It is well known that when a Runge-Kutta method is used for the time integration of the ordinary differential equations coming from the spatial discretization of partial differential equations, the classical order of the Runge-Kutta method is not attained. More precisely, in [4] it is shown that the order of convergence is goberned by the so called stage order of the Runge-Kutta method. In [2], a strategy to avoid this order reduction for linear initial boundary value problems has been proposed. The basic idea is to decompose the exact solution of the problem as the sum of two terms. The first one is computable directly in terms of the data and the second one is the solution of a suitable initial value problem which does not suffer from order reduction. In this poster we show the applicability of such a technique to deal with nonlinear problems when the time-integrator is the third order linearly implicit Runge-Kutta method of [1]. The main difficulty is that in the nonlinear case the new initial value problem explicitly depends on the exact solution of the original one, while this is not the case for linear equations. However, the use of a linearly implicit Runge-Kutta method allows us to have approximations to both solutions at a reasonable computational effort for some interesting test problems [1], [3].