Environmental traffic management: A review of factors affecting cold start emissions

Transcription

1 TRANSPORT RESEARCH LABORATORY Environmental traffic management: A review of factors affecting cold start emissions Prepared for DITM Division, Department of the Environment, Transport and the Regions P G Boulter TRL REPORT 27

2 First Published 1997 ISSN Copyright Transport Research Laboratory All rights reserved. The information contained herein is the property of the Transport Research Laboratory. This report has been produced by the Transport Research Laboratory under a contract placed by the Department of the Environment, Transport and the Regions. Any views expressed in it are not necessarily those of the Department. Whilst every effort has been made to ensure that the matter presented in this report is relevant, accurate and up-to-date at the time of publication, the Transport Research Laboratory cannot accept any liability for any error or omission. Transport Research Foundation Group of Companies Transport Research Foundation (a company limited by guarantee) trading as Transport Research Laboratory. Registered in England, Number TRL Limited. Registered in England, Number Registered Offices: Old Wokingham Road, Crowthorne, Berkshire, RG45 6AU.

3 CONTENTS Page Executive Summary 1 1 Introduction 2 2 General background Effect of cold starting on engine performance Effect of cold starting on catalyst performance Integration of cold starts into emissions models and inventories Defining and measuring cold start emissions Effects of temperature Estimating the number of vehicles operating in cold start mode 4 3 Cold start emissions measurements Engine and catalyst temperature Engine Temperature Catalyst temperature Summary of engine and catalyst temperature effects Drive cycle Distance and time required for stabilisation of emissions Changes in drive cycle Summary of drive cycle effects Soak time Effect of soak time on cold start emissions Summary of soak time effects Vehicle category Cold start emissions from different categories of vehicle Summary of cold start emissions according to vehicle category Vehicle mileage Effect of vehicle mileage on cold start emissions Summary of vehicle mileage effects Engine size Influence of engine size on cold start emissions Summary of engine size effects Ambient temperature Influence of ambient temperature on cold start emissions Summary of ambient temperature effects 22 iii

4 4 Cold starts & emission models Cold start emission modelling Input parameters affecting cold start modelling Spatial and temporal distribution of cold start emissions Spatial distribution according to road type Temporal distribution Summary of modelling of cold start emissions 25 5 Cold starts, traffic management and future technology Traffic management Future technology 26 6 Summary and discussion Defining and measuring cold start emissions Factors affecting cold start emissions Modelling of cold starts Cold starts and traffic management Future work areas Future technology 3 7 Conclusions 3 8 References 32 Appendix A : Glossary 35 Appendix B 36 Appendix C 38 Abstract 39 Related publications 39 iv

5 EXECUTIVE SUMMARY This Review was produced as part of project UG93, Environmental Management - Main Study, for Mr E H E Wyatt of the DITM Division of the Department of the Environment, Transport and the Regions. The objectives were to investigate the factors affecting the definition and measurement of cold start emissions and to suggest how emission models might better account for them. For each factor that was thought or known to affect cold start emissions the importance of its effect, and the level of understanding, was considered. These factors were: l Engine and catalyst temperature l Drive cycle l Cooling-down period l Vehicle mileage l Engine size l Ambient temperature Cold start emission measurements for different vehicle categories were also considered, both independently and in relation to some of the above factors. For petrol cars, cold start emissions originate fundamentally from problems associated with the combustion process at temperatures below that of normal operation. Until this normal operational temperature is reached in the engine, emission levels will be elevated. For catalyst-equipped cars, even if the catalyst reaches its working temperature before the engine, emission levels will not be minimised until the engine ceases to run rich. For diesel-powered cars and diesel commercial vehicles cold starts emissions are of less importance. The rate at which an engine warms up may be dependent on the nature of the drive cycle, although for passenger cars engine temperature has been shown to evolve similarly with time over different drive cycles. For catalyst equipped cars some studies have shown that the point at which the catalyst reaches its working temperature could be linked to drive cycle characteristics. The associated effects of changes in the drive cycle on cold start emissions are still uncertain. The temperatures of the engine and catalyst of a particular vehicle at the outset of a trip are determined in part by the duration of the cooling-down period after the previous trip. The cooling periods, starting temperatures and therefore the associated cold start emissions of in-use vehicles will be widely distributed in practice. Vehicle mileage and engine size have not been shown to be great contributors to cold start emissions, although some studies have shown high mileage vehicles to have increased cold emissions, and engine size could affect cooling rates. Ambient temperature has a considerable impact on cold start emissions, particularly from petrol cars. Cold start emissions tend to increase as ambient temperature decreases, and emission levels take longer to stabilise. Although emissions of gaseous pollutants and particulate matter from diesel vehicles are affected by a reduction in ambient temperature, the increase in the cold start emissions penalty seems to be less marked. The relationship between cold start emissions and the cooling period (or soak time ) will also be complicated by ambient temperature. As ambient temperature decreases, the time taken for a vehicle's engine to cool down to a particular temperature decreases. Simple cold start emission factors have usually been determined by calculating the ratio of the total cold-start to hot-start emissions, or the absolute excess emission, from bag samples collected over the same drive cycle. These factors have usually been incorporated into emission models that operate on an area-wide scale. The use of such factors has been aided by the relative simplicity of data acquisition, the availability of such data from emission tests and the ease of application. However, these factors lack flexibility. For example, cold start emission factors based on aggregate emissions measured over a given cycle after a particular cooling period will not be representative of trips after different cooling periods. In future emission models, a more complex cold start sub-model might be employed to account for ambient temperature and cooling periods. To make such a model feasible, vehicle usage parameters derived from social surveys would be combined with more detailed vehicle-based measurements. Consideration might also be given to the representation of cold start emissions using engine and catalyst temperature as a basis. The magnitude of the rates of emission resulting from cold starting may be influenced by certain types of traffic management scheme, but for all types of scheme the effects are unknown at this stage. Where cold starting will be of particular significance is in the assessment of parking schemes. If parking times are being affected, then this will influence the length of cooling periods and hence the magnitude of cold start emissions. New technologies are currently being developed, and older technologies refined, in order to reduce the cold start emissions penalty of catalyst equipped cars. The newer technologies are still being tested and it is difficult to predict when they might become more widely available. 1

