Hello Amirhossein,
We generally do not recommend using the fully rotated crack model (i.e., "Fixed crack" parameter set to 0.0) as it is not really realistic. In reality, once the crack localises, its direction stays more or less the same.
On the other hand, during the crack propagation/growth, its direction can change a bit as the direction of principal stress changes. This is the case especially when larger finite elements are used in the model. So, when using this reasoning, it can make sense to set the "Fixed crack" parameter to, for instance, 0.7, meaning that the crack can rotate until the tensile softening reaches 0.7 of the initial tensile strength. After that, the crack direction is fixed.
Does this make sense to you?