Combining the Classical and Lumped Diesel Particulate Filter Models

Abstract

The control of PM that emanates from SIDI and CI engines is of significant concern for the automotive industry. This is because PM can infiltrate the alveoli in humans and transport toxic chemicals absorbed on the surface of the particle, subsequently leading to lung related illnesses [1]. As a result, the automotive industry now widely implements PM filtration in the exhaust. In particular, DPFs are used for CI engine exhaust with GPFs currently being considered for SIDI engines.The operation of a DPF (or analogous GPF) includes the storage of PM in and on top of a porous wall with its eventual regeneration in an oxidative environment. This can be accomplished actively utilizing O 2 or passively by employing NO 2 ; often through synchronization with an upstream Oxidation Catalyst. This need to modulate device operation has led to the development of simulation tools in order to reduce the number of experiments needed by predicting the pressure drop through the device and its temperature evolution. Broadly speaking, the models typically fall into one of two categories; those designated to operate within an Engine Control Unit (ECU) [2] and the classical two-channel approach [3].The ECU approach is designed for computational speed and is still often based on the two-channel model originally posed by Bissett [4]. In a previous effort by the author, the research in this area was reviewed and a derived model was generated that used dynamically incompressible flow to simulate the flow through the device while creating lumped zones in order to predict the temperature evolution. Utilizing a high-level coding language that runs significantly slower than traditional programming languages [5], the filter temperature evolution could be predicted relatively accurately in a faster than real time manner. This was because only one term in the governing energy equations was influenced by compressible flow. However, decoupling temperature and pressure by using dynamically incompressible flow (effectively negating the ideal gas law), the model was unable to simulate the pressure drop during warm-up and PM regeneration experiments.As a result, this effort seeks to merge the two models by simulating temperature using the ECU approach while estimating the pressure drop using the two-channel methodology. In specific, the temperature profile through the device is generated by the ECU version that is then utilized as a constant within the two-channel model removing the need to solve for the channel temperatures. Moreover, deep bed filtration is included in the model formulation in order to simulate the pressure drop evolution in the device more accurately. Finally, model run times are given in order to document the computational expense of the combined model. DPF MODEL DERIVATIONIn a previous effort, instead of simulating both the inlet and outlet channels along with a porous wall, model derivation followed a lumped approach through the creation of a variable number of zones Combining the Classical and Lumped Diesel Particulate Filter M...

Development of a 2D SCR Catalyst on a Diesel Particulate Filter Model for Design and Control Applications to a Ultra Low NOx Aftertreatment System

“…This model used a lumped approach to solve the temperature equation in each of the radial and axial zones that are part of the 2D mesh developed using resistance node methodology. This model was implemented in real time in an ECU and combined with the classical filtration model [32] Konstandopoulos et al [12] developed a multidimensional CPF model with multiphase approach to simulate non uniformity's in the filter. The model was based on several single channel descriptions involving a 3D mesh in which several partial di↵erential equations were solved in a CFD framework.…”