Compressor

Compressor control menu
Main Menu Help Fwd Previous Rew

Cost of operating turbo compressors
Compressor operation
Maintenance cost
Operating cost
Commissioning cost
Putting it in perspective
Centrifugal compressors
Axial compressors
Compressor system classifications
Developing the compressor curve
The surge phenomena
Compressor control
Acrobat
Repairs are expensive
$50,000
$25,000
$750,000
Costs of repairs - materials and labor
3,000 hp
Process gas compressor
20,000 hp
Axial air blower
Seals
Bearings
Rotor Assembly
$20,000
$10,000
$200,000
Fwd Previous Rew Compressors Help Menu
Operating costs are large
Cost to operate one turbo compressor per year:

Plant air compressor 1,000 HP (746 kW) $457,000
Wet gas compressor 4,000 HP(2,984 kW) $1,830,000
Propylene refrigeration comp. 40,000 HP(29,480 kW) $18,300,000

Assumes power at $.07 per kilowatt hour or $457 per horsepower per year.
Energy costs vary due to local conditions.

Energy Saving Examples Fwd Previous Rew Energy Savings Predictions Compressors Help Menu
Energy savings examples
resulting from reduced recycle or blow-off
Actual Energy Savings Result From
Improved Antisurge Protection and Capacity Control

$1,200,000
$155,000
$78,000
Compressor
shaft power
Actual achieved
savings
Propylene refrigeration
FCCU air blower
Centac air compressor
40,000 hp ( 29 MW)
15,000 hp (11.2 MW)
1,500 hp (1.1 MW)
Compressor
application
Fwd Previous Rew Energy Savings Predictions Compressors Menu Help
Available energy savings can be predicted
Less than one year pay-backs* typical by reducing recycle of blow-Off
Pay-back
Period
(Months)
Reduced Recycle
(Per Cent of Maximum Compressor Flow)
*Assumes electro motor power At $0.05 US per kilowatt hour or turbine power at $327
per horsepower per year. Tax consequences are not considered in pay-back period
due to varying tax policies around the world.
12
11
10
9
8
7
6
5
4
3
2
1
0 5 10 15 20 25 30 35
Pay-back approximately 1
Month with 15% Reduction
1000 HP
3,500 HP
20,000 HP
Energy Saving Examples Fwd Previous Rew
Pay-back less than 10
Months with 15% Reduction
Pay-back less than 6
Months with 15% Reduction
Compressors Menu Help
Downtime costs can be enormous!
60,000 BPD Cat Cracker: $90,000 per hour, lost sales plus
fixed expenses. The biggest units are twice this size!
Natural Gas Production, 100 MMSCFD: $12,500 per hour
lost sales plus fixed expenses.
Consequences of downtime: Lost profit, lost customer
goodwill, repair costs, attention of top management.
Fwd Previous Rew Compressors Menu Help
Includes lost sales plus fixed operating expenses.
Most turbo compressor control system design problems are
discovered during commissioning.
Delays due to turbomachinery control problems are not unusual.
Startups are faster with a properly designed turbomachinery
control system.
Commissioning costs are large
Fwd Previous Rew
$2,300,000
100 MMSCFD
Natural Gas Plant
60,000 BPD
Cat Cracker
Start-up Cost Per Day $375,000
Process
30-year life cycle costs for a 20,000 hp compressor
Energy Costs
$180 Million
Initial Cost
$1.5 Million
Maintenance Costs
$4.5 Million
97%
of total costs
Source: Experiences in Analysis and Monitoring Compressor Performance
Ben Duggan & Steve Locke
E.I. du Pont, Old Hickory, Tennessee
24th Turbomachinery Symposium
Costs in constant dollars
30-year cost per 1,000 hp
?
What can we control?
0.0
5.0
10.0
15.0
Initial Cost Maintenance Energy Lost
Production
$ Millions
Controllable
Uncontrollable
Source: Experiences in Analysis and Monitoring Compressor Performance
Ben Duggan & Steve Locke, E.I. du Pont, Old Hickory, Tennessee
24th Turbomachinery Symposium
Costs in constant dollars
Centrifugal compressors
Widespread use, many applications
Gas is accelerated outwards by rotating impeller
Can be built for operation as low as 5 psi, or operation
as high as 8,000 psi (35 kPa or 55,000 kPa)
Sizes range from 300 hp to 50,000 hp
Single Case Compressor Centrifugal Impeller
DIFFUSERS
IMPELLERS
Cross Section of Horizontal Split
Fwd Previous Rew
Picture of Horizontal Split
Cross Section of Barrel Type
Picture of Barrel Type
Cross Section of Integrally Geared
Picture of Gear and Impellers
Picture of Integrally Geared
Cross section of horizontal split
Picture of Horizontal Split Fwd Previous Rew Centrifugals
Compressor inlet nozzle
Thrust bearing
Journal bearing
Shaft and labyrinth seal
Impeller inlet labyrinth seals
Discharge volutes
Impellers
Drive coupling
Casing
(horizontally split flange)
Compressor discharge nozzle
Menu Help
Picture of horizontal split
Fwd Previous Rew Centrifugals Cross Section of Horizontal Split Menu Help
Cross section of barrel type compressor
Fwd Previous Rew Centrifugals Picture of Barrel Type Menu Help
Picture of barrel type compressor
Fwd Previous Rew Centrifugals Cross Section of Barrel Type Menu Help
Cross section of bull gear compressor
Fwd Previous Rew Centrifugals Picture of Integrally Geared Picture of Gear and Impellers
Compressor volutes
Gear casing
Pinion shafts
Journal bearing
Impellers
Drive coupling
Labyrinth seals
Main gear
Inlet guide vanes
Menu Help
Picture of bull gear compressor
Fwd Previous Rew Centrifugals Cross Section of Integrally Geared Picture of Gear and Impellers Menu Help
Picture of (bull) gear and impellers
Fwd Previous Rew Centrifugals Cross Section of Integrally Geared Picture of Integrally Geared Menu Help
Gas flows in direction of rotating shaft
Can be built for lower pressures only
10 to 100 psi (0.7 to 6.8 Bar)
High flow rate
Efficient
Not as common as centrifugals
Axial compressors
Stator Blades
Rotor
Blades
Casing
Rotor Blades
Stator
Blades
Casing
Shaft
Fwd Previous Rew Cross Section of Axial Picture of Axial Compressors Menu Help
Cross section of axial compressor
Fwd Previous Rew Axials Picture of Axial
Compressor outlet nozzle
Rotor blades
Labyrinth seals
Guide-vane actuator linkage
Compressor rotor
Compressor inlet nozzle
Thrust bearing
Adjustable guide vanes
Menu Help
Picture of axial compressor
Fwd Previous Rew Axials Cross Section of Axial Menu Help
Compressor system classifications
Single-Section, Three-Stage Single-Case, Two-Section, Six-Stage
Two-Case, Two-Section, Six-Stage
Series Network
Fwd Previous Rew
Parallel Network
P
d
Discharge Pressure (P
2
) AP
c
Differential Pressure

(P
d
- P
s
) or (P
2
- P
1
) R
c
Pressure Ratio (P
d
/P
s
) or (P
2
/P
1
) H
p
Polytropic Head R
c
Q
s, normal
Q
s, mass
Q
s, vol
Compressor curve
for a specific
speed N1
R
process,1
Q
1
R
c1
Fwd Previous Rew
R
process,2
Q
2
R
c2
Q
s, vol
R
c
minimum speed
maximum speed
surge limit
stonewall or
choke limit
power limit
process limit
Fwd Previous Rew
stable zone
of operation
adding control
margins
Actual available
operating zone
How an airplane wing develops lift
Fwd Previous Rew
Bernoullis law
p
static
+
1
/
2
v
2
+ gH = Constant
The term gH is negligible for the wing
Then: p
static
+
1
/
2
v
2
= Constant
Due to the shape of the
wing: v
2
< v
1
thus p
2
> p
1
v
1
,
p
1
v
2,
p
2
As a result there is Ap or
lift
Lift
And the plane can
fly
How the airplane develops stall
Fwd Previous Rew
Lift
As the wing tilts back the Av changes and thus the Ap
This leads to more lift
Lift
Lift
When the wing is tilted too much the streaming profile
suddenly changes from laminar to turbulent
Lift
The air no longer sticks to the wing and the lift is lost
The plane starts to fall down
Developing the surge cycle on the compressor curve
Fwd Previous Rew
Q
s, vol
P
d
Machine shutdown
no flow, no pressure
Electro motor is started
Machine accelerates to nominal
speed
Compressor reaches performance
curve
Note: Flow goes up faster because
pressure is the integral of flow
Pressure builds
Resistance goes up
Compressor rides the curve
P
d
= P
v
+ R
losses
A
Compressor reaches surge point A
Compressor looses its ability to make
pressure
Suddenly P
d
drops and thus P
v
> P
d
Plane goes to stall - Compressor surges
B
Because P
v
> P
d
the flow reverses
Compressor operating point goes to point B
C
Result of flow reversal is that pressure goes
down
Pressure goes down => less negative flow
Operating point goes to point C
D
System pressure is going down
Compressor is again able to overcome P
v

Compressor jumps back to
performance curve and goes to point D
Forward flow is re-established
P
d
P
v
R
losses
P
d
= Compressor discharge pressure
P
v
= Vessel pressure
R
losses
= Resistance losses over pipe
Compressor starts to build pressure
Compressor rides curve towards surge
Point A is reached
The surge cycle is complete
From A to B 20 - 50 ms Drop into surge
From C to D 20 - 120 ms Jump out of surge
A-B-C-D-A 0.3 - 3 seconds Surge cycle
Rapid flow oscillations
Thrust reversals
Potential damage
Rapid pressure oscillations
with process instability
Rising temperatures inside
compressor
Major process parameters during surge
FLOW
PRESSURE
TIME (sec.)
1 2 3
TEMPERATURE
Fwd Previous Rew
TIME (sec.)
1 2 3
TIME (sec.)
1 2 3
Surge description
Flow reverses in 20 to 50 milliseconds
Surge cycles at a rate of 0.3 s to 3 s per cycle
Compressor vibrates
Temperature rises
Whooshing noise
Trips may occur
Conventional instruments and human operators
may fail to recognize surge
Some surge consequences
Unstable flow and pressure
Damage in sequence with increasing severity
to seals, bearings, impellers, shaft
Increased seal clearances and leakage
Lower energy efficiency
Reduced compressor life
Factors leading to onset of surge
Startup
Shutdown
Operation at reduced throughput
Operation at heavy throughput with:
- Trips - Power loss
- Operator errors - Process upsets
- Load changes - Gas composition changes
- Cooler problems - Filter or strainer problems
- Driver problems
Surge is not limited to times of reduced throughput. Surge
can occur at full operation
Compressor control
Fwd Previous Rew
Objectives
Antisurge control
Performance control
Other topics
Loadsharing for parallel compressors Major challenges of compressor control
Location of the surge limit
High speed of approaching surge
Control loop interactions
Loadsharing of multiple compressors
Protection #1: PI control
Protection #2: Recycle Trip
Protection #3: Safety On
Output linearization
The tight shut-off line
Fall-back strategies
Compressor networks
Base loading parallel compressors
Equal flow division system
CCCs equidistant Loadsharing system
Limiting control
Pressure Override Control (POC)
Flow Measuring Devices (FMDs)
Antisurge control valve
Piping lay-out considerations
Dynamic simulation single compressor
Dynamic simulation parallel compressors
Basic antisurge control system
Influence of controller execution time
Menu Help
Major control system objectives
(user benefits)
1. Increase reliability of machinery and process
Prevent unnecessary process trips and
downtime
Minimize process disturbances
Prevent surge and surge damage
Simplify and automate startup and shutdown
2. Increase efficiency of machinery and process
Operate at lowest possible energy levels
Minimize antisurge recycle or blow-off
Minimize setpoint deviation
Maximize throughput using all available
horsepower
Optimize loadsharing of multiple units
Energy Saving Examples
Fwd Previous Rew Compressor control Menu Help
Calculating the distance between the Surge
Limit Line and the compressor operating point
The Ground Rule
The better we can measure the distance to surge, the closer we can
operate to it without taking risk
The Challenge
The Surge Limit Line (SLL) is not a fixed line in the most commonly used
coordinates. The SLL changes depending on the compressor inlet
conditions: T
s
, P
s
, MW, k
s

Conclusion
The antisurge controller must provide a distance to surge calculation that
is invariant of any change in inlet conditions
This will lead to safer control yet reducing the surge control margin which
means:
Bigger turndown range on the compressor
Reduced energy consumption during low load conditions
Typical compressor maps include: (Q
s
, H
p
), (Q
s
, R
c
), or (Q
s
, p
d
)
coordinates, where:
Commonly used (OEM provided) coordinate
systems of the compressor map
Q
s
= Suction flow and can be expressed as actual
or standard volumetric flow
H
p
= Polytropic Head
R
c
= Compressor Ratio (p
d
/ p
s
)
p
d
= Discharge pressure of the compressor
p
s
= Suction pressure of the compressor
k
s
= Exponent for isentropic compression
These maps are defined for (1) specific set of inlet conditions:
p
s
, T
s
, MW and k
s
The problem with commonly used (OEM provided)
coordinate systems of the compressor map
These coordinates are NOT invariant to suction conditions as shown
Fwd Previous Rew
For control purposes we want the SLL to be presented by a single
curve for a fixed geometry compressor
Compressor control Menu Help
Developing invariant coordinates
The following variables are used to design and to characterize
compressors
Fwd Previous Rew
Fundamental variables
characterizing compressor
operation
H
p
= f
0
(Q, e, , , a, d, o)
J = f
1
(Q, e, , , a, d, o)
where:
H
p
= Polytropic head
J = Power
Q = Volumetric flow rate
- e = Rotational speed
- = Viscosity
- = Density
a = Local acoustic velocity
d

= Characteristic length
- o = Inlet guide vane angle

Through dimensional analysis (or similitude) we can derive two
sets of invariant coordinates
Dimensional analysis
or Similitude
Set 1
h
r

q
r

N
e

o
j
r

Re
Invariant coordinates
Set 2
R
c

q
r

N
e

o
j
r

Re
where:
h
r
= Reduced head
q
r
= Reduced flow
N
e
= Equivalent speed
- o = Guide vane angle
j
r
= Reduced power
Re

= Reynolds number
R
c
= Pressure Ratio

Coordinates (H
p
, Q
s
) and (h
r
, q
r
2
)
(H
p
, Q
s
)
NOT invariant coordinates
(h
r
, q
r
2
)
Fwd Previous Rew
where:
H
p
= Polytropic head
Q
s
= Volumetric suction flow
h
r
= Reduced head
q
r
2
= Reduced flow squared
Coordinates (R
c
, Q
s
) and (R
c
, q
r
2
)
Fwd Previous Rew
(R
c
, Q
s
)
NOT invariant coordinates
where:
R
c
= Pressure ratio
Q
s
= Volumetric suction flow
q
r
2
= Reduced flow squared
Compressor control Menu
(R
c
, q
r
2
)
q
r
2
Help
Coordinates (R
c
, j
r
) and (R
c
, N
e
2
)
Fwd Previous Rew
(R
c
, j
r
)
(R
c
, N
e
2
)
where:
R
c
= Pressure ratio
j
r
= Reduced power
N
e
2
= Equivalent speed squared
Representing the SLL as a single curve
using reduced coordinates
A coordinate system that is invariant to suction conditions is:
h
H
(ZRT)
r
p
s
= and
q
Q
ZRT
r
s
s
=
( )
Squaring the flow will still keep coordinates invariant:
h
H
(ZRT)
r
p
s
= and q
Q
ZRT
r
s
s
2
2
=
( )
Fwd Previous Rew
q
r
2
h
r
Design Nitrogen Off-gas
MW MW MW
P
s
P
s
P
s
T
s
T
s
T
s
k
s
k
s
k
s
Calculating q
r
2
(reduced flow squared)
Fwd Previous Rew
q
r
2
=
Q
s
2
(ZRT)
s
=
K
.
Z
s

