Multiphase CFD for the water industry: status and applications

Today, Computational Fluid Dynamics (CFD) is known by most of the water industry professionals. Less known is the fact that today's CFD is much more than some 'flow simulation of water'. The simulations can include gas and solids as a 'separate phase', and there are many proven ways to do that. Translated to water applications, this means bubbles, flocs and particles of all kinds. This what we call 'multiphase CFD'. The three phases can be combined in a single simulation.This article gives you some basic explanation and three very different practical examples in which multiphase CFD was used to solve practical problems. After reading this small article, you'll be much more aware about the possibilities so you can harvest them.

What are the 'three phases'?

First, let us quickly summarize what these phases can be. Water and wastewater treatment is special: we remove gases and solids from the water in some steps, and add gases, solids and liquids in other steps. So as a result, most of these unit processes contain more than water. Figure 1 shows the plug and play of CFD 'gears'. Although you can 'play' with viscosity, mass transfer, kinetics, ... we highlighted the 'phases' for you: liquid, gas and solids, and some examples of each.

And which phases do we have to use when? It depends on the objective and the nature of the process. For example, if your question is 'will solids wash out, and if so, which fraction? Where do solids accumulate?' You might use a 2-phase model containing water and the solids with a given density distribution. In the same process, your question might be: 'will flow distribution be optimal with this inlet structure?'. In the latter case, you might even use a 1-phase model (water only). On the other hand, a 'heavy' fluidized bed should probably always modelled at least in 2-phase, as the solids impact the liquid flow too much. So this is what we mean with OBJECTIVE and NATURE OF THE PROCESS.

Figure 1: The plug and play of gears makes CFD extremely powerful - not many practitioners are aware of today's possibilities

To put things in practice, look at Table 1. It contains some well known water and wastewater treatment processes. All of them have been modelled multiple times with CFD, and the number of phases depended on the specific optimisation or design objective. For example, we have modelled DAF and anaerobic reactors in 1-phase, 2-phase and 3-phase. The last 2 years we see a strong growth in 3-phase modelling.

Table 1: water and wastewater treatment processes contain various phases - to treat water, we either remove gas and solids, or we add them

Example 1: Aeration modelling (2-phase)

The scale at which aeration modelling can be applied varies enormously. Figure 2 shows aeration modelling (2 coarse bubble aerators) in a large surface water basin. Despite the size of the basin, the aeration impacts mixing significantly. The model used was extremely well validated in the past, so we are confident about its accuracy. We typically use that model to answer the following questions:

• Can we save aeration energy? (i.e. is mixing still ok if we lower air flow rate, or switch off one aerator?)
• Are aerators installed at their optimal location?

So these questions typically relate to energy savings and drinking water quality/treatment.

Figure 2: 2-phase aeration modelliing in a large surface water mixing basin (color scale: velocities of water mixing (red= high velocity)

Figures 3 and 4 shows wastewater examples (two different plants). Again, the model was validated at many plants before. We typically use this model to answer the following questions:

• Can we save aeration energy by better air distribution?
• Can we switch off one or more mixers, or save mixing energy in a different way?
• Can we improve effluent quality by improving mixing, keeping air in aerobic zone, ...

So these questions typically relate to energy savings and effluent quality improvements. Sometimes it's about capacity upgrades (doing more with the same reactor).

Keep in mind that we use similar models for pure oxygen plants, or systems with a mixture of fine and coarse bubble aeration. So it goes wide.

Figure 3: the impact on a mixer on bubble distribution in a wastewater treatment plant (color scale: local gas fraction in water - the red 'hotspots' are aerators at the bottom)

Figure 4: aeration modelling in a rectangular bioreactor - 3D geometry (left) - pay attention to the aerators (black spots) and gas holdup (top right) and velocities (bottom right) - this plant had shorcircuiting over the surface!

Example 2: 'Granular systems' modelling (2-phase)

This is an example of 2-phase modelling of water and solids, whereby the solids behave more like individual granules and are heavy and concentrated enough to impact the water flow. Further, we added the real size/density distribution (different sizes in the same simulation),as granules do not have 1 single size. Figure 5 shows the example of pellet softening, a well known drinking water treatment process. 'Similar species' in the wastewater industry are granular sludge systems and struvite reactors. In this drinking water example, we used the model to answer the following questions:

• Can we reduce the size of the reactors without the risk of washout?
• Can we optimise the reactor shape to get the optimal pellet distribution? How does this distribution look like as function of height?
• How to design the inlet structure as such that feed is distributed evenly, and shortcircuiting avoided?

So these questions typically relate to saving OpEx and CapEx (smaller reactor),and process reliability (continuous drinking water quality).

Again, we started testing different changes virtually with the model. A 1-phase model in this case would have made no sense, because of the mass of the solids and their impact on the water flow...

Figure 5: distribution of granules with different sizes in a pellet softening reactor (same simulation, but we visualised the classes separately)

Example 3: Systems with random solids (2-phase)

The wastewater industry is full of such examples. But also drinking water treatment faces solids, such as (particulate) or flocculated natural organic matter. One wastewater example is a sludge blanket (Figure 6, left). Another example is primary solids in an A-stage process (Figure 6, right). In both cases, we want to know what the solids do, as they impact process performance in different ways.

Figure 6: the application of 2-phase CFD (water/solids) for the design of two different wastewater treatment technologies

Some plants (especially in the US) ferment activated sludge. For that specific process, the questions we are answering are:

• Where does sludge accumulate, and is it well spread throughout the fermenter?
• Which mixers to apply, and how to operate them (eg intermediate mixing) to mix up the sludge?

For the A-stage process, the questions were as follows:

• How to design inlet and outlet structure as such that sludge washout is prevented?
• How to equally distribute influent through upflow sludge blanket?

Here's a drinking water example: Figure 7 shows the transport of organic and inorganic particles in a large surface water storage basin. We introduced 20 different classes (sizes, densities) and studied what their risk was for reaching the outlet. The model was used to optimise the inlet location. We used similar modelling approaches for DAF and sand filtration.

Figure 7: how particles with different sizes and densities are transported through a large surface water storage basin (ca. 3Mm³)

Conclusions

Often, water practitioners have experience with 1-phase CFD modelling. This article shows only a few out of many examples where multiphase CFD modelling was applied to solve practical questions in the drinking water and wastewater fields. Which phases to include (1, 2 or 3-phase) depends on two factors:

1. The specific project objective (e.g. 'do you want to quantify solids wash out' vs 'do you want to know how flow distributes')
2. The specific technology (e.g. a fluidized granular bed cannot be modelled using 1-phase only given its impact on the water flow)

Hence, we want to pick the simplest approach, but not an oversimplified one. Today's computational power has removed barriers with regard to multiphase that still existed 5 years ago.

Related blog posts

• The value of CFD for the water industry
• Saving costs and improving effluent by knowing what is happening in your bioreactor - Part I
• Saving costs and improving effluent by knowing what is happening in your bioreactor - Part II
• How to minimise regrowth risk in water towers and large reservoirs by mixing optimisation - Part 1