Key Parameters for Up-Pumping in Fermentation, Part 1

By Thomas Post, Lightnin

Introduction
All axial flow impeller technology in fermentation (as well as in most other processes) is based on down pumping. The impellers circulate the fermentation broth in a downward motion to hold the gas down in the tank. In aerobic fermentation, therefore, the impeller pumps against the flow of the rising air. The shear sensitivity of most fermentations and the tendency of most fermentation broths to coalesce create an environment that does not respond optimally to down pumping. In fact, recent research suggests that high solidity hydrofoil impellers used in an up-pumping mode can increase mass transfer, improve blending and heat transfer, and decrease blend time. In the retrofit of actual systems, yield increases of 20 to 40% have been obtained. To better understand why up-pumping technology offers superior fermentation results, this paper discusses the key parameters to consider when designing a fermentation system and how these parameters are addressed by up-pumping technology.

The application dictates impeller design. Choose
the right impeller for lower horsepower,
to save energy...

...for low shear, or for anything
in between.

Process Objectives
Fermentation represents one of the most complex mixing processes. Like hydrogenation and wastewater aeration, fermentation is a multi-phase mixing application where the transition between two or more phases of matter - liquid, gas or solids - is required to obtain a desired process result. Multiple mixing objectives come into play and interact with the fermentation environment. Optimization, therefore, must address the mixing objectives of this multi-phase application and the interaction between these objectives and the fermentation environment.

Most fermentations involve bacterial, fungal, mycelial or yeast-based broths, all of which are sensitive to shear. This is especially true of more recent mammalian cell cultures and rDNA. In fermentation, the need to disperse gas to enable mass transfer between liquid, solid or gas phases must be balanced with the need to maintain the integrity of the biomass.

Because blending is considered less complicated than mass transfer, it is the most overlooked parameter of full-scale fermentation. In small lab scale fermenters, uniform blending is relatively easy to achieve, but upon scaleup staging is common in which dead zones develop between impellers. Moreover, if the time scale of mixing approaches the time scale of mass transfer, the rate limited factor shifts to blending. In full-scale fermentation optimal blending is essential to maximize yield and productivity.

Mixing objectives during the fermentation process are defined as follows:

• Solids Suspension: to suspend the media and biomass
• Blending: to blend uniformly the broth containing the media, micro-organisms, nutrients and any reagents, as well as to maintain temperature and pH constancy throughout the broth. Blending depends strongly on the mechanical arrangement of the fermenter.
• Dispersion: to disperse the gas uniformly through the broth (and thereby increase gas hold-up), reduce the size of the gas bubbles, and prevent flooding.
• Mass transfer: for dissolving oxygen into the broth (gas-liquid mass transfer), enabling the dissolved oxygen to pass through the cell walls of the microorganisms as well as enable carbon dioxide flux in the opposite direction as the microorganisms metabolize (liquid--solid mass transfer).
• Heat transfer: facilitates removal of heat from cell metabolism and through the dissipation of mixing energy via viscous shear.

Optimizing fermentation must take into account all these concurrent goals because improvement in one of these objectives may affect another goal. For example, mass transfer is often limited by poor blending, dispersion, solid suspension and heat transfer; improvement in gas dispersion in the fermentation broth can increase gas hold-up and in turn may allow for greater mass transfer. Likewise, improvements in heat transfer result in better heat distribution, which in turn improves the environment for cell metabolism (mass transfer).

Mechanical Considerations
All mixing parameters are affected by the mechanical design of the fermenter. The following are basic criteria for mechanical design:

• Z/T: Tank height (actually ungassed liquid level) to diameter ratio.

  The Z/T ratios of most fermenters range from 2:1 to 4:1, and in the case of airlifts can range as high as 10:1. Uniformity and a well mixed tank are more easily achieved in short squat tanks. Tall tanks often used in fermentation tend to create individual mixing zones within the tanks that can significantly reduce blending.

• D/T: Impeller diameter to tank diameter ratio, used to describe impeller sizing.

  Smaller D/T impellers need to spin faster relative to larger D/T impellers to impart the same power required for mixing. Increases in impeller speed, however, increase fluid shear within the tank and that shear can be detrimental to the microorganisms in the broth.

• Impeller type

All the power transmitted from the impeller to the fluid goes to flow and shear. The shear and turbulence generated by the impeller represent energy loses. For a given amount of power, there is a huge spectrum of impellers that generate varying amounts of flow and shear. Different processes obviously require different impellers and impeller combinations. As with the introduction of the up-pumping A340 impeller, much of the technical innovation in mixing comes from advances in impeller technology.

