Emission test report

8/12/2019 Emission test report

1/234


2/234


3/234

E T C ED E M C

D S I D

Evaluation and improvement of road vehicle

pollutant emission factors based on

instantaneous emissions data processing

Tesi Doctoral

Autor:

Vicente F G

Directors:

Dra. Mara Rosario V N

Dr. Daniel G C

Dra. Panagiota D

2014

Ispra (Itlia)Castell de la Plana


4/234


5/234

S T E SD M E CD I S E D

Evaluation and improvement of road vehicle

pollutant emission factors based on

instantaneous emissions data processing

a dissertation submitted in partial fullment of the requirements for thedegree of Doctor of Philosophy

Author:

Vicente F G

Supervisors:

Dr Mara Rosario V N

Dr Daniel G C

Dr Panagiota D

2014

Ispra (Italy)Castell de la Plana (Spain)


6/234


7/234

A mon pare, que tot ho arreglaTo my dad, who can fix anything


8/234


9/234

Resum

Els instruments actuals permeten mesurar les emissions dels vehicles amb elevadaresoluci temporal. Les installacions de dinammetres de xasss i motor, i tambels sistemes porttils de mesura demissions (portable emission measurement systems,PEMS), sn capaos denregistrar senyals de concentracions de contaminants icabdals de gas amb resolucions d1 Hz, i alguns analitzadors arriben ns i tot als10Hz.

Aquestes dades demissions es poden emprar per a produir factors demissiinstantanis, que sn relacions funcionals empriques entre lestat operatiu instantanidel vehicle (normalment denit pel rgim de gir i el parell motor, en el qual cas es

parlade mapes demissi)ielseuperldemissionsaleixidadelcatalitzador.Elsmapesdemissi es poden utilitzar en entorns de simulaciper a predir les masses demissionsde contaminants sota qualsevol cicle ds, tot reduint la despesa en experiments, pertamb la quantitat de dades realment mesurades. Per tant, s essencial que aquestsmapes es construsquen a partir de dades exactes.

Lelevada resoluci temporal dels senyals mesurats no implica una major exactitud.Un prerequisit per a la producci de factors demissi de qualitat s la correctaatribuci temporal de les masses de contaminants mesurades als estats del vehicle odel motor que shan enregistrat. A planteja un repte tcnic perqu les installacionsde mesura demissions de vehicles no han sigut dissenyades per a facilitar elmodelat instantani de les emissions. Primerament, trobem que les concentracionsdels contaminants, els cabdals de gas i les dades del motor (s a dir, els ingredientsdels mapes demissi) provenen dinstruments diferents que sovint no disposen dun

Aquest resum sinclou de conformitat amb larticle 24 de la normativa dels estudis de doctorat dela Universitat Jaume I regulats pel Reial Decret 99/2011.

v


10/234

RESUM

senyal de temps com. A ms a ms, degut a restriccions en la conguraci fsicadels sistemes de mesura, les concentracions de contaminants i el cabdal de gasos

descapament es mesuren a alguns metres de distncia del punt demissi (leixida delcatalitzador), de forma que aquests senyals mesurats es veuen afectats per retards en eltransportiperfenmensdemesclaquevariendeformanolinialambelcabdal.Per,els analitzadors qumics tenen una caracterstica de resposta dinmica que retarda idistorsiona ms encara els senyals. Totes aquestes distorsions es tradueixen en unsol efecte observable: els senyals mesurats sn versions dinmicament retardades,suavitzades i aplanades (s a dir, amb pics menys pronunciats) dels senyals vertadersque sobservarien a leixida del catalitzador.

Lobjectedaquesta tesi s millorar la qualitat de la mesura i el modelat instantani de

les emissions dels vehicles mitjanant laplicaci de tcniques de post-processamentde dades. Elsobjectius de recercaque sen deriven sn els segents:

1. Avaluar els efectes que distorsionen i afecten lexactitud dels senyalsinstantanis demissions emprats en el modelat de les emissions dels vehicles, idesenvolupar un marc metodolgic per a la compensaci dels senyals.

2. Desenvolupar una metodologia completa per a la compensaci dels senyals, idisseminar-la entre les parts interessades del grup ERMES (European Researchon Mobile Emission Sources), que aglutina els laboratoris i altres entitats de

recerca europeus dedicats a la mesura i modelat de les emissions dels vehicles.

Les distorsions que afecten lexactitud dels senyals instantanis es poden compensaramb una combinaci de modelat fsic i post-processament de dades. La principalaportaci original daquesta tesi s una nova metodologiaque shi descriucompletamentper a compensar els senyals instantanis demissions. Mentre queles metodologies anteriors es basaven en tcniques de modelat de teoria de sistemesque requereixen una caracteritzaci experimental exhaustiva (per sub-sistema) de lesdistorsions imposades per lequip de mesura, el mtode alternatiu utilitza el CO2

com a traador per tal de caracteritzar el sistema de forma agregada.Lesconclusionsprincipals daquesta tesi sn les segents:

1. La compensaci dels senyals instantanis s necessria per a la produccide factors demissi amb elevada resoluci temporal. Qualsevol esquema decompensaci de senyals instantanis es fonamentar en una combinaci demodelat fsic dels sistemes de mesura i post-processament de dades.

vi


11/234

RESUM

2. El mtode del traador de CO2del qual shi aporta una descripci detalladai una validaci experimental en aquesta tesis una metodologia completa

i exible que permet compensar els senyals produts pels sistemes de mesurahabitualment emprats en el modelat instantani de les emissions dels vehicles(dinammetres de xasss o de motor i PEMS) de forma semiautomtica,tot facilitant la creaci de mapes instantanis demissions i reduint la crregaexperimental en comparaci amb les metodologies anteriors.

Mots clau: emissions dels vehicles, model demissi, factor demissi, mapa demissions,

CO2, consum de carburant, contaminants regulats, senyals instantanis, exactitud, aliniament

temporal, post-processament de senyals, dinammetre de xasss, dinammetre de motor, PEMS,

harmonitzaci, grup ERMES, Comissi Europea, traador de CO2.

vii


12/234

viii


13/234

Abstract

Current instrumentation can measure vehicle emissions with high temporal

resolution. Chassis and engine dynamometer test facilities, and also portableemissions measurement systems (PEMS), can record instantaneous pollutantconcentration and exhaust ow data streams at 1 Hz, and some setups now provide10 Hz resolutions.

Vehicle emission modellers can use instantaneous emissions data to produceinstantaneous emission factors, which are empirical functional relations between theinstantaneous driving state of a vehicle (normally characterised by engine speed andtorque, in which case they are called engine emission maps) and its emission behaviour

at the catalyst-out point. Engine emission maps can then be applied in vehicleor enginesimulation environmentsto predict the mass emissions of pollutants of avehicle for any given duty cyclethus reducing the need for costly measurements,but also the amount of measured data. It is therefore crucial that these engineemission maps be derived from accurate data.

e increased resolution of emissions signals does not equate with increased accuracy.A prerequisite for the derivation of accurate emission factors from instantaneousvehicle emissions data is a ne allocation of measured mass emissions to recordedengine or vehicle states. is poses a technical challenge, because vehicle emission test

facilities are not designed to support instantaneous emissions modelling, and theyintroduce number ofdistorting effectsthat compromise the instantaneous accuracy ofthe measured signals. First of all, pollutant concentrations, ow rates and engine data(i.e., the building blocks of engine emission maps) come from different instrumentsoften lacking a clear common time reference signal for a proper alignment. Also,due to measurement setup constraints, pollutant concentrations and exhaust gasow rates will often be measured several metres away from the catalyst-out point,

ix


14/234

ABSTRACT

and so the measured concentration signals are affected by transport delays andgas mixing phenomena, which vary non-linearly with ow. Finally, exhaust gas

analysers have a dynamic response characteristic that further delays and distortsthe measured concentration signals. All of the aforementioned distortions have oneobservable effect: measured signals that are dynamically delayed, smoothed andattened (i.e., having less sharp peaks) versions of the true emissions that wouldbe observed at the catalyst-out point.

ese distorting effects can be compensated through a combination of physicalmodellingand data post-processing. e main original contribution of this dissertationis a novel methodology for the compensation of instantaneous emission signals,which is fully described herein. Whereas previous methodologies relied upon systems

theory modellingincluding a comprehensive experimental characterisationtomodel the sub-systems of the measurement setup, the alternative approach uses CO2as a tracer of the distortions brought about by the measurement setup, which ismodelled as a lump system.

