1: \begin{abstract}

2: Simulation of geothermal systems is challenging due to coupled physical processes in highly

3: heterogeneous media. Combining the exponential  Rosenbrock--Euler

4: and Rosenbrock-type methods with control-volume (two-point flux approximation) space

5: discretizations leads to efficient numerical techniques for simulating geothermal systems. In terms of efficiency

6: and accuracy, the exponential Rosenbrock--Euler  time integrator has advantages over standard time-dicretization

7: schemes, which suffer from time-step restrictions or excessive numerical diffusion when advection

8: processes are dominating. Based on linearization of the equation at each time step, we make use of 

9: matrix exponentials of the Jacobian from the spatial discretization, which %

10: provide the exact solution in time for the

11: linearized equations. This is at the expense of

12: computing the  matrix exponentials of the stiff

13: Jacobian matrix,

14: together with propagating a linearized system.

15: However, using a Krylov subspace or Leja points techniques make

16: these computations efficient.%, and it can also be implemented without explicit computation of the Jacobian. 

17: 

18: The Rosenbrock-type methods  use the appropriate rational functions of the Jacobian 

19: from the spatial discretization. The parameters in these schemes

20: are found in consistency with the  required order of convergence in time.

21: As a result, these schemes are A-stable 

22: and only a few linear systems are solved at each time step.

23:  The efficiency of the methods compared

24: to standard time-discretization techniques are

25: demonstrated in numerical examples.

26: \end{abstract}

27: