Gas Turbine Compressor System Design Using Dynamic Process Simulation PDF
Gas Turbine Compressor System Design Using Dynamic Process Simulation PDF
Gas Turbine Compressor System Design Using Dynamic Process Simulation PDF
ASME Turbo Expo 2007: Power for Land, Sea and Air
May 14-17, 2007, Montreal, Canada
GT2007-28270
Gas Turbine Compressor System Design Using Dynamic Process Simulation
Head Ft LB/LB
compressor surge during trip but rather minimize the number of 120000
shutdown events that could cause a surge on trip. A careful 100000
80000
evaluation should be made to determine if this control strategy
60000
could be utilized on a specific application. 40000
20000
Compressor Modeling 0
The modeling of a compressor may be as simple as 0 500 1000 1500 2000 2500 3000 3500
using classical fan laws defined by the normalized parameters ACFM
a function of flow divided by rotational speed (Q/N) and 100% RPM 90% RPM 80% RPM
70% RPM 60% RPM Fan Law 90% RPM
efficiency as a function of flow divided by rotational speed
Fan Law 80% RPM Fan Law 70% Fan Law 60%
(Q/N). This approach works well for single compressor stages
and some multistage arrangements. However, this can lead to
Figure 1 Comparison of fan laws to actual compressor flow
significant error in determining surge depending on the number
Map
of compression stages in series in the compressor body and the
resulting overall compressor performance map especially below
design speed.
Compressor Map
Predicted head and flow data for multiple stages in Overlay of Fan Law
series are derived from individual stage performance. The
combined performance prediction may be accurate over a very 180000
120000
more accurate compressor map must be generated at multiple 100000
speeds especially for start up and shutdown studies. During an 80000
ESD the shape of the surge curve is critical. Many systems 60000
model the surge limit as a linear relation between flow and 40000
20000
differential pressure developed across the compressor. The
0
surge curve for each compression system is unique and may 0 500 1000 1500 2000 2500 3000 3500
not be linear in many applications. Modeling the system must ACFM
try to be as accurate as possible. The same is required for Figure 2
overload operation. Many compressor systems are started in Surge Overload Fan Law Surge Fan law Overload
overload. As multiple sections of compression are part of many
processes the accuracy of the curves used modeling become Figure 2 Comparison of fan law estimation to actual
even more important. performance showing the shift of the overall compressor
The practice of normalizing the head flow curve from performance map
the Q/N vs. Head /N^2 relationship to represent the entire curve
from low to high speed will lead to modeling errors. As shown overload. The consequences of this could limit the equipment
in Figure 1, the performance prediction by the compressor life and process operating range.
manufacturer’s extended performance curves does not match Continuous operation in overload is not recommended
the extrapolated Q/N based on the normal design speed curve. by some compressor manufactures (3). While not as
As can be seen the Q/N speed curves have a much different detrimental as operating in surge, the mismatch between
shape as the speed decreases. The change in shape will cause components can cause damage to the internal parts of the
erroneous results for the dynamic simulation. Figure 2 shows compressor.
the overlay of the surge and overload lines. The Q/N predicted
overload at the end of the curve will be greater than is actually
available at lower speed based on the manufacturers data. The
Anti-Surge Valve Requirements
The proper protection of a compressor requires
surge points will also not be estimated correctly and could lead
analysis of steady state online requirements and transient
to inaccurate conclusions. The conclusions drawn from a
requirements of the valve capacity. Steady state requirements
system that is not modeled accurately can lead to either over
can be determined from compressor performance data.
design or under design of the compression system components.
Traditional sizing uses an over sizing factor that multiplies
Over design could restrict operating range, reduce efficiency,
surge flow by the sizing factor to determine a required Cv for
and provide poor control system response. Under design may
the anti-surge valve. This method does not accurately predict
lead to predicting safe operation in an area of the extrapolated
specific compressor requirements for surge protection during a
compressor performance map that is actually in surge or
high-speed transient event such as an emergency shutdown
30000
Compressor Map
105% RPM
Head Ft
25000
45000 100%
20000
40000 90%
15000
80%
35000 10000 70%
60%
30000 5000 50%
40%
105% RPM 0
Head Ft
25000
0 500 1000 1500 2000 2500 3000 3500 4000
100%
ACFM
20000
Figure 6
90%
15000
80% Figure 6 Compressor Flow Head map with operating path
10000 70% in response to an ESD event, no surge is encountered
60%
5000 50%
40%
0
Revised Valve Size
0 500 1000 1500 2000 2500 3000 3500 4000
ACFM 3000
Figure 4
1500
The total deceleration of the rotating components is shown in
Figure 8.
