Centrifugal Compressors

Chevron Corporation 200-1 November 2001

200 Centrifugal Compressors

AbstractThis section discusses engineering principles, types of machines and configura-tions, and performance characteristics. It contains sufficient information, when used in conjunction with Company specifications, to understand how to specify and apply centrifugal compressors including auxiliaries and support systems.

The discussion is primarily aimed at heavy-duty multistage units, but the informa-tion can be applied to smaller and less severe-duty compressors as well.

Contents Page

210 Engineering Principles 200-3

211 Gas Flow Path

212 Conversion of Velocity Energy to Pressure

213 Thermodynamic Relationships

214 Performance Related to Component Geometry

215 Compressor Types

220 Performance Characteristics 200-14

221 General

222 Impeller Performance Curves

223 Use of Fan Laws

224 Surge

225 Stonewall

230 Selection Criteria 200-26

231 Application Range

232 Horsepower and Efficiency Estimates

233 Head/Stage

234 Stages/Casing

235 Discharge Temperature

236 Selection Review

200 Centrifugal Compressors Compressor Manual

November 2001 200-2 Chevron Corporation

240 Machine Components and Configurations 200-31

241 Machine Components

242 Dry Gas Seals

243 Configurations

250 Application and System Considerations 200-72

251 Effect of System Changes on Performance

252 Stable Operating Speed Ranges

253 Power Margins

254 Series Operation

255 Weather Protection

256 Process Piping Arrangements

257 Lube- And Seal-Oil Systems

260 Instrumentation and Control 200-79

261 Typical Instrumentation

262 Compressor Control

263 Control System Selection

264 Surge Control

265 Machinery Monitoring

270 Rerates and Retrofits 200-84

271 Capacity

272 Pressure

273 Power

274 Speed

280 Foundations 200-87

281 Foundation Mounting

282 Design Basis for Rotating Compressors

290 Materials 200-90

291 Sulfide Stress Cracking

292 Stress Corrosion Cracking

293 Hydrogen Embrittlement

294 Low Temperature

295 Impellers

296 Non-Metallic Seals

297 Coatings

Compressor Manual 200 Centrifugal Compressors


210 Engineering PrinciplesThis section covers the fundamentals of centrifugal compressors, describing the gas flow path, conversion of velocity to pressure, thermodynamic relationships, and the effect of component geometry on compressor performance.

These fundamentals provide a foundation for troubleshooting performance prob-lems, making rerating or initial selection estimates, evaluating vendor proposals, engineering compressor applications, and assisting with overall process design.

211 Gas Flow PathA discussion of the flow path through the centrifugal compressor will provide a better understanding of the compression process.

There is often confusion about the term “stage” when applied to centrifugal compressors. The process designer thinks of a stage as a compression step made up of an uncooled section, usually consisting of several impeller/diffuser units. The mechanical engineer or machine designer defines a stage as one impeller/diffuser set, and a section as a single compressor casing containing several stages. In this section of the manual:

• Stage is defined as one impeller/diffuser set

• Process stage is defined as an uncooled section (or casing) containing several impellers/diffusers

Based on this, a centrifugal compressor is made up of one or more stages; each stage consisting of a rotating component or impeller, and the stationary components which guide the flow into and out-of the impeller. Figure 200-1 shows the flow path through a section of a typical multistage unit.

212 Conversion of Velocity Energy to PressurePressure is increased by transferring energy to the gas, accelerating it through the impeller. Note that all work on the gas is done by the impeller; the stationary components only convert the energy added by the impeller. Part of this energy is converted to pressure in the impeller and the remainder is converted to pressure as it decelerates in the diffuser. A typical pressure-velocity profile across a stage is shown in Figure 200-2.

Since the kinetic energy is a function of the square of the velocity, the head (not pressure) produced is proportional to the square of the impeller tip speed:

(Eq. 200-1)

H KU2

g-------=



where:H = head,

U = impeller tip speed in ft/sec

K = a constant

g = 32.174 (ft-lb: mass) / (lb: force) (sec2)

Note “Head” is a term often used for the work input to a compression process. The units of head are foot-pounds (force) divided by pounds (mass). In general practice, “head” is usually taken as “feet.”

Manufacturers generally define performance of individual impellers in terms of:

• Head coefficient µ - a function of actual work input and stage efficiency

• Flow coefficient φ - a non-dimensional function of volume flow and rota-tional speed

Figure 200-3 represents a typical individual impeller curve. The head coefficient typically varies from about 0.4 to 0.6. The surge line in the figure is discussed in Section224. Using the head coefficient, the head can now be shown as:

(Eq. 200-2)

Fig. 200-1 Compressor Section (Courtesy of the Elliot Company)

ft.-lb.f

lb.m---------------

HµU2

g----------=



Fig. 200-2 Pressure and Velocity Profile

Fig. 200-3 Performance of a Centrifugal Compressor



213 Thermodynamic RelationshipsReferring to the thermodynamic discussion in Section100, the geometric and thermodynamic head relationships may now be equated.

(Eq. 200-3)

where:

As mentioned in Section100, the polytropic process is typically used for centri-fugal compressors (rather than the adiabatic process).

Using the relationship for k, n, and ηp, polytropic efficiency is:

(Eq. 200-4)

214 Performance Related to Component GeometryEffects resulting from the geometric shape of the principle components of the compressor are shown in Figure200-4. Variables such as the impeller configuration and blade angle, inlet guide vane angle, diffuser size and shape, etc., can be adjusted by the machine designer for optimum performance under a specified set of oper-ating conditions. Figure200-5 shows impeller vector diagrams for various blade angles.

Impellers with backward leaning blades, are more commonly used for most centrif-ugal compressors because of their increased stable operating range ( Figure200-6). Forward and radial blades are seldom used in petrochemical applications.

Machine output is always affected by combined losses, such as:

• Mechanical loss• Aerodynamic loss• Friction and shock loss

HpolyµU 2

g---------- ZavgRT1

rn 1–

n------------

1–

n 1–n

--------------------------------------= =

ZavgZ1 Z2+

2-------------------=

average compressibility=

ηp

k 1–k

------------

n 1–n

------------------------=



Mechanical losses , such as those from a journal or thrust bearing, affect the power input required, but do not influence the head-capacity curve. Aerodynamic losses that do influence the shape of the curve consist mainly of wall friction, fluid shear, seal losses, recirculation in flow passages, and shock losses. Shock losses are the result of expansion, contraction, and change of direction associated with flow sepa-ration, eddies, and turbulence. Friction and shock losses are the predominant sources of the total aerodynamic losses.

Figure 200-7 illustrates the affect of these combined losses in reducing the theoret-ical head.

Friction losses can be reduced by improving surface finishes. Shock losses may sometimes be mitigated by further streamlining of flow passages. These techniques will improve efficiency and tend to reduce the surge point, but they are costly, and there is a point of diminishing returns. The Company specification does not allow the manufacturer's quoted performance to include efficiency improvements due to impeller polishing.

Fig. 200-4 Impeller Inlet and Outlet Flow Vector Triangles (From Compressors: Selection & Sizing, by Royce Brown 1986 by Gulf Publishing Company, Houston, TX. Used with permission. All rights reserved.)



Fig. 200-6 Effect of Blade Angle on Stability



215 Compressor TypesThere are two types of compressors, defined by either an axial or radial casing construction. Figure200-8 illustrates this construction, referred in the API 617 Standards as:

• axial, or horizontally split• radial, or vertically split

API 617 (Centrifugal Compressors) requires the use of the vertically-split casings when the partial pressure of hydrogen exceeds 200 psi.

Other factors which influence the horizontal/vertical split decision include the abso-lute operating pressure of the service and ease of maintenance for a particular plant layout.

The top half of the horizontally-split casing (Figure200-9) is removed to access the internals. The stationary diaphragms are installed individually in the top and bottom half of the casing. Main process connections may be located either in the top or bottom half.

Fig. 200-7 Typical Compressor Head



Fig. 200-8 Joint Construction (Courtesy of the Howell Training Group)

Fig. 200-9 Horizontally-split Casing (Courtesy of the Howell Training Group)



The horizontally-split down-connected casing has the advantage of allowing removal of the top half for access to the rotor without requiring removal of major process piping.

Vertically-split or barrel compressors have a complete cylindrical outer casing. The stationary diaphragms are assembled around the rotor to make up an inner casing, and installed inside the outer casing as a unit, contained by heads or end closures at each end. Some later designs hold the heads in place by use of shear rings (Figure 200-10).

On the vertically-split casing, maintenance of the rotor and other internal parts (other than bearings and shaft-end seals) involves removal of at least one head, withdrawal of the inner casing from the outer pressure containing casing, and then dismantling of the inner casing to expose the rotor (Figure200-11). The inner casing and rotor can be removed from either the up- or down-connected vertically-split outer casing without disturbing process piping.

Both the horizontally and vertically-split casing designs allow removal of bearings and shaft-end seals for maintenance without disassembly of major casing components.

Figure 200-12 gives a comparison of pressure vs. capacity for multistage horizon-tally- and vertically-split casing construction. The size/rating comparisons are general. Specific pressure/capacity ranges and casing configurations vary between manufacturers.

Overhung-Impeller TypesSingle-stage, overhung-impeller (impeller located outboard of the radial bearings, opposite the driver end) designs are available in pressure ratings to approximately 2000 psi and capacities to 50,000 cfm.

Fig. 200-10 Shear Ring Head Retainer (Courtesy of Dresser-Rand)



Another type of centrifugal compressor is the integrally-geared configuration. This is an overhung-impeller type built around a gear box, with the impellers attached to gear pinion shafts and the impeller housings mounted on the gear box. Possible configurations include two, three, four, and even five stage designs with capacities to 30,000 cfm and pressures to 250 psig. These have typically been used as pack-aged-air or nitrogen compressors. The overall arrangement of this type varies signif-icantly between manufacturers.

Fig. 200-11 Vertically-split Casing (Courtesy of the Howell Training Group)

Fig. 200-12 Pressure/Capacity Chart (Courtesy of Dresser-Rand)



Major features of the integrally geared design include:

• Open impellers—maximum head developed• volute diffusers for optimum efficiency• different pinion speeds to optimize impeller efficiency

220 Performance Characteristics

221 GeneralFigure 200-13 presents a centrifugal compressor performance map, using API 617 nomenclature. The family of curves depicts the performance at various speeds where N represents RPM, and:

• Vertical axis—Head: polytropic head, pressure ratio, discharge pressure, or differential pressure; and

• Horizontal axis—Inlet Capacity: called “Q” or “Q1” shown as actual inlet volume per unit of time ACFM or ICFM where “A” is actual, or “I” is inlet.

Note that inlet flow volume, or capacity, is based on a gas with a particular molec-ular weight, specific heat ratio, and compressibility factor at suction pressure and temperature.

The curve on the left represents the surge limit . Operation to the left of this line is unstable and usually harmful to the machine .

A capacity limit or overload curve is shown on the other side of the map. The area to the right of this line is commonly known as “stonewall” or “choke”. Operation in this area is, in most instances, harmless mechanically, but the head-producing capability of the machine falls off rapidly, and performance is unpredictable.

Surge and stonewall should not be confused. Although machine performance is seri-ously impaired in either case, they are entirely different phenomena. These are covered in more detail later in this section.

Terms frequently used to define performance are “stability range” and “percent stability”. Referring again to Figure 200-13, the rated stability range is taken asQD - QS where QD is the rated point and QS is the surge point along the 100% speed line. The percent stability expressed as a percentage is:

(Eq. 200-5)

222 Impeller Performance CurvesFor convenience, manufacturers usually base the performance of individual impel-lers on an air test. Figure200-14 represents a typical curve which characterizes a

% stabilityQD QS–

QD--------------------- 100×=



certain impeller design. The vertical axis is usually called the head coefficient µ; and the horizontal axis is called the flow coefficient, φ . (See Section212 for defini-tions of µ and φ). In this way, impeller performance data are concisely cataloged and stored for use by designers. When a compressor is originally sized, the designer translates the wheel curve data into ACFM, discharge pressure, and RPM in wheel-by-wheel calculations to select a set of wheels that satisfy the purchaser's requirements.

Fig. 200-13 Typical Centifugal Compressor Performance Map (Courtesy of the American Petroleum Institute)



Theoretically, an impeller should produce the same head, or feet of the fluid, regard-less of the gas weight. However, in practice, a wheel will produce somewhat more head (than theoretical) with heavy gases, and less with lighter gases. Gas compress-ibility, specific heat ratio, aerodynamic losses, and several other factors are respon-sible for this deviation. Manufacturers should apply proprietary correction factors when the effect is significant. This effect contributes to variance from the well-known fan laws or affinity laws. (See the next sub-section.)

Notice in Figure200-14 that the heavier gas causes surge at a higher Q/N, that is, it reduces stability. The opposite is true of a lighter gas. Similar non-conformance can sometimes be observed when the wheel is run at tip speeds considerably higher or lower than an average design speed. The higher tip speed would surge at higher Q/N, and the lower tip speed would surge at a lower Q/N.

Figure 200-15 illustrates the effects of using movable inlet guide vanes. Notice that as the head or discharge pressure is reduced, the surge volume (defined by the dashed line) is also reduced. The effect is similar to that of speed reduction on a variable speed machine. Inlet throttling, although less efficient, will produce similar curves.

Centrifugal compressors recognize actual inlet cubic feet per minute (ACFM at inlet conditions, or ICFM). Performance curves are most commonly plotted using ACFM. This means that a curve is drawn for a specific set of suction conditions, and any change in these conditions will affect the validity of the curve.

Performance curves often plot discharge pressure on the vertical axis, and flow (ACFM) on the horizontal axis. To estimate performance for varying suction pres-sures, the curve should be converted to pressure ratio on the vertical axis. This can

Fig. 200-14 Individual Impeller Performance Curve



be done by dividing the discharge pressures on the vertical axis by the suction pres-sure on which the original curve was based. The effect of a small variation in suction temperature can be estimated by using a ratio of absolute temperatures with the original temperature in the denominator. This ratio is used to correct the inlet capacity on the X-axis by multiplying inlet capacities by the temperature ratio.

For a rough estimate for molecular weight changes of less than 10%, the pressure ratio on the curve can simply be multiplied by the ratio of the new molecular weight over the original. Unless there are gross changes in the gas composition causing large changes in specific heat ratio, this estimating method will only have an error of 1–2% for pressure ratios between 1.5 and 3. For more accurate estimates, a curve with polytropic head on the vertical axis must be obtained.

Remember that any change that increases the density of the gas at the inlet will increase the discharge pressure and the horsepower. Also, the unit will tend to surge at a slightly higher inlet volume.

223 Use of Fan LawsFan laws can be used in many cases to estimate performance for small changes in speed and flow, but care and judgment must be used. Using these laws is risky, and should be done cautiously.

The fan laws state that inlet volume is proportional to speed, and that head is proportional to the speed squared. These laws are based on the assumption that the fluid is non-compressible. Fan laws may be inaccurate when testing the perfor-mance level of multistage compressors at off-design speeds. ( Figure200-16 illus-trates this error.) Similar errors could be incurred in estimating surge volumes using the fan laws..

Fig. 200-15 Constant Speed Machine with Variable Inlet Guide Vanes


Compressor M

anual

Novem

ber 2001200-18

Chevron Corporation

Fig. 200-16 Error in Fan Laws – Multistage Compressor



To illustrate, assume a 10% mass flow reduction to the first stage. If all other inlet conditions remain the same, volume flow will also be reduced by 10%. Since mass flow was reduced by 10%, the second stage will also see a 10% flow reduction. (Figure 200-13 shows that flow reduction results in an increased discharge pressure from the first stage.) Since volume is inversely proportional to pressure, the volume to the second stage will be reduced further in proportion to the increased discharge pressure from the first stage. The second stage will have a similar effect on the third stage and so on. Deviation from the ideal gas laws will increase significantly as the number of compressor stages increases.

224 SurgeSurge is a situation that can destroy a compressor. It is a critical factor in design of the compressor and its control system. It is also a critical operating limit.

Surge is a condition of unstable flow within the compressor, resulting in flow reversal and pressure fluctuations in the system. This occurs when the head (pres-sure) developed by the compressor is less than that required to overcome down-stream system pressure. At surge, continuous “forward” flow is interrupted.

While surge is caused by aerodynamic instability in the compressor, interaction with the system sometimes produces violent swings in flow, accompanied by pressure fluctuations and relatively rapid temperature increase at the compressor inlet. Surge affects the overall system and is not confined to only the compressor. Therefore, an understanding of both the external causes and the machine design is necessary to apply an adequate anti-surge system.

The compressor surge region was previously identified in Figure 200-13. In Figure 200-17 lines depicting three typical system operating curves have been added. The shapes of these curves are governed by the system friction, and pressure control in the particular system external to the compressor

A compressor will operate at the intersection of its curve and the system curve. To change the point at which the compressor operates:

1. Change the speed or variable geometry of the compressor, thus relocating the compressor curve; or

2. Change the system curve by repositioning a control valve or otherwise altering the external system curve.

Typical Surge CycleA typical surge cycle is represented by the circuit between points B, C, D, and back to B (Figure200-17). If events take place which alter the system curve to establish operation at point B, the pressure in the system will equal the output pressure of the compressor. Any transient can then cause reverse flow if the compressor discharge pressure falls below the downstream system pressure.

For reverse flow to occur, compressor throughput must be reduced to zero at point C which corresponds to a pressure called the “shut-off head”. When the system



pressure has decreased to the compressor's shut-off head at C, the machine will re-establish forward flow since the flow requirement of the compressor is satisfied by the backflow gas (compressor capability now greater than system requirements).

Now that the compressor has sufficient gas to compress, operation will immediately shift to the right in approximately a horizontal path to point D. With the compressor now delivering flow in the forward direction, pressure will build in the system, and operation will follow the characteristic speed curve back to points B and C. The cycle will rapidly repeat itself unless the cause of the surge is corrected, or other favorable action taken, such as increasing the speed.

Several internal factors combine to develop the surge condition. From the surge description, you can see that the domed shape of the head-capacity characteristic curve is fundamentally responsible for the location of the surge point at a given speed. On the right side of the performance map (Figure200-17) the slope of the curve is negative. As inlet flow is reduced, the slope becomes less negative until it reaches zero at the surge point. As flow is reduced further to the left of the surge point, the slope becomes increasingly positive.

Section210, “Engineering Principles” covers internal factors and their effect on location of the surge region.

Fig. 200-17 Typical Centrifugal Compressor Performance Map Showing Surge Cycle



Frequency of SurgeFrequency of the surge cycle varies inversely with the volume of the system. For example, if the piping contains a check valve located near the compressor discharge nozzle, the frequency will be correspondingly much higher than that of the system without a check valve. The frequency can be as low as a few cycles per minute up to 15 or more cycles per second. Generally, the higher the frequency, the lower the intensity. The intensity or violence of surge tends to increase with increased gas density which is directly related to higher molecular weights and pressures, and lower temperatures. Higher differential pressure generally increases the intensity.

Design Factors Affecting SurgeA greater number of impellers in a given casing will tend to reduce the stable range. Similarly, so does the number of sections of compression, or the number of casings in series.

The large majority of centrifugals use vaneless diffusers, which are simple flow channels with parallel walls, without elements inside to guide the flow. The trajec-tory of a particle through a vaneless diffuser is a spiral of about one-half the circum-ferential distance around the diffuser (Figure200-18). If this distance becomes longer for any reason, the flow is exposed to more wall friction which dissipates the kinetic energy. As flow is reduced, the angle is reduced which extends the length of the trajectory through the diffuser (Figure200-19). When the flow path is too long, insufficient pressure rise (head) is developed and surge occurs.

Fig. 200-18 Design Condition Velocity Triangles (Reproduced with permission of the Turbomachinery Laboratory. From Proceedings of the Twelfth Turbomachinery Symposium, Texas A&M University, College Station, TX, 1983)



Occasionally, vaned diffusers are used to force the flow to take a shorter, more effi-cient path. Figure200-20 shows the flow pattern in a vaned diffuser. The vaned diffuser can increase the aerodynamic efficiency of a stage by approximately 3%, but this efficiency gain results in a narrower operating span on the head-capacity curve with respect to both surge and stonewall. The figure also shows how the path of a particle of gas is affected by off-design flows. At flows higher than design, impingement occurs on the trailing side of the diffuser vane creating shock losses which tend to bring on stonewall. Conversely, flow less than design encourages surge, due to the shock losses from impingement on the leading edge of the vane.

Despite adverse effects on surge, the vaned diffuser should be applied where effi-ciency is of utmost importance, particularly with small high-speed wheels.

Stationary guide vanes may be used to direct the flow to the eye of the impeller. Depending upon the head requirements of an individual stage, these vanes may direct the flow in the same direction as the rotation or tip speed of the wheel, an action known as pre-rotation or pre-swirl . The opposite action is known as counter-rotation or counter swirl. Guide vanes set at zero degrees of swirl are called radial guide vanes.

Fig. 200-19 Flow Trajectory in a Vaneless Diffuser (Reproduced with permission of the Turbomachinery Laboratory. From Proceedings of the Twelfth Turbomachinery Symposium, Texas A&M University, College Station, TX, 1983)



The effect guide vanes have on a compressor's curve is illustrated in Figure 200-21. Note that pre-rotation reduces the head or unloads the impeller. Pre-rotation tends to reduce the surge flow. Counter-rotation increases the head and tends to increase the surge flow.

Fig. 200-20 Vaned Diffuser

Fig. 200-21 Effect of Guide Vane Setting (Stationary or Variable)



Movable inlet guide vanes are occasionally employed on single-stage machines, or on the first stage of multistage compressors driven by electric motors at constant speed. The guide vane angle can be manually or automatically adjusted while the unit is on stream to accommodate operating requirements. Because of the complexity of the adjusting mechanism, the variable feature can only be applied to the first wheel in almost all designs.

External Causes and Effects of SurgeBriefly, some of the usual causes of surge (other than from machine design) are:

1. Restricted suction or discharge such as a plugged strainer.

2. Process changes in pressures or gas composition.

3. Mis-positioned rotor or internal plugging of flow passages.

4. Inadvertent speed change such as from a governor failure.

The effects of surge can range from a simple lack of performance to serious damage to the machine and/or the system. Internal damage to labyrinths, diaphragms, thrust bearing and the rotor can be experienced. Surge often excites lateral shaft vibration. It can also produce torsional damages to such items as couplings and gears. Exter-nally, devastating piping vibration can occur causing structural damage, mis-align-ment, and failure of fittings and instruments.

Surge can often be recognized by check valve hammering, piping vibration, noise, wriggling of pressure gages or ammeter on the driver. Mild cases of surge are some-times difficult to discern.

225 StonewallAnother major factor affecting the theoretical head-capacity curve is choke or stonewall . The terms surge and stonewall are sometimes incorrectly used inter-changeably, probably due to the fact that serious performance deterioration is observed in either case.

A compressor stage is considered to be in stonewall, in theory, when the Mach Number equals one. At this point the impeller passage is choked and no more flow can be passed. Industry practice normally limits the inlet Mach Number to less than 0.90 for any specified operating point.

We are concerned with two important items in defining stonewall: the inlet-gas velocity incidence angle, and the inlet-gas Mach Number.

The vector diagram (Figure200-22) shows an inlet-gas velocity vector which lines up well with the impeller blade at design flow.

The ratio of the inlet gas velocity (relative to the impeller blade) to the speed of sound at inlet is referred to as the relative inlet Mach Number .



(Eq. 200-6)

where:

As flow continues to increase, the incidence angle of the relative gas velocity, with respect to the impeller blade, becomes negative as shown in Figure200-23. The negative incidence angle results in an effective reduction of the flow area and impingement of the gas on the trailing edge of the blade, contributing to flow sepa-ration and the onset of choke.

It is important to note the choke effect is much greater for high molecular weight gas, especially at low temperatures and lower k values. For this reason, maximum allowable compressor speed may be limited on high molecular weight applications, with a corresponding reduction in head per stage.

Fig. 200-22 Inlet Gas Velocity Vector – Design Flow (Courtesy of the Elliott Company)

Fig. 200-23 Inlet Gas Velocity Vector – Negative Incidence Angle (Onset of Choke) (Courtesy of the Elliot Company)

Mach No.V rela1

----------=

a 1 g k ZRT1=

speed of sound at inlet=



230 Selection CriteriaThis section concentrates on equipment selection. (Forms are also available in the Appendix to assist in the estimating process.)

231 Application RangeRefer to Figure200-12 for a chart of capacity vs. pressure for horizontally- and vertically-split centrifugal compressors.

Normally, manufacturers do not design a compressor to match an application, they fit the application to one of a series of existing compressor casings or frame sizes. Therefore, check the manufacturer's bulletins for data required to make selection estimates. Figure200-24 provides data for a series of compressor casings based on a comparison of data from the industry.

In addition, the minimum discharge CFM (DCFM) should be considered. Current impeller designs limit impeller inlet CFM to approximately 300-500 ICFM. Thus, process conditions resulting in a discharge volume of less than approximately 250 DCFM may be unacceptable.

232 Horsepower and Efficiency EstimatesOne of the major benefits in doing your own estimates, rather than turning every-thing over to a manufacturer, is that you develop a better understanding of the appli-cation. You are then in a better position to discuss it with the manufacturers, evaluate alternate selections, and even catch errors in manufacturer's estimates.

Figure 200-25 is a plot of polytropic efficiency vs. inlet volume flow. This chart may be used for estimating polytropic efficiencies.

As discussed in Section100, manufacturers use a computer to calculate compressor performance on a stage-by-stage basis. Performance is based on each preceding

Fig. 200-24 Preliminary Selection Values for Multistage Centrifugal Compressors



stage, new impeller inlet conditions, including compressibility (Z) and k values to determine the individual performance for each successive stage.

If specific stage data is unavailable, overall calculations using average compress-ibility and a k value based on the average flange-to-flange temperature, will provide reasonably accurate results. (Refer to Section100 for compressibility equations.)

Estimate overall efficiency from Figure200-25, using average CFM from:

(Eq. 200-7)

where discharge ACFM is determined using Equation200-14 and an efficiency of 75%.

Determine n-1/n from:

(Eq. 200-8)

Fig. 200-25 Polytropic Efficiency vs. Inlet Volume Flow (Courtesy of Dresser-Rand)

cfmavgInlet ACFM Disch. ACFM+

2---------------------------------------------------------------------=

n 1–n

------------ k 1–kηp------------=



Recalculate head, discharge temperature, and gas horsepower (GHP) from:

(Eq. 200-9)

where:Hp = Polytropic Head in feet

(Eq. 200-10)

(Eq. 200-11)

where:w = weight flow in lbs./min.

Estimate brake horsepower using:

BHP = GHP + bearing loss + oil seal loss

where bearing loss is determined from Figure200-26, and oil seal loss is deter-mined from Figure200-27. The casing size in the figures is selected by comparing the cfmavg with the flow range in Figure 200-24.

233 Head/StageAlthough special impeller designs are available for higher heads, a good estimate for the typical multistage compressor is approximately 10,000 ft/stage. This is based on an assumed impeller flow coefficient of 0.5 and a nominal impeller tip speed of 800 fps.

The actual head per stage varies between manufacturers and individual impeller designs, ranging from 9,000 to 12,000 feet for 28 to 30 molecular weight gas at normal temperatures.

Head per stage is limited by:

• impeller stress levels• inlet Mach Number

H p zavg RT1rn 1–

n------------

1–n 1–

n------------

---------------------=

T2 T1rn 1–

n------------

=

GHPwH p

33000 ηp,-----------------------=



Impeller Stress LevelThe following speed margins are defined by API:

Fig. 200-26 Bearing Losses vs. Casing Size and Speed (Courtesy of Dresser-Rand)

Fig. 200-27 Oil Seal Losses vs. Casing Size and Speed (Courtesy of Dresser-Rand)

• Rated (Design) Speed: 100%

• Maximum Continuous Speed: 105% of Rated Speed

• Trip Speed: 110% of Maximum Continuous

• Overspeed: 115% of Maximum Continuous



Figure 200-28 identifies the impeller stresses at various rotational speeds. Reduced yield strengths required for corrosive gas will correspondingly reduce maximum head per stage through reduction in speed.

Inlet Mach NumberAn increase in gas molecular weight, or a decrease in k, Z or inlet temperature will result in an increase in inlet Mach Number. For high molecular weight or low temperature applications, Mach Number may limit head per stage for a given design.

234 Stages/CasingThe maximum number of stages per casing should normally be limited to eight. It is usually limited by rotor critical speeds, although in a few cases temperature can be a limiting factor.

Most multistage centrifugal compressors operate between the first and second criti-cals (flexible shaft rotor). Figure200-29 shows the location of critical speeds in relation to the operating speed range. API specifies the required separation between critical speeds and the compressor operating range. As the bearing span is increased to accommodate additional impellers, the critical speed decreases, with the second critical approaching the operating range. While some manufacturer's bulletins indi-cate as many as 10 or more stages per casing, designs exceeding eight impellers per case should be carefully evaluated against operating experience from similar units.

