How TRI got into the fluid drive businessFebruary 5, 2010
While documenting the IT system, I came across an article written in 1981. It explains how TRI improved the reliability of existing variable speed fluid drives.
Test, redesign, rebuild: Fluid-drive study concepts point way to upgrade equipment
Improvement of overall reliability demands scrutiny even of machinery that gives relatively little trouble. Learn how a program for large variable-speed boiler-feed-pump drives improved an already dependable element; even more important, how you can extend to your components the principles involved
By Dr. Melbourne F Giberson, P.E., Turbo Research Inc
The road to operating reliability in the central station is a difficult one, littered with breakdowns of equipment that went part of the way before a disappointing premature outage. The steam generator, the feed pump, the main turbine are a few of the major elements that have required searching and costly attention. Improvement in reliability is already clear for several categories of large machines, indicating that the goals are attainable.
For some of the humbler components, however, there has been scarcely a start toward constructive and cooperative work to widen the intervals between unforeseen outages. This work will have to be done if unit and station reliability and availability are to draw maximum benefit from the efforts on major elements. Perhaps new studies being planned can derive help from a series of programs covering shaft-driven fluid drives for boiler feed pumps.
These programs, for work done over a six-year period, resulted in modification recommendations to upgrade equipment that had by and large been satisfactory but was thought to be capable of some improvement. Keep; in mind that program details for other equipment will change. It is the program philosophy and the open cooperation among manufacturer, utility, and consultant that should be part of any new study of reliability enhancement.
Bettering good machinery
Shaft-driven fluid drives or hydraulic couplings with ratings from 10,000 to 30,000 hp at 3600 rpm have driven boiler feed pumps since the 1950s. Although many of the installations have been trouble free, the operating history of others has identified areas where improved mechanical reliability, machinery protection, and operating control would be valuable. The testing, redesign, and rebuilding programs, performed from 1974 to 1980, developed seven recommendations to upgrade these drives. Review briefly the machinery systems.
Two principal types of fluid drives in the ratings mentioned above are single and
dual-circuit designs, manufactured by American Standard Inc. The units may or may not have a brake on the output rotor. The schematic in Fig 1, of the single-circuit drive for a boiler-feed pump, outlines the four-bearing design, an alternative that has no pilot bearing and no brake.
The fluid drive, between prime mover and pump, connects to them through flexible couplings, which became part of the program study. The prime mover is the main turbine, of course.
The dual-circuit unit of Fig 2 has a pilot bearing, which is an intershaft bearing that supports one end of the output rotor. Note that there is now in operation a dual-circuit unit that has no pilot bearing. It is a four-bearing design instead. An optional brake is part of the Fig 2 configuration. Dual-circuit design increases the power-transmission capability of the unit
Scanning past history
The multifaceted program structure consisted of:
- Survey of operating history of the units-to identify the areas where improvement would help most.
- Analytical engineering-to predict rotor-bearing dynamics and bearing performance.
- Instrumenting and testing several operating drives and pumps.
- Redesign of mechanical components, followed by test.
- Change of machinery-protection and control systems, including sensor types and locations, alarm and trip circuitry, and operators’ controls.
- Change in maintenance and operating procedures.
- Seven modification packages developed from this effort. Review these packages, summarizing the recommended changes, in the table above.The operating-history survey covered units in several utilities, making possible a selection of areas calling for closest attention and an identification of problems for which answers were not at first available.
- In units experiencing difficulties, design areas identified as requiring most attention were:
- Need to reduce contaminants in the lobe oil, with both particulate matter and water figuring here.
- In-place balancing procedures.
- Design and installation of the pilot bearing.
- Repair of tank and foundation.
- Alignment procedures.
- Design of the flexible couplings.
The survey confirmed that the leading items on the list of fluid-drive problems were pilot bearings and brakes, in that order. It should be emphasized that the problems did not necessarily begin with those two elements, but rather that they were the elements where problems first revealed themselves most often.
When pilot bearings fail, damage to other parts can follow. For example, some pilot journals have been severely heat-checked after a pilot-bearing failure. High-amplitude vibrations, which may precede failure of a pilot bearing, can lead to catastrophic failure in other ways, too, unless operating personnel recognize the condition. Such a failure may follow demands by the utility system operator that the machinery be run regardless of high-vibration condition. High-amplitude vibrations, over long periods of time, have extensively cracked tank welds and have even injured foundations around units to the point where major reconstruction was needed.
