Andrew Creathorn is the Director of the Customer Service and

Support Team at Lightnin Mixers, in Rochester, New York. In his

current position he has responsibility for order related engineering
functions and product development. Mr. Creathorn joined Lightnin
in 1991, as a Project Engineer, and has 11 years of direct
involvement in the mechanical design and application of mixing
equipment. During this time he has designed numerous agitators
for a wide range of process requirements, and successfully led
initiatives to change Lightnins approach to mixer design to take
advantage of developing technology.
Mr. Creathorn received his B.Sc. degree (Mechanical
Engineering, 1987) from Teesside Polytechnic, UK, and spent four
years with a centrifugal pump manufacturer in a design
engineering capacity, prior to joining Lightnin.
ABSTRACT
Industrial agitator or mixer configurations and associated appli-
cations present the design engineer with a unique set of challenges.
This was especially true when a major chemical company required
an agitator for a large multistaged column. A traditional approach
to mixer design, in this case, would attract high vessel and agitator
cost due to the nature of the machine design and the associated
mounting/operational loads.
This paper reviews the design constraints and problems, along
with the solutions developed to provide the end user with
equipment that is easily maintainable, and had an attractive capital
cost. In particular, it compares the design solutions with the
traditional design approaches, and discusses how finite element
modeling was used to verify operational limitation due to natural
frequency constraints. The paper concentrates on the key
component associated with the mixer shaft, and specifically
excludes discussion relating to the design and feature requirements
relative to gearbox selection.
INTRODUCTION
In 1997 a large project was funded to produce an oil for use in
food production. Part of the process necessitated two large agitated
columns. As with any new project, delivery and price were key
drivers in the decision making process, but mechanical reliability
and installation/maintenance were also identified as equally
important. The process requirements demanded extremely violent
agitation dispersing sparged gas at rates in excess of 10 times those
seen in usual high gas rate applications.
The application of the mixer, in the extreme operational
environment, demanded that the design needed to employ a high
degree of conservatism. However, traditional approaches used in
the design of column agitators were unattractive given the ease of
maintenance requirements, the physical limitations of the site, and,
of course, the equipment cost.
With this in mind, each critical area of the design was
undertaken using a clean sheet of paper approach.
DESIGN CONSTRAINTS AND OVERVIEW
The column requiring agitation consisted of six equal stages.
Each stage required agitation using a single radial flow impeller.
The basic layout is shown in Figure 1.
Figure 1. General Arrangement of the Mixer Vessel System.
83
DESIGN CONSIDERATIONS FOR A
LARGE MIXER USED IN AN AGITATED COLUMN APPLICATION
by
Andrew Creathorn
Director of Customer Service and Support Team
Lightnin Mixers
Rochester, New York
The application demanded a vessel that was 90 ft long, which
resulted in a shaft length of 88 ft and a rotational speed of over 100
rpm. (Note: The total height of the assembly was over 102 ft.) In
addition, the shaft was required to transmit 1000 hp. Despite the
resultant 9 inch diameter, a review of the natural frequency of the
shaft, and the bending loads associated within the overhung
condition, it was determined that the shaft required support in two
places via steady bearings.
The application required that the vessel had a design temperature
and pressure of 400F and 150 psi, respectively. This requirement
demanded the use of a double mechanical seal. The seal
components and barrier fluid were required to be FDA approved.
Minimal leakage was specified between stages, and this
requirement necessitated the design of a stage seal. The design of
this seal was complicated by the requirement that the seal
components be replaced without removing the shaft. (It is
obviously not practical to remove a shaft 88 ft in length.) This
requirement to replace wear parts also needed to be addressed in
the design of the shaft/vessel seal and the two steady bearings.
TYPICAL MACHINE DESIGN
REVIEW AND DISCUSSION
Traditional mixer design involving mechanical seals typically
incorporate the motor/gearbox/seal combination into a single
machine. This machine would then be mounted via a mounting
flange to the vessel. This typical arrangement is shown in Figure 2.
The obvious advantage of this design is that the location of the seal
relative to the gearbox shaft can be carefully controlled. In
addition, a traditional mixer seal cartridge design can be employed
to seal the mixer shaft.
Figure 2. Traditional Mixer Arrangement for Sealed Vessels.
The traditional mixer seal cartridge is shown in Figure 3. The
seal cartridge has a fixed bearing to accommodate both the axial
load, due to the shaft down weight, and the radial load resulting
from the mixing action. The upper radial load is reacted into the
gearbox upper bearing.
Obviously, this design approach offers the user a reliable mixer
with a proven machine design satisfying the requirement for ease
of installation. However, the design is expensive, requiring
multiple shaft connection points (flange couplings) and a relatively
complex seal cartridge design. Moreover, due to the load acting
Figure 3. Traditional Mixer Seal Cartridge.
directly on the vessel nozzle, the design significantly increases the
cost of the vessel. This approach also necessitates a difficult seal
cartridge change that involves raising the entire shaft system to
expose the in-tank coupling. Figure 4 shows this typical sequential
seal change procedure.
Figure 4. Sequential Seal Change for Traditional Mixer Seal
Cartridge.
During the preparation of the bid, given the unusual nature of the
design, it was determined that if a nonconventional solution were
to be offered, a partnership with a vessel manufacturer would be
highly desirable. This was driven primarily by the obvious
requirement of removing the operational mixer loads from the
vessel head to the vessel shell, a common practice for mixers that
do not require mechanical seals.
Two alternative design solutions were developed using configu-
rations that allowed the mixer loads to be redistributed into the
PROCEEDINGS OF THE TWENTIETH INTERNATIONAL PUMP USERS SYMPOSIUM 2003 84
vessel shell. This alternative design significantly reduced the cost
of the vessel, allowing the vessel manufacturer to design the vessel
head to accommodate the pressure and temperature loads only.
At this point, a decision had been made to move away from a
conventional seal cartridge design and move toward a seal on a
sleeve type approach. Elimination of the bearing from the
assembly would allow the reacted loads to be transmitted directly
into the gearbox/machine assembly and further reduce the weight
of the seal arrangement.
Figure 5 shows the first alternative design considered. This
arrangement allows the loads to be removed from the vessel head.
It also accommodates the differential thermal expansion of the
vessel head relative to the vessel body. The growth of the shaft, due
to the thermal expansion between the fixed location in the gearbox
and the seal, was determined to be insignificant, and easily accom-
modated by the seal. Unfortunately, from a cost standpoint, the
configuration was very close to the design previously shown in
Figure 2. The third design configuration, shown in Figure 6, details
the arrangement ultimately selected for this application.
Figure 5. Alternative Machine/Mounting Arrangement.
This configuration, while offering significant cost savings and
ease of maintenance to the customer, also posed a number of
difficult design problems:

Perpendicular and parallel alignment of the mixer shaft through

the vessel flange

Thermal expansion of the vessel head relative to the mechanical seal

Seal change methodology

Accommodation of shaft deflection at the seal

In addition, irrespective of the design configuration chosen, the
overall vessel mixer arrangement had a number of additional
design hurdles to be overcome.

Steady bearing and tray seal alignment

Cantilevered vessel with relatively high slenderness ratio,

compounded by high overhung mixer mass, resulting in a low
natural frequency of the system (first natural lateral frequency of
the entire spring mass system)

Variable speed requirement of motor

Figure 6. Final Machine Design Arrangement.
DETERMINING SYSTEM NATURAL
(RESONANT) FREQUENCIES USING FEA
In many mixing applications the mixer mounting surface can
safely be considered infinitely stiff. Having made this assumption,
the designer can then calculate the first natural lateral frequency
(critical speed) using simple numerical methods. Safe operation for
the majority of applications is considered when the speed of the
mixer shaft is 80 percent of the calculated natural frequency. In
this instance, given the significant mass, and to fully consider the
interaction of the mixer/vessel to make sure that a natural system
frequency was not excited by forcing frequency, a more complete
analysis was necessary. Software to complete a finite element
analysis (FEA) was used to extract the systems natural frequencies
and determine the extent to which these frequencies would be
excited by the operational frequencies.
Mixer application experience has shown the operational
frequencies that have the highest energy are:

Shaft speedGearbox output shaft speed

Blade passingThe shaft speed multiplied by the number of

impeller blades

Baffle passingThe frequency associated with the impeller

blades passing the vessel chamber baffles

Motor speedOperational speed of the mixer

Figure 7 shows the FEA model developed to determine the
natural frequencies of the mixer/vessel system.
The results of the analysis are shown in Table 1. It was
immediately noted that the first natural frequency of the mixer/vessel
system (vessel rocking) was lower than the shaft speed. Under
conditions of single motor speed (i.e., no variable speed drive), the
separation between the two values would be considered sufficient to
prevent excitation. However, given the customer requirement to use
a variable speed drive to control the operation, it was important to
prevent excitation by specifying a speed block out range.
Given the mixer configuration, it was also important to review
the torsional natural frequencies. Excitation of torsional
frequencies is not always easy to detect in operation. Failure to
detect/predict these frequencies can lead to premature coupling and
gear failures. In this case the torsional frequencies were not
DESIGN CONSIDERATIONS FOR A LARGE MIXER USED IN AN AGITATED COLUMN APPLICATION 85
Figure 7. FEA Model Developed to Determine the Natural
Frequencies.
Table 1. Results of the Mode Shape/Natural Frequency Finite
Element Analysis.
coincident with any forcing frequency, and further
consideration/review was not required.
The results of the analysis allowed the detailed design of the
vessel and mixer to continue without change to the major
vessel/mixer sizes.
MECHANICAL SEAL DESIGN AND SEAL CHANGE
Basic decisions relating to the seal design had already been
made. The major challenges were associated with the alignment of
the mechanical seal and the differential expansion of the vessel
head relative to the gearbox/bearing member. Figure 8 shows the
mechanical seal design.
Figure 8. Mechanical Seal Design.
In order to overcome the thermal expansion calculated to be on
the order of 0.08 inch, it was important to allow the shaft to move
axially relative to the seal. In order to achieve this axial movement,
and to prevent the operational length of the seal from being
impacted, the seal was driven using a key, located in an extended
keyway.
A high degree of confidence existed that the mixer shaft could
be positioned concentric to the seal mounting flange. The
advantage of the partnership with the vessel manufacturer was that
we could work closely together in specifying mixer/vessel
interface details and dimensions. This relationship ensured that the
mixer mounting surface/structure could be carefully fabricated and
machined to minimize the amount of required site adjustment. The
vessel manufacturer used laser alignment techniques to ensure that
the surface of the mounting structure was parallel to the face of the
mounting flange.
The exact axial location of the seal on the final assembled shaft
could not be accurately determined, due to typical fabrication
tolerances and any shimming used during installation. It was
important to accurately position the shaft support collar (used
during the seal change) to make sure that it did not hit either the
seal or flange during operation. This position needed to be
determined by field measurement. To allow adjustment, a purpose-
fully made collar assembly was developed using a taper lock
design (Figure 9). Supporting the weight of the shaft and impeller
assemblies on a collar means that safe seal changes are achieved.
A slight shaft relief was machined in the approximate location of
the collar. In the event that the shaft slipped during a seal change,
the assembly would tighten onto the shaft, preventing the shaft
falling through the vessel. The seal change feature is shown in
Figure 10.
STEADY BEARING DESIGN
Steady bearings, as they apply to mixers, refer to bearings
located inside the vessel. Steady bearings provide added stability
for long in-tank shafts on which the impellers are mounted. They
are always of the plain bearing type, and in many respects are
similar to journal bearings. There are, in fact, several important
differences. Steady bearings are not true bearing arrangements.
The low rotational speeds associated with mixer applications mean
that a liquid film is not generated between the bushing and the
bearing sleeve.
PROCEEDINGS OF THE TWENTIETH INTERNATIONAL PUMP USERS SYMPOSIUM 2003 86
Figure 9. Shaft Support Collar.
Figure 10. Sequential Seal Change Procedure.
STEADY BEARING APPLICATION REVIEW
As previously discussed in the FEA section of this paper, mixers
in their most basic form can be analyzed as a simple spring mass
system. Traditionally, the majority of mixers are operated
subcritical. This means that, with respect to their first lateral
natural frequency or more simply stated their critical speed, mixers
are restricted to operation below a range of 0.4 to 0.80 times the
first bending critical speed. The exact ratio is largely determined by
specific application experience derived from operation in a number
of different conditions, as follows:

Operation in an open flow field of widely ranging severity,

i.e., the nonuniform/asymmetric flow, leading to uneven
impeller/shaft loading

Sparged gas, i.e., the introduction of a gas into the vessel,

usually directly beneath the impeller

Varying liquid levels, including operation with the vessel empty

Draw-offWhere the vessel is emptied with the mixer running

and the impeller operates at the surface of the liquid. (The steady
state condition, i.e., steady liquid level, is referred to as low level
coverage.)
Given the unique operating conditions, which differentiate
mixers from other fluid handling equipment, different levels of
damping are encountered during these different operating
conditions. At the draw-off condition essentially zero fluid
damping exists. This necessitates an inherently stable condition,
therefore defining the need to operate below the critical speed.
In this application dramatic shaft loading resulted from the
asymmetric flow associated with the open flow field and sparged
gas. For design purposes these loads are considered to act at the
centerline of the impeller, perpendicular to the shaft axis. Due to
the long shaft lengths, resultant bending moments in the overhung
condition are very high and would result in excessive deflections,
bent mixer shafts, and catastrophic seal deflection. In order to
increase the allowable shaft length additional shaft supports are
required in the form of one or more steady bearings. Steady
bearings allow the designer to effectively double the overhung
shaft length from a critical speed standpoint.
In summary, for this application, the use of steady bearings
provided the following benefits:

Increased critical speed of the in-tank shaft and impellers

Absorbs lateral loading

Reduced bending moments on in-tank shafting and mixer

support structure

Reduced seal shaft deflection

Reduced shaft diameter

Reduced shaft weight

Reduced cost

Reduced vibration
PRACTICAL DESIGN CONSIDERATIONS
While the application of a steady bearing provides the design
engineer with obvious advantages, its use also increases the
number of wearing parts, and therefore potentially the reliability of
the installed equipment and mean time between failure (MTBF). In
this case, given the physical size of the equipment, it was important
to maximize the time between replacement of the bushing, and
extend the life of the bushing holder. Introducing clean vessel
product as a lubricant, into the bearing in this case given the
properties of the liquid, provided an obvious opportunity for
significant positive impact.
MATERIALS OF CONSTRUCTION
The selection of the bushing material in general steady bearing
design is critical. It is necessary to select a material that can
provide corrosion resistance, toughness to handle the impact loads,
wear resistance, heat conductivity, high temperature resistance,
thermal stability, and ideally low cost. A common choice for
relatively low temperatures, up to 250F, is Teflon

. For tempera-
tures up to 700F and higher, carbon graphite is a likely choice. In
turn the sleeve must exhibit adequate strength, ideally high
hardness (i.e., Rc > 55), ability to be finished to a very smooth
surface (i.e., 8 to 12 Ra), and have good wearing characteristics for
long life. In this case, given the relatively high operating
temperature and the need for FDA approval, a material manufac-
tured from a polytetrafluoroethylene (PTFE) Kevlar