Keywords:vehicle emissions, emission model, emission factor, engine emission map, CO2, fuel

consumption, regulated pollutants, instantaneous signals, accuracy, time alignment, signal post-

processing, chassis dynamometer, engine dynamometer, PEMS, harmonisation, ERMES group,

European Commission, CO2tracer.

x


15/234

AcknowledgementsBerlin, May 2014

My thesis years will likely go down in history as the worst of a global nancial crisis, andyet I will remember them as a time of steady personal growth and transformation. I owe a

large debt of gratitude to many people (incidentally quite a few Greeks and nationals of otherdebt-laden countries) that helped me along the way: Rosario Vidal and Daniel Garranbesides being the reason I started doing research in the rst placefor supervising me overthe long distance; Penny Dilara for giving me the opportunity to join the Joint ResearchCentre of the European Commission and allowing me to pursue my own research interestseven when the prospects of success were uncertain; Georgios Fontaras for giving my researcha real purpose and helping me navigate my periods of obfuscation; Savas Geivanidis andMartin Weilenmann for being an inspiration and generously sharing their expertise; StefanHausberger and Leonidas Ntziachristos for performing the external review of the dissertationmanuscript; Urbano Manfredi, Franz Mhlberger and the rest of the VELA laboratories stafffor their help during the experimental campaign, and Konstantinos Agnanostopoulos for his

commitment to improving the algorithm of the CO2tracer methodology.e funding of the European Commission during my three years in Ispra as a doctoralresearch grantholder is gratefully acknowledged, as is the funding and loving support of myextended family during the preceding twenty-seven years. In no particular order, a word ofappreciation also goes to my beloved friends and colleagues Nana Amoateng, onorevoleMirko(Vittorio) Busto, David Pamies, Lorena Hojas, Julio Gonzlez Romero, Pepa Prez Barral,Guadalupe Sepulcre, Pascual Dions, Irene Pinedo, Lara Vilar, Georgina Harris-Lpez,Marina Kousoulidou, Cinzia Pastorello, Fabio Dalan, Biagio Ciuffo, Alessandro Marotta,Laura Lonza, Yoannis Drossinos, Covadonga Astorga, Pierre Bonnel, Denise Pernigotti,Lorenzino Vaccari, Teresita Freddi, Fulvio Ardente, Andrea Marsano, Vera iemig, Martha

Dunbar, Alison Pridmore, Ben Murphy, Andrew Singleton, Alexandros Nikolian, NafsikaStavridou, Tiberiu Antoe, Simone Russo, Hugo Carro, Silvia Cirillo, Giovanna Indrio,Martin Weiss, Monica uuianu, Elena Monea, Zuzana Koritschanov, Carme Calduch,

Javier Sanflix, David Cebrin, Enrique Moliner, Marta Royo, Carlos Muoz, Peter Mockand many others for making these years thoroughly enjoyable and productive.

From the bottom of my heart,Many thanks / Muchas gracias / Moltes grcies / Grazie tante /

xi


16/234

xii


17/234

Contents

1 Introduction 1

1.1 Goal and scope . . . . . . . . . . . . . . . . . . . . . . . . . . . 21.2 Justication . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.4 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61.5 Research methodology . . . . . . . . . . . . . . . . . . . . . . . 71.6 Structure of this dissertation . . . . . . . . . . . . . . . . . . . . 8

2 Measuring the emissions of road vehicles 112.1 Emission models and emission factors . . . . . . . . . . . . . . . 11

2.1.1 European emissions modelling: the ERMES group . . . . 122.1.2 ERMES models: COPERT and HBEFA . . . . . . . . . 14

2.2 Chassis and engine dynamometer testing . . . . . . . . . . . . . 152.2.1 Test cycles . . . . . . . . . . . . . . . . . . . . . . . . . 172.2.2 EF development from dynamometer laboratory data . . . 202.2.3 Chassis and engine dynamometer applications . . . . . . 232.2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 26

2.3 Remote sensing . . . . . . . . . . . . . . . . . . . . . . . . . . 282.3.1 EF development from remote sensing data . . . . . . . . 29

2.3.2 Remote sensing applications . . . . . . . . . . . . . . . . 302.3.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 322.4 On-road (chase) measurements . . . . . . . . . . . . . . . . . . 32

2.4.1 EF development from on-road measurement data . . . . . 332.4.2 On-road measurement applications . . . . . . . . . . . . 332.4.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.5 Tunnel studies . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

xiii


18/234

CONTENTS

2.5.1 EF development from tunnel studies data . . . . . . . . . 352.5.2 Tunnel study applications . . . . . . . . . . . . . . . . . 35

2.5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 362.6 On-board measurements (PEMS) . . . . . . . . . . . . . . . . . 372.6.1 EF development from PEMS data . . . . . . . . . . . . . 392.6.2 PEMS applications . . . . . . . . . . . . . . . . . . . . 402.6.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 43

2.7 Summary of EF development . . . . . . . . . . . . . . . . . . . 432.7.1 Instantaneous vehicle emissions modelling . . . . . . . . 45

3 Distorting effects of instantaneous vehicle emission signals 473.1 Vehicle emission measurement signals and systems . . . . . . . . . 48

3.1.1 Vehicle emission measurement signals . . . . . . . . . . . 483.1.1.1 Physical magnitudes . . . . . . . . . . . . . . 493.1.1.2 Time domain characteristics . . . . . . . . . . 503.1.1.3 Stochastic components . . . . . . . . . . . . . 50

3.1.2 Vehicle emission measurement systems . . . . . . . . . . 513.1.2.1 Sub-system division . . . . . . . . . . . . . . 51

3.1.2.1.1 Vehicle exhaust system . . . . . . . . 513.1.2.1.2 Raw gas line . . . . . . . . . . . . . 523.1.2.1.3 Dilution tunnel . . . . . . . . . . . 52

3.1.2.1.4 Analysers . . . . . . . . . . . . . . . 523.1.2.2 Dynamicity . . . . . . . . . . . . . . . . . . . 543.1.2.3 Linearity . . . . . . . . . . . . . . . . . . . . 553.1.2.4 Time variance . . . . . . . . . . . . . . . . . 553.1.2.5 Stability and mass conservation . . . . . . . . . 56

3.2 Distorting effects and instantaneous accuracy . . . . . . . . . . . 573.2.1 Systematic signal misalignments . . . . . . . . . . . . . . 58

3.2.1.1 Primary misalignment . . . . . . . . . . . . . 583.2.1.2 Secondary misalignment . . . . . . . . . . . . 60

3.2.2 Variable transport times . . . . . . . . . . . . . . . . . . 603.2.3 Exhaust gas mixing . . . . . . . . . . . . . . . . . . . . 623.2.4 Dynamic response of gas analysers . . . . . . . . . . . . . 643.2.5 Signal aliasing . . . . . . . . . . . . . . . . . . . . . . . 65

3.3 Summary of distorting effects . . . . . . . . . . . . . . . . . . . 66

xiv


19/234

CONTENTS

4 Compensation of vehicle emission signals 71

4.1 Methodological framework for signal compensation . . . . . . . . 71

4.1.1 Compensation algorithm . . . . . . . . . . . . . . . . . 734.1.2 Original signals . . . . . . . . . . . . . . . . . . . . . . 734.1.3 Reference signals . . . . . . . . . . . . . . . . . . . . . 734.1.4 Metrics of signal similarity . . . . . . . . . . . . . . . . . 74

4.1.4.1 Sum of squared residuals . . . . . . . . . . . . 754.1.4.2 Sum of absolute deviations . . . . . . . . . . . 754.1.4.3 Signal cross-correlation . . . . . . . . . . . . . 75

4.2 Design specications of compensation methods . . . . . . . . . . 774.2.1 Peak reconstruction and time alignment capabilities . . . . 77

4.2.2 Simplicity . . . . . . . . . . . . . . . . . . . . . . . . . 774.2.3 Broad scope and applicability . . . . . . . . . . . . . . . 784.2.4 Data sampling rates . . . . . . . . . . . . . . . . . . . . 79

4.3 Modelling emission signal distortions . . . . . . . . . . . . . . . 794.3.1 System identication . . . . . . . . . . . . . . . . . . . 804.3.2 Modelling a complete vehicle emissions measurement setup 82

4.4 Literature review . . . . . . . . . . . . . . . . . . . . . . . . . . 854.4.1 Partial methods based upon data post-processing . . . . . 854.4.2 Complete methods based upon physical modelling . . . . 87

4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

5 e CO2tracer method 91

5.1 Modelling distortions with a tracer gas . . . . . . . . . . . . . . . 915.1.1 Properties of CO2as a tracer gas . . . . . . . . . . . . . 92

5.2 eoretical background . . . . . . . . . . . . . . . . . . . . . . 935.2.1 Discrete impulse response and discrete convolution . . . . 94

5.2.1.1 Discrete impulse response . . . . . . . . . . . 945.2.1.2 Discrete convolution . . . . . . . . . . . . . . 97

5.2.2 Signal binning . . . . . . . . . . . . . . . . . . . . . . . 995.3 Experimental method . . . . . . . . . . . . . . . . . . . . . . . 1015.3.1 Instantaneous fuel consumption measurement . . . . . . 102

5.4 Signal post-processing method . . . . . . . . . . . . . . . . . . . 1035.4.1 Derivation of mass signals . . . . . . . . . . . . . . . . . 104

5.4.1.1 Derivation of the original CO2signal . . . . . 1055.4.1.2 Derivation of the reference CO2 signal . . . . . 105

xv


20/234

CONTENTS

5.4.2 Global alignment . . . . . . . . . . . . . . . . . . . . . 1065.4.3 Local alignment . . . . . . . . . . . . . . . . . . . . . . 110

5.4.3.1 Signal preparation: signal padding . . . . . . . 1115.4.3.2 Computation of time-shifted signals . . . . . . 1125.4.3.3 Selection of optimal bins . . . . . . . . . . . . 1145.4.3.4 Computation of the aligned signal . . . . . . . 1145.4.3.5 Alignment loop . . . . . . . . . . . . . . . . . 115

5.4.4 Local sharpening . . . . . . . . . . . . . . . . . . . . . 1165.4.4.1 Computation of sharpened signals . . . . . . . 1175.4.4.2 New selection of optimal bins and computation

of the compensated signal . . . . . . . . . . . 1185.4.5 Blind compensation of non-tracer gaseous pollutants . . . 1205.4.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . 120