Pilot-bearing failure can result from loss of Tube oil to the pilot bearing. The Tube oil comes to the bearing through holes in the input shaft (Fig 3). The oil must have enough pressure to overcome the head that the centrifugal reverse pumping action of the shaft produces. If, however, high-amplitude rotor vibration should damage the pilot-bearing Tube-oil seals, then the oil will tend to flow through the gap between shaft and seal, rather than into the shaft holes. With inadequate lubrication, which is the resulting condition, the pilot bearing fails. Of course, there are other causes of loss of Tube-oil supply pressure, such as loss of oil pumps.
A few seconds of operation with high-amplitude vibrations, regardless of whether they are synchronous or non-synchronous, are enough to destroy the fixed-geometry labyrinth seals of the original design.
American Standard engineered and festal a substantially improved seal (Fig 3, right) for all dual-circuit units with pilot bearing. The new Huhn seal is a floating-ring type. Its advantages are:
(1) lower seal leakage, because of close fit, so that pump pressure can be reduced considerably; (2) ability to follow vibra-tions as high as 0.010 in. in amplitude for extended time without failure.
Other sources of pilot-bearing failure include:
• Extended periods of high-amplitude vibration.
• Loosening of the bronze bearing, often because its coefficient of thermal expansion is higher than that of the steel into which it is inserted. With tempera-ture cycling, the bronze expands to its yield point, changes, dimensions, and ultimately comes loose.
• Particulate matter in the lube oil, which can score babbitt and journal. This particulate matter is one of the commonest sources of bearing failure. The original pilot bearings were especial-ly susceptible, with only a half-mil of babbitt flash coating over the bronze. Particles cannot imbed in babbitt this thin, and the babbitt wears away. The underlying bronze then runs directly on the steel pilot journal, a situation often disastrous even with lubrication.
• Water in the lube oil. With a viscos-ity about 5% that of lube oil, water is an inadequate lubricant for any of the bear-ings, including thrust bearings.
• Misalignment of the two rotors, which results in localized loading on the pilot bearing, wearing one end inordinately.
• The brake, the second major problem source, is an air/hydraulic type, with separate mechanical pin for added assurance against output-shaft rotation.
The brake had three reasons for existence: (i) to allow maintenance on out-put shaft, flexible coupling, and boiler-feed (bf) pump while the turbine shaft was rotating; (2) to keep the pump from rotating at low speed, a condition that can cause pump overheating and seizure if recirculation is deficient; and (3) to prevent reverse rotation if the discharge-line check valve fails to close.
This brake has often malfunctioned, however, resulting in wear or burnout.
The disc has even cracked, causing unbalance that made the output rotor vibrate at high amplitude. There have been many sources of brake failures:
• Inadequate air pressure because of dirt, water, or ice in air lines.
• Inadequate hydraulic-fluid supply, usually because of leaks in cylinder seals or damage to lines and fittings.
• Worn brake shoes.
• Loose or broken scoop-tube link-ages.
• Broken scoop-tube tips.
• Scoop tubes not being completely in declutch position when brakes go on.
• Improperly adjusted brakes.
• Malfunctioning controls, sometimes because of ac power interruption.
• The operating error of moving the scoop tube to “load” position while the brake is set.
Note that when the brake is applied, it is in 2n active state-not a de-energized fail-safe state. Consequently, all parts of the braking system, circuit-oil control system, and monitoring system must be in top condition and functioning properly. Failure of any one item in the brake system could cause brake failure.
The mechanical-brake pin is a’/z-in.-diameter one, through the disc. At best, it can restrain only 5% of the normal operating torque. In one case, a pin sheared when the scoop tube was exercised by station personnel who thought that the pin would hold.
At present, after application of the modifications described here, the brake, if retained, is the principal source of difficulties with the fluid drives.
Simulations aid analysis
In study of rotor-bearing dynamics, American Standard and Turbo Research separately examined four configurations: single and dual circuits, with and with-out pilot bearings. Bearing-film performance calculation for each existing bearing and several proposed ones determined film thickness, temperature distribution, pressure distribution, power loss, and dynamic characteristics.
Computer-simulation results were the engineering basis for setup of test procedures and test-result interpretation, be-side being an invaluable aid to understanding of rotor-bearing dynamics observed during operation.
Of the typical vibratory-mode shapes for the first three critical speeds (Fig 4), the one of. primary concern is the first critical, a conical or rocking mode, in which the rotor remains almost unbent. The spring action of the two bearing films, along with that of the supporting structure down to the foundation, chiefly determines the speeds at which critical speeds occur for these rotors.
