Particle-laden swirling flow

Application Challenge AC3-12 © copyright ERCOFTAC 2013

Introduction

The special features of swirling flows are utilised in combustion systems in order to provide flame stabilisation and good mixing between fuel and oxidiser. This is achieved by the central recirculation bubble developing in front of the burner exit. Swirl burners are usually operated with liquid (spray) or pulverised fuels.

In order to obtain a better understanding of the particle behaviour in such a complex swirling flow, detailed experiments were conducted on particle-laden swirling flow emanating into a pipe expansion (Sommerfeld and Qiu 1991). The gas-particle mixture was injected centrally without swirl together with a co-flowing swirling annular gas jet yielding a swirl number of about 0.5. Downstream of the inlet simultaneous measurements of gas and particle velocities (all three components) were conducted by phase-Doppler anemometry, which also provided local particle size distributions and the stream-wise particle mass flux. Two cases with different injection flow rates, but roughly identical swirl number were considered. Both cases showed a closed central recirculation region. Inlet conditions are available from highly resolved profiles 3 mm downstream of the edge of the inflow tubes. Since the particle mass loading is rather small two-way coupling effects are of minor importance. Numerical computations performed with the finite-volume code FASTEST in connection with the k-ε turbulence model showed reasonable good agreement with the measurements (Sommerfeld and Qiu 1993). The particle phase was simulated by Lagrangian tracking also yielding a quite good agreement with measured velocity profiles, the particle mass flux and the number mean particle diameter.

Relevance to Industrial Sector

Particle- or droplet-laden swirling gas flows are found in numerous technical applications. In spray or coal fired combustion systems, swirling flows are used to establish very high mixing rates between fuel and swirling air stream and to ensure the required flame stability. This is achieved by the developing central recirculation region, where the hot reaction products are convected backward and mix with the fresh fuel and air.

In addition particle-laden swirling flows are found in numerous different types of particle separation devices, such as air cyclones. The considered basic flow configuration allows detailed validation of computations for particle-laden swirling flows, especially with respect to Reynolds-stress turbulence modelling or LES applications and particle dispersion in anisotropic turbulence.

Design or Assessment Parameters

The assessment parameters for this test case are the mean velocity profiles as well as those for the rms values along the test section for both phases. Additionally, profiles of the stream-wise particle mass flux could be estimated from the measurements revealing the centrifuging effect of a swirling flow. From the numerical calculations also the particle residence time as a function of particle size may be used to assess the performance of numerical calculations.

Flow Domain Geometry

The swirling flow was realized in a kind of pipe expansion flow (Figure 1).

Through the central inlet tube (diameter 32 mm) the gas-particle mixture was injected into the test section without swirl. The co- flowing swirling flow was produced by a vane-swirl generator located upstream the inlet. The annular inlet tube has an inner diameter of 38 mm and an outer diameter of 64 mm. The test section has a diameter of 194 mm and a length of 1,500 mm. The end of the test section is connected to a sufficiently wide stagnation chamber.

Figure 1: Swirl Flow test section made from Plexiglas including the main dimensions (length 1,500 mm).

Flow Physics and Fluid Dynamics Data

The considered swirling flow is highly turbulent, but may be considered as incompressible. The measurements were done under ambient conditions with a temperature of about 300 K yielding an air density of 1.18 kg/m³ and a dynamic viscosity of 18.4×10^-6 kg/(m·s). The characteristic non- dimensional parameters are:

The flow Reynolds number, which is calculated with the total volume flow rate, the outer diameter of the annulus and the effective inlet cross- sectional area A_in:

{Re={\frac {\rho D_{3}{\dot {Q}}}{\mu A_{\text{in}}}}}

The swirl number, which is the ratio of the axial flux of annular momentum to the axial flux of linear momentum obtained by integration across the primary and annular inlets:

{S={\frac {\displaystyle 2\int _{0}^{D_{3}/2}\rho \ WUr^{2}\ dr}{\displaystyle D_{5}\int _{0}^{D_{3}/2}\rho \ U^{2}r\ dr}}}

Here U and W are the mean axial and tangential velocities and r is the radius. The values of the outer diameter of the inlet, D₃, and the inner diameter of the test section D₅ are given in Figure 1. The Reynolds numbers of the two cases considered are 52400 and 54500 and the swirl numbers are around 0.5.

As mentioned above, the particle mass loading was rather low so that the effect of the particles on the flow field may be neglected. The particles are spherical glass beads with a material density of 2,500 kg/m³ and a relative wide size distribution ranging between about 10 µm to 80 µm. The number mean diameter (mean diameter based on particle counts) is 45 µm yielding a mean Stokesian response time of about 15 ms.