In recent years, microfluidic biochips have been widely applied in various fields of human society. The emergence of distributed channel-storage architecture enables fluid to be directly cached within the flow channels without being transported to the dedicated storage. However, the volume of the flow channels is limited, it is important to carefully consider the fluid volume to be cached to ensure it can be accommodated within the appropriate flow channels. In this paper, we propose a high-level synthesis method considering actual volume management and channel storage through integer linear programming, providing a rational analysis of fluid caching and the temporal relationships between different fluid-handling tasks. Experimental results conducted on multiple benchmark tests demonstrate the effectiveness of the proposed method in reducing the completion time of bioassay, the number of caches, the maximum fluid caching volume, and the number of fluid-handling tasks.