.
R
u

.
T
s
MW
Ap
o,s
.
p
s
where:
R = R
u
/ MW
R
u
= Universal gas constant
R = Specific gas constant
MW = Molecular Weight of the gas
p
s
= Suction pressure
K = Orifice plate constant
- Ap
o,s
= Differential pressure across orifice plate
T
s
= Temperature of the gas in suction
Z
s
= Compressibility of gas in suction of compressor

(ZRT)
s
=
Ap
o,s
p
s
The antisurge controller calculates q
r
2
using p
s
and Ap
o,s
transmitters
Calculating h
r
(reduced head)
For polytropic compression R
t
= R
c
o
thus o =
R = R
u
/ MW
R
t
= T
d
/ T
s
Temperature ratio
R
c
= p
d
/ p
s
Pressure ratio
where:
R
u
= Universal gas constant
R = Specific gas constant
MW = Molecular Weight of the gas
p
d
= Discharge pressure
p
s
= Suction pressure
Z
s
= Suction compressibility
- o = Exponent for polytropic compression
log(R
t
)
log(R
c
)
The antisurge controller calculates h
r
using p
d
, p
s
, T
d
and T
s
transmitters
(ZRT)
s
h
r
=
Z
s

.
R
u

.
T
s
MW
R
c
o
-1
o
H
p
=
.
(ZRT)
s
=
R
c
o
-1
o
Building the Surge Limit Line
Any curvature of the Surge Limit Line can be characterized as a
function of the ordinate h
r
Fwd Previous Rew
q
r
2
h
r
The surge parameter is defined as:
S
f
q
s
r
r op
=
1
2
(h )
,
The function f
1
returns the value of q
r
2
on the SLL for input h
r
h
r
q
r,SLL
2
The surge parameter S
s
The function f
1
returns the value of q
r
on the SLL for input h
r
2
q
r
2
h
r
Fwd Previous Rew
h
r
q
r,SLL
2
As a result: S
s
=
q
r,op
2
q
r,SLL
2
q
r,op
2
OP
OP = Operating Point
S
s
< 1 : stable operating zone
S
s
< 1
S
s
= 1 : surge limit line (SLL)
S
s
> 1 : surge region
S
s
> 1
Introducing the distance between the operating
point and the Surge Control Line
Step 1: Introduce parameter d = 1 - S
s
q
r
2
h
r
S
s
< 1
S
s
> 1
S
s
= 1
Fwd Previous Rew
d

= 0
d

> 0
d

< 0
Step 2: Introduce parameter DEV = d - surge margin
DEV

= 0
Surge margin
DEV

> 0
DEV

< 0
The parameter DEV is independent of the size of the compressor
and will be the same for each compressor in the plant
Benefits:
One standard surge parameter
in the plant
No operator confusion:
DEV > 0 Good
DEV = 0 Recycle line
DEV < 0 Bad
Simplifying the surge parameter
by replacing h
r
with R
c
Reduced Head h
r
can be replaced by R
c
while keeping the
coordinate system invariant to suction conditions
Fwd Previous Rew
The surge parameter S
s
now becomes
S
s
=
f
1
(R
c
)
q
r,op
2
where the function f
1
( ) returns the value of q
r,SLL

on the SLL for the input R
c
2
This eliminates the need for T
d
and T
s
transmitters for control
Important Note: CCC still strongly recommends T
d
and T
s

transmitters as well as rotational speed N for compressor
monitoring purposes
The simplest CCC surge parameter
An antisurge algorithm can be designed around two transmitters:
Ap
o
and Ap
c
The parameter
Fwd Previous Rew
S
s
=
f
1
(R
c
)
q
r
2
is invariant to inlet conditions and speed
For two transmitters choose the function f
1
to be (R
c
- 1)
S
s
=
f
1
(R
c
)
q
r
2
=
R
c
- 1
Ap
o
p
s
p
d
p
s
- 1
( )
.
p
s
Ap
o
= =
p
d
- p
s
Ap
o
=
Ap
o
Ap
c
Selecting the specific function for f
1
(R
c
) to be (R
c
- 1) keeps the
map invariant to inlet conditions and speed
Disadvantage of the Ap
c
/Ap
o
surge parameter
The SLL is rarely a straight line in the coordinates q
r
2
and R
c

Fwd Previous Rew
q
r
2
R
c
Actual Surge Limit Line (SLL)
The parameter Ap
c
/Ap
o
represents a straight line in the invariant
coordinates q
r
2
and R
c
SLL calculated by antisurge
controller using
Ap
c
/Ap
o
= constant
The Ap
c
/Ap
o
approach results in loss of turn down and
unnecessary recycle
loss of operating envelope
Actual field data showing disadvantage of
Ap
c
/Ap
o
surge parameter
Surge parameter for compressor with sidestream
Problem definition
Ap
o,1

P
1

T
1
Fwd Previous Rew
Ap
o,2
P
2
T
2
1 2 3
q
3
and T
3
are internal to the compressor and cannot be measured
Derive a new surge parameter for compressor with
sidestream: q
r
N
e
Any combination of invariant parameters results in another
invariant parameter
Derive equation for surge parameter that does not require
measurement of T and q
r
at point 3
Step 1: Reduced flow
q
r
m
p
ZRT
=
.
.
where:
m = mass flow
Z = Compressibility
R = Gas constant
N
e
= Equivalent speed
q
r
= Reduced flow
N = Rotational speed
p = Pressure
T = Temperature
Step 2: Equivalent speed N
e
N
ZRT
=
Step 3: Combine q
r
and N
e
Fwd Previous Rew
q
r
N
e
m
p
ZRT
=
N
ZRT
.
.
m N
=
p
.
Calculating the invariant parameter q
r
N
e
q
3
Ap
o,1

p
1

T
1
Ap
o,2
p
2
T
2
Fwd Previous Rew
m
1
.
m
2
.
(m
1
+ )
.
N
.
m
2
.
p
2
=
m
3
.
N
.
p
3
= N
A p
p
T
A p
p
T
N
p
e,3
1 o,1
1
1
2 o,2
2
2
2
=
+
|
\
|
.
|
|
\
|
.
|
|
(
(
(
(
(
(
A A
1 2 3
r
Developing invariant surge patameter
R
c
vs. q
r
N
e
The approach to surge is fast
Typically, performance curves are
extremely flat near surge
Even small changes in compressor
pressure differential cause large
flow changes.
The speed of approaching surge is
high. In only 0.4 seconds, AP
O

dropped by 14%, with a 2% change
in AP
c

P
d
Q
s
A
D
100%
0
100%
0
AP
c
100%
P
d
0
1 SEC.
AP
o
Fwd Previous Rew
A
C
D
B
A
C
B
The approach to surge is fast - another example
For a 2% increase in pressure differential
(APc), flow rate AP
o
dropped 9% in 0.3 sec.
100%
0
0
AP
o
AP
c
100%
1 sec.
Surge parameter based on invariant
coordinates R
c
and q
r

Flow measured in suction (AP
o
)
P
s
and P
d
transmitters used to
calculate R
c
Basic antisurge control system