In the conical vibratory mode of the input rotor, each bearing can see a high level of force and yet-because of the 180-deg phase angle between forces applied to pillow blocks and pedestals- the net forces to tank and foundation can be quite small until major damage, such as severely wiped bearings, occurs.
The input rotors for all of the size-250 and -270 single- and dual-circuit units have similar mode shapes, although the critical speeds for the modes vary, depending on weight of flexible coupling, flexibility of tank and foundation, and amount of oil in the circuit element. Some dual-circuit units have the first critical speed below operating speed.
For continuous -operation, all of these units can and should be balanced so that each journal has less than 2 mils of synchronous motion (once per revolution vibrations) relative to the bearing, as measured by Bently-style noncontacting probes. In any event, running with over 4 mils of relative unfiltered vibration at any journal should be prohibited.
High-amplitude vibration sources need a close look at this point. The input rotor is more important here, because it is heaviest, is more complex in design, holds the circuit oil, and connects to the turbine/generator, a machine whose rotor axial growth at the input flexible coupling can be inches rather than the usual thousandths of an inch.
High-amplitude rotor vibrations can be synchronous (once per revolution), from unbalance-or nonsynchronous, from fluid-dynamic destabilizing forces or bearing-film-induced oil whip. Here is a more specific rundown on sources:
• Built-up mechanical design of input rotor, which surrounds the fluid-power transmission section. With a full load of oil in the circuit, the situation at full power transmission, the shell components are highly stressed and at elevated temperature. Result is a measurable stretch and a shift relative to one another at the bolted joints. Individual components often take a permanent set.
Field balancing of the input rotor is therefore almost inevitable, no matter how precise the shop balance. Once permanent set occurs, however, vibratory conditions should stay constant. To reduce the amount of field balancing, the input and output rotors are shop-balanced dry. Then the input rotor is balanced wet, with water in the shell to simulate-the mass unbalance during field operation with oil in the shell.
• Misalignment of the fluid drive relative to the turbine/generator, across the input flexible coupling. The forces and moments produced can change bearing loading and introduce oil whip, leading to high-amplitude nonsynchronous vibration. There are many causes for the misalignment, including loose keys in the front standard or pedestal of the turbine/generator. These keys control the lateral position of the shaft extending toward the fluid drive.
A cracked fluid-drive tank, rust in the tank foot pads, cracked or swollen foundation grouting under the drives, from oil, water, or caustic soaking, and flexible couplings that are too short, have inadequate tooth design, or are worn are other causes. Misalignment at assembly and thermal growth of adjacent units not properly considered during alignment are further reasons.
• Centrifuged dirt, in either the circuit elements or the gear-type flexible couplings that connect the fluid drive to the turbine/generator, is a source, too. Centrifuged out nonuniformly in elements, dirt changes input-rotor unbalance.
In the gear coupling, a temperature change of the turbine/generator rotor changes the axial distance across the coupling to the fluid drive. Dirt between gear teeth increases the coefficient of sliding friction, so the fluid-drive rotor is pushed axially until its thrust bearing becomes heavily loaded. When load is high enough, the flexible-coupling teeth suddenly slide axially, non-uniformly, applying a heavy bending moment to the input rotor. The net effect is equivalent to an unbalance, perhaps a large one. When the coupling-gear teeth are freed or the thrust-bearing load decreases, this equivalent unbalance disappears.
∎ High-power transmission at low output-shaft speeds. When, the output shaft is in the speed range of 2200-2700 rpm, the input-shaft vibratory amplitudes are known, as one test of a single-circuit drive proved, to be susceptible to the level of transmitted power. If transmit¬ted power was high, high-amplitude vibrations at the inboard bearing of the input rotor approached full clearance of the bearing, 10-12 mils.
Simultaneously, however, readings on the tank body did not exceed 1.5 mils, and vibrations at the outboard end of the input rotor did not exceed 5 mils. This is an outstanding example of a situation where failure can be avoided if internally mounted noncontacting displacement probes alert the operator to a dangerous operating condition.
∎ Broken parts of brass, bronze, or stainless steel. Parts broken in fatigue failure have ended in the circuit-element casings, to induce high-amplitude vibrations.