6 Results and discussion 1236.1 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

6.1.1 Experimental campaign . . . . . . . . . . . . . . . . . . 1246.1.2 Compensation of the tracer pollutant . . . . . . . . . . . 1296.1.3 Blind compensation of other pollutants . . . . . . . . . . 1406.1.4 Software implementation: esto. . . . . . . . . . . . . . 144

6.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146

6.2.1 Discussion of the CO2 tracer method . . . . . . . . . . . 1476.2.1.1 Operation with mass signals . . . . . . . . . . 1476.2.1.2 Signal post-processing methods . . . . . . . . . 1516.2.1.3 Flexibility and robustness . . . . . . . . . . . . 1526.2.1.4 Method limitations . . . . . . . . . . . . . . . 153

6.2.2 Validation of the CO2tracer method . . . . . . . . . . . 1546.2.2.1 Consistency of results . . . . . . . . . . . . . . 1556.2.2.2 Mass conservation . . . . . . . . . . . . . . . 1566.2.2.3 Inspection of the compensation process . . . . 159

7 Conclusions 169

xvi


21/234

List of gures

1.1 Prioritisation of emissions modelling research issues within ERMES 4

2.1 Structure of the ERMES group . . . . . . . . . . . . . . . . . . 142.2 Schematic representation of a chassis dynamometer emissions test

facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162.3 Time-velocity prole of the ERMES chassis dynamometer test cycle 192.4 Illustration of the variability of chassis dynamometer test results . . 212.5 Example of engine emission map . . . . . . . . . . . . . . . . . 222.6 Passenger car instrumented with PEMS . . . . . . . . . . . . . . 38

3.1 Schematic representation of a chassis dynamometer measurementsetup with dilution (CVS) . . . . . . . . . . . . . . . . . . . . . 53

3.2 Illustration of the distorting effects upon pollutant concentrationsignals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

3.3 Application of xed time shifts via the software settings ofmeasurement instruments . . . . . . . . . . . . . . . . . . . . . 59

3.4 Illustration of the variability of transport times within the exhaustsystem of Diesel and gasoline vehicles . . . . . . . . . . . . . . . 61

3.5 Effect of gas mixing upon the total pollutant mass reported by rawgas measurement congurations . . . . . . . . . . . . . . . . . . 63

3.6 Illustration of the effect of the dynamic response characteristic ofanalysers upon measured mass emissions . . . . . . . . . . . . . . 65

3.7 Illustration of signal aliasing . . . . . . . . . . . . . . . . . . . . 66

4.1 Reference framework for emission signal compensation methodologies 72

xvii


22/234

LIST OF FIGURES

4.2 Sub-system division of a chassis dynamometer measurement setup(CVS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

5.1 Flattening of signal peaks by gas mixing (axial diffusion) . . . . . . 965.2 Example discrete impulse response of a linear system . . . . . . . 975.3 Illustration of edge effects in the production of rolling data bins . 1015.4 Portable FC meter mounted inside the cabin of a passenger car . . 1025.5 General scheme of the CO2 tracer method . . . . . . . . . . . . . 1045.6 Reference and original mass CO2signals after global alignment . . 1085.7 Signal zero-padding before post-processing . . . . . . . . . . . . 1115.8 Computation of the time-shifted instances of the diluted signal . . 1135.9 Reconstruction of the locally time-aligned signal from the optimal

bins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1155.10 Computation of the sharpened instances of the aligned signal . . . 119

6.1 Snapshots of the experimental campaign . . . . . . . . . . . . . . 1266.2 Driving cycles of the experimental campaign . . . . . . . . . . . . 1286.3 Scatterplots of theoriginal, aligned and fully compensated CO2signals1316.4a Compensation of the diluted mass CO2signal (NEDCadataset) . 1326.4b Compensation of the diluted mass CO2signal (ERMESadataset) . 1336.4c Compensation of the diluted mass CO2signal (CADCadataset) . 134

6.5 Global assessment of compensation at 5-second bins (all datasets) . 1366.6a Assessment of compensation of NEDCadataset at 5-second bins . 1376.6b Assessment of compensation of ERMESadataset at 5-second bins . 1386.6c Assessment of compensation of CADCadataset at 5-second bins . 1396.7a Blind compensation of the diluted mass NOxsignal (NEDCadataset)1416.7b Blind compensation of the diluted mass NOxsignal (ERMESadataset)1426.7c Blind compensation of the diluted mass NOxsignal (CADCadataset)1436.8 estooverview . . . . . . . . . . . . . . . . . . . . . . . . . . . 1456.9 Illustration of dilution in a CVS setup . . . . . . . . . . . . . . . 149

6.10 Effect of the alignment and sharpening processes upon absoluteinstantaneous error . . . . . . . . . . . . . . . . . . . . . . . . . 1566.11 Illustration of the effect of time-shifting and sharpening upon mass

conservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1586.12 Computation of the inspectors of the compensation process . . . . 1616.13 Inspectors of the alignment process (by test cycle) . . . . . . . . . 1626.14 Inspectors of the sharpening process (by test cycle) . . . . . . . . . 163

xviii


23/234

LIST OF FIGURES

6.15 Fuel cut-off events identied by the alignment inspector duringNEDC testing . . . . . . . . . . . . . . . . . . . . . . . . . . . 164

6.16 Volume ow rates at the dilution tunnel reported by the test bench(all datasets) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1656.17 Scatterplots of estimated relative delaysvsOBD engine load (all

datasets) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167

xix


24/234

xx


25/234

List of tables

2.1 Selected ERMES group stakeholders . . . . . . . . . . . . . . . . 132.2 Summary of vehicle emission measurements in chassis dynamo-

meter laboratories . . . . . . . . . . . . . . . . . . . . . . . . . 272.3 Summary of vehicle emission measurements in engine dynamo-

meter laboratories . . . . . . . . . . . . . . . . . . . . . . . . . 272.4 Summary of remote sensing of vehicle emissions . . . . . . . . . . 322.5 Summary of on-road (chase) measurement of vehicle emissions . . 342.6 Summary of tunnel studies for the measurement of vehicle emissions 372.7 Summary of vehicle emissions measurements using PEMS . . . . . 43

2.8 Sources of instantaneous and aggregated data, by measurement setup 45

3.1 Air ow measurement methods in vehicle emissions modelling . . 493.2 Measurement setup sub-system division . . . . . . . . . . . . . . 513.3 Time and ow variability in sub-systems of measurement setups . . 563.4 Reestimates for different measurement chain sub-systems . . . . . 643.5 Summary of emission signal distortions: primary misalignment . . 673.6 Summary of emission signal distortions: secondary misalignment . 673.7 Summary of emission signal distortions: variable transport times . 68

3.8 Summary of emission signal distortions: gas mixing . . . . . . . . 683.9 Summary of emission signal distortions: dynamic response of analysers 693.10 Summary of emission signal distortions: signal aliasing . . . . . . 69

4.1 Recommended compensation scopes, by user type . . . . . . . . . 794.2 Partial models that rely (mostly) upon statistical signal post-processing 864.3 Models that rely (mostly) upon physical modelling . . . . . . . . 88

xxi


26/234

LIST OF TABLES

5.1 Summary of the data post-processing sequence of the CO2 tracermethod . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

6.1 Technical specications of the portable fuel consumption meter . . 1256.2 Characteristics of the test cycles of the experimental campaign . . . 1276.3 Details of the compensation process of the experimental datasets . 1306.4 Overall mass delity of the compensation process . . . . . . . . . 157

xxii


27/234

Acronyms and abbreviations

ADVISOR: Advanced Vehicle Simulator (vehiclesimulation software).AM: Alignment Margin.ARTEMIS: Assessment and Reliability ofTransport Emission Models and InventorySystems (EU-funded research project).BIBO: Bounded Input - BoundedOutput.CADC: Common ARTEMIS DrivingCycle.CFV: Critical Flow Venturi.CI-MS: Chemical Ionisation MassSpectrometry.CLD: Chemiluminiscent Detector.CMEM: Comprehensive Modal EmissionsModel.CNG: Compressed Natural Gas.CO: Carbon monoxide.COPERT: Computer Programme to calculateEmissions from Road Transport.CRT: Continuously Regenerating Trap (particlelter).CVS: Continuous Volume Sampling.

DACH-NL-S: Germany, Austria, Switzerland,the Netherlands and Sweden (group of countriesthat use HBEFA).DG: Directorate-General (administrative branchof the European Commission).DOC: Diesel Oxidation Catalyst.DPF: Diesel Particulate Filter.Dyno: [chassis or engine] dynamometer.

EC: European Commission.ECU: Electronic Control Unit.EEA: European Environment Agency.EF: Emission Factor.EMPA:Eidgenssische Materialprfungs undForschungsAnstalt(Swiss Federal Laboratories forMaterials Testing and Research).ERMES: European Research on MobileEmission Sources.ESC: European Stationary Cycle.esto: emission signals synchronisationtool.ETC: European Transient Cycle.EV: Electric Vehicle.FC: Fuel Consumption.FTIR: Fourier Transform Infrared spectroscopy(chemical analysis technique).FTP: Federal Test Procedure (USA standarddriving cycle).GPS: Global Positioning System.GRPE: Global Regulation working party onPollution and Energy.HBEFA:Handbuch fr Emissions-faktoren des

Strassen-verkehrs(Handbook of emission factorsfrom road transport).HC: Hydrocarbons.HDV: Heavy-Duty Vehicle.HEV: Hybrid Electric Vehicle.HFID: Heat Flame Ionisation Detector.HILS: Hardware-In-the-LoopSimulation.

xxiii


28/234

ACRONYMS AND ABBREVIATIONS

ICCT: International Council on CleanTransportation.IET: Institute for Energy and Transport (part of

Directorate-General Joint ResearchCentre).IPCC: Intergovernmental Panel on ClimateChange.

