7f58282790b0523b.tex
1: \begin{abstract}
2: We present a high-resolution convergence study of detonation initiated
3: by a temperature gradient in a stoichiometric hydrogen--oxygen mixture
4: using the {\sc Pencil Code} and compare with a code that employs
5: a fifth order weighted essentially non-oscillating (WENO) scheme.
6: With Mach numbers reaching $10$--$30$, a certain amount of shock viscosity
7: is needed in the {\sc Pencil Code} to remove or reduce numerical pressure
8: oscillations on the grid scale at the position of the shock.
9: Detonation is found to occur for intermediate values of the shock viscosity
10: parameter.
11: At fixed values of this parameter, the numerical error associated with
12: those small wiggles in the pressure profile is found to decrease with
13: decreasing mesh width $\delta x$ like $\delta x^{-1.4}$
14: down to $\delta x=0.2\um$.
15: With the WENO scheme, solutions are smooth at $\delta x=10\um$, but
16: no detonation is obtained for $\delta x=5\um$.
17: This is argued to be an artifact of a decoupling between pressure
18: and reaction fronts.
19: \begin{keywords}
20: Combustion; numerical methods; detonation; shock waves; chemical reaction.
21: \end{keywords}
22: \end{abstract}
23: