Skip to content

A review of past projects at TRI

TRI solved some very perplexing problems over the years. As is often the case, the first step is to identify accurately what the problem is, and then to design options to solve the problem. The following case studies are among many that have been solved by TRI. Should you have a similar problem, or a new problem, that is related to high amplitude vibrations – synchronous (“1-X”) or sub-synchronous, bearing damage, and/or fluid drive and flexible coupling issues, please contact TRI. TRI is interested in addressing such problems for our clients.

Often a client realizes he has a maintenance problem, and has been living with it for years believing that “that is just the way it is”. In most cases, TRI can identify the problem, and work with the client to bring TRI’s engineering and manufacturing skills to bear on the issue, and resolve it in a timely and cost-effective manner. No machine is too large or too small for us.

Unevenly Heated Rotor

The “shorted turns” of a generator field caused uneven heating of the rotor, which caused the rotor to bow and have high unbalance. The bow in the rotor caused very high amplitude synchronous (at running speed) rotor vibrations. These high unbalance forces effectively unloaded the bearings. Because the bearings were cylindrical bore, when unloaded, they would develop sub-synchronous (near first critical speed near 800 rpm) vibrations. The vibrations at mid-span of this 65 ton rotor were calculated to be on the order of 0.140 inches peak to peak.

TRI performed a computer simulation of the entire rotor-bearing system using TRI’s proprietary time-transient non-linear model and duplicated the vibration pattern. TRI then changed the bearing design for both bearings to elliptical with some non-standard features, and the sub-synchronous vibrations were suppressed and the synchronous vibrations were limited.

Tilting pad bearings were also simulated and these had even better performance characteristics. Because the elliptical bearings appeared to be sufficient and could be made more quickly, they were chosen.

The results were quite satisfactory.

Gear Box Bearing Upgrades

Gear Box Bearing

Westinghouse Gas Turbine-Gearbox-Generators use bearings in the gearbox that are easily damaged if the unit is operated at low MW Load. The journals of a gear shaft contact the bearings in a manner to limit the oil flow into the bearings, causing the bearings to overheat. 

TRI used the TRI Proprietary Bearing Computer simulation program which incorporates completely non-linear time transient solution techniques to model the bearings. The existing bearings were modeled and the problems with them were identified. New TRI bearings were designed and the orientations of the split lines of the bearings were optimized. Temperature sensing instrumentation was installed to be able to measure the bearing metal temperatures in the load zones. Then the bearings were manufactured and installed with the expected satisfactory performance, including acceptable temperatures.

Turbine-Condensate Correlations

TRI has been asked by a number of parties about the correlation between changes in LP Turbine Rotor Vibrations and changes in condenser back-pressure and between changes in LP turbine rotor vibrations and changes in condensate temperature, as well as what to do to reduce the sensitivity of the LP Turbine rotor vibrations to these changes in back-pressure and temperature.

The purpose of this case study is to address both the correlations and to provide recommendations to reduce the rotor vibration sensitivity.

Yes, there can be a correlation in each case. The way it works is as follows:

There can be, and usually is, a correlation between LP Turbine back-pressure and elevation of the cone support structures for the bearings in the LP turbine hoods. Also, there is usually a correlation between the LP Turbine condensate temperature and the elevation of the cone support structures for the bearings in the LP Turbine hoods. 

Dropping the elevation of one bearing shifts the loading on several bearings in both directions along the shaft line. In particular, the two adjacent bearings will pick up loading, but others are affected, carrying slightly less or more load. The load shift for the various pairs of adjacent bearings of a unit, for example, Bearings (4,5), (6,7), and, if they exist, Bearings (8,9), may be on the order of 1000 to 4000 lbs per mil (0.001 inches) of elevation change between adjacent bearings.

A change of loading on a bearing changes the film stiffness of that bearing: Increasing the loading increases the stiffness, and vice versa. Changing the stiffness of the bearing film of a turbine shaft can change the effective critical speed, often not much, but in the extreme, the critical speed can actually pass through the operating speed. Furthermore, almost all rotors in service are unbalanced to some degree. If a critical speed is near the operating speed and that critical speed changes, the vibratory response (amplitude and angle) at the operating speed due to the existing unbalance usually changes.  

In order to understand the actual degree of change of 1X, or synchronous, vibration in a given situation, it is helpful to plot the vibratory response on a polar plot, including the slow roll data point and the roll up data for each vibration probe of the machine. The most reliable data for understanding what is going on at each bearing in a GE machine is the “left side” data, even though the “right side” is usually the side with the larger amplitude. For a counter-clockwise rotating machine (GE), the left side is what is called the “hard side” because the thinnest film location is approximately directly in line with this probe on the lower right of the journal, and 1 mil of motion in the direction parallel to the left side probe typically represents greater vibratory force on the bearing than 1 mil of motion measured on the right side probe. The right side is the “soft side” because the journal can slide along the thin film zone, which is why the vibrations on this side are usually larger, but less important than the vibrations measured on the left (hard) side.

