Flow Mixing Induced Thermal Fatigue Damage in Power Plant Piping Studied Using High Performance Computing

Principal Investigator:
P. Karthick Selvam

Institute of Nuclear Technology and Energy Systems (IKE), University of Stuttgart (Germany)

Local Project ID:

HPC Platform used:
Hazel Hen of HLRS

Date published:

Understanding the nature of the turbulent flow mixing behavior in power plants which induce thermal fatigue cracking of components is still an unresolved challenge. Aside from measurements being performed at realistic power plant conditions (e.g. at 8 MPa pressure and temperature difference of 240°C between the mixing fluids) numerical calculations involving high-performance computing could throw more light into the complex fluid flow at any location of interest to investigators. Thus a combination of measurements coupled with numerical calculations could positively contribute towards realistic assessment of thermal fatigue damage induced in power plant components.

Piping systems perform the essential function of facilitating coolant transport within a power plant. Being a vast infrastructure, power plants have an extensive network of piping serving different purposes. Ensuring the integrity and functional capability of piping systems throughout their service life are important for the safe operation. Thermal loading imposed on the piping structure by the underlying fluid flow results in some of the unexpected material degradation and failure. Thermal fatigue was identified as a challenge to power plant safety during the 1970s and 1980s when new loading conditions (e.g. thermal stratification, valve leaks) that were not considered during the design stage resulted in fatigue cracks at different locations during the operation. Research efforts were subsequently made to understand these issues by the research community. This lead to thermal fatigue caused by clearly identifiable thermal loading on the structure to be well understood over the years.

On the other hand, a new challenge emerged whereby thermal fatigue in structures caused by thermal loading that could not be monitored using conventional plant instrumentation systems resulted in frequently reported damage of piping systems. Predominantly caused by mixing between flows (e.g. at T-junctions) at significant temperature differences (∆Ts) as shown in Fig. 1, the issue is still being widely investigated and no consensus exists internationally on assessing the fatigue damage caused by such thermal mixing events.

The Challenge

While experimental studies of T-junction flow mixing was conducted around the world to understand the factors leading to thermal fatigue, advances in high-performance computing over the past two decades enabled computational fluid dynamics (CFD) investigations of T-junction flows in much greater detail than was deemed possible before. CFD has complemented experiments in providing reliable information and description of the flow field at locations that are otherwise inaccessible using instrumentation. Thus a combination of measurements at high ∆T representing realistic power plant temperatures and high-fidelity CFD calculations validating the measurement data would provide a very complete description of the underlying flow behavior enabling a more accurate assessment of thermal fatigue. So the fluid-structure interaction (FSI) facility was commissioned at University of Stuttgart with capabilities to conduct flow mixing measurements at ∆Ts comparable to power plants (see Fig. 2) while large-eddy simulation (LES) method was chosen to simulate the turbulent mixing behavior. This represented the first time a LES calculation at such a high ∆T (up to 
200 °C) was performed in the literature and is the main highlight of this project work.

The need for high-performance computing (HPC)

Since LES was chosen to be the numerical methodology, highly intensive computations are to be expected making it necessary to opt for HPC to numerically investigate a wide range of inflow conditions (e.g. temperature, velocity) to understand their effect on flow mixing behavior. The computational domain on average contained about 8 – 15 million nodes and the LES was planned to simulate at least 30 seconds of physical flow mixing with a small time step (0.2 ms) to resolve the relevant flow scales and making it easier to compare the relevant parameters (mean temperature, temperature fluctuations and its frequency spectrum) against measurement data. About 300 – 1000 CPU cores are required for each computation and nearly 100 such computations were performed in the frame work of this project. Thus HPC using the CRAY XC40 supercomputer Hazel Hen provided the highest possibility of performing the vast number of simulations intended to be performed within the time frame of this project.

Findings from the work

• Flow mixing is observed to be incomplete in all the investigated cases due to the low Reynolds number flow from the branch pipe being insufficient to cause complete mixing of fluids. Thus a thermally stratified flow pattern emerges after the mixing between fluids at the T-junction.

