The mixing of turbulent and laminar flow inside a rectangular channel T-junction is investigated using experimental and numerical methods. A non-intrusive, optical measurement technique (Laser Doppler Velocimetry) is applied to gather point-wise velocity information of the flow. This information is used to calculate the turbulent kinetic energy k, a measure of velocity fluctuations. Computational Fluid Dynamics (CFD), allows the detailed investigation of mixing flows based on finite volume method. However, because of the complexity of turbulent flow, economical simulations are restricted to the use of turbulence models to account for the velocity fluctuations. Comparison of measurements and simulation indicates that while velocities are predicted correctly, the simulation severely underpredicts the turbulent kinetic energy. Aim of this work is to use the obtained measurement data to calibrate the turbulence model, reducing the error in the prediction of k, and therefore enable a better simulation of the mixing fluid flow. This is achieved by modifying the default values of eight model constants present in the equations of the turbulence model. In order to reduce the computational effort a statistical method, Design of Experiments (DoE), is applied to identify the best possible value combinations. Using this method, a stationary point is predicted, reducing the error between measured and simulated turbulent kinetic energy in the turbulent flow from initially 90 % to 45 %.