Typically, if the overall vibration is substantially more than the 1X vibration, then the remainder is generally sub-synchronous. If there is subsynchronous vibration, then this is important to note. A small amount of sub-synchronous is usually not a problem, but if it appears that the amplitude of the sub-synchronous component could jump to large amplitude motion, the situation should be evaluated for further action. If the unit has experienced large amplitude sub-synchronous vibratory motion, then it should be evaluated for remedial action, which TRI can offer in various forms.

There are typically two ways to support condensers. For reference, older condensers were fixed to, that is, hung from, the turbine deck and they had spring supports underneath the condenser. In this case, no flexible seal was necessary at the LP Turbine on the turbine deck. In the older units of this design, it was absolutely critical to have the condenser filled with the proper amount of water when cold alignments of the turbine-generator were made. Too much water in the condenser, and the turbine deck would be pulled down too far, and the alignment would not be right. Not enough water, and the turbine deck would not be pulled down enough, and the alignment would not be right.

The newer design, which began in the 1950s, is to have the condensers supported rigidly underneath from the basement, and to have a flexible sealing connection (dog bone) around the top of the condenser to the LP turbine at the turbine deck. The larger units, which are generally the subject of this Case Study, are made with the flexible sealing connection to the LP Turbine hood. When vacuum is drawn in the condenser, the entire LP Turbine hood is pulled down, as if a weight equivalent to (14.7 psia – actual condenser pressure in psia) times the projected area of the condenser connection to the LP Turbine (sq inches) is loaded evenly on top of the LP Turbine hood. The turbine hood and surrounding foundation drop accordingly. For an LP Turbine, the reinforced concrete (or fabricated steel) support structure must carry the LP Turbine hood, the LP Turbine rotor, and the entire vacuum loading. In addition, the ends of the turbine hood are pushed inward, and this helps to rotate the cone structures that support the LP turbine rotors. The internal support structure (struts, gussets, and the like) within each hood affects how the bearing elevation and bearing rotation are changed as a result of the forces due to the changes in vacuum loading.  

The vertical component of the change in vacuum loading in the present case is calculated as follows: When the backpressure in the condenser is on the order of 1.4 in Hg, which corresponds to 0.7 psia, the total vacuum loading on the foundation is (14.7-0.7 = 14 psi) acting over the entire area of the LP Turbine shell. If each LP Turbine shell is 20 ft x 20 ft, this is 400 sq ft or 57,600 sq inches, the total force downwards is 806,000 lbf. This force will drop the elevation of the entire LP Turbine structure. If the overall stiffness of the LP hood support structure is approximately 100 million lbf/inch, then the 14 psi pressure will cause the turbine deck to drop by 0.008 inches (8 mils).

Dropping the backpressure from 1.4 inches Hg (0.7 psia) to 0.8 inches Hg (0.4 psia) increases the downward pressure on the Hood by 0.3 psi, which corresponds to another 17,280 lbf downward loading. This 0.3 psi change of loading would drop the turbine deck by an additional 0.17 mils.

Each of the bearings of an LP Turbine is in a structure called a cone. It is, in effect, a tunnel that extends out into the LP Hood. These cone/tunnel structures are designed to move the bearings toward the center of the LP Turbine rotors, reducing the bearing spans of the LP Turbine rotors. When vacuum is drawn, the hood changes shape and each cone/tunnel structure rotates. 

A change of temperature can cause a significant change in the shape of a large steel structure, such as an LP hood. For instance, a 10 degree F change in a steel bar that is 16 inches long will change the length by 0.001 inches. If a steel structure that is 48 inches in length is subjected to a temperature change of 30 degrees (105 – 75) deg F, it will change length by 0.009 inches. 

Experience over many years indicates that it is very difficult to know how the bearings and the shaft line are going to respond to a change in vacuum (back-pressure) and/or to a change in temperature of the LP hood before a machine and its mounting structure are built. Similar units react in different ways. How the LP hood is constructed including what gussets exist, how the hood is supported by the foundation, and how the foundation is built (concrete or steel) affect how the bearing elevations will change with a change in vacuum and/or condensate temperature. 

Usually the foundation is a reinforced concrete structure down to bedrock or to substantial pilings, but some foundations are steel structures. The long vertical piers under the bearings are cooled and change length when the ambient temperature changes, in an amount as described above. Some columns change length more or less than others because some are more exposed to radiant heat from steam pipes than are others, and some are more exposed to changes in ambient temperature than others. Some units are so sensitive to ambient air temperature that opening a big door in cold weather will cause the vibrations to change almost immediately, and in some extreme cases, the sound of the machine vibrations will change, and will do so in a repeatable manner.

The alignment of the LP turbine rotors relative to each other and to the adjacent turbine or generator rotors is often intentionally adjusted from time to time by turbine engineers, so that at normal operating temperatures, the bearings are fairly evenly loaded as represented by the bearing metal temperatures. It is imperative to retain plant records of what alignment conditions provide the best operating conditions, and to use the latest records of what works best for each unit in the process of aligning the rotors for that unit during each subsequent outage.

