In the current technological uprise, it is important for systems to connect the physical and the digital world. For this purpose, Cyber Physical Systems (CPS) were introduced. These systems are able to collect data in the physical world, decide on the proper approach to solve problems in the digital world and communicate this result back to the physical world. An example for a Cyber Physical System is a smart grid, where the distribution of the energy in the grid is adjusted, based on the consumption of each consumer. Often it is important to understand the decision process of a CPS to either find the root cause of an appearing problem or explain the taken procedure to an end user to increase the transparency of the system. The extension of a CPS with the possibility to explain certain aspects of interest about the system, (e.g., unusual and possibly faulty states and behaviors) is called an explainable Cyber Physical System (expCPS). An expCPS is based on causal relations between either objects in the system (e.g., an electrical vehicle charging station, a building...) or events happening during the operation of a system (e.g., reduced amount of available energy, lack of sunshine,...). Therefore, when designing and building an expCPS, the following topics need to be considered: (1) which formalisms to use in order to represent causality relations and (2) how to automate the acquisition of such causality relation knowledge? While several research fields have investigated these topics in very diverse settings and for several application domains, in this thesis we focus on which of these techniques can be applicable in the context of an expCPS System based on Semantic Web technologies and geared towards smart grid settings.To that end, we follow the following methodology: In the start we perform a literature research to identify causality representation methods, causality acquisition algorithms and use cases of causal information in expCPS. Afterwards we select, the most suitable causality acquisition algorithms, define quality metrics to evaluate these algorithms and then implement these algorithms in a framework. In the end we compare the algorithms, based on the defined quality metrics, using controlled experiments. First the algorithms are evaluated on their own and then they are compared to each other.In the thesis we make the following contributions: We provide an overview of important aspects to consider, while choosing or creating a causality representation and causality acquisition algorithm. Further, the thesis develops a definition of causality in the context of expCPS throughout the thesis. The definition starts with a collection of important points described in the literature and is developed into a formal problem statement. Afterwards we propose a causality representation for an expCPS. The focus of the implementation is on four different causality acquisition algorithms: Granger Causality, Transfer Entropy (with a Kernel Estimator and a Kraskov Estimator) and the Peter-Clark Algorithm. In the evaluation of the implementation, we show the increased performance of the Transfer Entropy with a Kraskov Estimator compared to the other three algorithms.