Skip to main content

Unsteady transitional flows in axial turbomachines (UTAT)

Exploitable results

The main goal of the UTAT project was to contribute to the improvement of global performance in turbomachinery by the means of developing new unsteady design methods. One particular key issue is the unsteady transition from laminar to turbulent flow. Laminar to turbulent transition typically occurs in LP turbines at Reynolds numbers lying between 80 000 and 300 000. It can be due to natural boundary layer development (natural transition), to turbulence diffusion from the main flow into the boundary layer (bypass transition), to unsteady effects such as wake passing along the suction side (wake induced transition) or downstream potential blade row interaction, or may be tripped by dedicated blade surface roughness elements. The original research objectives resulted from the need to improve physical understanding as well as analytical and numerical modelling capacities of this phenomenon in axial turbomachines with a special emphasis on wake or potential interaction induced transition in LP turbines. To address these issues, the project was divided into four main areas of interest (Work Packages): assessment and development of RANS-based transition models and CFD methodologies The objective was the investigation of advanced RANS/URANS and hybrid URANS/LES models to study boundary layer transition caused by wake-blade interactions, potential interactions and blade surface roughness. The developed transition-models have been distributed to the industrial partners in order to test and validate industrial CFD codes against selected test cases. The following models have been newly developed and collectively tested and evaluated: - PUIM tested by UCAM, TUCz, Snecma and Alstom: this method is based on various empirical correlations with a given validity range. It performed well to predict attached and separated transition process. Wake boundary layer interaction was predicted with good qualitative and quantitative agreement for variety of geometrical configuration. - Integrated approximate eN method tested by KTH, MTU, ITP and VAC: this method is simple to implement and works well in attached and slightly separated flow conditions. However, the method fails if separation is getting large with low Reynolds numbers. The results depend on the accuracy of the BL parameters evaluated by the main flow solver. - Dynamic intermittency transport model tested by Ugent and TUCz: this two-dimensional model is easy to implement. It relies on the empirical Mayle criterion and works well for a range of transitional flows, including steady flows with bypass transition and wake-induced transition. - Non-linear eddy-viscosity model of AJL tested by model has a strong fundamental foundation. It is e ICSTM: this non-linear LRN k- capable of resolving with reasonable accuracy the steady and unsteady transition phenomena occurring due to combined action of free-stream turbulence and passing wakes. It relies however on a transition-specific modelling for the 'laminar fluctuation energy'. - Turbulence models with surface roughness tested by ONERA, Snecma, AVIO and ITP: If the effect of distributed roughness on friction coefficient can be well predicted, the models are not yet valid for an isolated trip element. - Detached eddy simulation (DES) tested by ONERA: this method is easy to implement. It allows a good prediction of the upstream wakes and is suited to fully turbulent flows with separation. However, the results do not allow us to conclude that DES can bring interesting improvements. The major results of this project have been exploited and/or disseminated in different ways by both industrial and academic partners. Dissemination by research institutes materialised through technical publications, reports, seminars, workshops and short courses available to third parties in line with EU policy. Partners concerned with education have made the use of UTAT knowledge on a regular basis for education of undergraduate and graduate students as well as of professional engineers. Exploitation by the industrial partners comprised intensive use of the newly generated technology and design procedures in their own low Reynolds number applications, where the benefits of transition control were incorporated at the design stage.