Combinatorial assignment problems involving finite resource allocation have traditionally been studied in static, single-instance settings. This thesis introduces SEQUCA, a sequential framework that extends a state-of-the-art Utilitarian Combinatorial Assignment (UCA) algorithm to dynamic environments. The framework evaluates social welfare across simulated time steps, accommodating stochastic task generation and fluctuating agent availability. Two implementations are compared: a baseline model that reapplies the UCA algorithm at each step, and a greedy extension that caches and reuses prior results when applicable. The results show that the greedy extension achieves comparable utility, with only minor reductions, while significantly improving execution time. Furthermore, the study identifies a performance bottleneck in the UCA algorithm when utility values across coalitions are identical, which reduces pruning effectiveness