Direct numerical simulation of a turbulent hydrogen-air premixed plane jet flame is performed to investigate fractal characteristics and to evaluate the fractal dynamic subgrid scale (FDSGS) combustion model. The DNS results show that the fractal dimension of flame surfaces increases with the downstream distance, and the fractal dimension computed using a 3D box-counting method reaches about 2.54 in the region where turbulence is developed by mean shear. An inner cutoff representation employed in the FDSGS combustion model could be used in large eddy simulations (LES) of complicated combustion problems. Static tests show that the procedure applied in the FDSGS combustion model adequately predicts the fractal dimension, Kolmogorov length scale, and flame surface area despite the presence of strong mean shear. Dynamic model evaluations are also carried out by conducting a series of LES using the FDSGS and other combustion models. In the dynamic tests, the mean temperature distributions and peak positions of variations of a reaction progress variable fluctuation in the transverse direction obtained from the LES with the FDSGS combustion model show good agreement with the filtered DNS fields. The present evaluation also revealed that one of the strengths in the FDSGS combustion modeling approach is that the model does not require SGS turbulent velocity fluctuation, since modeling of this quantity is straightforward for neither homogeneous turbulence nor the turbulent shear flows.