Scaling Smarter: How Advanced Modeling Accelerates Chemical Process Scale-Up

Process scale-up is the effort to take a chemical process that works at a smaller scale and design a version that will work similarly well at a larger scale. Oftentimes, the smaller process is in a laboratory and exhibits excellent product yields in a reactor measuring less than one inch in diameter; meanwhile, the larger process is envisioned for a chemical plant and may involve a reactor measuring something like 10 feet in diameter. However, one cannot simply multiply all the lab reactor dimensions and flow rates by a scale factor to yield an equivalently performing commercial reactor. The impact of scale is much more complex and not amenable to a “pinch-to-zoom” approach.

What is the reason for this complexity? In short, it is the interplay between

• transport phenomena (or hydrodynamics), including mass, heat, and momentum transfer; and
• reaction kinetics

and the fact that this interplay changes with scale.

Chemical reactions consume feed molecules and generate product molecules; they also either absorb or release heat, depending on the reaction. The rate at which a reaction can occur, accordingly, is generally influenced by the rate at which mass and heat can be moved in and out. This turnover is achieved principally by advection – that is, mass and heat are carried by fluid flow, which is in turn governed by momentum transfer within the fluid. These transport phenomena are all inherently dependent on scale – as examples, consider that diffusion takes more time over longer distances, and that turbulent eddies in a pipe flow grow larger with increasing pipe diameter. As a result, it is the scale-dependent coupling of reaction and flow that makes scale-up so complex.

This complexity, in combination with the sizable financial risk a company faces when building a new chemical complex (which may cost $100 million or more), is the reason that process scale-up is typically a slow and expensive step in commercializing a new process technology. Historically, companies attempted to mitigate scale-up risk using a brute-force (“empirical”) approach outlined in Figure 1:  building a series of, say, 3 or 4 intermediate-sized reactors in between the laboratory scale and envisioned commercial scale. For example, this series might include a small pilot plant followed by a large pilot plant and finally a demonstration plant. The hope with this approach is to reduce the amount of extrapolation – and the associated risk – en route to the commercial scale. However, this approach is very slow and costly and, ironically, not very effective since it does not consider the fundamental causes of scale-up gaps [1].

Figure 1: Traditional (“brute force” or “empirical”) scale-up approach

So, is there a way to study the impact of scale on reacting flow without building so many plants? Yes, via physics-based computer modeling.

There are two types of modeling particularly helpful for scale-up.

• Computational Fluid Dynamics (CFD) Modeling: This is a numerical approach to solving the fundamental equations for fluid flow and heat/mass transfer. If properly constructed, CFD modeling can predict hydrodynamic behavior at different scales with reasonable accuracy. The output of these simulations includes the velocity, temperature, and phase fraction at every “location” (computational cell) in the system—that is, mapped out in three dimensions. Because of this high level of detail, CFD modeling can be expensive and is usually reserved for probing behavior in key design scenarios but not for scoping or parametric study.
• Phenomenological Modeling: This is a broad term encompassing many types of system-level models that are reduced-order but still fundamentals-based. Instead of resolving 3D profiles, this approach relies on simplifying assumptions about the geometry and flow field. For example, a continuously stirred tank reactor (CSTR) might be assumed to have uniform temperature and concentration, eliminating the need for any spatial resolution. Since they are simpler than CFD, phenomenological models are much faster to run and as such are appropriate for process scoping. However, on their own, they generally require more model tuning to match plant data.

Model-assisted scale-up is a modern approach that addresses the issues of empirical scale-up by complementing the experimental pilot work with the above modeling tools to capture the underlying phenomena giving rise to scale-up gaps. As shown in Figure 2, a typical scale-up plan could look as follows:

• a pilot plant to obtain base catalyst and process data;
• a computational fluid dynamics (CFD) model to probe hydrodynamics at pilot and commercial scales; and
• a phenomenological model utilizing simplified hydrodynamics (informed by the CFD simulations) and detailed reaction kinetics to predict product yields.

Each of these three tools is then iteratively improved on the basis of lessons learned from the other two. For example, after an initial pilot plant design is developed, an initial CFD model can be run to test whether flow behavior will be as intended and make design corrections accordingly. The CFD model can also predict a residence time distribution or other flow-field data to inform the phenomenological model. After its design is updated, the pilot plant can be constructed and run to generate data that validate the CFD model and the phenomenological model. The validated phenomenological model can then be run for a wide range of conditions (since it is cheap and fast, unlike the pilot plant and CFD) to identify additional conditions of interest to run in the pilot plant.

Simultaneous to this pilot-plant-focused study is a study of the commercial plant. An initial design (encompassing process and high-level equipment design) is developed and then tested in the validated CFD and phenomenological models. As differences in flow and reaction behavior across scale are encountered, design modifications can be made iteratively to move the commercial reactor towards its target performance. Then, after the pilot plant and commercial studies have concluded, the commercial reactor design can move forward to FEED (front end engineering and design) and eventual construction and start-up.

Figure 2: Model-accelerated scale-up approach

Notice that in this example there is only a single pilot plant and no demonstration plant. This reduction in physical testing scope significantly reduces timeline and cost, leading to faster and cheaper commercialization.

If you are working on scaling up a new chemical process, be sure to harness the power of advanced modeling, including CFD. If in need or in doubt, reach out to an expert in chemical process fluid dynamics and scale-up to understand how modeling can help you.

Reference
[1] Harmsen, Jan. Industrial process scale-up: a practical innovation guide from idea to commercial implementation. Elsevier, 2019.