1000 The estimation of the required size of the valve to
500 avoid surge and prevent possible equipment damage during an
ESD event can be accurately predicted though the use of
0 dynamic simulation.
0 2 4 6 8 10
Time Sec
Figure 5 Example 2
Compressor Flow Surge Flow
Sudden discharge blockage with ASV system for pressure
Control
Figure 5 Flow vs. time map for the ESD event showing A Sudden blockage of the discharge side of a
surge limit line and compressor operating path during ESD compression system can be controlled by several methods. The
anti-surge system could open the recycle valve when it senses a
the event. A larger anti-surge valve capacity will be required to reduction of inlet flow. This causes the compressor operating
prevent this excursion into surge. An 80% larger ASV was point to move away from surge hence reducing discharge
selected after a number of iterations. This resulted in the pressure. The control system can also be configured with a
compressor staying out of the surge region during the discharge pressure override control. This system will sense a
8000
25000
Speed
ACFM
6000 20000
4000 15000
10000
2000
5000
0
0
0 5 10 15 20 25
Time Sec 0 10 20 30 40 50 60
Figure 8 Tim e S ec
Figure 10
315 25000
310 20000
305 15000
300 10000
0 10 20 30 40 50 60
5000
T ime Sec 0
Figure 9
0 5 10 15 20 25 30
Time Sec
Figure 9 Pressure vs. time response to a discharge blockage Figure 11
Compressor Flow Relief Valve Flow
small increase in pressure and recovery to set point in 10
seconds. Figure 10 shows the reduction in flow to a minimum Figure 11 Flow vs. time for undersized relief valve
value to maintain the discharge pressure for this event. While
the pressure recovery was quick (10 seconds) the time required The valve goes through a number of open close cycles due to
for the flow to stabilize took longer due to the system volume. the volume of the compressor discharge and the flow supplied
by the compressor. The flow through the relief valve, when
open, may allow the pressure in the discharge volume to
400
450
350
400
350 300
Pressure Psia
Pressure Psia
300 250
250 200
200
150
150
100 100
50 50
0 0
0 5 10 15 20 25 30 0 5 10 15 20 25 30
T ime Sec T ime Sec
Figure 12 Figure 14
Figure 12 Pressure vs. time for an undersized relief valve Figure 14 Pressure vs. time design relief valve.
decrease and thus the valve will close. This is purely a mass
imbalance relationship. The control limit is 360 psia however
the relief settle out was over 422 psia. The 422 psia was the Appendix
pressure where the compressor flow and the relief valve flow Simulation Requirements
matched. Figure 13 and 14 demonstrate the response of the 1. Process and Instrument Diagram
properly sized relief valve. Once full open the valve allowed 2. Process Flow Diagram
the pressure to reach 362 psia. 3. Gas analysis
4. Volumes of piping , coolers, and vessels
Mass Flow
5. Control valve design including Cv, and stroke time
Design Relief Valve 6. Block valve design including Cv and stroke time
7. Compressor map for flow, pressure , temperature and
35000 speed
30000
8. Design operating point for compressor system
9. Driver characteristics, power, speed, rate of change, startup
Mass Flow Lb/sec
25000
speed torque data
20000 10. Inertia of the rotating shaft system
15000 11. Specific conditions at process boundaries
10000
12. Definition of relief valves
5000
0
0 5 10 15 20 25 30
Time Sec
ACKNOWLEDGMENTS
Figure 13 The authors thank Dresser-Rand Company for providing
the opportunity to present this document. Also Quality
Compressor Flow Relief Valve
Monitoring and Control Company for providing training and
insight into dynamic process simulation..
Figure 13 Flow vs. time design relief valve.
REFERENCES
1. Quality Monitoring and Control Corporation; DYFLO
Conclusion Users Manual
Dynamic process simulation is a valuable tool that can 2. Franks. Roger G.E.; Modeling and Simulation in
be used to validate process design to ensure trouble free Chemical Engineering, R. , John Wiley & Sons, 1972
operation from initial start up. It can also assist in the 3. Sorokes, J.; Miller, H.; Koch J.; The Consequences Of
resolution of existing plant operational problems. This tool has Compressor Operation in Overload, Texas A&M
evolved from a complex computer program, involving many Turbomachinery Conference, September 2006
hours to set up and run, to the point where it can now be 4. McKee, R.J.; Blodgett, L.L.; Transient Gas Flow &
effectively used as part of initial design of petrochemical Control System Modeling for the Design of
processes. Compressor Surge Control Systems for Trips &
Upsets, Gas Machinery Resource Council 2004