For compound, or sidesteam loads, additional stage spacing may be required to allow for intermediate exit and/or entry of the gas. In these applications, the number of impellers would be reduced accordingly.

235 Discharge TemperatureIf the calculated discharge temperature exceeds approximately 350°F, cooling should be considered to avoid problems with compressor materials, seal compo-

Fig. 200-28 Impeller Stress Levels at Various Speeds



nents, and clearances. The exact temperature limit is dependent on factors such as the gas compressed, compressor materials, allowable temperature of the seal oil, and the type of seals. Also, note that discharge temperature will increase as flow is reduced toward surge.

236 Selection ReviewRefer to Section 2100 for centrifugal compressor checklists, which provide typical items covered during the review of any centrifugal compressor quotation.

240 Machine Components and Configurations

241 Machine ComponentsCentrifugal compressors are made up of a casing with stationary internals, containing a rotating element, or rotor, supported by bearings. Shaft end-seals are provided to contain the process gas. Figure200-30 shows a typical multistage

Fig. 200-29 Rotor Response Plot (Courtesy of the American Petroleum Institute)



compressor and identifies the basic components. (Refer to Figure200-1 for details of the gas flow path.

The main machine components are:

• Casings• Nozzles• Stage• Diaphragms• Impellers• Rotor• Shaft• Radial Bearings• Thrust Bearing• Balance Piston• Interstage Seals• Shaft-end Seals

CasingsThe following is a summary of casing materials and their applications.

1. Cast Iron• Limited to low pressure applications for non-flammable, non-toxic gases.

• Limited in location and size of main and sidestream connections to available patterns.

2. Cast Steel• Quality is difficult to obtain.• X-ray inspection requirements increase costs.• High-rejection rate or involved repairs can extend deliveries.

3. Fabricated Steel• Used for both horizontally- and vertically-split casings.

• Improved quality control possible.

• Delays associated with rejection or repair of castings are avoided.

• Variable stage spacing provides minimum bearing span for required stages.)

• Main and sidestream nozzle size and location are not limited by pattern availability.

4. Forged Steel• Used for small vertically-split casing sizes where application involves very

high pressures.

Compressor M

anual200 Centrifugal Com

pressors

Chevron Corporation200-33

Novem

ber 2001

Fig. 200-30 Centrifugal Compressor Nomenclature (Courtesy of Demag Delaval)



All centrifugal compressor casings used to be cast. But, due to the problems associ-ated with quality control on large castings, coupled with improved fabrication tech-niques and costs, many manufacturers converted to fabricated steel casings, especially on the larger frame sizes.

NozzlesInlet and outlet nozzles are available in a variety of configurations, depending on the manufacturer. They are normally flanged. (Typical arrangements are shown later in this section.) API 617 covers requirements for flange type, and ratings of main and auxiliary connections.

The increased use of fabricated cases has provided additional flexibility in nozzle orientation.

If the installation permits, the following should be considered:

1. Horizontally-split units with process connections in the lower half (down-connected) allow removal of the top half, and internals including rotor, without disturbing the process piping.

2. If overhead process piping is required, the use of vertically-split barrel compressor casings still allow removal of the inner casing and access to the internals without removing process piping. Fabricated casing design makes the vertically-split unit a cost-effective alternative for larger medium pressure applications.

StageThe heart of the centrifugal compressor is the impeller “stage”. The stage is made up of the following parts (illustrated in Figure200-31):

• inlet guide vanes• impeller• diffuser• return bend (crossover)• return channel

The stage can be separated into two major elements:

• The impellers which are mounted on the shaft as part of the rotor.

• The stationary components including the inlet nozzle and other components mentioned above.

The inlet volute, or return channel, guides the gas to the eye of the impeller, and aided by the guide vanes, distributes the flow around the circumference of the impeller eye.

One method of adjusting the stage performance, is to use different guide vane angles. This changes the angle of incidence on the impeller which in turn varies the head, efficiency, and stability. There are three types of fixed guide vanes; radial,



against-rotation, and with-rotation. The influence of various guide vane angles on a given impeller head characteristic is shown in Figure200-32.

Diaphragms

The stationary members inside the casing are called diaphragms. The diaphragm includes a diffuser for the gas as it leaves the impeller, and a channel to redirect the gas through the return bend and return channel into the next stage. Diaphragms can be either cast or fabricated, with cast diaphragms normally made of iron. Normally, diaphragms are not exposed to high pressure-differentials, and therefore are not highly stressed. Diaphragms should be made of steel where high-differentials may exist (such as back-to-back impellers).

ImpellersThe impeller is the most highly stressed component in the compressor. Available types vary widely, although the three basic types are designated as open, semi-open and closed :

Open impellers have the vanes positioned in a radial direction and have no enclosing covers on either the front or back sides.

Semi-open impellers usually have the vanes positioned in a radial or backward leaning direction and have a cover on the back side which extends to the periphery of the vanes. The radial blade, semi-open impeller provides for a maximum amount of flow and head in a single stage, even in large diameter impellers (Figure200-33).

Closed impellers have enclosing covers on both the front and back side. This is the most common type in our large process compressors. The blades are usually back-

Fig. 200-31 Centrifugal Compressor Stage Components (Courtesy of the Elliott Company)



ward leaning, although they may be radial. Forward leaning blades are normally used only in fans or blowers. (See Figure200-33)

Single-inlet impellers take the gas in an axial direction, on one side of the impeller only, and discharge the gas in a radial direction.

Double-flow impellers take the gas in an axial direction, on both sides of the impeller, and discharge the gas in a radial direction. They are, in effect, the equiva-lent of two single-inlet impellers placed back-to-back and, in general will handle

Fig. 200-32 Head-Capacity Characteristics of Constant Speed Centrifugal Compressor with Capacity Regulated by Variable Inlet Vane Angle (Courtesy of Dresser-Rand)

Fig. 200-33 Impeller Types – Closed and Semi-Open Backward Leaning (Courtesy of Dresser-Rand)



twice the flow at the same head as a single-inlet impeller of the same diameter oper-ating at the same speed.

Some impeller designs utilize a three-dimensional blade or vane configuration, which varies the inlet blade angle from hub to outside diameter, thereby providing optimum aerodynamic geometry, and improved performance over that of two-dimensional designs.

Centrifugal compressor impellers discharge gas radially, but the gas enters in an axial direction. An axial flow element called an inducer is sometimes incorporated into the impeller. This combination is called a mixed-flow impeller. This configura-tion results in increased efficiency in high-flow applications.

In the past, riveted impeller construction was used in a large number of applica-tions. Today, construction with welded components is more common.

RotorThe rotor is made up of the shaft, impellers, impeller spacers, thrust collar, and the balance drum. Figure200-34 shows several rotor configurations with various impeller types.

If a rotor always operates below the lowest critical speed, it is known as a stiff-shaft rotor. In contrast, a rotor with a normal operating range above one or more of its criticals is a flexible-shaft rotor. Most multistage centrifugal compressors have flex-ible-shaft rotors; and therefore, must pass through at least one critical during start-up or shutdown. From an operational point of view, stiff shafts would be preferable. However, it is not practical since the shafts would become prohibitively large.

ShaftsShafts are made from alloy steel forgings, finished by grinding or honing to produce the required finish. Special requirements are detailed in API 617 for balancing and concentricity during rotor assembly. Impellers are normally mounted on the shaft with a shrink fit with or without a key, depending on the particular manufacturer and compressor frame size. Most manufacturers use shaft sleeves to both locate impellers and provide protection for the shaft in the event of contact with internal labyrinth seals.

Special attention must be given to minimizing mechanical and electrical runout at the shaft area observed by proximity probes. See the General Machinery Manual for more information on mechanical/electrical mount.

Radial BearingsRadial bearings on centrifugal compressors are usually pressure lubricated. For ease of maintenance, they are horizontally- split with replaceable liners or pads. The liners or pads are usually steel backed with a thin lining of babbitt.

Since centrifugal rotors are relatively light, bearing loads are low. This often leads to instability problems which must be compensated for by the bearing design. Due to instability, the straight-sleeve bearing is used only in some slow-speed units with



relatively short bearing spans. The pressure-dam sleeve bearing, and the tilting-pad bearing are two commonly used designs which improve rotor stability.

Fig. 200-34 Centrifugal Compressor Rotor Configurations (Courtesy of the Elliot Company)



The top half of the pressure-dam design is relieved as shown in Figure200-35, creating a pressure point where the dam ends. This conversion of oil-velocity into pressure adds to rotor stability by increasing the bearing load.

The tilting-pad bearing shown in Figure200-36 is usually made up of five indi-vidual pads, each pivoted at its midpoint. By adjustments to the shape of the pads and bearing clearance, bearing stiffness and damping characteristics can be controlled. This bearing is successful in applications where the pressure-dam design is inadequate.

Thrust BearingThe tilting pad is the most common thrust bearing used in centrifugal compressors. The flat land and tapered land bearings are used less frequently. Figure 200-37 shows a tilting-pad bearing, consisting of a thrust collar (collar disk) attached to the rotor shaft, and a carrier ring which holds the pads. A button on the back of the pad

Fig. 200-35 Pressure Dam Sleeve Bearing Liner (Courtesy of the Elliott Company)



allows the pad to pivot freely, thus allowing adjustment to varying oil velocity at different compressor speeds. A further refinement to the basic design is the self-equalizing bearing shown in Figure200-38. An equalizing bar design allows the bars to rock until all pads carry an equal load.

Balance PistonFigure 200-39 represents the pressure profile acting on a centrifugal compressor impeller, showing net pressure and net thrust pattern. This pressure pattern on the impeller results in a net thrust force towards the suction end of the machine. The total net thrust is the sum of the thrusts from all the individual impellers.

The rotor's thrust is handled by the thrust bearing. However, in most multistage compressors, a very large, if not impractical, thrust bearing would be required to handle the total thrust load, if not otherwise compensated. Therefore a thrust compensating device, or balance piston (or balancing drum) is normally provided as part of the rotating element.

As shown in Figure200-40, compressor discharge pressure acts on the inside end of the balance piston. The area on the discharge side (outside) is vented, usually to suction pressure. The resulting differential pressure across the balance piston develops a force which opposes the normal thrust force, thus greatly reducing the net thrust transmitted to the thrust bearing.

Thrust compensation can be regulated by controlling the balance piston diameter. However, there are usually physical and design limitations. Normally a balancing force less than the total impeller thrust (approximately 75%) is selected to maintain

Fig. 200-36 Tilting-Pad or Pivoted Shoe Radial Journal Bearing (Courtesy of the Elliott Company)



the rotor on one face of the thrust bearing for all operating conditions. Otherwise, the rotor could bounce back and forth between the thrust faces as process condi-tions vary.

Interstage SealsInternal seals are installed on multistage centrifugals to prevent leakage between stages, thereby improving performance. Labyrinth seals are commonly used, being located at the impeller eye and at the shaft between stages. Figure200-41 illustrates internal labyrinth seals.

Fig. 200-37 Button-Type Tilting-Pad Thrust Bearing (Courtesy of the Elliott Company)



Shaft End-SealsCentrifugal compressors use shaft end-seals to:

1. Restrict or prevent leakage of air or oil vapors into the process gas stream.

2. Restrict or prevent leakage of process gas from inside the compressor.

Fig. 200-38 Self-Equalizing Tilting-Pad Thrust Bearing (Courtesy of the Elliott Company)

Fig. 200-39 Impeller Pressure and Thrust Patterns (Courtesy of the Elliott Company)



Various types of seals are used, depending on the gas being compressed, the pressures involved, safety, operating experience, power savings, and process requirements.

Shaft end-seals are separated into two broad categories:

• the restrictive seal which restricts but does not completely prevent leakage; and

• the positive seal designed to prevent leakage.

Restrictive seals are usually labyrinths. They are generally limited to applications involving non-toxic, non-corrosive, abrasive-free gases at low pressures. In some cases, ports for injection or withdrawal of the gas are used to extend the range of effectiveness. Some possible arrangements are shown in Figure200-42.

Another form of the restrictive seal is the dry carbon ring seal, often used on over-hung single-stage compressors where maximum sealing and minimum axial shaft spacing are important. Since this seal can be held to close clearances, leakage is less

Fig. 200-40 Centrifugal Compressor Balance Drum (Balance Piston) (Courtesy of the Howell Training Group)

Fig. 200-41 Interstage Seals (Courtesy of Dresser-Rand)



than with the labyrinth seal. Also, less axial shaft space is required (see Figure 200-43).

Positive seals, while varying somewhat in design between manufacturers, are either liquid-film or mechanical contact type.

The liquid-film type is shown in Figure 200-44. A schematic of a seal system is shown in Figure 200-45. Sealing oil is fed to the seal from an overhead tank located at an elevation above the compressor set to maintain a fixed five psi (typically) differential above “seal reference” pressure. (Seal reference pressure is very close to suction pressure.)

The oil enters between the seal rings and flows in both directions to prevent inward leakage to the process gas or outward leakage of the gas to the atmosphere. “Buffer ports” are often available for injection of an inert gas to further ensure separation of the process from the sealing medium. The oil-film seal is suitable for sealing pres-

Fig. 200-42 Ported Labyrinth Seals (Courtesy of the Elliott Company)

Fig. 200-43 Buffered Dry Carbon-Ring Seal (Courtesy of the Elliott Company)



sures in excess of 3000 psi. (See Figure200-46 for an illustration of a buffer-gas injection.)

Fig. 200-44 Liquid (Oil) Film Seal (Courtesy of Dresser-Rand)

Fig. 200-45 Oil Film Seal Schematic (Courtesy of Dresser-Rand)

Fig. 200-46 Oil Film Seal with Buffer to Separate Seal Oil from Bearing Oil (Courtesy of Dresser-Rand)



The tilting-pad oil seal (shown in Figure200-47) is a design that recognizes that in some cases the seal operates as a bearing. It can be used in high-pressure, high-pres-sure-rise applications to improve rotor stability.

The mechanical contact seal (Figure200-48) is used at pressures up to 1000 psi, and has the added feature of providing more positive sealing during shutdown. Sealing is provided by means of a floating carbon ring seal riding between a stationary and a rotating face. The seal medium (oil) functions primarily as a coolant. Seal oil differential is controlled by a regulator rather than an overhead tank.