In 1973, I took a considerable amount of test data on a 500 MW tandem compound GE LST-G which showed the following trends. When cold and when no vacuum was drawn, the LP Turbine Bearing 5 was substantially unloaded. As the temperature in the condenser increased, the LP hood expanded, raising Bearing 5. When the unit was at normal operating temperatures, the bearing metal temperatures evened out showing that the loading on these two bearings evened out.

On the other hand, other GE units of a slightly larger size have Bearing 4 unloaded and Bearing 5 loaded when cold. Yet, at operating conditions, the bearing metal temperatures are approximately the same.

In any case, it is typical that the largest mismatch between couplings in cold conditions along a GE tandem-compound train occurs between the last IP Bearing and the first LP Bearing. In other words, the first LP Bearing sees more variation in loading than any other LP turbine Bearing. The last Turbine bearing, adjacent to the generator, also sees variation in loading, but usually not so much variation as the first LP Bearing experiences. 

It is difficult to try to modify the hood structures in such a manner as to reduce the sensitivity of bearing loading to changes of vacuum condition or to condensate temperature. If these changes become concern items on a continuing basis, then the least costly options are (a) to change the first LP Bearing from an elliptical bearing to a tilting pad bearing, such as a TRI Align-A-Pad ® Bearing, and/or (b) to use a variable speed pump to control the circulating water flow rate to maintain uniform condensate temperature and backpressure.

Summary of Recommendations:

  1. Because most of the GE LP turbine bearings in the original configurations are marginally stable, that is, they are not far from experiencing sub-synchronous vibrations such as “oil-whip”, it is very helpful to have the LP Bearings mounted as rigidly as possible in the cone structures, or standards, as the case may be. 
  2. For those LP bearings that are adjacent to the IP turbine or to the generator and experience substantial elevation changes, and have demonstrated any tendency toward sub-synchronous vibrations, consideration of changing to a tilting pad bearing such as the TRI Align-A-Pad ® Bearing is appropriate.
  3. The standard GE elliptical bearings can be modified using a TRI proprietary design method in a way to improve the stability of the bearings while keeping the original Babbitted bearing length. In some cases, converting to shortened elliptical bearings is advantageous. In other cases, the bearings should not be shortened because the Babbitt becomes overloaded, and in this case, other design modifications can be made by TRI proprietary methods to reduce the amplitudes of synchronous rotor vibrations as well as to improve stability (resistance to sub-synchronous vibrations).
  4.  It is very important that the turbine engineers maintain accurate records of what alignment works for each machine. Trying to follow the OEM recommendations in the original mechanical design data (instruction) book is not necessarily advisable because this data may have been written before the unit was actually built and does not take into consideration how the unit is being operated, peculiarities of the foundation design, whether it is in indoor or outdoor unit, the range of ambient temperature conditions to which the unit is exposed, or what has been learned from this unit about what makes it work best. Bearing metal temperatures, vibratory characteristics, orbit shapes, and wear patterns should be used to adjust the alignment from time to time to optimize the performance of the machine. In many cases, bearing designs have been changed to permit a wider tolerance on the alignment data that is used. This is a clear benefit of using the TRI Align-A-Pad ® tilting pad bearings in places that are susceptible to a range of alignment conditions during operation, such as result from high temperature conditions of the standard when at high load and cooler temperature conditions when at low load.
  5. Where over-cooling of the condensate has occurred with the normal operating conditions for the circulating water pumps, it is appropriate to consider variable speed motors. Under certain circumstances, two-speed motors can be used. In extreme circumstance, turning on or off the circ water pumps may be appropriate to control the cooling of the condensers. Exercising any one of these options may help to optimize the cycle efficiency under various plant conditions and/or ambient conditions.

Because most Large Steam Turbine-Generators were built to operate at or near full load, and because capital costs were (and are) always a critical factor in new equipment, most LST-G units were made with a certain number of constant speed circulating water pumps, and these were intended to be on all of the time, and not cycle on and off. Nevertheless, some units have been modified to have two speed condenser cooling water pumps. For some units, variable speed circulating water pumps are being considered. Incidentally, in some cases, variable speed pumps are being considered to minimize water removed from a river primarily for external reasons, but would provide the possibility of cycle efficiency optimization.

There is a compromise in establishing the preferred condenser back-pressure: In almost all cases, the colder the cooling water, the lower the back-pressure, and the more MW load that is generated, with all other conditions being the same. However, the colder the condensate, the more heat that is required to heat the condensate to make the steam for the turbine. The objective should be to cool the condensate no more than is necessary to optimize the efficiency of the cycle. For those units that are operating at maximum firing rate, over-cooling the condensate will actually reduce the MW generated.

An issue to be considered in selecting an option for varying the circulating water flow is this: The motors for circulation water pumps are huge, and these motors and the associated switchgear typically were not designed to be started often. Stopping one of these motors and restarting it in a cyclic manner is not wise, even if it might optimize the cycle efficiency. 

