n any process, numerous n any process, numerous compo-nents are required to make the nents are required to make the en-tire concept work. As with any tire concept work. As with any ve-hicle, processes require an engine hicle, processes require an engine and, where gases are involved, the motive force and, where gases are involved, the motive force is usually a compressor. This article will is usually a compressor. This article will at-tempt to shed some light on the different types tempt to shed some light on the different types of compressors, how they can be selected, and of compressors, how they can be selected, and the differences between air and gas units.
the differences between air and gas units.
The gas compressor holds a special place in The gas compressor holds a special place in most mechanical and chemical engineers’ most mechanical and chemical engineers’ minds, and thoughts can range from admiration minds, and thoughts can range from admiration to hatred. The reason for this is that to hatred. The reason for this is that compres-sors are usually the most complex mechanical sors are usually the most complex mechanical units in a process; as a result, they must be units in a process; as a result, they must be un-derstood and selected carefully
derstood and selected carefully. A. A poorly select-poorly select-ed unit will not only be unreliable in its own ed unit will not only be unreliable in its own right, but also it will be less ﬂexible to the right, but also it will be less ﬂexible to the in-evitable changes that occur when most evitable changes that occur when most process-es evolve from theory to practice.
es evolve from theory to practice.
Compressor engineers will tell you that the Compressor engineers will tell you that the majority of problems associated with the units majority of problems associated with the units in service can be traced to the process, the in service can be traced to the process, the se-lection of the compressor itself, or
lection of the compressor itself, or to poor pack-to poor pack-aging. For the uninitiated, packaging refers to aging. For the uninitiated, packaging refers to all of the components that allow a compressor all of the components that allow a compressor to operate. These include drivers, couplings, to operate. These include drivers, couplings, lu-brication and sealing systems, controls, and brication and sealing systems, controls, and ﬁl-tration. If a plant is operating as planned, and tration. If a plant is operating as planned, and the proper compressor is chosen, most issues the proper compressor is chosen, most issues
that do arise are that do arise are from items such as from items such as poorly sized poorly sized cool-ers, pump failures, ers, pump failures, and so on. and so on. The process The process engineer’s role engineer’s role The packaging The packaging decisions should decisions should be left to the be left to the rotat-ing-equipment, ing-equipment, me-chanical, or chanical, or mainte-nance engineers on nance engineers on any project. The any project. The pro-cess itself and the cess itself and the appli-cation of the right cation of the right com-pressor fall squarely on the pressor fall squarely on the shoulders of the process shoulders of the process en-gineer. If a
gineer. If a ﬂﬂow scheme is al-ow scheme is al-ready
ready ﬁﬁxed, and the conditionsxed, and the conditions required of a compressor are required of a compressor are unre-alistic, then there is very little that the alistic, then there is very little that the mechanical engineer can contribute after mechanical engineer can contribute after the fact. Thus, a rudimentary knowledge of the fact. Thus, a rudimentary knowledge of what existing, commercially available what existing, commercially available compres-sors can do is necessary for a good process sors can do is necessary for a good process de-sign. Keeping the numerical possibilities in sign. Keeping the numerical possibilities in mind while forming pressure/temperature/ mind while forming pressure/temperature/ ﬂﬂowow relationships can save a company a tremendous relationships can save a company a tremendous
Making the best choice means
Making the best choice means
understanding how flows and
understanding how flows and
pressures relate to available
pressures relate to available
machines, and seeing if your
machines, and seeing if your
process can be adjusted to
process can be adjusted to
meet the capabilities of units
meet the capabilities of units
that are readily available,
that are readily available,
reliable, and inexpensive.
reliable, and inexpensive.
Select the Right
Select the Right
CompressorD. Gregory Jandjel, D. Gregory Jandjel, Gardner Denver Gardner Denver
Engineered Packaging Center Engineered Packaging Center
I© ©Copyright 2000Copyright 2000 American Institute American Institute of Chemical Engineers. of Chemical Engineers. All rights reserved. All rights reserved. Copying and Copying and
downloading permitted downloading permitted with
amount of both its time and money. Compressor inlet pressures P1 are as important as the discharge pres-suresP2, since many commercial ma-chines are limited on the inlet side. For example, there are numerous rel-atively inexpensive, oil-injected screw (OIS) compressors suitable for process gas duties that are limited to values of P1 of 100 lb/in.2 (psi) (all
pressures here are gage,i.e., psig, un-less otherwise noted). Specialized OIS compressors can accept a P1 of 700 psi, but these are more costly and have longer deliveries.
Most oil-free screw (OFS) units designed for gas applications are lim-ited to 150 psi at the inlet. There are special cases when they can achieve a
P1 as high as 225 psi, but these must be checked on a job-to-job basis. Re-ciprocating (recips) and centrifugal types are available at nearly all inlet pressures, but the centrifugals can be-come costly under P1 and P2
condi-tions where special pressures or mul-tistaging is required. These will be discussed in the centrifugal section below.
The compressor industry is seg-mented into numerous sections and subsections, and recent economic conditions have created considerable crossover as compressor suppliers jockey for position on their next sale. Fundamentally, there are two major areas: gas and air.
Gas compression is the more diffi-cult, since the processes they drive are normally very expensive and gases must be handled delicately to avoid leakage, unwanted condensa-tion, other phase problems, and ﬂash points. Different gases also have varying market demands, which af-fect the way a compressor is built and purchased.
Industries with high demand and volume will have gas compressors
“built for duty” and numerous com-petitors, resulting in low prices. Re-frigeration is the best example. While the exact numbers are difficult to
ac-curately predict, the global number of standardized refrigeration units sold (for ammonia, propane, propylene, Freon-replacements, and Freon) fall well into the thousands annually. Other industries have fair demand
and volume, but standardizing the machines is difficult, since their
P1 / P2 / ﬂow relationships are widely varied and inconsistent. Hydrogen is a good example where there are many applications, but disparate ﬂows and
s Figure 1. Example of “curved” performance on centrifugal compressors.
Inlet Volume D i s c h a r g e P r e s s u r e , P 2 -12 30 45 60 75
Approximate I GV Positions, deg. Rated Point E x p e c t e d S u r g e Preswirl 0
s Figure 2. Cost-effective sizes for commercially available gas compressors.
Flow Rate, acfm
D i s c h a r g e P r e s s u r e , P 2 , p s i g 0 100 200 400 600 800 1,000 3,000 5,000 15,000 0 500 1,000 5,000 10,000 50,000 100,000 Diaphragm Rotary Vane
Oil-Injec ted Rotary Screw
Oil-Free Rotary Screw 15,000 psig/100 acfm 6,000/6,000 Centrifugal 5,000/4,000 (Start 200) 3,000/10,,000 2,000/30,000 1,500/100,000+ 870/10,000 (Start 150) 400/50,000 (Start 250) 150/4,000
pressures make it very difficult to de-sign one set of off-the-shelf equip-ment to cover the variations.
Some of the major gas compressor markets, and uses, include vapor re-covery (which can also be subdivid-ed into many sections), production, transmission, fuel gases (power gen-eration and boilers), refriggen-eration, and processing. The refrigeration, fuel gas, and gas transmission indus-tries are very large and consistent, hence, many standardized compres-sors have been created to reduce cost and gain market share. Even these in-dustries ﬁnd that conditions upon which gas compressors are sized vary to such an extent that it is diffi-cult to produce equipment to cover the entire market.
