Weakly nonlinear asymptotics of the \(k-\theta\) model of cellular flames: the Q-S equation (Q2387127)

The problem of flame propagation and mathematical study of the flame front goes back to the 1960's work of Zeldovich and the intense activities at North-Western University of Bernard J. Matkowski and his colleagues, which produced many papers and a widely accepted monograph. In 1983 G. I. Sivashinsky initiated the study of diffusive instability of flames causing formation of unstable cellular structures. The difficulties of producing a mathematical model are clear. Here we are dealing simultaneously with a complex, temperature sensitive, multi-step chemical process, with a fluid flow of such complex gas mixture, combined with simultaneous radiation and convection heat exchange. Before discussing different models, let us list what is in the authors' experience important, and also list symbolism used here. \(T\) is the temperature of unburned gas, \(T_{ad}\) is adiabatic temperature of unburned gas, \(\sigma= T_u/T_{ad}\) while, \(T_a\) is the activation temperature. The crucial variable in authors' theory is the scaled temperature: \(\theta=(T- T_u)/(T- T_{ad})\). \(C\) is the concentration of a deficient reactant, \(C_u\) is value of \(C\) in the unburned gas, \(Y= C/C_u\). Also \(U\) denotes the velocity of undisturbed flame restricted to 2-dimensions, and \(D_{th}\) denotes thermal diffusivity. As in all problems of fluid flow it has been discovered that some dimensionless numbers predict important modes of behavior of the system. For example: the Reynolds number predicts the onset of turbulence. Here, we have the Lewis number \({\mathcal L}e\), which is the ratio of thermal and molecular diffusivity, while \(\beta= (1-\sigma)T_a/T_d\) is called the Zeldovich number. Instability of the process occurs when the Lewis number \({\mathcal L}e\) falls below a certain value. The set of corresponding 2-spatial dimensional state equations is: \[ \theta_t= \theta_{xx}+ \theta_{yy}+ \Omega(Y,\theta), \qquad Y_t={\mathcal L}e^{-1}(Y_{xx}+ Y_{yy})- \Omega(Y,\theta), \] where \(\Omega= {1\over 2}{\mathcal L}e^{-1}\beta^2 Y\exp[\beta(\theta- 1)/(\sigma+ (1-\sigma)\theta)]\). Strangely, close to instability the flame structure becomes almost steady and planar. After a change of units it is described by a simple geometrically invariant fourth order PDE: \(V_n= 1+ (\alpha- 1)k+ k_{ss}\). Here \(V_n\) is normal velocity of the flame interface, \(s\) the arc length along the (curved) interface surface, \(k\) is the curvature, and \(\alpha= {1\over 2}\beta(1-{\mathcal L}e)\). However, this is not the only model available, and linear analysis in the adiabatic case produces the system \(V_n= 1+ k+\theta\), \(\theta_{yy}- \theta=\alpha k\). This is known as the quasi-steady \(k-\theta\) model. Following this line of argument the authors obtain the following model called quasilinear using a coordinate \(\Phi(x,t)= t+y\) for flame propagating along the \(y\)-axis: \[ \Phi_t+\textstyle{{1\over 2}} \Phi^2_x= \Phi_{xx}- \theta,\quad \theta_{xx}+ \alpha\Phi_{xx}- \theta= 0. \] They rewrite this system of equations as a single equation and proceed to examine it, while establishing links with the K-S (Kuramoto-Sivashinsky) model of a fourth-order PDE: \(U_\tau+{1\over 2}(U_\xi)^2+ U_{\xi\xi}+ U_{\xi\xi\xi\xi}= 0\). Here a change of scale relates the \((\xi-\tau)\) coordinates to the \((x-t)\) coordinates. For \(\varepsilon<<1\) we have \(\tau= t\varepsilon^2\), \(\xi= x\varepsilon^{1/2}\). Thus the anticipated solution \(\Phi(x,t)\) is of the order \(\varepsilon U(\xi,\tau)\). The authors show that in the \(H^s\) Sobolev space the weak solutions of their system are asymptotically close to those of the K-S model. This paper should be of interest to applied mathematicians, chemists and engineers studying the propagation of flame, and particularly the dynamics of flame's boundary.