It is not often that an FD Fan, ID Fan, BFP, or other large auxiliary is forced out of service. Consequently, when the MW load is dropped for a period of several days to repair one of these items, it might be worthwhile to take a circulation water pump off, so long as the condenser function remains reasonably uniform for all of the condenser and LP Turbine sections.

Repair of severely damaged journal bearings

Problem: A Journal and mating Babbitted Bearing are both found to be significantly damaged. The situation addressed here is focused on a journal that is grooved, scored, or severely pitted. Causes may include a large solid particle in the lube oil, loss of lube oil, or extensive pitting from electrolysis. In severe cases, the journal may be damaged to a radial depth of 1/8 inch or more. 

Solution:  TRI offers a “turn-key solution” wherein the plant places one contract with TRI to provide the services to remachine the journal and to refurbish the bearing to suit the new journal size.

TRI is available to (a) provide overall project supervision, (b) provide site technical direction, (c) hire a company that specializes in machining and finishing journals or seal surfaces in place, and (d) TRI shops will refurbish the bearing, or manufacture a new bearing with undersize bore. TRI has been able to have all of the required resources moving and sub-contracts placed within 30 minutes of the phone call from the customer.

Benefits: 

  1.  Coordination of the repairs to both components – Journals and Bearings – is maximized, helping to assure that efforts are directed to a common goal – a fully functional journal-bearing. 
  2. TRI Engineering specializes in designing Babbitted Bearings (Circular bore, elliptical bore, and tilting-pad designs) for turbine-generators, rather than simply reverse engineering the bearing. Even on an emergency basis, TRI can ensure that whatever the application, the bearing design supplied will be suitable for the application. In certain situations, upgraded bearing bore designs can be machined into the surface of the Babbitt bore without changing the return time.
  3. Emergency repairs often involve adding a Babbitt layer to the existing Babbitt, and machining a new bore. However, bearings with Babbitt layers that are over 1/8 inch thick should be replaced with new steel backed bearings with a maximum of 1/8 inch Babbitt. TRI technical department creates CAD drawings that can be used to manufacture new bearings with the desired Babbitt thickness that can be installed at a later date. 
  4. When high vibration sub-synchronous vibrations (oil whip, oil whirl, steam whirl) occur and damage machinery, TRI Engineering can design bearing bores that suppress such dangerous vibrations, often during emergency repairs.

Babbitt-copper intermetallic compound layer

A simple literature review, or search on the Internet for “intermetallic compounds” would lead one to believe that this represents the state of the art in material science.  While the study of many new compounds is on the leading edge of technology, the intermetallic nature of copper and tin in an alloy form has been studied for more than 5000 years.  Without the aid of high tech tools, the Chinese developed the beta bronze alloy form of tin and copper some 1400 years ago.  This was the first metal that could be intentionally heat treated to provide a wide range of mechanical properties.  In more recent research, much attention has been paid to the formation of Cu6Sn5 and Cu3Sn intermetallic compound layers, and their effect on solder joints in electronic assembly.  Unfortunately, little, or no attention has been paid to the identical reaction that occurs when bonding a tin based Babbitt to a copper alloy backing material typical of many fluid film bearings used in industry today.

My first direct exposure to the resultant phenomenon of the formation of these compounds came about 10 years ago.  During the dis-assembly of a high speed gas compressor, the thrust pads were removed from the unit for inspection.  In this particular bearing, the pads were designed with ASTM-B23 Grade 2 babbitt bonded to a copper alloy containing approximately 2% chrome for increased mechanical strength.  In this application, the high sliding velocity present in the oil lubricated thrust bearing would have yielded unacceptably high bearing temperatures if conventional steel backing material had been used.  The copper alloy backing material was used due to its high thermal conductivity to provide improved bearing performance.  In this instance, following successful dimensional checks, and ultrasonic inspection of the babbitt bond, the pads were returned to the compressor deck to be re-installed in the machine.  During the installation process, one of the pads was inadvertently dropped from a height of about three inches on to a steel workbench.  As a result of this minor impact, the babbitt completely separated from the copper alloy backing material.  This was indeed somewhat disturbing that the babbitt could fall off of an otherwise acceptable part that was ready for installation in a very expensive machine that operates in excess of 10,000 RPM.

Intermetallic compounds differ from simple metal alloys in a few basic, but important ways.  A metal alloy consists of a base material to which certain percentages of other elements have been added.  Most alloys are simply a disordered solid solution of individual components.  The atoms in a conventional alloy link together with relatively weak metallic bonds.  In an intermetallic compound, there is a discrete, narrow range of chemical composition.  The atoms bond to one another with a combination of ionic and covalent bonds.  As a result, the individual atoms begin to take up preferred positions within the crystal lattice of the intermetallic compound.  This helps to explain the general tendencies of intermetallic compounds: higher strength, higher melting points, and poor ductility.

