Falling Films in Vessel CIP Applications

Applying an understanding of flow regimes and turbulence to clean-in-place (CIP) processes can optimize CIP cleaning efficiency, reducing water, solvent, energy and chemical use, while maintaining cGMP cleaning performance

For vessels and bioreactors in bioprocessing applications, designing clean-in-place (CIP) cycles that are optimized for lower flows, shorter times and reduced chemical usage are sought-after optimizations for both aqueous and chemical-solvent cleaning processes. Although CIP flow requirements are well established through industry guidelines and rules of thumb, examining the origins of these criteria can reveal opportunities to tailor cleaning cycles for improved efficiency in water, solvent, energy, and chemical usage.

Coverage criteria

The coverage criteria for CIP vessels is traditionally defined by a standard flowrate per length of vessel circumference. The origins of the coverage criteria are explained in Ref. 1. The first use of a coverage criteria expressed as a flowrate per unit width was with dairy silos in the early 1960s. Silos with aspect ratios that are taller and narrower than a typical bioprocess vessel could generally be cleaned with 2.0 gal/min per ft of vessel circumference. These vessels did not have many top head penetrations, so as long as a spray cleaned the top head of the vessel, the walls would also be cleaned by the sheeting action of the liquid flowing down the side.

The recommended flowrates increased to accommodate the greater complexity and penetration count of bioreactor heads. An additional 1.0 to 1.5 gal/min of flowrate was added per vessel penetration, and 2.0 gal/min per ft of manway circumference was recommended, in addition to the 2.0 gal/min per ft of vessel wall. For a 5-ft-dia. vessel with a 2-ft manway and 10 vessel penetrations, the effective overall flowrate is about 3.6 gal/min per ft of circumference.

These wetting rates align with the guidelines for static-spray-device requirements in the American Society of Mechanical Engineering (ASME; New York, N.Y.; www.asme.org) BPE-2026 Bioprocessing Equipment standard [2].

Reynolds numbers in falling films

Given the solution’s flowrate and physical properties, it is possible to calculate a Reynolds number and gain insight into the flow characteristics of the falling film. The Reynolds number is defined by Equation (1):

The diameter D must be converted to the hydraulic diameter to apply appropriate pressure and friction forces to the liquid. The hydraulic diameter is four times the cross-sectional area divided by the perimeter. (This formulation assures that the hydraulic diameter of a liquid-full pipe is equal to the actual diameter.) For a falling film, the cross-sectional area is defined by the difference between two circular areas (Equation (2)).

 

The diameter of the inside circle is equal to that of the outside circle minus two film thicknesses (Equation (3)).

 

The film thickness squared is eliminated because it is assumed to be insignificant. After simplifying, the equation becomes Equation (4):

Thus, the Reynolds number for this system is shown in Equation (5):

The Reynolds Number may be calculated to characterize the flow against the wall if the volumetric flow per unit wall width equals the average velocity over the film thickness.

Fluid properties of 20°C water flowing at 2.0 gal/min per ft may be substituted into the formula above to determine the Reynolds number of the cleaning liquid flow down the inside wall of the vessel, as shown in the caculations below.

There are three flow regimes defined for Reynolds numbers [3]:

• Laminar flow with negligible rippling corresponds to Re less than 20

• Laminar flow with pronounced rippling corresponds to Re between 20 and 1,500

• Turbulent flow corresponds to Re greater than 1,500

Based on the Reynolds number values in this case, it is apparent that the flow has turbulent characteristics. An astute observer will realize that the threshold values separating the flow regimes differ from those commonly seen on a Moody Diagram (Figure 1) for fully developed flow inside a circular pipe. While a hydraulic diameter will correlate pressure forces on a cross-section of flow with the friction forces on a perimeter of flow, it is not adequate to correlate the degree of turbulence across different geometries. Thus, the flow regimes, determined by experiment, have different numerical values in the different geometries.

While the numeric values of the Moody Diagram do not apply to this geometry, we can still appreciate that the degree of turbulence becomes insensitive to the Reynolds number once complete turbulence is reached (as observed in the upper-right portion of Figure 1). The viscosity of water at 70°C is 0.4035 Cp, and the density is 0.9778 kg/L. Under these conditions, Re = 4,011. The viscosity of acetone at 25°C is 0.306 Cp, and the density is 0.789 kg/L. Under these conditions, Re = 4,268. In each of these conditions, the observed turbulence is as chaotic as the flow of water at 20°C.

Cleaning at turbulent flow

The cleaning behavior of falling liquid films has been studied under different conditions [4]. These studies indicate that turbulent flow has better cleaning results than laminar flow. The analysis of the Moody Diagram suggests that for a given level of debris on the sidewall, analogous to surface roughness, the friction factor is insensitive to increases in the Reynolds number after passing the threshold of complete turbulence. Since the friction factor indicates resistance to flow, it also indicates a measure of energy loss.

In summary, energy is supplied to provide fluid flow and overcome interactions with the rough surface, but once the flow is turbulent, additional energy (flow) does not cause a more turbulent interaction with the surface. Theoretically, additional flow does not improve surface cleaning at fully turbulent flow. There are legitimate reasons to flow at higher than minimally turbulent conditions. Additional flow is often required to ensure complete coverage of pipeline tees; however, these conditions do not apply in free surface flow.

Designing CIP cycles to be optimized for lower flows, shorter times and reduced chemical usage are sought-after optimizations in both aqueous and chemical-solvent cleaning processes. Are there flow optimizations that can be applied to vessel solvent cleaning as is often used in active pharmaceutical ingredient (API) manufacturing? There are a few challenges to implementing lower flowrates on a cleaning cycle.

First, most CIP cycles will include a water circuit to remove solvents to permit vessel entry and maintenance. It is crucial to ensure complete coverage when removing solvents to comply with health and safety requirements. In addition, the targeting and validation of the spray devices are often accomplished with water and riboflavin. Since the validated flowrate cannot be correlated to a reduced flowrate simply based on the Reynolds number, the flow used during CIP cycle testing must be maintained. Also, while additional solvent volume may not improve turbulence, it may aid in the diffusion of soils from near the surface to the bulk flow. For these reasons, and potentially others, it is best to stick to the fundamentals of cleaning chemistry and cleaning device design to ensure that the minimum flowrate, time, temperature and chemical that satisfies the cleaning endpoint is specified for cGMP processes.

Edited by Scott Jenkins

References

1. Seiberling, D.A. (Ed.), “Clean-In-Place for Biopharmaceutical Processes,” 1st ed., CRC Press, 2008.

2. American Society of Mechanical Engineers, Bioprocessing. Equipment, ASME BPE-2026, international standard, (Revision of ASME BPE-2024), Section SD-3.9.2.1, www.asme.org.

3. Bird, R., Stewart, W. and Lightfoot, E., Transport Phenomena. 2nd ed., John Wiley and Sons, New York, 2002.

4. Fuchs, E., Boye, A., Murcek, R., and Majschak, J.P., An experimental comparison of film flow parameters and cleaning behaviour of falling liquid films for different Tilt Angles, Food and Bioproducts Processing, 93, 318–326, 2015.

Author

Howard G. Sneider is a principal process/specialty engineer with Fluor Corp. (6700 Las Colinas Blvd., Irving, TX 75039; Phone: +1.864.281.6278; Email: howard.sneider@fluor.com). He has more than 25 years of experience in biotechnology, pharmaceutical and advanced manufacturing process design. Sneider specializes in unit operations, facility design and large-scale manufacturing systems. Sneider holds an M.S.Ch.E. from the University of Connecticut and a B.S.Ch.E. from Columbia University.