Polar plots help visualize rotor motion (Figs 5,6). Fig 5 gives the synchronous response for a. rebuilt input rotor, size 270, dual circuit, on the run up from 600 to 3600 rpm, followed by rundown to 600 rpm. The plot makes clear the change in unbalance response because of stretching or shifting of rotor parts while reaching operating speed, full load, and running temperatures. The drive “doughnut,” or circuit elements of the input rotor, is involved in the stretching process. Non-contacting vibration probes on bearings gave the data.
With stretch complete, the rundown data to 600 rpm can be corrected for mechanical and electrical runout seen by vibration probes. Shifting the “600-dn” (speed decreasing) values to the center of the polar plot does this (Fig 6).
To treat the high-amplitude vibratory conditions, nonlinear bearing-film models were the basis of computation (box, next page). From these techniques came the design for tilting-pad bearings (Fig 7) to replace the existing bearings of a single-circuit fluid drive.
The purpose of these bearings was the desensitization of the input-rotor vibratory characteristics, both nonsynchronous and synchronous, to misalignment across the flexible coupling, to Tube-oil supply pressure, and to any bearing loading changes stemming from change in lifting force of the scoop tube as tube position changes.
An actual test
An example of how study and principle go to work in practice is a program for identifying problem sources in a case of nonsychronous vibration of the input rotor in a four-bearing, single-circuit unit.. At the beginning, the inboard journal of the rotor was vibrating 10 mils relative to the pedestal. The tank, however, was vibrating only 1.5 mils.
A first step, installation of tilting-pad bearings on the rotor, helped control vibration amplitudes, but did not eliminate the source of nonsynchronous vibration. The testing program initiated to identify problem sources included an in¬place performance test of the b-f pump. For this, instrumentation was added to the spool-piece of the output flexible coupling, transforming it into a torsiometer for measurement of power to the pump. Test results indicated that the pump was operating as designed and specified. More data revealed that:
• Conventional sources of nonsynchronous vibration were not the cause
0 A no-load mechanical test run had been successful.
• The nonsynchronous vibrations had occurred at a frequency between input¬shaft and output-shaft rotating frequen¬cies (Fig 8).
• The output shaft had only minor vibrations.
• Vibrations worsened at a given output-shaft speed when transmitted torque increased.
• The problem occurred only in the 2200-2700 rpm output-shaft speed range, which includes the speed at which the circuit-oil discharge temperature reached a maximum.
All these clues focused attention on the fluid-power transmission mechanism and suggested that the problem resulted from nonuniform flow-plus corresponding unbalanced inertia effects of the oil retained in the circuit elements of the input shell of the input rotor.
Inspection of the circuit-element internal surfaces along which the oil flowed revealed significant waviness, caused by localized polishing that had removed damage from a previous failure. Re-contouring of the wavy surfaces was a major contribution in reducing the nonsynch¬ronous excitation of the input rotor.
Next, by bringing the output shaft through the 2200-2700 rpm speed range with lower power-which for the particular application meant lower b-f pump flow and higher discharge pressure-the nonsynchronous high-amplitude vibrations were avoided. Once output-shaft speed was above 2700 rpm, the power level of the application did not appear to influence significantly the nonsynchronous vibration levels. It is quite natural to see a beat effect (Fig 9).
In general, it is difficult to detect the onset of high-amplitude vibrations unless rotor-vibration monitoring protection is adequate. Up-to-date non-contacting vibration-monitoring instrumentation, with properly located probes, will detect the condition however.
Redesign of several components of the fluid drives was helped by analysis when modifications affected bearing perform¬ance and rotor dynamics.
Relatively simple changes to output rotor and output-bearing pedestal allowed conversion of the single-circuit units to four-bearing design. Consequently, no attempt was made to retain the pilot-bearing configuration for these.
For the dual-circuit units, on the other hand, a major redesign of the output pedestal was necessary to accommodate a second bearing for the output rotor. As an interim step, redesign of the pilot bearing itself brought in a steel back and a thicker babbitt lining.
Component testing on several designs of pilot bearing indicated optimal bearing-surface design and the best way to bond babbitt to steel (Figs 10, 11). The best test results came when there were no oil grooves on the journal surface.
The service record for the steel-backed pilot bearings has been excellent. In inspection after more than two years of operation that had many start-stop cycles of the turbine, the original manufacturing tool marks were still visible over most of the babbitt surface. There was no measurable change in bore size.