As stated earlier, there are two discreet intermetallic compounds formed as by-products of the combination of copper, tin, heat, stress, vibration, and time.  Cu6Sn5 forms the instant molten tin comes in contact with the copper material.  This compound grows rapidly while the tin is liquid, and slows considerably once the metals solidify.  The presence of Cu6Sn5 is essential for good adhesion of the two materials.  Cu3Sn does not form while the tin is molten.  It forms between the copper and the Cu6Sn5 layer through solid state diffusion once the metals have hardened.  Both of the intermetallic compound layers will grow when exposed to heat, stress, and vibration.  The mechanical problems associated with this phenomenon are multi-fold.  The intermetallic compounds are extremely brittle, and have different coefficients of thermal expansion then those of either the tin based babbitt, or the copper backing material.  This alone makes the bond susceptible to cracking from mechanical or thermal stress.  In addition, the depletion of the tin from the layer of babbitt adjacent to the intermetallic compound leaves behind an alloy of questionable mechanical properties.

In a testing program, it was established that this problem could be reproduced, discretely and repeatedly simply by heat cycling the subject parts.  The babbitt bond on perfectly good parts could be rendered dangerously brittle, simply by cycling the parts between 250 and 80 degrees F for a couple of weeks, for a total of less than 300 cycles.  The heat cycled parts all had the following properties, identical to those found in the field:

1. Without additional stress, the babbitt bond remained intact
2. The babbitt can be separated from the backing by inserting a sharp object at the bond     line with minimal force
3. As the babbitt is separated from the backing, an audible “crackling” can be heard
4. Un-magnified inspection of the copper surface following removal of the babbitt shows a matte grey color
5. 100X magnification of the copper surface shows a combination of areas where the babbitt has been removed clean to the copper, as well as areas of remaining babbitt.  The areas of remaining babbitt do not appear as a crystalline fracture surface, but more as smooth shiny, almost polished areas.

Now that the problem was identified, and shown to be repeatable, the question remained as to the appropriate solution.  A similar problem had been solved in the early days of satellite construction.  Apparently, solder joints would fail after some of the early satellites would have made numerous orbits around the earth.  The formation of these intermetallic compounds was halted by applying a thin barrier plating to the copper prior to soldering.  Testing was then undertaken, and electroless nickel was found to provide an excellent barrier between the tin and the copper, as well as providing a surface that would readily accept the bonding of the babbitt.  The application of a nickel barrier layer to all copper and copper bearing alloys prior to babbitting, has prevented this phenomenon from recurrence.  This step should be taken with all new design, as well as any re-babbitting operations with copper alloy backed fluid film bearings.

TRI Transmission and Bearing Corp.
Fred Wiesinger
July 2000

Vibration Patterns Expected for Variable Speed Fluid Drives

High Powered Variable Speed Fluid Drives are used to provide power to Boiler Feed Pumps and Fans. Each of these Fluid Drives has an input shaft assembly that rotates at the speed of the input power source, which may be an electric motor or a steam turbine-generator that operates at constant speed. The output shaft assembly rotates at a variable speed, typically controllable from 20% to 97.5% of the input shaft speed.

The typical design of a variable speed fluid drive uses an input rotating assembly that is coupled to and rotates with the speed of the driver (motor or turbine-generator). The output rotating assembly is connected to and rotates with the driven load, a pump, fan, or compressor.

fluid drive schematic

The input rotating assembly consists of an input coupling, a shaft supported by the two journal bearings and a thrust bearing, and a fluid drive element that consists of an impeller and a casing that surrounds the impeller and a runner, all rotating together.

The output rotating element consists of a runner, an output shaft supported by two journal bearings and a thrust bearing, and an output coupling hub.

The impeller and runner are structures with vanes and pockets between the vanes shaped much like a half of a grapefruit with the linings that separate the fruit intact and the fruit removed.

The impeller rotates at the input shaft speed and the runner drives the output shaft, or pump (or fan) speed. The two rotating elements are connected hydraulically, but not mechanically. The speed of the output rotating assembly is controlled by hydraulic forces. A scoop tube is used to remove fluid from rotating assembly and to control the level of fluid in the element, thereby controlling the hydraulic forces between the impeller and runner. A scoop tube actuator, is driven by a signal from the plant control system, such as the central DCS (Distributed Control System), for the unit.

The casing that surrounds the impeller and runner is attached to the input rotating assembly so it rotates with the input shaft speed. The casing contains a fluid, usually an oil, being thrown outward by centrifugal force so that it is in the shape of a torus adjacent to the inside surface of the spinning casing. The thicker the torus of liquid that exists, the greater the coupling between the impeller and the runner. Oil, called “circuit oil”, is pumped into the spinning element, or “circuit” and is removed by a scoop tube at the rate that it is pumped in, removing heat from the process. The inside diameter of the torus is adjustable by the radial movement of the scoop tube as it removes circuit oil from the torus, thereby controlling the coupling between impeller and runner.

