Increasing complexity and shorter innovation cycles require a rethink in the development of mechatronic products. A function-oriented development process using model-based systems engineering (MBSE) methods promises to identify technical solutions via solution-neutral functional architectures. Since each function can typically be realized by multiple, alternative solutions the combinatorial number of overall system solutions explodes. Existing MBSE approaches enable to establish functional architectures as well as support the verification of solutions. However, evaluating different solution combinations with the current state of research involves repetitive and time-consuming modeling which leads to a heuristic selection of potentially suboptimal solutions. In this paper, we present a method for the systematic and function-oriented composition and model-based evaluation of solution combinations. Alternative solutions are structured based on a functional architecture, so that each possible solution combination can be composed and simulated with a minimum of effort, considering physical interactions and the efficiency. Based on simulation results, solution combinations can be verified against requirements and evaluated striving for the best solution combination. The elaborated method extends an existing MBSE approach and is illustrated by the running example of an electrical coolant pump of a passenger vehicle.