Friction in control valves and inadequate controller tuning are two of the major sources of control loop performance degradation. As friction models are needed to diagnose abnormal valve operation or to compensate such undesirable effects, process models play an essential role in controller design. This paper extends existing optimization-based methods that jointly identify the process and friction model parameters, so that a nonlinear process model structure is considered. The procedure is based on data generated from closed-loop experiments with an external test signal. A simulation example indicates that the method accurately quantifies the valve friction, the process dynamic and the nonlinear steady state characteristics, even when the system is subjected to different level of disturbances.