A fluid drive is a constant torque machine, which means that the torque transmitted by the output shaft to the load (pump or fan) is equal to the torque absorbed from the driver (motor or turbine) by the input shaft. Because there is always a difference in speed between the input shaft and output shaft, the power into the fluid drive is equal to the power out plus the power lost in the fluid drive process within the element. Power-in equals input speed times torque, Power-out equals output speed times torque, and power lost equals torque times the slip speed (input speed – output speed). It is the heat generated by the lost power that requires the circuit oil to be exchanged continuously.

The circuit oil flow pattern within the element consisting of the impeller pockets and runner pockets and the axial gap between them is extremely turbulent. For reference, for a fluid drive impeller of 25 to 27 inches in diameter, the impeller and runner pockets are about 6 inches deep axially and the gap is on the order of 1/4 inch. The pockets are larger or smaller in proportion to the size of the fluid drive element.

The turbulence of the circuit oil is affected by the speed difference, the torque transmitted, and the amount of circuit oil in the rotating element. The speed at which the greatest turbulence occurs is when the output shaft is at approximately 2/3 of the input shaft speed. This is in the range of 2400 rpm for a fluid drive input speed of 3600 rpm. The turbulence at a given speed increases with increased torque at that speed. Generally, when the circuit oil is the most turbulent, the fluid drive is generating the highest amount of heat and vibration in the element.

Four performance charts are presented that are based on the head-capacity plot for a main boiler feed pump for a 375 MW Turbine-Generator. The curves on the charts are made of constant speed lines (generally curving down to the right). The fluid drive power loss chart also has curves of constant efficiency lines (generally curving upward). The grids are the same on all of the charts. Data is presented at the intersection points of the grids.

For instance, following the constant speed line for 2400 rpm from left to right, it can be seen in these charts that power transmission increases, the torque increases, the losses increase, and the circuit oil flow rate required to maintain a constant temperature rise increases. All of this is associated with increased turbulence of the circuit oil when traversing the constant speed line going from low pump flow to higher pump flow.


A Size 270 Single Circuit Fluid Drive has a 27.0 inch diameter circuit, and it holds approximately 15 gallons, or 105 lbs) of oil when the output shaft speed is in the 2400 – 3000 rpm range. The radial acceleration that the oil experiences can be measured in the number of “g”s by dividing the acceleration by the gravitational acceleration constant (386 in/sec^2). At 3600 rpm, and at a diameter of 27 inches, the “g field” is approximately 4600.

To explain the features of the vibratory forces acting on the impeller and casing as well as on the runner, consider the following:

  • The impeller is not perfectly uniform in shape from pocket to pocket and
  • neither is the runner;
  • The impeller and runner have some vibration and therefore, for each one,
  • the center of rotation is not the geometric center of the respective part so
  • that the radial distance between their centers varies continually between 0
  • and 0.005 inches, or more;
  • The circuit oil exiting the impeller and entering the runner near the OD
  • has a frequency of the impeller (input);
  • The circuit oil exiting the runner and entering the impeller near the ID
  • has the frequency of the runner (output); and
  • The circuit oil exiting the gap between the impeller and runner and
  • passing between the input casing on the outside and the back of the runner
  • on the inside has a mixed flow influenced by both the runner and the
  • impeller positions and hence their speeds.

Under these circumstances, it is very likely that the oil flows can be very turbulent and non-uniform circumferentially. With over 100 lbs, or over 1600 ounces, of circuit oil, for every 1 ounce of oil that is out of balance, a radial force of approximately 300 lbs is exerted radially on the inside of the input casing. If there are 5 ounces of oil that are unbalanced, then the radial unbalance force is on the order of 1500 lbs.

Given that the entire input rotating element weighs only 1500 lbs, and it is the element carrying the circuit oil, the input rotating element can be expected to respond to these forces by vibrating at the frequencies of excitation: Input shaft frequency, output shaft frequency, and a mixed frequency that is usually slightly above the output shaft frequency.

The runner has a vaned structure on the inside and has a smooth surface on the outside. There is a gap between the outside runner surface and the inside of the input casing that contains circuit oil. Both the vaned structure on the inside of the runner and the smooth surface on the outside of the runner are exposed to circuit oil, and the pressures in the oil on both sides are proportional to each other. Consequently, the runner does not experience the radial force levels that the impeller and input casing experience due to unbalanced oil caused by turbulence.

However, for both the impeller and casing assembly as well as for the runner, the following applies: The greater the turbulence in the circuit oil, the greater the unbalance of the oil, the greater the radial forces on the impeller and casing as well as for the runner, and the higher the vibration of the rotating assemblies, particularly the input assembly. It also follows, that the greater the amplitudes of vibration of both the input and the output shafts, the greater the distances between their respective centers, the greater will be the non-uniform flows of circuit oil from each one to the other and the greater will be the unbalanced amount of circuit oil.

The radial forces acting on the impeller and casing and on the runner by the circuit oil are transferred to the respective input and output shaft assemblies. Each rotating assembly is supported by two journal bearings and a thrust bearing. The journal bearings resist the radial forces applied to the respective rotating assembly.

