Development of Diagnostic Algorithms for Exhaust Gas Recirculation Systems in Heavy-Duty Diesel Engines, 03-R9697Printer Friendly Version
Inclusive Dates: 04/01/07 03/31/08
Background - In 2006, the California Air Resources Board adopted the so-called heavy-duty, on-board diagnostics rule. This rule requires that all emissions-relevant failures on heavy-duty engines be detected, creating a large set of technical challenges for heavy-duty diesel manufacturers. The rule has a phase-in period beginning in 2010 and extending through 2016. Many of the recently introduced heavy-duty diesel aftertreatment components have not been used on production engines, so the failure modes and their effect on emissions are not fully understood. The goal of this project is to develop methods to simulate emissions-related component faults and design diagnostic algorithms to detect these faults.
The number of diagnostic monitors required to meet the heavy-duty on-board diagnostics rule is enormous. Depending on the type and number of emission-related components on a given engine, the number could be as high as 250. In order to reduce the scope to a manageable size for a one-year project, it was decided to focus on the exhaust gas recirculation (EGR) system, used primarily to reduce emissions of oxides of nitrogen (NOx). This system was chosen because of its expected inclusion on most on-road, heavy-duty diesel engines produced from 2010 until the foreseeable future, and because it provided a manageable work scope for the project.
Approach - The program plan was designed around a set of assumed failure modes. Three of the four components in a high-pressure exhaust gas recirculation system valve, cooler, and pipe had a variety of simulated failures introduced. For example, pipes had small holes drilled to simulate cracks and the cooler had some passages plugged to simulate clogging. While a comprehensive list of real-world failure modes can only come from in-use testing, the project included a list of likely failures for each component. The program included three phases: component characterization, metric development, and algorithm development.
Accomplishments - Methods to produce repeatable simulated faults in the EGR system were developed for each failure mode: leaks, restrictions, and reduced cooling. Steady-state engine emissions tests were conducted for each fault type with varying degrees of failure to generate data required to design the diagnostic monitors. Various methods of developing diagnostic monitors were evaluated including static maps, physics-based dynamic models, and neural-network models.