JRC: Joint Research Centre (Directorate-Generalof the European Commission).LAT: Laboratory of Applied ermodynamics(Aristotle University of essaloniki,Greece).LCV: Light Commercial Vehicle.LDV: Light-Duty Vehicle.

LPG: Liqueed Petroleum Gas.LTI: Linear, Time-Invariant (system).MEL: Mobile Emissions Laboratory (of theCenter for Environmental Research &Technology of the University of CaliforniaRiverside).MOVES: Motor Vehicle Emission Simulator(USEPA emission model).MSE: Mean Squared Error.NDIR: Non-dispersive Infrared (gasconcentration measurement principle).NDUV: Non-dispersive Ultraviolet (gasconcentration measurement principle).NEDC: New European Driving Cycle.NMVOC: Non-Methane Volatile OrganicCompounds.NOx: Nitrogen oxides (NO and NO2).NTE: Not-to-exceed (emission limits associatedwith delimited engine operating areas).OBD: On-board Diagnostics unit.PC: Passenger Car.PEMS: Portable Emissions MeasurementSystem.

PHEM: Passenger car and Heavy duty vehicleEmission Model.PHEV: Plug-in Hybrid Electric Vehicle.

PM: Particulate Matter.PM2.5: Particulate matter having a diameterbelow 2.5 micrometres.Re: Reynolds number.RPM: Revolutions per minute (unit of enginerotating speed).SAD: Sum of Absolute Deviations.SCR: Selective Catalytic Reduction.SFTP: Supplemental Federal Test Procedure(USA standard driving cycle).SSR: Sum of Squared Residuals.

STU: Sustainable Transport Unit (administrativesubdivision of the Institute for Energy andTransport).SULEV: Super Ultra Low EmissionVehicle.THC: Total Hydrocarbons.TUG: Technical University of Graz(Austria).UNECE: United Nations EconomicCommission for Europe.USEPA: United States Environmental Protection

Agency.VELA: Vehicle Emissions Laboratories of JRC inIspra.

VSP: Vehicle-specic Power.WHSC: World Harmonised StationaryCycle.

WHTC: World Harmonised TransientCycle.

WLTC: Worldwide harmonised Light-dutydriving Test Cycle.

WLTP: Worldwide harmonised Light-dutydriving Test Procedures.

xxiv


29/234

xxv


30/234

xxvi


31/234

Chapter 1

Introduction

Pollutant emissions from road transport need to be accurately estimated to ensurethat air quality plans are appropriately designed and implemented. is is no easytask, because road vehicles are operated under a wide range of conditions and theiremissions exhibit a highly transient behaviour. Moreover, their emission proles havea strong dependency on vehicle class, on operating and environmental conditions,and on the characteristics of the fuels used. Last but not least, road vehicle emission

regulations are frequently updated, which drives constant technological changes inpowertrains, fuels and after-treatment devices.

Emission modelsare used to perform the calculations needed to estimate emissionsfrom road transport at large geographic scales (e.g., for the compilation of nationalemission inventories). ese range from models which only require mean travellingspeeds to estimate emissions (e.g., COPERT) and models that require a descriptionof the traffic situationi.e., qualitative assessments of driving conditionstopredict emissions (e.g., HBEFA) to models that require second-by-second engineor vehicle state data (e.g., PHEM, MOVES) to derive emission information forthe complete driving prole. Whatever the case, modelling road vehicle emissionsrequires intensive testing to produce emission factors that cover all the relevantdriving situations. On the other hand, vehicle emissions tests are costly experiments,and not every single vehicle model is tested for modelling purposes. is makestime-resolved, accurate emissions data a very valuable resource for vehicle emissionmodellers.

1


32/234

1.1. GOAL AND SCOPE

1.1 Goal and scope

is academic dissertation aims to improve the quality ofinstantaneousroad vehicleemissions measurement and modelling through the applicationofdata post-processingtechniques.

is document provides a description of the distorting effects that compromisethe accuracy of instantaneous emissions signals reported by conventional emissionmeasurement equipment, and a theoretical framework for their compensation. edistorting effects covered include the time misalignments among signals output bydifferent instruments, the mixing of exhaust gas within the exhaust system of vehiclesand within pipe elements of the measurement setup, the variable transport times ofexhaust gas associated with non-steady exhaust ow rates and the dynamic responsecharacteristics of gas analysers.

e most successful existing methods for the compensation of instantaneous vehicleemission signals rely upon systems theoryconcepts. ese methods individually modelthe constituting elementsor sub-systemsof the measurement chain (e.g., rawgas line, diluted gas line, chemical analysers) through unit step characterisationexperiments. ese methods, which require access to an ultra-fast analyser andadditional apparatus to produce the step inputs, will be reviewed and explained in

simple terms so that no expert knowledge is required.e main original contribution of the research work presented herein is theproposal of anew, comprehensive method for the compensation of instantaneous roadvehicle emission signals through data post-processing. e method can be applied toemission signals obtained from vehicle emissions testing on chassis and enginedynamometer facilities, or with portable emissions measurement systems (PEMS),therefore covering the three main setups used in vehicle emissions measurement formodelling purposes.

e new method uses CO2as a tracer of the distortions suffered by mass emissionsignals, and models the complete measurement setup as a lump system. In principle,the method could be made to operate with other gaseous pollutants (e.g., NOx),but the advantages of using CO2(which will become apparent in chapter5,wherethe method is described in detail) are too many to warrant the choice of anyother tracer gas. us the method is (rather unimaginatively) calledthe CO2tracermethod. In comparison to previous methodologies, the new method sacrices some

2


33/234

CHAPTER 1. INTRODUCTION

insight regarding the behaviour of emission signals within each sub-system of themeasurement setup in exchange for much lower experimental work and reduced

maintenance and technological requirements. It also brings improvements in termsof exibility, since it can be adapted to the specic measurement setups of theindividual vehicle emission laboratories with only minor effort.

1.2 Justication

e ERMES (European Research on Mobile Emission Sources; see section 2.1.1on page 12 for a description of its activities) is a European network of researchers,

regulators and other stakeholders in vehicle emissions modelling. ERMES iscoordinated by Directorate-General Joint Research Centre (JRC) of the EuropeanCommission (EC).1 e development priorities of the two leading vehicle emissionmodels in Europe (COPERT and HBEFA) are set during the ERMES annualplenary meetings. To that avail, the executive board of ERMES proposes a listof research issues to all stakeholders, and following an open discussion theseare assigned a priority level (high, medium or low). e work programmesof participating research bodiesas well as the measurement programmes ofparticipating laboratoriesare then adjusted according to these priority levels, and

a pool of measured data is made available to participating vehicle emission modellersas a result (gure1.1).

Amongst the high-priority research issues included in the ERMES work programmefor the year 2012 was the development of tools for data collection and for correctionof instantaneous test results. is resulted in an allocation of research efforts andfunds from the Sustainable Transport Unit (STU) of the Institute for Energy andTransport (IET, a part of JRC) to the development of a new instantaneous signalcompensation method that could be disseminated within the ERMES network oflaboratories.

1e author of this dissertation had the pleasure of serving as scientic secretary of the ERMESgroup between the years 2010 and 2012 before passing his duties onto the much more efficient (anddear friend and colleague) Mr Fabio Dalan.

3


34/234


35/234


2. Instantaneous vehicle emission signals reported by conventional vehicleemissions measurement instrumentation are affected by a number ofdistorting

effectsthat compromise their accuracy.(a) ese distorting effects have little impact upon aggregated vehicle

emission test results (i.e., reported pollutant mass emissions over timeperiods larger than a few seconds), and so they have been largelyoverlooked by emission modellers.

(b) ese distorting effects can have a large impact upon the accuracy ofinstantaneous vehicle emission test results (i.e., pollutant mass emissionsover time periods smaller than a few seconds), and also upon thereliability of emission factors and models developed from them (e.g.,engine emission maps).

3. e aforementioned distorting effects can be compensated through acombination ofphysical modellinganddata post-processing, thereby improvingthe accuracy of the signals and the reliability of emission factors developedfrom them.

(a) Signal compensation approaches that rely upon physical modellingperform well, but the compilation of a complete physical model of a

vehicle emissions measurement setup is a difficult task that requiresexpert knowledge andad hoccharacterisation experiments.

(b) Signal compensation approaches that rely upon data post-processingrequire fewer experiments and resources for their implementation, butthey are less sophisticated and provide less information about the ways inwhich instantaneous emission signals become distorted throughout themeasurement setup.

(c) A new methodology for the compensation of instantaneous emissionsignals could be developed by combining new, advanced data post-

processing methods with existing knowledge from previous physicalmodelling efforts.

