Given n points continuously moving in 2d space, we want to maintain their convex hull each time. In this paper, first we use the kinetic data structure (KDS) framework and define a new KDS named Spiral. After that, we turn to both theoretical and experimental evaluations of our KDS; in theoretical evaluation, consider the four quality measures of kinetic data structures in the worst case, but in the experimental evaluation results of a simulation program are used to estimate the average case. We suggest an alternative efficiency parameter instead of previously defined and define a new responsiveness measure for the average case. The experimental factors are much better than the theoretical worst case (this is especially true for the efficiency parameter; log2n instead of n.); hence conclude that we can’t reject a KDS with rather large theoretical worst case parameters. However, this study shows that when working with random positions, the worst cases used to evaluate a KDS aren’t always sufficient because these are rarely occurred and the expected average is so much better. In the worst case, from point of view of responsiveness, efficiency, locality and compactness our KDS belongs to O(n), O(n), O(C) and O(n) respectively (C is a constant number) and in the average case, these parameters for our KDS are O(log n), O(log2 n), O(C) and O(n) respectively. Note that previously, the two last parameters was O (log n) and O (n log n).