Fwd Previous Rew Compressor control
1
UIC
VSDS
Compressor
1
FT
1
P
s
T
1
P
d
T
The antisurge controller UIC-1 protects the compressor against surge by
opening the recycle valve
Discharge Suction
R
c
q
r
2
Opening of the recycle valve lowers the resistance felt by the compressor
R
process
R
process+valve
This takes the compressor away from surge
Menu Help
Antisurge controller operation
Protection #1: The Surge Control Line (SCL)

Fwd Previous Rew
A
R
c
B
When the operating point
crosses the SCL, PI control
will open the recycle valve
PI control will give adequate
protection for small
disturbances
PI control will give stable
control during steady state
recycle operation
Slow disturbance example
SLL = Surge Limit Line
SCL = Surge Control Line
q
r
2
Adaptive Gain
Enhancing the effectiveness of the PI controller

A
R
c
B
When the operating point
moves fast towards the SCL,
adaptive gain moves the SCL
towards the operating point.
This allows the PI controller
to react earlier
As a result a smaller steady
state surge control margin
can be achieved without
sacrificing reliability
Fast disturbance example
Fwd Previous Rew
q
r
2
Protection #2: The Recycle Trip
Line (RTL)
Disturbance arrives the Operating
Point (OP) moves towards the SCL
Fwd Previous Rew

R
c
q
r
2
RTL = Recycle Trip
Line
Output
to Valve
Time
When OP hits SCL the PI controller
opens valve based on proportional
and integral action
Operating point keeps moving
towards surge and hits Recycle
Trip Line (RTL)
When the operating point hits the
Recycle Trip Line (RTL) the
conclusion is:
We are close to surge
The PI controller is too slow to
catch the disturbance
Get out of the dangerous zone
An open loop response is
triggered
Operating point Moves back to the
safe side of the RTL
The RT function decays out the
step response
PI controller integrates to stabilize
OP on SCL
Recycle Trip
Response
PI Control Response
Total response of controller is the
sum of the PI control and Recycle
Trip action
PI Control
Recycle Trip

Action
+
To antisurge
valve
Total Response
Benefits:
Energy savings due to smaller
surge margin
Compressor has more turndown
before recycle or blow-off
Surge can be prevented for
virtually any disturbance
Improving the accuracy of Recycle Trip

open loop control
Recycle Trip
is the most powerful method known for antisurge

protection
But, open loop control lacks the accuracy needed to precisely
position the antisurge valve
Open loop corrections of a fixed magnitude (C
1
) are often either too
big or too small for a specific disturbance
The rate of change (derivative) of the compressor operating point
has been proven to be an excellent predictor of the strength of the
disturbance and the magnitude required from the Recycle Trip

response
Therefore, the magnitude of actual step (C) of the Recycle Trip
response is a function of the rate of change of the operating point
or d(S
s
)/dt
Recycle Trip
based on derivative of S
s

d(S
s
)
dt
C = C
1
T
d
Output
to valve
Time
where:
C = Actual step to the valve
C
1
= Constant - also defines maximum step
T
d
= Scaling constant
d(S
s
)/dt = Rate of change of the operating point
Medium disturbance
PI Control
Recycle Trip

Total
Large disturbance
Output
to valve
Time
PI Control
Recycle Trip

Total
Benefits
Maximum protection
No surge
No compressor damage
Minimum process disturbance
No process trips
Fwd Previous Rew
Recycle Trip

Response calculation
100%
0%
What if one Recycle Trip
step response is not enough?

After time delay C
2
controller checks if Operating Point is back to
safe side of Recycle Trip

Line (RTL)
If Yes: Exponential decay of Recycle Trip

response
If No: Another step is added to the Recycle Trip

response
Fwd Previous Rew
Output
to valve
Time
One step response
PI Control
Recycle Trip

Total
100%
0%
C
2
Multiple step response
Output
to valve
Time
PI Control
Recycle Trip

Total
C
2
C
2
C
2
Protection #3: The Safety On
Line (SOL)
If Operating Point crosses the Safety
On
Line the compressor is in surge

Fwd Previous Rew
R
c
q
r
2
RTL = Recycle Trip
Line
The Safety On
response shifts the

SCL and the RTL to the right
New SCL
New RTL
Additional safety or surge margin is
added
Additional surge margin
PI control and Recycle Trip
will
stabilize the machine on the new SCL
SOL = Safety On
Line
Compressor can surge due to:
Transmitter calibration shift
Sticky antisurge valve or actuator
Partially blocked antisurge valve or
recycle line
Unusual large process upset
Benefits of Safety On
response:
Continuous surging is avoided
Operators are alarmed about surge
Pressure and Flow Variations
During a Typical Surge Cycle

Built in surge detector
100%
100%
0%
0%
P
d
AP
o
20 to 50
milli-seconds
1 TO 2 SECONDS
Fwd Previous Rew
Surge signature is recorded during
commissioning
Rates of change for flow and pressure
during surge are determined
Thresholds are configured slightly more
conservative than the actual rates of
change during surge
Surge is detected when the actual rates of
change exceed the configured thresholds
The following methods can be used:
Rapid drops in flow and pressure
Rapid drop in flow or pressure
Rapid drop in flow only
Rapid drop in pressure only
When surge is detected a Safety On

response is triggered
A digital output can be triggered upon a
configurable number of surge cycles
Increase compressor system reliability and
availability with fall-back strategies
Over 75% of the problems are in the field and not in the controller
The CCC control system has fall-back strategies to handle these field
problems
The controller continuously monitors the validity of its inputs
If an input problem is detected the controller ignores this input and
automatically switches to a fall-back mode
Benefits
Avoids nuisance trips
Alarms operator of latent failures
Increases machine and process availability
Fall-back strategies for the antisurge and
performance controller
Antisurge controller
If a pressure transmitter fails, a minimum q
2
r
algorithm is used
If a temperature transmitter fails, h
r
is characterized as a function
of compression ratio
If the speed transmitter fails, a conservative speed setting is used
If the flow transmitter fails
Redundant transmitter is used
Output is driven to:
Last value OR
Last Value selected: If Last Value >Pre-selected fixed value OR
Pre-selected fixed value selected: If Pre-selected fixed value>Last
Value
Performance controller
Switches to redundant transmitter upon primary transmitter
failure
Output goes to pre-selected value if all transmitters fail or is
frozen
All transmitter failures are alarmed
Output linearization
Controller
output
Flow rate
through
valve
For antisurge control a linear valve
is preferred
Linear valve gives the same
dynamic flow response over its
complete stroke
Existing valve has equal percentage
trim
Valve trim
equal percentage
Controller output is characterized as
mirror image in the linear valve line
Controller output
Dynamic flow response becomes
linear
Existing valve has quick opening
trim
Valve trim
quick opening
Controller output is characterized
as mirror image in the linear valve
line
Controller output
Dynamic flow response becomes
linear
Notes
Used to improve controllers operation when non-linear valves are
used
Used on retrofits to avoid additional investment in new valve
Works well with equal percentage characteristics
Works less satisfactory with quick opening characteristics
The Tight Shut-off Line (TSL)
Controller
output
Flow rate
through
valve
Many antisurge valves have the
following characteristic:
from 0% to low clamp value the
flow rate through the valve is
(almost) zero and does not
change
Once the low clamp is reached
the characteristic is linear
Typical low clamp value can be
5% - we will use the 5% as the
value throughout in this example
0% to the valve
Low clamp on controller output
For dynamic control we want to use
the range 5% - 100% on the valve
Dynamic control range
The 5% or low clamp value represents
the closed position for control
purposes
At the low clamp value the valve
Usually still leaks which results
in energy waste
Makes an annoying noise
Typical for worn valves and valves
with Teflon seat
CCC antisurge controller has a Tight
Shut-off Line (TSL) that eliminates
the disadvantages
TSL = Tight Shut-off Line
When the operating point is to the
right of the TSL the controller closes
the valve at 0% - point A
This is below the low clamp value
A
A
When the controller crosses the TSL
the output of the controller jumps to
the low clamp value - point B
B
B
The controller is now ready to go
when the operating points hits the
SCL - point C
C
Fwd Previous Rew

