Mixer CFDs: A powerful tool for improving mixer design and performance


Key points:

Mixer CFDs: A powerful tool for improving mixer design and performance

"Mixing is like alchemy," was the plant manager's reply when I said I was joining a mixer company. After working in the mixing industry for ten years, I see why outsiders may think that. As an insider, I have more visibility on the design rules and envelopes that guided designers, ensuring their designs would work.

Given mixers ubiquity in industry, there is extensive research, experience and knowledge available. Much of this has been gained through test-work, plant experience and many mistakes. CFDs are a powerful tool assisting in improving mixer design, performance and undestanding. Ultimately, with the goal of adding value to the process.

Lets first provide a brief introduction to mixers and then briefly discuss computational fluid dynamics (CFD) before we start getting into the nuts and bolts of mixer CFDs and the value that they can add.

Mixers, Stirrers and Agitators

Mixers, stirrers and agitators are generally all the same and consist of one or more impeller on a shaft that rotate within a tank, reactor or basin. The rotation of the impeller creates motion in the tank. How much motion is dependent on the type of impeller, impeller rotation speed and liquid properties. The type of mixer will depend on the duty required for that particular process. Whether the mixer is called a mixer, agitator or stirrer tells alot about the kind of duty that is expected of the machine and how it is thought of within the process.

Mixers are the unsung workhorses of almost every industry where fluids have to be handled. Which is just about every industry. Below is a list of some industries where mixers are used.

Clearly, mixers are used in many industries for different many different processes. It is therefore beneficial to understand their behaviour in greater detail.

Computation fluid mechanics

Computational fluid mechanics forms a branch of fluid mechanics. Many real world fluid dynamics problems can't be solved analytically and therefore are solved computationally. The process of solving the fluid dynamic problems computationally or numerically is called computational fluid mechanics (CFD). It involves breaking down the liquid domain into mesh and solving the equations of motion for the fluid. Typically, the Navier-Stokes equations are solved, which is a set of partial differential equations and requires initial and boundary conditions to find a unique solution. The fluid domain is divided into millions of mesh cells, the more cells the greater the accuracy. Given the vast number of cells and the number of equations solved, the simulation is normally solved on multiple computers.

Once a simulation is completed, it provides detailed velocity, pressure and turbulence information that is analysed to understand the behaviour of the equipment and the fluid. Depending on the simulation, other parameters like temperature may also be included in the simulation. The goal of performing a simulation is to understand the existing fluid flow and to improve the performance the equipment or part. It is this potential optimisation that we find so exciting. CFD can be applied anywhere where fluids are involved but has found widespread use in the aeronautics and automotive industries. It can be used for mixers too as we will discuss below.

Why is a mixer CFD essential?

Like any CFD, the purpose for a simulations has to be known and defined. If these are not known, then CFD just becomes an expensive way to create pretty pictures. Like in any business, the goal must ultimately be to make more money, more efficiently and these days in a more environmentally friendly and ethical way.

Since mixers are often involved in production processes, the simulation of a mixer tank, vessel or reactor should create at least some of the following benefits.

The question is then: what possibilities does a CFD simulation open up to achieving some of the above goals. To answer this question one needs to consider the different relationships that you have to the mixer. Are you the R&D engineer, sales engineer or operator?

An operator is not intimately involved with the design of equipment, but could require improvements when production is below expectations. This is common when the mixer duty has changed due to new products or processes. A CFD study can determine the nature of the problems specific to the reactor and provide recommendations. This may include the evaluating the settling out of precipitates or particles, excessive mixing damaging flocculants or shear sensitive products, excessive blend-times or insufficient reagent contact.

An R&D engineer may want to test new impeller designs or configurations without needing to do detailed tests for every design permutation. CFD allows many iterations to be tested rapidly with greater insight than can easily be achieved with test work.

A sales engineer might need to convince a customer that a new design will meet the required duty. This may occur when new impeller draw less power than the customer expects. This is related to the issues that the operator may have that were listed above. This is useful to the sales engineer as it provides some credibility.

In the sections that follow we discuss the insights that CFD can give into the mixing process. We have only chosen a few as it depends on the nature of the investigation.

Evaluate the velocities for deadspots, low flows and recycle loops

To troubleshoot underperforming mixers, the velocity in the tank should be evaluated. This will show where dead-spots occur, particles settle out. Impellers are sized and located do general rules. However, unusual tank sizes and aspect ratios or high required input powers can result in unusual configurations. With CFD, these are easy and cost effectively evaluated. The cost of the CFD before supplying the equipment is less than the cost of trial an error when the plant is commissioned or the cost of lost produciton.

This flow pattern analysis allows impeller configurations, number of baffles, positions of sparge ring and inlets and outlets to be optimised. These have implications on the degree of mixing but also the residence times and gas dispersion. We will discuss this below.

Impeller pumping and power characteristics

Characterizing new impellers is an important step in the design of efficient and effective mixers. For existing impellers these are well established. For new impeller designs, CFD provides a convient method for evaluating these without the need for complex equipment. Furthermore, impellers are generally tested in flat-bottomed tanks on their own. The performance of an impeller working in conjunction with a second mixer in the reactor or in alternative tank types is generally unknown. Characterising and understanding these configurations can give better optimised designs. This is a massive opportunity for improving the performance of a reactor that can easily be exploited with CFD.

