Read Microsoft Word - Q51760 RD_10_320201_4 ACEA_Euro6_OBD_Report.doc text version

EURO6

Light­Duty Vehicle OBD Project

Evaluation and Assessment of Proposed EOBD Emission Thresholds

Q51760 2nd November 2010 RD.10/320201.4

Authors:

Simon Finch Bohumil Hnilicka Hector Sindano Dr Andy Noble Project Director

Approved:

Ricardo UK Limited Cambridge Technical Centre Telephone: +44 (0) 1223 223200

400 Science Park

Milton Road

Cambridge

CB4 0WH UK

UK

Facsimile: +44 (0) 1223 223300

Registered in England No. 2815682 BN43 5FG

Registered Office: Shoreham Technical Centre

Shoreham-by-Sea West Sussex

RD.10/320201.4

TABLE OF CONTENTS

EXECUTIVE SUMMARY........................................................................................................3 1. INTRODUCTION ........................................................................................................6 2. TEST OBJECTIVES...................................................................................................6 3. TEST STRUCTURE ...................................................................................................7 3.1 Emission Thresholds ......................................................................................8 3.2 Test Vehicles ..................................................................................................9 3.3 Test Cycle.....................................................................................................10 3.4 Component Degradation...............................................................................10 3.5 In-Use Performance......................................................................................17 4. TEST RESULTS.......................................................................................................23 4.1 Upstream Oxygen Sensors (Gasoline)..........................................................23 4.2 Upstream & Downstream Lambda Sensors (Diesel LNT) .............................28 4.3 Downstream Lambda Sensors (Gasoline).....................................................43 4.4 3-Way Catalytic Converter (Gasoline)...........................................................48 4.5 Misfire Monitor (Gasoline).............................................................................54 4.6 Fuel System Monitor (Gasoline)....................................................................61 4.7 VVT System Monitor (Gasoline)....................................................................69 4.8 MAF Sensor Monitor (Gasoline)....................................................................75 4.9 Exhaust Gas Recirculation (Diesel)...............................................................79 4.10 EGR Cooler / ByPass (Diesel) ....................................................................104 4.11 NSC/LNT Monitor (Diesel) ..........................................................................107 4.12 Oxidation Catalyst (Diesel) .........................................................................118 4.13 Particulate Filter (Diesel).............................................................................129 4.14 SCR Monitor (Diesel) ..................................................................................157 5. SUMMARY.............................................................................................................169 5.1 Gasoline Results.........................................................................................170 5.2 Diesel Results.............................................................................................171 6. Conclusions..........................................................................................................172 6.1 Gasoline .....................................................................................................172 6.2 Diesel .........................................................................................................172

2 November 2010 © Ricardo plc 2010

Page 2

RD.10/320201.4

EXECUTIVE SUMMARY

The purpose of this report is to provide independently assessed information that will support the necessary legislative process to confirm the Euro 6 OBD thresholds. The capability of sensor technology and technology currently under development to achieve reliable and robust detection of failures at the Euro 6 OBD thresholds which have been proposed by the Commission, but which the Commission notes should be subject to impact assessment by September 2010, are assessed in this report. ACEA is of the opinion that the feasibility of technical measures to monitor against the proposed Euro 6 OBD thresholds must be reviewed as soon as possible in order to support the legislative process that will implement the appropriate Euro 6 OBD thresholds while also providing sufficient industrial lead-time for compliance. This report details the procedures and results of tests that were carried out by ACEA members within confidential OBD development programs with their suppliers, in order to evaluate and assess the feasibility of technical solutions needed to monitor against the Euro 6 OBD thresholds proposed by the Commission and within the available lead-times running up to the Euro 6 emissions stage which is mandatory from 2014. Five vehicles from five different manufacturers were tested as part of this specific program. Details of the vehicles tested are provided in Section 3.0 of this report. The conclusions that can be drawn from the tests on spark-ignition engines are as follows: 1. The proposed Euro 6 OBD thresholds were exceeded for NOx and NMHC with the degraded catalysts used. Compared to the proposed Euro 6 OBD thresholds of 100 mg/km (NMHC) and 90 mg/km (NOx), detection was achieved with the current calibration at 277 mg/km (NMHC) and 136 mg/km (NOx). However, further work is required with less aggressively aged catalysts in order to make a definitive statement about the feasibility of the monitoring techniques to detect failures at, or close to, thresholds corresponding to those proposed for Euro 6. 2. The proposed Euro 6 OBD threshold was exceeded for NMHC with the 6 % misfire level tested (277 mg/km versus proposed 100 mg/km). However, further work is required with a reduced calibration threshold and lower misfire rate to make a definitive statement about the robustness of the monitoring technique to detect failures at the proposed Euro 6 threshold. 3. VVT System Monitoring successfully detected a Total Functional Failure (TFF), however emissions exceeded the proposed EU6 EOBD thresholds (177 mg/km versus proposed 100 mg/km). Further investigation would be required with revised monitor thresholds and partial VVT failure to confirm that detection at the proposed threshold level would not result in false fault detection. 4. Although PN and PM measurements were taken, in each case a specific trend was not obvious and the proposed OBD and type approval thresholds were not exceeded in any of the test measurements made.

2 November 2010 © Ricardo plc 2010

Page 3

RD.10/320201.4

The conclusions that can be drawn from the tests on compression-ignition engines are as follows (note that these conclusions reflect the results from the different compression-ignition vehicles tested): 1. The EGR low flow monitor was able to distinguish the baseline result (without a malfunction) from a failure at the proposed EU6 EOBD emissions level (140 mg/km NOx). However, to reduce the risk of false failure detection with partial degradation and emissions within the type approval emission limit, more separation is required between the type approval and EOBD thresholds. 2. None of the EGR high flow faults resulted in significant increase in emissions due to the DPF effectively removing any additional PM. This failure is not a significant factor in the determination of the EU6 OBD thresholds. 3. Neither the proposed Euro 6 NOx OBD threshold (140mg/km) nor the type approval threshold (80 mg/km) was exceeded during an EGR Cooler / By-Pass total functional failure. This failure is not a significant factor in the determination of the EU6 OBD thresholds. 4. The NOx threshold achieved on the vehicle tested for NSC/LNT monitor was approximately the proposed EU6 threshold (140mg/km). However, on the application tested this represented a large scale failure of the NSC/LNT because of the relatively low engine out NOx level. Additional work is required to confirm a threshold that can be robustly detected by all engines. 5. In the case of DOC monitoring, emissions results for complete failure were below the proposed Euro 6 OBD thresholds. However, this corresponded to total failure of the DOC because tailpipe emissions were below 140 mg/km in this condition. The spread of emissions results suggests that a partially failed DOC could not be robustly detected for this vehicle. 6. DPF monitoring based on the current method of measuring differential pressure across the DPF produced a best achievable PM threshold of 33.4 mg/km, which is over 3 times the proposed Euro 6 PM OBD threshold of 9 mg/km. 7. The best achievable PN threshold for DPF monitoring based on differential pressure across the DPF was 41×1012/km, which is more than 30 times the proposed Euro 6 PN OBD threshold of 1.2×1012/km. 8. DPF monitoring based on using prototype resistive PM sensor technology produced a best achievable PM threshold of 14.5mg/km, which is 1.6 times the proposed Euro 6 PM OBD threshold of 9mg/km. 9. The best achievable PN threshold for DPF monitoring based on prototype resistive PM sensor technology was 85×1012/km, which is more than 70 times the proposed Euro 6 PN OBD threshold of 1.2×1012/km.

2 November 2010 © Ricardo plc 2010

Page 4

RD.10/320201.4

10. The NOx threshold achieved on the vehicle tested for SCR monitor was approximately the proposed EU6 threshold. However, on the application tested this represented a large scale failure of the SCR system because of the relatively low engine out NOx level. Additional work is required to confirm a threshold that can be robustly detected by all engines. In conclusion from the test program: 1. The Euro 6 OBD thresholds (CO, NMHC and NOx) proposed by the commission for positive-ignition (gasoline) engines appear achievable for the majority of systems and components, however, further testing of some monitors (e.g. misfire, 3WC, VVT) is recommended to confirm robustness for production. 2. Where PN and PM measurements were acquired for positive-ignition engines, they showed that the proposed OBD and type approval thresholds were not exceeded for the faults considered (misfire, fuel system rich/lean). Therefore it is a low priority to set PM and PN OBD thresholds for these faults. 3. The proposed PN OBD threshold for compression-ignition (diesel) engines is more than an order of magnitude too low for monitoring by current and near-term sensor technologies of differential pressure and resistive soot sensors. 4. Analysis of monitoring using prototype soot sensors appears to suggest that such sensors could achieve a lower OBD PM threshold than differential pressure sensing monitoring. However, the achievable threshold is higher than the proposed EU6 OBD level. In addition 1. PM (soot) sensors are in development and although some durability data was provided in this report, a longer term evaluation of sensor durability in real driving conditions and the effects of fuel quality and biodiesel content requires further consideration. 2. The effect of "intrusive" monitors on CO2 (and possibly other) emissions, has not been addressed in this study. If intrusive monitors are required, it is recommended that the effects on CO2 and other emissions be assessed as part of any impact assessment. 3. In addition, the effect of reducing OBD thresholds on the achievement of In Use Performance Ratio (IUPR) requirements needs to be further considered, particularly given the trend towards certain applications (e.g. hybrids, start-stop systems) expected to have less frequent monitoring opportunities.

2 November 2010 © Ricardo plc 2010

Page 5

RD.10/320201.4

1.

INTRODUCTION

Regulations 715/2007(1) and 692/2008(2) set the Euro 5 and 6 emission requirements for light duty vehicles. One major part of the Euro 5 and Euro 6 legislation relates to On Board Diagnostics (OBD). These Regulations specify the Euro 5 OBD requirements and also a Euro 6- diesel NOx OBD threshold that might be used for incentivising early Euro 6 diesel vehicles. In mid2008, the Commission proposed(3) a set of positive-ignition and compression-ignition OBD thresholds for Euro 6 that it intended to review by September 2010 based on an impact assessment on the technical feasibility of achieving the proposed thresholds. ACEA is of the opinion that the feasibility of technical measures to monitor against the proposed Euro 6 OBD thresholds must be reviewed as soon as possible in order to support the legislative process that will implement the appropriate Euro 6 OBD thresholds while also providing sufficient industrial lead-time for compliance. This document details the procedures and results of tests that were carried out by ACEA members as part of confidential OBD development programs in order to evaluate and assess the feasibility of technical solutions needed to monitor against the OBD thresholds proposed by the Commission and within the available lead-times running up to the Euro 6 stage that applies from 2014.

2.

TEST OBJECTIVES

The purpose of the tests described in this document was to provide the sensitivity of the monitored parameter and the tailpipe emissions in relation to the EOBD thresholds. This is summarised in the following list: · The change in the OBD judgement variable in relation to the change in emissions, resulting from component degradation. This is indicated by the trend line and in the Figure below.

Judgement variable used for OBD

Trend

NOT Detectable failure Emissions certification limit Potential OBD limit

Detectable failure

NEDC Emissions

(1) (2) (3)

Official Journal of the European Union, L171, 29.6.2007, p.1. Official Journal of the European Union, L199, 28.7.2008 p.1. Official Journal of the European Union, C182, 19.7.2008, p.17.

2 November 2010 © Ricardo plc 2010

Page 6

RD.10/320201.4

·

The capability of existing sensor technology and technology currently under development to achieve detection at the Euro 6 OBD thresholds aspired to by the Commission, taking into consideration the likely spread (dispersion) in the data as indicated by in the Figure above. Where the proposed EU6 thresholds cannot be achieved, the report has sought to highlight what is the `best achievable' OBD thresholds taking into account the change in emissions, , compared to the likely dispersion, , as shown in the Figure above.

·

OBD monitors can be either "passive", meaning that they perform within natural operating conditions dictated by the driver's behaviour or "intrusive", meaning that they perform under actively generated and precisely controlled test conditions. This study considers mainly "passive" monitors. For most of the systems tested here, to investigate whether OBD thresholds could be improved with intrusive monitoring techniques would require development and prototyping of new software strategies and is outside the scope of this study. It is expected that intrusive monitors may result in an increase in CO2 and potentially other emissions, however this impact is not examined as part of this report. Durability evaluation of sensors was not part of this study but remains a significant aspect to consider for the application and integration of sensors in vehicle emission control systems, especially in relation to real-driving performance and capability with different biofuel blends.

3.

TEST STRUCTURE

As part of this project, five ACEA members conducted engine test programs following test procedures produced by Ricardo, in conjunction with ACEA to investigate the relationship between · · Degradation of critical emissions control systems and the resulting increase in emissions over the NEDC. Performance of the OBD monitoring techniques currently available and the degradation of the emissions control system

2 November 2010 © Ricardo plc 2010

Page 7

RD.10/320201.4

3.1

3.1.1

Emission Thresholds

Gasoline Vehicles

LIMITS CO (mg/km) 1000 1500 THC (mg/km) 100 NMHC (mg/km) 68 100 NOx (mg/km) 60 90 PM (mg/km)* 5 / 4.5 9 PN 12 (x10 /km) + ++

EU6 Type Approval EU Commission Proposed EU6 EOBD EU6 Interim EOBD EU5 EOBD

1900 250 300

50 50

-

+ PM limit only applies to GDI engines ++ ×2 threshold to be considered once limit is set * The 4.5 mg/km emission limit for PM shall be effective from 1 September 2011 for the type-approval on new types and from 1 January 2013 for all new vehicles. A new measurement procedure shall be introduced before the application of the limit value 3.1.2 Diesel Vehicles

LIMITS CO (mg/km) 500 750 THC (mg/km) NMHC (mg/km) 140 NOx (mg/km) 80 140 PM (mg/km)* 5 / 4.5 9 PN 12 (x10 /km) 0.6 1.2

EU6 Type Approval EU Commission Proposed EU6 EOBD EU6 Interim EOBD EU5 EOBD

1900 1900

-

320 320

240 540

50 50

-

2 November 2010 © Ricardo plc 2010

Page 8

RD.10/320201.4

3.2

Test Vehicles

Five vehicles from 5 Manufacturers were tested as part of this program. These are detailed below Vehicle Manuf 1 Components Tested

Oxygen Sensor 3WC Fuel System VVT MAF

· ·

Euro 5 certified Emission control & After-treatment Equipment: EGR, 3WC

Manuf 2

· ·

Euro 6, Emission Emission control & After-treatment Equipment: EGR, NSC/DOC, DPF Euro 6 (certified) Emission control & After-treatment Equipment: EGR & Cooler, DOC+DPF+SCR

-sensor LNT/NSC

Manuf 3

· ·

EGR EGR Cooler Bypass DOC DPF SCR

Manuf 4

· ·

Euro 6 development state Emission control & After-treatment Equipment: DOC+DPF+SCR Euro 4 emissions level (matches Euro 6 PM) Emission control & After-treatment Equipment: EGR, DOC, DPF

DPF

Manuf 5

· ·

DPF

2 November 2010 © Ricardo plc 2010

Page 9

RD.10/320201.4

3.3

3.3.1

Test Cycle

Pre-Conditioning Before each test, pre-conditioning was carried out. This was dependent on the type of vehicle and also the algorithm used to detect the malfunction. The following is a list of the pre-conditioning cycles used:

Vehicle Manufacturer Manuf 1 Manuf 2 Manuf 3 Manuf 4 Manuf 5 DPF Regen No. of NEDC 1 1 No. of EUDC 3 3 3 Cold Soak

3.3.2

Actual Test · · · · Two NEDC tests from cold start were carried out for each level of system degradation Cold Start NEDC = 20 ­ 30°C Additionally, in some cases each level was tested over 3x NEDC cycles starting with a warm engine Also to ensure accuracy of results in some cases 2 iterations of testing were carried out.

3.4

3.4.1

Component Degradation

Oxygen Sensors (Gasoline & Diesel) Real world experience suggests that the most likely failure mode that is not detectable by electrical circuit continuity monitors is a gradual degradation of the sensor with age resulting in slow response. · · · · Manipulation of sensors in order to generate a slow response can be done by installing a thimble over the tip of the sensing element. Slower response sensors are generated by having fewer/smaller holes in the thimble. Level of degradation is defined by the number and size of holes in the thimble. A range of parts are tested until the emissions threshold is exceeded.

2 November 2010 © Ricardo plc 2010

Page 10

RD.10/320201.4

3.4.2

Downstream Lambda Sensors (Gasoline) · · Slow response lambda sensors are generated by using hardware simulation (Slug box = 1.3) A set of 2 or 3 tests was repeated to attain more consistent results

3.4.3

3-Way Catalyst (Gasoline) · · Limit Catalysts - Artificially aged catalyst A set of 3 to 6 tests was repeated to attain more consistent results

3.4.4

Misfire Monitor (Gasoline) The following levels of degradation was implemented through internal EMS misfire generator function · · Misfire 2 % - Intermittent walking misfire across all cylinders Misfire 6 % - Intermittent walking misfire across all cylinders

A set of 3 to 4 tests was repeated to attain more consistent results 3.4.5 Fuel System Monitor (Gasoline) The following levels of degradation were implemented · · Lean Fuel - Injector characteristic curve was multiplied by 0.8 to simulate fuel system leaner than target by 20 % Rich Fuel - Injector characteristic curve was multiplied by 1.2 or by 1.24 to simulate fuel system richer than target by 20 %. ­ PN testing was conducted at a later date and so corresponds to a different rich shift 1.2.

A set of 3 tests was repeated to attain more consistent results 3.4.6 Variable Valve Timing (VVT) Monitor (Gasoline) The following levels of degradation were implemented

2 November 2010 © Ricardo plc 2010

Page 11

RD.10/320201.4

·

Slow response ­ Inlet and exhaust cam VVT set to 1 % ­ ­ Closed loop control maps calibrated to simulate slow response to changing target position Principal effect observed on exhaust cam since inlet cam is parked at NEDC operating points Inlet cam target position map calibrated to 100% to simulate a cam phaser stuck in the most advanced position Inlet cam moves to maximum advance about 2 seconds after start following sufficient oil pressure to mechanically unlock the cam phaser mechanism Slave exhaust cam phaser (stuck at 0%) fitted to test vehicle

·

Stuck 100 % - Inlet cam VVT stuck 100 % ­ ­

·

Stuck 0 % - Exhaust cam VVT stuck 0 % ­

3.4.7

MAF Sensor Monitor (Gasoline) The following levels of degradation were implemented · Sensor drift (cold tests) ­ MAF sensor characterisation calibration was multiplied by 1.3 to simulate MAF sensor signal drift

2 November 2010 © Ricardo plc 2010

Page 12

RD.10/320201.4

3.4.8

Exhaust Gas Recirculation (Diesel) The following levels of degradation were implemented · High flow: ­ EGR valve position manipulated by an additional simulation block between EGR valve and the ECU to: · Limit closing position of EGR valve by modifying the demand PWM signal from the ECU, to simulate permanently open valve or valve not seating properly Modify the position feedback signal to the ECU to mimic the valve being in the correct position and avoid a position sensor fault.

·

·

The following degradation levels of EGR low flow tests were evaluated: + 15% => Valve closing point limited to 15 % of full travel + 25% => Valve closing point limited to 25 % of full travel

· ·

Higher levels of degradation could not be tested due to DPF clogging and damage

Low flow ­ alternative methods used: ­ Orifice plate was inserted after EGR valve to restrict flow

2 November 2010 © Ricardo plc 2010

Page 13

RD.10/320201.4

·

The following degradation levels of EGR low flow tests were evaluated: base (0 %) 14 mm restrictor (-33 % reduction in EGR flow rate) 9 mm restrictor (-67 % reduction) no flow, i.e. 0 mm restrictor (-100 %)

Two NEDC tests from cold start were carried out for each level of system degradation 3.4.9 NSC/LNT Monitor (Diesel) Ageing of catalyst by extended high temperature operation on test bench engine · NSC's were tested with the following levels of degradation: ­ ­ ­ ­ · · 0 % (new NSC) 1 % loss of efficiency 60 % loss of efficiency 95 % loss of efficiency

Emissions degradation was measured over 3x NEDC cycles Additionally, each level was tested over 3x NEDC cycles starting with a warm engine

3.4.10 EGR Cooler By-Pass (Diesel) Two levels of EGR cooler performance were tested: · · Baseline normal operation (0 % degradation) Major/Total functional failure (100 % degradation)

Total function failure of the EGR cooler was simulated by bypassing the coolant circuit of the EGR cooler

3.4.11 Oxidation Catalyst (Diesel) A single level of DOC degradation was tested, simulated by an empty DOC can Testing was carried out over the normal NEDC for measurement of emissions and separately over an NEDC cycle with a DPF regeneration active in order to assess the monitor judgement

2 November 2010 © Ricardo plc 2010

Page 14

RD.10/320201.4

3.4.12 Particulate Filter (Diesel) To degrade the DPFs plugs were removed from the back face of the DPF. Different removal strategies were followed by each Manufacturer. These are shown in the subsections below. 3.4.12.1 Manufacturer 3

The following levels of degradation were evaluated · · · 3.4.12.2 9 % rearward area 50 % rearward area Empty DPF Can, i.e. a major failure

Manufacturer 4

Plugs were removed from the face of the DPF as shown below:

DPFs were tested with the following levels of degradation applied:

2 November 2010 © Ricardo plc 2010

Page 15

RD.10/320201.4

DPF Ageing Baseline: No plugs removed (aged to FUL) 180 plugs removed 365 plugs removed 512 plugs removed 722 plugs removed 1800 plugs removed Empty DPF can

% Degradation 0 4 9 12 17 43 100

3.4.12.3

Manufacturer 5

To degrade the DPFs used for testing, plugs were removed from the rear face of the DPF. DPFs were tested with the following levels of degradation: · · · · Baseline, no plugs removed Plugs removed to target 6 mg/km PM emissions Plugs removed to target 12 mg/km PM emissions Empty DPF can (approximately 16 mg/km expected)

3.4.13 SCR Catalyst (Diesel) It is assumed that low catalyst conversion efficiency has the same symptoms, regardless of whether this is caused by degradation of the catalyst, dosing unit or regent quality. The SCR catalyst conversion degradation is simulated by reducing the urea dosing quantity without feedback to ECU.

The following levels of degradation were evaluated · · 50 % dosing 10 % dosing

2 November 2010 © Ricardo plc 2010

Page 16

RD.10/320201.4

·

0 % dosing, i.e. major functional failure

3.5

In-Use Performance

The Euro 6 legislation requires that OBD monitors meet minimum "in-use performance ratios" are met and reported in service. The In-Use Performance ratio is defined as the number of drive cycles in which the monitor makes a judgement (numerator) to the number of drive cycles that meet legislated minimum conditions (denominator) · Numerator conditions: ­ · Monitor has completed all decisions that would be necessary to make a fail judgement on this drive cycle 10 minutes duration 5 minutes accumulated above 40 km/h 30 seconds uninterrupted (continuous) idle And, other conditions for some specific monitors:, e.g. · · for diesel particulate filter and oxidation catalyst, cumulative 800 km driven since last time drive cycle was denominator was incremented

Denominator conditions: ­ ­ ­ ­

Minimum average ratio must be achieved: ­ ­ 0.336 for most monitors Alternative targets for some specific monitors, e.g. gasoline evap/purge and SAI

3.5.1

Model-Based Evaluation In order to assess the capability of any OBD monitoring strategy, it is important to consider whether the conditions required to make a judgement will allow the numerator to be incremented with sufficient frequency in the typical range of real world driving Ricardo have used a model based approach to assess this capability · Parametric model of vehicle coded in Simulink ­ · · Data used for a `typical' diesel passenger car Model of OBD monitor enable conditions for each monitor The model does not take into account the effect of changes in ambient conditions (i.e. model conditions are restricted to 20° constant altitude, and other ambient C, variables). ­ ­ · The effect of real ambient condition changes on engine operation may reduce the frequency with which monitor enable conditions can be met At some ambient conditions, in-use enablement could be significantly worse than predicted by this model

Model is run over operating profiles from Ricardo database, developed to represent different typical drive cycles:

2 November 2010 © Ricardo plc 2010

Page 17

RD.10/320201.4

­

City driving cycles

· ­ ­ · ·

London, Paris, Stuttgart, Turin

Rural drive cycle Highway drive cycle

The simulation will show whether we would expect a monitor to increment the numerator on each drive cycle, under the assumed ambient conditions. If the numerator can be incremented in all the cycles, then we would expect that the numerator will be incremented with sufficient frequency in real world use under similar ambient conditions. The following figure shows typical results that are obtained for each simulated drive cycle (city specific or rural or highway) ­ in this case the results are based on use of Manufacturer 2's vehicle in London City Cycle

·

The graph which shows the duration when the monitor is enabled is the most important one from the four graphs and therefore this is the only one that is provided in this report, together with the detailed results for each monitor where data to allow this evaluation to be carried out was available. The reason for choosing the 4 European cities is as follows: · Paris ­ ­ · well known for its severe daily congestion commuting patterns has been labelled the most congested city in Europe

London ­

2 November 2010 © Ricardo plc 2010

Page 18

RD.10/320201.4

­ · Turin ­ ­ ·

has implemented congestion charging zone representative of a large Italian city sophisticated traffic control system traffic calming in the centre of town a good public transport network congestion in Stuttgart is typical of a German city of this size

Stuttgart ­ ­ ­

2 November 2010 © Ricardo plc 2010

Page 19

RD.10/320201.4

The following parameters have been used to construct the city cycles:

Paris Total distance (km) Total time (sec) Driving time (%) Cruising time (%) Standing time (%) Avg driving speed (km/h) Avg trip speed (km/h) Maximum speed (km/h) No. of stops per km Avg. (+) accel (m/s ) Avg. (-) accel (m/s ) No. of accels per km

2 2

London 5.3 1076.5 83.2 4.72 16.8 20.9 17.4 52.9 2.37 0.651 -0.555 18

Stuttgart 5.77 1075.7 82.9 5.67 17.1 22.8 18.9 61.7 2.18 0.502 -0.473 16.4

Turin 5.7 1075.3 76.6 4.72 23.4 23.5 18.4 63.3 2.77 0.534 -0.500 13.3

10.1 1745.0 84.8 5.25 15.2 22.7 19.4 72.2 2.39 0.490 -0.451 15.7

2 November 2010 © Ricardo plc 2010

Page 20

RD.10/320201.4

Client Confidential

The graphs below show an example of typical results for a specific monitor.

