The 1998 Combustion group of the CTR summer program gathered more than twenty scientists to work on seven projects. Summer combustion programs are evolving rapidly from one year to another, exploring new fields of research in many cases. In 1998, new tools were developed and tested while new applications fields for existing tools were opened. The main new tool studied by three groups in 1998 was Large Eddy Simulation (LES) for reacting flows. Even though it is obvious that LES will be an important tool to both study combustion on a fundamental level and address practical applications in the near future, the status of LES for combustion is still far from mature. Multiple questions linked to the fundamentals of LES in reacting flows and to the practical feasibility of LES for such flows (in comparison with existing Reynolds-averaged formulations) remain open. During this program, the fundamental aspects of LES for combustion were studied for premixed, partially premixed, and diffusion flames. Cook and Bushe investigated one of the building blocks of all LES models for diffusion flames: the modeling of the scalar dissipation which measures the rate at which fuel and oxidizer are mixed by turbulence. Using existing CTR DNS data, they analyzed the validity of existing models. Trouvé and Vervisch also addressed a central problem for LES of diffusion flames: most LES of such flows are performed using infinitely fast chemistry assumptions which lead to nonphysical results in many cases. Relaxing this assumption is a necessary but complex task. Using DNS data, Trouvé and Vervisch developed a description of the ignition zones of turbulent diffusion flames and proposed modeling approaches for such simulations. In the field of premixed and partially premixed flames, Angelberger, Poinsot, Veynante, and Egolfopoulos focused on the development of simplified realistic chemistry and its coupling with LES. A thickened flame model based on DNS of flame vortex interactions was used to describe flame/turbulence interaction, and the final LES tool was shown to be efficient in computing flame transfer functions in combustion instabilities. The effect of pulsating equivalence ratio in lean flames was also investigated. New topics were also studied using DNS codes coupled to particle solvers: Réveillon and Vervisch studied the importance of fuel vaporization on the variance of fuel mass fraction, which is a crucial quantity for turbulent combustion models. They showed that, in addition to fuel vapor, vaporizing droplets were also creating high levels of fuel vapor variance, which must be included in LES or RANS models. Along the same lines of research (coupling DNS solvers for gas with particle tracking codes), Smith, Oefelein, Ruetsch, and Ferziger studied the formation of reactive particles in a turbulent flow with a specific emphasis on soot formation, which is a key problem in multiple diffusion burners. New practical questions were also addressed using existing tools: for example, the NTMIX CHEMKIN code, which can perform DNS with complex chemistry and transport, was used by Haworth, Cuenot, Poinsot, and Blint to study flame propagation in direct injection engines. In these engines, gasoline is injected directly inside the combustion chamber so that the flame propagates into a highly stratified mixture, leading to multiple new fundamental challenges for combustion research. Haworth et al. performed the first DNS of such propagation phenomena using a 29 species chemical scheme for propane air flames. Finally, totally new fields for CTR were studied this year: DNS tools developed for hydrocarbon flames at CTR were adapted by Niemeyer, Bushe, and Ruetsch to investigate flame propagation in thermonuclear flames. These flames have many common features with flames studied in the combustion community even though the parameter range and the chemical mechanisms differ by orders of magnitude. Using CTR DNS tools for such flames led to new insights into their physics and to an efficient interaction between two communities which do not meet often. Being able to introduce more realistic chemistry into DNS and LES is one necessary ingredient of CFD for combustion. Interestingly, the collaboration between chemists and CFD experts was very intensive and fruitful in 1998: Bowman, Blint, and Egolfopoulos proposed and modified chemical schemes in direct interaction with DNS or LES scientists to find the best compromises. This iterative procedure demonstrated that the stiffness of these schemes may be decreased significantly while preserving accuracy simply by promoting an efficient interaction between chemistry experts and DNS/LES users.