R
c
q
r
2
SLL RTL
SCL
SOL
C
PI Control
Benefits
No leakage and noise when controller
is far away from surge - point A
Eliminates noise and energy waste
Eliminates dead time in the response
of the antisurge valve when the
operating point is close to the SCL
Time
Controller
output
Compressor performance control
Also called:
Throughput control
Capacity control
Process control
Matches the compressor throughput to the load
Can be based on controlling:
Discharge pressure
Suction pressure
Net flow to the user
Performance control by blow-off or recycle
Compressor operates in point A

P
d
q
r
2
Shaft
power
Fwd Previous Rew
q
r
2
Curve 1
A
R
process
+ R
valve
Required power in point A is P
1
Curve 1
P
1
Pressure is controlled by blow-off
PIC - SP
Point B represents the point that
would deliver the pressure for R
process
Curve 2
R
process
B
Required power in point B is P
2
Curve 2
P
2
PT
1
PIC
1
Process
Power loss is P
1
- P
2
Q
loss
represents energy waste
Q
loss
Notes
Most inefficient control method
Regularly found in plant air systems
Rare in other systems
Not recommended
Compressor control
Notes
Curve 2 represents:
Lower speed on variable speed
systems
IGVs closed on variable geometry
compressors
Inlet throttle valve closed on fixed
speed compressors
Menu Help
Performance control by discharge throttling

P
d
q
r
2
Shaft
power
Fwd Previous Rew
q
r
2
Curve 1
A
R
process
+ R
valve
Required power is P
1
Curve 1
P
1
Pressure is controlled by
pressure drop over valve
PIC - SP
Pressure loss
across valve
Opening of valve would reduce
resistance to R
process
R
process
Lower resistance would require
less speed and power
Curve 2
Curve 2
P
2
PT
1
PIC
1
Process
Power loss is P
1
- P
2
Notes
Extremely inefficient (consumes
approx. the same power for every load)
Rarely used
Not recommended
Compressor control
Notes
Curve 2 represents:
Lower speed on variable speed
systems
IGVs closed on variable geometry
compressors
Inlet throttle valve closed on fixed
speed compressors
Menu Help
Performance control by suction throttling
Inlet valve manipulates suction
pressure

P
d
q
r
2
Shaft
power
q
r
2
Changing suction pressure
generates a family of curves
Suction valve open
Suction valve throttled
Pressure is controlled by inlet
valve position
PIC - SP
for given R
process
A
R
process
Required power is P
1
P
1
PT
1
PIC
1
Process
Fwd Previous Rew
Notes
Common on electric motor machines
Much more efficient than discharge
throttling
Power consumed changes proportional
to the load
Throttle losses are across suction valve
Performance control by adjustable guide vanes
Change of guide vanes angle o
results in different compressor
geometry

P
d
q
r
2
Shaft
power
q
r
2
Different geometry means
different performance curve
o
min
o
OP
o
max
Pressure is controlled by inlet
guide vane position
PIC - SP
for given R
process
A
R
process
Required power is P
1
P
1
PT
1
PIC
1
Process
Fwd Previous Rew
Notes
Improved turndown
More efficient than suction throttling
Power consumed is proportional to the
load
Power loss on inlet throttling is
eliminated
Performance control by speed variation
Changing speed generates a
family of curves

P
d
q
r
2
Shaft
power
q
r
2
N
min
N
OP
N
max
Pressure is controlled by speed of
rotation
PIC - SP
for given R
process
A
R
process
Required power is P
1
P
1
PT
1
PIC
1
Process
Fwd Previous Rew
SIC
1
Notes
Most efficient (Power ~ f(N)
3
)
Steam turbine, gas turbine or variable
speed electric motor
Typically capital investment higher than
with other systems
No throttle losses
Limiting control to keep the machine in its stable
operating zone
While controlling one primary variable, constrain the
performance control on another variable
CONTROL BUT DO NOT EXCEED
Discharge Pressure
Max. Motor Current
Suction Pressure
Max. Discharge Pressure
Net Flow
Min. Suction Pressure
Suction Pressure Max. Discharge Temperature
Exceeding limits will lead to machine or process damage
Performance controller controls one variable and can limit
two other variables.
Power limiting in the performance controller
an example
Primary variable P
d
PIC-SP
Limiting variable Power
Power limit
Compressor operates in point
A for R
1
at N
1
N
1
A
R
1
Q
s, vol
R
c
Fwd Previous Rew
Process resistance changes
from R
1
to R
2
B
R
2
PIC will speed machine up to
N
2
in order to control pressure
P
d
N
2
Machine hits power limit
Compressor operates in point
B for R
2
at N
2
Process resistance decreases
further to R
3
R
3
PIC would like to speed
machine up to N
4
and operate
in point D
N
4
D
However power limiting loop
takes control and controls
machine at speed N
3

Compressor will operate in
point C for R
3
at N
3
N
3
C
Benefits
Maximum protection
No machinery damage
Maximize production
Machine can be pushed to the
limits without risk of damage
Note: Same approach for other variables (pressures, temperatures, etc.)
Limiting P
s
or P
d
using the antisurge controller
1
UIC
VSDS
Compressor
1
FT
1
P
s
T
1
P
d
T
The antisurge controller can be configured to limit:
Maximum discharge pressure (P
d
)
Minimum suction pressure (P
s
)
Both maximum P
d
and minimum P
s

This does NOT conflict with antisurge protection
Fwd Previous Rew
Discharge Suction
Interaction starts at B
Performance controller on
discharge pressure reduces
performance to bring
pressure back to setpoint
Unless prevented, PIC can
drive compressor to surge
Antisurge controller starts to
operate at B
Even if surge is avoided,
interaction degrades pressure
control accuracy
Results of interaction
Large pressure deviations
during disturbances
Increased risk of surge
Interacting antisurge and performance loops
A
C
AP
o
PIC-SP
R
c
P
s
Fwd Previous Rew
B
The performance controller interacts with
the antisurge controller
Both controllers manipulate the same variable - the
operating point of the compressor
The controllers have different and sometimes
conflicting objectives
The control action of each controller affects the other
This interaction starts at the surge control line - near
surge - and can cause surge
Ways to cope with antisurge and
performance loop interactions
De-tune the loops to minimize interaction. Result is
poor pressure control, large surge control margins and
poor surge protection
Put one loop on manual so interaction is not possible.
Operators will usually put the Antisurge Controller on
manual. Result - no surge protection and often partially
open antisurge valve
Decouple the interactions. Result - good performance
control accuracy, good surge protection and no energy
wasted on recycle or blow off
Interacting antisurge control loops
1
PIC
2
UIC
R
c,2
q
r,2
2
R
R
c,1
q
r,1
2
R
Disturbance comes from the
discharge side
P
d,2
increases
P
s,2
remains constant
R
c,2
increases
Section 2 moves towards surge
Disturbance
R
Antisurge controller UIC-2 will open
the recycle valve to protect section 2
against surge
P
d,2
decreases
P
s,2
increases
R
c,2
decreases
Section 2 moves away from surge
Opening of recycle valve on section 2
caused P
s,2
= P
d,1
to increase
Result:
P
d,1
increases
P
s,1
remains constant
R
c,1
increases
1
UIC
VSDS
Section 1 Section 2
R
Antisurge controller UIC-1 will open
the recycle valve to protect section 1
against surge
P
d,1
decreases
P
s,1
increases
R
c,1
decreases
Section 1 moves away from surge
Opening of recycle valve on section 1
caused P
d,1
= P
s,2
to decrease
Result:
P
s,2
decreases
P
d,2
remains constant
R
c,2
increases
The system is oscillating
Slowing down the controller tuning would lead to:
Increased risk of surge
Compressor damage
Process trips
Bigger surge margins
Energy waste
Menu Help
Loop decoupling between multiple antisurge controllers
1
PIC
2
UIC
1
UIC
VSDS
Section 1 Section 2
All CCC controllers are connected on a serial network
Serial
network
Serial
network
This allows them to coordinate their control actions
When UIC-2 opens the recycle valve:
Section 2 will be protected against surge
Section 1 will be driven towards surge
How much section 1 is driven towards surge depends on how much the
recycle valve on section 2 is opened
The output of UIC-2 is send to UIC-1 to inform UIC-1 about the
disturbance that is arriving
UIC-1 anticipates the disturbance by immediately opening its valve
Note: The same applies when the antisurge
valve on section 1 is opened first Menu Help
Loop decoupling simplified block diagram
FA
Mode
PI RT
Loop
Decoupling
Antisurge
Controller 1
Analog Inputs
DEV1
FA
Mode
PI RT
Antisurge
Controller 2
Analog Inputs
DEV2
2
UIC
1
UIC
VSDS
Section 1 Section 2
Serial
network
Antisurge controller UIC-2 opens its valve to protect
section 2 against surge
To antisurge valve 2
+
UIC-1 is protecting section 1 against surge using PI
and Recycle Trip

