The number and complexity of industrial wind turbine installations have increased significantly over the last decades. As maintenance costs are high and down-times lead to substantial revenue loss, increasing the reliability and optimizing the maintenance process are crucial tasks from an industrial perspective. However, many of the proposed diagnosis systems merely focus on parts of the turbine or only locate a portion of the faults. Model- based diagnosis has been applied successfully in industrial settings and further provides a solid theoretical background. Therefore, we propose a model-based approach depending on automatically retrieved health variables and on an extensive expert knowledge on specific component-oriented failure modes as well as their effects on measurable signals. As the expert assessment provides causal links between faults and their manifestations, we formally create a Propositional Horn Clause Abduction Problem (PHCAP). In this paper, we present a modeling concept taking advantage of existing expert knowledge and show how it can be used for wind turbine diagnosis employing already existing algorithms and structures. Our models enable us to directly determine root causes from the links between malfunctions to observable turbine signals on a system level with a relatively low effort compared to other approaches.