NOx emissions in direct injection diesel engines – part 1: Development of a phenomenological NOx model using experiments and three-dimensional computational fluid dynamics

Abstract

There exists a well-established correlation of exhaust NOx emissions arising from diesel engines with the adiabatic flame temperature, in particular for conventional (i.e. short ignition delay, diffusion combustion-dominated) operating conditions. Most published NOx emission models rely on this correlation. However, numerous experimental studies have identified operating conditions where this correlation fails to capture the exhaust NOx trend. In this work, a novel phenomenological NOx model concept is introduced, including a first successful validation against experimental data. The model development is based on experimental observations and is supported by three-dimensional computational fluid dynamics computations, strengthening the understanding of the underlying mechanisms leading to the discrepancy between the adiabatic flame temperature and exhaust NOx trend. For long ignition delay operating conditions, the improved mixture preparation before ignition leads to reduced mixing rates during and after combustion. Both the improved mixture preparation before ignition and the instantaneous increase of mass observed above 2000 K after start of combustion are due to compression heating of the burned gases. Key features of the model are improved description of mixture distribution at start of combustion, NOx formed in products of premixed burn, different physical treatments of premixed and diffusion sourced products, and inherent consideration of burned gas compression heating. Model results capture the NOx emissions for conventional diesel combustion, as well as for operating conditions where the NOx emissions do not follow the adiabatic flame temperature trend. Moreover, the results show that the contribution of NOx from products from premixed burn and the consideration of compression heating effects on burned (postflame) gases are essential to capture the NOx emissions under the latter conditions.

NOx emissions in direct injection diesel engines: Part 2: model performance for conventional, prolonged ignition delay, and premixed charge compression ignition operating conditions

“…In this section, the phenomenological NOx model that has been introduced in Part 1 22 is validated against a broad range of operating conditions and data from the two engines. First, the model calibration from Part 1 is extended for the influence of charge dilution (i.e.…”

“…Based on the successful calibration of the SCE (see Part 1 22 and previous section), in this section, the model performance is discussed with respect to different single parameter variations. The operating conditions of the different variations are summarized in Table 5.…”

“…The following section points out the key aspects of the phenomenological NOx model used, which is presented in detail in Part 1. 22 The model belongs to the family of quasi-dimensional package models 14,23 and differentiates between the products of premixed and diffusion combustion. Furthermore, it is coupled to a one-dimensional (1D) diesel jet model 24 to compute the fuel distribution in mixture fraction space at start of combustion (SOC), and therefrom the ratio at which combustion products are allocated to multiple premixed combustion product zones with varying air-to-fuel ratio.…”

“…The development and validation of the new phenomenological NOx model is structured in two parts; the validation being presented in this article (Part 2). In Part 1, 22 the understanding of the correlation of premixed combustion and NOx emissions in DI diesel engines has been strengthened by a comprehensive analysis of three-dimensional computational fluid dynamics (3D-CFD). It was concluded that at operating conditions with long ID, leading to large proportions of premixed combustion, compression heating of the burned gases contributes substantially to the NOx trend-reversal, supporting the assumption postulated by Musculus.…”

“…The article starts with an outline of the key components of the phenomenological NOx model, which are addressed in more detail in Brückner et al 22 This is followed by an overview of the experimental facilities used to obtain the experimental data. Then, the model calibration from Part 1 is extended to account for the influence of EGR.…”