The two journal bearings, known as the input-outboard and the input-inboard bearing, support the input shaft. Two journal bearings, known as the output-inboard and the output-outboard support the output shaft. The input-inboard journal bearing is the bearing that primarily supports the impeller and input casing as well as the spinning circuit oil, and therefore, it is the one that is subjected to the greatest force levels due to the dynamic forces of mechanical unbalance and unbalance of the turbulent circuit oil as described above, as well as the static weight of the inboard end of the input rotating assembly.

In the Size 270 Single Circuit Fluid Drive, the centerlines of the journal bearings are approximately 25 inches apart. The impeller is overhung from the input-inboard bearing by about 12 inches, and the runner is overhung from the output-inboard bearing by approximately 18 inches.

The original pressure dam journal bearings were relatively small bearings, being 7 inches in diameter and about 2.75 inches long. These bearings were generally adequate to resist the mechanical unbalance forces of a well balanced fluid drive and they could resist a minimum level of radial forces due to the unbalanced circuit oil flow when the boiler feed pump was operated in the 2300 to 2700 rpm speed range, such as would occur during start-up – if the feed pump flow was kept at the very minimum permitted flow, or if the start-up pump were used to bring the boiler to full pressure and then the transfer to the main pump was made.

However, most operators would not operate the unit this way. The typical start-up is to get on the main pump as soon as is practical, and while limiting the pressure to the silica limit, say 1200 psi at first, the flow would be increased to the maximum possible by operating the throttle valves to the Valves Wide Open condition. There are two benefits: The greater the flow, the faster the boiler water will clear and the faster the boiler pressure could be raised, and the second is that the power generated would be maximized during the water clean-up process.

Problems arise for the fluid drive when it is operated this way, even for a few hours during a start-up, because the output shaft speed is gradually raised and it traverses the 2300-2700 rpm speed range causing (a) high heat generation with the fluid drive getting quite hot, and (b) there is a considerable amount of turbulent circuit oil flow with consequent high rotating radial forces in the element that are applied to the bearings, particularly, the input-inboard journal bearing. The original small pressure dam bearings could not withstand the forces, and would be damaged causing increased bearing clearances. In turn, this would permit larger distances between the impeller and runner centers which lead to bigger unbalanced circuit oil flows which led to higher forces that the bearings had to resist. Even when the fluid drives were built with great precision, the original bearings, particularly the input-inboard bearing, would fail repeatedly.

The root cause of the continued bearing failures was discovered via a considerable amount of testing and redesign work. Once it was realized that the radial forces resulted from unbalanced circuit oil flow in the element, and that they were torque dependent and an inherent part of the fluid drive process, then the solution became apparent: Build the rotor and journal bearing system so that it could easily resist these unbalance forces without degrading.

In 1984, using its high tech bearing design capabilities, TRI designed and manufactured a heavy duty input shaft and tilting pad journal bearing system that was extremely robust. The upgraded tilting pad journal bearings are the same as those that support an IP turbine rotor weighing some 40,000 lbs.

From the first start-up, this upgraded TRI Heavy Duty journal bearing system proved to be the solution to this vibration problem. Subsequently, the journal bearings for the output shaft were upgraded to tilting pad bearings. Typically, these fluid drives are run for approximately 10 years between planned disassembly for inspection, cleaning, and reconditioning.

The various frequencies that were observed in the vibration traces from the proximity probes that were installed in the original fluid drives, described above, can still be seen in the vibration traces of the upgraded fluid drives. The only difference is that the upgraded journal bearings can resist the forces on a continuous operational basis.

Fluid Drive for ID Fans

fluid drive for ID fans

A 400 MW lignite fired power plant has two 6000 hp, constant speed ID Fans that were controlled by dampers on the inlet sides of the fans. Extensive testing, including gas flows and pressures, as well as electrical testing using power factor meters, provided a good “baseline” of the electrical power consumed for the fan operation at different flows and boiler conditions. Using the “test block” head – capacity plots and established fan laws, TRI made calculations as to the electrical power that would be consumed at the various load points of the test data. These calculations indicated that considerable power would be saved across the entire operating range of the fans, particularly at high loads and at low loads. Another part of the justification was that the heat exchanger for cooling the circuit oil in the fluid drives was to be located inside the Forced Draft Fan Room, thereby recovering all of the heat generated by the fluid drives in the slip process. 

With this background, a project was undertaken wherein TRI designed and built two fluid drives, one for each fan, along with a single oil conditioning system for both fans. 

This is a back to back fan arrangement, and the reinforced concrete foundations for the motors were extended to connect to each other, making the space to move the motors back so that the fluid drives could be installed between the motors and the fans. These fluid drives were made with impellers and runners with 70 inch working diameters. TRI is not aware of any fluid drive impeller/ runner size on the North American Continent that is as large or larger than these are. Due to the short space permitted, the fluid drives were built with rolling element bearings and with compact flexible couplings.