6 1 Introduction The term cold start can often be misleading. It is usually used to describe the section of a journey which is driven with the engine below its optimum operational temperature, and to express a contrast with the subsequent hot operation of the vehicle. It might easily be confused with low temperature starting, which actually refers to the ambient temperature and not the thermal condition of the engine. It has been suggested by Lenner (1994) that a more suitable term would simply be starting, but as most studies refer to cold starts, the description cold start will be retained in this Review. Fuel consumption and emissions of the main automotive pollutants: CO, HC, NO x and CO 2 are generally elevated during the cold start period, and these excess emissions are therefore termed cold start emissions. These elevated emission levels are causing concern because it has become apparent that a large proportion of trips by car in urban areas are being driven under cold start conditions. For example, Laurikko (1996) showed that the CO emission rate for a well-functioning catalyst-equipped vehicle is well below 1 g/km during hot operation. This can be compared with an emission rate of 1 g/km during the first 5m of cold-start driving. In order to gain a better understanding of the factors affecting cold start emissions, the Driver Information and Traffic Management Division of the Department of the Environment, Transport and the Regions has asked TRL to review the subject. The main objectives of the Review are to investigate the factors affecting the definition and measurement of cold start emissions and to suggest how emission models might better account for them. The Review is divided into a number of sections which deal with different aspects of the subject. In Section 2, a general background to the cold start problem is provided. Section 3 of the Review covers the body of information generated from studies on cold start emissions. This includes coverage of the fundamental importance of engine and catalyst temperature, the approaches that have been adopted in finding a definition of cold starts and cold start emissions, and measurements obtained according to drive cycle, vehicle characteristics and external conditions. For each factor that is thought or known to affect cold start emissions measurement, the relative importance is given consideration. Section 4 describes the ways in which cold starts, when incorporated into emissions inventories and models, are reliant upon the relationships between travel survey data and the parameters and measurements described in Section 3. Where a traffic management scheme is introduced, resultant changes in vehicle operation will affect certain aspects of cold starting, and these effects are outlined in Section 5. Since some of the terms used in this review will not be widely understood a glossary is given in Appendix A. 2 General background 2.1 Effect of cold starting on engine performance Irrespective of whether or not a petrol-fuelled car is fitted with a catalyst, cold start emissions originate fundamentally from problems associated with the combustion process in the cylinders of the engine. When a petrol engine is running efficiently around its optimum operational temperature and under low loads, there is a near-stoichiometric balance in the combustion chamber between the fuel and the oxygen in the air. Ideally, the fuel vapour would use up all the oxygen during complete combustion to form carbon dioxide and water vapour, and when approaching this condition the engine could be considered to be running hot. This normal operational temperature is well above common ambient temperatures, and so it takes some time for the engine to warm up. During the warm-up phase, a fraction of the fuel will condense on cold surfaces within the inlet manifold and cylinder, thus reducing the amount available for combustion. Until the engine has warmed sufficiently an excess of fuel must be delivered to compensate for the loss, to avoid misfire and maintain driveability. This causes the engine to run rich, that is to say the air:fuel ratio in the cylinder decreases, and emissions of carbon monoxide (CO) and unburned hydrocarbons (HC) rise accordingly. Cole (1992) revealed that sufficient quantities of the light fuel fractions must reach the cylinder, as these vaporise more readily during the warm-up phase. It is the large quantity of heavier fractions reaching the cylinder that do not vaporise and are left to burn partially, resulting in poor overall fuel economy and high levels of emissions. Nitric oxide (NO) formation is favoured by a high combustion temperature and pressure, and high oxygen content in the combustion mixture. Nitrogen dioxide (NO 2 ) is not formed during engine combustion, but when the hot exhaust gases pass through the manifold and exhaust system in the presence of oxygen. For conventional petrol cars measurements of NO x (the sum of NO and NO 2 ) emissions over standard drive cycles have tended to yield similar values for both cold start and hot operation. The reasons for this are not entirely clear. During the cold start period greater engine loads are required to overcome friction in the engine and gear system. Higher engine loads lead to the higher peak temperatures and pressures that enhance NO production. This seems to be compensated by conditions in the combustion chamber, such as low temperature, which are unfavourable for NO formation. These opposing factors tend to result in smaller changes for NO x emission than CO/HC emission during cold start. Variation in measured values is invariably a problem when analysing the results from emission tests, and this problem is exaggerated during the cold start period. Laurikko (1996) considered the emissions output from around 4 catalyst cars over a given drive cycle at o C. It was found that the worst emitter produced around 7 times more CO than the best and HC varied by a factor of 5. There was found to be no direct link between emissions of 2

7 CO and HC early in the cycle. The correlation between cold start emissions of CO and HC was characterised with R 2 value for the linear regression being less than.5. NO x emissions varied by a factor of 3, but this was due to a few vehicles being particulary high and low emitters of this pollutant. In petrol engines the air/fuel mixture is ignited by a spark plug (spark ignition or SI), whereas in diesel vehicles ignition results from an increase in pressure and associated increase in temperature (compression ignition or CI) after fuel has been injected directly into the cylinder. Cold start emissions from diesel engines are generally much smaller than those from petrol engines, although the effect is still significant. 2.2 Effect of cold starting on catalyst performance For cars equipped with three-way catalysts, running rich to counteract the effects of low engine temperature is still the primary reason for elevated emissions of CO and HC during the cold start period but, in addition, the catalyst is not converting exhaust pollutants as efficiently as during hot operation. The reasons for this are (i) catalyst inefficiency below a particular temperature (known as the light-off temperature) and (ii) deviation from the optimum air:fuel ratio in the exhaust gas resulting from the enrichment of the fuel in the engine. The catalytic convertor is heated by the exhaust gases from the engine, and needs to reach a minimum light-off temperature of around 3 o C in order to function adequately; below this temperature removal of pollutants is minimal. The exhaust gas composition is strongly dependent on the combustion mixture in the engine, and until the engine is sufficiently warm that it no longer needs an excess of fuel in the mixture, the air in the exhaust will lack sufficient oxidising agents to convert the hydrocarbons and carbon monoxide to water and carbon dioxide. Therefore, until the catalyst is functioning efficiently, catalyst equipped cars behave similarly to non-catalyst cars in terms of emissions of CO and HC, although levels are generally lower from the more modern cars. However, for catalyst cars, cold start emissions of CO and HC are much higher in relation to hot emissions, and therefore make up nearly all the emissions during a test. The situation regarding NO x emissions from three-way catalyst cars is again complex. Although NO x emissions from the exhaust are usually lower after the engine and catalyst have warmed up, the decrease is not as pronounced as it is for CO and HC emissions from the same vehicles. This is because NO production in the engine is often similar during cold start and hot operation, whereas CO and HC emissions are already lower during hot operation. 2.3 Integration of cold starts into emissions models and inventories For examination of the changes in vehicle emissions following the introduction of a road scheme, modelling tools need to be as accurate as possible. Cold starts contribute significantly to emissions over short-medium urban trips and must therefore be given due consideration in the modelling process until ameliorative measures are in place. By compiling emissions inventories and constructing emissions models, the impact of different assumptions or developments relating to cold starting can be investigated. However, in a list of developments taking place regarding emissions inventories compiled by Hutchinson et al (1995), methodologies were described as crude and cold starts were identified as an area that was poorly understood. Therefore, if cold starts are to be more accurately incorporated into urban emission models, it is necessary to identify the parameters which affect their measurement, magnitude and duration. 2.4 Defining and measuring cold start emissions In order to quantify cold start emissions for a particular vehicle, workers have usually viewed the cold start itself as a definite event bounded by time or distance travelled, and have thus attempted to define the cold start period. In fact, no agreement has been reached on a standard definition in terms of these parameters. A variety of definitions have been used in the literature, and they have largely been based on simple assumptions about when the cold period actually ends. Measured emissions have varied substantially, even for vehicles of the same type and where the same cold start definition has been employed. There are two general approaches that have been adopted for expressing cold start emissions for a given pollutant i: (a) an absolute cold excess emission (E cold(abs) ) i (b) a ratio of cold to hot emissions (E cold /E hot ) i (or relative cold start emissions) These emission factors are referred to throughout this Review, and where the text refers to cold start emissions, the terms absolute, absolute excess, and relative are used in this context. They have usually been determined through the collection of bag samples containing the exhaust gases from vehicles driven from both hot and cold start over the whole of a drive cycle or cycle section. Hot start emissions have been subtracted from cold start emissions to give the excess cold start emission. Alternatively, workers have determined the ratio of the total cold start to hot start emissions over the same distance. That these simple factors have been used extensively in emission models is probably due to the relative simplicity of such an approach in terms of data acquisition and application of the method, and the availability of such data from emission tests. Where continuous emission measurements have been taken over a given drive cycle, the end of the cold start period has been related to the point of emission stabilisation with cold-start and hot-start emissions being summated up to this point. However, the point of stabilisation is defined arbitrarily, if it is defined formally at all, and there is an inherent uncertainty associated with 3

8 its determiniation. Another problem concerning emission factors derived in this way is their lack of flexibility. The range of temperatures of in-service vehicles at trip origin is approximately bounded by the ambient temperature at a given location and the normal operational temperature(s) of the vehicles. Vehicle starting temperature affects cold start emissions, but simple emission factors cannot account for this without the need for extensive testing. 2.5 Effects of temperature A few studies have also shown ambient temperature to be an important factor governing the magnitude of cold start emissions. On a hot day a car may have to be driven for several kilometres in an urban area before the engine is fully warmed and operating efficiently. In similar conditions a diesel car may warm up more quickly, but on a cold day both diesel and petrol cars will take longer to warm up. The rate of change of engine and/or catalyst temperature may be dependent on the driving pattern. However, the strength of this relationship, and the resultant impact on cold start emissions is unclear at present. A variety of vehicles and test cycles have been used to define cold start emissions, and each cycle has a distinct speed and acceleration profile which may affect measured values. If the effect is a major one, then this could complicate analyses considerably. 2.6 Estimating the number of vehicles operating in cold start mode In order to determine the percentage of vehicles operating in the cold start mode, vehicle usage parameters such as trip purpose must also be accounted for. Trip purpose can give some indication of the cooling down period or soak time and probable engine temperature before restart, although when an engine is switched off there is no obvious point at which its temperature may be described as specifically cold except when it has cooled to the ambient temperature. Nor can it be assumed that the engine had reached its full operational temperature on the previous trip. On longer trips this temperature should be attained, except during the initial stages. However, most car journeys in the UK are relatively short: about 4% of journeys are less than 5km (QUARG, 1993), and almost 6% of them less than 5 miles (8km) (Department of Transport, 1995). On these journeys the engine will be operating below its optimum temperature for a large fraction of the time. 3 Cold start emissions measurements In this section, information from the review of literature relating to different aspects of cold start emissions is presented. Short summaries, suggesting the relative importance of each factor considered, are given at the end of each sub-section. 3.1 Engine and catalyst temperature Engine Temperature It has already been noted why cold start emissions are a direct result of low engine temperature, and workers have used engine oil temperature or, more often, coolant (water) temperature to define the duration of the cold start period. From the beginning of the cold-start period engine temperature gradually increases until it reaches that of normal operation, after which point it remains relatively stable. Consequently, pollutant emissions gradually decrease until they reach the fairly steady levels observed during hot operation. It has generally been assumed that for conventional (non catalyst) petrol cars the end of the engine warm-up period coincides with the point at which emission levels become steady, and therefore defines the cold start period. Eggleston et al (1993) defined cold engines as those with a water temperature below 7 o C. In an evaluation of trip start activity, Kishan and Defries (1993) analysed instrumented vehicle behaviour during cold,warm and hot engine starts. The trip start period was deemed to end when the coolant reached 6 o C (or when the engine was turned off if that occurred sooner). During cold-start tests carried out at TRL, vehicles were considered to be fully warmed once the coolant temperature had reached 85 o C. Potential problems with using engine temperature as a measure of the cold start period have been encountered at INRETS. Joumard et al (1995) measured the cold-start emissions of several pollutants from ten petrol noncatalyst, ten petrol catalyst and five diesel passenger cars over the following drive cycles: A standard European NEDC cycle. This is comprised of the ECE 15 cycle followed by the Extra-Urban Drive Cycle (EUDC). The United States Federal Test Procedure (FTP75 cycle) Three composite drive cycles, each composed of 15 repetitions of a short cycle. The short cycles had average speeds of 8, 19 and 48 km/h ( short slow urban, short free-flow urban and short road respectively), and the duration of each of these cycles was approximately 2 to 3.5 minutes. The composite cycles therefore lasted between around 3 and 5 minutes. The authors measured temperature changes for coolant, oil and exhaust gas as functions of time and distance travelled. Some results of this study are shown in Figures 1 to 3. Overall, the results demonstrated that an equilibrium temperature was achieved more quickly for coolant than for engine oil. The engine warm-up profiles for catalyst and non-catalyst petrol cars over various drive cycles were similar when considered as a function of travel time, but not when compared as a function of distance travelled. Most of the increase in engine temperature for each of the three classes of vehicle was generally seen to have occurred by around 8 minutes, although the study did not account for differences in ambient temperature between the different tests. The important relationship between engine temperature 4

9 1 8 Water temperature o C Short slow urban Short free-slow urban Short road NEDC FTP Time (s) Figure 1 Changes in coolant temperature as a function of time for petrol, non-catalyst cars (Source: Joumard et al, 1995) 1 8 Oil temperature o C Short slow urban Short free-slow urban Short road NEDC FTP Distance (km) Figure 2 Engine oil temperature for petrol, non-catalyst vehicles as a function of distance (Source: Joumard et al, 1995) 5

10 25 2 Exhaust gas temperature o C 15 1 Short slow urban Short free-slow urban Short road NEDC FTP Distance (km) Figure 3 Exhaust gas temperature as a function of distance for petrol, catalyst cars (Source: Joumard et al, 1995) and pollutant emissions was considered. Some results for CO emissions are shown in Figure 4. Plots of emission changes against coolant temperature demonstrated that the drive cycle, studied pollutant and vehicle type were significant factors. Such plots enabled the authors to determine minimum temperatures for which emissions could be considered steady. These are listed in Table 1. It was noted that water temperatures seemed to be a better indicator of engine thermal condition in terms of emissions since, for each technology, the dispersion of the limit temperature values was lower for water than for oil. The authors adopted an original approach to defining the cold-start period. The end of the cold period was determined by working back in time from the end of the test. In the case of an approximately steady hot emission value, the sample standard deviation on the measurements decreased as the number of measured points, going back in time, increased. When reaching the cold period, a clear emission variation which was not a random value was recorded. From this point back, the sample standard deviation continually increased. This technique yielded average hot and cold emission values, and the cold distance travelled- ie from vehicle start-up to emission stabilisation. Then an average cold unit emission was obtained over the same distance, followed by a hot unit emission beyond the cold period. The value of (E cold /E hot ) i Table 1 Water, oil and exhaust gas limit temperatures ( o C) beyond which vehicle emissions are relatively steady. (Source: Joumard et al, 1995) Vehicle CO HC NO x CO 2 FC Range Water Petrol Catalyst Diesel Oil Petrol Catalyst Diesel Exhaust gas Catalyst

11 5 Urbain Lent Court - Short slow urban Urbain Fluide Court - Short free-slow urban 4 Route Court - Short road CO relative emission Water temperature o C Figure 4 Evolution of relative CO emissions vs. cooling water temperature for vehicles equipped with a catalyst (Source: Joumard et al, 1995) (see section 2.4) could then be calculated, along with a value for the absolute cold excess emission. It was concluded that to define the start of the hot emission period in terms of thermal equilibrium of oil or water temperatures was a flawed procedure, and it was difficult to determine whether an engine was specifically hot or cold from an emissions standpoint as emissions reached their equilibrium levels before engine temperature stabilisation. This study illustrates some of the problems involved in attempting to define the cold start period as a definite event Catalyst temperature Engines and catalysts warm up at different rates, and this complicates the analysis of the cold start period. In catalystequipped vehicles, part of the delay between engine start-up and the point at which exhaust emissions stabilise is due to the warming-up of the catalyst. However, it has been suggested that the rate at which the catalyst warms up is not as important as the rate at which the engine warms up (Bardon et al, 1992). This is because the catalyst has often reached its operational temperature (i.e. maximum efficiency) before the exhaust gas composition stabilises. Experiments with pre-heated catalysts have resulted in quite small improvements in emissions of CO and HC due to the rich mixtures involved during cold start. Laurikko (1991) also noted that the key factor for cold emissions was excessive enrichment of the air/fuel mixture and not catalyst temperature. Catalyst temperature (exhaust gas temperature) was measured by Joumard et al (1995) as a function of cycle distance for various drive cycles, and it was found that catalysts reached an equilibrium temperature of around 23 o C (Table 1, Figure 3). This is in general agreement with the statement by Larson (1989) that the oxygen sensor in the catalyst must be warmer than 25 o C in order to operate correctly. More recently Laurikko (1996) suggested that the threshold temperature for catalyst light-off was between 25-3 o C. However, it can be difficult to equate the stabilisation of catalyst temperature to stabilisation of emissions. In a test on one vehicle, Laurriko (1991) found that the temperature of the catalyst rose to around 4 o C after one minute, but no conversion was possible before the air:fuel ratio reached the stoichiometric value. Lenner (1994) found that the time period required for the catalyst to attain its light-off temperature was typically less than 2 minutes at around 15 o C, and this period was fairly independent of ambient temperature. Again, the time required for complete warm-up of the engine, with the ensuing stoichiometric operating mode that renders the catalyst fully operative, was considered to be longer and dependent on ambient temperature. 7