As a result of the different gas conditions found in the marketplace, broad ranges of customized compres-sors are available. These will typical-ly cost more than standard units, such as those built for air or the more com-mon refrigeration and wellhead gas service. The delivery times for these units are also longer, as most are built from the moment a purchase order is placed, unlike air units, which are built ahead of time based on market estimates. All API-based compres-sors, except for American Petroleum Institute (Washington, DC) API 11P recips, are custom builds, so that the user should expect deliveries in ex-cess of 35 weeks for a package. Cen-trifugal and OFS options, built to API standards, often have deliveries ap-proaching one year from order. (Since many API standards will be men-tioned here, their titles and content will not be mentioned. For more in-formation, see API’s Web site: www.api.org/.)
Although air is a gas, the incredi-ble volume and competition in the air market has bred an entirely different line of compressors. From a corrosion standpoint, air is actually a very diffi-cult gas. It contains two oxidants (O2 and CO2), and water is normally
pre-sent. However, air is drawn from the atmosphere, and if a compressor leaks, there is little harm. With a P1
variance of roughly 11.2–14.7 psia and well over 90% of applications using aP2 of 100–150 psi, it has been relatively easy for the world’s air compressor manufacturers’ to “ dial-in”their designs. They have complete lines of products that are built with considerable capacities, using compo-nents that are only acceptable to air or nitrogen. Companies such as Gard-ner Denver (GD) produce thousands of OIS compressor air-ends per year.
Air compressors are normally lightweight and use materials, such as bronze, which would not be accept-able for most gases. Thus, air units should always be treated separately
from those handling process gas. Only some reciprocating units (in-cluding some made by GD) have found success in crossing over. How-ever, despite using the same frame, even these units have differences in internal construction, depending on whether air or gas is used.
An aero-derivative screw com-pressor has made inroads in the gas production arena, speciﬁcally well-head gas. Once again, these units are constructed differently from the stan-dard air units; they are engineered such that they service wells that typi-cally have 5–7 year lives.
Process gas compressors should be selected from the moment the process design begins. Operating companies that are successful with their
com-Table 1. Minimum sizing information
required by compressor vendors.
Gas Compressors • Site elevation above sea level
• Gas inlet pressure
• Gas composition (or molecular weight (MW)), heat capacity at constant pressure Cp, k -value, compressibility Z (for budgetary considerations); also indicate the water content in the gas
• Gas ﬂow
• Gas discharge pressure required • Discharge temperature limitations • Drive selection:
For gas engines or t urbines, provide the fuel gas lower heating value (LHV), temperature, and pressure
For steam turbines, provide the steam temperature, inlet pressure, and acc eptable backpressure
• For air or water cooling, provide the design temperature of the medium • Level of speciﬁcation:
Manufacturer’s standard “ Near” API
API speciﬁcations with client inputs Client’s speciﬁcations
Oil level permitted in discharge gas
Air Compressors • Site elevation
• Ambient air temperature range (winter low and summer high) • Relative humidity at high temperature
• Discharge pressure required • Discharge temperature limitations • Oil level permitted in discharge air • Drive selection
• Air or water c ooling • Level of speciﬁcation
pressors normally try an iterative pro-cess. The engineer designs an initial process that ﬁts the range of what the market has available, then these con-ditions are sent to various compressor vendors and feedback is provided on issues of feasibility, availability, price, and delivery. The process engi-neer than incorporates the revised pressures, temperatures, and ﬂows into a workable process model.
Naturally, this procedure may in-volve mechanical and maintenance engineers, a corporate
rotating-equip-ment engineer, or an engineering con-sultant. In the end, process engineers should know a little about what com-pressors can do, and mechanical engi-neers should know a little about what a process’needs and limitations are. Compressor sizing
Important sizing information, re-quired by compressor vendors and packagers, is listed in Table 1. Com-pressors are sized on absolute inlet pressure (psia), inlet temperature,
inlet ﬂow, and gas characteristics. Positive displacement (PD) types, which consist of all models other than centrifugals, are sized on actual inlet volume (actual ft3 /min (acfm)).
For most reciprocating and rotary screw applications, the discharge pres-sure becomes almost secondary in the selection of the physical size of the unit. However, knowing the discharge pressure is necessary, since it will dic-tate whether the compressor is capable of the desired pressure, whether the gas will remain gaseous throughout
Table 2. How the four basic types of compressors stack up.
Type of Compressor OIS OFS Centrifugal Reciprocating Principle of operation PD-OIS PD-OFS Dynamic PD
Maximum P 1,psi 700 225 Closeto P 2 Close to P 2(Depends on rod load, as well) Maximum P 2,psi 865 400 1,500 –5,000 6,000
Maximum ﬂow, acfm 10,000 47,000 100,000+ 6,000 Maximum T 2,°F 250* 400 –500 350 –500 300 –400
Pulsation None None None Large
Surging No No Yes No
First critical speed Below Below Above Below Maximum pressure ratio/stage 23:1 5:1† 1.5 –3:1 5:1
Design point efficiency 70 –85% 70 –85% 70 –88% 75 –92% Off-design efficiency Excellent Fair Poor Fair Oil-free status Filters needed Oil-free Oil-free Filters‡
Polymer gas applications No Yes Difficult No API continuous run, h 16,000 24,000 24,000 8,000 5-yr reliability/uptime 98 –99.5% 99 –99.5% 97 –99.5% 90 –95%
Standby required§ No No No Yes
Installationarea X 2X 2X 4X
Noise level (with enclosure) 85 dB 85 dB# 85 dB 90 dB
Vibration level Small Small Small Large Sensitivetovibration No No Yes No
Capacity control 15 –100% Recycle 70 –100% Step or recycle Discharge accumulator No No Yes Yes
Discharge temperature control Yes Possible No No Variable inlet pressure Yes Yes No Limited Method of P 1variance Slide valve Recycle PCV¶ SVU** /PCV¶
Part-loadpower Low High High High
Gas composition effect Small Small Large Small –medium
Lowest molecular weight 2.0 2.0 10.0 2.0 (Special distance piece) Startingtorque Low Low High High
Installation costs Low Medium Medium High Spare parts cost Low Medium High High
Operationcost Low Low Low –medium Medium –high Maintenance required Low Low Low Medium –high
* OIS compressors are cooled by oil injection. Most lubricants break down at 280°F.
†OFS units can achieve 8:1 ratios with liquid injection. ‡Or nonlubricated with distance piece and purge.
§Standby requirement can also depend upon level of speciﬁcation. #OFS noise level w ith silenc ers.
¶PCV is pressure cont rol valve.
the compression cycle, the number of stages required, and what the power requirement for the unit will be.
Centrifugal compressors are dy-namic, which means that their perfor-mance is “curved” and depends tremendously upon inlet ﬂow and dis-charge pressure requirements. Figure 1 shows the typical pressure volume relationship for these devices. Note the P2 curves downward as the ﬂow increases; this is found in all dynamic compressors. The multiple curves are for different inlet guide vane (IGV) settings on a single-stage compressor. IGVs are pneumatically actuated vanes at the entrance of the compres-sor that alter the gas ﬂow and create a pressure drop. Unlike butterﬂy valves, they are quite efficient at 80–100% of compressor-design ﬂow rates. The amount of turndown ob-tained by using IGVs depends upon staging. Single-stage units will typi-cally be capable of 65–105% of ﬂow range with IGVs, while the effect is closer to 90–100% on multistage ma-chines. The 105% is achieved on
sin-gle-stage units by “preswirl,” which is possible by reversing the IGVs.
