Cen­trifu­gal com­pres­sors: op­er­a­tion, re­li­a­bil­ity, start-up and main­te­nance

AMIN AL­MASI is a lead me­chan­i­cal en­gi­neer in Aus­tralia. He is char­tered pro­fes­sional en­gi­neer of En­gi­neers Aus­tralia (MIEAust CPEng – Me­chan­i­cal) and IMechE (CEng MIMechE) in ad­di­tion to a M.Sc. and B.Sc. in me­chan­i­cal en­gi­neer­ing and RPEQ (Reg­is­tered Pro

DEMM Engineering & Manufacturing - - MAINTENANCE PLANNING - BY AMIN AL­MASI


There have been many types of com­pres­sors such as re­cip­ro­cat­ing, screw, ax­ial, etc; how­ever, cen­trifu­gal com­pres­sors are known as main compressor op­tions for many mod­ern in­dus­trial and man­u­fac­tur­ing plants. In past three decades, cen­trifu­gal com­pres­sors have re­placed many other compressor types in many dif­fer­ent ser­vices and ap­pli­ca­tions. They are now most widely used compressor type in mod­ern plants.

In gen­eral, cen­trifu­gal com­pres­sors have high tech­ni­cal de­mands and re­quire elab­o­rate re­li­a­bil­ity knowl­edge, main­te­nance and op­er­a­tional knowhow. The safe, re­li­able, and sus­tained op­er­a­tion of cen­trifu­gal com­pres­sors rep­re­sent a key com­po­nent for max­i­miz­ing on-stream pro­duc­tion in many plants. Re­li­a­bil­ity, op­er­a­tion and main­te­nance of cen­trifu­gal compressor trains are dis­cussed.


Cen­trifu­gal com­pres­sors have been used in many dif­fer­ent ap­pli­ca­tions from light gases to heavy gases and in a wide range of op­er­a­tional tem­per­a­tures from cryo­genic to hot gases. Most chal­leng­ing ones are those with high head per im­peller. Op­er­a­tors have usu­ally tended to limit the head per im­peller within cer­tain lim­its for bet­ter op­er­abil­ity, op­er­a­tional flex­i­bil­ity and re­li­a­bil­ity. As a very rough in­di­ca­tion, 2500 – 4500 m per im­peller might be noted as such limit de­pend­ing on de­tails of each ser­vice. As an av­er­age fig­ure, 4000 m per im­peller might be con­sid­ered. For ex­am­ple, for a fluid with den­sity of around 1 kg/m3, this means 0.4 – 0.5 Bar per each im­peller. When gases com­pressed, the den­sity is in­creased and higher pres­sure per im­peller is ex­pected. As an­other rough in­di­ca­tion, for a fluid with den­sity of around 10 kg/m3, the above­men­tioned means 3- 4 Bar per each im­peller. Ob­vi­ously for rel­a­tively heavy weight gases at high pres­sures (say above 80 Bar), the den­sity is rel­a­tively high, say around or above 50 kg/m3, and high pres­sure rise by im­peller (above 20 Bar per im­peller) might be ob­tained for 3000 – 4000 m per im­peller.

There have been com­pres­sors which can achieve a higher head per im­peller; for ex­am­ple, even 6500 m or 7000 m per im­peller or more. High head per im­peller can be at­tained with the com­bi­na­tion of high tip speeds, proper de­sign us­ing high strength ma­te­ri­als, and high head co­ef­fi­cient (3D so­phis­ti­cated im­pellers). These ex­tremely high head im­pellers are not pop­u­lar be­cause they of­fer nar­row op­er­at­ing ranges, risks of surge and op­er­a­tional prob­lems. Rel­a­tively high head per im­peller re­quires a high ro­ta­tional speed, op­er­a­tion at high Mach num­bers, a lim­ited op­er­at­ing range, re­duced ef­fi­ciency, rel­a­tively low head rise-to-surge and dif­fi­cul­ties in the con­trol and op­er­a­tion. An op­ti­miza­tion should be done as lower head per im­peller means more im­pellers re­sult­ing in a long bear­ing span and po­ten­tials of ro­tor­dy­nam­ics is­sues; also, this is more ex­pen­sive and re­sults in heaver and larger com­pres­sors. Usu­ally, any compressor for above 4500 m head per im­peller should be dealt with a great care. Compressor’s ef­fi­ciency is a crit­i­cal mat­ter. For most mod­ern cen­trifu­gal com­pres­sors, the ef­fi­ciency falls in the ef­fi­ciency range of 80 – 88 per­cent. As more im­pellers are added to a compressor, op­er­at­ing ranges get ef­fected and re­duced; also, the com­bined ef­fi­ciency, over­all re­li­a­bil­ity and sta­bil­ity in­her­ently might be re­duced. The sta­bil­ity re­duc­tion is crit­i­cal which is of­ten re­lated to dy­nam­ics, surge, stall, etc. One rea­son is there could be in­ter­ac­tions be­tween im­pellers; also, where there are many im­pellers, over­all, the op­er­at­ing range of the assem­bly can be lim­ited and there is more chance that any de­vi­a­tion or op­er­at­ing point move­ments lead to surge or other in­sta­bil­ity.

