Engine exhaust emissions are normally measured using gas analysis equipment which determines component gas concentrations on a volumetric basis (% or ppm). Emissions legislation is generally written in terms of the mass of emissions produced for a given amount of work done by the engine (kWh or distance driven over a defined drivecycle). The calculation of mass emissions from volumetric data requires a measure of exhaust flow rate on a molar (volumetric) basis. Emission measurements are used in development/approval of engines including new engine type.

There are two basic known approaches to emissions measurement in routine use, namely dilute exhaust sampling and raw exhaust sampling.

The dilute approach is required for most legislative testing because it is non- intrusive and requires no instrumentation on the engine-the exhaust gas simply flows into a sensing space allowing the exhaust flow rate to be derived from the volume collected in a given time, and the emissions to be measured. In this case the total dilute exhaust flow is measured directly, often relying on the fair assumption that the properties of diluted exhaust are similar to those of air.

Dilution systems of this type are normally controlled so that the total flow of dilute exhaust remains constant throughout the test, hence the term Constant Volume Sampling, or CVS. Dilute sampling is well suited for measurement of the total (integrated) emissions over a transient drive cycle.

However, there are serious problems with the technique. The vehicle tested needs to be placed under strictly controlled conditions and run through a rapidly defined drive cycle, both of which, in practice, can nevertheless differ significantly from test to test. Furthermore the hardware requirements including test bed, controlled environment and monitoring equipment are costly. Accordingly this system requires expensive facilities and suffers from poor reproducibility.

The second approach, raw emissions sampling, is used for most development testing, particularly on the engine testbed. This approach requires additional instrumentation on the engine, for example fuel/air mass flow meters. The exhaust molar flow rate is then inferred from measured flows and exhaust gas composition data as described below. Standard testbed flow measurement systems have very limited dynamic capability, so these methods for raw sampling are generally unsuitable for transient testing.

For raw emission sampling in diesel engines, where the air-fuel ratio (AFR) is generally high, the conventional approach is to assume that density of exhaust is the same as that of air (as the proportion of fuel is comparatively low), so the molar exhaust flow rate-the base value measured in emissions testing-can be derived from the sum of air and fuel mass flow rates and the molar mass of air according to equation (1): <BR> <BR> <BR> <BR> <BR> <BR> gAir + gFuel<BR> qmExh =<BR> <BR> <BR> MAir (European Heavy Duty Regulations Euro 3 (EU directive 1999/96/EC))

External air flow meters are not generally fitted for gasoline engine tests, and because of the lower AFR the exhaust density is different from that of air. The established approach for gasoline engines therefore uses carbon balance which leads to the following equation (2): <BR> <BR> <BR> <BR> <BR> <BR> gFuel 1<BR> <BR> <BR> qmExh MFuel (CO2 + CO + HC) (2) (see for example the US Federal regulation (CFR 40 § 86.345.79) However, a problem with this formulation is that as C02 is dominant over CO and HC, this result is very sensitive both to errors in C02 and fuel flow. Also, as fuel flow and emissions are measured at different locations, combination of values taken simultaneously will not provide an accurate value for qmExh as the formula will combine, say, the fuel flow at a given instant with the emission arising from the fuel flow at a slightly earlier time. In addition, if the fuel mass flow goes to zero (e. g. during overrun conditions) then the CO2, CO and HC could also fall to zero. In this case the molar flow of exhaust is indeterminate, even though there continues to be positive flow. This demonstrates a fundamental weakness in the methodology when dealing with extreme transients, but more important is the fact that any errors in the time alignment of the emissions data relative to the fuel flow will be amplified into large features in the exhaust flow trace.

A further problem with raw emissions sampling is that it is carried out in a sequence of steady state condition of engine speed and load in order that a 3- dimensional mapping can be constructed. As a result, this approach is overly rigid and time consuming and deals poorly with transients.

According to the invention there is provided a method of obtaining a measure of exhaust flow rate in an engine comprising the steps of obtaining air flow, fuel H: C ratio, C02, CO and HC data and deriving the exhaust flow rate measure therefrom and an exhaust flow rate monitoring system comprising air flow, fuel H: C ratio, COs, CO and HC measurements extractors and a processor arranged to process said measures to obtain the exhaust flow rate.

The invention further extends to a vehicle control system comprising a monitoring system as described above and a controller arranged to control a vehicle engine based on monitored values and a method of evaluating a fuel or engine comprising the steps of driving an engine with a fuel and obtaining a measure of exhaust flow value according to a method as described above.

The invention provides a range of advantages. It allows emissions evaluation without the extensive facilities or poor reproducibility of the CVS system making it more cost and time effective for emissions development as a result of the improved mathematical modelling used.

Embodiments of the invention will now be described by way of example with reference to the drawings, of which: Fig. 1 which is a simplified diagram of an engine and monitoring system; and Fig. 2 is a diagram of an on-vehicle monitoring system.