• The nature of the thermally stratified flow illustrates the difference caused by ∆T between the mixing fluids. An unstable stratified flow being subjected to extreme oscillations is seen at low ∆T (e.g. 65 °C) whereas the opposite case of stable stratification is observed at higher 
∆T (> 140°C) between flows (see Figs. 3 and 4). This could be attributed to the significant increase in buoyancy forces that is brought about by the substantial increase in ∆T between fluids.

• Strong turbulent penetration of hot fluid into branch pipe and vice versa was observed at 
∆T >140 °C. Stable stratification also gives rise to strong thermal gradients between the top and bottom of the pipe leading to bending of the structure at ∆T beyond 140 °C. Mean temperature predictions by LES showed good agreement with measurement data.

• Thermal fluctuations are seen to have the highest amplitudes near the stratification layer as seen from both LES and measurement data. Comparison between measurement and LES data exhibit reasonable agreement with one another with the exception of a few positions where LES predictions either over- or understate measurement data.

Who will benefit from the insights and in what way?

Thermal fatigue in commercial power plant piping is a highly multidisciplinary problem and the research community as a whole will benefit from the insights gained during this study enabling further advancements in this field. It goes without saying that power plant utilities (e.g. coal or gas based, nuclear plants) will be the biggest beneficiaries since they can optimize their operations based on the overall findings to extend the service life of the power plant resulting in fewer incidents of thermal fatigue damage.

Publications within the framework of the project:

• P. Karthick Selvam. Thermal mixing characteristics of flows in horizontal T-junctions. Doctoral thesis, University of Stuttgart.

• P. Karthick Selvam, John Kickhofel, Horst-Michael Prasser, Rudi Kulenovic, Eckart Laurien. Thermal mixing of flows in horizontal T-junctions with low branch velocities. Nuclear Engineering and Design, 2017, Vol. 322, pp. 32 – 54.

• Patrick Gauder, P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Large eddy simulation studies on the influence of turbulent inlet conditions on the flow behavior in a mixing tee. Nuclear Engineering and Design, 2016, Vol. 298, pp. 51 – 63.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Experimental and numerical analyses on the effect of increasing inflow temperatures on the flow mixing behavior in a T-junction. International Journal of Heat and Fluid Flow, 2016, Vol. 61, pp. 323 – 342.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Large eddy simulation on thermal mixing of fluids in a T-junction with conjugate heat transfer. Nuclear Engineering and Design, 2015, Vol. 284, pp. 238 – 246.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Experimental Investigation of High Cycle Thermal Fatigue in a T-Junction Piping System. International Journal for Nuclear Power, 2015, Vol. 60, Issue 10, pp. 606-608.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Numerical analyses of influence of branch flow on thermal mixing in a T-junction piping system. In: Proceedings of the 16th International Topical Meeting on Nuclear Reactor Thermalhydraulics (NURETH-16), August, 30 – September 4, 2015, Chicago, USA, Paper – 12873.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Experimental and numerical analyses of turbulent mixing of coolant streams in a mixing tee. In: Proceedings of the Fourth International Conference on Fatigue of Nuclear Reactor Components, September 28 – October 1, 2015, Seville, Spain, Paper – 40.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Large Eddy Simulation on Thermal Fluid Mixing in a T-Junction Piping System. International Journal for Nuclear Power, 2014, Vol. 59, Issue 11, pp. 648-649.

• P. Karthick Selvam, Rudi Kulenovic, Eckart Laurien. Large eddy simulation of fluid mixing at high temperature differences in a T-junction piping system. In: Proceedings of the International Congress on Advances in Nuclear Power Plants (ICAPP), April 6 - 9, 2014, Charlotte, NC, USA.

Scientific Contact:

Karthick Selvam 
Institut für Kernenergetik und Energiesysteme (IKE) 
Universität Stuttgart 
Pfaffenwaldring 31, D-70569 Stuttgart (Germany)
e-mail: karthick.selvam [at]

Tags: Universität Stuttgart HLRS CSE