The use of a partial flow sampling system (PFSS) to measure nonroad steady-state diesel engine particulate matter (PM) emissions is a technique for certification approved by a number of regulatory agencies around the world including the US EPA. Recently, there have been proposals to change future nonroad tests to include testing over a nonroad transient cycle. PFSS units that can quantify PM over the transient cycle have also been discussed. The full flow constant volume sampling (CVS) technique has been the standard method for collecting PM under transient engine operation. It is expensive and requires large facilities as compared to a typical PFSS. Despite the need for a cheaper alternative to the CVS, there has been a concern regarding how well the PM measured using a PFSS compared to that measured by the CVS. In this study, three PFSS units, including AVL SPC, Horiba MDLT, and Sierra BG-2 were investigated in parallel with a full flow CVS. The focus was on the ability of PFSS units to determine PM emissions steady-state and transient engine operation relative to a full flow CVS.
Although PFSS units are in common use as a substitute for CVS PM measurement under steady-state conditions, the correlation obtained in the baseline tests conducted in this study using state-of-the-art PFSS units were far from satisfactory. Agreement between PFSS and CVS PM emission values was improved when PM emissions with the PFSS units were calculated by applying CO2-based dilution ratio, based on real time dilute and exhaust CO2 measurements, rather than using flow-based dilution ratio that is typically used with PFSS units. Further improvements for the steady-state data, particularly for SPC and MDLT, were also made by orienting the sampling probes in the exhaust pipe to face upstream to match CVS sample probe direction, and by modifying sample probe and exhaust pipe diameters to achieve near isokinetic sampling. For light load engine operation and the ISO 8-Mode Test, however, discrepancies between PFSS and CVS results remain an issue. Although volatile material content of the PM is one factor that may be responsible for such discrepancies, further work is required in this area.
For transient engine operation, none of the PFSS units, when used in real time, compared well on PM emissions relative to the CVS. However, operating the PFSS in a look-ahead mode, where the sample probe response was advanced relative to a pre-recorded exhaust flow trace of the same kind as the transient cycle to be tested, improved the PM results for the SPC to within 5 percent of the CVS. Improvement was also noticed with the MDLT, but PM emissions remained 10 percent below the CVS. The BG-2 was not equipped with a look-ahead function and instead used a pre-production transient dilution air controller (TDAC) to improve the PM results, but only for a limited number of test points. Further testing with the BG-2 equipped with TDAC is required.
This project was a highly focused research program where modifications and improvements to PFSS units were implemented as needed. While it served to identify the improvements that could be made to PFSS steady-state and transient PM emission measurement, further investigation is required to improve consistency of total PM measurement in the presence of volatile material and to improve PFSS ability to handle real time transient emission testing without relying on a “look-ahead” function. In addition, this work indicates that a step-by-step validation criteria are needed to qualify a variety of PFSS units as valid systems to be used for steady-state and transient PM emission measurement.
An important conclusion of the work is that it does show that some PFSS units can become CVS substitutes for transient PM measurements. Additional refinements, however, are needed but should be achievable through more work.