Centrifugal units can be seen as
ﬁxed-ratio compressors, which means that only a ±5% change in inlet pres-sure is normally allowed for a ﬁxed discharge pressure. Dynamic com-pression results from the conversion of gas velocity to pressure. Thus, molecular weight (MW) plays a key role, since the selection of the com-pressor is directly related to the head calculation for the process. This is why centrifugals are not used for hy-drogen, or other low-MW gases. The head is simply too high for a cost-ef-fective solution when the MW dips below 10. When the MW is below 5, a centrifugal selection is almost im-possible, since the head may be at or well over 100,000 ft.
The main difference that a user sees between PD and dynamic com-pression is that the former’s compres-sors do not provide a signiﬁcant in-crease in ﬂow with a drop in dis-charge pressure, whereas dynamic compressors do (this is called riding
out the curve). In fact, if a PD blower is sized properly for the pressure ratio required, the ﬂow might drop by a few percentage points if the ratio is increased or decreased by more than 10%, due to volumetric efficiency. Table 2 compares the four basic types of compressors.
Forestalling against surge By the same token, while PD units can produce 20% higher discharge pressures (or more) with only per-centage-point changes in ﬂow, dy-namic units can only rise in pressure marginally with a heavy payment in
ﬂow and a very great danger of ap-proaching surge. Surge is when a cen-trifugal compressor approaches the end of its curve at the left (Figure 1). It is best to keep the compressor op-eration at least 5% (by ﬂow) to the right of this line.
Physically, as the compressor ap-proaches surge, vibration begins and increases as you approach the surge line (a line drawn by connecting all the left-most points on the IGV curves). Vibration is quite severe and can heavily damage a compressor whose impeller(s) may be spinning anywhere from 7,000–50,000 rpm. For this reason, all centrifugals should be purchased with a surge system, which unloads the machine via recir-culation, as conditions approach surge. Reciprocating, screw, and other PD compressors will push the inlet gas against whatever system resis-tance exists at the discharge. This is why high-pressure shutdown and re-lief valves are so important. PDs will continue to push until the system is satisﬁed, or something else gives (system resistance, unloader setting, recycle setting, alarms/shutdowns, or the relief valve). Apart from volumet-ric efficiencies, which can vary from 70–95% depending upon the selec-tion vs. requirement relaselec-tionship, a PD unit will compress the same amount of actual ﬂow through its cylinder, regardless of pressure, tem-perature, or MW.
Naturally, one could get into
se-s Figure 3. Air compressors — cost-effective units that are commercially available.
Reciprocating 1,500 psig/300 acfm
Flow Rate, acfm
D i s c h a r g e P r e s s u r e , P 2 , p s i g 0 500 2,500 3,000 5,000 200 400 600 800 1,000 1,200 1,400 1,600 10,000 20,000 30,000 40,000 Centrifugal
Oil-Injected Rotary Screw 250/100-3,000
300 100 Dry Screw 200 400 1,500 600/400 300/2,000 200/150-2,000 250/200-3,000 150/3,000-40,000
mantics over the issue of volume con-sistency in PD equipment, but when you compare a 25% variance with the 100+% found in dynamic units, the argument holds true. Thus, a com-pressor system vendor must know the inlet absolute pressure and inlet tem-perature, along with the ﬂow to cal-culate the acfm and peg the proper compressor sizing.
Reciprocating unit ﬂow results from the multiplication of cylinder volume (based on stroke and diame-ter), number of cylinders, actual rpm, and volumetric efficiency. Screw unit
ﬂow is based on swept volume times rpm times volumetric efficiency. This illustrates the need for an acfm calcu-lation by the vendor and the fact that the voltage must be known for motor drives, since the calculation depends on rpm, and many PD compressors are direct driven.
The compressor vendor, due to skid pressure drops across each side of the compressor, should calculate the acfm. Vendors prefer to be given the process
ﬂow requirement in standard ft3 /min
(scfm), (normal) nm3 /h, lb/h, or kg/h,
to ensure accuracy in sizing. This is why the gas composition or character-istics are so important.
Effect of k-value
If the buyer’s gas composition is yet to be ﬁnalized, the buyer’s guess on MW, the ratio of heat capacity at constant pressure to that at constant volume Cp / Cv, and the compressibili-ty Z will still be much better than the vendor’s. The MW and Z are obvious-ly necessary for ﬂow calculations, particularly if the ﬂow is provided in the desired standardized or mass-ﬂow format. However, Cp / Cv, often known as thek -value, is crucial as well.
Thek -value is used in various cal-culations, including horsepower and mechanical volumetric requirements for PD compressors. On centrifugal types, the MW is critical, since the head calculation is so heavily affected by this number (the larger the MW, the lower the head), and the head di-rectly affects the power requirement.
On PD units, the k -value directly af-fects the power requirements, and the calculation of it, for bidding purposes. Heavy hydrocarbon gases, such as propane, butane, and propylene, have
k -values around 1.14 under most inlet pressure/temperature conditions. Air, nitrogen, and hydrogen are typically at 1.4. Helium can be higher than 1.6. If there were two processes, one with propane and one with air, using iden-tical numbers except for gas compo-sition, the power used by a PD com-pressor could be 5–10% higher for the air than the lower k -value propane. This can often mean a frame break in the compressor driver, and this has a signiﬁcant impact on the temperature rise across the compres-sor. A frame break is a jump from one size to another. If the jump is from smaller to larger, then it can be costly. Temperature rise affects oil cooling and gas cooling, so the entire system can be affected if cost estimates are based on a 1.15 k -factor, and the ﬁnal design balloons to 1.3, for example.
There is one ﬁnal note about the gases lighter than 10 MW, speciﬁ cal-ly hydrogen and helium. Although PD compressors, notably reciprocat-ing and screw units, are the best for these low-MW gases, these lighter gases are“slippery”and tend to result in lower volumetric efficiencies in the 70–85% range. “
Slip-pery” means that the gas molecules are small and their density is light, so it is difficult to “capture”
the gas and make it go where you want it to go. Thus, when compared to a typical natural gas (MW of 17 and k -value of 1.28), hydrogen and helium will require a roughly 10% larger ma-chine, due to lower volu-metric efficiency, and 5–10% more horsepow-er, due to the higher k -values.
For the purposes of this article, some rules of
thumb will be presented to help guide the process engineer through the availability and selection process. This becomes necessary since nearly every compressor manufacturer has equipment with different pressure and
ﬂow capabilities. However, the mar-ketplace will always dictate what is purchased and what is a fringe player, and any good engineer must have bottom-line sensitivity in this day and age. Figure 2 shows the general area where the various types of gas com-pressors are cost-effectively available in terms of pressure and ﬂow. Figure 3 does this for the air machines.
The main types of compressors used in the industry are reciprocating, rotary screw, and centrifugal; lesser selected are rotary-vane, liquid-ring, and diaphragm machines. Only the major ones will be discussed at length in this article.
Reciprocating compressors In the U.S., from the 1900s to the 1960s, the reciprocating compressor became the workhorse for all com-pression (Figure 4). These have been used in nearly every application, in al-most every conceivable way. Although their population is dwindling, and an-nual sales volumes have been in steady decline since the 1970s, these machines’ capability in handling large
s Figure 4. Bare Y-type reciprocating compressor.
pressures and small ﬂows, along with single-body/multistage availability, will keep them a necessary part of the landscape for the foreseeable future.
