• Three-dimensional 
• Steady state
• Multicomponent-gas (air and humidity)
• Realizable K-Epsilon with Two-Layer type: Xu (Buoyancy Driven)
• Solid (concrete obstacles)
• Energy with CHT
• Fluid Film (liquid water)
• Film Evaporation/Condensation
• S2S-Radiation

Boundary Conditions

The boundary conditions are a warm wall to the right, a cold wall to the left and a puddle of water at the bottom.

The cold wall is set to 1.2 °C, the hot wall is exposed to the environment with 21 °C, the bottom puddle is set to 13.7 °C and the top is set to 14.4 °C.
The other walls are set adiabatic, but contribute to the radiation.

The puddle temperature is particularly influential on the humidity level in the system as it prescribes the evaporation rate. Small deviations in the temperature of 1 K lead to 10% more or less evaporation.

The cold and bottom walls have shell regions attached to allow for the fluid film model with evaporation/condensation. 
The bottom puddle is not allowed to empty by applying a mass source to it. The film on the left wall is allowed to exit at the bottom edge.

Radiation Settings

As described in the paper, setting the correct radiation settings is crucial. The overall temperature and therefore humidity level in the system is highly dependent on the energy balance. And this balance is dependent on the emissivity of all the surfaces in the experiment.

There are four types of surfaces in the experiment: Plastic (PVC) for most walls, Aluminum for the cold wall, Glass for the warm wall, and Plaster for the rods.
The emissivity is set to PVC: 0.5 - Alu: 0.4 - Glass: 0.7 and Plaster: 0.85

Physics and Solver Settings

The fluid film model and solver settings are based on Best Practice for Fluid Film Evaporation/Condensation.

To fit the emissivity to the experiments a curve fit optimization can be used.
To get the error the methodology described here was used: Example: Calibration Study in Design Manager using Curve Fit and NRMSE (normalized root-mean-square error)

Results

Unhumidified Cavity

There are two variants - the unhumidified and the humidified cavity. The main interest is on the case with fluid film and humidity but here is one comparative result for the case without humidity.
In the middle the temperature from the simulation is shown. On the right is the deviation from the experimental measurements - keep in mind there are only measurements on the probe positions - the rest is interpolated.
On the left is the comparison for the vertical temperature profile - the dots are experimental measurements and the lines from simulation.

Overall agreement is very good. One can clearly see the difference in solid and fluid temperature in the 'humps' of the lines. 

Humidified Cavity

In the humidified case the bottom puddle is filled and in the simulation the fluid film is turned on.

The plots on the right-hand side are from horizontal cuts at the middle of the cavity and plot the temperature and velocity, respectively.
The experimental measurements are shown in green, simulation results from the paper are shown in blue and results of this work are shown in red. These colors are used for all following plots.

The temperature level fits the experiment quite well. For the velocity the Ansys Fluent result are much sharper towards the wall. This is likely due to the laminar modeling compared to the turbulence model used in this work. 
The Realizable K-Epsilon with Two-Layer type: Xu (Buoyancy Driven) has shown in many natural convection cases to provide good results.
The larger velocity boundary layer fits the experimental results better. The hump in the second gap is missing in both simulations.
It is also worth mentioning that the paper uses only a 2D simplification of the model whereas we modeled the full 3D cavity.

Comparing the vertical temperature and relative humidity profiles shows a good match between simulation and experiment. 
These profiles are from the left side of the cavity, close to the cold wall, cutting through the first plaster rod row.

Here are some more results showing the mass fraction of water, the temperature once more, and the relative humidity in the middle of the cavity.

Future Work

In following articles we want to discuss the flow or turbulence modelling for natural convection cases on this example. The animation at the top is a preview to this work.

As described in the fluid film evaporation best practices sometimes the film model is not actually needed to model wall evaporation and condensation. A simplified setup without fluid film model will be shown in another article.

Full refence to the original paper: