC: Diesel Oxidation Catalyst

18

DPF: Diesel Particulate Filter

ECU: Engine Control Unit

EGR: Exhaust Gas Recirculation

EPA: Environmental Protection Agency

FTP: Federal Test Procedure

OBD: On-Board Diagnostics

ORNL: Oak Ridge National Laboratory

PCCI: Pre-mixed Charge Compression Ignition

PM: Particulate Matter

RF: Radio Frequency

TEOM: Tapered Element Oscillating Microbalance

ULSD: Ultra-Low Sulfur Diesel

ε : Permittivity

Rε : Dielectric Constant

DPFM : Mass of soot in DPF

inm& : Engine-out soot emissions rate

oxm& : Soot oxidation rate

outm& : Soot loss rate

δtan : Loss tangent