Tow-phase Flows in Fluent and Estet-Astrid
Said GHALIMI (mfn06) and Samir Karaaa (cshp01)
Two-phase flow problems can be modeled in several ways. For this reason we aimed to study one two-phase flow problem using FLUENT and ESTET-ASTRID software. Those simulations could be compared to better understand how the two software models this problem.
It is known that bubble column reactors have a wide range of applications such as absorption, catalytic slurry reactions etc. One of the BEI projects in which we participate aims to estimate the design parameters for a bubble column reactor. Bubble columns are contactors in which a discontinuous gas phase in the form of bubbles moves in relation to a continuous phase. The continuous phase can be a liquid or a homogeneous slurry. The reaction between the two phases can take place on the bubbles' interface. Beside this estimation that makes use of empirical correlation, it is important to model the problem.
The chemical process industries are increasingly turning to bubble reactors for gas-liquid contacting operations such as the oxidation of the ortho-xylene to toluene acid. This reaction takes place in a liquid phase which in our case is the acetic acid (solvent), the reactive component is dissolved in the solvent when the oxygen is continuously injected into the mixture.
Fig. 1 - Bubble column, with the gas injector.
We need to produce a fixed amount of Toluene acid annually, the reaction is :
Making use of FLUENT or ESTET-ASTRID, we hope to calculate different physical parameters such as :
Volumetric oxygen-melting mass transfer coefficient K.
The slip velocity.
The superficial gas velocity.
Mean bubble diameter.
The Hatta number and the Acceleration coefficient.
The modeling capabilities of FLUENT allow to simulate a wide range of discrete phase problems including particle separation and classification, spray drying, bubble stirring of liquids, liquid fuel combustion, and coal combustion. Indeed, in addition to solving transport equation for the continuous phase, FLUENT allows you to simulate a discrete second phase in a Lagrangian frame of reference. This second phase consists of spherical particles, which may be taken to represent droplets or bubbles, dispersed in the continuous phase. FLUENT computes the trajectories of these discrete phase entities, as well as heat and mass transfer to/from them. The coupling between the phases and its impact on both the discrete phase trajectories and the continuous phase flow can be included. You can include the discrete phase flow in your model by creating an Injection in the Set Injection Properties.
There are three particle types we can define :
Inert : the active laws are Inert Heating or Cooling, and are applied while the particle temperature is less than the vaporization temperature.
Droplet : the active laws here are Heating, evaporation, and boiling. Those laws are applied while the particle temperature reaches the vaporization temperature.
Combusting : the active laws here are Heating, evolution of volatile/swelling, and an Heterogeneous surface reaction.
Difficulties encountered :
The components brought into play don't exist in FLUENT database, nor the reaction rate or the reaction enthalpy. All the following must be defined.
A Gas-liquid Mass Transfer Model must be defined e.g. the two films model which models the transfer occurring in a gas-liquid interface.
The first model (Mass Transfer) must be modified to take the reaction in the mass transfer balance into account.
The energy produced should be let out to keep the operating conditions constant.
The continuous phase must be a gas.
As you can see, in every case, the continuous phase must be a gas. So we conclude that it isn't possible to model easily this problem with FLUENT.
Astrid is able to simulate the bubble column problem, Indeed, we have learn ESTET-ASTID at the IMFT, by testing some examples, such as :
The jet problem.
The flow between two plane plaques.
But, in the IMFT, there isn't a post process to visualize results. And there is no way to exploit the output binary files.
Despite the difficulties encountered, this task will continue to model the BEI problem, likely with ESTET-ASTRID. By the way, we must find a post-process able to visualize ESTET-ASTID outputs result e.g. RUBENS.