12 3.1.3 Summary of engine and catalyst temperature effects Engine temperature and catalyst temperature are fundamental factors determining cold start emission rates. Most other factors affecting cold start emissions are secondary, in that their importance relates to how they influence engine and/or catalyst temperature. It has been found that, generally, engine temperature evolves similarly over different drive cycles when measured as a function of time, but that the drive cycle influences the warm-up rate measured as a function of distance. At around 15-2 o C most of the increase in engine temperature for each vehicle class has generally been seen to have occurred by around 8 minutes, or over a distance of 3-6km, although these factors will change with ambient temperature (i.e. longer periods would be needed with low ambient temperatures). For petrol catalyst cars, the different rates at which the engine and catalyst warm up complicate the analyses. However, it appears that catalysts normally reach their full operational temperature well before the stabilisation of engine temperature and exhaust emissions. Consequently, cold start emission levels will be affected more by the rate of engine temperature stabilisation than by the catalyst reaching its operational temperature. 3.2 Drive cycle For passenger cars driven over cold start cycles, continuous measurement of emissions has yielded approximate times and distances covered during the cold start period. Such times and distances have related specifically to the point at which, within reasonable limits, emissions were seen to stabilise. However, the more common approach to quantifying cold start emissions has been to compare aggregate hot and cold emissions over the same standard drive cycle. This has therefore contributed to discrepancies in measured values from different studies because the cycle phases employed have not always been representative of the cold start period, as determined more accurately via continuous measurement. The drive cycle employed has often been the Federal Test Procedure (FTP) used for type approval in the United States. Figure 5 shows the speed/time profile of the FTP type approval test. The first phase is taken to represent a cold start and lasts for the first 55 seconds of the cycle. The pattern of the third phase ( seconds) is identical to the initial phase, although during this period the engine is normally running hot. The difference in total emissions from phase one and phase three has often been used to describe cold start emissions Distance and time required for stabilisation of emissions It was noted by Laurikko (1996) that typical trip lengths in urban driving tended to be around 3-5km (the author is based in Finland; the corresponding UK figure may be different), and that this was similar to the distance required for engine warm-up (see also section 3.1.1). From the previous discussion regarding the apparently overriding influence on emissions of engine temperature (section 3.1.2), it might be expected that emissions from both noncatalyst and catalyst cars would stabilise after a similar distance. This concept seems to be supported by some results, but not by others. In the study by Joumard et al (1995), the definition of the equilibrium emission levels over the three short drive cycles employed (section 2.1) allowed the authors to determine the distances and times travelled in the cold start mode. The distance covered during the cold start period was shown to generally increase with increasing average cycle speed, whereas the time period was shown to generally decrease as the average speed of the drive cycle increased (Table 2). Table 2 Distance travelled in cold start mode (top in km) and cold duration (bottom, in min.) according to pollutant type, vehicle type and average cycle speed (Source: Joumard et al, 1995) Speed (km/h) Distance covered during cold start period (km) CO HC NO x CO 2 Petrol Catalyst Diesel Speed (km/h) Time spent during cold start period (minutes) CO HC NO x CO 2 Petrol Catalyst Diesel Joumard et al (1995) found that in terms of emission stabilisation of the three main pollutants, CO, HC and NO x, the average distance travelled for petrol non-catalyst, catalyst and diesel cars averaged 4.5km, 6km and 6.5km respectively. The maximum values for the distance covered and time elapsed during the cold start period were determined to be 13.5km (Catalyst, CO 2 ) and 35 minutes (Diesel, NO x ) respectively. However, compared to the cold start periods found in other studies, the average values were quite high and the maximum values seemed extremely high at the lower and intermediate speeds, considering that engine temperatures had generally stabilised by around 8 minutes (see Figure 1). There is a question to be raised at this point, and that is 8

13 Phase 1 (1 55s) Phase 2 ( s) Phase 3 ( s) minute pause Speed (km/h) Times (s) Figure 5 The FTP test cycle whether these long cold start periods were a consequence of the technique used to define the cold start period (i.e. the standard deviation method). The cold start period may have appeared so protracted because of the precise nature of its definition. It is difficult to determine when emissions stabilise because the stability is approached gradually rather than abruptly. At this point, small improvements in emission stability will therefore take a disproportionately long period of time to be achieved. By referring to the data given in Figure 1, which shows changes in engine temperature as a function of time over different drive cycles, it seems that different drive cycles resulted in similar warm-up profiles when measured as a function of time. In terms of elevated emission levels, the bulk of the cold start effect may well have occurred considerably earlier than some of the times quoted for the cold start period. Research by Volvo on a single catalyst vehicle (reported by Ajne et al, 1992) showed that, at 2 o C, emissions of CO, HC and NO x stabilised after less than 2km, and other studies have concluded that a large proportion of CO and HC emissions on a trip are emitted during the first kilometre of travel. According to Joumard et al (1995), the distance travelled to reach the equilibrium point of exhaust gas temperature varied with drive cycle, but was generally between around 3 and 6km. From the Volvo study data, emissions of CO at 2 o C could be seen to stabilise after around 3 minutes, and HC and NO x after about 1.5 minutes (Ajne et al, 1992). Cold start emission measurements taken by TRL have shown CO and HC levels from catalyst cars to stabilise after around 3-4 minutes. In cold start emissions tests over the FTP drive cycle, Hassel et al (1993) found that for catalyst passenger cars around 8% of the of the total emissions of CO and HC occuring over a drive of 17.8km in length were emitted in the first 2km. For non-catalyst petrol cars 6% of total emissions occurred over the first 2km. However, emissions were not seen to stabilise fully until a few kilometres further. Because of the relatively small increase in emissions of HC and CO during cold starting for diesel cars, cumulative emissions of CO and HC were seen to vary in a similar way to summated NO x, that is to say linearly with distance. Lenner (1994) found that the time period required for the catalyst to light-off was typically less than 2 minutes at around 15 o C, and also fairly independent of ambient temperature. However, the time required for complete engine warm-up, with the ensuing stoichiometric operating mode that renders the catalyst fully operative, was considered to be dependent on ambient temperature and to vary from 2-3 minutes at 2 o C ambient, up to 5-6 minute - 2 o C ambient. In experiments using an engine test bed, Laurikko (1989) measured continuous emissions of CO, HC and NO x from petrol non-catalyst, petrol catalyst and diesel engines under constant and variable loads. For the catalystequipped vehicles, nearly all of the CO and HC was emitted during the first 3-5 minutes, and the effect of ambient temperature on the duration of the cold start period appeared to be lower with variable loading than with constant loading. At 2 o C, the end of the cold start period for all three types of engine could be seen to be approximated by a fall in emission rates of CO and HC around the 3 minute point. Discrepancies in the measured values for catalyst lightoff time were partially explained in terms of drive cycle characteristics by Laurikko (1996). For around forty current-technology passenger cars the author determined catalyst light-off times at o C by following the CO traces over the FTP drive cycle and determining the point at which the concentration fell below.1%. This was assumed to represent the start of conversion. All the measured light-off times fell between 5 and 29 seconds, with an average of around 18 ± 2 seconds. The results seemed to form two groups, and it was argued that these groups could be associated with the high-speed, 9

