Phenomena: CFD of Large Structures
Potential Industry Applications:
- Oil & Gas
- Environmental
- Chemical
- Manufacturing
Main Software: STAR-CCM+ v7.04 to v8.04
Secondary Software: Matlab, Mathmatica, Excel, and Microstation
Analysis Type: CFDModels:
-
Volume of Fluid (VOF) multiphase
-
Transient state
-
k-ε double layer turbulence
-
Surface Tension
Mesh:
- Iterative Anisotropic
- 30-40 Million cells
Solver: PISO solver
Findings:
-
CFD has a place in small production facilities for optimizations
-
Backflow in circular distribution chambers, optimized by CFD on a case-by-case basis, is beneficial to even distribution
-
Channel distribution chambers can be accurately studied using 1D simulation
-
CFD of non-circular distribution chamber shows optimization of baffle to reduce wall effect
-
CFD needed to assess pressure drop in many chambers as literature can be an order of magnitude off from reality
For obvious reasons, the processes used to treat public water have been standardized for quite some time. However, this has led to a ‘rule of thumb’ mentality (at the plant level) among waste water professionals for even non-regulated processes. Additionally, the reduced budgets and returns for water purification plants, in comparison to other production industries like oil and gas, has limited the use of optimization tools like simulation.
Nonetheless, R&D professionals in water purification are using CFD simulations to better understand the flow between standardized processes. If applied to the production plants, perhaps one could better optimize the system beyond the ‘rule of thumb’? This is what Pedro Fonseca, Suez Environment’s Hydraulic Engineer, hopes to test with the help of CFD specialists from BlueCAPE.
Simulation Setup
Channel (left), Circular smoking pipe, Non-circular smoking pipe (right) Distribution chambers.
“A treatment plant is normally a set of complete tanks where water goes in and come out called distribution chambers. Each chamber has its functionality,” explains Fonseca. “These distribution chambers follow, from a design engineering point of view, the same ‘rules of thumb’ as pipelines. But what we see is that when we get to the 3D domain these guidelines can be greatly improved upon. That’s where CFD becomes a great weapon.”
Distribution chambers come in various shapes and sizes, but typically they are large, spanning dozens of meters in diameter, and made of concrete. Due to orifices, baffles and weirs, however, the flow is influenced on a scale in the order of millimeters.
Due to the sludge and sewage that may be present in the process, the system was setup using the volume of fluid (VOF) multiphase model. Additionally, surface tension was assumed to have an influence on the results. Finally, the system was modeled as a transient state, with free surface CFL<0.5 and double layer turbulence k-ε and y+ wall treatment.
As the day-to-day sewage characterization is impossible to predict or validate, it was assumed that the fluidic characteristics of the system were close to water. The water was therefore not simulated as a Lagrangian solid liquid multiphase as one might expect given the particles present in the real life flow.
Due to the large model size and small water surface, 30-40 million cells were needed to mesh the various distribution chambers. This is an interactive anisotropic mesh that used refinement field functions to assess the free surface details of the system.
Circular (smoking pipe) distribution chamber simulation.
This ‘smoking pipe’ looking geometry involves water flowing through the bottom inlet pipe. The fluid fills the circular chamber until the weirs direct the flow into separate processes. It is suspected by ‘rule of thumb’, however, that the inlet will force the fluid to prefer some weirs over others.
“There are a lot of ‘rules of thumb’ in this kind of design. They are correlations between the kinetic energy of the entry, the diameter of the chamber, the length of the vertical section,” says Fonseca.
However, “We can see immediately that there is an interesting backflow at the bottom of the chamber where the water comes in. This will ensure that the water doesn’t just impact the weirs in front as some of the water is deflected backwards. Therefore, the jet is much smaller than you would predict with outside knowledge. This is one of the reasons to use CFD in our business,” Fonseca stresses.
In other words, the simulation was able to provide an optimum relation between the diameter and inlet pipe to ensure the backflow of the jet. This backflow will improve the equality of fluid distribution.
Variations of the parameters showed varying results on a case-by-case basis, further proving that CFD has an important place in the individual production facility.
Channel Distribution Chamber simulation.
A channel distribution chamber will divide the flow between six parallel processes using weirs. By looking at it, “We can almost say that it is 1D reducible, that is, we can imagine that there is just one path to reduce the complexity model.”
As predicted, the comparison between the 1D and CFD models saw a flow rate distribution difference of less than 0.5%. This suggests that for this particular chamber, CFD isn’t needed as 1D simulations are much faster.
Therefore, individual facilities shouldn’t feel obligated to simulate every subsystem of the process with CFD. If a simulation can be simplified to a 1D or 0D process then a complete CFD analysis may not be necessary. Skipping the simulation will save time and money.
Non-circular Distribution Chamber
Looking almost like a combination of the circular and channel chambers, this setup uses an inlet pipe, a baffle at the inlet, and weirs to distribute the flow. The inlet baffle makes use of CFD imperative to assess the flow.
Non-circular (smoking pipe) Distribution Chamber simulation.
The baffling effect is seen within the simulations. Additionally, the wall effect of the chamber has been reduced to have minimal effect on the distribution. Using CFD, the individual production facility can benefit from assessing the best location of the baffle to ensure proper distribution.
Pressure Drop
When designing a distribution chamber, or any complicated fluid flow process, the typical coefficients seen in literature may not be applicable to the current geometry. Therefore uncertainty can become a factor.
“It is my experience that we can have an order of magnitude error between common literature and reality. This is a little bit too much when you talk of a business where we treat water at 10 – 20 m3/s. So this is another example where we use a lot of CFD,” explained Fonesca.
By assessing the pressure drops of subsystems, individual production facilities can reduce the power usage of their pumps at the design stage. This can represent a significant reduction to the budget and a good step towards a green label for carbon reduction.
Conclusion
Fonesca found that, “We were able to reduce a lot of the outside capital costs, proof that we are optimizing the designs and going further than the ‘rule of thumb’.” This suggests that the daily use of CFD in water treatment, or any other production industry, is feasible, despite the lack of funds the average company has in comparison to the Oil and Gas industry.
In the future, the team hopes to include Eulerian multiphase in the simulations. This will allow the optimization of bubble sizes and sand filters in various purification processes. In addition, a UV simulation will help ensure water disinfections.