2 November 2010 © Ricardo plc 2010

Page 21

RD.10/320201.4

Client Confidential European Automobile Manufacturers Association (ACEA)

A summary table is produced showing the number of times the monitor was enabled during each of the cycles considered. The following notes should be taken into consideration when interpreting these results: · The model considers monitoring enable conditions not only for enabling of the monitor, but also that the conditions are enabled for sufficient time to make a judgement. At least one judgement made on each cycle: ­ ­ ­ · The monitor can be enabled over a sufficiently wide rage of conditions We would expect it to be able to achieve minimum ratio requirements Multiple judgements improves confidence The required monitoring conditions may be too selective Minimum ratio requirements may be difficult to meet

·

No judgements on a particular drive cycle: ­ ­

2 November 2010 © Ricardo plc 2010

Page 22

RD.10/320201.4

Client Confidential

4. 4.1

TEST RESULTS Upstream Oxygen Sensors (Gasoline)

The table below gives NEDC emission results:

The tables below shows emissions for CO, NOx, NHMC, and PM components did not exceed proposed EU6 OBD emission threshold for all levels of degradation tested

2 November 2010 © Ricardo plc 2010

Page 23

RD.10/320201.4

Client Confidential

Emissions vs. 'level of degradation' 300 250 NOx [mg/km] 200 150 100 50 0 Base Degradation Slow response

Emissions vs. 'level of degradation' 2000

Manuf 1 EU6 prop OBD EU5 OBD

1500 CO [mg/km] 1000 500 0

Manuf 1 EU6 prop OBD EU5 OBD

Base Degradation

Slow response

Emissions vs. 'level of degradation' 300 250 NMHC [mg/km] 200 150 100 50 0 Base Degradation Slow response Manuf 1 EU6 prop OBD EU5 OBD

50 40 PM [mg/km] 30 20 10 0 Base

Emissions vs. 'level of degradation' Manuf 1 EU6 prop OBD EU5 OBD

Slow response Degradation

2 November 2010 © Ricardo plc 2010

Page 24

RD.10/320201.4

Client Confidential

4.1.1

Evaluation of OBD Monitor Strategy Upstream lambda sensor monitoring is achieved by inducing fuelling oscillation and measuring sensor response. A fault is set if the calculated lambda rate of response is above a calibratable threshold. The figure below shows the target lambda values for response rate thresholds measured over an NEDC.

Vehicle speed [km/h] Vehicle speed 100 50 0 0 1.2 Lambda [-] Base Slow response

200

400

600

800

1000

1200

Target lambda (bank 1)

1

0.8 0 1.2 Lambda [-]

200

400

600

800

1000

1200

Target lambda (bank 2)

1

0.8 0

200

400

600 Time [s]

800

1000

1200

2 November 2010 © Ricardo plc 2010

Page 25

RD.10/320201.4

Client Confidential

The figure below shows the lambda sensor response rate monitor parameter measured over an NEDC

Vehicle speed 100 50 0 0 2 Lambda [-] 1.5 1 0.5 0 2 Lambda [-] 1.5 1 0.5 0 200 400 600 Time [s] 800 1000 1200 200 400 600 800 1000 1200

Vehicle speed [km/h]

Base Slow response

200

400

600

800

1000

1200

Upstream O2 sensor lambda (bank 1)

Upstream O2 sensor lambda (bank 2)

2 November 2010 © Ricardo plc 2010

Page 26

RD.10/320201.4

Client Confidential

The figure below shows the lambda sensor response rate monitor output measured over an NEDC

2 November 2010 © Ricardo plc 2010

Page 27

RD.10/320201.4

Client Confidential

4.1.2

Conclusion · · · · · · The proposed EU6 EOBD emissions thresholds were not exceeded for any emissions species (NOx, CO, NMHC and PM) with slow response faults on the upstream lambda sensor A monitoring strategy (using an induced fuelling oscillation and measuring the sensor response) was capable of identifying the degraded component for upstream sensor Sensor accuracy has not been studied because of data were not available Analysis of the likely monitor enabled conditions over typical real world driving cycles has not been carried out because of data were not available The proposed EU6 OBD limits would be achievable for monitoring of lambda sensor Where the work indicates that a lower fault threshold is required then in order to set diagnostic thresholds for production vehicles, it will be necessary to consider the robustness of the monitor results on correctly functioning vehicles

4.2

Upstream & Downstream Lambda Sensors (Diesel LNT)

In this case both lambda sensors are used for after-treatment emission control. · · · Upstream lambda sensor is used for Closed-loop control of air/fuel ratio for the rich mixture required during regeneration Downstream lambda sensor is used for the Diagnosis of degraded NSC/LNT efficiency for OBD by comparing with upstream lambda measurement during regeneration Downstream lambda sensor is used to sense the end state of NSC regeneration in normal operation

The table below gives NEDC emission results.

2 November 2010 © Ricardo plc 2010

Page 28

RD.10/320201.4

Client Confidential

All emissions species remained within the type approval limits with the degraded components installed.

2 November 2010 © Ricardo plc 2010

Page 29

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 30

RD.10/320201.4

Client Confidential

4.2.1

Evaluation of OBD Monitor Strategy Analysis of the monitor capability is based on Manuf 2 existing lambda monitoring algorithm. This outlined below: · Monitor enable conditions ­ ­ · DeNOx regeneration activated Dew detection released Evaluation of lambda sensor delay time if monitoring is enabled If delay times ( t UsDs , t Ds Us) exceeded the threshold values (t thres, Us Ds , t thres, Ds Us) then the error counter is incremented A given number of increments is needed to detect slow response sensors

Monitor strategy ­ ­ ­

­

Lambda monitoring is done immediately after switching into rich mixture for NSC regeneration

2 November 2010 © Ricardo plc 2010

Page 31

RD.10/320201.4

Client Confidential

­

If the value of one lambda sensor falls below lambda threshold a timer starts. Time will be stopped if the value of the second lambda sensor also falls below lambda threshold. The passed time is a judgement parameter for lambda monitor and is compared with a given threshold time. If the delay of the slower lambda sensor is too high a lambda control fault will be set (monitoring strategy up/down counter).

­

For the purpose of this report a judgment parameter is defined by the time delay of the first regeneration because the following time delays (corresponding to the next regenerations) are influenced by the time length between the following regenerations. The following figures show analysis using this monitor judgement parameter. The first figure shows results of the baseline test, with upstream and downstream lambda sensors set to 0 % degradation

2 November 2010 © Ricardo plc 2010

Page 32

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 33

RD.10/320201.4

Client Confidential

The next figure below shows results with upstream lambda sensor degraded by 100 %

2 November 2010 © Ricardo plc 2010

Page 34

RD.10/320201.4

Client Confidential

And finally the figure below shows results with downstream lambda sensor degraded by 100 %

2 November 2010 © Ricardo plc 2010

Page 35

RD.10/320201.4

Client Confidential

All these results are summarised in the following set of graphs to illustrate performance against the proposed OBD thresholds. Calculated judgement parameter shows good separation between normal and degraded components for both upstream and downstream lambda sensors. Ability to detect faulty components on the NEDC cycle is good.

4.2.1.1

Upstream Sensor Monitor Robustness We have assumed a typical lambda sensor tolerance of up to 0.05 at lean air/fuel ratios The effect of lambda sensor tolerance on the time delay of the first regeneration can be seen below to be small compared to the separation between the normal and faulty component.

2 November 2010 © Ricardo plc 2010

Page 36

RD.10/320201.4

Client Confidential

·

The tolerance effect is much smaller than the other test to test variations in the data ­ The effect of sensor tolerance cannot be seen in the graph There is good separation between results with baseline and faulty parts

·

2 November 2010 © Ricardo plc 2010

Page 37

RD.10/320201.4

Client Confidential

­ 4.2.1.2

Monitoring should be robust to tolerance effects

Downstream Sensor Monitor Robustness We have assumed a typical lambda sensor tolerance of up to 0.05 at lean air/fuel ratios The effect of the lambda sensor tolerance on the time delay of the first regeneration can be seen below to be small compared to the separation between the normal and faulty component

2 November 2010 © Ricardo plc 2010

Page 38

RD.10/320201.4

Client Confidential

·

The tolerance effect is much smaller than the other test to test variations in the data ­ The effect of sensor tolerance cannot be seen in the graph Monitoring should be robust to tolerance effects There is good separation between results with baseline and faulty parts ­

·

4.2.2

Evaluation of Monitor IUMPR The Ricardo model was calibrated over the NEDC test cycle so that the modelled monitor enable conditions were met at the same time in the cycle as the Manuf 2 monitor is enable in the test data · · · The lambda sensor monitor is enabled only during NSC regeneration The in-use performance model was modified to simulate NSC regeneration frequency and estimated NSC regeneration enable conditions to match the NSC regeneration conditions encountered in the NEDC test data The enable conditions for lambda sensor monitoring were: ­ ­ ­ ­ ­ DeNOx regeneration activated 900 rev/min < Engine speed < 3000 rev/min Vehicle speed > 30 km/h Torque up to just below full power, torque < 240 Nm Dew detection of lambda downstream (lambda controlled DeNOx mode), enable time > 450 s · ­ ­ · Dew release is strongly dependant on general vehicle load and driving conditions Gear > 1 Time required for monitor judgement 3 s

The following figures show the results for each simulated drive cycle

2 November 2010 © Ricardo plc 2010

Page 39

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 40

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 41

RD.10/320201.4

Client Confidential

The table below summarises in-use performance ratio assessment from the above results.

CYCLE London Paris Stuttgart Turin Rural Highway

Number of Judgements 1 1 4 2 3 7

The monitor was enabled over all the conditions evaluated. We would therefore expect this monitor to achieve minimum ratio requirements robustly. Best monitoring robustness is achieved under highway driving conditions. 4.2.3 Conclusion · · · The proposed EU6 EOBD emissions threshold were not exceeded for any emissions species with slow response faults on either the upstream or downstream lambda sensors used for NSC control A monitoring strategy based on time delay measurement between lambda output signals during NSC regeneration was capable of identifying the degraded components for both upstream and downstream sensors Sensor accuracy has a minimum influence on the robustness of the suggested strategy indicating that the judgement parameter is capable of making a robust detection over a real fleet in real world conditions

2 November 2010 © Ricardo plc 2010

Page 42

RD.10/320201.4

Client Confidential

· ·

Analysis of the likely monitor enabled conditions over typical real world driving cycles suggests that the monitor is capable of meeting the legislated minimum in-use performance ratio requirements The proposed EU6 OBD limits would be achievable for monitoring of lambda sensors upstream and downstream of NSC

4.3

Downstream Lambda Sensors (Gasoline)

This refers to HEGO sensors located downstream of the catalyst for the purpose of monitoring the performance of the catalyst. The table below gives NEDC emission results:

The tables below shows emissions for CO, NOx, NHMC, and PM components did not exceed proposed EU6 OBD emission threshold for all levels of degradation tested

2 November 2010 © Ricardo plc 2010

Page 43

RD.10/320201.4

Client Confidential

Emissions vs. 'level of degradation' 300 250 NOx [mg/km] 200 150 100 50 0 Base Degradation Slow response

0 Base 2000

Emissions vs. 'level of degradation'

CO [mg/km]

Manuf 1 EU6 prop OBD EU5 OBD

1500 1000 500

Manuf 1 EU6 prop OBD EU5 OBD

Slow response Degradation

Emissions vs. 'level of degradation' 300 250 NMHC [mg/km] 200 150 100 50 0 Base Degradation Slow response Manuf 1 EU6 prop OBD EU5 OBD

Emissions vs. 'level of degradation' 50 40 PM [mg/km] 30 20 10 0 Base Degradation Slow response Manuf 1 EU6 prop OBD EU5 OBD

2 November 2010 © Ricardo plc 2010

Page 44

RD.10/320201.4

Client Confidential

4.3.1

Evaluation of OBD Monitor Strategy Downstream O2 sensor monitoring is performed by evaluating sensor response during rich to lean transitions when fuel cut is activated. A fault is set if the measured sensor response time is greater than a calibratable threshold. The figure below shows the target lambda values for response rate thresholds measured over an NEDC.

2 November 2010 © Ricardo plc 2010

Page 45

RD.10/320201.4

Client Confidential

Vehicle speed [km/h]

Vehicle speed 100 50 0 0 Base Failure catalyst

200

400

600

800

1000

1200

Catalyst monitor fuelling control active (bank 1) 1 Flag [-] 0.5 0 0

200

400

600

800

1000

1200

Catalyst monitor fuelling control active (bank 2) 1 Flag [-] 0.5 0 0

200

400

600 Time [s]

800

1000

1200

2 November 2010 © Ricardo plc 2010

Page 46

RD.10/320201.4

Client Confidential

4.3.2

Conclusion · · · · · · The proposed EU6 EOBD emissions thresholds for NOx, CO, NMHC and PM were not exceeded for all slow response faults induced in the downstream O2 sensor A monitoring strategy (evaluating sensor response during rich to lean transitions when fuel cut is activated) was capable of identifying the degraded component for downstream sensor Sensor accuracy has not been studied because data were not available Analysis of likely monitor enabled conditions over typical real world driving cycles has not been carried out because data were not available The proposed EU6 OBD limits would be achievable for monitoring of the downstream O2 sensor Where the work indicates that a lower fault threshold is required then in order to set diagnostic thresholds for production vehicles, it will be necessary to consider the robustness of the monitor results on correctly functioning vehicles

2 November 2010 © Ricardo plc 2010

Page 47

RD.10/320201.4

Client Confidential

4.4

3-Way Catalytic Converter (Gasoline)

The table below gives NEDC emission results:

The next two sets of results show that while emissions for CO and PM did not exceed proposed EU6 OBD emission threshold for all levels of degradation evaluated, nevertheless those for NOx and NMHC emissions exceeded proposed EU6 OBD emission threshold when failure cats were fitted.

2 November 2010 © Ricardo plc 2010

Page 48

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 49

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 50

RD.10/320201.4

Client Confidential

4.4.1

Evaluation of OBD Monitor Strategy The catalyst monitor operates by cycling fuelling rich and lean under certain operating conditions. Upstream and downstream oxygen sensor signals are analysed to determine the level of oxygen storage in the catalyst hence inferring catalyst efficiency

2 November 2010 © Ricardo plc 2010

Page 51

RD.10/320201.4

Client Confidential

Vehicle speed [km/h]

Vehicle speed 100 50 0 0 Base Failure catalyst

200

400

600

800

1000

1200

Catalyst monitor fuelling control active (bank 1) 1 Flag [-] 0.5 0 0 1 Flag [-] 0.5 0 0

200

400

600

800

1000

1200

Catalyst monitor fuelling control active (bank 2)

200

400

600 Time [s]

800

1000

1200

2 November 2010 © Ricardo plc 2010

Page 52

RD.10/320201.4

Client Confidential

4.4.2

Conclusion · · · · · · · · The proposed EU6 EOBD emissions thresholds for NOx and NMHC were exceeded for tests carried out with degraded catalysts Proposed EU6 EOBD emissions thresholds for CO and PM were not exceeded for all tests carried out A monitoring strategy (based on the analysis of the upstream and downstream oxygen sensor signals) was capable of identifying the degraded catalyst used in these tests Catalyst monitor accuracy has not been studied because data were not available Analysis of the likely monitor enabled conditions over typical real world driving cycles has not been carried out because data were not available A less aggressively aged catalyst (giving NOx and NMHC just below the proposed EU6 OBD threshold) was not available for testing In order to make a definitive statement about the feasibility of monitoring techniques for EU6 OBD tests, a less aggressively aged catalyst is required. Where the work indicates that a lower fault threshold is required then in order to set diagnostic thresholds for production vehicles, it will be necessary to consider the robustness of the monitor results on correctly functioning vehicles

2 November 2010 © Ricardo plc 2010

Page 53

RD.10/320201.4

Client Confidential

4.5

Misfire Monitor (Gasoline)

The table below gives NEDC emission results:

2 November 2010 © Ricardo plc 2010

Page 54

RD.10/320201.4

Client Confidential

The set of results below show that emissions for NOx and CO did not exceed proposed EU6 OBD emission threshold for all levels of degradation evaluated.

Emissions vs. 'level of degradation' 300 250 NOx [mg/km] 200 150 100 50 0 Base Misfire 2 % Degradation Misfire 6 %

0 Base Misfire 2 % Degradation Misfire 6 % Emissions vs. 'level of degradation'

CO [mg/km]

Manuf 1 EU6 prop OBD EU5 OBD

2000 1500 1000 500 Manuf 1 EU6 prop OBD EU5 OBD

Results for PM emissions were also similar in that they did not exceed proposed EU6 OBD emission threshold for all levels of degradation evaluated. However results for NMHC exceeded the proposed EOBD thresholds at the level of degradation of 6% misfire.

2 November 2010 © Ricardo plc 2010

Page 55

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 56

RD.10/320201.4

Client Confidential

PN results were also measured although the EU have not as yet proposed a limit for this in GDI applications. These results are given in the graph below.

Emissions vs. 'level of degradation' 6 5 PN [x1012/km] 4 3 2 1 0 Base Manuf 1 EU6 prop OBD (to be defined) EU5 OBD (not defined) Misfire 2 % Degradation Misfire 6 %

4.5.1

Evaluation of OBD Monitor Strategy Misfire detection is made by analysis of changes in Crankshaft Speed Fluctuations (CSF), since a misfire will cause a fall in speed after a faulty firing event. An adaptive algorithm is used to reduce engine to engine effects. Multiple methods of detection are used to improve detection capability.

2 November 2010 © Ricardo plc 2010

Page 57

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 58

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 59

RD.10/320201.4

Client Confidential

4.5.2

Conclusion · · · Misfire at 2 % and 6 % did not cause emissions to exceed proposed EU6 OBD emission threshold limits (CO, PM, NOx) NMHC emissions exceeded proposed EU6 OBD emission threshold with a 6 % misfire rate A misfire monitor strategy based on the analysis of changes in crankshaft speed was shown to be capable of identifying the 6 % misfire fault. No fault detection was possible in the case of 2 % misfire because this was below the calibrated threshold (although misfires were correctly identified) The results suggest that a monitor calibration target of less than 6 % misfire rate may be necessary. This increases the risk of false fault detection, however, misfire monitor accuracy and robustness under normal driving conditions has not been studied in this work. PN emissions are not sensitive to this failure mode

·

·

2 November 2010 © Ricardo plc 2010

Page 60

RD.10/320201.4

Client Confidential

4.6

Fuel System Monitor (Gasoline)

The table below gives NEDC emission results:

2 November 2010 © Ricardo plc 2010

Page 61

RD.10/320201.4

Client Confidential

Results below show CO emissions exceeded proposed EU6 OBD emission threshold when a rich fuelling fault is induced. PM and NOx emissions did not exceed proposed EU6 OBD emission threshold limit under both lean and rich fuelling conditions.

2 November 2010 © Ricardo plc 2010

Page 62

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 63

RD.10/320201.4

Client Confidential

Although NMHC emissions did not exceed proposed EU6 OBD emission threshold limit under both lean and rich fuelling conditions, nevertheless the results were very close to the limit under rich conditions ­ this requires further investigation to ensure robustness of any monitoring techniques. Again PN results are included for completeness ­ but no thresholds are defined yet.

4.6.1

Evaluation of OBD Monitor Strategy Fuel system monitoring strategy is based on analysis of the long-term fuel system adaptations. If long-term adaptations exceed the calibrated thresholds for a pre-determined time then a fault is set. Separate thresholds are calibrated for rich and lean sided faults. The figures below show examples of Fuel System monitor performance:

2 November 2010 © Ricardo plc 2010

Page 64

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 65

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 66

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 67

RD.10/320201.4

Client Confidential

4.6.2

Conclusion The following general trends were observed for the lean and rich air/fuel mixture faults: · · · · · · · · · CO emissions increase when the rich air/fuel mixture fault occurs, and exceed proposed EU6 OBD threshold limits There is no significant change in NOx and PM emissions when either rich or lean faults are introduced, in comparison to the base configuration NMHC emissions increase when the rich air/fuel mixture fault occurs Emissions of NOx, NMHC and PM did not exceed proposed EU6 OBD thresholds for either rich or lean air/fuel mixture faults The monitoring strategy based on analysis of long-term fuel system adaptations is capable of identifying lean and rich fuel system malfunctions Monitor accuracy and robustness over a wide range of typical real world driving cycles has not been analyzed due to data not being available Based on the results in this report, however, the proposed EU6 OBD limits would be achievable for fuel system monitoring based on analysis of long-term fuel system adaptations PN emissions are not sensitive to rich mixture malfunctions tested Where the work indicates that a lower fault threshold is required then in order to set diagnostic thresholds for production vehicles, it will be necessary to consider the robustness of the monitor results on correctly functioning vehicles

2 November 2010 © Ricardo plc 2010

Page 68

RD.10/320201.4

Client Confidential

4.7

VVT System Monitor (Gasoline)

The table below gives NEDC emission results:

NMHC emissions exceed proposed EU6 OBD emission threshold when inlet VVT cam is stuck at 100 % . CO emissions did not exceed proposed EU6 OBD emission threshold limit under all levels of degradation for VVT cam. The CO emissions are quite high when 100 % stuck degradation is introduced, although less than the proposed OBD threshold. NOx and PM emissions did not exceed proposed EU6 OBD emission threshold limits under all levels of degradation for VVT cam

2 November 2010 © Ricardo plc 2010

Page 69

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 70

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 71

RD.10/320201.4

Client Confidential

4.7.1

Evaluation of OBD Monitor Strategy If the cam position error is higher than a threshold then a time counter is incremented. When the time counter value exceeds a calibrated threshold (set to 10 seconds presently) then the VVT monitor detects a malfunctioning system. The first set of results show performance of VVT monitor when valve is stuck open at 100%

Note that both banks detect a fault during the pre-conditioning cycle.

2 November 2010 © Ricardo plc 2010

Page 72

RD.10/320201.4

Client Confidential

The next set of results below show performance of VVT monitor when valve is stuck open at 0%. In this case the fault gets detected within 500 seconds of the NEDC test cycle.

The final set of results show the effect of a slow actuator response malfunction. The monitor was able to detect the fault on bank 1 in 726 seconds, but the fault was not detected on Bank 2.

2 November 2010 © Ricardo plc 2010

Page 73

RD.10/320201.4

Client Confidential

4.7.2

Conclusion · · · · Faults caused by the VVT slow response and VVT stuck 0 % do not cause emissions to exceed proposed EU6 OBD threshold limits for all emissions (NOx, CO, NMHC, PM) Fault caused by the VVT stuck in 100 % position causes NHMC emissions to exceed proposed EU6 OBD threshold limit A change to the monitor detection threshold may be all that is required. However this needs to be verified A VVT monitor strategy is inherently robust because it is based on a direct measurement of cam position error. However monitor accuracy has not been studied in this report because data were not available

2 November 2010 © Ricardo plc 2010

Page 74

RD.10/320201.4

Client Confidential

· · ·

Analysis of the likely monitor enabled conditions over typical real world driving cycles has not been done because of data were not available A VVT monitoring strategy based on the analysis of cam position error is capable of detecting faults to the level of proposed EU6 OBD thresholds Where the work indicates that a lower fault threshold is required then in order to set diagnostic thresholds for production vehicles, it will be necessary to consider the robustness of the monitor results on correctly functioning vehicles

4.8

MAF Sensor Monitor (Gasoline)

The table below gives NEDC emission results:

2 November 2010 © Ricardo plc 2010

Page 75

RD.10/320201.4

Client Confidential

The results are plotted below showing that NOx, NMHC, PM and CO emissions did not exceed proposed EU6 OBD emission threshold limit for all levels of degradation evaluated (base and sensor reading drifted high).

Emissions vs. 'level of degradation' 300 250 NOx [mg/km] 200 150 100 50 0 Base Degradation Emissions vs. 'level of degradation' Rich Manuf 1 EU6 prop OBD EU5 OBD

250 NMHC [mg/km] 200 150 100 50 0 Base Degradation Rich Emissions vs. 'level of degradation' Manuf 1 EU6 prop OBD EU5 OBD

Emissions vs. 'level of degradation'

2000 1500 CO [mg/km] 1000 500 0 Manuf 1 EU6 prop OBD EU5 OBD

50 40 PM [mg/km] 30 20 10

Manuf 1 EU6 prop OBD EU5 OBD

Base Degradation

Rich

0

Base Degradation

Rich

2 November 2010 © Ricardo plc 2010

Page 76

RD.10/320201.4

Client Confidential

4.8.1

Evaluation of OBD Monitor Strategy MAF sensor monitoring is achieved by comparing the measured airflow with an estimated airflow which is calculated from a model. A fault is detected if the measured value deviates from the modelled value by more than a calibratable threshold

Engine air flow (from air charge estimation) 70 60 Air flow [g/s] 50 40 30 20 10 0 0 200 400 600 Vehicle speed 120 Vehicle speed [km/h] 100 80 60 40 20 0 0 200 400 600 Time [s] 800 1000 1200 800 1000 1200 Base Rich

2 November 2010 © Ricardo plc 2010

Page 77

RD.10/320201.4

Client Confidential

4.8.2

Conclusion · · · · · · All pollutants are insensitive to a rich airflow malfunction on the MAF sensor A monitoring strategy based on comparing the measured airflow with an estimated (modelled) airflow is capable of detecting a drift in MAF sensor signal reading high by 30% or more Sensor accuracy has not been studied because data were not available In order to fully evaluate the capability of MAF sensor monitoring techniques with regard to their feasibility for detecting proposed EU6 OBD limits, data covering a wider range of failure needs to be analysed (e.g. MAF sensor signal drifting low) However based on the results presented in this report, MAF sensor malfunctions to the level proposed in EU6 OBD can be diagnosed using model-based diagnostic strategies Where the work indicates that a lower fault threshold is required then in order to set diagnostic thresholds for production vehicles, it will be necessary to consider the robustness of the monitor results on correctly functioning vehicles

2 November 2010 © Ricardo plc 2010

Page 78

RD.10/320201.4

Client Confidential

4.9

Exhaust Gas Recirculation (Diesel)

The tables below give NEDC emission results from manufacturer 3 that carried out EGR flow monitor tests.

