Affect dynamics in digital learning systems (DLS) research investigates how students' emotions evolve while using these systems. This information can reveal the most common emotions experienced by students and how they transition between them. Recent studies indicate that affect dynamics in DLS vary depending on students' demographic and personal characteristics, such as nationality. However, most research has focused on students from the United States and the Philippines, leaving a gap in understanding Latin American learners and the potential differences in affect dynamics based on gender and emotion duration. This study introduces an affect dynamics model for Brazilian students using a Math DLS in a school setting, encompassing emotions like confusion, frustration, boredom, and engagement. We also present evidence that affect dynamics models differ based on students' gender and emotion duration. Analyzing affect dynamics across cultures allows for better understanding and generalization of the resulting models while also demonstrating the influence of emotion duration and learners' gender on affect dynamics. This knowledge enables tailoring more adaptive interventions for students' emotions in DLS.