5


36/234

1.4. OBJECTIVES

1.4 Objectives

e stated goal of the research work presented in this document is to improve thequality of instantaneous road vehicle emissions measurement and modelling throughthe application of data post-processing. e specic research objectivesthat will beaddressed in relation to the aforementioned goal are structured as follows:

1. To assess the distorting effects that compromise the accuracy of instantaneousvehicle emissions signals used for vehicle emissions modelling, and to developamethodological frameworkfor the compensation of instantaneous emissionsignals. It is assumed that this could be accomplished by achieving thefollowing secondary objectives:

(a) To identify and describe all the relevant distorting effects and the signalsaffected.

(b) To assess the impact of the identied distorting effects upon the accuracyof instantaneous emissions signals reported by conventional emissionmeasurement equipment and upon emission factors derived from non-compensated signals.

(c) To make a classication of these effects with regard to vehicle emissionsmeasurement and modelling, presenting them in a clear manner and

providing simple, best-practice data compensation strategies.2. To develop a comprehensive, exible methodology for the compensation

of the distorting effects, and disseminate it among the stakeholders of theERMES group. It is assumed that this could be accomplished by achievingthe following secondary objectives:

(a) To study the existing instantaneous emission signal compensationmethodologies, identifying potential shortcomings and reasons whyexpert compensation methodologies are not standard practice among

vehicle emission laboratories.(b) To formulate and validate a new signal compensation methodologythat relies heavily on data post-processing. Said methodology should beexible and easy to apply, and yet have performance levels comparable tothose of advanced compensation strategies based on physical modelling.

(c) To produce a software implementation of the new methodology toencourage its adoption by the laboratories of the ERMES network.

6


37/234


1.5 Research methodology

e research methodology followed is outlined next:

An extensive literature reviewof the techniques available for vehicle emissionmeasurement, and the ways in which the resulting emissions data can beused for the development of emission factors (i.e., of functional relationsbetween vehicle operation parameters and their emission behaviour) wasperformed. A strong focus was placed on those techniques that are able toproduce continuous streams ofinstantaneous data. e special characteristicsof instantaneous emission signals were investigated, as well as the distorting

effects that compromise their accuracy.

An expert consultation process2 was initiated by ECJRC to study the problemof instantaneous emission signal accuracy and to explore the possibility ofestablishing harmonised approaches to be shared among ERMES laboratories.

As a result of this process, ECJRC took the initiative of drafting amethodological report on the topic and coordinating future developments.

Astudy of signal compensation strategiesbased on data post-processing was performedwith the aim of developing a new, comprehensive signal compensation

method. Early methodology concepts were initially applied to historicaldatasets extracted from the database of the VELA laboratories of IET.

An experimental campaign was designed and executed once the compensation meth-odology had reached a sufficient level of maturity. e experimental campaignincluded the measurements required in order to obtain the necessary data forfurther development and validation of the new compensation methodology,as well as those experiments needed for the physical modelling of theVELA 1 test facility (a chassis dynamometer laboratory used to measure the

emissions of internal combustion engine passenger cars). ese experimentswere performed with the technical assistance of Dr Savas Geivanidis.

2Said process involved Dr Martin Weilenmann of the Swiss Federal Laboratories for MaterialsTesting and Research (EMPA), Dr Savas Geivanidis and Dr Leonidas Ntziachristos of the Laboratoryof Applied ermodynamics (LAT) at Aristotle University of essaloniki (Greece), Dr StefanHausberger of the Technical University of Graz (TUG, Austria), and Dr Panagiota Dilara and DrGeorgios Fontaras of ECJRC.

7


38/234

1.6. STRUCTURE OF THIS DISSERTATION

1.6 Structure of this dissertation

is academic dissertation is divided into seven chapters, the rst of them being thisintroduction which now concludes.

Chapter 2, Measuring the emissions of road vehicles, critically reviews thedifferent techniques available for the measurement of road vehicle emissionsin relation to the development ofemission factorsfound in emission models.Each technique is assessed from a modelling perspective, emphasisingthose techniques that produce continuous streams of instantaneous data(namely engine and chassis dynamometer measurements, and PEMS). is

review is adapted from a journal article co-authored by the author of thisdissertation (Francoet al.2013).

Chapter3,Distorting effects of instantaneous vehicle emission signals, discussesthe distorting effects that compromise the accuracy instantaneous vehicleemission signals. e most relevant distorting effects (namely the timemisalignments among signals output by different instruments, the variabletransport times of exhaust gas associated with non-steady ow rates, themixing of exhaust gas within the exhaust system and the pipe elements of themeasurement setup, and the dynamic response of gas analysers) are describedand assessed with regard to their impact upon the accuracy of the emissionsignals themselves, and that of the emission factors derived from them.

Chapter 4, Compensation of instantaneous vehicle emission signals, beginswith a proposal of amethodological frameworkfor the advanced or expertcompensation of instantaneous vehicle emission signals. is framework isused to support the subsequent analysis of its constituting methodologicalelements. e chapter concludes with a review of existing signal compensationmethodologies. is review places a special focus on compensation methods

based upon physical modelling followingsystems theorypractice, which wereconsidered as a benchmark for eventual new developments.

Chapter 5, e CO2 tracer method, contains the complete description ofthe instantaneous signal compensation methodology that constitutes themain original contribution of this dissertation. e description covers boththeoretical and practical aspects.

8


39/234


Chapter6,Results and discussion, presents the results of an actual application ofthe CO2tracer method to instantaneous emission datasets obtained from an

experimental campaign carried out at the VELA facilities (Vehicle EmissionsLaboratories) of IET. It also discusses the CO2tracer method in relation toearlier methodologies, and presents a validation of the method.

Chapter7,Conclusions, presents the conclusions of our research and is the nalchapter of this dissertation.

9


40/234

1.6. STRUCTURE OF THIS DISSERTATION

10


41/234

Chapter 2

Measuring the emissions of road

vehicles

Air pollution is a signicant risk to human health and to the environment.Outdoor air pollution is estimated to cause 3.7 million annual premature deathsworldwide (WHO 2014). Road transport often appears as the single most importantsource of urban pollutant emissions in source apportionment studies (e.g., Maykutet al.2003;Querolet al.2007). In the coming decades, road transport is likely toremain a large contributor to air pollution, especially in urban areas.

For this reason, major efforts are being made by regulators, the automotive industryand the scientic community in order to reduce polluting emissions from roadtransport. ese include new powertrains and emission after-treatment technologyimprovements, fuel renements, optimised urban traffic management methods, andthe implementation of tighter emission standards (EC2011b).

2.1 Emission models and emission factors

Pollutant emissions need to be accurately estimated to ensure that air qualityplans are designed and implemented appropriately. Road vehicle emissions areparticularly challenging in this regard because they depend on many parameters,such as vehicle characteristics and emission control technology, fuel specications,

11


42/234

2.1. EMISSION MODELS AND EMISSION FACTORS

and ambient and operating conditions (gearshift strategy, temperature of engine andafter-treatment devices and others). Due to this complexity, and to the variety of

vehicle types available in the market,emission modelsare necessarily used to performthe calculations of road transport emissions that support regional or national emissioninventories.

Smit, Ntziachristos and Boulter (2010) proposed a classication of vehicle emissionmodels in ve major categories according to theinput datarequired. ese rangefrom models which only require mean travelling speed to estimate emissions andmodels that need traffic situations (i.e., qualitative assessments of driving conditions)to express emissions, to models which require second-by-second engine or vehiclestate data to derive emission information for the complete driving prole. Regardless

of the differences among practical implementation, every road vehicle emissionmodel will be based onempirical emissions data, and it will provide a collection ofemission factors(EFs), which are functional relations between pollutant emissionsand the transport activity that causes them.

is chapter is not intended as a review of the existing vehicle emission models,1 butrather as a review of how emissions data may be collected and presented inthe form of EFs. We will therefore examine the experimental techniquesused tomeasure road vehicle emissions in relation to the development of EFs found inemission models. e emission measurement techniques covered include those most

widely used for road vehicle emissions data collection, namely chassis and enginedynamometer measurements, remote sensing, road tunnel studies and portableemission measurements systems (PEMS). An earlier, similar review was performedby Faiz, Weaver and Walsh (1996). Strong points and limitations are presentedfor each method, together with literature examples of successful implementations.e main advantages and disadvantages of each method with regard to emissionsmodelling are also presented.

2.1.1 European emissions modelling: the ERMES group

e development of accurate EFs found in road vehicle emission models is ajoint enterprise among several parties that requires intensive testing to adequatelycover all the relevant vehicle types and driving conditions, and substantial research

1Kousoulidouet al.(2010d) recently performed a review of vehicle emission models and inventorytools used worldwide.

12


43/234

CHAPTER 2. MEASURING THE EMISSIONS OF ROAD VEHICLES

and modelling efforts to keep up with technological advances. Measurement andmodelling methodologies need to be improved to accurately reect real-world

emissions, and this needs to be accomplished with limited resources.In the European context, these circumstances led to the creation of the ERMESgroup (European Research group on Mobile Emission Sources) in 2009 in aneffort to bring together all European research groups working on transport emissioninventories and models, as well as funding agencies and other stakeholders, underthe coordination of the European Commission Joint Research Centre (ECJRC).e ERMES group aims to support a permanent network of mobile emissionmodellers and users, to coordinate research and measurement programmes for theimprovement of transport emission inventories in Europe and to become a referencepoint for mobile emissions modelling and related topics in Europe while maintainingand improving international contacts (table2.1).

