A. Zeckendorf, J.W. Altena
Fluor Daniel B.V.
Haarlem, The Netherlands
A combination of design and dynamic simulation early in a project to design a 38-mw gas-injection compressor for storage resulted in an anti-surge and recycle layout that avoided surge and reduced costs.
In February 1991, Nederlandse Aardolie Maatschappij (NAM) requested Fluor Daniel BY to participate in a study team for the NAM underground gas-storage injection facilities.
For this project, an existing gas field in the northeast Netherlands was to be used. For injecting surplus natural gas, a gas-injection compressor station with total installed power of 76 mw was required.
To obtain buffer gas (cushion gas), the initial discharge pressure of the compressor station during fill-up is approximately 150 barg. After fill-up, the compressor discharge pressure would vary between 230 barg at the beginning of the injection period (in spring) and 290 barg at the end of the injection period (in autumn).
The injected gas would be withdrawn from storage during periods of high demand, that is, at low ambient temperature in the winter.
TEAM'S CONSIDERATIONS
The task of the study team was to select, based on economic, technical, operational, and environmental conditions, the optimum number, size, and type of drivers for centrifugal gas-injection compressors with two process stages.
The driver types reviewed were simple cycle gas turbines, fixed-speed electric motors (with and without hydraulic torque convertor for speed variation), and thyristor controlled variable-speed electric motors.
Because the storage facilities would be installed in an area preserved for a natural park, every precaution was taken to limit impact on this area.
Special attention was paid to the following aspects:
- The driver-compressor combinations shall be suitable for the maximum expected annual 250 starts and stops and a total life of 25 years.
Moreover, the driver-compressor combinations shall be capable of frequently running on partial load.
- Noise emissions shall be held extremely low.
- Possible NOx, CO, and CO2 emissions must be acceptable.
- The height of the installation must be kept to a minimum in order to limit the so called "horizon pollution."
- Emission of hydrocarbons to atmosphere shall be minimal. Therefore, the installation must be protected against overpressure by a high integrity pressure-protection system (hipps) instead of a large relief valve connected to a flare system.
Two rapidly closing (< 2 sec) hipps valves in the compressor discharge line substitute for a relief valve. The hipps control is hard wired and completely separated from the control and safeguarding system.
In case the compressor control and safeguarding system cannot prevent the pressure from rising, the hipps' pressure sensors in the compressor's first and second-stage suctions reach their setpoints.
The hipps' valves are triggered and will close so rapidly that the settle-out pressure will fall within the design pressure of the system.
A detailed study, part of the subject of this article, led to the selection of two parallel operating 38-mw electrical thyristor controlled variable-speed synchronous electric motors (VSDS) as drivers for the gas-injection compressors.
Speed variation is also a simple and effective method of controlling the capacity of the compressor and thus saving on energy consumed.
The technical feasibility of the 38-mw VSDS-gas injection compressors was verified in detail. Each compressor is rated to compress 12 million standard cu n-t/day (MMscmd; 420 MMscfd) from an inlet pressure of 63 bara and an inlet temperature of 15 C. to a discharge pressure of 291 bara and a discharge temperature of minimum 80 C.
A typical performance map is shown in Fig. 1 (6828 bytes).
Part of the study was the review of the antisurge and recycle system. The details of that review are outlined presently, followed by a discussion of the dynamic simulation study.
ANTISURGE, RECYCLE CONTROL
Initially, the compressor comprising two process stages was provided with an interstage cooler and an attercooler.
Each stage was provided with its own combined antisurge/recycle control loop. The recycle gas was cooled by the intercooler for the first stage and by the aftercooler for the second stage.
Because the intercooler and aftercooler have large volumes, the combined antisurge/recycle loop was considered too slow to avoid surging. For this reason, fast acting hot gas antisurge loops were added (Fig. 2) (13442 bytes).
In this configuration, the hot gas antisurge valves are used to protect each individual compressor stage from surge due to sudden or short-term disturbances such as sudden closure of the hipps valves.
If constant recycling becomes necessary, the gas is routed through the cold recycle line, which is equipped with a heat exchanger, to cool the recycle gas. Thus, heat build-up, so called "temperature runaway, can be avoided without compromising surge protection.
In the course of the study, the aftercooler was deleted because a minimum gas-injection temperature of 80 C. is required to prevent cooling down of the reservoir. The consequence was that the antisurge recycle system, as proposed in Fig. 2 (13442 bytes), could not be applied.
