Dr. Dirk Lucas

Head Computational Fluid Dynamics
Phone: +49 351 260 2047

Modeling of churn-turbulent flows

==> Click on the pictures/videos for full view.

While Computational Fluid Dynamics (CFD) codes for single phase flows has been widely used for industrial applications, in the case of two-phase flows the applicability is still very limited due to the complexity of the interface.
One of the most convoluted flow-pattern that can be found in gas-liquid two-phase flow systems is the so-called churn-turbulent regime. The churn-flow can be reached in large pipes when high superficial gas velocity and low superficial liquid velocity are used. It is characterized by high gas volume fraction (in excess of about 30%), and a tri-modal broad distribution of the bubble sizes. The bubbles occurring in such a flow can be classified in small, large, and distorted bubbles. The small bubbles are generally spherical or elliptical and are encountered in a major concentration in the wake of large and distorted bubbles and close to the walls. Large, -ellipsoidal or cap bubbles can be found in the core region of the flow as well as the distorted bubbles with a highly deformed interface (see Fig. 1).

Churn-turbulent flow is commonly encountered in industrial applications, and so a necessity to understand the physics behind it, and model it in an efficient and accurate manner arises. A typical example is boiling flow in nuclear reactors. Especially for accident scenarios, boiling may lead to high void fraction including churn-turbulent flow. The flow structure may have a strong influence on the safety. For these reason, reliable predictions of such flows is an important issue for safety analyses.

Fig. 1: Examples of the churn-turbulent flow regime

The computational modeling of the flow is done using an Eulerian multi-fluid approach with the ANSYS software package CFX, while ICEM CFD is generally utilized for generating the mesh. The starting point which is based on existing models for bubbly flows include drag force, turbulent dispersion, bubble induced turbulence, among others. Some examples of these are the particle induced turbulence model developed by Rzehak (Rzehak and Krepper, 2012), the bubble-bubble interaction model established by Liao (Liao et al, 2011), and the inhomogeneous MUSIG (MUltiple SIze Group) approach developed by Krepper (Krepper et al., 2008). The last one allows to defined different bubble sizes groups with different velocity fields for both large and small bubbles. For the simulation of churn-turbulent flows, the gas phase is represented by 3 different gas fields corresponding to the 3 types of bubbles mentioned above. Such a multi size group approach allows creating a more realistic approximation of the churn-phenomenon (see Fig. 2a, 2b, and 2c). Experimental data for upwards vertical pipe flow obtained at the TOPFLOW facility are used for the validation of the CFD models.

Fig. 2a: 6 mm ≤ DB ≤ 11 mm Fig. 2b: DB > 11 mm Fig. 2c: Velocity of the liquid

The Eulerian multi-fluid approach is widely used to describe dispersed flows like bubbly or droplet flow since such flow patterns are characterized by scales of interfacial structures which are smaller than the grid size. For flow situations with large-scale interfaces like film-, annular or horizontal stratified flows usually interface tracking methods are used. In principle, the interface capturing methods should be used for interfacial length scales several times larger than the grid size, while for an averaged two-fluid approach bubble sizes smaller than the grid size are required (Hänsch et al., 2012). Since in case of churn-turbulent flows dispersed flows and large interfaces occur simultaneously, a combination of these modeling concepts would be needed. Such an approach, the GEneralized TwO Phase flow (GENTOP) concept, was recently developed in our CFD group. This concept which consider dispersed and continous gas phases is applied to churn turbulent flow (see Fig.3).

Fig. 3: Continuous (DB > 7 mm) and dispersed (DB ≤ 7 mm) void fraction representation on churn-turbulent flow

Each simulation is carefully validated against experimental data acquired from the TOPFLOW vertical test section facility, where water and air are use as test fluids. Experiments using large (DN 200) and small (DN 50) pipes are used and an upward flow is study. The radial profiles at different heights are measured using a two-level wire-mesh sensor in order to obtain gas velocities, bubble size distribution, total gas holdup, among other parameters (see Fig. 4 and 5).

Fig. 4: Bubble Size Distribution and Gas Velocity Profile at L/D = 39.9 (JL = 1.017 m s-1; JG = 0.342 m s-1)
Fig. 5: Bubble Size Distribution and Gas Velocity Profile at L/D = 39.9 (JL = 1.017 m s-1; JG = 0.342 m s-1)

Current studies are being devoted on the improvement of the modeling capabilities for high void fraction regimes, such as the development and modifications of different closure laws for large distorted gas structures (churn- and slug bubbles), and the constant improvement of the GENTOP –concept for the physically accurate full resolution of such large structures in highly turbulent flows (see Fig. 6).

Fig. 6: Videos of full 3D transient simulation using the GENTOP –concept. (JL = 1.017 m s-1; JG = 0.342 m s-1)



This work is carried out in the frame of a current research project funded by E.ON, and in cooperation with the Nuclear Science and Engineering Department of the Massachusetts Institute of Technology (MIT).



Dr. Dirk Lucas

Head Computational Fluid Dynamics
Phone: +49 351 260 2047