Investigation of Errors in NOx Recovery with a CVS Using a Vehicle Exhaust Emission Simulator (VEES) as a Cross Check Tool 2007-01-0312

The accuracy of low-level emission measurements has become increasingly important, due to the development and implementation of SULEV and PZEV vehicles. One technique to test the low-level measurement performance of a CVS is to inject a known mass of a trace gas, such as propane, into the sample system and verify that substantially all of the mass injected is recovered, typically within 2% of the total injected mass. A Vehicle Exhaust Emission Simulator has been used to inject precise amounts of trace gases with a known accuracy in the range of 0.5% to 1.0%. Recoveries for propane, carbon monoxide, and carbon dioxide are typically 98% or higher, while recoveries for nitrogen oxide are sometimes as low as 95% to 96%. In other words, as much as 5% of the injected nitrogen oxide mass is not recovered by the CVS. This represents an unexpected loss of 3% to 4% of the injected nitrogen oxide.

A test program was developed to characterize the loss of nitrogen oxide in this type of CVS injection test. The key variables studied include the nitrogen oxide concentration, total NO mass, CVS flow rate, residence time between injection and arrival at the sample bag, as well as CVS dilution ratio and interferences in the chemiluminescent NOx detector. An analysis of the results indicates that the mass lost is related in a simple fashion only to the concentration of the diluted nitrogen oxide. A correction function based on this result correctly predicts the mass of NOx recovered in the CVS bags over a wide range of CVS operating conditions. This suggests that the nitrogen oxide undergoes a chemical or physical reaction while in the CVS, and further experiments are underway to test this theory.