high accuracy nanoparticle sensor for combustion engine exhaust gases

The ensuing improved CNC nanoparticle sensor is now based on a vertical, annular design.. This gives the new design a uniquely high discriminatory power by nanoparticle diameter, a preci

Procedia Engineering 168 ( 2016 ) 35 – 38

1877-7058 © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference

doi: 10.1016/j.proeng.2016.11.129

ScienceDirect

30th Eurosensors Conference, EUROSENSORS 2016 High-Accuracy Nanoparticle Sensor for Combustion Engine Exhaust Gases

M Krafta,*, J Kaczynskia, T Reinischb, M Ungerb, A Bergmannb,c

a CTR Carinthian Tech Research AG, Europastraße 12, 9524 Villach, Austria

b AVL List GmbH, Hans-List-Platz 1, 8020 Graz, Austria

c Institute of Electronic Sensor Systems, Graz University of Technology, Inffeldgasse 10/II, 8010 Graz, Austria

Abstract

Nanoparticles are one indispensable analyte in the characterisation of any combustion engine or exhaust aftertreatment system

To improve the presently prevalently used condensation nucleus magnification-based particle counters (CNC) in performance and footprint, a dedicated, comprehensive all-in-one CFD model was developed Following experimental validation, the model was successfully used to simulate and better comprehend the internal functionings of the present standard system, identify critical design parameters and develop a new, improved sensor system design The ensuing improved CNC nanoparticle sensor is now based on a vertical, annular design Besides being highly compact, the new layout yields an almost perfectly homogenous (super-)saturation of the aerosol stream and superior temperature control of all relevant components This gives the new design a uniquely high discriminatory power by nanoparticle diameter, a precise controllability of the effective particle detection cut-on size and a significant reduction in the effective particle losses in the sensor

© 2016 The Authors Published by Elsevier Ltd

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference

Keywords: nanoparticle sensor; condensation nucleus counter; automotive exhaust analysis; system design; system simulation

1 Introduction

Among the various airborne pollutants, in particular nanoparticles < 100 nm are presently regarded with increasing concern As internal combustion engines in general – and Diesel engines in particular – are one relevant

* Corresponding author Tel.: +43 4242 56300; fax: +43 4242 56300 400

E-mail address: martin.kraft@ctr.at

© 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license

( http://creativecommons.org/licenses/by-nc-nd/4.0/ ).

Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference

particle source, the particle emission is a parameter that has to be reliably and efficiently measured [1] Similarly, the efficiency and effectiveness of exhaust aftertreatment systems, like (nano-)particle filters has to be controlled The present, legally mandated standard measurand for the analysis of such automotive exhausts is the particle number concentration, i.e the actual count of nanoparticles ≥ 23 nm per time unit As these particles are way too small for a direct, individual detection e.g by optical means, the standard approach is to exploit the condensation nucleus magnification principle to build a condensation nucleus counter (CNC) The aerosol stream is first saturated with working fluid vapours, typically n-butanol, and then cooled down to achieve a super-saturated atmosphere Under these conditions, the nanoparticles may act as condensation nuclei, i.e working fluid condenses on them and grows them to typically µm-sized particles that can then easily be counted [2] The critical particle diameter dK

above which this happens can be derived from the Kelvin equation according to

S T R

M

dK

ln

4

˜

˜

˜

˜

˜ U

V

(1)

The cut-on size of the detection method thus depends on i) the material properties of the working fluid, in

particular its liquid density U, surface tension V and molar mass M, and ii) the temperature T and gas (super-)

saturation S near the particle CNC-based particle sensing thus depends on carefully balanced processes to achieve

reliable, reproducible particle detection The increasing demand for ever more precise and reliable sensors and smaller footprints now calls for new, integrated approaches For this reason, in a first step a comprehensive, dedicated in-depth simulation model of a CNC sensor was developed, implemented and validated for an established, routinely used CNC device

2 Nanoparticle Sensor Modelling

2.1 Sensor Simulation Model

The entire sensor simulation model was established in ANSYS-Fluent Each simulation was based on a detailed 3D geometry model of the respective CNC sensor The flowing medium was modelled as air using the standard Fluent model; for all other gaseous, liquid or solid components, all relevant parameters were modelled as polynomial functions of temperature for the range 10 – 90°C, based on literature data Regarding the working fluid n-butanol, this included gas and liquid density, vapour pressure, surface tension, gas viscosity, gas heat capacity, gas thermal conductivity, evaporation enthalpy and the binary diffusion coefficient in air Air – working fluid mixtures were modelled using the ideal gas approximations in Fluent All solid parts, i.e Al for the structural parts and the respective polymers for the thermal insulation elements separating the different temperature zones and the saturator element, were defined by their material density, heat capacity and thermal conductivity The set temperatures were defined at the outer surfaces of the aluminium structural blocks All gas – solid interfaces in the saturator and downstream of it were modelled as working fluid vapour equilibrium sources / drains, as defined by the surface temperature and the related working fluid vapour pressure The simulation model thus accounts for the gas flow, heat exchanges between and resulting thermal distributions in all solids and the gas phase, the evaporation and subsequent distribution of n-butanol, and its super-saturation in the condenser zone of the CNC The Kelvin diameter distribution is then calculated from these values in post-processing using Eq 1