An engine emissions monitoring system according to the invention is shown generally in Fig. 1. An engine 10 is controlled by an engine control unit (ECU)

12 controlling, inter alia, air flow, fuel flow and air to fuel ratio (AFR or A) into the engine. The engine exhaust gases are exhausted through an exhaust outlet 14 to a monitoring space 15 having CO, HC and CO2, sensors mounted thereon. These various components will be well known to the skilled person and any appropriate known individual components can be employed.

Accordingly a further detailed description of each of the components is not included here.

A processor 20 is provided for processing the various data required to monitor exhaust emissions. The processor 20 is shown as a separate component but can be part of the ECU, separately mounted in a vehicle or a stand-alone system in a test-bed environment.

As discussed in more detail below, the system allows the determination of transient mass emissions from raw sampled emissions data. The transient mass emissions can then be integrated over a drive cycle to give a result comparable to that from a CVS test. The system shown in Fig. 1 provides dynamic instrumentation for flows and emissions and minimises the effects of errors in the dynamic"time-alignment"of measured signals. Dynamic measurements of air flow, fuel flow and AFR are generally available from the ECU 12 interface, based on the control sensor information available to the ECU. These signals can be calibrated against testbed instrumentation under steady-state conditions.

This ECU data is, by definition, synchronised with engine combustion events, which means that the time alignment problem is only concerned with the variable transport times between the engine and emissions analyser. The system, however, assumes constant transport delay between engine and emissions sampling points or that the transport delay is known.

The system further makes use of more sophisticated processing of the data to provide more accurate representations of mass emissions.

Equation (2) can be rewritten dependent on airflow rather than fuel flow data where the fuel flow is derived from air flow and AFR as equation (3): <BR> <BR> gais 1 1<BR> MFU AFR (CO. +CO+HC) This suffers from the same problems, however, as it is merely a rewriting of the equation. However AFR can be derived from the exhaust composition. One known way of doing this is using the Spindt and Brettschneider algorithms.

SAE technical papers 650507 and 972989 respectively. This well known algorithm makes use of all the measured emissions including 02. It can be shown that substituting the Spindt expression in sequence (3) gives equation (3a) as follows: A problem with this approach is that it relates to obtaining a value for 02which requests an additional sensor and one that is known to be affected by measurement error. Although 02 tends toward 0.21 as COs goes to zero, any time errors will lead to large erroneous features in exhaust flow.

The invention addresses this by expressing AFR in terms of CO2, CO and HC only, and discussed in more detail below.

As a result the molar flow rate dry exhaust can be calculated from air mass flow and emissions data and used, with emissions volume concentrations, to calculate emissions mass flow rates as follows: The inputs to the system are: Intake Air Mass Flow (kg/s) 9Air Fuel H: C Ratio (-) n C02 Dry Volume Fraction () [CO2] CO Dry Volume Fraction (-) [CO] HC Dry Volume Fraction (-) [HC] Water-Gas Reaction EquilibriumConstant (-) K and a constant is: Molar Mass of Air (mmAir) 0.028964 (kg/mol) MAir The fuel H: C ratio-the average number of hydrogen atoms per carbon atoms in the fuel molecules is of importance because it determines the ratio of carbon dioxide to water resulting from complete combustion of the fuel.

We assume that hydrocarbon fuel can be characterised by CHn and that the combustion process of fuel in dry air does not produce any Nox.

The combustion of hydrocarbon fuels in dry air can be described in terms of the chemical reaction : CH, + A(O2 + 79/21N2)#aCO2 + bCO + cH2O + dH2O + dH2 + eO2 + fCHn + gN2 where the mols of air A is related to mass air-fuel ratio by

MC + n#MH<BR> A = AFR# (4)<BR> <BR> Mo2 + 79/21#MN2 Four atom balance equations can be derived from the reaction equation: Carbon 1 = a + b + f Oxygen 2A = 2a + b + c + 2e 5 (a) - (d)<BR> Hydrogen n = 2c + 2d + nf Nitrogen 79/21A = g Now if the number of mols of dry exhaust is Nd = a + b + d + e + f + g, then the volumetric fractions of the measured emissions can be expressed as : <BR> <BR> <BR> <BR> <BR> <BR> [CO2]dry = a [CO] dry = b = f (6)<BR> <BR> <BR> <BR> Nd Nd Nd The water-gas reaction equilibrium equation is also known: <BR> <BR> <BR> <BR> <BR> [CO][H2O] b#c<BR> = = 3.5 (7) [CO2][H2] a#d Now using the carbon balance, note that a+b+f= ( [CO,] + [CO] de + [HC] dry)#Nd = 1 (8) <BR> <BR> <BR> <BR> <BR> <BR> and hence<BR> a = [CO2]dry b = [CO]dry f = [HC]dry X X X (9) <BR> <BR> <BR> whereX= [CO2]dry +[CO]dry +[HC]dry = 1/N@<BR> <BR> Nd Note also that as b = [CO] dry the water gas reaction equation becomes a = [CO2] dry [CO]dry c [COJ d' Solving this with the hydrogen atom balance gives expressions for d and c:

In (l-f) [CO] d<BR> <BR> d =<BR> <BR> [CO] dry + K#[CO2]dry (11) & (12) 1/2n#(1 - f)#K#[CO2]dry [CO] dy + K#[CO2]dry Solving the oxygen and nitrogen atom balances gives: e + g = A#(1+79/21) -a - 1/2b - 1/2c<BR> 1/4n#(1 - f)#K#[CO2]dry<BR> = A#(1+79/21) - a - 1/2b - (13)<BR> <BR> <BR> [CO]dry + K#[CO2]dry Eliminating a, b and f using expressions derived earlier gives: which by partially substituting for Nd, rearranges to:

and thus Now Nd is the number of mols of dry exhaust that are produced per mol of CH, fuel, so the molar flow rate of dry exhaust is given by: <BR> <BR> <BR> qmExhdry = Nd#gFuel where gFuel is the mass flow rate of fuel (19)<BR> MC+n#MH and hence This formula for the molar flow rate of dry exhaust (mol/s) is less sensitive to the measured emissions and so is correspondingly less sensitive to errors in the time alignment between emissions and air flow data. The equation can be simplified further if: CO « C02, or k > 0.8 to give and for stoichiometric or lean operation a good estimate is given by

Equations 20 to 22 provide a measure of dry exhaust flow rate which is independent of the value of oxygen measurement. In addition to avoiding the problems inherent in measuring O2 levels, a "1/(1+X)" form of equation is obtained which is representative of the physical situation, namely that dry exhaust is expected to be similar to, but less than, the volume of air. Further still, because X is typically less than approximately 0.07, the result is insensitive to errors in X and hence to fluctuations in transport time and emissions.

It will be appreciated that an alternative form of equation (20) can be obtained for hydrocarbon fuels characterised by CHnOm using the chemical reaction CH2nOm + A(O2 + 79/21N2) # aCO2 + bCO + cH2O + dH2 + eO2 + fCHn + gN2 (23) In that case equation (4) becomes <BR> <BR> <BR> <BR> <BR> <BR> A = AFR#MC + n#MH + m#MO (24)<BR> <BR> <BR> <BR> MO2 + 79/21#MN2 and the oxygen atom balance becomes <BR> <BR> <BR> <BR> Oxygen 2A+m=2a+b+c+2e<BR> <BR> (2j) As a result equation (14) and (15) become

(27) Following the same derivation as for equations (16) to (19), we arrive at an adjusted equation (28) The possibility of formulating the exhaust flow equation in this form arises because it expresses a relationship between the volume of the dry products of combustion (QMExhdry) and air consumed in the reaction Gai, JMair. This relationship depends on fuel properties, air/fuel ratio and the completeness of combustion. These same factors uniquely determine the composition of the exhaust. It is therefore possible to express the relationship between dry exhaust volume and air flow in terms of fuel properties and exhaust composition alone without explicit determination of the air fuel ratio or completeness of combustion. By deriving the equations in the form independent of the 02 value, the equation further represents intuitively the physical situation in which the 02 has effectively disappeared from the air leaving the major component of nitrogen in both air and exhaust which has minimal involvement in combustion reactions. In contrast equation (3a) is of the form 1/ (X+Y) where X and Y are of similar magnitude and measurement error, giving rise to a similar uncertainty in the derived result.

Use of the new formula reduces the effect on exhaust flow of errors in time alignment between emissions and airflow traces. It is also important to minimise the magnitude of these errors. The problem is that during transient operation the time taken for exhaust to travel from the engine to the emissions sample point is variable and may be unknown.

However if the emissions sample point is close to the engine (e. g. pre/post a close coupled catalyst) then the transport time is always small, and the absolute magnitude of the variation is small compared with the response time of the emissions analysers, and this is implemented in the preferred embodiment. This means that a simple fixed time shift is acceptable for alignment of traces.

A further implementation of the invention is shown in Fig. 2 where like reference numerals relates to like elements. In this embodiment the system forms part of an in-car unit for real-time monitoring. CO, HC and COz sensors 16,17, 18 respectively are mounted on the exhaust gas outlet (exhaust pipe) 14 and provide monitoring signals to the ECU 12. The emission data can be used for a range of purposes including on-board diagnostics for emissions, engine management, catalyst failure and so forth.

It will be appreciated that the invention can be implemented in any of a number of ways. Mass air flow can be measured using testbed instrumentation or engine management system sensors, or determined from fuel flow and air fuel ratio (universal exhaust gas oxygen sensor or UEGO) data. The method can be applied to both engine-out and post-catalysts/tailpipe emissions data and can be used in any appropriate type of combustion engine including gasoline and diesel engines.