Fig. 200-47 Tilt-Pad Oil Film Seal (Courtesy of Dresser-Rand)



242 Dry Gas SealsDry gas seals represent the latest technology for compressor shaft end sealing, and are currently the preferred sealing technology for most centrifugal compressor applications. Under dynamic (rotating) conditions, dry gas seals function as restric-tive seals. Depending on the design and conditions, dry gas seals can behave either as restrictive or positive seals under static conditions. Similar to pump mechanical seals, dry gas seals use mating faces to create the sealing interface between the rotating and stationary parts. The seals depend on a fine balance between pressure forces, closure spring forces and aerodynamic forces that are created by very

Fig. 200-48 Mechanical Contact Seal (Courtesy of the Elliot Company)



shallow grooves or depressions on one of the seal faces, as shown in Figure 200-49). This balance results in a face gap of about 0.0001" to 0.0002", through which the seal leaks at very low rates. Leak rates are approximately propor-tional to seal size, sealing pressure and rotational speed, and are influenced to a lesser extent on gas conditions. Depending on these parameters, leakage rates gener-ally range from fractional SCFM to about 4 SCFM. Although the dry gas seal design concept first achieved significant commercial use in the early 1980’s, it can be traced back to the early 1950’s. Dry gas seal technology is presently also applied in both steam turbines and pumps, but this section will address only centrifugal compressor applications. Dry gas seals are an advancing technology in the petro-chemical industry, so it is important to be aware of the age of information (including this Gray Manual section), as well as the duration of successful field experience for any given advance.

In general, dry gas seals offer the following primary advantages compared to other sealing technologies:

• lower leakage rates and improved pressure capability vs. other restrictive type seals, and

• simpler, more efficient and lower cost operation and auxiliaries vs. other posi-tive type seals.

Dry gas seals can offer additional advantages as well, all of which should be consid-ered in the economics if justification for gas seals is needed (see Application Considerations section). Justification is usually an issue for retrofits, but on new compressors, economics are favorable, especially if the alternative design requires expensive and/or inefficient auxiliaries (seal oil systems, eductor systems, etc.).

The primary advantages of gas seals are the result of an advanced and precise design that relies heavily on the proper operating environment. Reliable operation is extremely dependent on having seal gas (the gas seen by the seal faces) which is free of particulates and liquids. In addition, the reliability of designs can be compro-

Fig. 200-49 Dry Gas Seal Rotating Face Segment, Shown with Exaggerated Depth Groove Geometry. (Second cross-sectional view shows operating face gap.) Courtesy of Flowserve Corporation

Face Rotation

Gas Path

Rotating Face

StationaryFace



mised when approaching current experience envelopes in sealing pressure, sealing temperature and seal face surface speeds, either singularly or in combination. All of these issues focus on assuring the proper gas film and stress levels at the seal faces. Other vulnerabilities include seal face hang-up (caused by sticking of o-rings used for secondary sealing of the faces), reverse rotation, reverse pressurization and lube oil contamination of the seal faces. Many of these vulnerabilities are associated with earlier gas seal designs, and have been reduced or eliminated with design advances.

ArrangementsDepending on the application, one or two pairs of faces may be used in various arrangements, usually in conjunction with labyrinth seals, to achieve the desired process gas containment level. One pair of faces (a single seal) may be used for moderate pressure applications that are neither flammable, toxic nor environmen-tally harmful (air, nitrogen), since the normal seal leakage will be to atmosphere. However, low pressure services suitable for a single seal are also suitable for a laby-rinth seals, which offer greater simplicity and reliability, as well as significantly lower initial cost. A single seal arrangement is shown in Figure200-50.

More typical applications require a dual seal arrangement to further limit or prevent leakage to atmosphere, as well as to provide a back-up (or secondary) seal in the event of a failure of the inner (or primary) seal faces. Dual seals can be provided in either a double seal arrangement or a tandem seal arrangement. Double seals are oriented in an opposed fashion to contain seal gas (sometimes called barrier gas in

Fig. 200-50 Simplified Single Seal Arrangement, Shown Without Primary Seal Labyrinth Courtesy of Flowserve Corporation

Clean Seal Gas

PROCESS ATMOSPHERE

Leakage



double seals) supplied between the two seals from an external source (see Figure 200-51).

The seal gas must be available at all times at a pressure higher than the process gas pressure at the seals (or the sealing pressure). Although the sealing pressure is usually very close to suction pressure during operation, a compressor trip can cause sealing pressure to rise to a settle-out pressure in some compressor circuits. The double arrangement is most desirable when nitrogen can be used as the seal gas, especially when emissions containment is of primary concern. The double arrange-ment is also desirable when there is a high potential for primary seal reverse pres-surization in a tandem arrangement (see the Seal Gas Supply and Venting Systems section). The double arrangement allows a small amount of seal gas leakage both into the compressor across the primary seal, and also to atmosphere across the secondary seal. When using properly filtered nitrogen as the seal gas, it provides both dry and clean conditions for both seals, prevents harmful emissions to atmo-sphere and requires a relatively simple auxiliary system. In services where the process gas is either wet or dirty, it may still be necessary to use a purge gas to keep liquids and solids away from the primary seal. It is important to consider that a reduction of nitrogen pressure below the sealing pressure will result in process gas emission and possible damage to the primary seal faces, so some back-up or safety provisions may be needed to avoid these consequences (see the Seal Gas Supply and Venting Systems and Shutdown Protection sections). Furthermore, if the nitrogen supply is known to have poor reliability, a tandem seal arrangement may be the best choice. Since nitrogen is not always available at high enough pressures, double seal arrangements are usually limited to lower pressure services such as FCC or coker wet gas.

The tandem seal arrangement is the most commonly used on compressors, espe-cially in moderate to high pressure services. The seals are oriented in tandem to

Fig. 200-51 Simplified Double Seal Arrangement Shown Without Primary Seal Labyrinth Courtesy of Flowserve Corporation

PROCESS ATMOSPHERE

Seal Gas(Barrier Gas)

Leakage Leakage



restrict outward leakage (see Figure200-52), although the primary seal normally provides essentially all of the sealing duty. In addition to acting as the back-up seal,

the secondary seal provides emission containment under normal conditions. The cavity between the two seals is typically vented to flare (or safe location) through porting in both the seal housing and compressor. If the seal gas is environmentally harmful, toxic or has the potential to be toxic, a tandem seal with an intermediate (or interstage) labyrinth should be selected, provided there is an inert gas available for buffering. The intermediate labyrinth is located between the primary and secondary seals, so pressure in this cavity is normally very low. A port between the labyrinth and the secondary seal allows the buffer gas (typically nitrogen) to flow across the labyrinth, preventing seal gas from reaching the secondary seal. Most of the buffer gas exits the seal through the primary seal vent, which is piped to the flare, while a smaller amount leaks across the secondary seal. The tandem seal arrangement generally requires the most extensive auxiliary system, which must deliver seal gas, deliver buffer gas (if needed), and monitor seal venting conditions. The tandem arrangement allows for seal gas to be supplied from either the compressor discharge or an external source, provided the external source pressure exceeds the sealing pressure. As a result, tandem arrangements are currently the only choice for moderate to high pressure services.

A labyrinth seal just inboard of the primary seal is often included in the design of any of the above three arrangements (see Figure200-53). This inner or primary seal labyrinth:

• limits leakage to atmosphere in the event a primary seal failure (this function is mostly for single seals)

Fig. 200-52 Simplified Tandem Seal Arrangement with Intermediate Buffered Labyrinth Courtesy of Flowserve Corporation

PROCESS ATMOSPHERE

Inert Buffer GasLeakage

Inert BufferGas

Intermediate labyrinth



• prevents large amounts of seal gas from flowing into the compressor, and

• minimizes the chance of solids and liquids from getting close to the primary seal faces.

The inner labyrinth seal can either be integral to the seal assembly or provided as a separate compressor component. Similarly, labyrinths can be used on the outboard side of the seal assembly to prevent bearing lube oil from contaminating the seal faces (this and other options are described in better detail in the Separation Seal section). For either application, the use of abradable seals (rotating labyrinth teeth running within a soft, non-metallic, close-clearance stationary ring) should be avoided, as users have experienced failures due to excessive heat generation and particulates generated from the abradable material. Properly engineered abradable seals continue to be acceptable for interstage and balance piston sealing.

Seal FacesSeal face materials and designs vary between different suppliers. Since the seal faces are the components that have the greatest influence on the operating enve-lope, reliability and leakage rate, they are the focus of ongoing design improve-ments. Face designs must be optimized to address numerous issues, including:

• Hydrostatic lift (slight separation of the faces caused by pressure while rotor is static)

• Dry running tolerance • Lift-off properties• Gas film stiffness

Fig. 200-53 Simplified Tandem Arrangement Showing Shrouded Seal Face Design, Primary Seal Labyrinth, and Separation Gas Arrangement Courtesy of Flowserve Corporation

Seal Gas Primary Seal Vent Inert Separation GasSecondarySeal Vent

Primary SealLabyrinth

Separation Gas Labyrinth Seal

Seal Face Shrouding



• Seal gas properties and their variability• Stresses and deflections due to sealing conditions (pressure and temperature)• Stresses and deflections due mounting/driving forces and dynamic forces• Tolerance to reverse rotation

Some of these issues are addressed with the face materials. Earlier designs of dry gas seals typically used tungsten carbide for the rotating seal face and carbon for the stationary seal face. Current designs are making greater use silicon carbide, silicon nitride or in some cases, a coated, ductile steel for rotating faces. Silicon carbide has also become the popular alternate material for the stationary face, especially when high pressures raise deformation to unacceptable levels in carbon materials. At present, and depending on the supplier, low to moderate duty services use either tungsten carbide/carbon or silicon carbide/carbon face combinations, while high pressure, high speed services use silicon carbide/silicon carbide or silicon nitride/silicon carbide combinations. Other material combinations have been used, especially at extreme conditions.

It is important to note that except for coated ductile steel faces, rotating face mate-rials are very brittle, making them vulnerable to excessive stress with the potential to break up very quickly. Although silicon carbide and silicon nitride tend to disin-tegrate in to very small pieces, these pieces can still upset or damage the secondary seal. In contrast, once tungsten carbide is broken, sizable fragments can cause significant secondary damage to the entire seal assembly and even the compressor. In order to mitigate damage or unsafe conditions in the event of a failure, a shrouded face design (see Figure200-53) should be specified for a tungsten carbide rotating face, if not provided as the standard. In addition to providing burst contain-ment, the shroud also offers the ability to drive the rotating face at its outer diam-eter, which results in reduced face stresses. For silicon-based faces, burst containment is of less value, and reduced heat transfer is a trade-off. However, the stress reduction provided by outer diameter drive methods may be desirable or necessary for some applications.

Seal face groove geometry also varies between suppliers, and has evolved over the years. Most suppliers offer both unidirectional and bi-directional face designs. Unidirectional faces typically have a spiral groove geometry, although L-shaped grooves have also been used (see Figure200-54). Unidirectional seals generally offer the best performance with regard to lift-off, gas film stiffness and stability , making them the best choice for many of the more difficult applications. Unidirec-tional seals have the disadvantage of being intolerant of reverse rotation, which can cause dry running damage to the faces. Because of this vulnerability, it is important to incorporate assembly features (both labeling and geometry differences) which can help prevent the installation of the wrong seal parts or assemblies (inboard vs. outboard) on a between bearings compressor design.

Bi-directional face designs have a greater variety of groove geometries among suppliers, including U-shapes, “spruce tree”-shapes and T-shapes, (see Figure 200-55). Bi-directional designs offer equal performance in both directions of rotation, but this performance is generally less than that of a unidirectional seal. They are acceptable for services where gas film stiffness and face separation is not



marginal (best evaluated by the supplier). They may be most attractive in services where compressor flow reversal potential exists (i.e., back pressure services rather than recycle services), especially if there is a history of compressor discharge check valve problems. One other advantage of bi-directional designs is that one seal assembly can be used to spare both sides of the machine. However, this should play little or no role in selecting a bi-directional design, especially considering that seal assemblies are usually changed out in pairs, and critical services warrant having a full set of spares.

Secondary Sealing ElementsSecondary sealing elements (different than the secondary gas seal in a dual arrange-ment) provide sealing between the seal assembly and the compressor, as well as between various seal components. Typically, elastomeric o-rings are used as the secondary sealing elements, although other seal types are used to address specific problems. Most secondary sealing elements are static (once parts are assembled, there is no movement of the parts that form the joint). As with other machinery

Fig. 200-54 Unidirectional Seal Face Groove Geometry Courtesy of Flowserve Corporation

Fig. 200-55 Bi-direction Seal Face Groove Geometry Courtesy of Flowserve Corporation



applications, it is important to select materials that are compatible with the normal and potential gas streams seen by the seals. In addition, high pressure applications must be evaluated for the potential of extrusion and explosive decompression (the latter is a function of sealing pressure, gas composition and compressor system decompression rate). High pressures may require the use of high Durometer elas-tomers or polymer (such as PTFE) materials. The polymer seals often use metallic springs to provide the proper contacting or energizing force.

In addition to static secondary sealing elements, there are also dynamic secondary sealing elements, which seal the moving joints between the stationary seal faces and their retainers or housings. The stationary seal faces must move axially to accom-modate lift-off, gas film thickness changes and axial movement or thermal growth of the rotor. The dynamic capability of the stationary face secondary seal is another critical performance and reliability aspect of dry gas seals, since sticking or hang-up of this seal can result in either excessive leakage or damaging face contact. Poten-tial design options to minimize seal hang-up include spring energized polymer seals or spring energized o-rings, both of which reduce o-ring contact forces (spring ener-gized o-rings are shown behind stationary seal faces on prior seal arrangement drawings). Some spring energized designs are also claimed to provide at least some degree of reverse pressurization tolerance. Although this may be a benefit for some applications, at this time, there is insufficient data and experience to support relying on this feature to eliminate or even reduce measures for preventing reverse pressur-ization.

Installation and maintenance should always be considered in the secondary sealing element joint design, especially those between the seal housing and compressor casing, and the seal sleeve and shaft. Optimum o-ring placement and tapered diam-eter changes can minimize or eliminate the potentially damaging action of sliding o-rings across components during installation, as well as reduce potential for o-rings falling out of ID grooves during installation. Besides sealing, o-rings between the shaft and seal sleeve may also serve to center the rotating parts of the seal on the shaft. Since seal sleeves are part of the seal assembly, and must have sufficient clearance for a cold slip fit, o-rings can act as a low friction centering device. More critical applications may require metallic centering devices.

Separation Seal and Separation Gas SupplyPreventing bearing lube oil from contaminating the seal faces is a key element of seal reliability. Provisions are often necessary to accomplish this when the bearing and seal are in close proximity, especially if the span between the bearing and seal is contained within a housing. A restrictive seal in conjunction with inert gas purging (separation gas), are typically used to form a barrier for the gas seal assembly (see Figure200-53). The restrictive seal is usually a radial clearance seal in a lantern ring arrangement (separation gas enters between a pair of close clear-ance seals), and may either be a labyrinth or close clearance carbon ring design. A labyrinth design will typically consume in the range of 5 SCFM of separation gas per machine end, while a carbon ring can reduce this rate by at least one half.