14 high-load modes of the FTP. Some catalysts achieved light-off in less than 1 seconds (or less than 1km), corresponding to the first acceleration mode ( hill ) of the FTP. Most of the vehicles, however, needed the high-speed mode during the second hill of the FTP in order to warm up the engine and the convertor, and the associated lightoff times varied between 2 and 6 seconds (1.5-3km). The author also found that if a slower-speed test cycle was employed, such as the European test cycle ECE15, warmup was slower and light-off times were longer Changes in drive cycle The nature of the drive cycle might affect the warm-up rate of the engine and/or catalyst, and consequently measured cold start values. Hard accelerations during the cold start period have the potential to substantially affect emissions. Continuous emissions and vehicle operation parameters were measured over specified drive cycles by TRL. Acceleration events in the early part of the drive cycle were critical to the formation of CO/HC in both gasoline and diesel vehicles. Figure 6 shows the characteristic rapid increases in emissions associated with early-cycle accelerations (i.e. at around 15s). NO x levels were seen to be strongly related to engine loading. However, at higher engine loads, optimum engine and catalyst temperatures will be reached more quickly. Thus, from the standpoint of cold start emissions, one problem is which type of drive cycle has the greater overall impact: (A) high engine loading and high emissions over a short time period, or (B) low loading and low emissions, but over a longer period. It has been found in some studies (e.g. Joumard et al, 1989) that hot emissions during constant speed (i.e. lower engine loading) tests are lower than over transient (i.e. higher engine loading) cycles and, although it would be tempting to assume that this effect would also apply to cold starts, there is the complicating factor of engine warm-up rate. At present, although there is little evidence indicating which of the two scenarios has the least impact on emissions. The work of Joumard et al (1995) showed changes in engine warm-up rate as a function of time did not differ greatly between the cycles employed. For diesel and non-catalyst petrol cars at least, this suggests the possibility of reducing the priority of the drive cycle variable during modelling. For catalyst cars, the warm-up rate of the catalyst might also have to be considered independently to that of the engine, as some studies (e.g. Laurikko, 1996) have shown that the light-off point could be linked to drive cycle characteristics. Using 25 catalyst-equipped cars with high odometer readings, Kutscher (1991) compared emissions from the first and second phases of the FTP cycle. At the start of each cycle, vehicle engines were cold. The two phases were taken to represent different driving modes, with the first phase consisting of more speed variation than the second phase. For example, the average speed for the first phase was 41.2 km/h with a maximum speed of 91.2 km/h, whereas for the second phase the average speed was 25.8 km/h and the maximum 55.2 km/h. It was found that the phase representing smoother driving behaviour, phase 2, appeared to give lower cold start emissions, on a g/km basis, than phase 1. The results are summarised in Figure 7. The differences in emissions between the two tests were found to be comparatively small (around 1%). However, factors other than the differences in driving pattern might also have been influential. For example, the second FTP phase is longer than the first (6.2 km compared to 5.78km). Both phases could well have included end sections driven under hot operating conditions. For the second phase, this section may have been longer and therefore cold emissions could have been reduced compared to the first phase. Grant (1992) used the two identical phases of the FTP drive cycle to compare cold and hot emissions. Phase one at the start and phase three at the end were used to measure the differences in cold and hot emission levels for 7 pairs of vehicles. Each pair consisted of two vehicles of a given model, one equipped with a petrol engine and a catalytic converter, the other with a diesel engine. The selection criteria for the study vehicles were: (a) both vehicles would have been produced by the same manufacturer (b) petrol/diesel engines of similar maximum power (c) current models in widespread circulation in Europe (d) use of modern technology. Figure 8 (upper diagram) shows the average values of the 7 ratios (i.e. emission levels in the first (cold) phase divided by emission levels in the third (hot) phase (base 1) obtained over the full 5.78km of each phase). The lower diagram shows the results obtained when following only the first 4.24 km of the same phase. Regulated emissions ( Reg.Em ) refers to total regulated emissions (CO, HC, NO x and particulates). For catalyst cars in particular, the results suggest that most of the pollutants that had been emitted after 5.78km had already been emitted after 4.24km, and possibly even sooner. Similar conclusions could be drawn from work undertaken at Volvo (reported by Ajne et al, 1992) in which the cumulative emissions of CO, HC and NO x from a single catalyst-equipped vehicle were also monitored over the first cycle of the FTP. The results showed that catalyst light-off caused emissions to reach their hot levels well before the end of the cycle used to measure cold start emissions at a range of ambient temperatures. If this effect were to be exhibited by most catalyst vehicles it would mean that the length of the cold start period would be overestimated and the relative cold start emission factor (E cold /E hot ) would be underestimated. This suggests that a shorter drive cycle than the first phase of the FTP cycle would often describe the cold start period adequately. However, as long as the drive cycle contains the entire cold start period, the absolute cold start emission (E cold(abs) ) does not change with cycle length, and is probably therefore a more reliable indicator. Other problems associated with the using the FTP cycle to determine cold start emission estimates have been identified. Between the second and third phases of the FTP there is a 6 second period during which the engine is switched off and allowed to cool a little. Lenner (1994) described research conducted on 25 closed-loop (i.e. threeway) catalyst vehicles in which the 6 second pause was 1

15 1.2 1 Petrol catalyst.8.6 CO(g/s).4.2 Diesel Cycle times (s) (a) Example carbon monoxide at -1 o C (gasoline & diesel) Petrol catalyst.2 HC(g/s) Diesel Cycle times (s) (b) Example hydrocarbons at -1 o C (gasoline & diesel).4.35 Petrol catalyst.3.25 NO(g/s) Diesel Cycle time (s) (c) Example oxides of nitrogen at -1 o C (gasoline & diesel) Figure 6 Continuous CO, HC and NO x emissions from diesel and catalyst-equipped gasoline vehicles at -1 o C 11

16 g/km FTP-1, cold start FTP-2, cold start driving than with the FTP. It was also noted that the FTP limited accelerations to 3mph/s (1.34 m/s 2 ). The study showed the presence of accelerations significantly exceeding this value. Over 9% of the vehicles tested had at least 1% of their accelerations greater than 3.3mph/s (1.47 m/s 2 ) and observed accelerations reached 6mph/s (2.68m/s 2 ). 1 CO HC NO x Figure 7 Cold start emissions from the first phase of the FTP compared with cold start emissions from the second phase of the FTP (Source: Kutscher, Diesel cars CO 2 CO HC NO x PART. REG.EM Petrol catalyst cars Summary of drive cycle effects It is often difficult to determine when emissions stabilise because the stability is approached gradually rather than abruptly. However, the results from several studies where emissions have been measured continuously from start-up have shown that for catalyst cars at least, the bulk of cold start period typically occurs before around 3 minutes or 2km. For the quantification of cold start emissions from individual vehicles, bag samples have usually been taken over a standard drive cycle (commonly the FTP) whereby all of the cold start emissions have been assumed to be consigned to an initial phase of the cycle. The FTP drive cycle, at 5.78km, will be useful for defining cold start emissions from some cars, though it may result in the underestimation of the (E cold /E hot ) ratio for vehicles whose emissions stabilise rapidly. However, as long as the drive cycle contains the entire cold start period, the absolute cold start emission (E cold(abs) ) does not change with cycle length, and is probably therefore a more reliable indicator. The effects of subtle variations in the speed profiles of urban drive cycles on actual cold-start emission levels are uncertain at present, although for catalyst-equipped cars the large amount of variation in the light-off time from vehicle to vehicle has been partially explained in terms of drive cycle characteristics omitted. The data was compared with that obtained from a different set of 24 cars driven over the standard cycle. It could be seen that the standard cycle yielded average relative cold emissions of CO, HC and NO x that were higher by 75%, 31% and 15% respectively. In a summary of the Fourth Annual CRC-APRAC On- Road Vehicle Emissions Workshop, Cadle et al (1994) referred to on-road driving surveys (Kishan and Defries, 1993) in four U.S. cities during which more than 23 hours of second-by-second driving data was collected. The data showed that speed distribution and cold-start driving pattern were very different from those in the FTP cycle. There were many more short distance trips than had been assumed, a broader distribution of soak times and CO, HC and NO x emissions were higher in hot, stabilised in-use CO 2 CO HC NO x PART. REG.EM. Figure 8 Effect of engine temperature on FTP emissions, (top) comparison of results from cold phase 1 and hot phase 3 (base 1) over full 5.78km, (bottom) as above but with a shortened phase length (4.24km) (Source: Grant, 1992) 3.3 Soak time Effect of soak time on cold start emissions Soak time refers to the period during which a vehicle remains inoperational and is left to cool down at the prevailing ambient temperature. The temperature of the engine and/or catalyst of a particular vehicle at the outset of trip, and consequently the duration of the cold start period and magnitude of cold start emissions, is determined in part by the duration of the cold soak period. When a vehicle remains inoperative the engine cools down and, to some degree, cold start effects become important once more on restart. As the soak times for all the vehicles on the road will be widely distributed, then it follows that the distribution of associated cold start emissions will also be large. Cold start emission factors based on aggregate emissions over a given cycle after a particular cooling period will not be representative of trips after different soak times. The soak time/cold start emissions relationship will be further complicated by the ambient temperature, and possibly engine size. As ambient temperature decreases, the time taken for a vehicle s engine to cool down decreases, and the time elapsing between successive cold starts becomes shorter. Simple physics dictates that lighter engines should cool down more rapidly than heavier ones 12