Another beneﬁt that reciprocating units provide is their ability to be used as nonlubricated. This means that no oil is injected into the cylin-der. If the distance pieces are long enough and purged, then there should be no introduction of oil vapors into the process either (reciprocating units produce a tremendous amount of oil vapors, which can slip around the rings, if the cylinder is not designed to stop them). The result is that non-lubricated recips are often a cost-ef-fective solution in providing a low-oil content (vapors), or oil-free gas. (Dis-tance pieces deﬁne the length be-tween where compression takes place (the piston area) and the location of the oil (the crankcase).)
These devices are also the most
ef-ﬁcient method of compressing gas at a speciﬁc condition (under a given
P1, T 1, and P2). They can be multi-staged easily, and if the unit is lubri-cated, then adiabatic efficiencies in the 80–92% range can be achieved. Figure 5 shows three machines staged together in parallel, used for boosting natural gas in a cogeneration facility. Figure 6 presents two recips used for natural gas peak –demand shaving.
Unfortunately, if the process has many off-design points or if the com-pressor is oversized, the efficiency gains at design are lost at part-load. This is why variable-speed drivers and unloading valves have become popular, and sometimes expensive, alternatives to the standard recycle system setup.
The one major negative, which maintenance people are well aware of, is that reciprocating compressors are maintenance-intensive. This means that their reliability percentage
(oper-ation percent-age over 5 yr) is among the lowest of the compressors available, and that the repairs themselves can be costly. Most reciprocating units fall in the 92–95% reliability/availability range, while screws and centrifugals can achieve 98–99.5% levels.
Another problem that users have encountered is the pulsation and un-balanced forces created by the recipro-cating, or piston motion. This requires special foundations and pulsation sup-pressors, while the other types do not.
Mainly, reciprocating compressors are most cost-effective when the pro-cess P2 is above 865 psi at the dis-charge and the ﬂow is less than 2,000 acfm. These devices are also good for pressures above 500 psi, if the ﬂow is less than 300 acfm. In general, recip-rocating units are the most competi-tive type at any pressure, if the ﬂow is less than 200 acfm.
The nonlubricated forms are very competitive across the range, as well. Recent innovations in OFS and ﬁ ltra-tion technology have resulted in the use of OFS and OIS compressors on many oil-free applications.
In processes that involve low MWs, typically below 10, with pres-sures above 350 psi, recips are still very popular. Above 870 psi, they are the only solution, due to the above-mentioned difficulties that centrifugal units have.
A variation of the reciprocating compressor is the diaphragm machine. This derivation uses reciprocating
mo-tion, but has no cylinder. The piston
rod actually moves a plate back and forth, creating compres-sion via the moving diaphragm. These movers can achieve the highest avail-able pressures in the marketplace, roughly 15,000 psi. However, the largest unit can barely handle 100 acfm. Small ﬂow is simply the nature of this unique form of machine.
Diaphragm units are used mostly on high-pressure gases in R&D or other small-ﬂow processes. They are sometimes used as theﬁnal booster in a chain of compressors to achieve a pressure above 5,000–6,000 psi.
Capacity control is normally han-dled through recirculation of gases, or blowing off of air. As the compressor only sees the full design ﬂow under these circumstances, part-load power is the same as for full load. Stepped-unloading is available, through the use of various suction and cylinder valving techniques. By disabling or enabling suction unloader valves in different cylinder combinations, vary-ingﬂows can be achieved.
Unloading normally come in three-step (0–50–100%) or ﬁve-step (0–25–50–75–100%) jumps, and the method is especially popular for the API 618 compression units. Unfortu-nately, ﬁeld personnel are forced to
s Figure 5.
Three identical, horizontal recips, packaged on a common skid, illustrate how large and customized a packaged unit can be.
live with the decision of using step un-loading, which is often unreliable. It is not unusual for operations personnel to disable or remove reciprocating un-loaders sometime during their operat-ing life, to reduce maintenance costs and increase running time.
Another solution for the capacity control issue is using a variable speed drive. Atﬁrst glance, this is the best so-lution for efficiency and power savings. However, 21st century budgets are tighter than ever and variable frequency drives (VFDs) for motors are expen-sive. Also, when using VFDs, it is im-portant that a transient torsional study be done to avoid rod load problems on the compressors at certain speeds. Mechanical loading
Reciprocating compressor frames are no longer rated by horsepower, since the piston rod load rating is a far more accurate predictor of the me-chanical strain on the unit. It is com-mon for corporate speciﬁcations to state that rod load shall not exceed 90% of the continuous operating de-sign for the frame. Rod load capabili-ties increase with the size of com-pressor, so actual numbers would not be relevant or helpful to this article. What is helpful is knowing that cer-tain combinations of pressure ratio
andﬂow will load a compressor more or less. By verifying the load of a speciﬁc application against the pub-lished maximum for the compressor chosen, the mechanical engineer can verify mechanical loading.
Typically, reciprocating compres-sor cylinders can handle pressure ra-tios of up to 5:1 in a single stage, and can generally be offered in up to six stages, depending upon the gas and rod loading. Both gas temperature and rod loading become issues, which as-sist the compressor engineer in decid-ing whether multistagdecid-ing is required. Note that API 618 does not allow any cylinder to have a discharge tempera-ture above 300°F. However, many wellhead gas applications have used lubricated cylinders to a maximum of 350°F per cylinder. In the air market, some vendors have approached 400°F per cylinder, on lubricated cylinders, to reduce staging and increase cost beneﬁts to the client.
The main problem with elevated temperatures is that ring-life and valve-life are reduced; in some cases, dramatically. Newer materials have allowed the recent elevations, but any lubricated cylinders operating over 325°F, or nonlubricated cylinders above 300°F, should be examined carefully.
It is not unusual for piston rings to wear and be replaced annually. The cylinder valves should last longer, somewhere between 1–2 yr. In both cases, the nonlubricated cylinder parts would have shorter lives than the lu-bricated ones in the same application. Reciprocating compressors come in three formats:
1. Low-speed (300–500 rpm), long-stroke (7 in. and more), per API 619.
2. Medium-speed (500–800 rpm), medium-stroke (5–7 in.), GD/Joy-type units.
3. High-speed (900–1,800 rpm), small-stroke (2–6.5 in.), per API 11P.
Reliability is not too different be-tween the types, but preferences and features will guide the buyer toward one or the other. The important thing to remember with recips is that aver-age piston speed is important, not the actual rpm. Piston speed is a function of rpm and stroke length. Normally, users will accept piston speeds in the 600–800 ft/s range. Nonlubricated process applications would normally be sized below 600 ft/s, per API 618.
The GD-type and API 11P units are the most cost-effective and have steadily gained market share over the traditional “slow rollers” that API 618 speciﬁes, but the traditional ma-chine is still popular in end/dollar processes. As high-end processes have been the last to see recent budget demands, and the dwindling API 618 market has caused ferocious competition between the numerous suppliers that remain, this may not change for some time.
The centrifugal compressor is ex-tremely popular, mostly because near-ly all are oil-free. Centrifugals also have oil-vapor and aerosol problems, but most process units can be consid-ered oil-free, due to the special seal arrangements that are used. Another tremendous advantage these machines have over all others is the enormous
ﬂows (100,000+ acfm) that some units can compress in a single body,
with a high-pressure (1,500–5,000 psi) capability.