When available, nitrogen is preferred as the separation gas for compressors in combustible gas services. Although air has been used in some applications, it has



the potential for creating combustible mixtures in the cavity between the separation seal and the gas seal (the outer seal vent area). At present, excess purge gas (25 or more SCFM) either to the separation seal or directly to the outer seal vent cavity is a solution used by at least one compressor OEM. Membrane units that generate nitrogen from an air supply may provide an acceptable alternative solution provided the membrane system is sized and designed to achieve the proper nitrogen purity.

Seal Gas Supply and Venting SystemsDepending on the service and seal arrangement, seal gas can either be supplied from the compressor gas stream or from an external source. As previously described, double seals will inherently require externally supplied inert seal gas. Tandem seals or single seals can use either compressor discharge gas or an external supply of gas. Examples of the latter include nitrogen, hydrogen, fuel gas and other by-product gases. Determining factors include the availability and cost of a suitable and reli-able external gas supply and the characteristics of the gas from each source (cleanli-ness, liquid/moisture content, toxicity, thermodynamic properties, etc.).

Note In order for dry gas seals to operate reliably, it is essential that a constant and sufficient supply of seal gas be delivered in a clean and dry condition. Although this appears straightforward, gas seal failures are often a result of not meeting this requirement.

Some of the unanticipated conditions that may be encountered include:

• Loss of externally supplied seal gas

• Insufficient seal gas while compressor is at idle speed or stopped

• Excessive supply system pressure drop in low pressure (close to 1 atmosphere) services

• Reverse pressurization if suction pressure and seal gas fall below vent pressure (possible during startup in low pressure services)

• Reverse pressurization if vent pressure rises above seal gas pressure (possible during flare system excursions in low pressure services)

• Reverse pressurization on loss of seal gas supplied to double seals

• Saturated seal gas due to changes in the process

• Liquid formation in the seal gas due pressure letdown cooling

• Oil mist contamination from external seal gas source compression system

• Filter element failures due to pressure pulsations from an external seal gas source supply system

Seal gas supply systems should include the following features, where pertinent:

Seal gas back-up supply for external seal gas. If there is any chance of loosing seal gas supply from an external source, the seal gas system should include auto-matic cut-in of a back-up source of gas. Tandem seals should typically use



discharge gas as a back-up. In this case, it might be necessary to design for poten-tial differences in liquid removal requirements between the different types of seal gases. Double seals using nitrogen can be backed up with nitrogen bottles, although this supply will be limited. If bottles are used, pressure monitoring/alarming of the bottles should be included in the design in order to assure their readiness over many years. In addition, an alarm and a compressor shutdown are also recommended in the event that seal gas pressure approaches the actual sealing pressure. The shut-down set point should be selected to allow safe coast down before the primary seal differential pressure reverses enough to cause damaging face contact.

Seal gas back-up supply for compressor discharge seal gas during idling. At slow roll speeds or even when stopped, there may be enough gas force to provide lift-off of the primary seal faces. Current gas seal designs are pressure balanced to provide hydrostatic lift-off at a target pressure without rotation, in order to mini-mize rub damage on start-up. Without sufficient discharge pressure to provide seal gas flow, compressor gas stream particulates can enter the seal faces during idle time and cause damage once rotating speeds are sufficient. In order to prevent this contamination, an external seal gas supply can be used, again with proper attention to liquid removal requirements. As an alternative, packaged pressure boosting systems can be used to raise the seal gas pressure when compressor discharge pres-sure is inadequate. Either alternative should be designed for automatic cut-in, as discharge pressure falls. Note that a back-up seal gas supply may not be necessary if the compressor is not capable of slow roll, sealing pressure is below the hydrostatic lift-off pressure, and the process gas is relatively clean.

Seal gas supply filtration. The filtration system should include five micron (nominal) duplex filters, arranged in parallel with individual isolation valves to allow for on-line element changes. A differential pressure indicator and high DP alarm should also be included for monitoring filter element condition. If seal gas is to be provided from an external source fed by reciprocating compressors, filter elements should be robust enough to withstand pressure pulsations in the system.

Seal gas supply control for tandem and single seals. The preferred seal gas supply method for tandem seals is currently with differential pressure control above a reference pressure. Measurement point options for the reference pressure include the seal balance line, the thrust balance line or a seal cavity port. When the system is supplying two seals, the supply flow to each seal can be balanced by installing an orifice in each of the individual supply lines. Individual flow indicators and throt-tling valves that are parallel to each of the restriction orifices can also be installed to provide some degree of adjustment. Valve arrangements that can completely cut off seal gas flow to a seal should be avoided. The proper set point range for the differ-ential pressure control should be determined by the compressor and seal suppliers, based on the design of the compressor and seal gas labyrinths, the reference gas measurement point and optimization of the seal gas consumption rate. Additional considerations must be made for lower pressure services to prevent pressure rever-sals due to either venting system pressure excursions or vacuum conditions that can occur during compressor start-up. The seal gas supply control system should also include differential pressure alarms, and provisions for on-line maintenance of the



differential pressure control valve (manual bypass with local DP indicator and isola-tion valves). Single seals can be similarly arranged.

Seal gas supply control for double seals. In most cases, a simple back pressure control regulator or valve can be used to deliver inert seal gas, such as nitrogen, to the seals. The set point should be sufficiently high to assure a positive pressure differential above the maximum expected sealing pressure, including the settle-out pressure. The control regulator or valve can have a manual bypass and isolation valves to allow on-line maintenance. As previously described, a low pressure alarm and shutdown are also required to avoid seal damage.

Seal gas liquid removal. Depending on the nature of the potential liquid in the seal gas streams, different methods of liquid removal and control may need to be employed. Liquids can be carried into the seal gas system directly, or they can condense out of the gas stream in various parts of the seal gas system, including at the primary seal faces. Careful thermodynamic evaluation of each seal gas stream (normal and back-up) under existing and estimated future conditions is key to iden-tifying problematic conditions and designing the appropriate facilities. External seal gas supplies can also contain lubricating oil if reciprocating compressors are part of the supply system. For retrofits, confirm process conditions by analyzing actual gas samples, with special attention to capturing and identifying liquids. On new installa-tions, pessimistic expectations for a clean and dry gas are recommended. It is impor-tant to design the entire seal gas supply system to achieve the desired goal of delivering clean and dry seal gas to the seals. Design strategies include:

• Use of stainless materials for supply system lines and components downstream of the filters.

• Use of, and proper location of, coalescers: Depending on how the gas behaves as it is let down, the location of the coalescer could be upstream or down-stream of the supply control valve. The coalescer should be properly selected and sized for the normal and extreme conditions (note that over sizing can allow gas to cool excessively or reduce coalescer efficiency). The device should include a sight glass, a differential pressure indicator and valving to allow element changes on-line without disrupting flow. Services that are normally dry may use a bypass rather than a duplex arrangement, but the bypass should be routed to avoid low points.

• Use of heaters and tracing downstream of the liquid removal device: The intent is to provide adequate margin (some users target 35-50 degrees F) from satu-rated conditions. Note that conventional steam tracing may not provide the needed reliability, necessitating either temperature monitoring or alternative tracing/heating methods.

• Prevention of liquid accumulation in lines/vessels: Piping should be routed without low points and also sloped to avoid liquid accumulation. Idle lines (bypass loops, duplex filter legs) should also be arranged to prevent liquid accumulation, or should include low point bleeders where this is not practical. In addition, liquid removal vessel low points should be piped to the compressor



suction for continuous blow down operation. The blow down line should include an orifice to avoid excessive recycling of gas back to suction.

• Promoting controlled condensation: In cases where it is not practical to heat the seal gas to achieve adequate superheat, a cooler upstream of the liquid removal device can be installed to knock out additional liquid. A lesser amount of gas heating downstream of liquid removal can then provide the desired super-heating.

Seal gas monitoring. The seal gas should be monitored for proper pressure, flow and temperature. The following devices and locations are suggested:

• Pressure indication (and recording for critical services) downstream of pres-sure regulator or control valve. Measurement should be consistent with control method (differential vs. absolute). Low and high pressure (or differential pres-sure) alarms should be included at the same location.

• For double seal arrangements, a low, low pressure shutdown (see Shutdown Protection section).

• Flow indication (and recording for critical services), with high and low alarms on seal gas supply header. Orifice-type flow meters are recommended.

• Except when using nitrogen, temperature indication (and recording for poten-tially saturated gases), with high and low alarms on seal gas supply header downstream of all pressure significant pressure drops.

• High temperature alarm, if there is a possibility of seal gas temperature approaching mechanical limits of seal.

Vent system and monitoring. Leakage from the primary seal in a tandem seal arrangement must generally be vented to a safe location such as a flare system. The venting system should allow for safe disposition of leakage under normal and emer-gency situations, and also have capability for leakage monitoring and excessive leakage warning for each primary seal. In low pressure services, any check valves used in the vent system must have sufficiently low opening pressures to avoid creating excessive back pressure that could result in reverse pressurization of the primary seal faces. The following devices and locations are suggested for leakage monitoring:

• Flow indication (and recording for critical service) at each primary seal vent. Orifice-type flow meters are recommended. Sizing should be for normal leakage rates, up to 3-4 times the normal design rate.

• A pressure switch located between the seal and an orifice in the vent line. Sizing of the orifice and setting of the pressure switch should trigger a high pressure alarm at about 4 times the design leakage rate.

• A pressure switch with a setting to trigger a high, high pressure alarm at a rate or pressure that would allow for a safe manual or automatic shutdown (see the Shutdown Protection section).



Design of the above systems should also take into consideration keeping the inter-mediate seal cavity pressure low enough to allow continued inert buffering of an intermediate labyrinth, if so equipped. Likewise, the normal flow of the buffer gas out the primary seal vent should be considered in sizing of the above devices.

Note Seal gas supply and venting system panel design has been very variable and a potential source of reliability, operability and maintainability problems. It is important to specify requirements, review designs, and where possible, test the auxiliary panels to better assure satisfactory performance.

Intermediate Labyrinth Buffering and Separation Gas SystemsClean nitrogen must be supplied to intermediate labyrinths in a fashion similar to providing external seal gas to a double seal. A five micron (nominal) duplex filter arrangement with differential pressure indication and high alarm, pressure regula-tion, flow indication and a low supply pressure alarm are required. To reduce components and complexity, the filtration for the buffer system can also serve the separation gas system, however, independent flow measurement, pressure regula-tion and low pressure alarming are recommended. If separation gas is to be supplied independently (use of air or no buffering of intermediate labyrinth), filtration can be relaxed if a labyrinth-type separation seal is used.

Shutdown ProtectionThe gas properties, seal arrangement and the criticality of the service may dictate a need for a protective shutdown. The purpose of the protective shutdown would be to either prevent a hazardous release of gas and/or prevent damage to the seal. In the case of a double seal arrangement, loss of seal (barrier) gas could result in reverse pressurization of the primary seal, leading to possible face damage and/or release of gas. A protective trip on low seal gas supply pressure is strongly recommended for this application. An exception can be made if there is specific testing to assure reverse pressure capability, and the process gas is non-toxic.

On tandem seals in toxic services, a protective shutdown is also strongly recom-mended for personnel protection. The shutdown should activate when the primary seal vent pressure exceeds the capabilities of the nitrogen buffer to the intermediate labyrinth. For other tandem seals or a single seal, a protective shutdown on high primary seal vent pressure can be used to minimize seal damage or limit gas leakage to atmosphere. Since gas seals have demonstrated rapid failure in some instances, the impact of a compressor shutdown with little or no warning must be considered when deciding on protective shutdowns and their set points.

Application ConsiderationsThe dry gas seal is the preferred method of shaft end sealing for nearly all new centrifugal compressors in hydrocarbon services. However, there are limits that warrant additional expert review, confirmation of user experience and perhaps special design and testing activities. These cautionary limits include:

• Surface speeds in excess of 400 feet per second• Sealing pressures in excess of 2000 psig, and



• Sealing temperatures in excess of 300°F (move pointer to yellow to see comments)

In addition, combinations of two or more values approaching the above warrant similar scrutiny. When checking user experience for similar services or conditions, it is important to go beyond the vendor installation list and actually contact at least some of the end users to confirm success and/or retrieve lessons learned. In most cases, the seal gas auxiliary system will require more scrutiny than the design of the seal assembly.

Retrofit applications can be more difficult to justify, since the cost of the auxiliaries for the original seal are already sunk. Typically, retrofits are justified primarily by reliability, and in some cases, seal performance. If credits associated with these improvements are insufficient, the following other factors should be included in the justification, if applicable:

• Elimination of venting gases from an oil seal trap system

• Elimination of driver power for seal oil pumps

• Reduction of main driver power draw due to lower seal parasitic loses (20-25 HP/seal for oil bushing seals)

• Elimination of make-up seal oil and disposal or reconditioning of contaminated seal oil

• Reduction of seal system auxiliaries maintenance and testing

• Elimination of motive gas consumption (steam and/or process gas) for laby-rinth seal eductors

• A potential reduction in other utilities used by auxiliaries (air, N2, steam, cooling water, etc.)

The above must be offset by costs associated with dry gas seals, which are gener-ally much less.

Retrofits on compressors with oil seals can have a significant impact on rotor dynamics. It is essential to have a rotor dynamic analysis conducted with expert review to identify potential problems and solutions. Retrofits on compressors with labyrinth seals will generally result in less rotor dynamic effect, but depending on geometries and existing rotor dynamic margins, there may be a potential issue. A pre-analysis expert review of these situations is suggested.

The selection table (Figure200-56, at the end of this section) is to assist in selecting an appropriate sealing arrangement and preliminary auxiliary scheme for most services. When using this table, also refer to the text of this section to better eval-uate decisions and options.

Figure 200-57, Figure200-58 and Figure200-59 are schematics of typical dry gas seal auxiliaries used for the specific arrangements and schemes.



Maintenance ConsiderationsThe complexity and sensitivity of dry gas seals generally requires two maintenance provisions that are not typically needed with other compressor seal designs. These include:

• Sparing of the complete seal assemblies• Inspection, overhaul and testing by a qualified repair facility (generally the

OEM).

Once an assembly is tested, it should not be altered prior to installation.

Testing should include the following measurements on both primary and secondary seals:

• Static leakage rate at various pressure levels, up to settle-out or discharge pres-sure

• Break-away torque at the same above pressures

• Dynamic (rated speed) leakage rates at various pressure levels, up to suction pressure plus 25% or 10 psi (whichever is greater)

• Leakage rates at rated speed and sealing pressure after 1 hour run

• Static leakage at settle-out or discharge pressure immediately after 1 hour run

• Leakage rates during 15 minute run following a “hot restart” (within 5 minutes of 1 hour run, at settle-out pressure).