Map of each in­di­vid­ual im­peller af­fects the fi­nal map and the over­all surge points; there­fore, the fi­nal op­er­at­ing map and mar­gin to the surge of a com­bined im­peller assem­bly could be lim­ited where there are many im­pellers in a compressor. Care­ful stag­ing can al­low tak­ing ad­van­tage of the best op­er­at­ing range of each im­peller, as there have been many suc­cess­ful com­pres­sors with many im­pellers. How­ever, where there are many im­pellers, each can limit the fi­nal op­er­at­ing map of the compressor to some ex­tent; there­fore, the fi­nal map may be re­stricted. Each im­peller should be care­fully se­lected in con­cert with other im­pellers. In­ter­ac­tions be­tween dif­fer­ent compressor im­pellers and stages should be con­sid­ered and matched. Over­all, any compressor should have an ac­cept­able op­er­at­ing range/map; prefer­ably the range should be rel­a­tively wide range, con­sid­er­ing rated con­di­tion(s) and al­ter­na­tive op­er­at­ing cases.

An­other im­por­tant op­er­a­tional con­sid­er­a­tion is the gas ve­loc­i­ties in dif­fer­ent part of a compressor. Compressor noz­zle ve­loc­i­ties should be care­fully eval­u­ated, and the noz­zle should be prop­erly sized. Noz­zle ve­loc­i­ties in 20 - 40 m/s range are com­monly used de­pend­ing on dif­fer­ent fac­tors such as gas molec­u­lar weight, etc. The higher the molec­u­lar weight and the lower the tem­per­a­ture, a lower ve­loc­ity limit should be em­ployed. For spe­cial sit­u­a­tions such as re­vamp/ren­o­va­tion projects, rel­a­tively high ve­loc­ity lim­its (even 40 m/s or more) might be ac­cepted af­ter thor­ough ver­i­fi­ca­tions. Rel­a­tively high noz­zle ve­loc­i­ties can trans­late into high pres­sure drops, op­er­a­tional prob­lems, lower op­er­at­ing range, risks of some vi­bra­tions and more noise.


The compressor load at the start-up should usu­ally be kept as low as pos­si­ble to keep the start­ing torque at min­i­mum pos­si­ble. The low pres­sure of gases at the compressor start-up was of­ten achieved by de­pres­sur­iz­ing the whole com­pres­sion sys­tem; this is known as the de­pres­sur­ized star-up which was a very tra­di­tional and old-fash­ioned way of op­er­a­tion. Com­pres­sors then can be loaded once the driver reaches a suit­able con­tin­u­ous speed. The de­pres­sur­ized star-up has al­lowed re­duc­ing the size of driver or start­ing de­vice. How­ever, this start-up method is not pop­u­lar

any­more be­cause of en­vi­ron­men­tal and op­er­a­tional is­sues. In a plant, this de­pres­surised start­ing method could re­sult in:

• The loss of gases to de­pres­surise com­pres­sion loops af­ter any plant shut­down.

• Flar­ing or de­struc­tions of a large amount of gases.

• In­creased plant down­time; some­times 5 – 12 hours re­quired to com­plete plant start-up for a de­pres­surised start-up.