Centrifugals are dominant in the large gas-pipeline transmission indus-try, where hundreds of thousands of cfm must be compressed to 1,000 psi after considerable line losses. They are typically placed every 100 miles of pipe, or closer, and are critical to the transmission of natural gas throughout the world.
In process use, they are used for large ﬂows, high pressures, and heavy-MW gases, and are popular on the critical process paths found in most petroleum reﬁneries and petro-chemical plants. These are locations where you ﬁnd the larger ﬂows and, with the extensive use of catalysts, the oil-free feature is important.
From a cost standpoint, centrifu-gals seem to do best when ﬂows are larger. There is not much choice with the other types above 50,000 acfm. As Figure 2 shows, there are also many combinations of high pressures with even larger ﬂows that simply make the centrifugal the only choice.
In the psi/acfm coverage areas where all the types meet, centrifugals are normally used in speciﬁc process-es that have ﬁxed pressure ratios and require oil-free gas, particularly with pressures above 400 psi. Normally, if a centrifugal is similar in cost to a re-ciprocating unit, the centrifugal should be chosen for its reliability in service and relatively large mainte-nance intervals. Still, while centrifu-gals are costly to ﬁx, particularly any-thing involving the rotor/impeller or seals, lifetime costs are normally below those of the recips.
The reliability advantage is a sig-niﬁcant point for both screw com-pressors and centrifugal units. Recip-rocating compressors must often be
“spared,” i.e., with a standby spare unit, due to their low reliability per-centage. Centrifugal packages built to API standards are intended to be used without spares, and have done so with reliability in the 99%+ range. These high-reliability units are nor-mally built to API 617, using API 613
gear boxes, API 541 motors, API 671 couplings, and API 614 lube/seal sys-tems. When steam turbines are used, API 611 (general purpose) or 612 (special purpose) turbines come into the equation.
API 617 centrifugal compressors typically operate in the 8,000–15,000 rpm range and use gear boxes driven by the main driver. There is another speciﬁcation, API 672, for integrally geared centrifugals. Machines made to this standard are intended for air service only, but some users have purchased them for clean gas applica-tions, such as pipeline-quality natural gas boosting. The main problem with the 672-type units is that they are bullgear driven by 2-pole motors and the impeller/pinion operates at around 30,000–50,000 rpm. The in-ternal gearing, high speeds, and limit-ed shaft seal designs make this type of centrifugal less reliable (98–99% range) than the traditional API 617 variety.
Bullgears are large gears where the shaft is driven by the driver. They have smaller gears, often, multiple gears, called pinions that can spin very quickly due to their size vs. that of the bullgear.
Centrifugal units are the latest type of compressor to be released by the manufacturers, in bare form, to pack-agers. The use of packagers has great-ly reduced the ﬁnal cost of the com-plete skid to the user and may bring about greater use of centrifugal units in more competitive industries such as refrigeration and fuel gas boosting. Capacity control is achieved by inlet throttling (valve or IGV), vari-able speed, or recirculation. On sin-gle-stage machinery, IGVs are the most cost-effective solution, coupled with a recirculation system that can also handle surge. On multistage units, recirculation and variable speed are the best options.
The only problem with using vari-able speed on centrifugals is that, in most applications, the compressor cannot operate over the entire speed range available from the driver. These
units run above their ﬁrst critical speed; the critical speed is where the
ﬁrst harmonic vibrations occur for ro-tating equipment, and is normally in the 8,000–12,000 rpm range for most equipment. If you reduce the speed on a centrifugal operating at its de-sign point, you will often ﬁnd a criti-cal disturbance at around 70–75% of the run speed. The variable speed driver would then be programmed to ramp through this speed quickly and avoid it by at least 5% on each side. The difference would have to be re-circulated from a greater speed, if process off-design of normal ﬂow or other part-load point were needed.
The main negatives associated with centrifugal compressors are their comparative costs in pressure/ ﬂow ranges covered by screws and recips, their pressure ratio inﬂexibilities, en-ergy consumption due to difficulties at part-loads, extreme sensitivity to vibration, and the size of the installa-tion required if full API auxiliaries are required.
Typical stages can only operate to roughly 2:1, so multistaging is often required, but intercooling need only take place every second or third stage. These units are available in up to eight stages in a single body. Thus, pressure ratios of 28 are possible.
Many vendors also have drive-through designs that allow two com-pressor bodies to be connected and use a single driver. This allows for more than the standard eight stages. Stage pressure ratio balancing and impeller selection are critical when using two bodies at one speed.
Standard centrifugal air compres-sors normally come in a three-stage format (Figure 7). This allows dis-charge pressures up to 150 psi, from a normal sea level inlet (14.7 psia); for longevity, it is best to use them at 125 psi. High-pressure ratios can be achieved across each air stage, since the bullgear design allows for pinion velocities above 20,000 rpm. Al-though this is fast, and would be questionable on all process gas appli-cations, it is very normal on air and
has been the standard design for air centrifugal units for 20 years.
Oil-free rotary screw compressors
The original screw com-pressor was oil-free. Designed in the 1930s by Dr. Lyceum of
Swedish Rotary Machines
(SRM), the ﬁrst working pro-totype was built in Europe in 1939. Commercial use of the screw compressor did not begin until 1946, due to World War II, but its popularity was
enormous in post-war Europe during the rebuilding process.
In 1948, a group of American busi-nessman obtained a license to build screws from SRM. Americans used these compressors predominately in the steel industry, for coke-oven and kiln gas applications. The steel indus-try was a dominant player in the post-war U.S., as the country built its infras-tructure and the car industry developed into the major player it is today.
The OFS compressor was, and still is, favored for these “dirty” applica-tions since it has the unique ability to pass 200 micron-sized particles contin-uously and without incident. Particles would settle into reciprocating cylin-ders and cause severe damage to the rings and cylinder walls, and sludging of the oil. The particles in these gases would create a sandblast effect on cen-trifugal impellers and cause excessive wear, vibration, and downtime.
OFS compressors can also be built using exotic metals. The standard machine normally has a cast iron cas-ing with forged steel rotors. Howev-er, casings can be made from ductile iron, cast steel, and Type 316 stain-less steel. More importantly, an ordi-nary cast iron or steel casing can be nickel-plated to provide a high level of corrosion resistance and hardness.
The rotors can be
made from different grades of steel, different grades of stainless steel (such as 13-4 and 17-4) and even more exotic materials such as In-conel. It is all a matter of what the buyer can afford. API 619 is the speciﬁcation that covers OFS units.
In contrast, recips have little ﬂ exi-bility in materials choices, with the only major changes normally found in cylinder liners, if they are used in the design of the unit. Centrifugals can also be made in a variety of mate-rials similar to OFSs. OIS models normally use cast iron casings, with most companies offering options to use ductile iron (sometimes called nodular iron) and cast steel. OIS ro-tors are manufactured in either duc-tile iron or steel.
OFSs with polymers
Polymerizing gases are truly where OFS units shine. Gases such as styrene or butadiene tend to coat any contacted metal with polymer over time. This is disastrous to re-ciprocating units, since the cylinder and crankcase become overgrown with polymer, and there is a sludg-ing effect of polymer in the oil. Centrifugals have vibration
prob-lems, since the coating either unbal-ances the impeller or grows out from the casing walls to meet the coating on the impeller, causing a rubbing effect. The coating also re-duces diffuser and volute areas in centrifugal units, which hinders per-formance. For this reason, centrifu-gal compressors used in low-density
and high-density polyethylene
(LDPE and HDPE) service must continually be shut down and cleaned, often chemically. This downtime is expensive, not to men-tion the actual cleaning costs in ma-terials and labor.