The main emission species of interest are NOx for EGR low flow and CO and NMHC for EGR high flow. PM is not the main focus here because the DPF is not damaged. · · · NOx emissions were very close to proposed EU6 EOBD threshold with 9 mm EGR flow restrictor CO emissions are well below proposed EU6 EOBD threshold NMHC emissions are well below proposed EU6 EOBD threshold. There is overall a slight increase in NHMC with EGR high flow, although the results are not consistent over both test iterations

2 November 2010 © Ricardo plc 2010

Page 79

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 80

RD.10/320201.4

Client Confidential

4.9.1 4.9.1.1

Evaluation of EGR OBD Monitor Strategy Manufacturer 2 Details of the Manuf 2 EGR flow monitoring strategies were not available and monitor parameters were not logged during the testing · · · Manuf 2 have previously certified the same EGR monitor to detect the levels of malfunction tested in the program Both the EGR high and EGR low flow monitors successfully detected the degraded components tested during this program Tailpipe NOx emissions do not increase significantly until NSC overload occurs: ­ ­ ­ Significant levels of EGR flow reduction can be tolerated before the OBD limits are exceeded Monitoring of the EGR flow rate at this level over the NEDC is possible to detect Real world emissions may be considerably worse since the NEDC is mainly low speed and load operation · The NSC will become overloaded sooner at higher speed load conditions

4.9.1.2

Manufacturer 3 Analysis of the monitor capability based on Manuf 3 existing EGR high and low flow monitoring algorithms is outlined below · EGR Low flow: ­ ­ Monitoring using mass air flow sensor Judgement parameter is difference between desired and actual air mass flow

2 November 2010 © Ricardo plc 2010

Page 81

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 82

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 83

RD.10/320201.4

Client Confidential

·

EGR High Flow ­ ­ Monitoring using mass air flow sensor during overrun Judgement parameter is ratio of modelled air flow (using intake air pressure and temperature, engine speed and volume) to measured mass air flow from sensor

2 November 2010 © Ricardo plc 2010

Page 84

RD.10/320201.4

Client Confidential

Four steady state speed and load conditions of the NEDC cycle were considered for evaluating the performance both the low flow and the high flow monitors: · · Vehicle speed stable conditions defined by Ricardo: a), b), c) and d) These conditions do not necessarily reflect Manuf 3 current calibration of monitor enable conditions but were chosen to assess the potential for the monitor to be calibrated in these condition

The figures below show EGR low flow monitor judgment parameter versus NEDC NOx at each steady state condition

2 November 2010 © Ricardo plc 2010

Page 85

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 86

RD.10/320201.4

Client Confidential

In the figure below the low flow monitor judgment parameter versus NEDC NOx is presented showing results from only three steady state conditions a), c) and d) from the previous figure Condition b) is excluded as the fault could not be distinguished and a false pass would be possible. There is good separation between 9 mm and 14 mm restrictors to allow detection at the EU6 threshold at conditions a), c) and d)

2 November 2010 © Ricardo plc 2010

Page 87

RD.10/320201.4

Client Confidential

The air mass governor deviation was generally below the currently calibrated lower threshold limit of ­100 mg/stroke when the monitor was enabled during the NEDC cycle with 9 mm restrictor. The monitor would have detected the fault during this test with the current calibration

2 November 2010 © Ricardo plc 2010

Page 88

RD.10/320201.4

Client Confidential

The EGR high flow monitor is enabled during overrun fuel cut-off with a warm engine. It is enabled three times during NEDC: conditions a), b) and c). The results are shown below:

2 November 2010 © Ricardo plc 2010

Page 89

RD.10/320201.4

Client Confidential

The figure below shows the EGR high flow monitor judgment parameter (ratio) during conditions (a) and (b) of NEDC:

2 November 2010 © Ricardo plc 2010

Page 90

RD.10/320201.4

Client Confidential

The figure below shows the EGR high flow monitor judgment parameter (ratio) during conditions (c) of NEDC:

2 November 2010 © Ricardo plc 2010

Page 91

RD.10/320201.4

Client Confidential

The figures below show EGR high flow monitor judgment parameter versus NEDC NOx at each steady state condition, giving Mean and standard deviation of judgment parameters at each separate occurrence of monitor enabling. Results shown are from 1st iteration of testing. Emissions measurement was inconsistent but significantly below EOBD limit

2 November 2010 © Ricardo plc 2010

Page 92

RD.10/320201.4

Client Confidential

4.9.1.3

Monitor Robustness The following tolerance data was provided by Manuf 3 for the OBD monitor input parameters · EGR low flow monitor System Air Flow Sensor Engine Displacement Volume Tolerance 7% 5%

·

EGR high flow monitor System Air Flow Sensor Engine Displacement Volume Intake Manifold Air Temperature Sensor Intake Manifold Pressure Sensor Tolerance 7% 5% 2% 2%

The next figure is a repeat of the results presented earlier, but with tolerance bands applied

2 November 2010 © Ricardo plc 2010

Page 93

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 94

RD.10/320201.4

Client Confidential

· ·

Plot below shows tolerance analysis of mean value of air mass governor deviation for a vehicle speed of 32 km/h Assuming that the Worst Performing Acceptable (WPA) is at the type approval emissions limit and the Best Performing Unacceptable (BPU) at the EU6 emissions limit, there is overlap between the upper and lower envelopes of the judgement parameter There is no overlap between the BPU and the lower tolerance envelope of the baseline data Presented monitor strategy is capable to detect low flow at the EU6 emissions limit but there may be risk of early MIL activation for a partially degraded system, at less than the type approval threshold, with components near to limit tolerances

· ·

2 November 2010 © Ricardo plc 2010

Page 95

RD.10/320201.4

Client Confidential

· · ·

Plot below shows mean value and error bars based on the given tolerances for High Flow EGR At the maximum EGR high flow fault tested, there is significant overlap between the Best Performing Unacceptable (BPU) and Worst Performing Acceptable (WPA) The maximum EGR high flow fault that could be tested without risk of clogging DPF was well below the proposed EU6 EOBD thresholds for all emissions species

2 November 2010 © Ricardo plc 2010

Page 96

RD.10/320201.4

Client Confidential

4.9.2 4.9.2.1

Evaluation of Monitor IUMPR Low Flow EGR Monitor Ricardo IUMPR model was calibrated to monitor enable conditions similar to Manuf 3 EGR low flow monitor in the test data · · · Engine speed > 1000 rev/min Accelerator pedal position > 1.5 % Steady state operation : defined as follows ­ ­ · engine speed rate of change within ±40 RPM/s accelerator pedal rate of change < 5 %/s

Monitor enable conditions met for a continuous (un-interrupted) time of 5 seconds in order to make a judgement

2 November 2010 © Ricardo plc 2010

Page 97

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 98

RD.10/320201.4

Client Confidential

The table below summarises EGR Low-Flow in-use performance ratio assessment from the above results.

CYCLE London Paris Stuttgart Turin Rural Highway

Number of Judgements 23 18 8 12 19 46

The monitor was enabled several times over all the conditions evaluated. We would therefore expect this monitor to achieve minimum ratio requirements robustly. Best monitoring robustness is achieved under highway driving conditions. 4.9.2.2 High Flow EGR Monitor The following monitor enable conditions were programmed into the Ricardo IUMPR Model to analyse Manuf 3's strategy: · · Coolant temperature > 80 degC In overrun ­ ­ Engine speed > 1100 rev/min Accelerator pedal position < 1.5 %

2 November 2010 © Ricardo plc 2010

Page 99

RD.10/320201.4

Client Confidential

·

Monitor enable conditions met for a continuous (un-interrupted) time of 5 seconds in order to make a judgment

The following histograms show the high flow EGR monitor enable profiles for Rural and Highway drive cycles:

Corresponding profiles for city driving are given on the following page for the 4 representative European cities.

2 November 2010 © Ricardo plc 2010

Page 100

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 101

RD.10/320201.4

Client Confidential

The table below summarises EGR High-Flow in-use performance ratio assessment from the above results.

CYCLE London Paris Stuttgart Turin Rural Highway

Number of Judgements 14 12 10 11 10 14

The monitor was enabled several times over all the conditions evaluated. We would therefore expect this monitor to achieve minimum ratio requirements robustly. Best monitoring robustness is achieved under highway driving conditions.

4.9.3 4.9.3.1

Conclusion Low Flow EGR Monitor · For Manuf 3 the proposed EU6 EOBD emissions threshold of NOx was exceeded with the low EGR flow system degraded with a restrictor of less than 9 mm diameter.

2 November 2010 © Ricardo plc 2010

Page 102

RD.10/320201.4

Client Confidential

·

A monitor judgment parameter using MAF, intake air temperature and pressure was shown to be capable of distinguishing the baseline result (without a malfunction) from the proposed EU6 EOBD emissions threshold result over the NEDC test. ­ ­ Analysis of likely tolerances on the judgement parameter suggests that the judgement parameter is capable of making a robust distinction at the proposed EU6 EOBD threshold over a real fleet in real world conditions There may be a risk of early MIL activation at close to the type approval NOx emissions level when component tolerances are close to the limits

· ·

Analysis of likely monitor enable conditions over typical real world driving cycles suggests that the EGR low flow monitor is capable of meeting the legislated minimum in-use performance ratio requirements The proposed EU6 OBD limit of 140 mg/km NOx would be achievable for EGR monitoring using difference between desired and actual air mass flow

4.9.3.2

High Flow EGR Monitor · The proposed EU6 EOBD emissions thresholds were not exceeded for any emissions species with the maximum level of EGR high flow degradation evaluated. ­ · · Higher levels of EGR flow could not be tested due to risk of damage (clogging) of the DPF A monitor judgement parameter using the MAF, intake air temperature and pressure was shown to be capable of distinguishing the baseline result from the proposed EU6 EOBD emissions threshold result over the NEDC test Analysis of likely tolerances on the judgement parameter suggests that the judgement parameter is not capable of making a robust distinction between the level of degradation tested and the baseline system ­ ­ · Higher levels of EGR degradation may be robustly detectable High EGR flow may be indirectly monitored by detection of clogging in the DPF system, as this was the limiting factor during the testing

Analysis of the likely monitor enabled conditions over typical real world driving cycles suggests that the Manuf 3 EGR high flow monitor strategy is capable of meeting the legislated minimum in-use performance ratio requirements

2 November 2010 © Ricardo plc 2010

Page 103

RD.10/320201.4

Client Confidential

Both EGR low and high flow monitors were capable of detecting the levels of degradation tested. The monitor strategies used have been previously certified and are in production with this level of threshold detection. For the Manuf 2 system (EGR with NSC and DPF after-treatment), monitoring of the EGR system for high and low flow to the proposed EU6 thresholds is feasible with current monitoring strategies and technology

4.10 EGR Cooler / ByPass (Diesel)

The table below gives NEDC emission results.

The proposed EU6 EOBD emissions threshold for NOx was not exceeded even with an EGR cooler/bypass major failure

2 November 2010 © Ricardo plc 2010

Page 104

RD.10/320201.4

Client Confidential

The figure below is a zoom-in of the above graph to show the increase in emissions between the baseline (no malfunction) and the totally failed cooler.

2 November 2010 © Ricardo plc 2010

Page 105

RD.10/320201.4

Client Confidential

4.10.1 Evaluation of OBD Monitor Strategy Details of an EGR Cooler flow monitoring strategies were not provided and therefore monitor capabilities were not assessed. The results presented are only an assessment of the impact of a major functional failure on NOx emissions on the specific vehicle tested by Manuf 3 4.10.2 Conclusion · · The proposed EU6 EOBD emissions threshold for NOx was not exceeded even with an EGR cooler/bypass major failure For the vehicle tested, monitoring of the EGR cooler would not be required to meet the proposed EU6 EOBD limits as emissions do not exceed the OBD threshold

2 November 2010 © Ricardo plc 2010

Page 106

RD.10/320201.4

Client Confidential

·

Feasibility of a monitoring strategy to detect major functional failure of the EGR cooler has not been assessed as part of this report

4.11 NSC/LNT Monitor (Diesel)

The table below gives NEDC emission results.

NOx emissions were very close to proposed EU6 EOBD threshold at 95 % degradation of NSC CO emissions exceeded the proposed EOBD threshold with the 95 % degraded NSC. Both CO and NMHC emissions also showed significant increases with degraded NSC.

