Integrated Management of Experimental Research- and Meta-Data for Fan Test Rigs

1.6 Test Rig & Instrumentation

Experimental studies still remain the most reliable validation technique, as well as important tools during fan design and analysis, development of prediction methods and understanding of physical phenomena. A considerable amount of time, money and know-how is invested conducting such studies, generating and evaluating large amounts of experimental data in digital form.
During these experimental product-validation- or research-projects different tests of varying complexity on different test rig configurations are carried out. Additionally, it is very common for the experimental setup, and thus the generated data, to be subject to a lot of enhancements and adjustments over time. The collective long-term usability of these heterogeneous datasets is largely dependent on the availability of descriptive metadata. This includes information about utilized measurement instrumentation and corresponding calibration data, test environment conditions, and the unit under test. Especially when facing unplausible results, the ability to trace evaluated data back to the measured raw data, metadata and evaluation steps is key. Hence it is imperative to invest into thorough documentation of this data, metadata and its provenance.
In this paper, we focus on the recently at TU Darmstadt developed integrated documentation of descriptive test information by means of data aquisition- and evaluation-software used for fan test rigs. A generic data model based on the HDF5 file format is proposed, enabling flexible storage of measurement data in different processing stages and corresponding metadata in the same file. Each raw and derived dataset is linked to metadata of the utilized device or analythical model, as well as the input dataset. On this basis, the postprocessing of raw measurement data is streamlined, using the integrated calibration data or model parameters for automatic computation of derived process variables. As a result, the need for customized postprocessing software per test configuration can be limited. Thus most further individual computational analysis is built upon the process variables, retaining full information about their origin. One of the fan test rigs at the Chair of Fluid Systems at Technische Universität Darmstadt serves as application example.