An article by David Price first published in Engine Technology International
Diesel aftertreatment systems are now commonplace in every type of on-road vehicle, from passenger cars to heavy-duty trucks. In order to meet emissions and diagnostic regulations, sophisticated control systems are required. These should have a detailed understanding of the various catalysts and filters within the aftertreatment system.
In conjunction with its customers, Pi Innovo uses production-ready OpenECU electronic systems to speed up development of advanced model-based control solutions for diesel exhaust aftertreatment. Pi’s M250 and M460 OpenECU platforms are suited to meeting the needs of aftertreatment applications.
The latest emissions regulations can only be met with control systems that understand what is happening within the various catalysts and filters. It is proving to be extremely difficult and costly to directly measure these parameters. To overcome this challenge, Pi Innovo recommends creating a mathematical model of catalytic elements, such as DOC, DPF, and SCR. Sensors can then be placed both upstream and downstream of each model to calculate parameters.
Extensive research is currently taking place to analyze the complex interactions that occur deep within aftertreatment systems. Selecting the mathematical model complexity is a key issue in all systems. Model complexity has been categorized from the simplest (0D) to the most complex (3D).
The majority of today’s model-based mass production systems incorporate 0D models, and these models use a single set of equations for each complete catalyst/filter element, and assume that the entire element has the same average conditions.
Even with this basic complexity, models require considerable time and effort to develop and calibrate, often involving everyone from the substrate chemists and the exhaust system supplier to the vehicle manufacturer.
Exhaust systems for the smaller volume manufacturers and the retrofit market pose the additional challenge of tuning a model and control system to fit a variety of low volume applications.
The current trend is to migrate to 1D models, which consider variations in one dimension, usually along the axial flow of the exhaust. They have the capability to provide virtual sensor information, thus providing data as if an additional sensor is located at a point within the catalyst. For example, a 1D model of a DOC enables the control of the considerable changes in temperature that can occur when exothermic reactions take place. Furthermore, when moving from 0D to 1D, improvements in the temperature tracking accuracy from 20°C to 5°C have been achieved.
The increased performance and cost ratio of the latest electronics, such as OpenECU, has removed any production cost penalty for using embedded 1D models. As a result, the performance of the 32bit-based OpenECU products is sufficient to execute multiple 1D models, control strategies, and full production diagnostics. Standard versions of these ECUs are available off-the-shelf, in prototyping or low volume production quantities.
There is a great deal of extra effort required when developing 1D models. However, this effort results in several benefits. The choices made in optimizing the model will depend on the number and location of sensors in the exhaust (pressure, temperature, and gas sensors, such as NOx or O2). The virtual sensor capability of 1D models enables a trade-off between development cost and the piece price cost of additional sensors in the system. A more complex model may also allow a reduction in the amount of calibration effort required to tune the system to the vehicle. There is a complex trade-off between the costs of the system development and potential reductions in production volume costs.
The OpenECU platforms allow users to develop control systems more rapidly and tailor them to a variety of production applications. This rapid development cycle can provide additional savings when assessing the cost benefits of developing new control systems.