ON ROAD PARTICULATE EMISSIONS FROM A DIESEL PASSENGER CAR AROUND CAMBRIDGE Jonathan P.R. Symonds, Mark G. Rushton and Kingsley St.J. Reavell. Cambustion Ltd., J6 The Paddocks, 347 Cherry Hinton Road, Cambridge CB1 8DH, U.K. Introduction A Peugeot 406 2.2l HDi passenger car was driven around Cambridge and its environs whist monitoring the particulate size spectrum in real-time using a Cambustion DMS50 mobile fast particulate spectrometer. As well as the full spectrum, nucleation and accumulation mode number, CMD and GSD and accumulation mode mass were calculated. Using the logged road speed, the tests were then repeated on a more traditional chassis dynamometer (“rolling road”) facility to compare real world emissions with those in a laboratory environment. Instrumentation and Experimental Details The DMS50 is a new differential mobility analyser which classifies aerosol particles from 5 nm to 560 nm from their electrical mobility, producing both a size spectrum and multi-lognormal fits to instrument response in real time, with a time response of around 500 ms and data rate of up to 10 Hz. Particles are charged with a unipolar diffusion charger and pass into a classification column, where a radial electric field diverts them from a sheath air flow until they land on any of 22 detection rings, each connected to an electrometer circuit (Fig.1). The instrument transfer function is used generate a 34 channel size spectrum from the 22 electrometer signals, assuming a smooth continuous spectrum. High voltage electrode Aerosol Charger Sheath air flow Grounded Electrode Rings (11 of 22 shown) Charged Particle Trajectories Space charge guard Figure 1: DMS50 Schematic A Bayesian statistical algorithm fits up to two lognormal functions, chosen to maximise a posterior probability given by the product of the likelihood of currents synthesised from these functions matching the measured currents and an assumed prior probability based upon the CMD and GSD of the particles [1]. By tuning the map of prior probabilities, the algorithm can distinguish between the nucleation (volatile) mode (~<30nm, narrow GSD) from the accumulation (soot) mode (~>30nm, wider GSD) in diesel combustion aerosols. Only statistically significant modes are returned, based upon the noise base of the instrument. The use of lognormal functions helps to suppress noise in the spectrum, which when present in the large end tail of a size distribution can lead to large errors in volume or mass weighted spectra. Accumulation mode particulate mass is calculated directly from the lognormal parameters by application of the Hatch-Choate equations, using a fractal dimension of 3.19 for Diesel agglomerates. This value was arrived at by considering existent literature data on DMA