Table 2.1:Selected ERMES group stakeholders

Laboratories and member state representatives

Umwelt Bundesamt(Germany) IFSTTAR (France) TNO (the Netherlands) EMPA (Switzerland)

ARPA Lombardia (Italy) University of Leeds (United Kingdom) INSIA-UPM (Spain) UmweltbundesamtGmbH (Austria) Trakverket (Sweden) KLIF (Norway) CASANZ (Australia)

European Commission and others

European Commission DG CLIMA European Environment Agency (EEA)

Vehicle emission modellers

LAT, Aristotle University of essaloniki(Greece)

Technical University of Graz (Austria) INFRAS (Switzerland)

Industry representatives

Conservation of Clean Air and Water inEurope (CONCAWE)

European Automobile Manufacturers Associ-ation (ACEA)

Ford Motor Company AVL MTC

ERMES group chair

European Commission DG JRC

e three-level structure adopted by the group (schematically represented ingure 2.1) comprises an executive boardcomposed of a reduced number ofsenior scientists in charge of drafting the work programme and monitoring itsprogress, a working group on models and emission factorswhich includes the

JRC, the nancing EU Member States, laboratories and national expertsand abroad contact group including the European Environment Agency (EEA), relevant

13


44/234

2.1. EMISSION MODELS AND EMISSION FACTORS

Directorate-Generals of the European Commission, industry representatives and allother interested parties (Franco, Fontaras and Dilara2012).

Executive board

Working groupon models andemission factors

Members: who? Roles: what?

Contact group

Relevant DGs ofEC, industry, EEA,others (annualplenary meeting)

EC-JRC, Member States,laboratories andnational experts(periodic meetings)

Reduced number ofexperts (frequentcontacts)

Proposal of workprogramme, progressmonitoring

Development of models,emission factor updates

Review of emission factors,demands/funding for futureresearch, reports on specialtopics

Figure 2.1:Structure of the ERMES group

2.1.2 ERMES models: COPERT and HBEFA

COPERT (Gkatzoias et al. 2007) and the Handbook of Emissions Factors(HBEFA; de Haan and Keller2004)are the two leading emission models in Europe.COPERT is the main road transport emissions model of the EMEP/CORINAIR

Atmospheric Emissions Inventory Guidebook (AEIG), and it is used by severalEuropean member states to compile their official national inventories of emissionsfrom road transport. HBEFA is mostly meant for use at ner geographic scales (downto street canyon level) and requires more detailed traffic data inputs than COPERT.HBEFA is mostly used in the DACH-NL-S European countries (Germany, Austria,Switzerland, the Netherlands and Sweden).

Both models estimate traffic emissions in three large blocks, namely emissionsproduced during thermally stabilised engine operation (hot emissions), excessemissions occurring during engine start from ambient temperature (cold-startand warming-up effects) and non-methane volatile organic compound (NMVOC)emissions due to fuel evaporation. Total emissions are calculated as the combinationof vehicle eet and activity data selected by the user and the libraries of emissionfactors included in the models.

14


45/234


In the case of COPERT, emission factors are speed-dependent, whereas HBEFAprovides emission factors for different traffic situations,i.e., qualitative descriptions

of the traffic environment. Even though the approaches behind COPERT andHBEFA are somewhat different, both models are largely underpinned by the sameexperimental data, and further methodological convergence is expected under thesteering of ERMES.

HBEFA and COPERT are backed up by extensive amounts of vehicle emissions testdata, and they provide emission factors for a comprehensive set of vehicle categoriesof current and past technology. e emission factors in both COPERT and HBEFAhave achieved a high level of quality over the past decade thanks to common projectssuch as ARTEMIS (Keller and Kljun2007) or COST 346 (Sturm and Hausberger

2005), together with the national activities and joint funding actions between theDACH-NL-S group and JRC. Nevertheless, both models are in need of variousmethodological improvements (such as the one presented in this dissertation) andof frequent updates to keep up with new vehicle technologies.

2.2 Chassis and engine dynamometer testing

A chassis dynamometer simulates the resistive power imposed on the wheels of avehicle during real driving. It consists of a dynamometer that is coupled via gearboxesto drive lines that are directly connected to the wheel hubs of the vehicle, or to a setof rollers upon which the vehicle is placed, and which can be adjusted to simulatedriving resistance. A separate roller will be required for each drive axle (i.e., each axlebeing driven by the engine).

During chassis dynamometer testing, the vehicle is tied down so that it remainsstationary while a professional driver operates it according to a predetermined time-

speed prole and gear change pattern shown in a monitor (the so-called driver'said). e driver operates the vehicle to match the speed required at the differentstages of the driving cycle (Nine et al. 1999). Chassis dynamometer test cycles(cf.section2.2.1)are typically transient (Yanowitz, McCormick and Graboski2000)and therefore the driver must anticipate and comply with changes in the requiredspeed within a specied tolerance (Wanget al.1997). Experienced drivers are ableto closely match the target speed prole.

15


46/234

2.2. CHASSIS AND ENGINE DYNAMOMETER TESTING

e load applied to the vehicle via the rollers can be controlled by the laboratoryoperators to simulate aerodynamic and rolling resistance (road load) for the vehicle

under test, while the size of the rollers and the use of ywheels accounts for vehicleinertia. e exhaust ow rate is continuously monitored, and a proportional sampleof the vehicle exhaust gas is collected in polymer bags for later analysis, or processedby chemical analysers attached to the sampling line (see gure 2.2). On-line analysersare increasingly being installed in test benches in order to record vehicle emissionswith a ne temporal resolution. ese analysers may either sample alone or in parallelwith the bag sampling technique at the diluted gas line (that is, at the end of thedilution tunnel, where the total ow is quasi-steady and easier to measure).

Dilution airDrivers aid

Roller bench

Particle measurement system

Control unit

Heat exchanger

Dilution tunnel NOx

CO2

CO

HC

Exhaust gas

sampling bags

Blower

Gas analysers

Figure 2.2: Schematic representation of a chassis dynamometer emissions test facility

Because dynamometer facilities are designed to meet regulatory standards, theirresults are viewed as highly accurate as long as proper calibration and maintenanceprograms are established (Traver et al. 2002). Also, they may be enclosed inclimatically controlled test cells to simulate driving under a wide range oftemperatures, including sub-zero tests.

A disadvantage of a chassis dynamometer testing is that it does not necessarily

represent real-world emissions of individual vehicles. is is due to the limitedrange of test conditions (e.g., the set ambient temperatures and the preconditioningroutines, the absence of road gradients) and to the fact that a dynamometer isimplemented instead of actual driving. In particular, the driving resistance valuesthat simulate road load are obtained from vehicle coast-down tests often performedunder articially favourable conditions, thus frequently yielding fuel consumptionand emission results below real-world levels (Mellioset al.2011). Moreover, chassis

16


47/234


dynamometer test results may not be representative of the emissions of entire vehicleeets because typically only a few vehicles from each technology class are tested for

modelling purposes.Anengine dynamometeris a device that simulates the resistive power directly at theengine power output. In an engine dynamometer test cell, the dynamometer shaftis directly connected to the engine shaft and it measures power at the ywheel ofthe engine, where no transmission or driveline losses inuence the results. Fullytransient dynamometers may absorb or place any specied load (within limits) onthe engine, even during load and speed change conditions. Engine test cells mayalso be climatically controlled. Emission measurements on an engine dynamometerrequire removing the engine and the exhaust gas after-treatment system from the

vehicle (Oh and Cavendish1985; Arteltet al.1999).Heavy-duty vehicle (HDV) engines can be coupled to many different chassis andbody types. Heavy-duty type-approval regulations have historically relied uponengine dynamometer testing because it would be impractical to type-approve allthe possible combinations. A disadvantage of this approach is that the emissions ofthe complete vehicle are not reected in engine testing, even though modern enginetest benches can be made to run any real-world engine load test cycle by simulatingthe vehicle to get torque and engine speed curves, either off-line or as hardware-in-the-loop simulations (HILS;cf. Lee2003). In the past few years, the increasingtechnological sophistication of engine and after-treatment control systems of HDVshas made it cumbersome to perform engine dynamometer tests independently ofmanufacturers, which in turn continue to use this technique for engine and after-treatment device development, both for heavy-duty and light-duty vehicles (LDVs).Chassis dynamometer testing has thus become the primary source of emissions datafor the development of EFs for recent-technology HDVs (e.g., the current EUROVI technology class).

2.2.1 Test cycles

A (standard) emissions test cycle (or driving schedule) is a predened driving prolethat the vehicle or engine under test has to follow. Test cycles last forseveral minutes,and often comprise several parts (or sub-cycles) that represent different drivingconditions (e.g., urban or highway driving). Test cycles are an integral part of allchassis and engine dynamometer tests, and their representativity and completeness

17


48/234


(i.e., their ability to statistically represent the driving conditions under study) areessential to achieve good testing results (Andr and Rapone2009). e number

of engine and vehicle dynamometer test cycles used worldwide for emission andfuel consumption measurements is continuously expanding to cover regulatoryneeds, while also trying to simulate real-world driving conditions (Andr et al.2006).

