Sixth DOE Crosscut Workshop
on Lean Emissions Reduction Simulation
September 23rd and 24th, 2003
GM R&D Center, Warren, Michigan

Agenda | Overview | Abstracts | Summary | Presentations | Discussion

To read/view presentations' summaries/abstracts click on abstracts link or go to agenda and click on presentation's title

---

The two day workshop was held at GM Research. Presentations are to be posted on the CLEERS web site in the near future, http://www.cleers.org/.

CLEERS = Crosscut Lean Exhaust Emissions Reduction Simulation. It is a subgroup of the DOE Diesel Crosscut Team. The goal is to promote lean exhaust emission control simulation. Subgroups or Focus Teams have been formed for LNT, DPF, and SCR modeling. CLEERS can help coordinate and expedite modeling efforts, and provides a mechanism for industry feedback to DOE. Participants include Ford, GM, DC, DDC, Caterpillar, Cummins, International, Mack/Volvo, JMI, Northwestern University, University of Wisconsin, Wayne State University, EPA, TACOM, ORNL, PNNL, LLNL.

This is the second workshop since forming the LNT Focus Group, and the first since forming the DPF and SCR groups. The LNT and DPF groups have been having monthly teleconferences; the SCR group is just getting organized.

Presentations were made by a number of people. Several themes emerge:

1. The data needs on LNT and DPF performance to permit modeling need to be defined in a standardized way, so that catalyst suppliers can run one set of data to meet everybody's needs. The data format needs to be efficient and easy to run, and complete enough to cover the needs.
2. Many different models will be developed by the various participants to meet their own needs. This coordination can help reduce overlap and repetition. Models range from 0 dimensional to 3-D.
3. In complex, dynamic systems like LNT and DPF, models are needed both for system design and optimization, and for embedded model based controls. Of course, different model types are appropriate for the different uses.
4. Basic models of kinetics, mass transfer and so on will be an important part of developing the understanding of how these catalysts behave. As such, they will contribute to development of improved catalyst formulations in the long run.

Fundamental kinetic models are beginning to be published for LNTs. However, they are not very mature. Basic reaction mechanisms are still not certain, and many rate constants and the like are not known. There isn't even a good base of published data against which models can be evaluated.

ORNL presented DRIFTS analysis of model LNT catalyst work. This is a start toward identifying surface intermediates and thus reaction mechanisms. LLNL has begun some work with ab initio calculations of NOx storage on BaO.

There are five chemical steps involved in LNT NOx storage (more for SOx issues!):

1. NO to NO2
2. NO2 storage (thought to be NO2 only) on BaO, BaCO3 or similar sorbents
3. Reductant evolution - depending on reductant source; H2, CO, HC
4. NOx release - thermal and reductant driven
5. Reduction NOx to N2

The presence of CO2 and water can strongly affect these steps. Mixture speciation is complex, and important.

The speciation of engine exhaust arriving at the LNT can be changed quite a lot by calibration effects such as post injection timing and EGR. There can be a lot of aldehydes, and up to 8% CO/4% H2 is possible.

Some research on a model LNT showed that aging reduced surface area of both Pt and sorbate, but the damage levels out after initial degradation. P and Zn poisons are present in significant quantities although their importance is not certain. They did not see migration of Pt or sorbate into the cordierite washcoat.

Cummins tested integrated LNT/DPF (4-Way) systems based on cordierite or on a fiber based filter. At equal washcoat loading the performance was similar, although the fiber was better at 200C due to lower thermal inertia. The fiber filter could hold 50% more washcoat without excessive backpressure, and this improved LNT performance.

Wayne State University (Henein) has done extensive measurements of emissions and flows from a DIATA engine varying injection timing, EGR, injection pressure, and swirl ratio (but not post injection). This comprehensive data set is useful to help match engine and aftertreatment.

DPF models are more complex than flow through models because

1. Flow through the walls must be included
2. Dynamics of filter cake are not understood, and are very important
3. Chemistry of soot oxidation are complex and not well defined.

Some key issues are

• Particle morphology and oxidation characteristics
• Soot distribution and impact on Tmax
• Ash creation and composition and transport
• Gas emissions during regeneration
• Local properties of soot cake - permeability, density etc

Cummins (Yezerets) has developed a clever method of measuring soot oxidation rates. Their data shows that the kinetics are faster initially than steady state, apparently due to pre-exposure of soot to air in the experimental method. When that effect is removed, the soot oxidation in air is well described by a simple Arrenhius function. Soot from different sources has significantly different kinetic behavior.

The presentations were followed by general discussion and wrap-up.

Back to Top

---

Agenda | Overview | Abstracts | Summary | Presentations | Discussion