A unique characteristic of these fluid drives is that the flow of circuit oil into the element of each fluid drive is controlled in order to maintain a constant Circuit Oil Discharge (COD) Temperature. This maintains the entire fluid drive at a constant temperature of approximately 190° F and minimizes the circuit oil flow, especially at the upper end of the speed range and at the lower end of the speed range.

An unanticipated benefit was found to be this: This plant is located in a colder portion of the United States, and the ID fans are located outside. When they have been sitting for a period of time, they develop a bow and the rotors are cold. In the original configuration, when they were started, they went to full speed and had very high vibrations until the rotors warmed up and the bow relaxed. With the fluid drives, the motors and input portion of the fluid drives start and are up to speed within about 2 seconds, and then the output section of the fluid drive and the fan gradually roll up in speed. The fans can be held at low speed, so they can warm up permitting the bow to relax. Then, the fans can be brought up to speed in the operating speed range and put into service. 

Before the fluid drives were installed, the fan rotors would be brought up to max speed, and if the rotor were bowed, the vibrations would be very intense. They could be felt in offices some 200 feet away. Now, with the fluid drives, the fan rotors remain balanced as they go up in speed, and the vibrations are low, even during the start-up mode. This helps the balancing process greatly, because the bow is removed at low speeds and any vibration at high speed is due to either actual mass unbalance of the fan or deposits on the fan wheels which need to be removed.

Fluid Drives for Large Compressors and Pumps

TRI designs and manufacturers variable speed fluid drives for large pumps and compressors. TRI has provided retrofit packages to existing applications. Many of these units have been made to the available space envelope. In other words, they are “custom made” for the specific application.

Consider an existing application that consists of a constant speed motor driving a constant speed compressor. In such a case, it is typical to move the motor back away from the load (pump or compressor), and install a fluid drive with new flexible couplings. In some applications, the same flex couplings can be used with new coupling halves of the same design and size that are mounted onto the fluid drive input and output shafts. Then, the existing flex couplings do not have to be removed from the existing equipment.

In most cases, there is no need for the manufacturer of the pump or compressor or the driving motor to be involved. TRI has considerable expertise in the required engineering skills such as rotor-dynamics and stress analysis. If the compressor or pump is to have a new application (process fluid or pressure) then perhaps the manufacturer might be involved. We can provide drawings for the foundation modifications and your civil engineers can create the necessary foundation drawings. TRI can supply the soleplates, or we can provide soleplate specifications and you can manufacture soleplates.

TRI has many variable fluid drives in service for boiler feed pumps from 3,000 to 28,000 hp at 3600 rpm input in power plants, as well as fans at 600 rpm. 

TRI manufactures all of our products in our U.S. facility. In almost all applications, we machine the impellers and runners from forged steel. The consequence is that our equipment has a very high reliability record. Most of our fluid drives have operational records exceeding ten years between inspections.

TRI can supply complete oil systems, or we can provide the specifications for the actual hardware of the oil systems.

There is considerable energy savings that will result from application of fluid drives with large pumps and compressors. In addition, the starting operation for all components, motors, compressors, pumps, and any gears, is much easier, so life expectancy of the components is greatly increased. Removal of any discharge control valves permits improved ease of operation, and the valves that are removed no longer need maintenance. The time period that it takes to pay back the capital cost of the installation is usually measured in months. The reduced maintenance cost and increased production time (reduced downtime) should also be factored into the “benefit to cost” ratio.

New Emergency Lube Oil Pumps

This existing Lube Oil System was provided by a major US supplier and was less than ten years old. It had a number of weak points, the most serious of which was that the DC powered Emergency Lube Oil Pump almost always failed to start pumping oil when it was turned on.

In this case, the focus was on the DC powered emergency lube oil pump that was a vertical centrifugal pump mounted on top of the oil reservoir with the pump suction bell in the oil. The suction bell filled with vapor from bubbles in the oil that gathered and coalesced, displacing the oil. When the centrifugal pump started, it could not pump oil until the vapor had been removed, which took many seconds. In that time period, a serious “Loss of Lube Oil” Event had occurred, with considerable damage to the bearings of the turbine-generator, requiring full disassembly, rotor machining for seals and journals, and repaired bearings.

This sequence of events was not fully understood until it occurred multiple times. A contributing factor was that the entire oil skid was built on top of a small reservoir and it was installed so that it was almost impossible for an operator to get to any of the major components. Maintenance on critical components had to wait for a major outage.

TRI designed and supplied a Complete Replacement Lube Oil System. This system was built like many other TRI oil systems, with duplex AC powered pumps, duplex filters and duplex coolers that were located separately so that most maintenance could be performed without a major outage. Almost always TRI designs use DC powered pumps that are mounted horizontally, are positive displacement, and are fully flooded, with considerable attention paid to isolation valves, pressure regulator valves, and pressure relief valves. Where possible, a separate battery is used for the DC Motor of the emergency lube oil pump. TRI uses a combination of hard-wired and DCS or PLC programmable controls.

While this case involved a 30 MW Turbine-Generator, the same design features apply to any size machine, whether new or in-service.