A key simplification made was modelling the aerosol particles as mass-less flow trajectories evenly spaced across the gas inlet, rather than individual items This assumption was validated by separate simulations using two-phase discrete particle models, in particular the Stokes-Cunningham approach for sub-micron particles These simulations showed that the reciprocal effect of the presence of solid nano-particles, and later of liquid micro-droplets, on the flow regime is negligible for the particle volumes occurring in a CNC sensor However, the work also showed that the approximation becomes increasingly unreliable for even smaller nano-particles ≤ 12 nm While not affecting the flow, for this particle size range enhanced simulations models – and probably a fundamentally different CNC sensor layout – would be required

2.2 Reference System Simulation

For model validation, the CNC simulation model was applied to the geometry and operation conditions of a TSI

3790 engine exhaust condensation particle counter (TSI Inc., USA) The TSI 3790 is based on a vertically aligned saturator – condenser – detector layout (Fig 1, left) The aerosol flow is split into 8 separate, cylindrical channels to increase the surface / volume ratio and passes through a micro-porous polyethylene wick element saturated with n-butanol After a thermal insulator, the gas streams are cooled in the condenser, creating the required supersaturation, and recombined before being passed through a nozzle into an optical stray-light detector

Fig 1 left: Basic layout of the TSI 3790 system; centre: temperature distributions; right: Kelvin diameter distribution in the condenser section

For model validation, the simulation-predicted response function to different particle sizes was compared to the measured data acquired with a TSI 3790 under identical operation conditions, yielding a virtually identical shape with a ~ 2 nm offset of the model towards lower particle diameters The subsequent detailed investigation yielded a few interesting points First, while thermal equilibration of the gas to the desired temperature is reliably achieved throughout the device, the internal section of the saturation element is significantly cooler than the outside (Fig 1, centre) This effect is caused by the solvent evaporation in combination with the rather low thermal conductivity of the solvent-saturated porous polymer This low thermal conductivity furthermore is a main cause of the systems inertia when being warmed-up to operating conditions Secondly, the aerosol flow is slightly inhomogenously split into the eight channels, with a higher flow velocity and hence lower gas saturation in the channels opposite to the inlet This flow variability reflects in the super-saturation and hence the Kelvin diameters in the single channels in the condenser section (Fig 1, right) Altogether, while not detrimental to the overall function of the device, the resulting inherent variability of essentially 8 parallel CNCs effectively broadens the response function (Fig 3)

Making use of the validated simulation model, a number of key optimisation parameters for the design of a CNC-based nano-particle sensor were identified, including high surface/volume ratios in saturator and condenser, possibly homogeneous, fully laminar flow, an optimised heat transfer to the evaporation surface and a minimised heat flow between saturator and condenser Based on these premises, a vertically oriented annular system was developed (Fig

2, left) In this layout, the aerosol stream enters the saturator section radially through an aerodynamically optimised tapered spiral that ensures a perfectly homogeneous distribution of the gas flow across the vertical flow channel The saturator located above features microporous polymer wicks on both the inner and the outer surface to optimise the saturation behaviour Using two separate, thin cylindrical elements in connection to metal supports warrants an efficient heat transfer, reducing the temperature gradient across the saturator is by more than 75% in comparison to the standard device This was found to greatly enhance system stability and controllability, and significantly reduce the warm-up time

Given the homogenous flow distribution, the aerosol stream is almost perfectly homogenously saturated with working fluid vapour (Fig 2, right) From this immediately results a much sharper size-discrimination function (Fig 3) and a slightly higher total particle transmission The cut-on point shifts by ~ 1 nm/K temperature difference of saturator and condenser, allowing a precise control of the sensor sensitivity criterion These effects could be experimentally validated, and may be a new starting point for future sensing uses

Fig 2 left: Basic layout of the annular high-accuracy CNC nanoparticle sensor; right: Relative gas saturation of the aerosol stream in the

saturator section; the inset illustrates the saturation at the saturator – isolator interface, showing an inhomogeneity less than 0.7%

Fig 3 Sensor response curves to identical combustion aerosols of various sizes, with both CNCs operated under identical standard conditions

Acknowledgements

This work was performed within the Competence Centre ASSIC - Austrian Smart Systems Research Center,

co-funded by the Austrian Federal Ministries of Transport, Innovation and Technology (bmvit) and Science, Research and Economy (bmwfw) and the Federal Provinces of Carinthia and Styria within the COMET - Competence Centers for Excellent Technologies Programme

References

[1] B Giechaskiel, et al., Calibration of Condensation Particle Counters for Legislated Vehicle Number Emission Measurements, Aerosol Science and Technology 43(12): 1164-1173 (2009), doi: 10.1080/02786820903242029

[2] Y.S Cheng, Condensation Particle Counters, in P Kulkarni, P A Baron, K Willeke (Eds.), Aerosol Measurement: Principles, Techniques,