Additional testing can be done to simulate emergency shutdown, depressurization and reverse pressure scenarios.

Seals that have been stored for long periods of time may need to have secondary sealing elements changed to counter relaxation and degradation that can occur during long storage durations or the upcoming time in service. The OEM should be consulted for specific recommendations for a particular seal design. The sealing element changes should also be done by a qualified repair facility, and the seals should again be tested to verify the integrity of the parts and the re-assembly work.

The suggested initial seal inspection interval is five years, until inspection results and performance demonstrate that longer intervals are possible. Immediate or earlier inspections should be considered if there is an event that compromises the environ-ment of the seal faces. Some event examples, in approximate rank order of urgency (high to low), include:• Filling of the compressor case with liquid• Loss of filter integrity• Loss of seal gas or purge gas• Degradation of leakage performance• Extreme compressor vibration levels• Loss of separation gas



• Prior inspection results that indicate problems not yet resolved. A qualified manufacturer’s representative should be present to perform or oversee, installation of seals, at least until plant personnel are very comfortable with proce-dures and methods. Use of the representative for seal removal is also advised, espe-cially if it has never been performed by plant personnel.


Compressor M

anual

Novem

ber 2001200-64

Chevron Corporation

Fig. 200-56 Seal Arrangement and Scheme Selection Chart

Single seal(also consider

labyrinth seals forlow pressures)

Discharge gas

Tandem seal

Double seal(see section text

for potentialaltering factors)

Tandem sealw/ intermediatelabyrinth & N2

buffer

ARRANGEMENTPRIMARY SEAL GAS

(filtered and dried)

Discharge gas

Nitrogen.Also use a dried/filtered purge

gas to keep process gascontaminants away from

primary seal.

Discharge gas

BACK-UP SEAL GAS(filtered and dried)

External supplygas or amplifiedprocess gas (2)

Bottled Nitrogenfor b/u

(optional, if lowN2 pressure s/d is

tolerable)

External supply(primary) gas

IDLE SPEED SEAL GAS(filtered and dried)

None

None

Discharge gas

None

External supplygas

or amplifieddischarge gas (2)

None

Optional(high ventpressure)

Amplifieddischarge gas (2)

PROTECTIVESHUTDOWN

Optional(high primary vent

pressure)

Low seal gas (N2)supply pressure

High primary ventpressure for toxic

gases


gases


gases

OR

NOTES:(1) At normal gas seal leakage rates.(2) Idle speed seal gas may not be necessary ifprocess gas is clean, compressor does not slow rolland sealing/settle-out pressure is less than lift-offpressure.(3) Consider membrane unit option also.(4) Compatible to process and available atpressures consistently above sealing and settle-outpressures.

External supplygas or amplifiedprocess gas (2)

YES

YES

YES

YES

NO

NO

NO

NO

EXTERNAL

DISCHARGE

None

Processgas non-flamable,

non-toxic andenvrionmentally

acceptable?

Process gasnon-toxic and

environmentallyaccteptable?

(1)

Reliable N2 supply 25 or more

psi greater thansealing & settle-out

pressures?

Weigh pros/consof external supplyvs. discharge gas

External supply ofreliable/compatible

gas? (4)

N2 available forbuffer gas?

(3)

YES

Use alternativeseal technology.

NO

Compressor M


pressors


Novem

ber 2001

Fig. 200-57 Seal Gas Supply Schematics

DPI

TIFI

DPI

CompressorDishcarge

Continuous blowdownto compressor suction

Coalescing device w/DP indicatorand high DP alarm (1). Level

indicator is optional. Bypass can bereplaced by second coalescing

device.

NC

Duplex filters w/DP indicatorand high DP alarm.

Flow indicator w/high & low alarms

Temperature inicator w/high & low alarms (1)

Sealing pressurereference

DPX/DPI

NC

Differntial pressure controlvalve & transmitter w/

indication, low and highalarms.

Supplies to primary sealswith check valves and fixed

restriction. Options caninclude individual flow

indicators with restrictionadjustment (2).

Seal Gas Supply for Single or Tandem Seals Using Only Discharge Gas Supply

DPI

TI

FI

DPI

CompressorDishcarge

External Supply(controlled below compressor

discharge pressure)

Emerg. Isolation Valve (3)

Notes:(1) Needed for gases with potential to besaturated. Coalescer location may bedownstream of major pressure drops,depending on flash properties of gas.Lines downstream of coalescer shouldbe sloped and traced. A heater may berequired to assure sufficient margin fromcondensation.(2) Restriction adjustment arrangementsmust be designed to always assure aminimum flow.(3) Needed only for flamable gases.Bypass for testing is optional.

NC

NC


Flow indicator w/high & low alarms

Temperature inicator w/high & low alarms (1)

Sealing pressurereference

DPX/DPI

NC

Differntial pressure controlvalve & transmitter w/

indication, low and highalarms.

Supplies to primary sealswith check valves and fixed

restriction. Options caninclude individual flow

indicators with restrictionadjustment (2).

PI

Pressure indicator w/ low& high alarm

Seal Gas Supply for Single or Tandem Seals Using External Gas Supply as Primary Seal Gas or as Idle Speed Seal Gas

PressureAmplifying System

for Idle SpeedSeal Gas

Fixedopeningpressureon bothcheckvalves

Continuous blowdownto compressor suction

Coalescing device w/DP indicatorand high DP alarm (1). Level

indicator is optional. Bypass can bereplaced by second coalescing

device.


Compressor M

anual

Novem

ber 2001200-66

Chevron Corporation

Fig. 200-58 Seal Gas and Separation Gas Supply Schematics

Seal Gas (Barrier Gas) Supply for Double Seals Using Nitrogen orBuffer Gas Supply for Tandem Seals with Intermediate Labyrinths

Notes:(1) Needed for double seal installationsin critical service or highly toxic services,and tandem seal installations with highlytoxic seal gas. May also be used forother services if plant nitrogen is notreliable.(2) Sepate DP regulators (located onindividual suppy lines) are required foreach seal when controlling buffer gaspressure differential relative to primaryseal vent pressure.

DPI

PrimaryNitrogen

Supply


NC

Pressure regulator.Alternate is DP control

relative to seal referencepressure (double seals) orprimary seal vent pressure

(tandem seals)(2).Valves for on-line

maintenance are optional.

Supplies to each seal withcheck valves and flow

indicators w/ high & lowalarms.

PIPI PI

PI

Pressure indicator w/ highalarm, low alarm. Low, lowshutdown for double seals.

FI F I

To Purge GasSupply Circuit

for double seals(if applicable)

To Separation GasSupply Circuit

Nitrogen Bottles

Pressure indicatorw/ high & low alarms

Back-up System OnlyWhen Needed (1)

DPI

Gas Supply


NC

Pressure regulator.Alternate is DP control relative to seal

reference pressure (for purge gas)or vent system pressure (for separation gas).Valves for on-line maintenance are optional.

Supplies to each seal withcheck valves and flow

indicators w/ high & lowalarms.

PIPressure indicator w/high alarm, low alarm.

FI FI

Check valve needed onlywith back-up system

Not required if supplied from buffer gas orbarrier gas system, as shown above.

Separation Gas Supply for All Seals orPurge Gas Supply for Double Seals

Compressor M


pressors


Novem

ber 2001

Fig. 200-59 Seal Venting System Schematics

FI

Flow indicator w/high alarm

PI

Pressure indicator w/ highalarm. High, high shutdown

for toxic tandem sealservices. Shutdown can be

replaced with high, highalarm for single seal or other

tandem seal services.

To Flare or othersafe location

Seal Vent Port

To Flare

Optional gassampling point

Rupture diskHigh Leakage

Relief Circuit (1)

(2)

Venting and Leakage Measurement System for Tandem Primary Seal (Each Seal)

FI

Flow indicator w/high alarm fortoxic services.

(3), (4)

Seal Vent Port

To Flare or other safelocation

Optional gassampling point fortandem seals w/

intermediatelabyrinth.

(4)

(2), (5)

Venting System for Single Seal, Double Seal or Tandem Secondary Seal (Each Seal)

Low point drainw/ sight glass

(3)

Notes:(1) Needed only for high pressures or whenflare header pressure limit can be exceeded.(2) Check valve must open at low enoughpressure to prevent reverse pressurization ofgas seal.(3) Only for installation with separation seal.(4) Can also be used when air is used asseparation gas in combustible services.(5) Check valve not required for single sealsvented to atmosphere.



243 ConfigurationsConfiguration refers to the relationship between the inlet, discharge, and side streams to the mechanical arrangement of the compressor. This will be clarified by the following examples.

Figure 200-60 shows a typical cross-section of a multistage centrifugal compressor. This is called a “straight-through” compressor because flow goes in one end and out the other.

Another common configuration is the “compound,” or “Out-and-In” type (Figure 200-61). This arrangement allows removal of the total gas stream for inter-cooling, power savings, or processing, and re-entry for additional compression. Note the additional spacing required for flow extraction and re-entry. Although some designs can minimize the effect, this reduces the maximum number of impel-lers available for compression.

The “sidestream compressor” shown in Figure 200-62 allows the introduction or extraction of partial flows at intermediate levels to satisfy various process require-ments. The number of sidestreams in a single casing is limited only by available spacing. This arrangement adds the complexity of requiring mixed temperature calculations to determine impeller performance downstream of sidestream inlets.

The “double-flow” configuration effectively doubles the capacity of a given frame size (Figure200-63). The compressor is divided into two sections, the inlet flow entering at either end, and discharging through a common discharge nozzle at the

Fig. 200-60 “Straight-Through” Centrifugal Compressor (Courtesy of the Elliot Company)



center of the casing. The impellers in each section face in opposite directions, achieving thrust balance at all operating conditions. While flow is doubled, the number of stages available for increasing head is cut in half. The use of the double-flow option should be carefully evaluated against other alternatives.

The compressor in Figure200-64 utilizes what is commonly called the “back-to-back” impeller arrangement. This type has advantages in high pressure-rise applica-tions where thrust balancing becomes difficult using a conventional thrust bearing and balancing drum. Since the back-to-back impellers produce opposing thrust forces, the net thrust is significantly reduced, eliminating the need for a balance piston to provide thrust compensation. This arrangement must, however, be care-fully reviewed with respect to division wall-flow disturbances, bearing span, and seal design on rotor stability.

One other configuration to note is a combination series/parallel unit, Figure 200-65. Eastern Region has one of these in booster-compression service, and reports good performance, and flexibility switching back and forth in order to obtain higher flows, or discharge pressure, as needed for system operation.

Fig. 200-61 Compound Centrifugal Compressor (Courtesy of Dresser-Rand)



Fig. 200-62 Centrifugal Compressor with Side-stream Connections (Courtesy of Dresser-Rand)

Fig. 200-63 Double Flow Compressor (Courtesy of Dresser-Rand)



Fig. 200-64 Back-to-Back Impeller Arrangement (Courtesy of Dresser-Rand)

Fig. 200-65 Series/Parallel Compressor



250 Application and System Considerations

251 Effect of System Changes on PerformanceA centrifugal compressor operates at the intersection of its performance curve and the system resistance curve. For constant inlet conditions, the operating point of a variable-speed unit can be changed by either a change in speed or by altering the system curve. Constant-speed unit performance can only be modified by changing the system curve.

ExampleIn Figure 200-66 a typical system resistance curve has been added to performance curves indicating the effect of a change in inlet pressure. The solid curve shows original performance while the lower curve shows the effects of a reduced inlet pressure. Calculations using fan laws (assuming a constant inlet volume flow) would indicate revised operation at point C. However, since the compressor would actually seek a new operating point at the intersection of its revised performance curve and the system curve, the resulting operation would be at point B.

Fig. 200-66 Effect on Performance Due to Change in Pressure (From Compressors: Selec-tion & Sizing, by Royce Brown 1986 by Gulf Publishing Company, Houston, TX. Used with permission. All rights reserved.)



If the effects of the system curve are large, estimates made using the fan laws will be significantly in error.

252 Stable Operating Speed RangesThe compressor stability range is discussed in connection with performance curves and surge in Section220. This is very important.

In addition to performance stability, a satisfactory margin must be maintained between the operating speed range and the critical speeds of both the compressor and driver.

Although API 617 defines these required margins, the following can be used as a general guideline:

• lateral critical—should not fall in the range from 15% below any operating speed to 20% above the maximum continuous speed.

• torsional criticals—(complete train) no torsional critical should fall in the range from 10% below any operating speed to 10% above maximum contin-uous speed.

253 Power MarginsThe rated horsepower for centrifugal-compressor drivers should be a minimum of 110% of the maximum horsepower required for any specified operating point.

For motor drivers, it is necessary that the motor be carefully matched to the compressor, and items reviewed such as:

• motor speed-torque characteristics,• accelerating-torque requirements of the compressor, and• motor supply voltage during acceleration.

(See the Motor section of the Driver Manual.) Steam turbines should have a maximum continuous speed 105% of rated compressor speed.

Driver requirements are further detailed in API 617. API Standards 611 and 612 cover general purpose and special purpose steam turbines.

254 Series OperationWhen two or more casings (or sections) are operated in series, the manufacturer usually furnishes two performance maps: one for each casing, and one showing overall casing performance. For determination of the surge volume, use the overall curve.

In most situations, it is desirable to have an individual anti-surge recycle line around each casing (or around each section of compression of compound casings). It is not practical for one anti-surge control to accommodate two casings or sections at oper-ating conditions significantly removed from the rated point. In addition, the overall



operating stability range can be improved because the anti-surge controls can be set for the stability range of each casing rather than the overall range for all casings.

255 Weather ProtectionAlthough centrifugal compressors are generally suitable for unprotected outdoor installations, daily temperature fluctuations can affect equipment alignment. Cold temperatures, heavy rains, salt atmosphere, blowing dirt or sand can make mainte-nance difficult, and maintenance of equipment cleanliness impossible.

Most equipment specification packages include detailed requirements for weather protection of controls and instrumentation. However, conditions vary between loca-tions. Therefore, get specific input from site personnel. Also, make sure the specifi-cations accurately reflect what the field has found to be most trouble-free.

256 Process Piping ArrangementsThe inlet piping configuration is an important factor that must be carefully evalu-ated to ensure satisfactory compressor performance. Performance predictions are based on a smooth, undisturbed flow pattern into the eye of the first impeller. If the flow has any rotation or distortion as it enters the compressor, performance will be reduced.

Figure 200-67 may be used as a guideline to establish the minimum length of straight pipe run ahead of the compressor inlet.

