What is GRI-Mech?
How to use GRI-Mech
Please tell us what you learn, and about your problems
Some cautionary notes


The results reported on these Web pages are products of computational and experimental research sponsored by the Gas Research Institute. It has been carried out at The University of California at Berkeley, Stanford University, The University of Texas at Austin, and SRI International.

GRI-Mech is an optimized detailed chemical reaction mechanism capable of the best representation of natural gas flames and ignition that we are able to provide at this time. Our ongoing program comprises development of extensions to include more chemistry, more robust parameterizations, and additional tools and documentation to make working with GRI-Mech more effective for everybody.

In order to use the input files directly you need the Sandia National Laboratory Chemkin-II programs. Ignition profiles calculated with the GRI-Mech reaction mechanism and thermochemical data should be independent of the program used to compute them; noticeable small differences in flame profiles should be expected, however, if you are using a flame code other than Chemkin, because the transport calculation, the numerical method of solving the differential equations, and so on, differ from one flame modeling program to another.

More details are available on the optimization process.

Before telling you more we are obliged to say:

LEGAL NOTICE These files, both the ones intended for use as computer input as well as those comprising documentation, were prepared by The University of California, Berkeley, Stanford University, The University of Texas at Austin, and SRI International as a result of work sponsored by the Gas Research Institute (GRI). Neither GRI, members of GRI, nor any person acting on behalf of either:

  1. Makes any warranty or representation, express or implied, with respect to the accuracy, completeness, or usefulness of the information contained in these files, or that the use of any data, method, or process disclosed in these files may not infringe privately owned rights; or
  2. Assumes any liability with respect to the use of, or for damages resulting from the use of, any information, data, method or process disclosed in these files.


What follows is a very brief overview of GRI-Mech. A paper document giving a more detailed account of the way GRI-Mech is developed and maintained is available. Send us an e-mail or regular mail request to get a copy of this document.

GRI-Mech is essentially a list of elementary chemical reactions and associated rate constant expressions. Most of the reactions listed have been studied one way or another in the laboratory, and so the rate constant parameters mostly have more or less direct measurements behind them. In order to get successive releases out in a timely manner we do not continually prepare comprehensive hard copy documents explaining the sources of information that were consulted and evaluated, the rate constant theory that went into constructing the set, and the details of each step of the optimization process. We do update the documentation provided on this Web site with such information regularly. You are welcome to make whatever use you like of this information and have our permission to cite this Web documentation if you so desire. (But see the foregoing disclaimer.)

Once we have a starting set of rate constant parameters, we undertake extensive sensitivity tests on a variety of experimental data related to natural gas ignition and flames. These tests tell us which rate parameters should be considered closely to 'tune' the set so as to get an optimum representation of the data. We then go through a long process of simultaneous parameter optimization, most of it done automatically---i.e., by a computer adjusting the parameters rather than one of us---to get the parameter set for each successive release. This is carried out with strict constraints to keep the rate parameters being optimized within predetermined bounds that we set on the basis of evaluations of the uncertainties in measurements of the rates of elementary reactions and by applications of conventional reaction rate theory. Once an optimal parameter set is found, it is checked against the literature in an extensive 'validation round'.

In addition to reviewing the literature rate constant data we also examine the thermochemistry of the free radicals in the mechanism. Some of the thermochemical parameters in GRI-Mech differ from those in the Sandia data base. For this reason we provide a file of the complete GRI-Mech thermochemistry, represented as the coefficients of NASA polynomials.

Our mechanism files are specifically formatted for use with the Chemkin modeling programs. For the flame calculations in the optimization and validation process we use the Sandia transport package and database. (R.J. Kee, J. Warnatz and J. A. Miller, Sandia Report SAND83-8209) (Note: The symbol CH2* used for singlet methylene in GRI-Mech 1.1, was changed to CH2(S), as in the original Sandia notation, in later releases.)

More about the optimization process.


The easiest way to learn how to set up GRI-Mech for combustion modeling is to just do it! If you have Chemkin up and running, and have tested it using the sample files provided with the Chemkin distribution, all you need to do is substitute the GRI-Mech input files for the Chemkin samples. That's it! Whatever works for you to make the Chemkin sample files run will also run the GRI-Mech files.


The authors continue to expand their understanding of how GRI-Mech works in their own labs as well as from others who may see things that should be done from very different perspectives. Our project has benefitted over and over again from comments by GRI-Mech users; we look forward to more of them!

It makes us feel good to hear about successful applications of GRI-Mech, but it helps us more in our development work, and consequently all users of later releases, to hear about failures. We welcome suggestions of any kind.

Please be as specific as you can in telling us about your results and your problems. We will be happy to include the results of your tests, with appropriate citation to you, in our Web and printed materials. Just ask.

We try to keep our list of interested parties up to date so that we can quickly communicate changes, problems, and mechanism updates. Since our ftp daemon only records logins, not actual user names, we ask that people who set out to use GRI-Mech let us know their e-mail addresses. Please mail to


We want to warn you about several important aspects of GRI-Mech.


Or if you MUST so, be very careful. GRI-Mech has been optimized as a whole, and should be used just as you get it if you want to exploit its ability to model natural gas combustion. You will not surpass the overall performance we get for natural gas combustion by adjusting any 'sensitive' reaction rate constant expressions. Any substitution of 'better' rate constant expressions or thermochemical or transport coefficient values, or outright removal of species or reactions, will put you at risk of getting significantly deteriorated performance of the mechanism when tested against the available spectrum of natural gas combustion data.

We recognize that GRI-Mech users are human and will be tweaking away on this or that aspect of our product anyhow, or making major changes for specific purposes, like doing sensitivity analyses of one kind or another. Keep in mind in doing so that we do not claim that GRI-Mech is a suitable starting point for mechanism building by patchwork means. You can use it that way if you like, but we cannot predict what the consequences may be.

Numbers of species and reactions

The list of reactions and species in GRI-Mech contains numerous entries that are 'unimportant' for natural gas combustion.

There are several reasons why we have them there anyway. One is that there are special purposes (like models of flame radiation or ionization) where elementary reactions that are otherwise negligible become very important, and we want to have these reactions on hand for occasions where somebody is checking these aspects. A second reason is that the combustion of some other fuels (methanol, acetylene, ...) can be modeled using GRI-Mech as a subset, with the knowledge that the part of the mechanism relevant to natural gas has been optimized in the manner described above. There are techniques for shortening reaction and species lists, which produce what are called reduced mechanisms. We invite others to 'submit' their reaction mechanisms reduced from GRI-Mech for inclusion on this Web document.

Back reactions

We treat all reactions as reversible even though it is clear that negligible reverse flux will ever occur in many of them. If your modeling program requires explicit inclusion of reverse reactions, GRI-Mech as presented in this directory will require additional calculations to find out which of them are really needed for your application.

Computer time

The numerous species and reactions in GRI-Mech which really do not need to be included for modeling natural gas combustion increase the demand on computer resources for doing the chemical part of the model by a large factor. We accept this in order to avoid coping with the numerous problems that arise in streamlining such computations.

Computer time has not been a problem for us even when using GRI-Mech on relatively small workstations. If you are using an older PC, then you may encounter longer integration times than you want to live with. We are thinking about ways to deal with this and will welcome suggestions about how this could best be done.

GRI-Mech main menu