Nevertheless, any unit with a pilot bearing is still subject to loss or short¬term interruption of the high-pressure pilot-bearing tube-oil supply. The loss or interruption can result from loss of pumps, from air in the filters or coolers during transfer, from high Tube-oil temperature if the coolers become fouled, or even from turnoff of cooling water.
An effective design change is separation of lube-oil/circuit-oil reservoir and supply system for drive and coupling from the corresponding system for the turbine/generator and b-f pump. This change reduces the amount of particulate matter, sludge, and water in the fluid drive and coupling. Separation from the b-f pump is especially important where water or steam can enter the bearing-oil cavities of the pump.
Setting alignment specifications and allowable tolerances, and thinking in terms of alignment retention and ease of making alignment moves, are other important considerations for continued successful operation of fluid drives and b-f pumps.
For the alignment specifications, it is necessary to understand the alignment changes that the rotors-not the surrounding housings-experience going from cold to hot condition. Tolerances on the specifications depend on the capabilities of the flexible couplings to accommodate misalignment without transmitting significant shear forces or bending moments between adjacent rotors.
Where misalignment repeatedly occurs, the source must be ferreted out and correction made. Perhaps redesign to improve the flexible coupling’s ability to take misalignment will be needed. Longer coupling spool-pieces and improved tooth shapes are choices here.
Improved bracketry can make align-ment moves easier. The alignment of fluid drive and b-f pump needs checking during each annual outage.
Heat dissipation may be a significant operating loss in the fluid drive when the b-f pump shaft speed is below 3200 rpm at maximum power on the turbine/generator. The loss can be substantially reduced by reducing pump capacity. Trimming the impeller diameters, removing a stage, or redesigning the impellers can do this. The result will be an increase in pump speed, allowing the
fluid drive to operate at a better efficiency. A pump speed range of 3420-3510 rpm is a good objective, with the higher end of the range preferable.
Protection and control
Original protection and control system for the fluid drive had a few trips; an operator’s manual trip, a high circuit-oil discharge temperature trip, and a low bearing-oil pressure trip. Sometimes trips were found for open generator breakers, or for low b-f pump pressure or brake status.
The most important added protection is for high-amplitude vibrations of rotors relative to bearings. Other recommended protection involves the brakes, bearing tube-oil supply temperature and pres¬sure, circuit-oil divert valve position, and feedwater system status. Move the sen¬sors for oil to better locations if possible. This means at the unit itself, rather than in the pump discharge/cooler/filter headers in the basement.
Improved displays for the alarm and trip lights will help, along with improved operator control switch and operational status indicators.
A manual for operation and maintenance procedures is needed for each system, consisting of upgraded fluid drive, b-f pump, and feedwater-flow monitoring, in each specific application. The procedures help in diagnosing and correcting problems, and also are a training document for operators.
In addition to the cooperation of American-Standard Inc.’s Industrial Products Div, two utilities were active in support. At Consolidated Edison Co of NY, W B Warner and G Kraus gave valuable assistance, and Virginia Electric & Power Co also aided the effort.
As a result of the work, forced outages should be greatly reduced when stations apply the recommendations and follow reasonable operating and maintenance procedures.
Even More Info…
Penetrate nonlinear-bearing secrets with computers
Nonlinear techniques are important for film bearings and seals. Proprietary programs can predict unbalance response, resonant whirl, and oil whip for multi-bearing machines. The programs model a wide range of film-bearing geometries, such as circular bore with or without axial grooves, and elliptical, multilobe, and tilting pad bearings. Turbulence and variable film viscosity are included. If viscosity is variable, programs calculate a temperature profile through the film, and an extremely valuable attribute of these techniques -include the fully nonlinear and cross-coupled stiffness and damping characteristics of the film.
One example is a program for synchronous lateral-vibratory rotor response for critical speeds. Another, for rotor stability, addresses nonsynchronous time-transient lateral-vibratory response. Journal bearings, fixed-bore and tilting-pad, benefit from variable-viscosity and turbulence inclusion in a third program
Consider eliminating dovetails if babbitt is heavily loaded;
Sometimes a babbitt bearing in which dovetail grooves mechanically lock the babbitt in place experiences cracking, even when chemical: bonding, is good. Circumferential cracking appear¬ing over the sharp edge of the dovetail anchor of the steel bearing shell, starts at the stress-concentration point of the steel/babbitt interface and finally comes to the inner surface. Uniformly thick babbitt, with a good chemical bond, has. proved superior
Published in Power Magazine, June 1981
Edited by William O’Keefe