A clearing houses acting as central counterparty collects collaterals via margin calls to cover a portfolio's market risk. Over the past few years exchange trading volumes in commodity trading and in particular for energy related products have grown massively. At the same time the margin models applied by clearing houses have hardly changed. The margin systems SPAN and TIMS are based on simple calculation methodologies from the 1970s but are still by far the most used systems by clearing houses. This work introduces the concept of copula in which marginal distributions of the single risk factors can be modelled separately from their multivariate dependence structure. It compares a T-copula model and a variation of it, the grouped T-copula, with a Gaussian copula and the TIMS system and proves the superiority of the T-copula models.