Two categories of test cycles may be used in chassis or engine dynamometertests, namely steady-state (or modal) and transient cycles: steady-state test cyclesinvolve running the engine or vehicle under a number of xed operating pointsor modes, each one of them dened by a certain constant engine speed andload. For each mode, the engine or vehicle is operated for a sufficient amount

of time to produce relatively stabilised emission rates. When two or more modesare included in the test cycle, the emissions measurements from each mode aretypically combined using a weighted averaging scheme, with specic denitions ofeach mode and weighting schemes differing from one test cycle to another (Arteltet al. 1999). On the other hand, transient test cycles include variations in theoperating conditions as part of the test procedure and they are regarded as morerepresentative of real-world operation because they can be designed to account forreal-world situations such as idling, acceleration and deceleration. Detailed technicalinformation on the most commonly used standard driving cycles can be found in

the literature (CONCAWE2006a,2006b; Barlowet al.2009).Chassis dynamometer test cycles are predominantly transient. is is the case of type-approval cycles such as the FTP and SFTP (Federal Test Procedure and SupplementalFederal Test Procedure, used for emission certication of LDVs in the USA) and theNEDC (New European Driving Cycle, used for emission testing and certicationof all Euro 3 and later LDV models in Europe). e latter has often been criticisedfor being too smooth and underloaded for typical vehicle operation, as it covers onlya small area of the operating range of engines (Kgeson1998;Mellioset al.2011;

Weisset al.2011a).

In general, type-approval cycles underestimate real-world emissions because theyexhibit low speed dynamics to ensure that their trace can be followed by the lesspowerful vehicles, and also because manufacturers are able to optimise the emissionperformance for specic operating points (Ntziachristos and Samaras 2000). Inorder to address some of the shortcomings of current type-approval test cycles, anew transient chassis dynamometer test cycle (Worldwide harmonised Light-duty

18


49/234


driving Test Cycle, WLTC) and its accompanying test protocols (WLTP) are beingdeveloped within the framework of a larger UNECE project set to produce a global

technical regulation for harmonised testing of LDVs.Besides standard driving cycles used for regulatory purposes, there are many othersthat characterise driving in specic areas (Esteves-Boothet al.2001)or a particulartechnology (Rapone et al. 2000). From the EF development perspective, the so-calledreal-world cyclesprovide the most valuable emissions data thanks to a widercoverage of engine operating points in comparison to type-approval cycles. Examplesof real-world cycles are the test cycles for LDVs and HDVs included in the defaultdatabase of the MOVES model (USEPA 2012), the ARTEMIS suite of LDVcycles (Andr2004) and the recently developed ERMES test cycle (gure 2.3) which

was specically designed for emission modelling purposes. e short duration of theERMES cycle is an advantage in terms of testing schedule exibility and costs ofindividual test runs (Knorr, Hausberger and Helms2011).

0

1

2

3

4

5

6

0

20

40

60

80

100

120

140

160

Gear

Velocity[

kmh

]

Time [s]

VelocityGear (Diesel 6 gears)

0 400 800200 600 1000 1200 1400

-1

Figure 2.3:Time-velocity prole of the ERMES chassis dynamometer test cycle

Engine dynamometer test cycles are predominantly modal. Some of the main modaltest cycles used around the world for the type-approval of current-technology heavy-duty engines include the European Stationary Cycle (ESC; applicable in Europe,and comprising 13 modes) and the Supplemental Emissions Test (applicable in theUSA, also comprising 13 modes). A few transient test cycles are also in use, suchas the European Transient Cycle (ETC) and the US heavy-duty engine Federal Test

19


50/234


Procedure (FTP). For newer technology engines (EURO VI and beyond), the WorldHarmonised Stationary Cycle (WHSC) and its transient counterpart (WHTC) have

been proposed by the UNECE GRPE group in an effort to create global cycles thatreproduce typical driving conditions in the EU, USA, Japan and Australia (Steven2001).

2.2.2 EF development from dynamometer laboratory data

Chassis dynamometertesting has achieved a high degree of standardisation and isarguably the most proven technology for vehicle emissions measurements. In orderto obtain robust EFs (i.e., ones that are unlikely to change within the accepted

uncertainty if there was repetition of the original measurement programme ormodelling activity;cf.IPCC2003), a sufficiently large number of vehicles shouldbe tested repeatedly under different driving cycles.

Engine dynamometertesting is somewhat less useful for EF development because itproduces results in units of quantity of pollutant emitted per unit of engine energyoutput (such asg kW h1), which are not directly relevant to real-world activitypatterns. To estimate total emissions using this type of EF, one needs to estimate orcalculate the engine power prole over a trip travelled and apply a relevant EF.

A straightforward approach to EF development is to plot the aggregatedor bagresults of various driving cycles with respect to the mean speed or another aggregatedkinematic parameter (e.g., mean acceleration or relative positive acceleration) of thespecic cycle and then t a polynomial trend line to the experimental data usingmathematical regression. e resulting formula of the trend line is the EF thatexpresses vehicle emissions as a function of the selected parameter. A disadvantage ofsuch a simple approach is that it may not adequately capture the impact of differentdriving cycles upon emission performance.

20


51/234


An illustration of this situation isprovided in gure 2.4(see right hand

column of this page), where the massemissions of NOx (in g km1) over

forty-one different driving cycles andsub-cycles for fteen different DieselEuro 4 passenger cars of similar enginecapacity are summarised as a functionof mean cycle or sub-cycle velocity.Figure2.4shows the mean, maximum,and minimum recorded value for eachsub-cycle to illustrate the variability ofthe emission levels within a given vehicleclass, and also the variability of averageemission levels from different test cycleswith similar average velocities.2

e observed variability is much higherfor CO and HC and lower for CO2and fuel consumption than it is forNOx (Boulter and McCrae 2007;

Zallinger 2010), and it is generallyhigher at lower mean velocities (Choiand Frey 2009, 2010). In any case,the development of robust EFs fromchassis dynamometer data requires themeasurement of a sufficient number ofvehicles and an adequate selection ofdriving cycles that are representativeof the driving conditions being

modelled.

0.0

1.0

2.0

3.0

4.0

5.0

6.0

7.0

8.0

HBR4II (StGoHW) (7.1)ART_urban_3-hot (8.6)

ART_urban_4-hot (11.7)


ART_urban-cold (17.3)

ART_urban-hot (17.3)

Legisl_ECE_2000-cold (18.7)


Legisl_US_FTP2-hot (25.7)


HB_R3_III (LE5)-hot (32.0)

HB_R4_I (LE6)-hot (33.6)

Legisl_NEDC_2000-cold (33.6)

Legisl_NEDC_2000-hot (33.6)

Legisl_US_FTP1-cold (41.1)

Legisl_US_FTP3-hot(41.1)

ART_rural_3-hot (43.1)


HB_R3_II (LE3)-hot (52.4)

HB_R3_I (LE2u)-hot (52.9)

ART_rural-hot (61.4)

Legisl_EUDC-hot (62.4)

HB_R2_III (LE2s)-hot (65.6)


HB_R2_II (LE1)-hot (77.2)



HB_R2_I (A4)-hot (89.5)

HB_R1_III (A3)-hot (100.3)

ART_MW_150_2-hot (102.2)

HB_R1_II (AE2)-hot (107.7)

ART_MW_130-hot (114)

EMPA_BAB-hot (117.4)

HB_R1_I (AE1)-hot (118.3)

ART_MW_150-hot (120.0)

ART_MW_150_1-hot (122.5)

ART_MW_130_3-hot (123.0)

ART_MW_150_3-hot (125.3)

ART_MW_130_4-hot (128.0)

ART_MW_150_4-hot (133.2)

HBR4III (StGoUrb) (4.1)

Subcycle ID(mean velocity in kmh )-1meanmin. max.

NO emissions per sub-cycle [gkm ]-1x

Figure 2.4:Illustration of the variabilityof chassis dynamometer test results

2Source: Artemis 300 database (many thanks go to Stefan Hausberger for providing the data anthe idea as to how to illustrate this point).

21


52/234


Whenever instantaneous data from on-line analysers (typically with a samplingfrequency of 1Hz or higher) are available in addition to aggregated (or bag)

values, other, more elaborate approaches to EF development may be considered.In such cases, measured emission values can be related to recorded instantaneouskinematic parameters or engine covariates. is is done in the MOVES model, whichuses the metric of vehicle-specic power (VSP) and instantaneous velocity to bininstantaneous data and estimate EFs, and also in PHEM (Passenger car and Heavyduty vehicle Emission Model), which uses modal chassis and engine dynamometertest data to produce engine emission maps that predict pollutant mass emissionsas a function instantaneous engine speed and engine power, normalised by theirmaximum rated values (see gure2.5).

Enginepower[kW]

11.19.38.99.18.67.1

6.46.35.95.85.96.6

4.15.34.93.93.5

3.43.63.83.64.1

3.33.54.1

6.5

Engine speed [rpm]

Pollutant mass emissions [gs ]-1

11.2

Figure 2.5: Example of engine emission map [adapted from Kousoulidou et al.(2010d)]

An advantage of EFs derived frominstantaneous emissions datais that they allow for

the simulation of fuel consumption and emissions for any driving pattern and vehicleconguration (Kousoulidou2011). For such applications, the instantaneous massemissions for a non-measured driving pattern can be interpolated from measureddata. However, the creation of engine emission maps requires additionaldata post-

processingefforts to ne-tune the results. For example, maps derived from modal datamay have to incorporate correction factors to account for the excess emissions typicalof transient states, or be calibrated with bag results (Hausbergeret al.2009).