2 November 2010 © Ricardo plc 2010

Page 107

RD.10/320201.4

Client Confidential

Emissions vs. 'level of degradation' 600 500 NOx [mg/km] 400 300 200 100 0 0 20 40 60 Degradation [%] 80 100 Manuf 2 EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

Emissions vs. 'level of degradation' (with hot starts) 600 500 NOx [mg/km] 400 300 200 100 0 0 20 40 60 Degradation [%] 80 100 Manuf 2 EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

Emissions vs. 'level of degradation' 350 300 NMHC [mg/km] 250 200 150 100 50 0 0 20 40 60 Degradation [%] 80 100 Manuf 2 EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

Emissions vs. 'level of degradation' 2000 1500 CO [mg/km] 1000 500 0 0 Manuf 2 EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

20

40 60 Degradation [%]

80

100

2 November 2010 © Ricardo plc 2010

Page 108

RD.10/320201.4

Client Confidential

4.11.1 Evaluation of OBD Monitor Strategy Analysis of the monitor capability is based on Manuf 2 existing NSC monitoring algorithm. Details of the monitoring strategy provided by Manuf 2 are given on the following pages. The basic functionality can be summarized as follows: · · · · Regeneration using rich fuelling mixture entering the NSC The NSC efficiency is calculated from the ratio between `slipped rich fuelling mixture' and `supported rich fuelling mixture'. This calculation is based on a measurement of upstream and downstream lambda values. If the NSC becomes less efficient the ratio value increases The value of this ratio is used as the judgement parameter of NSC efficiency

Calibration parameters were tuned to give the best monitor judgement possible with the available data · Monitor enable conditions ­ ­ · DeNOx regeneration activated `Reageant consumption' greater than a threshold Evaluation of rSlip2Supply if monitoring is enabled If rSlip2Supply exceeds a threshold rSlip2Supply, thres then an error counter is incremented The threshold rSlip2Supply, thres depends on the exhaust gas temperature

Monitor strategy ­ ­ ­

2 November 2010 © Ricardo plc 2010

Page 109

RD.10/320201.4

Client Confidential

Reagant consumption

Supplied reagant

Reagant slip

For the purpose of this report a judgement parameter is defined by the ratio value 3 s after the first starting regeneration This is done to allow a consistent comparison between different tests, since the ratio value is not constant throughout regeneration and the length of each regeneration event is not consistent across tests The following figures show analysis using this monitor judgement parameter. The first figure shows results of the test, with 1 % degradation

2 November 2010 © Ricardo plc 2010

Page 110

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 111

RD.10/320201.4

Client Confidential

The next figure shows results with 60% degradation

2 November 2010 © Ricardo plc 2010

Page 112

RD.10/320201.4

Client Confidential

And finally the figure below shows results of a component degraded by 95%

2 November 2010 © Ricardo plc 2010

Page 113

RD.10/320201.4

Client Confidential

Plot below shows calculated judgement parameter versus NOx emissions as NSC is degraded. As NSC is degraded there is a clear trend and a significant separation between the result at the type approval threshold and the results obtained at the OBD threshold.

Judgement parameter vs. emissions

0.7 0.6 0.5

ratio r Slip2S ply [-]

0.4 0.3 0.2 0.1 0 0

NSC - 1 % degradation NSC - 60 % degradation NSC - 95 % degradation EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

20

40

60

80

NOx [mg/km]

100

120

140

160

2 November 2010 © Ricardo plc 2010

Page 114

RD.10/320201.4

Client Confidential

Plot below shows calculated judgement parameter versus CO emissions as NSC is degraded. As NSC is degraded there is a clear trend and a significant separation between the result at the type approval threshold and the results obtained at the OBD threshold

Judgement parameter vs. emissions

0.7 0.6 0.5

ratio r Slip2S ply [-]

0.4 0.3 0.2 0.1 0 0

NSC - 1 % degradation NSC - 60 % degradation NSC - 95 % degradation EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

200

400

600

CO [mg/km]

800

1000

1200

2 November 2010 © Ricardo plc 2010

Page 115

RD.10/320201.4

Client Confidential

4.11.1.1

Monitor Robustness · · · · Ricardo experience is a typical lambda sensor tolerance of up to 0.01 at stoichiometric and rich air/fuel ratios Figure shows variations of judgement parameter when recalculated for upper and lower tolerance limits of both lambda sensors Distinction is still possible between worst performing acceptable and best performing unacceptable parts after taking into account sensor tolerances The minimal estimated damage rate that could be detected robustly is 93 % and it corresponds to 128 mg/km NOx emissions

Judgement parameter vs. emissions

0.8 0.6 0.4 0.2 0 0

ratio r Slip2S ply [-]

NSC - 1 % degradation NSC - 60 % degradation NSC - 95 % degradation EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD

20

40

60

80

NOx [mg/km]

100

120

140

160

2 November 2010 © Ricardo plc 2010

Page 116

RD.10/320201.4

Client Confidential

4.11.2 Evaluation of Monitor IUMPR The following parametric information were used run the Ricardo IUMPR model · · · The NSC monitor is enabled only during NSC regeneration The in-use performance model was modified to simulate NSC regeneration frequency and estimated NSC regeneration enable conditions to match the NSC regeneration conditions encountered in the NEDC test data The enable conditions for NSC regeneration and monitoring were: ­ ­ ­ ­ ­ ­ ­ 900 rev/min < Engine speed < 3000 rev/min Vehicle speed > 30 km/h Torque up to just below full power, torque < 240 Nm Dew detection of lambda downstream (lambda controlled DeNOx mode), enable time > 450 s Gear > 1 Regeneration duration > 4 s Time required for monitor judgement 3 s

The results here are exactly identical to those under Lambda Sensor monitor given in Section 4.2.2. Therefore the same conclusion can be drawn regarding robustness of the monitor.

2 November 2010 © Ricardo plc 2010

Page 117

RD.10/320201.4

Client Confidential

4.11.3 Conclusion · · · · · · The proposed EU6 EOBD threshold for NOx was reached (although the test results did not exceed the limit) with the NSC degraded to 95 % The proposed EU6 EOBD threshold for CO was exceeded with the NSC degraded to 95 % The raw emissions are below the proposed Euro 6 OBD threshold (see page 106) A monitor judgement parameter based on the NSC efficiency using the upstream and downstream lambda measurement is capable of distinguishing the baseline result from the proposed EU6 EOBD emissions threshold result over the NEDC test Analysis of likely tolerances on the judgement parameter suggests that the judgement parameter is capable of making a robust distinction at the proposed EU6 EOBD threshold over a real fleet in real world conditions Analysis of expected monitor enable conditions suggests that monitor enable conditions should be met over a wide range of driving conditions and so the monitor should be capable to meet the minimum in-use performance ratio requirements

4.12 Oxidation Catalyst (Diesel)

Only NMHC and CO emissions are considered in this case. The table below gives NEDC emission results.

2 November 2010 © Ricardo plc 2010

Page 118

RD.10/320201.4

Client Confidential

The table below shows NMHC emissions versus degradation level

4.12.1 Evaluation of OBD Monitor Strategy Analysis of the monitor capability is based on Manuf 3 existing DOC monitoring algorithm. The basic functionality of the monitoring strategy provided by Manuf 3 is summarized below. · · DOC provides oxidation of HC's during normal operation and is used to provide high temperature feed-gas to the DPF during regeneration. Monitor judgement is based on a comparison of temperature sensors at DOC inlet and outlet during DPF regeneration, where a significant temperature rise is expected at the DOC outlet.

2 November 2010 © Ricardo plc 2010

Page 119

RD.10/320201.4

Client Confidential

A functional diagram of the DOC monitor strategy is given below

2 November 2010 © Ricardo plc 2010

Page 120

RD.10/320201.4

Client Confidential

An example showing the monitor output during a regeneration of the DPF is illustrated in the figure below.

2 November 2010 © Ricardo plc 2010

Page 121

RD.10/320201.4

Client Confidential

The next 2 pages show examples of typical monitor results operating over an NEDC, first with a normal and then with a failed catalyst

2 November 2010 © Ricardo plc 2010

Page 122

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 123

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 124

RD.10/320201.4

Client Confidential

The figure below demonstrates the monitor capability over NEDC cycle.

2 November 2010 © Ricardo plc 2010

Page 125

RD.10/320201.4

Client Confidential

4.12.1.1

Monitor Robustness

The following tolerance data was provided by Manuf 3 for the DOC monitor input parameters System Tolerance (estimates) Temperature Sensor (Ups/Dns of DOC) Air Massflow Sensor Fuel Injection Mass Unburnt Fuel in Exhaust Gas 5% 7% 10% 20%

The tolerances were applied as follows: · Measured heat calculation (numerator) ­ ­ ­ · Air flow sensor Temperature sensor (upstream of DOC) Temperature sensor (downstream of DOC) Fuel injection mass Unburned fuel

Simulated heat calculation (denominator) ­ ­

Since the monitor is using temperature difference, the absolute temperature sensor tolerance for each sensor was calculated from the normal temperature during regeneration

2 November 2010 © Ricardo plc 2010

Page 126

RD.10/320201.4

Client Confidential

· · ·

The charts below show the results of the Ricardo tolerance analysis compared to the Manuf 3 tolerance analysis, showing the tolerance range of the judgement parameter compared to the emissions results The results are not identical, however are very similar and both confirm that there is sufficient separation in the judgement parameter between the normal and degraded DOC system Differences are probably due to the estimation of the absolute temperature sensor tolerance being slightly different

2 November 2010 © Ricardo plc 2010

Page 127

RD.10/320201.4

Client Confidential

4.12.2 Evaluation of Monitor IUMPR In order to assess capability of any OBD monitoring strategy, it is necessary to consider whether enable conditions required to make a judgement will be met with sufficient frequency in the typical range of real world driving · DOC monitoring only takes place during DPF regeneration

In case of DOC, the denominator may wait 800 km in between incrementing · To meet the target minimum in-use performance ratio of 0.336, monitoring is needed only once every 2400 km

Therefore, meeting the minimum in-use performance ratio should be possible provided that: · · The frequency of needing to regenerate the DPF is not very low There is a good chance for monitoring to be completed during each regeneration ­ ­ Regeneration is not easily interrupted Monitoring is enabled in most regeneration conditions

The monitor appears to complete a judgement during the 3x UDC parts of NEDC test, when vehicle speed is low and stationary · This gives good confidence that monitoring complete frequently in real world use

4.12.3 Conclusion · · · · The proposed EU6 EOBD emission threshold for NMHC was not exceeded with the total removal of the DOC A monitor judgement using the temperature sensors at DOC inlet and outlet during DPF regeneration was capable of distinguishing the baseline result from the total failure result over the NEDC test for the vehicle tested Analysis of likely tolerances suggests that the monitor would be capable of making a robust distinction between the baseline failure and the total failure The proposed EU6 OBD limit of 140 mg/km NMHC was achievable for DOC monitoring for the vehicle tested here, however:

2 November 2010 © Ricardo plc 2010

Page 128

RD.10/320201.4

Client Confidential

­ ­ ­

This corresponded to total failure of the DOC because tailpipe emissions were below 140 mg/km in this condition The spread of emissions results suggests that a partially failed DOC could not be robustly detected for this vehicle Data from vehicles with different after-treatment layouts and engine out emissions would need to be assessed, at DOC degradation levels in between normal and total failure in order to confirm the feasibility of a 140 mg/km limit for all vehicles

·

Analysis of the likely monitor enable conditions over typical real world driving cycles suggests that the DOC monitor is capable of meeting the legislated minimum in-use performance ratio requirements

4.13 Particulate Filter (Diesel)

The table and the two graphs below give NEDC emission results obtained by Manufacturer 3

The figures below show PM and PN increase due to DPF degradation

2 November 2010 © Ricardo plc 2010

Page 129

RD.10/320201.4

Client Confidential

The table and figures below give NEDC emission results obtained by Manufacturer 4

2 November 2010 © Ricardo plc 2010

Page 130

RD.10/320201.4

Client Confidential

Results obtained by Manufacturer 5 are given in the table and figures below. Emission test results are shown for NEDC test cycles from both cold and hot starts

2 November 2010 © Ricardo plc 2010

Page 131

RD.10/320201.4

Client Confidential

Note that NOx emissions in the above table are above the Euro 6 threshold for all tests. This is because the engine under test is certified to Euro 4 and so NOx is not considered in this analysis.

2 November 2010 © Ricardo plc 2010

Page 132

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 133

RD.10/320201.4

Client Confidential

4.13.1 Evaluation of Manuf 3 DPF Monitor Strategy The basic functionality of the monitoring strategy provided by Manuf 3 is summarized below. · Differential pressure measured directly by the ECU using a sensor connected immediately upstream and downstream of the DPF

An example showing the monitor output is illustrated in the figure below.

2 November 2010 © Ricardo plc 2010

Page 134

RD.10/320201.4

Client Confidential

The set of figures on the next 4 pages illustrate the operation of the monitor functionality over an NEDC cycle.

2 November 2010 © Ricardo plc 2010

Page 135

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 136

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 137

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 138

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 139

RD.10/320201.4

Client Confidential

The OBD judgment parameter defined as the ratio of measured to simulated values of soot is shown in the figure below vs. PM emissions.

