Severe surge incidents at Process

Air Compressor, their after effects

and problem resolution
A 37 years old centrifugal Process air compressor suddenly starts surging at its normal operating
point, after overhaul in 2004. Detailed investigation reveals a design issue in HP case of the machine
which was hidden since 1966. Surges on the machine results in damage of downstream secondary re-
former catalyst and fouling of waste heat boilers. This article explains the investigation details, root
cause and how to avoid such pitfalls in design/relocation of any centrifugal compressor.

Muhammad Majid Latif

Engro Chemical Pakistan Ltd.

urea capacity to 600,000 tonnes per year. In 1995, the

Introduction urea capacity was expanded to 750,000 tonnes per year.

E ngro Chemical Pakistan limited is the second In 1998 urea production capacity was increased from
largest manufacturer of Urea N46 in Pakistan. 750,000 to 850,000 tonnes along with an overall energy
The manufacturing site is located in a small rural efficiency of 7%. Further to this other projects were im-
town of Pakistan known as Daharki which is 600 plemented which have increased energy efficiency by
km north of famous port city of Karachi. The site was another 5%.
setup in 1965 after discovery of vast resources of natural Daharki site is currently operating one 1660 tonnes per
gas in vicinity and was named as Esso Chemical Paki- day steam reforming ammonia plant and two Toyo De-
stan, which was later named as Exxon Chemical Paki- sign Urea plants of combined capacity of 2830 tonnes
stan. In 1991 Exxon decided to divest its equity from per day.
global fertilizer manufacturing. At that time Pakistan
based employees of Exxon along with reputable interna- Background
tional investors decided to purchase major share hold-
ing. Exxon’s patent of brand name “Engro” was also a In original design of plant (when commissioned at
part of the deal. Pascagoula), process air compressor was designed
The Company's original plant was commissioned on to deliver 2050+ KSCFH (Kilo Standard Cubic
December 4, 1968, and since then, except for short Feet) of air to meet H2/N2 ratio of 3.0 for ammonia
maintenance shutdowns, the plant has been in continu- production. However after relocation to Daharki,
ous operation. The initial urea capacity was increased Pakistan the air demand reduced considerably (to
from 173,000 to 268,000 tonnes per annum till 1993.
around 1700 KSCFH) the main reason was that
In 1993, Engro relocated Bechtel design Ammonia and
a TOYO Total Recycle Urea plants from Pascagoula,
unlike Pascagoula, at Daharki the feed Natural Gas
USA and Bellingham UK respectively and increased its contains about 18% Nitrogen.

2005 320 AMMONIA TECHNICAL MANUAL

Compressor Details coupled with a steam helper turbine. (For more
details please refer to figure 1 ).
Process air compressor was originally designed
and manufactured by “Clark Brothers” in 1966, Machine is connected with Secondary reformer
the company owner ship and name changed sev- of Ammonia plant, through primary reformer
eral times after that and now it has became convection section air preheat coils. A simpli-
Dresser Rand. fied flow diagram of the circuit is given as fig-
The machine consists of four stages, which are ure 2.
contained in two casings (i-e LP casing and HP
casing). Original antisurge scheme on this machine con-
isisted of single antisurge blow-off valve (FR-
Compressor Driver consists of one 17000 HP 117) at the final stage of the machine.
GE frame 5 single shaft gas turbine which is-

Simplified Diagram of Process Air compressor, K-2501 Train.

(Figure 1)

Instt. Air
to UTY-1
E-2503 D-2508 ATM 60 KSCFH
Suction filter

Process Air
to Sec Ref.
118
FR
NH3

1700 KSCFH
Suction filter

Anti Surge PRC-335

Valve

116
FR
FRC-117
HP case

117
LP Case

FR
Helper
Gas
Steam GB 1st Stage 2nd Stage GB 3rd Stage 4th Stage
Turbine
Turbine