22


53/234


Additional complications arise iftransient dataare used to develop the instantaneousEFs. In this case, the ne allocation of instantaneous mass emissions to engine state

data poses a technical challenge because pollutant concentrations are affected by anumber of distortions including mixing of the exhaust gas, variable transport delaysand dispersion due to the nite dynamic response characteristics of gas analysers. Aspointed out in chapter 1, all of these effects result in measured emission signals whichare smoothed, dynamically delayed instances of the true signals at the catalyst-outpoint. A number of methodological proposals have been made to address this issuethrough data post-processing [see, for example, the work of Weilenmann, Soltic and

Ajtay (2003); Ajtay and Weilenmann (2004) and Geivanidis and Samaras (2008)],but they are not applied in routine measurements.3

2.2.3 Chassis and engine dynamometer applications

Chassis dynamometermeasurements are used for the type-approvalof road vehiclesand engines (i.e., to check and certify the compliance with legal emission limits).Scientic studies involving chassis dynamometer measurements are conducted forseveral other purposes, including the investigation of the emission characteristicsof specic types of vehicles or pollutants, the assessment of emission controltechnologies or the analysis of the emissions performance of different types of

fuels.Chassis dynamometer testing can cover a wide range of pollutants depending on thetype of analysers used to process vehicle exhaust. As far as the type of vehicles testedis concerned, chassis dynamometer tests are more commonly used for motorcycles,passenger cars and light commercial vehicles (Pelkmans and Debal2006;Fontaraset al.2007;Fontaras, Pistikopoulos and Samaras2008; Chianget al. 2008) thanfor HDVs (Wanget al.1997; Morawskaet al.1998;Whiteld and Harris1998;Ramamurthy and Clark1999; Clarket al. 2002), because only the more costly HDVchassis dynamometer laboratories can accommodate these larger vehicles.

Chassis dynamometer studies have been used to investigate the emission proles ofseveral pollutants from different types of vehicles under variousoperating conditions.For instance, Yanowitzet al.(1999)reported the emissions of regulated pollutantsfrom twenty-one in-use Diesel HDVs as measured on a chassis dynamometer for

3One such methodological proposaldescribed in chapter5is the main original contributionof this dissertation.

23


54/234


three different standard driving cycles. Mohr, Forss and Steffen (2000) carried out anexperimental study on particulate emissions of gasoline vehicles with three passenger

cars at a chassis dynamometer. Durbin et al. (2002) investigated the variation ofammonia emissions across different driving cycles. Soltic and Weilenmann (2003)studied the total amount and the partitioning of NOxemissions over different testcycles for sixteen Euro 2 light-duty vehicles. Heeb etal. (2003) reported the emissionsof methane, benzene and the alkyl benzene class compounds for gasoline passengercars from Euro 0 to Euro 3 technology class over the US urban driving cycle (FTP),measured by chemical ionisation mass spectrometry (CI-MS) under cold and hotengine conditions.

Inordertoassesstheinuenceofcold engine start eventsupon emissions, Weilenmannet al. (2005) measured and analysed benzene and toluene emissions at variousambient temperatures for Euro 3 gasoline cars, Euro 2 Diesel cars and pre-Euro 1gasoline cars over a repetitive urban real-world test cycle. More recently, Livingston,Rieger and Winer (2009) used a chassis dynamometer equipped with FourierTransform Infrared Spectroscopy (FTIR) analysers to measure tailpipe ammoniaemissions from a random sample of light and medium-duty vehicles, while Adamet al. (2011) investigated the time-resolved emissions of several hydrocarbonsand other unregulated pollutants of a medium-sized truck using state-of-the-artspectrometers, and Fontaraset al.(2013) investigated the emission prole of Diesel

and gasoline Euro 5 passenger cars over several test cycles.

Chassis dynamometer studies have been used to assess the performance characterist-ics of different emission control technologies. For example, Huai et al. (2004) measuredN2O emissions from different vehicle technologies ranging from non-catalyst tosuper-ultra-low-emission vehicles (SULEV) over the FTP and other, more aggressivecycles. Heeb et al. (2006a, 2006b) studied the efficiency of catalytic reduction of NOand the selectivity towards NH3 and analysed the parameters with impact uponthe NH3output of a three-way catalyst equipped gasoline LDV during real-world

driving. Olfert, Symonds and Collings (2007) studied the modal particle emissionsof a passenger car equipped with a Diesel oxidation catalyst (DOC). Bergmannet al. (2009) evaluated the efficiency of a passenger car Diesel particulate lter(DPF) over NEDC and a custom acceleration-deceleration chassis dynamometercycle, while Biswaset al.(2008, 2009) studied the particulate emissions of DieselHDVs retrotted with recent-technology after-treatment systems (DPF and selectivecatalytic reduction, SCR) over steady-state and transient cycles.

24


55/234


Repeated chassis dynamometer testing is an excellent way to assess the influence ofdifferent fuelsupon vehicle emissions, because the repeatable conditions allow even

small emission and consumption differences to be observed. Chaoet al.(2000)usedchassis dynamometer measurements to study the effect of an additive containingmethanol on the emission of carbonyl compounds from a HDV Diesel engine.Some chassis dynamometer studies (Wanget al.2000;Fanick and Williamson2002;USEPA2002) compared exhaust emissions from in-use heavy-duty trucks fuelledwith biodiesel blends with those from trucks fuelled with Diesel fuel.

e impact upon both particle and carbon dioxide emissions of alternative fuels suchas liqueed petroleum gas (LPG) and unleaded gasoline was analysed by Ristovskiet al. (2005). Peng et al. (2008) examined the effect of biodiesel blend fuelon aldehyde emissions in comparison with those from Diesel fuel. Aslam et al.(2006) studied the performance of compressed natural gas in comparison withgasoline. Nelson, Tibbett and Day(2008) reported the emissions of a range of toxiccompounds from twelve in-use vehicles which were tested using urban driving cyclesdeveloped for Australian conditions and Diesel fuels with varying sulphur contents.More recently, Kousoulidouet al.(2010b) studied the impact of biodiesel on theregulated pollutant emissions and fuel consumption of a modern passenger car,and Fontaras et al. (2012) assessed the on-road emissions of four Euro V Dieseland CNG waste collection trucks. A study by Kousoulidouet al.(2012)used chassis

dynamometer test data to provide correction factors for pollutant emissions whenbiodiesel is used on passenger cars at different blending ratios.

Chassis dynamometer measurement data have also been used as an input to specicemission models. For example, the study of Kear and Niemeier (2006)used chassisdynamometer test data to derive a model aimed at developing operational correctionfactors for distance-based HDV Diesel particle EFs measured on standard testcycles for real-world conditions. e study of Fontaras et al. (2007) presentedthe application of a tool for predicting CO2 emissions of vehicles as measured

on a chassis dynamometer under different operating conditions, and these resultswere directly used to represent hybrid EFs in COPERT. In the context of theDECADE projectcarried out under the 5th Framework Programme of theEuropean Commissiona software package was developed to predict vehicle fuelconsumption and emissions for a given distance-speed prole. Specic LDVs weresubjected to measurements on engine dynamometers in order to give input to themodel (Pelkmans and Debal2006).

25


56/234


Engine dynamometer test data have been found to be especially accurate for thesimulation ofinstantaneous fuel consumption. Moreover, the upcoming HDV CO2

monitoring regulation is expected to rely upon a combination of vehicle componentmeasurement and modelling and vehicle simulation (Hausberger et al. 2012).

Newer powertrain congurations such as Hybrid Electric Vehicles (HEVs) and full-electric vehicles (EVs) need modied test benches to evaluate the electric powerows among the driveline components. UNECE Regulation 101 (UNECE2005)denes standard test procedures for the measurement of fuel consumption andCO2emissions from passenger cars and LDVs, including HEVs and plug-in hybridelectric vehicles (PHEVs). is regulation also proposes a method to measurethe electric rangeof EVs and PHEVs. Silva, Ross and Farias (2009) proposed a

methodology based on the SAE J1711 standard to produce dynamometer-basedEFs for PHEVs, including life cycle assessment considerations to make comparisonsfair among different vehicle technologies. In general terms, chassis dynamometers(coupled with some degree of hardware simulation) capture the emissions of newerpowertrain congurations better than engine dynamometers (ICCT2012).

2.2.4 Summary

e characteristics of engine and chassis dynamometer measurements are summar-ised in tables2.2and2.3.

26


57/234


Table 2.2: Summary of vehicle emission measurements in chassis dynamometer laboratories

Applications Characteristics

General Type-approval and general testing of

LDVs. General testing of HDVs. More commonly used for LDVs than

for HDVs.EF development Provides high-resolution EFs. e quality of EFs is linked to the

representativity of the test cycle used. e use of standardised cycles leads to

good repeatability and comparabilityof results (slightly worse than enginedynamometer testing).

Advantages Proven technology, highly automated and economically

optimised. Good precision and repeatability. e complete vehicle is measured, leading to better

representativeness for on-road duty cycles (as comparedto engine dynamometers).

Emission measurements are obtained in units that aremore useful for emission inventory purposes.

Relatively inexpensive (especially for LDVs).Disadvantages Driving cycles/dynamometer load may not be repres-

entative of real-world conditions. Limited exibility totest alternative driving conditions.

Dilution conditions not representative of real world(especially for particle number emissions).

Table 2.3: Summary of vehicle emission measurements in engine dynamometer

Emission test report

Documents

Transcript of Emission test report