Liquid-handling instruments are indispensable tools in modern biomedical laboratories, streamlining compound and sample management tasks with precision and efficiency. Compound dispensing from large chemical libraries divided over hundreds of microwell plates can require substantial swapping of plates, particularly if compounds from multiple source plates are to be dispensed in each destination plate. Despite robotisation, plate swapping is a time-consuming necessity for high-throughput experiments, posing a significant bottleneck. In this paper, we explore the application of constraint programming (CP) to the planning of liquid-handling tasks to minimise plate swaps in automated dispensing. We formulate the problem as a combination of a set partitioning problem and the construction of a bipartite network. We present six CP models implemented in MiniZinc and evaluate their performance on synthetic benchmarks using three state-of-the-art constraint solvers. This work highlights the potential of CP to enhance the scalability and efficiency of automated compound transfer systems.