+
To antisurge valve 1
UIC-2 reports PI and Recycle Trip
output to UIC-1 Loop decoupling block multiplies reported PI and

Recycle Trip

values with decoupling gain M
2

PI
2

.
M
2
+
RT
2

.
M
2
Loop decoupling value is added to output to
antisurge valve 1
+
Loop decoupling values of other controllers
(performance and antisurge) are added to output to
antisurge valve 1
From other
controllers
PI
n

.
M
n
+
RT
n

.
M
n
Each controller has its own decoupling gain M
n
to
allow for tuning of relative loop gains between
different controllers
UIC-1 reports its PI and Recycle Trip values to UIC-2
Same decoupling takes place
Loop
Decoupling
PI
1

.
M
1
+
RT
1

.
M
1
PI
n

.
M
n
+
RT
n

.
M
n
+
From other
controllers
Benefits
Avoids control system oscillations
Allows faster tuning of control system
Reduced risk of surge
Allows closer operation to surge limits
without taking risk
Menu Help
Compressor networks
Compressors are often operated in parallel and, less
frequently, in series
The purposes of networks include:
Redundancy
Flexibility
Incremental capacity additions
Usually, each compressor is controlled, but the network is
ignored
Compressor manufacturers often focus on individual
machines
Control of the network is essential to achieve good surge
protection and good performance control of the network
Control system objectives
for compressors in parallel
Maintain the primary performance variable (pressure or flow)
Optimally divide the load between the compressors in the
network, while:
Minimizing risk of surge
Minimizing energy consumption
Minimizing disturbance of starting and stopping individual
compressors
Process Flow Diagram for base load control
Process
PIC
1
1
UIC
VSDS
Compressor 1
2
UIC
VSDS
Compressor 2
HIC
1
Suction
header
Fwd Previous Rew
Swing machine
Base machine
Notes
All controllers act independently
Transmitters are not shown
Parallel compressor control by base loading
R
c,1
q
r,1
2
R
c,2
q
r,2
2
Fwd Previous Rew
Compressor 1 Compressor 2
Machines operate at same R
c
since suction and discharge of both
machines are tied together
PIC-SP
Base load one or more compressors and let the other(s) absorb the
load swings
Swing machine Base machine
Base machine is fully loaded and runs without recycle
Q
C,2
=

Q
P,2
Swing machine can be running with recycle
Q
C,1
Q
P,1
where:
Q
P
= Flow to process
Q
C
= Total compressor flow

Q
C
- Q
P
= Recycle flow
Load could be re-divided to eliminate recycle
Q
P,1
Q
P,2
Q
P,1
+

Q
P,2
= Q
P,1
+

Q
P,2
Notes
Base loading is inefficient
Base loading increases the risk of surge since
compressor #1 will take the worst of any disturbance
Base loading requires frequent operator intervention
Base loading is NOT recommended
Process Flow Diagram for equal flow division
Process
PIC
1
1
UIC
VSDS
Compressor 1
VSDS
Compressor 2
Suction
header
Fwd Previous Rew
Notes
Performance controllers act
independent of antisurge control
Higher capital cost due to extra
Flow Measurement Devices (FMD)
Higher energy costs due to
permanent pressure loss across
FMDs
1
FIC
2
FIC
2
UIC
out
out
RSP
RSP
out
RSP
Parallel compressor control by equal flow division
R
c,1
q
r,1
2
R
c,2
q
r,2
2
Fwd Previous Rew
c
PIC-SP
Machine 2 operates with recycle while machine 1 still has turn
down
Q
P,1
Q
P,2
Q
C,2
Equal flow Equal flow
Q
P,1
=

Q
P,2
where:
Q
P
= Flow to process
Q
C
= Total compressor flow

Q
C
- Q
P
= Recycle flow
Equal flow division might work if both machines are identical Machines are never identical except by coincidence - different
resistance due to piping arrangments
Bias relay on remote setpoint would only work if curves have
same steepness
Notes
Requires additional capital investment in FMDs
Requires additional energy due to permanent pressure
loss across FMDs
Poor pressure control due to positive feedback in
control system (see next)
Equal flow division is NOT recommended
Dynamic control problem with pressure to flow
cascade system
Fwd Previous Rew
q
r
2
R
c
N
1
N
3
N
2
Pressure controller (PIC) provides
Remote SetPoint (RSP) for Flow
controller (FIC)
PIC
1
OUT
RSP
FIC
1
The FIC provides the RSP for the
speed controller, suction throttle
valve or guide vanes
OUT
RSP
SIC
1
The PIC is the master and the FIC is
the slave
Master Slave
In a typical master-slave control
scheme the slave needs to be
approx. 5 times faster than the
master
A
The machine is operating in point A
This is the intersection of 4 lines:
Resistance line R
1

Performance curve N
1

PIC-SP
FIC-SP = Output of PIC
R
1
PIC-SP
FIC-SP
Process disturbance causes the
resistance to change from R
1
to R
2
R
2
As a result the machine moves to
point B
B
Since the PIC is slow it does not
move its output yet which is the FIC-
SP
The FIC reacts fast and will try to
maintain its SP
The FIC will speed up the machine to
point C at speed N
3
C
The disturbance is amplified
Positive feedback system
Only as the PIC starts to reduce its
output to control pressure the FIC-
SP comes down and the pressure is
restored
D
Notes
Requires additional capital
investment in FMDs
Requires additional energy due to
permanent pressure loss across
FMDs
Poor pressure control due to positive
feedback in control system
Equal flow division is NOT
recommended
Process Flow Diagram for equidistant control for
parallel compressors
Process
1
UIC
VSDS
Compressor 1
VSDS
Compressor 2
Suction
header
Fwd Previous Rew
Notes
All controllers are coordinating
control responses via a serial
network
Minimizes recycle under all
operating conditions
1
LSIC
2
UIC
out
RSP
Serial
network
out
RSP
2
LSIC
1
MPIC
Serial
network
Serial
network
Parallel compressor control by equidistant operation
R
c,1
q
r,1
2
R
c,2
q
r,2
2
Fwd Previous Rew
c
PIC-SP
The DEV is a dimensionless number representing the distance between the
operating point and the Surge Control Line
Lines of equal DEV can be plotted on the performance curves as shown
.1
.2
.3
DEV = 0
.1
.2
.3
Machines are kept at the same relative distance to the Surge Control Line
(SCL)
This means in practice the same DEV for both machines
DEV
1
DEV
2
Recycle will only start when all machines are on their SCL
Since DEV is dimensionless all sorts of machines can be mixed: small,
big, axials, centrifugals
The DEV will be the same for all machines but they will operate at
different speeds and flow rates
Dev
1
= Dev
2