The nozzle loads, or forces and moments that the compressor can accommodate without misalignment are generally specified by the manufacturer.

API 617 specifies an arbitrary 1.85 times the limits defined by the NEMA SM-23 Standard. This results in limits which are not practical for all machine types. This criteria relates allowable loadings only to flange size. For example, a lightly constructed unit with 8-inch, 150-pound flanges would be expected to withstand the same loadings as a heavy barrel casing with 8-inch, 2500-pound flanges.

Specification, CMP-MS-1876, Centrifugal Compressors, specifies allowable load-ings related to the weight of the machine. This approach provides limits which are generally accepted within the industry.

The design and location of piping supports, and the accommodation of thermal expansion, is generally left to the piping designer, although it should also be reviewed by the project or machinery engineer. This should be checked in detail during construction to ensure correct installation of piping, and that the location and setting of supports is in accordance with design drawings and specifications. Section700 contains installation and precommissioning checklists which include piping installation review.



Fig. 200-67 Minimum Straight Pipe Run Ahead of Compressor Inlet (1 of 2)Note: Use the chart to determine Dimension “A”. (Courtesy of the Elliot Company)



Fig. 200-67 Minimum Straight Pipe Run Ahead of Compressor Inlet (2 of 2)Note: Use the chart to determine Dimension “A”. (Courtesy of the Elliot Company)



The following additional items should be considered when reviewing the overall compressor piping design.

1. High-velocity streams generate noise. Maximum velocity can be limited by the amount of noise that is allowed.

2. No side connections (such as the balance piston return line) should be put in the straight piping run ahead of the compressor inlet.

3. When a permanent strainer is used, specified compressor inlet pressure must include an allowance for strainer pressure drop.

4. To avoid problems prior to startup, the compressor manufacturer should be advised of the description and location of each strainer.

5. Woven wire mesh should not be used in strainers for centrifugal compressors. Wire mesh has the tendency to plug very rapidly, requiring frequent removal, and in some cases, it has been ingested into the compressor causing serious internal damage.

6. Inlet strainers should be located in the first pair of flanges away from the compressor's nozzle. Strainers should not be located right at the suction nozzle, since excessive flow distortion could result.

257 Lube- And Seal-Oil SystemsThe lubrication of centrifugal compressors is generally handled by a pressurized system, which also provides the seal oil and control oil in some cases. One system usually supplies all machines in a given train (such as the compressor, any gears, and the driver).

A basic pressurized lube system consists of a reservoir, pumps, coolers, filters, control valves, relief valves, instrumentation, and other auxiliaries specific to the application.

Seal oil may be provided from a combined lube and seal oil system, or from a sepa-rate seal oil system. Generally, combined systems are selected for sweet gas services. Separate seal oil systems are generally selected for compressors in services that contain hydrogen sulfide or other corrosive or toxic gases. In either type of system, the inner (sometimes called ‘sour’) seal oil leakage is normally not returned to the reservoir. The outer (sometimes called ‘sweet’) seal oil leakage is returned to the reservoir. Under certain conditions, it is possible for sour gas to migrate into the outer seal oil stream that is returned to the reservoir. Having a separate system posi-tively avoids contamination of the lubricating oil and subsequent corrosive attack of babbitt-lined bearings and other components served by the lubricating oil system.

API 614, Lubrication, Shaft-Sealing, and Control Oil Systems for Special Purpose Applications , and Specification CMP-MS-4762 cover the design, manufacture, and testing of the overall system, as well as individual components. Used as a reference, they provide guidelines based on user experience which can easily be scaled down or tailored to fit any requirement.



The system may be designed either as a console or baseplate-mounted package, with all components mounted on a single baseplate, or alternately as a multiple-package arrangement, with system components separated into individually packaged units. In this case the individual component packages are piped together in the field.

Oil return lines must slope toward the reservoir(s) to allow gravity draining. This is often overlooked when piping is being laid out. Also, be careful to avoid “head knockers” when laying out pipe.

Off-shore applications may require a system mounted integrally with the compressor/driver baseplate, with off-mounted air coolers.

The console arrangement, because of its compact layout, may limit or restrict access to various components making maintenance difficult. The multiple-package arrangement allows greater flexibility in locating the individual packages for improved maintenance access. A major disadvantage of the multiple-package arrangement is that the complete system is seldom shop tested and therefore perfor-mance is not verified prior to arrival on site.

Careful attention at all phases from initial specification through installation and startup will contribute significantly to trouble-free compressor train startup and operation. Historical maintenance data from many compressor installations indicate approximately 20 to 25% of centrifugal compressor unscheduled downtime results from instrument problems (many of these associated with operation and control of the lube and seal system).

When designing or modifying a system, obtain specific input from the field regarding site requirements, preferences, and operating experience. They may have already modified the basic system to correct problems experienced, found a partic-ular type or brand of instrument that functions better under their site conditions, or standardized on components to reduce spare parts inventories, etc.

The following highlights areas requiring special attention:

1. For critical or non-spared equipment, include a main and an identical full-sized auxiliary oil pump (not to be confused with an emergency oil pump which is normally of much smaller capacity, sized only to handle lube and seal require-ments during coast-down). A popular drive arrangement for turbine-driven compressors is a steam-turbine driven main oil pump with an electric motor driven auxiliary. This arrangement has the advantage that auto-start control of the electric motor driven unit is relatively simple and reliable with rapid accel-eration to full speed and rated pressure output. For installations where steam is not available, several alternate drive combinations are used, including motor, shaft-driven, and in a few cases air or gas expanders. With motor driven main and auxiliary pumps, each should be supplied by an independent power source.

2. Consider adequate oil-flow to bearings and seals during coast-down following a trip of the auxiliary pump. The two approaches used most often involve either an emergency oil pump or overhead rundown tanks.

Overhead rundown tanks are typically located to provide an initial pressure (head) equal to the low oil pressure trip pressure. API requires capacity to be



sufficient to supply oil for a minimum of three minutes. In the majority of cases this is adequate.

A second method is an emergency oil pump. This pump would probably be DC motor driven, with power supplied by a battery backed UPS system.

3. Manufacturers often insist that the response time of a motor driven auxiliary pump is sufficient to avoid pressure decay tripping the main unit, and therefore accumulators are not required. However, several tests have shown this not to be the case. The option should always be held open so that accumulator require-ments are based on the system demonstrating acceptable stability during the prescribed testing.

4. The system rundown tanks, and the accumulators are sometimes confused. The rundown tanks provide lubrication and cooling to bearings and seals during coast-down. The accumulator is designed to maintain system pressure within specified limits during transient conditions or upsets, thus avoiding machinery trips.

5. When oil seals are used, the manufacturer is normally asked to guarantee a maximum value for this inner seal-oil leakage. The guaranteed value is often found to be considerably lower than actual leakage on test or following startup. Since size of the degassing tank is based on this leakage rate, the tank often ends up being undersized.

API specifies that the degassing tank be sized for a minimum of three times the guaranteed inner seal oil leakage. Actual leakage, however, has in some instances exceeded quoted values by more than 10 times. The manufacturer's sizing criteria should be verified based on review of leakage-rate tests for similar seals.

6. For centrifugal lube-oil pumps, the pump head should be compared to the maximum allowable filter pressure drop (of dirty filters) to ensure that suffi-cient oil flow is provided to the machinery as the filters become dirty.

7. Shaft-driven main lube-oil pumps are not recommended, since any mainte-nance or repair of this pump requires the machine be shutdown.

260 Instrumentation and Control

261 Typical InstrumentationTypical instrumentation is shown in Figure200-68.

API 614 and 617 data sheets include several additional instrumentation options. These data sheets provide a good checklist for defining the requirements of a specific application.

Whatever alarms and shutdowns are chosen, it is very important to make sure they are installed with facilities to allow testing.



262 Compressor ControlThe control system must regulate compressor output to satisfy the process require-ments and must also keep the compressor from operating in surge.

Performance requirements are usually established during the process-design phase, based on a cooperative effort between the process designer and machinery engi-neer. Although control parameters for an existing process may already be set, (making selection of the compressor control system relatively straight forward), a process update or modification, a change in type of compressor or driver, or a need for improved efficiency, may dictate a change. Refer to the Instrumentation and Controls Manual for coverage of control system design.

Fig. 200-68 Typical Centrifugal Compressor Instrumentation

Indicator Alarm Shutdown

Lube and Seal System

Lube oil pump discharge pressure x

Oil header pressure (each level) x

Low lube-oil header pressure x x

Standby oil pump running x

Seal-oil pump(s) discharge pressure x x

Seal-oil differential pressure x

Standby seal-oil pump running x

Low seal-oil level x

Low seal-oil pressure x x

Run-down tank level x x

Compressor

Compressor flow rate x

Compressor suction pressure low and high (each section) x

Compressor discharge pressure low and high (each section) x

High compressor discharge temperature x WS(1)

Journal bearing temperature WS(1) WS(1) WS(1)

Thrust bearing temperature WS(1) WS(1)

High liquid K.O. levels x x x

Surge event x

Shaft Vibration x x x

Axial Position x x x

(1) WS = when specified



An understanding of the effect of varying gas conditions on compressor perfor-mance is necessary to properly evaluate control alternatives. Figure200-69 shows the performance curve for a centrifugal compressor operating at constant speed with varying inlet conditions.

263 Control System SelectionVariable-speed and constant-speed suction throttling are the two most common control methods. Adjustable inlet guide vanes are sometimes used, primarily on single-stage units.

Turbine driven compressors typically use variable speed, with either pressure or flow as the controlled variable. Suction throttling is generally used for motor-driven compressors. Variable-speed motors and hydraulic or electric variable-speed couplings are seldom applied to centrifugal compressors due to their added cost, and because they significantly lower the efficiency of the unit.

A review of centrifugal compressor characteristics highlights the differences between these two methods:

For variable-speed control, the capacity varies directly with speed and the head varies proportional to the square of speed. Therefore, as speed is reduced, capacity and head are reduced to meet the process requirements, with a corresponding reduc-tion of horsepower and a minimum loss in efficiency.

On the other hand, constant-speed operation essentially produces a constant head. Throttling reduces the inlet and outlet pressures but adds losses by introducing added resistance to the system.

Figure 200-70 shows typical constant-speed performance curves indicating the effect of suction throttling. Figure200-71 shows typical variable-speed perfor-mance curves. A comparison gives an indication of the difference in power require-ments between the two methods.

For a capacity requirement of 80%, suction throttling requires approximately 86% horsepower. For the same 80% capacity, control by variable speed requires approxi-mately 81% horsepower.

Parallel OperationParallel operation of two or more compressors adds additional complexity to the control system evaluation.

Slight variations in compressor performance characteristics, piping configuration, and instrument settings can cause one unit to take all the load, thus forcing the others into recycle, or alternately causing endless “hunting” between units.

For example, if one unit starts to recycle slightly ahead of the other and suction temperature is increased due to the recycle, its capability to produce head will be reduced, thereby locking this unit into recycle. Alternately, if suction temperature is reduced by recycle, head output is increased forcing the other unit into recycle, starting a back-and-forth swing between units.



Simulation studies are often necessary because of the complexity involved in matching parallel compressors. Direct your efforts toward developing the least complex control logic that will meet process and operating requirements. One common approach is to base load one unit, allowing the second unit to take process swings.

Fig. 200-69 Effects of Changing Gas Conditions at Constant Speed (Courtesy of the Elliott Company)

Fig. 200-70 Constant Speed Performance Curves (Courtesy of the Elliott Company)



264 Surge ControlIn the case of air compressors, surge control is often accomplished by a discharge blow-off valve, regulated to maintain the required minimum flow to the compressor. This is based on a minimum flow setting and is applicable only for units operating at constant inlet conditions. In most applications, however, it is necessary to recycle flow back to the suction, through a bypass cooler, in order to maintain stable opera-tion. Consult a company specialist for assistance in selecting an appropriate control system.

265 Machinery MonitoringMachinery monitoring systems are covered in detail in the General Machinery Manual . In summary:

• Monitoring systems are used to confirm that machinery is operating within specified design limits, to provide an indication of machinery condition, and to warn of changing conditions which might result in machinery damage or failure.

• Machinery monitoring varies from periodic manual recording of data, to auto-mated continuous computer data logging and performance analysis.

Fig. 200-71 Variable Speed Performance Curves (Courtesy of the Elliott Company)



The most common systems are those described in the General Machinery Manual. Virtually all new centrifugal compressors come with some monitoring system.

270 Rerates and RetrofitsIt is often desirable to modify process conditions to improve overall plant effi-ciency or to increase production. However, this often requires rerating an existing compressor.

Before spending a considerable amount of time and effort in redesigning the process, it is advisable to make a preliminary feasibility estimate to determine the rerate capabilities of the existing compressor. This will identify various limitations and help avoid completing a total process redesign only to find out that a compressor cannot meet these new requirements.

The major areas which require evaluation include capacity, pressure, speed, and power . Consider consulting the OEM, and/or a Company specialist before making significant changes to any critical (unspared) centrifugal compressor.

271 CapacityWhile impellers and internal stationary components can be relocated and new ones added, the casing nozzle sizes are fixed. The maximum capacity that can be handled with a reasonable pressure drop is therefore dependent on the nozzle size and related to inlet gas velocity.

Inlet velocity is dependent on gas conditions, allowable noise levels, and inlet piping configurations. An acceptable rule-of-thumb is a maximum of 140 ft/sec for air or lighter gases and approximately 100 ft/sec for heavier hydrocarbons.

The actual inlet gas velocity can be calculated from:

(Eq. 200-12)

where:Q = ACFM in ft3/minute at inlet pressure, temperature, Z, MW

D = inside diameter of the nozzle, in inches

If side load or compound inlets are involved, inlet gas velocity should be checked for all inlet connections.

272 PressureNext, check the pressure rating of the existing unit:

During manufacture, the casing was hydrotested to 1½ times the maximum oper-ating pressure (nameplate rating). If the pressures involved in the rerate exceed the

V 3.06 Q

D2-------=



nameplate rating, it will be necessary to re-hydrotest the casing for the new conditions.

Note the following items:

• It may be necessary to check with the manufacturer to confirm that the casing design pressure is adequate for rerating and rehydrotesting.

• Compressor operating characteristics, relief valve settings, or settle out pres-sures may set the maximum operating pressure.

• If set by compressor characteristics, use pressure rise to surge at maximum continuous speed.

• Side stream or compound compressors may have been hydrotested by sections with a different pressure for each. Check each section for compatibility with new conditions.

Check the compressor to determine its capability of producing the head required.

Use Equation200-3 to calculate the head for the rerated condition based on the desired pressure ratio. An attempt may be made to re-use some or all of the existing impellers, based on an overall polytropic efficiency of 70% for the initial estimate.