As a con­se­quence, the pres­sur­ized start­ing ca­pa­bil­ity can help in­creas­ing the over­all plant avail­abil­ity, per­for­mance, op­er­a­tion and ef­fi­ciency. The in­cen­tives of pres­sur­ized start­ing ca­pa­bil­i­ties led to sig­nif­i­cant de­vel­op­ment ef­forts for cen­trifu­gal com­pres­sors. For a compressor train, the driver siz­ing, de­sign and op­er­a­tional de­tails should be in a way to pro­vide the ca­pa­bil­i­ties for the pres­surised start-up. Since the re­quired start-up torque is mainly a func­tion of compressor fluid load (pres­sure, etc) and ro­tat­ing in­er­tia, it is crit­i­cal to fully un­der­stand the com­plete com­pres­sion sys­tem as well as the whole driver assem­bly for a proper start-up ar­range­ment. A rig­or­ous dy­namic sim­u­la­tion should also be per­formed to take into ac­count start-up tran­sients; this needs an ac­cu­rate dy­namic model and data, the ac­tual con­trol soft­ware and re­al­is­tic op­er­a­tional per­for­mances. The pres­surised start­ing is a mod­ern re­quire­ment and should be con­sid­ered for mod­ern cen­trifu­gal com­pres­sors.


An ac­cu­rate dy­namic sim­u­la­tion is needed in the dif­fer­ent stages of selec­tion and op­er­a­tion, for in­stance the sup­port of siz­ing of many spe­cific key el­e­ments in the com­pres­sion sys­tem, com­mis­sion­ing, and re­li­a­bil­ity in­ves­ti­ga­tions. Gen­er­ally, dy­namic re­li­a­bil­ity sim­u­la­tions are fun­da­men­tal to un­der­stand the op­er­a­tion of a compressor sys­tem in all op­er­at­ing sit­u­a­tions es­pe­cially tran­sient cases. Par­tic­u­larly such an ac­cu­rate dy­namic sim­u­la­tion is crit­i­cal for the op­er­a­tion as all im­por­tant tran­sient cases such as all cases of start-up, emer­gency shut­down, mal­func­tion, etc should be prop­erly sim­u­lated and re­quired ac­tions from op­er­a­tors should clearly be de­fined. Those el­e­ments which need help for siz­ing and ver­i­fi­ca­tions could be re­lief valves, con­trol valves, suc­tion/dis­charge valves, equip­ment siz­ing, tim­ing pa­ram­e­ters and oth­ers.

Among var­i­ous op­er­at­ing sce­nar­ios, a compressor emer­gency shut­down (or compressor emer­gency trip) can rep­re­sent one of high-risk cases which can ex­pose the compressor to a se­vere surge con­di­tion. Ma­jor plant up­sets can cause un­wanted emer­gency shut­downs of compressor sys­tems be­cause there is no time for a nor­mal shut­down pro­ce­dure; such a shut­down im­poses po­ten­tial risks of surge on a compressor. An anti-surge sys­tem for any cen­trifu­gal compressor should be pro­vided to pro­tect the compressor from surge dur­ing an emer­gency shut­down. The in­ves­ti­ga­tion of his­to­ries of some plants has in­di­cated that an emer­gency shut­down rep­re­sents the high­est risk to cen­trifu­gal com­pres­sors. In some cases, based on dy­namic sim­u­la­tion re­sults, a hot-gas-by­pass anti-surge valve should be con­sid­ered to avoid surge dur­ing such an emer­gency trip. Al­though such an anti-surge valve, which is in ad­di­tion of nor­mal anti-surge valve, should be con­sid­ered as the last so­lu­tion. Typ­i­cally, such a hot-gas-by­pass anti-surge valve is a fast-act­ing valve, to open in less than one sec­ond, to re­lieve the gas to the suc­tion side and avoid surge at emer­gency shut­down. The route of such valve is a short dis­tance be­tween the dis­charge and suc­tion and be­cause there is no cool­ing on such a short loop, it is known as hot-gas­by­pass; it can only be used for a short time of shut­down.

Gen­er­ally, anti-surge and con­trol in­stru­ments should be as close as pos­si­ble to the compressor noz­zles. Com­mon rec­om­men­da­tion is 1 – 5 me­tres from the compressor noz­zles for pres­sure and tem­per­a­ture mea­sure­ment in­stru­ments. The flow mea­sure­ment el­e­ment which is a key el­e­ment of any anti-surge sys­tem should usu­ally be within 4 - 12 me­tres from the compressor noz­zle. Too of­ten, a flow mea­sure­ment el­e­ment is more of a chal­lenge, given the ad­di­tional re­quire­ments for straight runs of com­monly-used flow el­e­ments; for ex­am­ple, “10 D” up­stream and “5 D” down­stream (D: di­am­e­ter of the pip­ing). Mount­ing in­stru­ments and sen­sors too far from compressor noz­zles might lead to some op­er­a­tional is­sues such as slower ac­tion on chang­ing vari­ables and there­fore slower re­ac­tion of the anti-surge sys­tem, which could re­sult in some in­sta­bil­ity and dam­ages.


Compressor per­for­mance test­ing at man­u­fac­turer’s shop and also at site are im­por­tant re­quire­ments for nearly any cen­trifu­gal compressor train. The test­ing ca­pa­bil­ity of com­pres­sors’ man­u­fac­tur­ers should be con­firmed dur­ing a bid­ding stage. It is nec­es­sary to ver­ify the per­for­mance, me­chan­i­cal run­ning, ro­tor­dy­nam­ics, tor­sional be­hav­iours and sta­bil­ity of op­er­a­tion of any compressor be­fore its de­liv­ery to the job site.

The head co­ef­fi­cient and ef­fi­ciency of an ad­vanced cen­trifu­gal compressor varies with many spe­cific pa­ram­e­ters and op­er­a­tional de­tails such as Mach num­ber, etc. There­fore, such co­ef­fi­cients, pa­ram­e­ters and de­tails are usu­ally unique in each ap­pli­ca­tion. On this ba­sis, a per­for­mance test­ing is re­quired to be con­ducted at the site rated con­di­tions or con­di­tions close to rated con­di­tions. This is specif­i­cally true for the Mach num­ber; there­fore, the test should be at the rated Mach num­ber or very close to it. Reynolds num­ber is a di­men­sion­less quan­tity that is used to help pre­dict sim­i­lar flow pat­terns in dif­fer­ent fluid flow sit­u­a­tions.

For com­pres­sors, like many other ma­chiner­ies, Reynolds num­ber closely re­lates to the boundary layer and fric­tional losses. These char­ac­ter­is­tics should be eval­u­ated very care­fully for com­pres­sors. Only ASME PTC type-1 test can of­fer an ac­cu­rate and use­ful per­for­mance test ar­range­ment for cen­trifu­gal com­pres­sors.

An­other im­por­tant is­sue is cor­rect pre­dic­tion of surge points dur­ing a shop per­for­mance test. Surge points could be dif­fer­ent if the pip­ing and in­ter-stage fa­cil­i­ties are var­ied. In other surge points are dif­fi­cult to be re­con­structed in the shop test. A smart de­ci­sion could be us­ing ad­di­tional ves­sels or pip­ing spools in the test ar­range­ment, to sim­u­late the dis­charge and suc­tion gas vol­umes as close as pos­si­ble to the site ar­range­ment. For a vari­able speed cen­trifu­gal compressor, the surge points at dif­fer­ent speed should be mea­sured.

A con­sid­er­able por­tion of a VSD elec­tric driven compressor string test­ing is usu­ally ded­i­cated to the fi­nal val­i­da­tion of the VSD sys­tem and its in­te­gra­tion with the main equip­ment as well as with the elec­tri­cal power sys­tem. Shaft torque rip­ple is a well-known weak point of vari­able fre­quency sys­tems; this rip­ple and the amount of har­monic dis­tur­bances in­jected by a VSD sys­tem into the sur­round­ing elec­tri­cal grid should closely be mon­i­tored with an ad­vanced record­ing sys­tem. The gen­eral ex­pec­ta­tion is that the test re­sult should val­i­date low har­monic ef­fects and low torque rip­ple, well within the rated val­ues, there­fore not re­quir­ing any mit­i­ga­tion mea­sure such as an ex­ter­nal sup­pres­sion, ad­di­tional har­monic fil­ters, etc.

There is al­ways con­cern about the vi­bra­tion be­hav­iour of a ma­chin­ery train in a per­for­mance test. In ad­di­tion to the job’s vi­bra­tion mon­i­tor­ing sys­tems in­clud­ing bear­ing vi­bra­tion mon­i­tor­ing sys­tems, cas­ing vi­bra­tion sys­tems and oth­ers (which is pro­vided for the compressor and shipped by the compressor), a thor­ough tem­po­rary vi­bra­tion mon­i­tor­ing sys­tem should be in­stalled on the pip­ing, cas­ings, bear­ing pedestals and sup­port struc­ture to get a bet­ter in­sight to all vi­bra­tion and dy­namic ef­fects gen­er­ated. This pro­vides a com­plete pic­ture of the train ro­tor­dy­nam­ics, over­all ma­chin­ery dy­namic be­hav­iours and whole struc­tural vi­bra­tions.


Newspapers in English

Newspapers from New Zealand

© PressReader. All rights reserved.