Q
1
=

Q
2
N
1
= N
2
Notes
Maximum turndown (energy savings) without recycle or blow-off
Minimizes the risk of surge since all machines absorb part of the
disturbance
Automatically adapts to different size machines
CCC patented algorithm
Compressors in parallel - the primary response
Master Controller
Loadsharing
Controller
Loop
Decoupling
FA
Mode
PI RT
Loop
Decoupling
+
Antisurge
Controller
Fwd Previous Rew
Analog Inputs
+
DEV
To antisurge valve To performance
control element
Master controller controls the main Process
Variable (PV) via its PID control block
PID
PV
SP
The output of the master controller PID goes
to the primary response block in the
loadsharing controller
Primary
response
In the primary response block the controller
checks if the machine is close to the SCL:
Yes: dont reduce capacity - keep output
constant
No: reduce capacity as necessary
Apply loadsharing gain M
0
The output of the master controller goes via
the primary response block directly to the
performance control element
DEV > 0
Dont change
output
x
Yes
No
Apply loadsharing
gain
To performance
control element
Primary response
In order to check if the machine is close to
the SCL the primary response block needs
the DEV
The DEV is reported by the antisurge
controller
DEV DEV
When the machine is close to the SCL the
master controller will no longer reduce
performance to control the primary variable
The master controller will start to open the
recycle valve to control the primary variable
Primary
response
If DEV <= 0 apply loadsharing gain
Output goes to antisurge valve
DEV < 0
Dont change
output
x
Yes
No
Apply loadsharing
gain
To antisurge valve
Primary response
The load balancing response
Master Controller
Loadsharing
Controller
Loop
Decoupling
FA
Mode
PI RT
Loop
Decoupling
+
Antisurge
Controller
Fwd Previous Rew
Analog Inputs
+
DEV
control element
The fast master controller controls the
primary process variable by directly
manipulating the final control elements
PID
In order to balance the machines they need
to be kept at the same DEV
The antisurge controller reports the actual
DEV to the load balancing block in the
This reported DEV becomes the Process
Variable (PV) for the load balancing PID
loop
Load
balancing
PV
PV
SP
Primary
response
The loadsharing controller reports this
DEV PV also to the master controller
DEV DEV
D
E
V

Other loadsharing controllers also report
their DEV PV to the master controller
DEV from other
loadsharing controllers
Primary
response
The master controller calculates the
average of all reported DEV PVs
Average
This average DEV is sent out to all
loadsharing controllers to become the SP
for all load balancing blocks
SP
The load balancing block is a slow
controller that will equalize all DEVs for all
parallel compressors
Its output is added to the total output to
the performance control element
The Pressure Override Control (POC) response
Master Controller
Loadsharing
Controller
Loop
Decoupling
Load
balancing
FA
Mode
PI RT
Loop
Decoupling
+
Antisurge
Controller
Fwd Previous Rew
Analog Inputs
Average
+
SP PV
DEV from other
DEV
D
E
V