Initially estimate the speed from the affinity law (see later discussion regarding speed limitations):

(Eq. 200-13)

where:N1 = original speed

N2 = rerated speed

Hp1 = head for rerated pressure

Hp2 = head for original pressure

This same procedure will work for applications involving side loads or intercooling between sections. The head for each section is determined based on the conditions for that section, and the total head is the sum of the individual section heads.

273 PowerSince motor drivers are seldom oversized, anything more than a minor power increase may require a new motor. This requires close evaluation of proposed process changes to see if necessary improvements can be achieved while still staying within the driver's capabilities.

N2 N1Hp2

Hp1

--------

12---

=



In contrast, turbines and gears can usually be modified to provide increased power. Although turbine data sheets will sometimes provide information regarding maximum steam flow or uprate capabilities, discussions with the manufacturer may be required.

From Equation200-11, you can see that gas horsepower (GHP) is directly propor-tional to weight flow (w) and head (H), or:

(Eq. 200-14)

For example, if weight flow is increased by 10% and head is increased by 10%, the power requirement is increased by:

1.10 × 1.10 = 1.21 or 21%

Furthermore, a driver power margin of 10% is recommended. Therefore, the total recommended requirement is increased by:

1.21 + 10% (1.21) = 1.33 or 33%

274 SpeedFinally, review the speed based on impeller stress and compressor critical speeds.

Impeller stress levels are related to the impeller tip-speed as discussed in Section240. While the maximum allowable tip speeds vary with manufacturer, impeller design, and material, a good rule-of-thumb for impellers with backward leaning blades is 900 ft/sec maximum tip velocity.

Determine impeller tip speed by:

(Eq. 200-15)

or, using the 900 ft/sec., maximum speed is:

(Eq. 200-16)

Maintain the following critical speed separation margins:

• Any critical speed at least 20% below any operating speed• Any critical speed at least 20% above maximum continuous speed

Revamping of the rotor may have some effect on critical speeds; however, ignore this effect for the initial feasibility estimate.

GHP2 GHP 1w2Hp2w1Hp1-----------------=

uDN

229---------=

N max299 900( )

D-----------------------=



280 FoundationsThis sub-section provides a basis for establishing the dynamic forces to be used by civil engineers in foundation design calculations. Soil mechanics, natural frequency calculations, bearing pressure, concrete strength, and other design factors are not covered here. Refer to the Civil and Structural Manual for such information. Foun-dations, anchor bolts, and grouting are discussed in the General Machinery Manual.

In addition to knowing the dimensions and weights of the machinery to be supported, engineers designing the foundation must know the magnitude, direction, and frequency of the dynamic forces that the machinery will exert on the foundation.

The importance of foundations to a compressor installation cannot be overem-phasized. Foundations attenuate vibratory forces generated by the machinery, and reduce transmission of these forces to the surrounding plant and equipment. Foun-dations also keep the machinery in alignment.

To perform these essential functions throughout the life of the installation, the foun-dation must be sized to support the weight of the machinery while imposing a toler-able bearing pressure on the soil or structure. It must be properly designed so that the system, consisting of the foundation, soil, machinery, and piping, is not at or near a resonant condition. It is particularly important on offshore structures, which may be susceptible to resonance from the machinery vibration.

The purchaser of the machinery is normally responsible for the design of the foun-dation. The vendor or manufacturer of the machinery will seldom take this responsibility because his expertise is not in this field. It would not be in his best interest to accept the risks associated with the design. Additionally, the vendor does not have specific knowledge about the soil conditions at the site.

281 Foundation MountingCentrifugal compressors are installed on either soleplates or fabricated steel base-plates. The baseplates may be of the non-self-supporting or self-supporting type, depending on site requirements. These intermediate supports provide a permanent mounting point for the machine feet, which can then be shimmied for final location and alignment. In many cases, the baseplate is extended to support both the driver and driven equipment, and in cases such as off-shore installations, it can also contain the lube and seal system. The baseplate simplifies installation.

Section700 contains a detailed checklist including foundation mounting. This checklist may be used in conjunction with Specification MAC-MS-3907, Grouting of Machinery for Foundation Mounting . (See the General Machinery Manual for more information on foundations, anchor bolts, and grouting.) Section100 includes criteria for establishing forces to be used in foundation design for centrifugal compressors.



282 Design Basis for Rotating CompressorsDynamic (centrifugal, and axial) and rotary compressors generally exert much smaller dynamic forces than reciprocating compressors. Nevertheless, these forces should be accounted for to avoid a potentially serious vibration problem during operation of the compressor. A fault in the design of a concrete foundation is extremely difficult to correct after the concrete has been poured. There is no easy way to add mass, alter the stiffnesses, or adjust damping to change the natural frequency of a concrete foundation in an effort to move the system away from a condition of resonance. In a few extraordinary cases, it has been necessary to break out an existing foundation and pour a redesigned foundation to solve a serious vibration problem. Obviously, such instances are exceedingly expensive and time consuming.

While guidelines have been developed over the years for the allowable vibration of the foundation itself, criteria for defining the forces to be used in foundation design have been lacking.

A misunderstanding between the foundation designer and the compressor manufac-turer regarding the unbalanced forces to be allowed for in the design has contrib-uted to many foundation vibration problems. These problems have commonly been caused by not designing for the actual dynamic forces, but rather for some lower value, due to communication problems between the foundation designer and the machine manufacturer.

Depending on how the question about unbalanced force is asked, the manufacturer might respond with the rotor's residual unbalance from the dynamic balancing machine. This balancing-machine tolerance is an extremely small number which might be only 1/20th of the actual force at rated speed. At other times, arbitrary values are assumed for foundation design, yet they may not be representative of actual machine operation.

Dynamic ForcesThe dynamic force generated by the rotor(s) of rotary and dynamic compressors is related to the running speed and the vibration of the rotor. Because of the complexity of the subject, it is impossible to accurately predict the behavior of a rotor system with one or two simple equations.

Fortunately, however, standards have been developed for allowable limits of vibra-tion for new machinery. One of the most widely used standards is the API limit for dynamic and rotary machines:

(Eq. 200-17)

where:Av = Peak-to-peak amplitude (displacement) of vibration in mils (0.001

inches)

Av 2 or 12000N

---------------12---

whichever is less,,=



N = Rated speed in RPM

Note This equation is valid for speeds down to about 3000 RPM. Below 3000 RPM the limit is 2 mils.

The following equation may be used for calculating the force used in foundation design. This equation is based on a vibration three times the amplitude calculated from Equation200-17. A safety factor of three is recommended because that is about the maximum vibration level where you would ever allow a compressor to continue to operate.

(Eq. 200-18)

where:F = Dynamic force, lbs

N = RPM

WR = Weight of rotor, lbs

The force calculated is actually a rotating vector, and it should be assumed that it is acting perpendicularly at the center of the rotor. It should also be assumed that there will be a 50% reaction at each bearing from the unbalanced rotating force. The reac-tions at the machine's hold-down bolts can then be resolved.

Figure 200-72 shows the resolution of these forces to bearing reactions. The latter reactions are transmitted to the foundation via soleplates or baseplate and anchor bolts. Note that Equation200-18 can also be applied to the rotors of turbine drivers and gearboxes.

Occasionally the foundation designer may want to add a factor above the dynamic force determined by Equation200-18, although Equation200-18 is quite conserva-tive. Five times the API vibration limit has been used as a design criterion in some cases where there were special concerns about the design. This would provide a safety factor of 1.67 beyond Equation200-18. To make the calculation, substitute 7.1 for 4.3 in Equation200-18.

Other ConsiderationsThe question sometimes arises about whether the foundation would survive if a large chunk of metal, such as a piece of an impeller or turbine blade(s), were thrown off the rotor while running at full speed. A second question might be whether the foundation should be designed to accommodate such an occurrence. Foundations usually will survive such accidents, although some repairs to anchor bolts, hold-down bolts, or bearing pedestals may be necessary. Generally, such occurrences are not taken into account in the design. The forces involved are extremely high, and it is impossible to predict their magnitude. It is suggested that bolting and structures be checked for adequacy at 10 times rated torque. This value is often used on turbine-generator foundations, because a short circuit can cause an instantaneous

F 4.3 10 8– N2 WR A v×=



torque increase to that level. Similarly, a compressor rotor might cause such a torque increase in the event of a severe rub.

It is recommended that the natural frequency of the foundation system be at least 30% above or below the frequency of any compressor or driver operating speed.

As a rule of thumb, the weight of the foundation should be no less than three times the weight of the rotating machinery it supports.

290 MaterialsSelection of casing material is influenced by the service involved. Steel casings are required by API 617 for air or nonflammable gas at pressure over 400 psig or calcu-lated discharge temperature over 500°F (anywhere in the operating range), and for flammable or toxic gas. Stainless steel and high nickel alloys are generally used for low temperature refrigeration units. A materials guideline which covers recom-mended materials for compressor components is included as an Appendix of API 617.

Although manufacturers have a background of experience in applying materials and manufacturing processes to special applications, never assume the manufacturer completely understands your process.

Include a complete process gas analysis, with emphasis on corrosive agents, and water vapor, together with any anticipated variation in composition, off-design or alternate operating conditions, or possible process upsets. Specifications should encourage the manufacturer to offer alternatives or comment based on their experience.

When defining the operating environment, also consider the possibility of contami-nant build-up during compressor shutdown which might contribute to subsequent component failure. For example, the addition of water or cleaning chemicals during a unit shutdown may add one of the components that lead to a sulfide stress cracking failure (see Section291).

API imposes specific design limitations for corrosive gas applications. However, actual operating experience may dictate addition or modification to these requirements.

API also contains an appendix of material specifications for major compressor component parts.

The following discussion will help you recognize applications where the potential for problems may exist. Detailed descriptions of the failure mechanisms mentioned is beyond the scope of this manual. (See the Materials Manual.)

291 Sulfide Stress CrackingA prevalent problem is sulfide stress cracking of highly stressed components, espe-cially impellers. It requires the presence of hydrogen sulfide, water in the liquid state, an acid pH, and tensile stress.

Compressor M


pressors


Novem

ber 2001

Fig. 200-72 Unbalanced Forces from Compressor and Turbine Rotors



The use of inhibitors has been investigated, although in most cases the practical solution for operation in this environment has been a change of material.

Studies indicate that for materials with yield strengths between 100,000 to 110,000 psi, stress levels required for sulfide cracking are near the yield strength. In contrast, materials with yield strengths of 140,000 psi exhibited susceptibility at stresses as low as 30,000 psi.

Continuing studies have resulted in establishing the generally accepted API 617 guidelines, which limit material yield strength to 90,000 psi or less, and a hardness not exceeding Rockwell C22.

Note that in 1987, sulfide cracking caused the loss of a critical compressor supporting a major hydroprocessing facility, costing several million dollars. The cause was impeller stage pieces with too high a yield strength.

292 Stress Corrosion CrackingMaterials operating where the combination of tensile stress, a corrosive medium present, and a concentration of oxygen are susceptible to stress corrosion cracking. The effects of stress and corrosion combine to produce spontaneous metal failure.

Because all conditions required for stress corrosion cracking are less likely to exist in a normal environment, corrosion cracking is not as common. Also, materials modified for sulfide cracking produce a material less susceptible to stress corrosion.

293 Hydrogen EmbrittlementCompressors handling hydrogen (hydrogen at partial pressures greater than 100 psig, or concentrations greater than 90 molar-percent at any pressure) are susceptible to hydrogen embrittlement. This embrittlement occurs when a metal is stressed in a hydrogen-rich atmosphere.

Metals highly prone to embrittlement include high-strength steels and high-strength nickel base alloys. Those having only a slight tendency include titanium, copper, austenitic stainless steels and aluminum alloys, with most materials commonly used on centrifugals falling in between.

As in the previous cases, the most practical solution has been found in selection of material properties compatible with the process involved.

API 617 limits impellers to 120,000 psi yield strength and a hardness less than Rockwell C34. Figure200-73 shows that this stress level is for overspeed RPM, and is therefore conservative at running speed.



294 Low TemperatureStandard compressor casing materials are generally good for temperatures of -20 to -50°F. Below these temperatures, standard materials become brittle, and materials with improved low temperature properties must be used.

Nickel based steel alloys are generally used, with suitable alloys available for both fabricated and cast casings, for temperatures to approximately -150°F. Special nickel alloys and austenitic stainless steels may be used for temperatures to -320°F.

Also review other component materials for compatibility with the operating temper-ature range. The materials appendix of API 617 is an appropriate guide for material selection since temperature limits specified indicate limits commonly applied by compressor manufacturers.

An unusual example of the application of low temperature material requirements is an air compressor located in a cold climate region. Although this compressor might be located in an enclosed (even heated) building, it could be exposed to inlet air temperatures well below -50°F. Suction throttling would further reduce inlet temperatures.

Where reduced maximum yield strength and hardness are specified, apply the same requirements to any welding and repair procedures.

295 ImpellersCentrifugal compressor impellers are most commonly made from alloy steel forg-ings of AISI 4140 or 4340. Materials such as AISI 410 stainless steel and precipita-tion hardened stainless steels (including Armco 17-4 pH or 15-5 pH) may be used in situations where corrosion resistance is required. Austenitic stainless steels, monel, and aluminum, although somewhat limited in their application, are used in some

Fig. 200-73 Impeller Stresses at Various Speeds of Rotation (Courtesy of the Elliott Company)



special cases. Figure200-74 identifies the chemical analysis of various impeller materials. Figure200-75 provides a listing of mechanical properties.

296 Non-Metallic SealsElastomeric seal requirements in centrifugal compressors are generally handled by O-rings. Since compressor applications seldom involve pure gases or fluids, selec-tion of the proper O-ring material can become quite difficult. Carefully evaluate the operating environment, considering factors such as temperature, pressure, and fluid composition (with special emphasis on corrosiveness of the gas).

Operating experience in the same or similar service is of prime importance.

Figure 200-76 provides “application charts” for typical O-ring materials.

297 CoatingsCoatings are not widely used to improve corrosion or erosion resistance of compressor internals. Problems include:

• surface preparation prior to coating• maintenance of critical tolerances• balancing coated components• protection of coating during handling• modification of established manufacturing procedures

Fig. 200-74 Chemical Analysis of Impeller Materials (Courtesy of the Elliott Company)



Selection of compatible materials or material properties is generally the most prac-tical approach.

Fig. 200-75 Mechanical Properties of Impeller Materials (Courtesy of the Elliott Company)



Fig. 200-76 O-Ring Application Charts (Courtesy of the Elliott Company)

Centrifugal Compressors

Documents

Transcript of Centrifugal Compressors