17 made of the same material, although there is a lack of information on the effect of engine size on cooling rate. Recently, in order to represent the cool-down period, workers have used values that approximate to the time span in question. In the US the Environmental Protection Agency has defined a cold start as any start that occurs 4 hours or more following the end of the preceding trip for non-catalyst vehicles and 1 hour or more for catalystequipped vehicles (EPA, 1993). For the Danish Road Data Laboratory, Bendtsen and Thorsen (1994) estimated the proportion of total emissions due to short journeys (up to 6km), especially those with cold engines. The cold soak period was defined as one where the car had been stationary with the engine switched off for more than two hours (at 7 o C), although it was also reported that a catalytic converter will cool down in about 3 minutes under these conditions. Few studies have assessed the soak time effect. One exception was a study for the US Environmental Protection Agency in which Srubar et al (1978) measured emissions from four petrol cars equipped with oxidation catalysts and one non-catalyst car over the first phase of the FTP cycle. The test vehicles used were all 1976-technology petrol-fuelled passenger cars, all except one of which were equipped with oxidation catalysts (see Glossary, Appendix A). Emission tests were conducted following soak periods between 1 minutes and 36 hours in length. The five vehicles tested had different emission levels, and in order to be able to discuss emission trends the authors used the emissions after a 16-hour soak as a baseline (i.e. 1%). Normalised values were therefore presented as the percentage of emissions following a 16-hour soak. It was found that HC and NO x emission rates and fuel consumption increased almost linearly with log 1 of the soak period length. The only non-catalyst car tested, the Civic, had fairly stable HC emission levels at soak periods of less than 2 hours, but did follow the prevalent trend after longer soak times. CO emission rates were found to be stable following soak periods of 1 minutes to 1 hour in length, but the level ranged from 1 to 7% of the 16-hour soak value. Following soak periods of 2 hours and longer, CO emissions increased approximately linearly with log 1 of the soak length. Soak period length was also seen to affect emission rates during the hot second phase of the FTP cycle, but the magnitude of the effect was small compared to that observed during the first phase. The normalised FTP-phase 1 emission rates are shown in Figure 9. It should be noted that the study by Srubar et al (1978) was conducted some time ago and, therefore, may not be representative of the current UK or European fleet Summary of soak time effects Soak time is a major factor affecting cold start emissions. Some early results showed that pollutant emissions and fuel consumption vary linearly with log 1 of the soak period length. However, more research would be required to either formulate a relationship between soak time and starting emissions which could be applied to existing emission factors, or to develop a cold start function that implicitly includes soak time. This might include an investigation of the effect on emissions of restart for vehicles of various engine size, and at various temperatures and times on the engine cooling curve. There is some indication that such work may already be underway in the United States, as Cadle et al (1994) reported that amongst changes in the Californian motor vehicle emissions inventory, a continuous start function based on soak time would be considered in the future, rather than only hot or cold starts. 3.4 Vehicle category Cold start emissions from different categories of vehicle Petrol and diesel-fuelled engines have inherently different efficiencies of combustion and consequently different emission levels. Most cold start testing has been carried out over standard drive cycles for petrol cars (with and without catalysts) and for diesel cars, and generally with the pollutants CO, HC and NO x. Joumard et al (1995) recorded average emission values during the cold period, calculated over the three short cycles monitored and weighted by the average cold distance travelled (from Table 2). These were compared with the emissions that would have been recorded for the same period under hot conditions to obtain absolute and relative cold start values. The results, given in Figure 1, showed relative cold start emissions of CO and HC from petrol catalyst vehicles as being the most significant - factors of around 1 and 16 over cold start distances of 5.9 and 3.6 km respectively. Figure 11 gives absolute excess cold start emissions. For conventional petrol cars, it was observed that NO x emissions decreased by 1.55g (equivalent to a 13% reduction on the hot emission value) over a cold start distance of around 3km. For diesel cars the reduction was.8g (4%) over a distance of 6km. The cold start emission values reported elsewhere for different types of vehicles have generally been obtained using a less sophisticated methodology than that employed by INRETS. The variation in the cold start distance according to pollutant and vehicle technology has not been taken into account elsewhere. The Transport Research Laboratory also undertook a programme of emissions measurements on a total of 18 cars under real driving conditions. Petrol catalyst,noncatalyst and diesel vehicles were driven under urban hot, urban cold, rural and motorway conditions, and emissions of CO,HC and NO x were measured and quoted on a g/km basis. Emissions from conventional petrol cars were dominant. For CO and HC, the urban cold cycle resulted in the highest emission levels for each type of vehicle. QUARG (1993) referred to measurements of CO, HC and NO x emissions made during an urban drive around Stevenage by Warren Spring Laboratory with both cold and hot-engined vehicles. The relative cold start emissions penalties, E cold /E hot, for non-catalyst petrol and diesel cars were similar for CO (1.6), and differed by a factor of 2 for HC. However, for petrol catalyst cars cold start emissions 13

18 16 14 Normalised rates Normalised rates Normalised rates Impala LTD Fury Vega Civic Soak length (hours) (a) Transient HC emission rates Impala LTD Fury Vega Civic Soak length (hours) (b) Transient NOx emission rates Impala LTD Fury Vega Civic Soak length (hours) (c) Transient CO emission rates Normalised rates Impala LTD Fury Vega Civic Soak length (hours) (d) Transient fuel consumption Figure 9 Normalised FTP-phase 1 emission rates of HC, NOx, CO and fuel consumption as a function of log 1 of the soak period length (Source: Srubar et al, 1978) 14

19 Absolute excess cold start emission (g) Cold emission/hot emission 2 15 ;;; Petrol, non-catalyst Petrol, cat. Diesel ; CO HC NO x CO 2 Figure 1 Cold/hot unit emission for the whole cold period according to pollutant and vehicle types (Source: Joumard et al, 1955) ;5.96; ;2 Petrol, non-catalyst Petrol, cat. Diesel -5 CO HC NO x CO 2 Figure 11 Absolute cold excess emission according to pollutant and vehicle types (Source: Joumard et al, 1995) of CO and HC were about an order of magnitude higher than the corresponding hot emissions from this type of vehicle. For NO x, the E cold /E hot ratio was similar (around 1.2) for all three types of vehicle. A noteworthy point here is that the constitution of the vehicle fleet is changing with time, with catalyst cars becoming more widespread. Holman et al (1993) recognised that the large cold start emissions penalty from catalyst cars will become increasingly important as cold starts will contribute a greater proportion of the total emissions. The authors estimated that by the year 2 around 5% of CO and volatile organic compounds (VOC) emissions will result from cold starts. The results from these studies and several others are shown in Appendix B. Where specified, the entries refer to tests on a number of vehicles at common legislative (or similar) test temperatures, over the stated drive cycle. The data includes results from several Swedish studies which were assimilated by Lenner (1994). The results in Appendix B show that absolute cold start emissions of CO and HC are usually highest from noncatalyst petrol cars (often around 5-15g for CO, and 3-15g for HC at around 2 o C). Although catalyst cars are virtually devoid of emission control during the early parts of a cold-starting drive cycle, in absolute terms cold emissions from catalyst cars are no higher than from noncatalyst cars, and are generally substantially lower (around 15-6g for CO, and g for HC at 2 o C). The high relative cold start emissions of CO and HC observed for catalyst cars (with factors usually around 5-1) are a consequence of the very low emission rates during hot operation. For diesel cars, relative cold start emissions of CO and HC (around at 2 o C) tend to be slightly lower than those from non-catalyst petrol cars ( ), although in absolute terms, they are much lower (.7-1.9g for CO and.2-.9g for HC). Another conclusion that can be drawn from the data is that cold start emissions of NO x for con-catalyst petrol and diesel vehicles tend not to change much compared with hot operation. The E cold /E hot ratio therefore averages around 1. For catalyst cars, absolute excess emission tends to be positive, at around.7-2.g at 2 o C, with a relative emission factor of around Pavlidis and Joumard (1995) analysed results from several European laboratories (based on a sample of 1828 vehicles) in what was claimed to be the first statistical analysis of cold start data. From FTP cycle results and other existing data, average emission factors for eight different vehicle technologies were studied: (a) noncatalyst petrol, (b) petrol- closed loop catalyst, (c) petrolopen loop catalyst, (d) diesel (total), (e) diesel without oxidation catalyst, (f) diesel with oxidation catalyst, (g) turbo diesel, (h) nonturbo diesel. The averaged results are summarised in the tables given in Appendix C. The data showed that noncatalyst petrol cars have the highest absolute emissions of any technology, although cold NO x emissions were lower than corresponding hot emissions. Diesels again proved to have very low emissions of CO and HC, with their cold emissions never exceeding the hot emissions from catalyst petrol cars. With oxidation catalysts applied to diesels, slight improvements were observed in absolute emissions for all pollutants in both the cold and hot cycle. There is very little experimental data relating to the cold start performance of heavy duty diesel vehicles, except for Holman (1996), who analysed TRL cold start data for 1 heavy-duty engines over the US heavy-duty cycle. The results, averaged for all 1 vehicles, are summarised in Table 3. The apparently high excess emission value for CO 2 does not imply that there is a particularly severe cold start emissions penalty for this compound. During hot operation emissions of CO 2 are also considerably higher than emissions of the other pollutants listed. Despite the apparent trends indicated in the Table, it was found that there were no strong systematic trends in the emission characteristics of the engines when comparing cold and hot emissions. For some engines hot emissions were greater than cold emissions for particular pollutants. It was concluded that the average cold start emission penalty for the 1 engines was overly influenced by the high excess emissions from two of the engines. Excluding 15