control element
When a large disturbance occurs it can
happen that the performance control
element (e.g. speed) is too slow to keep the
pressure under control
PID
PV
SP
The operating point rides the curve and the
pressure rises sharply
Primary
response
There is a high chance to exceed the relief
valve setting and trip the process
The CCC master controller has a Pressure
Override Control (POC) mode that will open
the antisurge valve to get the disturbance
under control quickly
DEV DEV
POC-SP
PI
(One-Sided)
SP
PV
Opening of the antisurge valve is much
faster than a reduction in speed
As soon as the operating point drops under
the POC-SP line the antisurge valves start
to close again
Primary
response
The primary PID loop will stabilize the
operating point on the PIC-SP line
R
c
q
r
2
PIC-SP
Relief valve
setting
Benefits
Fast response during fast upsets
Avoid process trips due to lack of
response in performance control elements
Allows closer operation to process limits
without taking risk
Loadsharing for multi-section compressors
Process
1A
UIC
VSDS
Section 1
VSDS
Section 1
Suction
Header
Fwd Previous Rew
A
LSIC
out
RSP
Serial
network
RSP
B
LSIC
1
MPIC
Serial
network
Serial
network
Section 2
Section 2
2A
UIC
1B
UIC
1B
UIC Serial
network
Serial
network
out
Train B
Train A
How to operate equidistant from the Surge Control Line (SCL) when there is
more than one section per machine ???
Select per train -- in the loadsharing controller -- the section closest to the
SCL
By selecting the section closest to the SCL it is guaranteed that the other
section on the same train is not in recycle
Share the load -- equal DEVs for both trains -- on the section closest to the
SCL
Selecting the section closest to SCL for parallel operation
Master Controller
Loadsharing
Controller
Loop
Decoupling
Load
balancing
FA
Mode
PI RT
Loop
Decoupling
+
Antisurge
Controller
Fwd Previous Rew
Analog Inputs
Average
+
SP PV
DEV from other
DEV1
To antisurge valve-1 To performance
control element
PID
PV
SP
Primary
response
DEV1 DEV2
Both antisurge controllers
report their DEV to the
PI
(One-Sided)
SP
PV
Primary
response
FA
Mode
PI RT
Loop
Decoupling
+
Antisurge
Controller
DEV2
To antisurge valve-2
Primary
response
The lowest DEV is selected:
the section closest to the
SCL
<
The selected DEV is
reported to:
Primary control
response blocks
Load balancing block
Master controller
averaging block
Flow Measuring Device (FMD) selection criteria
Main selection criteria for FMD in antisurge control system:
Repeatability
Sufficient signal-to-noise ratio
Accuracy of the FMD is not critical
FMD delays must be absolutely minimal
Present state-of-the-art limits the choice of FMD to head flow meters
or to other devices that are based on the principle of velocity
measurement:
Orifice plates
Venturis
Pitot tubes
etc.
Recommended flow range for FMD and transmitter is maximum
compressor flow
Recommended Ap corresponding to Q
max, compressor
is 10 WC (250
mmH
2
O) or more
Flow Measuring Device (FMD) location
Fwd Previous Rew
The preferred location of the FMD:
Suction of compressor
As close to the inlet flange as
possible
VSDS
Compressor
Discharge Suction
minimum possible
Less preferable location of the FMD:
Discharge of compressor
As close to the discharge flange as
possible
minimum possible
Selection of the location should be based on:
Necessity of surge detection
Often more difficult with flow measured in discharge
Capital cost of flow measuring device
Operating cost of the FMD (permanent pressure loss)
Response time of the FMD transmitter
The speed of approaching surge is high
Fwd Previous Rew
100%
0
100%
0
APc
100%
Ps
0
1 SEC.
APo
A
C
D
B
A
C B
In only 400 ms, AP
O
dropped by 14%,
with a 2% change in AP
c
The transmitter type and brand should be
selected based on two major factors:
Reliability
Speed of response
Desired rise time for Ap (flow) transmitters
is 200 ms or less
Pressure step is 100%
The first order response (63%) is less
than 200 ms
Time
Actual
pressure
Transmitter
output
63% response
1- (1/e)
t
1
is less than 200 ms
Desired rise time for pressure transmitters
is 500 ms or less
The effect of damping the Ap
o
(flow) transmitter
Knowing the flow is essential to determine the distance between the
operating point and the SCL
Damping the Ap
o
(flow) transmitter destroys essential information
50
0
-50
0
1.25 2.50 3.75 5
Time (seconds)
Fwd Previous Rew
Flow
Start of Surge
Actual Flow
t = 16.0 s
t = 1.70 s
t = 0.20 s
t = 0.03 s
Damping the Ap
o
(flow) transmitter can paralyze the
complete antisurge control system!!!
Sizing the antisurge control valve
Criteria for antisurge valve sizing based on CCCs experience
Provide adequate antisurge protection for worst possible disturbances
Provide adequate antisurge protection in all operating regimes
Sized to provide flow peaks greater than what is required in steady state to
operate on the Surge Control Line
Sized to avoid choke zone
Not be oversized from controllability point of view
Fwd Previous Rew
Take point A at the intersection of the
maximum speed performance curve and
the Surge Limit Line (SLL)
Calculate C
v,calc
(or equivalent) for point A
Select standard valve size using the
following criteria:
1.8
.
C
v, calc
< C
v,selected
< 2.2
.
C
v, calc
R
c
Q
vol
A
Sizing the antisurge control valve - alternative method
Fwd Previous Rew
R
c
Q
vol
An alternative method yielding
excellent results is:
Take design point of the compressor
point A
A
Draw a horizontal line through the design
point
Take point B at intersection of maximum
speed performance curve and the
horizontal line
B
Calculate C
v,calc
in point B
Select standard valve size using the
following criteria:
0.9
.
C
v, calc
< C
v,selected
< 1.1
.
C
v, calc
Stroke speed and characteristic of the antisurge valve
Fwd Previous Rew
Antisurge valve stroke speed
Antisurge valve must have speed of response adequate for antisurge
protection for all disturbances
Recommended full stroke times:
Size Close to open Open to close
1 to 4 1 second < 3 seconds
6 to 12 2 seconds < 5 seconds
16 and up 3 seconds < 10 seconds
Closing time needs to be the same order of magnitude to assure the same
loop gain in both directions
Antisurge valve characteristic
Normally control valves are selected to be open 80% to 90% for design
conditions
Antisurge valves can operate anywhere between 0% and 100%
In order to have an equal loop-gain over the whole operating range a linear
valve is required
This will allow for the fastest tuning leading to smaller surge margins
Improving the performance of the antisurge valve
Fwd Previous Rew
Most normal control valves can be made to perform as required for
antisurge control
The following steps help improve the performance of the valve
Install positioner
Minimize tubing length between I/P and valve positioner
Install volume booster
Minimize volume and resistance between volume booster and actuator
Increase air supply line to
3
/
4
or more
Increase size of air connection into the actuator
Drill additional holes in actuator - avoids pulling a vacuum
Piping lay-out consideration when designing an
antisurge control system
Piping lay-out influences the controllability of the the total system
The primary objective of the antisurge controller is to protect the
compressor against surge
This is achieved by lowering the resistance the compressor is feeling
The resistance is lowered by opening the antisurge valve
Dead-time and time-lag in the system needs to be minimized
This is achieved by minimizing the volume between three flanges
Discharge flange of the compressor
Recycle valve flange
Check valve flange
Fwd Previous Rew
VSDS
Compressor 1
volume to be
minimized
Using a single antisurge valve increases recycle lag time
Fwd Previous Rew
Section 1 Section 2
In order to protect section 1 the antisurge valve needs to be opened
The volume between compressor discharge, check valve and
antisurge valve determines the dead time and lag time in the system
Large volume
Large volume significantly decreases the effectiveness of the
antisurge protection
Result
Poor surge protection
Large surge margins
Energy waste
Process trips because of surge
Note
This specific piping layout is found on
many wet gas compressors in FCCUs
Sharing recycle coolers degrades surge protection
Fwd Previous Rew
Section 1 Section 2
The piping lay-out for section 2 is excellent for surge protection
Minimum volume between the three flanges
Small volume
The piping lay-out for section 1 is not ideal
Large volume to be de-pressurized decreases ability of the control system to
protect the machine against surge
Result
Poor surge protection
Large surge margins
Energy waste
Process trips because of surge
Installing recycle valve upstream from
cooler improves control response
Fwd Previous Rew
Compressor 1
Compressor 1 has ideal piping lay-out for surge protection
Minimum volume between the three flanges
Compressor 2
Minimum
volume
The piping lay-out for compressor 2 is commonly found in the industry
The cooler creates additional volume and decreases the effectiveness of the
antisurge control system
Increased
volume due
to cooler
The piping lay-out for compressor 2 can be acceptable if the additional
volume does not create excessive dead time and lag in the system
Result
Increased surge margins
Energy waste
Recycle lines configured for optimum surge protection
Fwd Previous Rew
Compressor has ideal piping lay-out for surge protection
Minimum volume between the three flanges for all sections
Section 2 Section 3 Section 1
Minimum
volume
Process Suction
Which antisurge piping configuration do you choose???
Fwd Previous Rew
These two piping lay-outs are most common for antisurge control Lay-out #1 has minimum volume between the flanges and is the best lay-out
for antisurge control purposes
Process Suction
Suction Process
Lay-out #1: Compressor with recycle lines optimally configured for antisurge control
Lay-out #2: Compressor with coolers upstream of recycle take-off
Lay-out #2 requires one cooler less and thus the capital investment is lower When selecting lay-out #2 the residence time of the gas in the surge volume
should be verified to check acceptable time delays are not exceeded
Lay-out #2 will require bigger surge control margins
Influence of controller execution time
Analog controller
SLL
SCL
100%
0%
Controller
output
100%
0%
Leading engineering contractor
performed evaluation of execution
time influence on ability to protect
compressor from surge
Dynamic simulation of compressor
was built
Digital controllers are compared
against analog controller on
simulation
Analog controller has no execution
time and is immediate
Analog controller tuned for minimum
overshoot
Digital controllers get exact same
tuning parameters
Digital controllers get exact same
disturbance
Operating
point
Time
Time
Menu Help
Analog vs digital controller at 2 executions per second
Analog controller
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
Digital controller
(2 executions per second)
Time
Time
Time
Time
Compressor surged
Large process upset would have resulted
Tuning same as analog controller
Menu Help
Analog controller
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
Digital controller
Time
Time
Time
Time
Compressor surged
Large process upset would have resulted
Menu Help
Analog controller
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
Digital controller
Time
Time
Time
Time
Compressor almost surged
Control system would have to be set up with bigger surge margins
Menu Help
Analog vs CCC controller at 25 executions per second
Analog controller
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
SLL
SCL
100%
0%
Controller
output
100%
0%
Operating
point
CCC antisurge controller
Time
Time
Time
Time
Response of CCC controller nearly indentical to analog controller
Adding Recycle Trip
to PI control will allow even smaller surge margins

Menu Help
Dynamic simulation single compressor
1
UIC
VSDS
Compressor
1
FT
1
P
s
T
1
T
s
T
Process Suction
1
P
d
T
1
T
d
T
1
ST
1
PIC
1
HIC
Load
Note: Speed transmitter for
indicating purposes only
Start simulation
Compressor is controlled on P
d
by PIC-1
HIC-1 controls the process load and can be used to create process disturbances
Controllers communicate via serial communication to computer running the
simulation
Serial
network
MODBUS Menu Help
Dynamic simulation parallel compressors
Process
HIC
1
Load
1
UIC
VSDS
Compressor 1
VSDS
Compressor 2
1
LSIC
2
UIC
out
RSP
Serial
network
out
RSP
2
LSIC
1
MPIC
Serial
network
Serial
network
MODBUS Start simulation Menu Help

Compressor

Uploaded by

Copyright:

Available Formats

Compressor

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Compressor

Uploaded by

Copyright:

Available Formats

What are some of the major costs associated with operating compressors?

What factors influence the operating costs of compressors?

Compressor control menu

Main Menu Help Fwd Previous Rew

is the most powerful method known for antisurge

step response is not enough?

Line the compressor is in surge

response shifts the

output to UIC-1 Loop decoupling block multiplies reported PI and

to PI control will allow even smaller surge margins

You might also like