Process characteristics

In this section, we will look at process characteristics such as shear rates and turbulent dissipation. Its important to consider the process results as a slight improvement in production yields may be worth more than electricity savings. While there are many processes and their duties are all unique, we will look at shear rates, turbulent dissipation and residence time.

Shear rates

Shear rates are typically important in the food, pharmaceutical, and chemical industy. To understand the shear rates in a vessel, it is only really possible to measure the outcome. That is to say, the quality and consistency of the final product. Alternatively, CFD allows the shear rate to be evaluated at every location throughout the tank. This can be used to better understand the outcome of the test-work rather than guessing at the mechanism.

Understanding shear rates in mixing vessels is important as it affects the product quality, yield and ultimately the bottom line on the process.

Turbulence

Turbulence is key in efficiently mixing a reactor. This is due to the turbulent eddies transferring momentum between flow streams instead of simply relying on a liquid viscous shear stresses. Although its clear that turbulence is an important, it is a complex topic an is typically modelled in engineering CFDs. In modelling turbulence, a parameter that frequently comes up is turbulence dissipation rate.

The turbulent dissipation rate is the rate turbulent energy is lost to the system via turbulence and is closely related to the mixer input power. This is why some mixer applications specify a specific input power - it is an indirect way of specifying the degree of turbulence in the system. Intuitively, this makes sense. When we add sugar to tea, stirring improves the rate at which the sugar dissolves. The general bulk motion and in addition to the turbulence have improved the rate of the desired reaction, i.e. to dissolve the sugar. Furthermore, the turbulent energy dissipation rate also appears in some models for mass transfer in gas dispersion applications, further highlighting the utility of interpreting turbulence in a mixer application.

In analysing a new configuration, reactor or impeller, often only the velocities, power and pumping are considered. Only considering these does not provide a full picture of the reactor and the turbulence distribution in a reactor should also be considered as it indicates where reaction are likely to occur.

Residence time

The residence time for a product in the reactor or the time for a product to adequately disperse through the reactor. This is ultimately the goal of a reactor and understanding the reactor performance in terms of this is extremely useful, both from an operator and mixer design point of view. Residence can be evaluated by introducing a passive tracer in the domain. This can be thought of as a dye that is allowed to disperse as the mixer mixes. The rate at which the dye disperses can be controlled in the simulation which provides a means of modelling the dispersion of different products with unique characteristics. The use of scalars can also show if there is short circuiting of product between the inlet and the outlet which allows the position to be optimised. Similarly it is possible to determine if the mixer is able to uniformally disperse the product. It should be clear that any of these investigations can lead to improvements in the performance of the reactor. It seldom happens like this is life, but fortunately modelling a scalar is also computationally cheap.

Multi-physics

A single phase simulation often gives a very good idea of the the reactor behaviour at a low computational cost. Consequently, it should not be overlooked as a starting point for understanding, interrogating, investigating real reactors. Real reactors however are never single phase and therefore further physics should be modelled. Over an above being multiphase, heat transfer, mass transfer, and reactions can also occur. Although each of these dimensions adds complexity, they also add additional opportunity for optimisation.

While modelling all the physics is possible, the complexity and solve time also increases. The approach should always be to start with the minimum model and build up complexity until there is sufficient confidence to make an improvement. If you are considering investing in a simulation, you should consider what details are important to consider and what details can be ignored. Some details are more obvious than others. For example, including bolting details on impellers are unlikely to greatly influence the overall flow pattern. However some details are more subtle. For example in a continuous reactor, it may not be necessary to include the inlet and outlet flow rates if these are small relative to the flow generate by the impeller. Which details to include and exclude will be problem specific.

Scale up

Following from the quote, "Mixing is like alchemy", we've heard, "Scaling is an art". This is not only relevant to mixing however. We are certain that process engineers will agree for most of the equipment that they need to consider. CFD could probably help there too though.

Scale-up in mixing is interesting because there are so many scale-up parameters such as equal tip speed, equal specific power, equal specific torque, equal Froude number or even in between depending on the application. While knowing which scale-up factor to use is part of the skill in designing mixer, CFD can help improve the confidence that the correct scale-up is correct.

Why not just test?

CFD is not a silver bullet and it is important to validate that the setup is correct and the correct sub-models are chosen. This is especially true when when multi-physic and multi-phase simulations are performed. Experience with the process an equipment can also help evaluate if a CFD makes sent. Once a model is validated for a reactor, impeller, or configuration that is known, there can be greater confidence that the new model that goes outside of the existing knowledge and experience, is correct.

There are of course limitations to test work and this is one of the places where CFD shines. Setting up a good test program is difficult and should never be underestimated. With reference specifically to mixing, CFD simulations are able to address the folling issues.

Talk to us

We would love to hear what you would like to accomplish with your reactors. We have developed mixer template simulations that we use for our analyse and we are sure can help you get the most out of your mixers.

Request a Meeting