20 Table 3 Average cold start excess emissions for heavyduty diesel vehicles (Source: Holman, 1996) Pollutant Approximate Approximate relative excess cold start cold start emission emission (grammes) CO HC < NO x -1 *.9 CO Particulate Matter * For NO x there appeared to be an average excess hot emission of around 1g. these two engines, and three engines that had not fully warmed during the cycle, there were, on average, no excess emissions from the remaining five engines. In contrast with the cumulative emission curves for passenger cars, cumulative emissions of CO, HC and particulate matter from the heavy duty engines were seen to increase virtually linearly with time Summary of cold start emissions according to vehicle category Most cold start testing has been performed on passenger cars, and generally with the pollutants CO, HC and NO x. There is clearly a great deal of variation of cold start emissions between individual vehicle types, and also between the average results from large groups of vehicles. However, some general trends have been identified from the test cycle data. Absolute cold start emissions are generally highest from non-catalyst petrol cars, followed by catalyst cars and then diesels. Cold start emissions of CO and HC from catalyst cars are high in relation to hot emissions. This is a consequence of the very low emission rates of these pollutants from catalyst cars during hot operation. Cold start emissions of NO x from all three categories of passenger car are usually small (ie less than 2g). Although the data are limited at present, cold start emissions from heavy-duty diesel engines appear to be minimal, with the possible exceptions of CO and particulate matter. 3.5 Vehicle mileage Effect of vehicle mileage on cold start emissions There is little evidence to suggest that either vehicle mileage and/or age are major factors contributing to cold start emissions. Mileage is considered to be a more important factor for closed-loop catalyst cars as it is associated with catalyst failure and deterioration, which will affect the level of pollutant emissions during both hot and cold start conditions. The most common reasons for high emissions from three-way catalyst-equipped cars are firstly the failure of the sensor controlling air:fuel ratio, and secondly failure of the catalyst itself (QUARG, 1993). The Dutch Ministry of Housing, Physical Planning and Environment (1992) (Source:QUARG, 1993) found that old catalysts take longer to reach their light off temperature, and deposits may build up on the surface of the catalyst which also increases the time taken for maximum efficiency to be achieved. The light off temperature itself can also increase with age (Ajne et al, 1992). Holman et al (1993) suggested that the rate of catalyst degradation varies with pollutant, and quoted figures which were based on discussions with industry experts. Table 4 gives the annual degradation rate for three pollutants as obtained by Holman. Table 4 Catalyst degradation rates (Source: Holman et al, 1993) Pollutant Nitrogen Oxides 1. % Volatile Organics 2.5 % Carbon Monoxide 3. % Annual Degradation Rate In a study of 25 catalyst-equipped vehicles with high odometer readings, Kutscher (1991) found that hydrocarbon emissions during cold start were about 3-4% higher for vehicles with an odometer reading above 8,km compared to those with a reading below 8,km. Pavlidis and Joumard (1995) considered the influence of vehicle mileage on cold start emissions. In Figure 12 absolute cold start emissions of CO, HC and NO x (y-axis) from closed-loop catalyst vehicles are plotted against vehicle mileage (x-axis). Average CO, HC and NO x absolute excess emissions showed a slight linear increase over the first 5,km (weak correlation, R<.1). After a given mileage (5,km for CO, 8,km for HC) an increase in absolute cold excess was detectable as the catalysts approached the end of their operational lives, although the authors added that more data was required in the high mileage region to establish this apparent trend. It was therefore assumed that as the catalysts approached the limit of their lifespan (8,-1,km), cold start emissions increased more rapidly than hot emissions. For CO, a regression analysis performed on the range above 5,km resulted in the line of best fit represented by: Y = X (R 2 =.55) where Y signifies the absolute cold emission (g) and X the respective odometer reading in km. Cold emissions from diesel engines failed to show any relationship with mileage Summary of vehicle mileage effects The effect of vehicle mileage on cold start emissions appears to be quite small, although it may become significant for high-mileage vehicles. Even so, it seems likely that catalyst degradation is a more significant factor in terms of increasing emissions at higher mileage. 16

21 [g] CO/1 Regres. CO/1 HC NO x [km] Figure 12 Relationship between odometer reading and cold start emissions of CO, HC and NO x from catalyst cars. Numbers at the top correspond to the number of cars in each mileage class (Source: Pavlidis and Joumard, 1995) 3.6 Engine size Influence of engine size on cold start emissions In general, smaller engines should cool down more rapidly than larger ones, although the warm-up rate is dependent on other parameters. There is little experimental data relating engine size to cold starting. The data analysed so far suggests that the effect is probably small and, even if an effect does exist, it may well be swamped by the variation in measured emissions between vehicles. However, one area that has not been investigated is the effect of engine size on cold start emissions after different soak times (section 2.3). A report by Klein (1993) for the Netherlands Central Bureau of Statistics describes a technique for determining emission factors from road traffic. Values for cold start emission factors were given (source: Rijkeboer and Van der Haagen (1992)). These values were averaged for the first 4km after starting and based on annual average temperature of 9 o C. The data included cold start emission factors for catalyst-equipped vehicles of two engine sizes: <1.4 litres and > = 1.4 litres. Table 5 contains the factors employed. For NO x the differences were small, but for CO and HC the urban emission factors were generally larger by a factor of two for the larger engined vehicles. Farrow et al (1992) measured emissions from forty noncatalyst petrol vehicles as part of a larger survey for the Department of the Environment and the Department of Transport. The objective was to assess the exhaust emissions of CO 2, CO, NO x and total hydrocarbons (THC), together with fuel consumption over a series of EC Table 5 Cold start emission factors (E cold /E hot ) for catalyst cars (Source: Klein, 1993) Engine Size CO VOC NO x Urban Non-urban Urban Non-urban Urban Non-urban < 1.4 l >= 1.4 l regulatory dynamometer tests, simulated road dynamometer tests and real on-road tests. The tests were carried out in the as received condition, as well as following engine tuning. It was found that tuning had little effect on overall results, with no statistically significant differences being recorded between the two conditions. At TRL an analysis was performed on the raw data provided in the report in order to identify any relationship between engine size and cold start emissions over the urban drive cycles. The plots of absolute excess cold start emissions of CO, THC, and NO x are given in Figure 13. The plot for NO x (and CO 2 -not shown) appeared to indicate that absolute cold start emissions of these components are independent of engine size. The results for CO and THC also point towards a similar lack of correlation. The variability of the measured parameters with respect to individual vehicles was large. Similar trends were also found for relative cold start emissions. Pavlidis and Joumard (1995) also found a similar lack of correlation between absolute excess cold start emissions of CO, HC and NO x and engine capacity over 8 capacity classes of diesel and closed-loop catalyst vehicles Summary of engine size effects No systematic links have been established between engine size and measured cold start emissions. However, engine size could potentially effect engine cooling rate at the end of a journey and therefore the effect on restart emissions at different and shorter soak times. 3.7 Ambient temperature Influence of ambient temperature on cold start emissions The automobile industry has successfully met the demands of progressive exhaust emission legislation mainly by the employment of catalytic convertors. Design guidelines have been determined by the nature of legislative drive cycles. Low ambient temperatures have not yet been considered in European test procedures, although such conditions have been found to increase emissions and to diminish the performance of catalyst systems (Laurikko, 1995). Bardon et al (1992) stated that delivered starting mixtures at Canadian winter temperatures are typically 1-3 times richer than stoichiometric, and mixtures during the warm-up phase remain well above stoichiometric until engine temperature exceeds 6 o C or more. Therefore it has also become necessary to examine the relationship between measured cold start values and ambient temperature, although there are several problems associated with this task. For example, it has been suggested that the only obvious point at which an engine can be described as specifically cold is when it is in thermal equilibrium with the ambient air. Cole (1992) suggested that this could represent one definition of a cold start, although the temperature of the ambient air could be anywhere between, for example, -2 o C and +3 o C. Gurney and Allsup (1989) found that emissions from different vehicles do not always correlate at different temperatures, and cars having the lowest emissions at high temperature 17