Extraction

70 KSCFH
Air

Air
91 1
RF

E-2502 D-2507 E-2504 D-2509

AMMONIA TECHNICAL MANUAL 321 2005

Overhaul History
HP case of the process air compressor, K-2501
was first time overhauled in 1993 (at the time of SEQUENCE OF EVENTS :
its commissioning at Daharki site). After 1993 al-
though small reduction in machine efficiency was During Ammonia plant Startup on 17th April 2004
noted, however only LP case of the machine was at 9:00am Process Air compressor Train, K-2501
overhauled in 1995 and HP case was never surged. The machine remained in surge for
opened. around 7 minutes. Ammonia plant back end was
It was decided to overhaul the machine during tripped and machine was offline from the process
annual shutdown of plant in April 2004. Both immediately by fully opening its 4th stage dis-
LP and HP cases of the machine were opened up charge vent valve FRC-117.
and completely overhauled. While only erosion K-2501 surged again on 28th June 2004, due to
was noted in LP case of the machine, consider- slight fluctuation in fuel control valves. The ma-

ATM Air preheat coil

FIGURE 2
Process Air Flow
Antisurge Blow
off valve

FR-117

FR-116

2°Reformer
10"
12"

D-2504
Air Intake

To HTS
FR-117 FR-116 Bee Hieve
coil protection

From Primary
Steam for air

Reformer

K-2501
6"-90 ft Dead Leg
Waste Heat
Process Air Boilers, E-2501A/B
Compressor

able damage was observed in HP case of the chine had never surged before at similar operating
machine. It was found that the IGVs (Inlet conditions.
Guide Vanes ) and rotor of the 3rd stage of the
machine has been damaged and eroded (Please Due to similar nature of both events, details of
refer to damage photos given at end ). first event only are given below :

Damaged rotor and IGVs were replaced and First Surge Incident :
complete overhaul of HP casing was carried out.
Ammonia plant was started on 16th April 2004 April 17th , 2004
(after turnaround) and production was achieved at
around 2100 hrs. After that ammonia FE load was
being increased till the time of the event (at which
it was 140%).

2005 322 AMMONIA TECHNICAL MANUAL

8:30 am 9: 05 am
Backend (BE) boardman observed that It was observed in field that FRC-117
ammonia loop pressure was going down. This is a transmitter needle was moving from zero to max
clear indication that plant now requires more air along with the wheezing sound and duration for
addition to maintain same pressure and H2/N2 ra- each surge cycle was around 5 sec. FE boardman
tio. opens FRC-117 from 15% to 75%. Field operator
and Shift Supervisor witnesses in the area that
8:31 am valve moves to 75% opening. Surge didn’t stop.
BE boardman requests Front End (FE)
boardman to increase air by increasing speed of 9:08 am
process air compressor, K-2501. FE boardman FE Boardman fully opens FRC-117 (veri-
instructs him to close air compressor vent valve fied in field), Ammonia back-end is tripped, air is
FR-117 which was 17% open at that time (this cut and normal shutdown sequence is followed.
valve remains 100% closed during normal plant
operation). 9:09 am
Air Compressor comes out of continuous
8:35 am surging Gas turbine rpm and exhaust temperatures
With permission of FE boardman, BE normalize. Air compressor vibrations revert back to
boardman starts closing FRC-117 in small steps of normal values.
1-2%. Gas turbine was being operated at 4805
rpm at that time.
TECHNICAL EVALUATIONS :
Detailed evaluations of data captured at Plant DCS
(Honey Well, TDC-3000) and at the Emergency
9:02 am
Shut Down system (PLC) were carried out and a
Shift Coordinator was informed by Field
fault tree was developed. (Please refer to Fault tree
Operator that wheezing noises are coming from
analysis given below).
process air compressor. Alarms started coming on
Gas Turbine Speed Control (Mark-V) panel. Tur-
Following are the highlights:
bine speed fluctuation from 4400 to 5200 rpm
started. Vibration circuit fault alarm (K-2501) ap-
A single antisurge /blowoff valve, FRC-117 is
peared on Bentely Nevada panel. All vibration in-
installed at 4th stage discharge of the machine.
dications for Process air compressor (especially
The antisurge system is based on a single point
HP case), K-2501 started fluctuating.
programmed in DCS, TDC-3000. i-e flow point
at maximum speed of process air compressor.
9: 04 am
Compressor discharge flow, FRC-117 On 17th April at the time of surge, the flow at
started varying from zero to max, it was also vary- compressor discharge was 1700 KSCFH at gas
ing on FRC-116 (air to 2°Reformer), low air flow turbine speed of 4805 rpm. As per compressor
to 2°reformer security, FSL-313 alarms were also map the surge flow value at this speed was
appearing but no action was taken as it was by- 1450 KSCFH .
passed. FE boardman opened FR-117 from 12 to
15% but compressor surging didn’t stopped. Gas Compressor i) stage temperature rise profiles ii)
turbine enters temperature control mode (cut Thrust Vibration trends & iii) Noise measure-
back) as its exhaust touches its limit (1000 °F) and ments in field all indicate that the surge initi-
alarm appears at Mark V. ated in HP case and the first stage to enter
surge was 3rd.. Discharge temperature profiles
(captured from TDC-3000) are attached as fig-
ure 3.

