Mixing (process engineering)

In the field of industrial process engineering, mixing is a fundamental unit operation. This process involves the manipulation of a heterogeneous physical system. A heterogeneous system is one where the different parts are not spread evenly. The goal of mixing is to make this system more homogeneous. A homogeneous mixture is one that is the same throughout.

Mixing allows for the transfer of heat or mass between different streams, components, or phases. Modern industrial processing almost always requires some form of mixing to function correctly.

The mechanism of mixing depends on the state of the materials involved. When mixing powders, engineers look at two different dimensions: convective mixing and intensive mixing. Convective mixing occurs when material is transported from one location to another within the mixer. This process distributes components over one another until they reach a randomly ordered state. For free-flowing and coarse materials, this is often sufficient. However, if materials are cohesive, they form lumps. To break these lumps, more energy is required through impact forces or shear forces. This process of breaking up lumps is known as deagglomeration.

Liquid-liquid mixing is another common operation. The equipment used depends on the nature of the liquids being blended. Single-phase blending involves liquids that are miscible, meaning they can dissolve in each other. An everyday example is adding milk to coffee. These liquids often have low viscosity, so the momentum of the liquid being added can cause enough turbulence to mix them. In contrast, blending in a viscous liquid like honey requires more mixing power per unit volume. Multi-phase mixing, where two liquids do not dissolve in each other, requires high-shear, low-flow mixers. These mixers create droplets of one liquid within the other.

Solid-solid mixing is one of the oldest unit operations in solids handling. Engineers use dry blenders to blend multiple dry components until they are homogeneous. These blenders can be used for pharmaceuticals, foods, chemicals, and cosmetics. They range in size from small half-cubic-foot laboratory models to massive 500-cubic-foot production units. Blending times for dry ingredients are often quite short, typically between 15 and 30 minutes. However, this time depends on the percentages of each component and their differences in bulk density. Mixing can be done in batch mixers or through more complex continuous dry-mix processes.

Liquid-solid mixing is used to suspend coarse solids or to break up fine lumps. For example, mixing granulated sugar into water involves suspending particles through the bulk motion of the fluid. When mixing fine powders like flour into water, the mixer must use a high shear field to disintegrate the lumps. In industry, concrete mixing is a vital liquid-solid process. It involves commingling cement, sand, gravel, and water into a homogeneous mass. Another specific goal is solid suspension, which improves mass transfer. This is often done by using axial-flow impellers to create momentum.

Engineers use specific measurements to track how well a solid-liquid suspension is working. They use the Relative Standard Deviation, or RSD, of the solid volume fraction. A perfect suspension would have an RSD of 0%. In many practical cases, an RSD of 20% or less is considered sufficient for homogeneity. Because measuring full-scale tanks is difficult, engineers often measure at a small scale. They then use a "scale-up" criterion to predict the results for a larger tank. They can also use computational fluid dynamics software to perform these calculations.

Understanding mixing requires complex mathematics. The math of mixing is highly abstract and is a part of ergodic theory and chaos theory. To estimate stirring speeds, engineers use correlations. One famous correlation was published by Zwietering in 1958. It provides a crude estimate for "bad" quality suspensions where particles do not stay suspended. Other researchers, such as Mersmann in 1998, have provided different correlations. More recently, machine learning has been used to build models that are even more accurate than these classical methods. This mathematical foundation allows engineers to design reliable equipment for global industries.