The OFS compressor actually im-proves over the life of the polymer coating. These machines are designed so that their timing gears ensure that the two rotors do not actually make physical contact. However, for effi-cient compression, the rotors must be as close to one another as possible. As a polymer coats the rotors and cylinder walls, these clearances close up and increase the efficiency. The rubbing-effect is not a problem, ei-ther. OFS units are solid and probably the least sensitive to vibration.
Thus, as the polymer coating rubs
s Figure 7.
GD-Turbo centrifugal for air contains three stages.
against itself, pieces smaller than 100
µm break off and are released
through the discharge without inci-dent. Polymer applications have built-in ﬁltration and scrubbing sys-tems throughout the piping, since polymerized pieces can be formed and released in the piping, vessels, and other components as well.
Recently, the OFS compressor has become popular in vapor recovery, par-ticularly offshore. Figure 8 shows an API 619 OFS bare unit, waiting to be packaged. The machine will be used for mixed-hydrocarbon recovery vapor recovery in the Gulf of Mexico. The compressor will be motor-gear driven at 6,260 rpm by an 800 kW motor.
Offshore heating tank vapors are no longer ﬂared, but, rather, they are drawn by a compressor and boosted to 100–150 psi. The resultant gas is cooled, ﬁltered, and further com-pressed to the local sales gas pressure (1,000 psi in the U.S.).
In locations such as Alaska and the North Sea, OIS compressors have dominated the vapor recovery unit (VRU) gas booster market that has developed through environmental concerns and regulations. As the North Sea generally was ﬁrst in most offshore developments, the Brazilian and Gulf of Mexico offshore plat-forms have been designed with OIS compressors, as well. Unfortunately, the hotter climates have caused high-er inlet temphigh-eratures to the ﬁrst stage of compression, which has caused problems in that gases that are
nor-mally condensed during cooling prior to the gas compressor are not at high-er temphigh-eratures.
On northern sites, the typical inlet gas temperature falls in the 60–90°F range with the majority of tempera-tures at or below 80°F. In the Gulf of Mexico and offshore Brazil, this tem-perature ranges from 80–130°F and is mostly 100–120°F. This temperature difference allows hydrocarbon heav-ies, such as decanes and higher, to re-main in the gas in sufficient quanti-ties to condense at the discharge and mix with the lubricating oil.
More importantly, since most of these applications are water-saturated, water becomes a big issue as the inlet temperature moves above 100°F. The water content of saturated gas at 120°F is nearly thrice that at 80°F. Typically, this means that the gas should not be compressed beyond 70 psi to avoid condensation at the low temperatures at which OIS units operate.
The combination of lower temper-atures and greater funds spent on the early North Sea platforms yielded process conditions that were not du-plicated in the Gulf of Mexico. How-ever, existing processes are hard to let go of.
This is why some major oil com-panies have had trouble with equip-ment in hotter climates, when the very same machines, using the same process (with lower temperatures) worked so well in colder regions. The solution is to correct the process tem-peratures and pressures, as well as
add better ﬁltration to the newer ap-plications, or use OFS compressors, despite their greater cost.
OFS machines can accept
mist/aerosol entrainment and the higher discharge temperatures associ-ated with oil-free units ﬂash liquids and keep the unsteady gases in the gaseous phase. Also, the pressure limitations would force the user into interstaging, with cooling and ﬁ lter-ing, creating a better environment for the compressor.
The National Oil Company of Mex-ico (PEMEX) has long used OFS units to boost the gas from near atmospheric to discharge pressures of 50–70 psi. Other companies are still trying to achieve the 100–125 psi levels they are accustomed to with their OIS compres-sors, but this would mean two stages in an OFS compressor and signiﬁcantly higher cost.
The solution is in the process engi-neers’ hands. If the process is de-signed to operate at the highest dis-charge pressure achievable by a sin-gle-stage OFS unit, then the ﬁnal dis-charge pressure can be achieved by modifying the already expensive equipment on the back end. Recircu-lation processes can be adjusted by increasing size of vessels and pipe to limit pressure drop.
By the same token, if the process re-ally needs the higher pressure, then the engineer should design to the limit of a two-stage OFS setup, which can be in the 175–225 psi area starting from at-mospheric conditions. This would allow full use of the more expensive two-stage OFS units and drop the cost of any compressors that follow, since they would be much smaller, due to a higher P1. In recirculation processes where there may not be further com-pression, the vessels and piping can be sized much smaller, since the compres-sor could now overcome much larger pressure drops across the system. OFSs and liquids
Liquid entrainment is also a prob-lem for all other compressors. In many saturated gases, there is a
Bare OFS compressor, awaiting packaging. The machine will be used in hydrocarbon vapor recovery.
steady stream of mist or aerosol in the inlet gas. On centrifugal units, this causes erosion similar to partic-ulate bombardment. On recips, this liquid forms in the cylinder and pro-duces a variety of problems from oil dilution to excessive corrosion and wear. OIS compressors may have problems when the liquid does not flash off in time and slowly accumu-lates in the lubrication system until it either replaces or dilutes the injec-tion oil.
In OFS systems, the liquid entrain-ment is easily ﬂashed off due to the 450–500°F maximum temperature for a compressor that can sometimes achieve 5:1 ratios (with low k -value gases) in a single stage. This misting liquid capability was used by engi-neers in the actual design of some process systems.
Capital costs for OFS machinery are always competitive to API cen-trifugal and API reciprocating units, as long as they can be kept in a single stage. What industry has found is that certain gases can accept the introduc-tion of liquids, such as water in styrene, since, many times, the liquid is a part of the process anyway.
For example, with styrene, it is commonplace in both the Badger (now Raytheon) and Lummus (now ABB-Lummus Global) processes to inject water into the OFS inlet. Water is shot in as a mist and the ﬂow rate is set to achieve a speciﬁc discharge temperature from the OFS compres-sor, while increasing the pressure ratio to 6:1 to 7:1, due to the dis-charge temperature drop created by the energy required to ﬂash the water during compression.
In low k -value gases such as iso -butane, 8:1 pressure ratios can be achieved by reinjecting condensed i -butane from the discharge into the suction. Each stage of an OFS com-pressor is basically a separate ma-chine, with its own compressor body and driver. Therefore, cramming the application into one stage by either using the temperature maximums or liquid injection can yield a capital
cost decrease of as much as 50% in certain instances. In addition, opera-tions needs only worry about one set of bearings and seals, not two.
OFS capacity controls
The advantage do not end there. OFS compressors are similar to mul-tistage centrifugal units in the meth-ods by which capacity control is achieved. The three major forms of control — inlet throttling, gas recir-culation, and variable speed — still apply. However, OFS compressors typically operate in the 3,000–8,000 rpm range, so they are well below their typical ﬁrst critical speed of 12,000 rpm. This makes them excel-lent variable speed machines. They can be operated down to roughly 50% of design speed, depending upon the application, and the entire range can be used unless an unusual vibration is found at one of the speeds in between (this is rare).
As mentioned previously, the styrene process uses water. It also uses steam, so the existence of boilers allows for over-sizing and use of steam turbine drivers. Unlike cen-trifugal compressors, which require gear speed increasers most of the time when steam turbines are used, OFS units often run near the steam turbine (backpressure, single-stage type) effective speed of 4,000–6,000 rpm. For styrene, this has allowed the use of direct-driven OFS compressors with steam turbine drivers.