• Heat transfer surfaces

Helical coils, vertical plates or vertical coil bundles are placed between the impeller blade tips and the tank wall to regulate the temperature of the broth. Placement of these surfaces impacts the flow pattern developed by the fermenter.

• Baffles

Baffles reduce the inefficient tangential velocity component of all impellers to produce a more efficient radial or axial flow. Positioning and design of baffles along with the placement, type and design of heating coils affects the flow patterns and thus overall process results.

If fermentation is kinetically or reaction rate limited, no change in the mechanical configuration of the mixer will improve process results. In most cases, however, fermentation purity and yield change with the mixer configuration. Specifically, yields change on scale-up, are different between tanks of approximately the same volume, and are different in the same tank with different impeller configurations. Only when fermentation and yield are the same regardless of the size or kind of the equipment or the location of equipment in the tank is the fermentation limited by the reaction rate and not by some mixing parameter. This paper assumes that the fermentation is not kinetically or reaction rate limited, and indeed this reflects the vast majority of actual fermentation installations.

The Effects of Gas

Flooding
If a mixer is flooded, the gas sparged into the tank controls the flow pattern in the fermenter rather than the mixer, and that flow pattern is random and ineffectual. The power of the mixer (defined as power per unit volume or P/V) relative to the isothermal expansion power of the gas (IEPG/V) determines whether a mixer is flooded. If IEPG/V is equal to or greater than the power of the mixer, then the gas controls the flow pattern and the impeller is flooded. In this case the flow pattern is random and chaotic, and if solids are present they will stay at the bottom of the tank. Geysering will be readily visible.

On the other hand, once the mixer power is 2-3 times greater than isothermal expansion power of the gas, the gas becomes well dispersed above the impeller. Although gas is dispersed to the sides of the tank walls, there is no distinct flow pattern below the impeller and most solids will still collect at the tank bottom. In this case, however, the liquid surface at the top of the tank appears smooth and the gas appears well dispersed, creating the illusion that the impeller is not flooded and that the mixer controls the flow pattern.

It is not surprising, then, that a vast majority of fermentation operations operate sub-optimally. In fact, the majority of fermentations that Lightnin has been asked to retrofit are actually flooded.

Only when the power of the mixer is at least 3 times greater than the isothermal expansion of gas does the mixer control the flow pattern and there is no flooding. In this case, the gas is completely dispersed throughout the tank and any solids present will follow the flow pattern created by the mixer. Only under these conditions is optimization possible.

Note that this applies to non-coalescing systems (where bubbles do not tend to aggregate to form larger bubbles). As we will discuss later in this paper, however, most fermentations are coalescing systems (where gas bubbles tend to converge to form larger bubbles). For coalescing systems, the P/V requirements are five-fold greater relative to the IEPG/V in order to prevent flooding. This may explain why so many full-scale fermentations are flooded.

Because A340 impeller technology creates an upward flow pattern that pumps with ,rather than against, the gas, up-pumping A340s do not flood at any power level or gas rate. The mixer rather than the gas is therefore always in control of the flow pattern regardless of whether the gas is coalescing or not.

Gas Holdup
When gas is introduced into the tank and is dispersed by the impellers, the liquid depth in the tank increases. This is known as gas hold-up and is defined by the difference in tank volume between the gassed and ungassed states of the fermentation. Normally, greater gas hold-up is associated with greater mass transfer because of the increased gas-liquid interface.

Gas hold-up is greater using A340s than in conventional designs for several reasons. Because the gas is pumped upward, it is mechanically sheared each time it passes through an impeller zone. This decreases bubble size and increases the gas-liquid interface. In addition, the strong radial flow of the uppermost A340 is so great that significant quantities of air are induced from the headspace. Even without sparge air, A340 impellers can induce air from the headspace at levels up to 2 vvm (volume of gas in scfm divided by the volume of the liquid per minute) at 1 kW/m³. Moreover, in experiments at Lightnin the standard aeration efficiency (as measured by the oxygen transfer rate in pounds of oxygen per horsepower - hour) of multiple A340s was found to be 5.5, almost double traditional standards.

In Part 2 of this paper we will cover K factor (the ratio of gassed horsepower to ungassed horsepower), mass transfer coefficient, shear rates, and coalescence as they relate to up-pumping in fermentation.

For more information: Thomas Post, vice president of technology, Lightnin, 135 Mount Read Blvd, Rochester, NY 14611. Tel: 716-436-5550.