Numerical Designs of Experiments (NDoE) are used in a product development process to optimize the product. A NDoE may combine a costly numerical model and numerous experiments. The NDoE process consequently becomes very expensive. However, some methods and algorithms were developed to shorten the NDoE process, as sensitivity analysis, surrogate modelling and adaptive DoE. Because of their complexity, advanced expert knowledge or a long preparation step is required to optimally choose and configure all of these methods, in order to run the most efficient NDoE process. To answer this issue, a knowledge management approach is proposed in this paper. It capitalizes and reuses knowledge about NDoE process. This solution is proposed because of the lack in term of models and standardized processes for this specific NDoE application. An ontology was developed to manage, share and reuse knowledge and enable queries for information retrieval in a database. The database lists every NDoE processes executed. Then, the knowledge is analysed by a decision-support system to help designers to choose the best configuration.