2 November 2010 © Ricardo plc 2010

Page 140

RD.10/320201.4

Client Confidential

4.13.1.1

Monitor Robustness

The following tolerances were provided by Manuf 3 for the inputs into the monitoring strategy System Differential Pressure Sensor Tolerances <5%: ±11-17 hPa depending on sensor type (@ 10 ­ 85 °C, 0-750 hPa) partially compensated by drift adaption 10% (estimation. Involved components: HFM, T5, T6, injection mass) 25% (estimation) <5%

Exhaust Gas Volume Flow Model Engine-out Particulate Matter Model Unrecognised CRT effects

Manuf 3 approach: The Gaussian failure calculation for failures with statistical distribution was applied as follows:

2 November 2010 © Ricardo plc 2010

Page 141

RD.10/320201.4

Client Confidential

The charts below show the tolerance impact on the judgement parameter using the data provided by Manuf 3 · · Left plot: Absolute tolerances

Right plot: Gaussian statistical analysis by Manuf 3

For the purpose of identifying the best achievable OBD threshold, the analysis will continue using the Gaussian statistical method

2 November 2010 © Ricardo plc 2010

Page 142

RD.10/320201.4

Client Confidential

The charts below show Gaussian tolerance analysis for the monitor judgement results versus PM and PN emissions. The following labels are used BPU = Best Performing Unacceptable WPA = Worst Performing Acceptable Robust monitoring can be achieved where there is no overlap between BPU and WPA. Based on this analysis the graphs show monitoring capability to detect particulate emissions levels for this vehicle of: Particulate Matter (PM): Particulate Matter (PM): 33.4 mg/km 33.3 % Particulate Number (PN): 41.3x1012 parts/km Particulate Number (PN): 18.2 % which is equivalent to a particulate reduction rate of:

2 November 2010 © Ricardo plc 2010

Page 143

RD.10/320201.4

Client Confidential

4.13.2 Evaluation of Manuf 4 & Manuf 5 DPF Monitor Strategy Monitor strategy assessed by both Manuf 4 and Manuf 5 were based on installation of a Particulate sensor located downstream of the DPF. Both manufacturers used prototype resistive-type Particulate Matter sensor technology in the recommended position in the exhaust for homogeneous soot dispersion. Sensor current increases as particulate matter accumulates on the sensor electrodes The particulate sensor is still under development and so a mature monitoring strategy is not available. Any monitor judgement of the soot level downstream of the DPF (and hence failure of the DPF) would be based on the sensor response time: · Time taken for the sensor current to reach regeneration threshold ­ ­ · After every engine start the sensor must first reach it's dew point and then it has to be regenerated (as preconditioning) before a measurement can be made The sensor is then regenerated by an internal heating element to burn off the accumulated particulate matter. After the regeneration, the sensor is ready to accumulate particulate matter and possibly make another judgement

The "response time" of the sensor decreases as DPF leakage increases

Key Particulate Matter (PM) Sensor Characteristics are illustrated on the figure below with respect to a cold start NEDC.

2 November 2010 © Ricardo plc 2010

Page 144

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 145

RD.10/320201.4

Client Confidential

The key criteria for OBD monitoring is that soot accumulation on the sensor reaches the regeneration threshold at least once within the NEDC test · · This must include the time for the sensor to reach it's dew point and complete the preconditioning regeneration as well as the actual sensor response time Ideally, two soot accumulation and regeneration events would be preferred within the NEDC test, to prevent false MIL activation

Data was provided from 2 sensor locations in the exhaust system as part of ongoing investigations by Manuf 4 to find the best position. The two locations are indicated on the schematic below.

· ·

Location S1 - Just after DPF Location S2 - Downstream of SCR catalyst

2 November 2010 © Ricardo plc 2010

Page 146

RD.10/320201.4

Client Confidential

The figures below shows sensor regeneration and response times vs Particulate Mass (PM) Emissions for the two sensor locations indicated above.

Similar results for Particulate Number (PN) emissions are shown for the NEDC over the same sensor locations.

2 November 2010 © Ricardo plc 2010

Page 147

RD.10/320201.4

Client Confidential

The test vehicle used by Manufacturer 5 was EU4 specification and therefore did not have NOx after-treatment equipment installed. Therefore the results presented below from Manufacturer 5 are equivalent to location S1 Manuf 4's results above.

2 November 2010 © Ricardo plc 2010

Page 148

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 149

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 150

RD.10/320201.4

Client Confidential

Based on the results from Manuf 5, the only fault level that was capable of being detected during the NEDC cycle from a cold start was the empty DPF can · This may not be representative of a leaking DPF with similar PM emissions, faster warm up to sensor dew point may be due to lack of DPF substrate thermal inertia

When started with a hot engine, a regeneration event was possible with all levels of degraded DPF tested · A number of the hot start tests achieved 2 regeneration events within the test cycle

The limiting factor appears to be the time taken for the sensor to reach it's dew point when the engine is started from cold · · A sensor location further upstream in the exhaust system would benefit, however, the position in this test was as per the sensor manufacturer's recommendation for homogeneous mixing The capability of the sensor is very dependant on the specific engine/vehicle on test, e.g. the thermal inertia of other exhaust after-treatment components ­ The test vehicle here was EU4 specification and had no NOx after-treatment that would likely delay the dew point further

The dew point times achieved in this set of tests appear longer than Ricardo would expect

4.13.2.1

Monitor Robustness

Additional testing (data not included in this report) was carried out by Manuf 4 to determine robustness of PM sensor response time to exhaust flow distribution · · Comparing locations S1 and S2 Comparing results with the same % failure of DPF but with plugs removed from different locations across the face

Based on that analysis, the following conclusions were made:

2 November 2010 © Ricardo plc 2010

Page 151

RD.10/320201.4

Client Confidential

· ·

Location S1 is most favourable for regeneration time, due to early dew point detection Monitor based on sensor location S1 would not be robust because: ­ ­ Results are inconsistent depending on the location of the removed plugs on the DPF face This may cause either false MIL or false pass due to poor soot dispersion

·

Manuf 4 recommendation is that a position further away from the DPF with good dispersion of soot, i.e. S2, is necessary. 12.3 mg/km Particulate Matter (PM): 1.4 times the proposed EU6 EOBD limit 38.3x1012 /km Particulate Number (PN): > 30 times the proposed limit

Based on the results from location S2, a regeneration during the NEDC cycle was first possible with 12 % DPF failure, equivalent to: · ·

In order to improve robustness the monitor result should be based on 2 successive regenerations. Using sensor location S2, this was only achieved on the NEDC with a DPF failure level of 100 % (i.e. empty DPF can) · · It does not seem likely that robust monitoring will be possible to an adequate emissions sensitivity if 2 regenerations are required in one NEDC Ricardo recommendation would be to use a single regeneration and then make best use of the pre-conditioning cycles allowed in the legislated test procedure ­ ­ Pending code set on 1st and 2nd cycles (2 pre-conditioning cycles allowed) MIL activated on 3rd cycle (actual test cycle)

Data provided by the particulate sensor supplier shows results of repeatability and durability testing of the sensor (on testbench only): · · Typical repeatability of single sensor of ±3 % response time 29 sensors aged to 3000 hrs showed response time calculations varied by ±25 %

The charts below show the effective regeneration times using sensor location S2 with error bars reflecting regeneration time tolerance of ±126 seconds · ±68 secs equivalent to ±25 % tolerance of the average response time

2 November 2010 © Ricardo plc 2010

Page 152

RD.10/320201.4

Client Confidential

·

Additional ±58 secs for the tolerance of sensor dew point time during the tests

Based on these results, the following observations are made: · · Best performing `malfunctioning' DPF may not be detected at 12 % DPF failure Best achievable monitor threshold would be approximately 14 % DPF failure, at which point estimated emissions would be: ­ ­ PM = 13.7 mg/km PN = 43.4x1012 parts/km

The analysis contained in this report only considers sensor response time during NEDC testing. · PM sensor response time will depend on engine out particulates, hence engine speed and load

2 November 2010 © Ricardo plc 2010

Page 153

RD.10/320201.4

Client Confidential

·

Therefore a monitoring strategy must implement a variable response time threshold ­ ­ ­ Necessary upper limit to give a test pass result for the monitor Will have to be a function of the speed/load `history' since the last PM sensor regeneration This may lead to additional real world tolerances not yet considered

Since the test data provided by manufacturer 5 do not show any regeneration events possible within the NEDC test cycle from cold start, a proper tolerance assessment of the robustness of any potential monitor strategy is not possible Accurate prediction of the sensor response times are difficult because they will depend on the speed/load conditions of the engine while the soot accumulation is in progress If a regeneration is made just before the end of the NEDC, the soot accumulation will mainly be during the highest speed part of the cycle · The best indication of sensor response time in this part of the cycle is the hot NEDC data for 75 % failed DPF: ­ ­ · Average response time = 327 s and 3 = 9 % Equivalent cold NEDC PM emissions were 13.2 mg/km average

Extrapolating for PM emissions at the proposed threshold of 9 mg/km this implies a worst case (mean + 3) sensor response time of 523 seconds ­ So for regeneration to occur before the final deceleration of the NEDC, the sensor dew point must always be achieved within the first 600 seconds of the cycle

With the sensor dew point not being reached until later than 1000 seconds from cold start, as in these test, a monitoring strategy using this prototype soot sensor is not feasible. If some development of the system was possible to improve the exhaust warm up and achieve a sensor dew point at 800 seconds, this would imply: · · An average sensor response time of 298 seconds, after applying a 3 tolerance Extrapolating from the 75 % failed DPF data, achievable OBD thresholds of: ­ 14.5 mg/km for PM

2 November 2010 © Ricardo plc 2010

Page 154

RD.10/320201.4

Client Confidential

­

84.8x1012 parts/km PN

The analysis contained in this report only considers sensor response in the NEDC · · Real world PM sensor response time will depend on engine out particulates, hence engine speed and load Therefore a monitoring strategy must implement a variable response time threshold ­ ­ ­ Necessary upper limit to give a test pass result for the monitor Will have to be a function of the speed/load `history' since the last PM sensor regeneration This may lead to additional real world tolerances not yet considered

4.13.3 Evaluation of Monitor IUMPR Manuf 4 & 5 monitoring strategy will essentially operate under all conditions after the sensor dew point is reached after each engine start. The likely difficulty with real world in-use monitor performance ratio will be the time taken to run the monitor: · Monitor takes around 1000 seconds to complete a judgement on the NEDC cycle ­ · This could be similar for vehicle warm up at slow driving conditions (same temperature) It may be possible for the denominator to increment before the monitor can complete The performance ratio denominator may increment after 600 seconds of driving ­

The concern above is mitigated by the following: · Denominator for DPF monitoring is allowed to wait 800 km driving in between incrementing ­ · Unlikely that a drive cycle longer than 1000 seconds will not have occurred in that time It seems unlikely that drive cycles of this specific combination of speed/load and duration to be frequent for a statistically significant number of drivers

2 November 2010 © Ricardo plc 2010

Page 155

RD.10/320201.4

Client Confidential

4.13.4 Conclusion · · · · · The proposed EU6 EOBD emissions threshold for PM was exceeded with approximately 8 % DPF degradation The threshold for PN was exceeded with the DPF degradation exceeding 0.14 % (estimated by linear interpolation) There was a measured difference in monitor judgement parameter between the baseline test and that at the proposed EU6 EOBD PM Analysis of likely tolerances on the judgement parameter suggests that the judgement parameter is not capable of making a robust distinction at the proposed EU6 EOBD threshold over a real fleet in real world conditions The table below summarises the test results obtained by all 3 manufacturers who carried out tests on DPF, showing the best achievable EOBD thresholds for monitoring of the DPF. PM (mg/km)

9 Manuf 3 Manuf 4 Manuf 5 33.4 13.7 14.5 PN 12 (x10 parts/km) 1.2 41.3 43.4 85 Comments

Proposed OTL

Using enhanced p strategy Using resistive PM sensor Using resistive PM sensor

·

Analysis of the likely monitor enabled conditions over typical real world driving cycles suggests that the DPF monitor is capable of meeting the legislated minimum in-use performance ratio requirements

2 November 2010 © Ricardo plc 2010

Page 156

RD.10/320201.4

Client Confidential

4.14 SCR Monitor (Diesel)

The table below gives NEDC emission results. NOx emissions were very close to EU6 threshold with 10 % or lower SCR dosing rate.

Emissions vs. 'level of degradation' 500 NOx [mg/km] 400 300 200 100 0 Base 50 % dosing 10 % dosing Degradation 0 % dosing Manuf 3 - 1st iteration EU6 TA EU6 prop OBD EU6 int OBD EU5 OBD Manuf 3 - 2nd iteration

4.14.1 Evaluation of OBD Monitor Strategy Analysis of the monitor capability is based on Manuf 3 existing SCR monitoring algorithm. The basic functionality of the monitoring strategy provided by Manuf 3 is summarized below. · · SCR efficiency monitor is based on the monitoring of high tailpipe NOx caused by poor SCR conversion efficiency. The efficiency is monitored by a single NOx sensor downstream of the SCR catalyst, compared to modelled engine out NOx

2 November 2010 © Ricardo plc 2010

Page 157

RD.10/320201.4

Client Confidential

A functional diagram of the SCR monitor strategy is given below

2 November 2010 © Ricardo plc 2010

Page 158

RD.10/320201.4

Client Confidential

The figure below shows a plot of the Judgment variable vs. NOx

2 November 2010 © Ricardo plc 2010

Page 159

RD.10/320201.4

Client Confidential

· ·

Results are plotted from tests with cold and hot engine start Judgement results for hot tests are plotted against the NOx measured with the equivalent level of degradation from the cold test

2 November 2010 © Ricardo plc 2010

Page 160

RD.10/320201.4

Client Confidential

4.14.1.1

Monitor Robustness

The following tolerances were provided by Manuf 3 for the inputs into the monitoring strategy System NOx Sensor Engine-out NOx Model Tolerances 15 ppm (c<100 ppm), 15% (c>100 ppm), aged 30% (Estimation)

Tolerances were applied to the judgement efficiency (ratio) calculation as follows: · · ±15 % tolerance to the numerator (measured value) ±30 % tolerance to the denominator (modelled value)

Worst case tolerances were considered when either the numerator is at its upper limit and the denominator is at its lower limit, or the numerator at the lower limit and denominator at the upper limit. The charts below show tolerance range of the judgement parameter compared to the emissions results, comparing Ricardo analysis with Manuf 3 analysis. The results are very similar and both confirm that there is sufficient separation in the judgement parameter between the normal and threshold degraded SCR system The differences are because: · Ricardo analysis uses 15 %, rather than 15 ppm NOx sensor tolerance at low concentrations, and so underestimates the tolerance for the baseline result. Using 15 ppm at low NOx concentrations would only affect baseline result and would make no difference to conclusion.

