Go with the flow
Better modelling of selective catalytic reduction will help to lower NOx emissions from diesel engines
- Published in Features.
Good mixer: Turbulence models are an essential element of the analysis
The next stages of emissions legislation in Europe, the US and Japan will place much tighter limits on NOx. Selective catalytic reduction (SCR) systems will be one of the key technologies employed to meet the standards.
SCR system development is very challenging. Ensuring that the right quantities of urea solution are injected, and at the right times, isn’t easy.
Nor is ensuring that all of the reactants are properly mixed before they reach the catalyst. If not, there won’t be a uniform distribution of ammonia – the reducing agent – over the surface if the substrate and NOx conversion efficiency drops.
Simulation is the best place to start, but SCR is a complex process involving thermal, fluid and structural analysis as well as modelling of the chemical reactions. A dedicated tool would help. CAE software firm Ansys started looking at this in 2004 after noticing Tier Ones and OEMs asking for better SCR simulations at symposia and conferences.
“Customers can use simulation to optimise designs and arrive at the best solution quickly – what angle the injector should be, where the mixer should be located and what type of mixer,” says Padmesh Mandloi, senior technology specialist at Ansys.
“Each of these variables could cause significant differences in the uniformity values.”
The physics behind SCR systems is complex. The starting point is the injection of the urea solution into the exhaust pipe.
High exhaust gas temperatures cause it to decompose into urea and water. This is modelled using computational fluid dynamics (CFD) as multi-component particles.
“Our existing tools had the capability to model this. However you still had to come up with a user-defined function to implement reaction rates – we added sub-routines to do this,” says Mandloi. “And the decomposition phase must be accurately modelled, and for this you need empirical inputs – generic CFD solvers alone would not be able to do this.” Then there’s mixing.
Increasingly, system suppliers are using stamped sheet-metal plates to increase turbulence, improving the process. But this makes the model more complex.
Compounding these issues is the fact that the plate designs are becoming more intricate to improve mixing still further. So it’s important that the mesh in this part of the model is fine enough and that the CFD tool can solve highly turbulent flows with sufficient accuracy.
