This paper presents a new method to describe, analyse and estimate production system performances. According to current literature, the basic performance measures are WIP (Work-in-Process), lead time and throughput. These are also in this article used as performance measures. Continual trade-offs between different performances are necessary to stay competitive in today's market. Different performances are archived by adjusting system parameters. Knowledge about relations between system parameters and system performances in existing systems, and about system performances of not yet implemented system alternatives, is essential for business. Queuing theory and simulation can estimate system performances of not yet implemented systems and help the decision makers, but when the complexity increases queuing theory becomes very difficult and heavy to use. A single simulation presents limited information. Multiple simulations are necessary to ensure that the best alternative is chosen. A high number of simulations demand a lot of computer time and resources. Reduction of runs is desirable even with cheaper computer equipment. Currently, traditional two-dimensional charts are the only tools to present and analyse system performances. This article presents a new surrogate model for easier estimation and presentation of system performances, their internal relations, and relations to the system parameters. With the new surrogate model, system performances based on simulations are presented as positions in a three dimensional environment. Parametric curves and surfaces of Bezier type are generated and adapted to these positions. System performances of other system alternatives are then estimated. The number of simulation calculations can thereby be moderated. The method is illustrated with a small production line system.