2 November 2010 © Ricardo plc 2010

Page 161

RD.10/320201.4

Client Confidential

·

The Manuf 3 analysis uses one set of test results while the Ricardo analysis is based on the average of all results

Engine-out NOx emissions vary in relation to the engine concept, the vehicle weight, and drive resistance. The vehicle tested in this study has quite low engine out NOx levels and so the proposed EOBD threshold appears to be achievable. With a rough graphical interpolation, the Best Unacceptable SCR catalyst, whose tolerance region would not overlay with the Worst Acceptable catalyst (at Type Approval Limits) would be 50 % + 24.4 % about 75 % damaged.

2 November 2010 © Ricardo plc 2010

Page 162

RD.10/320201.4

Client Confidential

Therefore, a simple estimation of the robust monitoring threshold can be calculated: (engine out emissions ­ EU6 standard) * 0.75 + EU6 standard To predict a realistic forecast of EU6 engine out emissions, all diesel participants of the ACEA EU6 study were requested to provide their expected range.

2 November 2010 © Ricardo plc 2010

Page 163

RD.10/320201.4

Client Confidential

·

The table below shows the range of engine out NOx emissions of manufacturers taking part in the study, along with the respective predictions of an achievable NOx OBD threshold using the calculation above

·

This simple prediction shows that a NOx OBD threshold that can be robustly achieved by the whole range of EU6 diesel vehicles has to be higher than the proposed 140 mg/km ­ For some vehicles, even the interim EU6 NOx OBD threshold of 240 mg/km may not be robust A more accurate prediction of the monitor capability for higher engine out levels of NOx is very difficult because of assumptions that must be made about the overall SCR system performance and the NOx emissions at the time that monitoring is active It is recommended that further testing would be required with a vehicle giving higher engine out NOx to confirm the accuracy of this prediction before considering a further reduction to the OBD NOx threshold This forecast is based on a very simple analysis ­

·

­

4.14.2 Evaluation of Monitor IUMPR Ricardo model was calibrated over the NEDC test cycle with enable conditions to match Manuf 3's enable conditions in the test data: · · · Engine speed > 1500 rev/min Engine temperature > 80 degC Engine torque: (note: a full load torque maximum is 270 Nm)

2 November 2010 © Ricardo plc 2010

Page 164

RD.10/320201.4

Client Confidential

­ ­ ·

> 62 Nm > 70 Nm

Monitor enable conditions met for a cumulative time of 60 s in order to make a judgement

The analysis was repeated with two threshold engine torque limits, as shown above, to assess the sensitivity during the city cycles where engine load is generally lower. The results below show the output using the 62 Nm torque threshold:

2 November 2010 © Ricardo plc 2010

Page 165

RD.10/320201.4

Client Confidential

2 November 2010 © Ricardo plc 2010

Page 166

RD.10/320201.4

Client Confidential

The table below summarises the in-use performance ratio assessment

CYCLE

Number of Judgements 62 Nm 70 Nm 1 0 0 0 3 7

London Paris Stuttgart Turin Rural Highway

1 1 1 0 3 8

The above analysis suggests that there is a high risk of the monitor not achieving minimum ratio requirements robustly under all operating conditions. · · · Monitor enable conditions must be matched to the SCR system dosing enable conditions This means that sufficient speed and load are required for operating temperature of the SCR catalyst Monitor enablement and final in-use performance will be dependent on the specific vehicle and it's SCR control system calibration

2 November 2010 © Ricardo plc 2010

Page 167

RD.10/320201.4

Client Confidential

4.14.3 Conclusion · · · Emissions of NOx were at or very close to the proposed EU6 EOBD threshold with the SCR system degraded to 10 % or less of the normal urea dosing rate A monitor judgement parameter using a NOx sensor downstream of the SCR catalyst was capable of distinguishing the baseline result from the proposed EU6 EOBD emissions threshold result over the NEDC test for the vehicle tested Analysis of likely tolerances on the judgement parameter for this vehicle suggests that the monitor would make a robust distinction at the proposed EU6 EOBD threshold of 140 mg/km NOx, however: ­ ­ ­ · This corresponds to near total failure of the SCR system because the basic engine out emissions from the vehicle tested are low compared to the proposed EOBD threshold Analysis of the best achievable EOBD threshold for the range of engine out NOx levels for the manufacturers involved in this study suggests that the best achievable threshold may need to be as high as 280 mg/km in some cases Further testing with several representative EU6 vehicles with higher engine out NOx is proposed, before considering a lower EOBD NOx threshold

Analysis of the likely monitor enable conditions over typical real world driving cycles suggests that there may be a risk of limited monitor enablement in some low speed/load driving conditions.

2 November 2010 © Ricardo plc 2010

Page 168

RD.10/320201.4

Client Confidential

5.

SUMMARY

The table below summarises results obtained for each of the components evaluated. The first column (sect) gives the section number within the document where results for this component are discussed. The IUMPR column is used to indicate whether the likely monitor enable conditions over typical real world driving cycles at normal ambient conditions will achieve minimum ratio requirements robustly under all operating conditions and, where indicated in the table below, in respect of the best achievable thresholds. This is only done where data was provided to make this analysis possible. In the case of gasoline engines, although analysis of the likely monitor enable conditions over typical real world driving cycles has not been carried out, nevertheless these monitors have been engineered for the California ULEV II emission standards and are fully compliant with the associated OBD regulations including minimum IUMPR requirements. However, this is not the case for compressionignition engines Table key for Sections 5.1 and 5.2: none n/a EU6 TFF Data not available Pollutant is not sensitive or degradation is not significant relative to the OBD threshold An emissions threshold is not applicable because the pollutant was not sensitive to the fault The proposed EU6 EOBD threshold is achievable Total Functional Failure

2 November 2010 © Ricardo plc 2010

Page 169

RD.10/320201.4

Client Confidential

5.1

Sect

Gasoline Results

Monitor Evaluated in this Report on NEDC Maximum Observed Degradation 12 (mg/km) except PN (x10 /km) CO NM HC none none NOx 47.6 none PM 2.69 none PN Achievable Threshold on NEDC (mg/km) 12 except PN (x10 /km) CO EU6 n/a NM HC n/a n/a NOx EU6 n/a PM EU6 n/a PN IUMPR -

Component (Fuel)

4.1 4.3

Linear O2 Sensors (upstream) Lambda Sensors (downstream) 3W Catalytic Converter Misfire Monitor Fuel System Monitor VVT System Monitor MAF Sensor Monitor

Inducing fuelling oscillation and measuring sensor response Evaluating sensor response during rich to lean transitions when fuel cut is activated Cycling fuelling rich and lean to determine oxygen storage Crankshaft Speed Fluctuation (CSF) method Analysis of long-term fuel adaptations Direct measurement of cam position error Measured vs Modelled Airflow

282 none

4.4 4.5 4.6 4.7 4.8

978 956 1764 1298

none**

277 216 90 177

none**

136 64 61 46

none**

none 3.6 3.9 none

none**

none none -

Not regulated

277* 216 EU6 n/a

136* EU6 EU6 EU6 n/a

Not regulated

Not regulated

EU6 1764 EU6 n/a

n/a EU6 n/a n/a

n/a n/a -

* NOx and NMHC were exceeded for tests carried out with degraded catalysts. A less aggressively aged catalyst (giving NOx and NMHC just below the proposed EU6 OBD threshold) is required in order to make a definitive statement about the feasibility of monitoring techniques for EU6 OBD tests ** Lambda Feedback is assumed to compensate for the MAF sensor having a rich bias. This makes the fault insensitive to small deviations in MAF sensor reading Regarding IUMPR, refer to explanation in Section 5 above

2 November 2010 © Ricardo plc 2010

Page 170

RD.10/320201.4

Client Confidential

5.2

Diesel Results

Sect

Component (Fuel)

CO 4.2 4.9 4.9 4.10 4.11 4.12 4.13 Lambda Sensors (linear) Low Flow EGR High Flow EGR EGR Cooler / ByPass NSC/LNT Monitor Oxidation Catalyst Particulate Filter Comparing sensor delay times against threshold values difference between desired and actual air mass flow using MAF, intake air temperature and pressure No monitor evaluated using upstream and downstream lambda measurement using temperature sensors Differential pressure across DPF Prototype resistive-type PM sensor 4.14 SCR Monitor By NOx sensor downstream of SCR catalyst, compared to modelled engine out NOx 174 none 163 none 1071 none none -

NM HC none none 23.5 none 102 none none -

NOx 56.2 1146 none none 133 none none 136

PM 2 2.5 none 58.8 50.3 -

PN

0.0014

CO EU6 n/a n/a TFF n/a n/a n/a n/a

NM HC n/a n/a n/a EU6 n/a n/a n/a n/a

NOx n/a EU6 n/a TFF§ n/a n/a n/a TFF§

PM n/a n/a n/a n/a n/a 33.4 14.5 n/a

PN n/a n/a n/a n/a n/a 41 85 n/a

none none 53 103 -

§ Note that this engine's engine out emissions are lower than the OBD threshold and therefore a TFF is specified.

2 November 2010 © Ricardo plc 2010

Page 171

IUMPR

Monitor Evaluated in this Report on NEDC

Maximum Observed Degradation 12 (mg/km) except PN (x10 /km)

Achievable Threshold on NEDC 12 (mg/km) except PN (x10 /km)

-

RD.10/320201.4

Client Confidential

6.

CONCLUSIONS

The impact of reduced OBD thresholds or intrusive monitoring techniques on CO2 emissions are not examined in this report. Durability evaluation of sensors was not part of this study but remains a significant aspect to consider for the application and integration of sensors in vehicle emission control systems, especially in relation to real-driving performance and capability with different biofuel blends.

6.1

Gasoline

The main points that need to be highlighted for Gasoline engines are as follows: 1. OBD Monitor thresholds were exceeded for NOx and NMHC with the degraded catalysts used. Work is required with less aggressively aged catalysts in order to make a definitive statement about the feasibility of the monitoring techniques to detect a threshold component. 2. Misfire detection using Crankshaft Speed Fluctuation (CSF) method successfully detected a 6 % misfire, however in some tests this caused emissions above the proposed EU6 thresholds. Further investigation would be required with revised monitor thresholds and a lower misfire rate to confirm that detection at the proposed threshold level would not result in false misfire detection. 3. VVT System Monitoring successfully detected a Total Functional Failure (TFF), however emissions exceeded the proposed EU6 EOBD thresholds. Further investigation would be required with revised monitor thresholds and partial VVT failure to confirm that detection at the proposed threshold level would not result in false fault detection. 4. Although PN and PM measurements were taken, in each case a specific trend was not obvious and the proposed OBD and type approval thresholds were not exceeded in any of the test measurements made.

6.2

Diesel

With regard to Diesel vehicle emissions, the following conclusions are drawn:

2 November 2010 © Ricardo plc 2010

Page 172

RD.10/320201.4

Client Confidential

1. The EGR low flow monitor was able to distinguish the baseline result (without a malfunction) from a failure at the proposed EU6 EOBD emissions level. However, to reduce the risk of false failure detection with partial degradation and emissions within the type approval emission limit, more separation is required between the type approval and EOBD thresholds. 2. None of the EGR high flow faults resulted in significant increase in emissions due to the DPF effectively removing any additional PM. 3. Total EGR Cooler / By-Pass failure did not cause the proposed EU6 EOBD thresholds to be exceeded. 4. The NOx threshold achieved on the vehicle tested for NSC/LNT monitor was approximately the proposed EU6 threshold. However, on the application tested this represented a large scale failure of the NSC/LNT because of the relatively low engine out NOx level. Additional work is required to confirm a threshold that can be robustly detected by all engines. 5. Emissions for DOC failures examined were below the proposed OBD thresholds. However, this corresponded to total failure of the DOC because tailpipe emissions were below 140 mg/km in this condition. The spread of emissions results suggests that a partially failed DOC could not be robustly detected for this vehicle. 6. The best achievable PM threshold for DPF monitoring based on differential pressure across the DPF was over 3 times the proposed EU6 threshold. 7. The best achievable PN threshold for DPF monitoring based on differential pressure across the DPF was over 30 times the proposed EU6 threshold. 8. The best achievable PM threshold for DPF monitoring based on resistive soot sensor was over 1.6 times the proposed EU6 threshold. 9. The best achievable PN threshold for DPF monitoring based on resistive soot sensor was over 70 times the proposed EU6 threshold. 10. The NOx threshold achieved on the vehicle tested for SCR monitor was approximately the proposed EU6 threshold. However, on the application tested this represented a large scale failure of the SCR system because of the relatively low engine out NOx level. Additional work is required to confirm a threshold that can be robustly detected by all engines.

2 November 2010 © Ricardo plc 2010

Page 173

Information

Microsoft Word - Q51760 RD_10_320201_4 ACEA_Euro6_OBD_Report.doc

173 pages

Find more like this

Report File (DMCA)

Our content is added by our users. We aim to remove reported files within 1 working day. Please use this link to notify us:

Report this file as copyright or inappropriate

547431


You might also be interested in

BETA
Microsoft Word - Q51760 RD_10_320201_4 ACEA_Euro6_OBD_Report.doc