AMMONIA TECHNICAL MANUAL 323 2005

 TDC-3000 DCS has a scan time of >250 mil- (as there was no N2 in their natural gas while
liseconds, which is not suitable for detection gas at Daharki contained around 18% Nitro-
and control of compressor surge. (< 100 milli- gen).
second scan time controllers are required). So
it captured only 4-5 surge cycles, which actu- During commissioning time at daharki (1993)
ally were much more than that. lots of trips happened on this machine During
those, very high vibrations were observed at
FSL-313T, low process air flow security shut- HP case (>5.0 mils) and Gas turbine tripped at
down initiated and re-setted (without board- high exhaust temperature (over-load). At that
man action) 81 times. This security was by- time it was attributed to critical speed of HP
passed during startup. This indicates that 81 case etc. and HP case was not opened till
cycles of surge happened. (Note : this security 2004. No further investigation of Gas turbine
is configured on a much faster PLC system). overload was carried out.

Generally in case of centrifugal compressors When HP case was opened in April 2004
if we open the antisurge valve of a compressor turnaround for overhaul, it was found that the
in surge, it always comes out of surge. How- temporary strainer installed at suction of 3rd
ever this was not true in case of Process Air stage had broken and it had damaged the 3rd
compressor, K-2501. Even though the anti- stage rotor. As such high vibrations on HP
surge valve was opened from control room case have not been reported till the time of
and field verified yet the machine remained in commissioning in 1993, it is apprehended that
surge for around five minutes. 3rd stage got damaged at that time. See photos
given below. Another fact which supports
It was interesting to note that this was the only this finding is that there is a step change in ef-
compressor at our site for which we had ficiency of 3rd stage of the machine which re-
EIGHT different compressor maps (surge mained around 69% from 1993 till April 2004
curves) which were obtained from 1966 till and which changed to around 74.5% after in-
2003. Final evaluations were done by Dresser stallation of spare rotor.
Rand in 2003 and last set of curves were is-
sued in May 2003. Unlike standard practice on Centrifugal com-
pressors, no check valve was installed at the
In most of the curve sets, it is clear that 3rd discharge of this machine.
stage is more susceptible to surge than any
other case. However in last set obtained in CONCLUSION :
May 2003, 4th stage has been highlighted as
the most susceptible stage. Surge at Process Air Compressor occurred :

Surge control point for this machine was set at • Compressor surge line moved towards right
1525 KSCFH from commissioning time side after compressor overhaul. Different
(1967) till July 2003. Based on new map, the compressor maps which are being used since
flow point for surge control was revised to a 1967, have became invalid after rotor re-
more conservative value of 1625 KSCFH. placement of HP case.

Before compressor was brought to Daharki, • In Pascagola USA, the compressor was al-
Pakistan this machine always operated above ways operated above 2100 KSCFH flow. So it
2100 KSCFH flow at Pascagola, USA refinery ran much away from surge curve. So it never
surged there.