combination
was selected as the bushing material.
Figures 11 and 12 show the final arrangement for both the inter-
mediate and lower steady bearing.
MAINTENANCE CONSIDERATIONS
A point commonly overlooked has to do with the replacement of
the worn parts. Occasionally, these parts become large and heavy.
Designs that ease their replacement will improve the total cost of
ownership due to quicker turnaround times at scheduled
shutdowns. It was obviously impractical to remove the shaft and
impellers to gain access to the bushing. The bushing, bushing
holder, and sleeve, therefore, need to be axially split, with bolted
DESIGN CONSIDERATIONS FOR A LARGE MIXER USED IN AN AGITATED COLUMN APPLICATION 87
Figure 11. Steady Bearing Arrangement (1).
connections at the split lines. This allows removal of the bushing to
be relatively simple and quickly achieved. While the sleeve is
relatively easy to replace, it is still relatively heavy. With this in
mind the sleeve was designed to have essentially two running
positions. This feature allowed the sleeve to be repositioned and a
new wear surface used, essentially doubling the life of the sleeve,
providing a relatively low cost way of extending the time required
between replacements, and negating the need for the customer to
carry a very low usage spare.
PROPORTIONS
Steady bearings have a wide range of length to diameter (L/D)
ratios between 0.25 to 2.00. Most common high temperature/long
life/aggressive applications will use an L/D of 1.00 and have
achieved generally satisfactory life. The flexibility of the mixer
shafts also sets up a requirement for generous diametrical
Figure 12. Steady Bearing Arrangement (2).
clearances on the order of 0.002 to 0.007 inch per inch of shaft
diameter. This will result in a bearing that is generally considered
to be looser than what would normally be considered for a
journal bearing. In the case of mixers this is necessary due to the
large slope of the shaft through the bearing. It is possible for even
a stiff shaft to exhibit more than 0.010 inch of elastic deflection
through the length of the bushing. Without this generous
clearance, the parts are likely to bind and prematurely fail. In
addition it is very important to accurately calculate installation
startup clearances for the bushing that result in the correct
operational clearances as noted above. The differences in the
thermal expansion rates of the bushing holder (316 series
stainless steel) and bushing material needed to be carefully
considered. Poor analysis at this point can lead to the bushing
expanding onto the shaft and/or the bushing pushing off the
retainer. The overall performance and reliability of the steady
bearing are very dependent upon the concentricity of the sleeve
relative to the shaft and bushing and the calculation of the correct
operating fits.
DESIGN OF THE VESSEL TRAY SEAL
The customers need to limit the leakage between the column
stages necessitated the design of a tray seal. Given the low
allowable leakage rates required and the wide temperature
gradient, a traditional simple seal solution (lip seals, etc.) was not
appropriate. The use of additional steady bearings (or similar
arrangement) was not practical from an alignment or cost
standpoint. Therefore a simple split seal needed to be designed and
applied at the tray locations that were not used to support the
steady bearings.
Figure 13 shows the design concept used to solve the problem.
As with the steady bearing, a bushing disk manufactured from a
combination of PTFE and Kevlar is used to seal against a wear
sleeve located on the in-tank shaft. The disk, as with the other
components of the tray seal assembly, was split to allow easy
installation and replacement. The two halves are joined using
retaining plates. Rotation of the disk against the sleeve is prevented
by the use of an antirotation pin. Close operational clearances
minimize the leakage between the retaining ring and the disk. As
with the steady bearing, the wear sleeve was made sufficiently long
to allow the relocation on the shaft relative to the position of the
disk, as opposed to requiring replacement.
PROCEEDINGS OF THE TWENTIETH INTERNATIONAL PUMP USERS SYMPOSIUM 2003 88
Figure 13. Split Tray Seal Design.
CONCLUSION
The design of the mixer vessel system provided a number of
opportunities to challenge traditional thinking. Key elements to the
project being successful were associated with careful review of
each key component, and, more specifically, attention to those
features that would provide relatively easy installation,
maintenance, and high reliability.
Despite the extreme operating environment, the mixer has
operated without the need to replace any of the wear parts.
DESIGN CONSIDERATIONS FOR A LARGE MIXER USED IN AN AGITATED COLUMN APPLICATION 89
PROCEEDINGS OF THE TWENTIETH INTERNATIONAL PUMP USERS SYMPOSIUM 2003 90