搅拌釜内混合液体的分离涡模拟英文文献和翻译
In the present work, the detached eddy simulation (DES) model was used to gain an insight into the
liquid-phase mixing processes in stirred tanks agitated by centrically and eccentrically located four
pitched-blade turbine. The impeller rotation was modeled using the sliding mesh (SM) method. Based
on the time variation of the scalar concentration, the mixing patterns and mixing time were analyzed
and compared with the published planar laser induced fluorescence (PLIF) experimental data (Hu YY,
Liu Z, Yang JC, Cheng Y. Liquid mixing in eccentric stirred tank. CIESC J 2010; 61(10): 2517–22). Results
show that the combination of DES model and SM method can characterize the transient mixing state in
stirred tanks and can provide a detailed spatial and temporal evolution of the scalar concentration. The
concentration recorded at several locations inside the domain revealed different mixing patterns under
concentric and eccentric agitation configurations. The differences in mixing time at different monitoring
points were small under eccentric agitations, which exhibit better mixing performance than the concen-
tric agitation. Overall, the predicted mixing time compared well, on average within 20%, with the results
obtained by the PLIF technique. The agreement shows that DES is a reliable tool to investigate the
unsteady mixing characteristics in stirred tanks.6313
2012 Elsevier Ltd. All rights reserved.1. Introduction
Stirred tanks are widely used in the chemical, mineral process-
ing, wastewater treatment and several other process industries.
Among these operations, the mixing of single and multiphase flu-
ids is one of the most common unit operations and is fundamental
to most aspects of process performance. The most crucial parame-
ter used to evaluate the mixing efficiency is mixing time and typ-
ically, the t95 mixing time is used. It is defined as the time required
from a non-equilibrium condition to achieve within a value of ±5%
of the final concentration. Extensive investigations of the fluidmix-
ing in stirred tanks have been carried out over the past several dec-
ades based on experimental techniques. Nere et al. [1] reviewed
such experimental techniques and pointed out their main features
and limitations. Some of these disadvantages, such as the difficult
applicability in the case of non-transparent stirred tanks, the se-
vere experimental conditions, and the fail to obtain very detailed
evolution of the scalar concentration, have confined their applica-
tion in the process industries and can only be used in the labora-
tory scale experimental measurements.
In recent years, owing to the great progress in the computer
technique, the computational fluid dynamics (CFD) technique is
being increasingly used as a substitute for experiments to carry
out detailed studies of the mixing process. The accurate prediction
of the liquid flow field, including its turbulent characteristics, plays
an important role in the numerical predictions of the mixing time
[2]. Previous studies have demonstrated that numerical simula-
tions based on the Reynolds-averaged Navier–Stokes (RANS)
approach can provide satisfactory results of the mean flow in
stirred tanks [3–7]. However, it is difficult to obtain accurate pre-
dictions of the turbulent quantities and fails to capture the instan-
taneous nature of the turbulent structures because of the
assumption of isotropy and ‘time-averaged’ modification of the
Navier–Stokes equations [4,6–8]. This will certainly affect the pre-
dictions of the mixing time in stirred tanks. As a matter of fact, it
has been identified that, in the turbulent flow regions, even when
the fully predictive simulation strategies, such as the sliding mesh