2005 324 AMMONIA TECHNICAL MANUAL

• We damaged 3rd stage of machine in 1993 machine exhibited excellent recovery be-
which changed its surge characteristics, and havior, during surge test (which can be at-
helped us operate it at low flow of 1500 to tributed to decrease in discharge volume
1700 KSCFH for next 10 years. After over- by installation of check valve and removal
haul of the machine the machine started surg- of dead-leg).
ing at 1700 KSCFH flow.
• A conservative surge protection curve
PROBLEM RESOLUTION : based on 3rd stage surge characteristics
was programmed in the controller, which
The following steps were taken for resolution of actuated 4th stage blow-off valve to pre-
the problem. vent machine from going into surge.
• A Check valve was installed at the dis-
charge of the compressor, in such a way After the above mentioned changes, the machine
that minimum volume remained between is being operated trouble free. To prevent this
compressor discharge flange, antisurge kind of problem the same design changes can be
valve flange and the check valve. (This re- applied while commissioning a new/old centrifu-
sulted in quick depressurization of ma- gal compressor.
chine in case antisurge valve opened).
AFTER EFFECTS OF THE SURGE INCIDENT
• All dead legs and extra piping were re- :
moved from the process air line. (for quick Background :
depressurization )
Secondary reformer, D-2504 catalyst was loaded
• Extra flow meter was installed along with during April 2004 turnaround. Loading was car-
additional pressure transmitters on 3rd ried out using sock loading method. After load-
stage of the machine.(In order to deter- ing, Top and bottom manhole below bee-hive(see
mine the surge characteristics of 3rd stage). picture below) of D-2504 remained open for next
two three days. Top nozzle was installed then and
• Dresser Rand was contacted and on their bottom man hole closed after that. No damage to
proposal, Inlet Guide Vanes, IGVs of HP bee-hive or fallen bricks or pieces of balls or
stage of the machine were changed. This white powder was observed at that time.
was done to create additional surge mar-
gin and improve efficiency.

• Compressor Control Corporation (ccc)

was contacted and their controller was in-
stalled on existing 4th stage of the ma-
chine. A surge test was carried out in
April 2005 using both 4th stage and 3rd
stage flow meters and it confirmed that 3rd
stage of the machine is surging before 4th
stage. However due to change of IGVs,
the machine now surged at much lower During the surge event, air flow and pressure kept
flow than before (around 1450 KSCFH) at on fluctuating inside Secondary reformer, D-2504.
same conditions and around 18% margin Air flow increased from 1700 KSCFH to beyond
was created. Also it was observed that 2450 KSCFH , became 200 KSCFH and then the
cycle was repeated. As per ESD log of FSL-313,

AMMONIA TECHNICAL MANUAL 325 2005

low flow at secondary reformer occurred 81 • As there seemed to be no other abnormal-
times which is consistent with observations in ity, it was decided to unload the catalyst in
field. order to find the problem source.

This kind of flow fluctuations caused flameout in • While unloading, top 1/3 of catalyst was
secondary reformer, which in turn resulted in air found in very good shape (no breakage, or
escape from the burner. dust) however after that, a cylindrical
segment of approx 3-4 ft diameter and 4 ft
Observations : height, just above Bee-Hive was discov-
ered in very bad shape, totally crushed and
• Outlet Temperatures of waste heat boilers, powdered. Except for that portion, all re-
E-2501A/B downstream Secondary Re- maining area contained normal catalyst.
former started increasing at a very very
high rate (upto 1°F/hr) after restart of • Going further down, it was discovered that
Ammonia plant, however after this initial the alumina balls at the top of the bee-hive
high rate, the outlet temp rise trend slowed have totally crushed and are in powder
down. Detailed investigation was car- form (Max quantity : <0.5%). Some of
ried.out the 2” balls which were 1 ft below 1” balls
have moved above them in bee-hive cen-
• It was apprehended that Alumina and re- ter. As noted by inspection, all rows of
fractory material was causing the fouling, bee-hive (aprox 5 ft height) have broken
however it was not easy to point out the their edges. This shows that major
source of this material. shakeup of bed has taken place.
• Ammonia plant was shutdown on 8th May • Bee hive roof center and top row of bricks
2004 and inspection + blowing of these have been broken, some of the alumina
exchangers was carried out. Top surface balls were embeded in this structure. Quite
of catalyst and burner was inspected but a few balls (<0.25% of total) became blue
no abnormality was found. Plant was re- from bright white original color.
started on 13th May 2004 but the fouling
phenomenon continued after the startup. Conclusion :
• Ammonia unit was again stopped on 30th These observations were then analyzed along
May 2004 in order to find the source of with Johnson Matthey representative and it
this refractory material. Catalyst and sup- was concluded that either localized explosions
port material vendor Johnson Matthey rep. have accrued inside the bed or the bed has
was also present at this time. been damaged by 60 to 80 psig pressure ham-
mering.
• Secondary reformer burner was removed
and top surface was examined, No abnor- As due to shock wave, gas could not escape
mality found. No refractory damage in from bee-hive top, it crushed the catalyst and
transfer lines was observed. However on balls present there.
opening the bottom manhole of the re-
former, water was found filled in side it. Also due to low pressure zone formed near bee-
Also inspection representative found some hive top, the balls moved in reverse direction
pieces of broken Bottom Bee-Hive bricks around bee-hive causing damage to edges of its
fallen in water. walls and moving 2” balls above

