Fundamental quality requirements for the generation of LES-specific validation data sets from systematic wind tunnel model experiments

Abstract

RANS-based models as well as LES-based codes have to be validated carefully in order to document their fitness for purpose and to document the quality of model predictions. For RANS codes, several test data sets are already available. Due to the higher demands on validation data for LES-based models which significantly exceed the extent and quality of the existing data sources, there is a need of model- and application-specific, temporarily and spatial-resolved reference data. Particularily in wind engineering and engineering meteorology such data can be obtained from measurements in boundary layer wind tunnels. Within the scope of a research initiative funded by the German Research Foundation, a concept for quality requirements and evaluation criteria has been developed jointly by wind tunnel modelers in close cooperation with numerical modelers. The good performance of physical modelling is mainly a result of the physical similarity of the simulation of the relevant flow and dispersion processes. However, wind tunnel modellers must follow stringent quality criteria to maintain accurate and reliable validation data for the investigation of urban flow and dispersion problems. The goal is to support the confidence of numerical modellers in using laboratory data and accepting them as a reference. The paper will introduce the model- and application-specific quality requirements defined for compiling LES-specific validation data sets. In addition to the commonly agreed on criteria such as the provision of test cases of different complexity or the desired systematic testing with respect to fundamental physical relationships, some - often underrated - criteria were identified. The required detail in data documentation or the needed completeness of the data bases are defined exactly. In particular, it will be illustrated, why and to what extent the required model specific test data have to meet distinct quality and quantity levels which significantly exceed standard laboratory measurements. As a related and very important aspect, the contribution will show that field data are of limited value in validating eddy-resolving flow and dispersion models. Furthermore, it will be illustrated, how numerical models can be evaluated with data generated in the wind tunnel laboratory. A comparison of averaged quantities is usually not sufficient for LES model validation purposes. It has been shown that even the validation of statistical measures for characterizing the turbulent structures is not satisfactorily. The next steps of validation exercises will be identified which include the comparison of correlation measurements as well as the evaluation of coherent structures.