OFS compressors depend upon their rotors’ability to capture gas and push it through a smaller space at the discharge. This means that the most important variable in the aerodynam-ic equation is the rotor tip speed. Users should not be concerned that
smaller OFS units run at
7,000–10,000 rpm, since the smaller rotor diameter reﬂects a desired tip speed. In most cases, OFS compres-sors should operate in the 80–110 m/s tip speed range. Anything over 110 m/s, on a standard 4/6 male/female lobe conﬁguration, should be exam-ined carefully and be proven by the
manufacturer that offers it. In the 4/6 conﬁguration, a male rotor has the input shaft, the female rotor is beside it. They are “mated” for ﬁt and must be replaced as a set.
Note that tip-speed limitations are more easily achieved by large male rotors, say 630 mm dia., than smaller ones, which can go down to 127 mm. Some manufacturers have offered 91 mm dia. versions, but these are very small and would have to be run in ex-cess of 12,000 rpm to achieve perfor-mance-efficient tip speeds.
OFS compressors are shown on the P2 / ﬂow chart (Figure 2). They are good to inlet pressures of 150 psi, and, in some cases, to 225 psi. They can be used to discharge pressures of 350 psi and, sometimes, 400 psi. Re-cent developments have allowed
ﬂows up to 80,000 m3 /h or 47,000
acfm. These higher ﬂows can only be realized with discharge pressures of 200 psi or less, but this has made a great impact on hydrogen recircula-tion and styrene.
One disadvantage that the OFS units have is that they use four shaft seals. Centrifugal compressors use one (single-stage) or two (multistage) seals, while OIS types need only one. Four seals are expensive, necessitat-ing the use of dynamic dry gas seals. Four of these and a buffer system can add $250,000 or more to the cost of a compressor.
OFS compressors are very good process machines. Unfortunately, they have been underused in the U.S., due to operators’ lack of experience with them as opposed to centrifugal and reciprocating types. They would
make excellent ethylene and
polyethylene units, compared with most centrifugals that are used in these processes today. In a recent comparison, an OFS screw package was bid at $800,000 for a single-stage, direct-driven unit vs. a $1.5 million centrifugal. However, the ro-tating engineer recommended the centrifugal, due to his comfort level. He had never used an OFS before, but had used centrifugal units often.
T h e c o m p a n y b a s i c a l l y p a i d $700,000 more for a compressor based on this inexperience.
Oil-injected rotary screw compressors
OIS compressors were introduced in 1958, and their most famous fea-ture, the slide valve, was invented in 1959. Once again, both developments occurred in Europe, so that the U.S. has been behind the times ever since. API 619 did not include this type of compressor until the very latest edition (3rd), which was released in 1997.
The OIS compressor was ﬁrst ap-plied in the air market in the 1960s, and, by the 1980s, these units were dominant. In the 1970s, they entered the refrigeration market, and became principal by the late 1980s. A decade later, acceptance grew in the fuel gas market, and, by the 1990s, they were used on fuel gas wherever they could, only limited by their own pressure ca-pabilities.
Figure 9 is a packaged OIS sys-tem, rated 250 hp, for compressing boiler off-gas at a liqueﬁed propane gas (LPG) facility. The compressor is at the right of the photo, and the box at the left is a fan cooler.
OIS compressors will likely be-come the machines of choice in most vapor recovery, fuel gas, and other process gas industries by 2015. Three engineering developments have oc-curred to make this so: lubricant ad-vancement, gas ﬁltration to levels below oil-free standards, and pres-sure/ ﬂow improvements.
Until the mid-1980s, OIS units were heavily limited by lubrication. The lubricants available would easily be oxidized, diluted, and broken down, unless used on gases that were innocuous, such as ammonia. Air would oxidize the oil, water would cause foaming, and hydrocarbons would dilute the lubricant. These ini-tial issues caused reliability problems. The fact that mechanical engineers and operators were accustomed to re-ciprocating compressors did not help either. For these reasons, screw
com-pressors received a bad reputation among U.S. engineers, and were not taken seriously until the 1990s.
Custom lubricants came about in the 1980s and the landscape changed. Now, air compressors run longer, cor-rode less, and require fewer oil change-outs than before. Hydrocarbon gases no longer dilute the lubricant beyond use-fulness and water can be controlled. Lubricants have evolved from hy-drotreated mineral oils to synthetics, such as polyalpha oleﬁns (PAOs), poly-ol esters (POEs), and ppoly-olyalkylene gly-cols (PAGs). Currently, there are also additives that can be used, such as an-tioxidants, antihydrates, and anticorro-sives. Lubricants are even available in various viscosity grades, so that dilu-tion can be planned and accounted for.
The industry maximum acceptable dilution rate is 20% by gas to oil. OIS compressors should not operate below 12 cSt, or higher than 300 cSt, so an established range can be main-tained by a speciﬁc lubricant and vis-cosity grade. This has allowed exten-sive use of OIS compressors in hy-drocarbon service.
Most recently, ﬁltration companies that were players in the medical ﬁeld have entered the commercial gasﬁ ltra-tion arena, and another evolultra-tion will result. In the past, effectively cleaning inlet gas to 0.3–1.0 µm, or reducing
oil carryover in the gas to less than 1
ppm were either not possible or ex-pensive —but this is no longer so.
With better inlet ﬁltration, OIS compressors can now be used more frequently with gases that contain particulates, such as carbon ﬁnes. These ﬁnes normally fall in the 1–10
µm range, so they would pass
through traditional inlet ﬁlters. Now,
ﬁlters can be designed for 0.5µm and
will stop the smaller 1µm particles.
Ensuring oil-free service
More importantly, oil/gas separa-tion has evolved to such an extent that gas compressor packages can now cost-effectively guarantee an oil con-tent of 0.01 ppm or less in the gas stream leaving the skid. A standard for oil-free was set by the International Organization for Standardization (ISO), Geneva (www.iso.ch/). The or-ganization’s ISO 8573-2, “ Com-pressed Air for General Use —Part 2: Test Methods for Aerosol Oil Con-tent,” (1996) is a speciﬁcation created to standardize what process engineers consider being oil-free and how the value can be veriﬁed. Although this speciﬁcation is written for air, the 0.01 ppm ﬁgure put forward by this docu-ment has widely become accepted as the value to which equipment should be designed to achieve oil-free status.
The ﬁgure is no accident; oil smokes (mostly aerosols) fall into the
0.01–1.0 ppm range. They could not be ﬁltered by coalescing and ﬁ ltra-tion elements until very recently. Thus, entrained oil would be cap-tured, but aerosols (or smokes), would get through and could contam-inate a gas up to 1.0 ppm by weight. If 0.01 ppm is achieved, the gas is basically oil-free.
free is a truly a misnomer. Oil-free screw compressors and centrifu-gals used on air are not necessarily oil-free. This is because most industrial air is drawn at a petroleum reﬁnery, or a chemical or petrochemical plant, and the atmospheric air is not free of oil there. Granted, oil is not introduced, but a sensitive process, such as phar-maceutical production or processes using catalyst beds, would be ruined by the oil that is drawn in air. Filtration is the only solution to ensure air, or even gas, quality. The goal is 0.01 ppm.
Reciprocating compressor vendors accepted their inability to achieve oil-free status long ago. That is why they are called nonlubricated reciprocating compressors, instead of oil-free. Of course, nonlube recips often do intro-duce oil into the process, since reciprocating units produce considerable amounts of oil vapors.