2005 326 AMMONIA TECHNICAL MANUAL

Problem resolution :

Damaged Bee-hive was repaired, additional balls

were filled in the bee-hive portion and the catalyst
was dedusted and loaded again.

Author(s) Profile :

Muhammad Majid Latif is Section

Head Process Engineering Am-
monia, at Engro Chemical Paki-
stan Ltd. Telephone: 92 723
641001, Email : mmlatif @ engro
.com. .He is responsible for new
projects, trouble shooting and optimization of
Ammonia unit. He has a varied background hav-
ing worked at IBM, FFC , CocaCola Export Corp
and EngroVoPak terminal Ltd. before joining En-
gro Chemicals Daharki. His areas of specializa-
tion includes Process design, Process Modeling &
Simulation, CFD modeling and troubleshooting.
He holds a BE in Chemical Engineering from
University of Engineering & Technology, Lahore.

Authors’ Acknowledgement

I would like to acknowledge the support provided

by Dresser Rand and Compressor Controls Corpo-
ration (CCC) in the successful completion of this
project. I would also like to thank the I&E, opera-
tions and maintenance teams of Engro Chemical
Pakistan, Plant in implementing the project from
concept to completion.

AMMONIA TECHNICAL MANUAL 327 2005

2005
Stage Temperature profiles during
K-2501 Surge incident on 17/4/04
2750
380 Flow, KSCFH 3rd Stage Surged First
2500
1st Stage
370
2nd Stage 2250
3rd Stage
360 4th Stage 2000
350 1750

340 1500

328
330 1250

320 1000

Temperature °F
750
310
500
300
250
290
0
-2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Time (20 sec intervals)

AMMONIA TECHNICAL MANUAL

 History of Surge Curves for Process Air Compressor, K-2501
(Only 5 out of eight are shown as other 3 are very similar to 2003 and 1996 cases)
2100
Safe Flow Above Curve

2000
1966 Clark Bothers Curve
(Canceled later)

1900

AMMONIA TECHNICAL MANUAL

Approximate Actual surge Locations
on 17th April and 28th June
1996 'Current case' curve
1800

2003 Dresser Rand final Curve

329
1700
1997 'Revamp case' Curve

1600

1500

(Please note that this includes 10% margin)

Flow at which machine will surge, KSCFH
1986 Dresser Clark Curve

1400
4650 4700 4750 4800 4850 4900 4950 5000 5050 5100

MML/-
Gas Turbine Speed, rpm

2005
2/7/04
AMMONIA TECHNICAL MANUAL 330 2005
2005 331 AMMONIA TECHNICAL MANUAL
 Very First Batch of catalyst
just after unloading start.
Note excellent condition of
catalyst with no dust.

Very First Batch of catalyst

in tarpaulin lined trolley.
Excellent condition, No
dust.

Start of Damaged balls,

just above Bee Hive. Note
that 2” Balls which were 1
ft below 1” balls have
come to top

2005 332 AMMONIA TECHNICAL MANUAL

 Small and large refractory
pieces found in catalyst
bed. Two sacks of these
were found mixed in
center of catalyst.

Healthy Bee Hive, Just

before loading in April
2004

Water filled in bottom

portion of secondary
Reformer. Note high level
of debris and Broken Bee-
Hive Bricks (indicated by
Arrows) identified by Mr.
Habib Pervez

AMMONIA TECHNICAL MANUAL 333 2005

 Damaged Top portion of
Bee- Hive

Damaged Top portion,

other side. Please note
how the balls have
crushed into powder and
also Embedded in Bee-
hive top.

Damaged Edges of the

Bee-hive due to upward
movement of Alumina
balls during pressure
swings.

2005 334 AMMONIA TECHNICAL MANUAL