Mixture Models and a New Form of Empirical Fundamental Equations of State by Means ofMolecular Simulation and Hybrid Data Sets

今天，经验多参数模型被成功地用来准确地描述测量良好的混合物的性质。然而，这类模型在描述多组分混合物方面受到限制，其中相关的二元子系统没有或在实验上描述得很差。观察到的问题既与基本的多流体对应状态方法有关，也与用于描述纯组分的状态方程的特征有关。从理论上讲，不同分子之间的相互作用显然不能用对应的态方法来描述。事实上，分子水平上的不同特征对纯流体和混合物中的总相互作用的贡献都不同。然而，根据真实流体的实验数据，这些贡献很难分开。在这一点上，分子模拟有一个严重的优势：分子参数可以系统地变化。基于Lustig的方法，可以直接分析分子特性对亥姆霍兹能量及其导数的影响。在此基础上，建立了反映分子参数影响的经验多参数状态方程的泛函形式。为了达到高精度，最终的方程必须符合混合数据集(包含实验和分子模拟数据的数据集)。利用这种新型的经验多参数状态方程混合模型，可以描述基于不同分子特征相互作用所产生的贡献的组合的相互作用。如果Lustig的方法能够适用于混合物的分子模拟，就有可能直接研究分子相互作用对亥姆霍兹能量导数的影响。这将为发展多参数状态方程的分子混合规则奠定基础。为了达到高精度的目的，还需要对混合数据集进行拟合。然而，根据分子参数或基于分子模拟数据，可以正确地(尽管不那么准确地)描述实验研究得很差或没有实验研究的二元子系统。该项目旨在开发新一代精确的(半)经验性质模型，该模型基于广泛和新颖的分子模拟结果，在描述实验研究较差的纯流体和多组分混合物的二元体系性质方面明显优于现有模型。

Today, empirical multiparameter models are successfully used to accurately describe properties of well measured mixtures. However, models of this kind are limited with regard to the description of multicomponent mixtures, in which relevant binary subsystems are not or poorly experimentally described. The observed problems are related both to the underlying multi-fluid corresponding states approach and to features of the equations of state used to describe the pure components. Theoretically it is obvious that interaction between unlike molecules cannot be described by corresponding states approaches. In fact, different features on a molecular level yield different contributions to the total interaction both in a pure fluid and in a mixture. However, these contributions can hardly be separated based on experimental data for real fluids. At this point molecular simulation has a severe advantage: molecular parameter can be varied systematically. Based on the approach by Lustig it is possible to directly analyse the impact of molecular features on the Helmholtz energy and its derivatives. On this basis it becomes possible to develop functional forms for empirical multiparameter equations of state, which reflect the influence of molecular parameters. To achieve high accuracy, the final equations have to be fitted to hybrid data sets (data sets containing both experimental and molecular simulation data). With this new type of empirical multiparameter equations of state mixture models can be developed, which describe interactions based on a combination of contributions that result from interactions of unlike molecular features. If the approach by Lustig can be adapted to molecular simulations for mixtures, it becomes possible to directly investigate the influence of molecular interactions on derivatives of the Helmholtz energy. This would establish the basis for the development of molecularly founded mixing rules for multiparameter equations of state. To achieve the aimed at high accuracy, fitting to hybrid data sets will be required for mixtures as well. However, poorly or not experimentally investigated binary subsystems can be described correctly (even though less accurately) on the basis of molecular parameters or on the basis of molecular simulation data. The proposed project aims to develop a new generation of accurate (semi-)empirical property models, which are based on results of extensive and novel molecular simulations, and which are clearly superior to existing models with regard to the description of properties of poorly experimentally investigated pure fluids and binary subsystems of multicomponent mixtures.