The third major reason is pressure and ﬂow design. Until 1990, OIS units could not ex-ceed 4,500 acfm in a single compressor body, now there are units that can do 10,000 acfm. Until the late 1980s, the discharge pressure was limited to 350 psi, however, recent innovations have brought standardized units up to 520 psi and special designs to 865 psi. The latter units accept a 700 psi inlet pressure, hence, their recent in-roads in the fuel gas market.
Depending upon the differential pressure across the compressor, and the ﬂow, the inlet pressure may be pushed to 120 psi on ratios below 4:1 or be limited to 70 psi on ratios above 7:1. In many instances, a process de-sign inlet pressure of 125 psi can be reduced to 100 psi without great im-pact. This small change can reduce
costs by one-half on the OIS com-pressor package. A bare OIS unit is shown in Figure 10.
The fuel gas market is a good ex-ample of how the newer OIS machines can be used to great advantage over re-cips and centrifugals. Most power gen-erating facilities built today use
gas-ﬁred turbines. The incoming natural gas from the utility normally ﬂuctuates and must be regulated to the turbine pressures. Unfortunately, the pressure often dips below the turbine require-ments and compression is required.
On a recent project, the incoming natural gas ranged from 310 psi to 600 psi and the desired discharge was
650 psi. A centrifugal must be de-signed for the worst case, so an inlet pressure regulator was required to en-sure that it always received 310 psi, ±5%. Thus, if the incoming gas were at 600 psi, energy would be wasted, since the gas would be reduced then recompressed.
On a reciprocating compressor, the entire inlet range could not be accept-ed, due to rod load problems at the high end. Again, a regulator or expen-sive unloaders would be required, and the latter are not always reliable.
The OIS compressor can take the entire range and unload at the high end to reduce energy consumption by 60%. A regulating valve is not required, ei-ther. Thus, the OIS compressor not only offers the same capital cost as an API 11P reciprocating unit, but the client could purchase a unit that is more efficient over the entire range of compression and more reliable, as well. The added ﬂexibility of the slide valve also reduces the size of the recir-culation cooler and saves water.
OIS units are sometimes called
oil-ﬂooded, but this term is entirely inac-curate. The oil is literally injected, at a pressure higher than discharge, into several key areas of the compressor to provide lubrication, sealing, and cool-ing. Nearly two-thirds of a fully in- jected screw compressor’s oil goes to-wards cooling, not lubrication. This is why OIS units can produce pressure ratios as high as 23:1 in a single stage. Most companies do not use these machines at ratios above 10:1, since they become extremely inefficient with adiabatic efficiencies dropping to 50% at 15:1 ratios. Normal adiabatic effi-ciencies for screws, at a single de-sign point, are 70–80%. Howev-er, their true installed efficiency remains close to those percent-ages, since OIS units are con-sidered among the most effi-cient compressors for overall operation at different points, due to the slide valve.
This valve is an internal ca-pacity control device, and is built into all reputable gas machines, without incurring an additional cost. In fact, it is part of the design. The slide valve literally recirculates the compressed gas before compression is completed. That means that only a little energy is required to boost the gas enough so that it can be internally recirculated to the internal suction of the machine.
Most slide valves can operate in the 10–100% range, so 0% ﬂow is achievable with a very small recircu-lation system; GD recommends a 25% size. This removes the need for inlet throttling, full-size recirculation,
s Figure 10. Bare OIS gas compressor with inlet valve mounted.
or variable speed control. In fact, OIS units do not make good VFD units, since their efficiency is so closely tied to tip speeds in the 50–60 m/s range, and reducing the machine’s speed would reduce its tip speed.
The slide valve can be instructed to load or unload based on either P1 or
P2. Thus, the process has a very sim-ple capacity control device, based on system pressures. The slide valve has also been proven to be very reliable.
The fact that screw compressors use injection oil for cooling is another positive for process design. If a pro-cess engineer would like to maintain a discharge temperature, or if the gas condenses at a known temperature, then the oil injection can be throttled to maintain a preset discharge temper-ature. Using a control valve on the main injection line and a temperature signal from the discharge does this. We recommend that a process stay 18°F (10°C) above the discharge dew point for water, and 25°F (14°C) for hydrocarbon gases. Once the gas makes it through oil/gas separation, there is no the danger of fouling the lubricant and harming the compressor. OIS compressors can operate up to 10,000 acfm in a single body unit, how-ever, the most cost-effective and com-petitive area is at 5,000 acfm and below. Standard compressors are generally one-half the price of custom units, but
custom pricing does compare favorably with most recips. The great advantage is that standard OIS units can allow for skid deliveries in the 20–26 week range. Figure 11 shows a two-stage OIS, preceded by a blower, all skid mounted. Special casings would push this out to 35 weeks, while custom (API 619) ma-chinery normally takes about 40 weeks. All these delivery times are shorter than those for centrifugal and API 618 recip-rocating machinery.
The inexpensive screw compressors can achieve discharge pressures of 520 psi, with some cost-effective modiﬁ ca-tions. Standard OIS gas units can achieve 350 psi at the discharge, and this pressure level provides the buyer with many market options and creates a healthy competitive bid situation.
The main negative found in OIS units is that they do require fairly clean gas at the inlet during normal operation. They handle upsets better than recips and centrifugals, but they cannot do so on a continuous basis. Oil carryover is really no longer an issue with the improvement in ﬁ ltra-tion technologies. However, polymer-izing gases are still a problem, so styrene and butadiene applications where these gases are above 20% content are deﬁnitely not
recom-mended. Even antipolymerizing
agents have been unsuccessful on OIS units.
To sum up
In the end, the process conditions and gas will dictate which is the best compressor for a particular applica-tion. If a process engineer properly uses the information presented here, then the choice for the right compres-sor will become abundantly clear.
The engineer must understand where the ﬂows and pressures are in relation to available equipment, and see if the process can be adjusted to meet the capabilities of units that are readily available, reliable, and inexpen-sive for the application under study. Working with the mechanical depart-ment and compressor vendors, the right compressor will lead to the right
ﬁnal process conditions, or vice-versa. The most important advice is to keep an open mind, and use an itera-tive selection process. Any systems engineer or economist will tell you that working in a closed loop without feedback will lead to ruination. Open that external loop, gain that external feedback, and a marginal process can
become a great one. CEP
D. G. Jandjelis product manager, gas compressor systems, for The Gardner Denver Engineered Packaging Center (formerly Allen-Stuart Equipment), Houston ((713) 6510 ext. 131; Fax: (713) 896-1154; E-mail: email@example.com). He is involved in all aspects of marketing, sales, and management of the company’s gas compressor division, and is responsible for major accounts, worldwide. His technical duties include reviewing engineering data, selecting the most appropriate systems for bidding, and client liaison throughout production and testing. He has conducted seminars for over 50 major clients and has extensive experience in specifying, costing, troubleshooting, and engineering
compressors. Prior to his current employment, he was a manager for both Howden Compressors and A-C Compressor Canada, Inc. He began his career as a mechanical engineer with Ingersoll-Rand Canada, Inc. Jandjel holds three degrees: a DEC in pure and applied science from John Abbott College, a bachelor’s in mechanical engineering from McGill University, and an MBA, also from McGill.
All photos courtesy of Gardner Denver, Inc. s Figure 11.
Landﬁll gas compression. First stage is handled by a blower; second and third by two-stage OIS unit.