22 2 Excess cold start emission (g) Excess cold start emission (g) Engine size CO : Ecold (absolute) Vs. Engine size 4 petrol, non-catalyst cars (over 3 cycles) Engine size HC : Ecold (absolute) Vs. Engine size 4 petrol, non-catalyst cars (over 3 cycles) 1 Excess cold start emission (g) Engine size NOx : Ecold (absolute) Vs. Engine size 4 petrol, non-catalyst cars (over 3 cycles) Figure 13 Cold start emissions relative to hot start emissions as a function of engine size for 4 non-catalyst petrol cars. Results from TRL analysis of data from Farrow et al (1992) 18

23 can sometimes be the worst emitters at low temperature. A limited number of studies have investigated the temperature dependence of cold start emissions, and these have shown that cold start emissions of CO and HC generally increase as ambient temperature decreases (e.g. Laurikko, 1995; Hassel et al, 1993). At lower temperatures emissions tend to increase at a greater rate during the cold start period. It also takes longer for emission levels to stabilise. This renders the problem of cold starts even more important in winter, and in countries at high latitudes. The Transport Research Laboratory holds a substantial data set relating to the effects of ambient temperature on cold starts (already referred to in Section 2.2). The data covers the second-by-second emissions from six catalyst and three diesel cars tested over a drive cycle specified by TRL. The drive cycle was taken to represent realistic urban driving conditions. The cycle was developed from data collected during the DRIVE-modem European project (André et al, 1991), in which 58 privately-owned vehicles were instrumented to measure real-world speed profiles over the period of one month. Analysis of the data has shown that the ambient test temperature significantly affected the mass of exhaust gas emissions and fuel consumption of the pre-soaked vehicles, an effect particularly noticeable during the first part of the drive cycle. Each vehicle was tested at five ambient temperatures, - 1,, 1, 2 and 3 o C. At each temperature the vehicle was driven over a specified drive cycle. Each test point was repeated twice, once with the vehicle fully soaked at the ambient temperature and again with the vehicle fully warmed (engine coolant temperature 85 o C.). In the test programme the maximum ambient temperature of 3 o C was significantly below the component temperatures expected during normal vehicle operation, where fluid temperatures are in the range of o C. A vehicle starting from soaked temperature conditions is therefore effectively starting cold irrespective of ambient temperature. Figure 14 shows how cumulative emissions from a catalyst car varied with time. The graphs emphasise once again the extent to which the cold start period influences the total amount of pollutants produced by catalyst cars over the course of a journey, but they also show how the total levels of pollutants produced during this period depend upon ambient temperature. The effects were greatest for CO and HC emissions. It can also be seen in the examples that the catalyst was alight and working at the first significant acceleration (i.e. after approximately 18s) irrespective of the ambient test temperature (see also section 3.2.1). In Figures 15 and 16 average cold start and hot start emissions over the TRL cycle are compared at each ambient temperature studied in the range -1 o C to +3 o C. Emissions of CO from catalyst cars were particularly high during the first kilometre or so of the driving cycle when started from cold. For diesels the effect of ambient temperature on emissions was less marked, although for CO a systematic decrease in total cold start emissions with increasing ambient temperature was observed. For petrol catalyst cars it was clear that for CO and HC, excess cold start emissions were substantially reduced when the ambient temperature increased from -1 o C to 1 o C. A further reduction was evident at 2 o C and again at 3 o C, although the effect became smaller. The cold start CO emissions from catalyst vehicles at an ambient temperature of -1 o C were, on average, 361% greater than at 3 o C (79% higher for diesels). The equivalent increase for HC emissions was 336% for petrol and 33% for diesel. NO x emissions from both catalyst and diesel cars were less affected by ambient temperature and the cold start penalty was less marked. Emissions of NO x were dependent on the engine load which was mostly affected by the drive cycle. Ambient temperature affected the quantity of particulates from the diesel vehicles. The weight of particulates formed during the cold start tests at -1 o C were on average 45% greater than at 3 o C. Hassel et al (1993) carried out cold start emission measurements over the FTP cycle at ambient temperatures of 2 o C, 5 o C and -1 o C. Regulated catalyst, unregulated catalyst, conventional petrol and diesel passenger cars were soaked at the appropriate temperature for 12 hours before the beginning of each test series. For a more realistic evaluation of emission levels at colder ambient temperatures, on-car heating and lighting were switched on. CO, HC, NO x and CO 2 emissions were monitored continuously. Particulate emissions from diesel vehicles were available as integral measurements for phases 1 to 3 of the FTP test cycle. For HC and CO, all types of vehicle showed a marked increase in emissions with decreasing ambient temperature during phase 1 of the test cycle, and a sharp fall in emissions during phases 2 and 3. Conventional passenger car emissions of HC and CO exceeded those of regulated catalyst cars by a factor of between 3 and 4. The authors pointed out that diesel passenger cars had HC and CO emission levels during the cold phase which the regulated catalyst cars only achieved in the later stages of the test. It could also be seen from these results that the phase 1 (cold start) emissions increased most significantly on a g/km basis at lower temperatures. Therefore, this would indicate that most of the increase in emissions at lower temperatures occurs in the cold start period. Although the changes in hot emissions of CO and HC (phases 2 and 3 of the FTP cycle) were not generally systematic, other studies (e.g. Larson, 1989) have shown that hot emissions can also increase at lower ambient temperatures, although to a much lesser absolute extent. Emission levels varied to a lesser degree over the 3 phases for NO x than for CO and HC. What was apparent was the lower emission levels of the regulated catalyst car compared to the diesel cars during the first phase. The study provided more information on particulate emission levels for cold start conditions. Figure 17 shows the particulate emissions of the diesel cars studied for phases 1 to 3 of the FTP cycle as a function of ambient temperature. Emissions were seen to increase with decreasing ambient temperature during phase 1. For phases 2 and 3 particulate emissions dropped to around 6-7% of 19

24 6 5-1 o C CO cum per test o C 1 o C 2 o C 3 o C Time (s) HC cum per test o C o C 1 o C 3 o C 2 o C Time (s) 2.5 NO x cum per test o C -1 o C 1 o C 2 o C 3 o C Time (s) Figure 14 Cumulative emissions of CO, HC and NO x as a function of time over a TRL cycle for a Ford 2.1 catalyst car (after cold soak) 2