Question about Cantera's h2o2.cti mechanism

[Mike Hansen]

Cantera users & developers,

I have been using Cantera-MATLAB to study auto-ignition of hydrogen-oxygen mixtures. I have noticed a significant disparity in the ignition delay when using the provided h2o2.cti mechanism and GRI 3.0. I initially thought that the h2o2 mechanism is simply gri30 with carbon chemistry removed. However I think this disparity disproves this - could someone provide insight and/or a citation for the h2o2 mechanism? I've been scouring documentation and the internet for days without gaining much ground.

Some context for the above: This occurs when I run 1 atm batch reactors with stoichometric H2/O2 near the autoignition temperature (~850 K). The further I get from the autoignition temperature, the closer the gri30 and h2o2 results become. The ignition delays are nearly identical when I run at 1400 K. This occurs with the batch reactor utility in Cantera as well as my custom solvers that use Cantera to compute RHS terms. In some cases the ignition delay of GRI 3.0 is ~80x that of h2o2.cti.

One final bit of info: when I use the O'Connaire 2004 mechanism, highly-optimized for H2/O2, provided by LLNL I get results that nearly match those of h2o2.cti. Ignition delay differs by about a factor of two.

Thanks in advance!

Mike Hansen

Chem. Eng. PhD student, University of Utah

Graduate Intern, Sandia National Labs

[Ray Speth]

Hi Mike,

It looks like h2o2.cti is missing the following reaction from GRI 3.0:

#  Reaction 33
three_body_reaction( "H + O2 + M <=> HO2 + M",   [2.80000E+18, -0.86, 0],
         efficiencies = " AR:0  C2H6:1.5  CO:0.75  CO2:1.5  H2O:0  N2:0  O2:0 ")

which if you remove the species not present in the h2o2.cti mechanism, would be written as:

three_body_reaction( "H + O2 + M <=> HO2 + M",   [2.80000E+18, -0.86, 0],
         efficiencies = " AR:0  H2O:0  O2:0 ")

This certainly looks like an error in h2o2.cti, which has been there since at least 2003, and probably as long as h2o2.cti has existed.

It looks like other H2/O2 mechanisms (e.g. O'Connaire et al. 2004 and Li et al. 2011) parameterize this reaction as a Troe falloff reaction. I would recommend against using GRI 3.0 for simulating hydrogen combustion, as nearly all of the experiments that it was validated against were for CH4 combustion.

Regards,

Ray

[Mike Hansen]

Ray,

Thank you for the detailed answer!

I know not to use mechanisms on problems for which they weren't optimized; I just felt like checking my assumption that h2o2.cti was the submechanism of gri30.

Mike