In order to avoid a recycle cooler on the second-stage recycle loop, the study team proposed the schematic referred to as "Configuration C" (Fig. 3c) (9953 bytes) in the discussion that follows.
The intercooler is now used for interstage cooling, both in the first and second-stage recycle loop. It is well understood that, with a common cooler for both recycle loops, the loops are no longer independent. This interaction complicates the system.
Verification with several compressor manufacturers and suppliers of compressor-control systems revealed that actual experience with the proposed system is limited and in combination with hipps protection not existing.
Therefore, a dynamic simulation during preparation of the project specification was decided upon. This study enables comparison of the previously recommended antisurge and recycle layout with other systems proposed by the compressor manufacturers and antisurge control suppliers.
The configuration recommended by the study team has the advantage that a second-stage recycle cooler is not required. In this way, the second-stage volume is reduced, resulting in a relatively low settle-out pressure.
SELECTION BASIS
Dynamic simulation was used to compare the surge behavior of the new design with three existing antisurge and recycle systems.
For each of the four systems, two extreme cases of getting into surge were dynamically simulated. The dynamic simulation was executed by Trident Consultants Ltd., London, as instructed by and in consultation with Fluor Daniel BV.
Configuration A (Fig. 3a) (9953 bytes) has one combined cold gas antisurge/recycle line for each compressor stage. The layout shows an interstage cooler with the capacity of the compressors' rated flow and a second-stage recycle cooler with a capacity of 70% of rated flow.
Configuration B (Fig. 3b) (9953 bytes) shows one cold-gas antisurge line for each compressor stage and a common recycle line for two compressor trains over both compressor stages.
The configuration has an interstage cooler also used for antisurge protection with a cooling capacity of 100% of rated flow. In addition, the system has a recycle cooler in the common recycle line over the two compressor trains.
Configuration C (Fig. 3c) (9953 bytes) shows the system proposed by the study team. This configuration has both a hot-gas antisurge line and a cold-gas recycle line for each compressor stage.
In this configuration, the intercooler is used as recycle cooler for the first, as well as for the second stage. For this reason, the intercooler has a cooling capacity of, 120% of the capacity required at rated conditions.
The cooler has a very low pressure drop (0.5 bar, maximum), and its design capacity is large enough to perform recycle and interstage cooling simultaneously,
Configuration D (Fig. 3d) (9953 bytes) shows the least complicated antisurge and recycle control. It has only one common cold-gas antisurge and recycle line over both compressor stages.
As in -Configurations A and B, the system consists of two coolers: an interstage cooler and a second-stage recycle cooler with a cooling capacity of 100% and 70%, respectively, of rated flow.
STARTING POINTS
Configurations A, B, C, and D were compared. The same compressor curves were used for all four configurations.
For all antisurge or combined antisurge/recycle valves, a stroke time of 2 sec was simulated. All valves with a dedicated recycle function had a stroke time of 5 sec.
For the dynamic simulation, the rated capacity of 12 MMscmd for each compressor train was used.
After verification of the validity of the dynamic simulation model, two extreme conditions for getting into surge were selected:
- Emergency shutdown (ESD) for one compressor: the inlet and outlet isolation valves close simultaneously, the motor trips, and the antisurge and recycle valves assume a fully open position.
The system is not depressurized.
- Blocked discharge caused by a hipps valve of Train 1, while both compressor trains are in operation: The motor of Train 1 is not stopped to simulate an extreme condition.
EVALUATION PARAMETERS
The four different configurations were evaluated by comparison of both the lay-outs' safety and economic aspects and by review of the results of the two dynamic simulation runs.
- Layout parameters: cost and safety
- Dynamic simulation parameters
- Surge behavior (Ss variable)
- Period of surge
- Maximum absorbed compressor power during surge
- Settle-out pressure (as a result of ESD)
- Maximum discharge pressure (as a result of hipps closure)
- Safety margin (energy consumption)
- Stability of the system after surge.
When a CCC FA32 control system is used, compressor surge can be expressed in the Ss variable. The point at which this parameter equals unity (Ss = 1) is defined as the point at which surging starts.
For each compressor stage the Ss variable is defined as shown in the accompanying equation box.
The control algorithm of the antisurge control system calculates a value of S, defined as S = Ss + safety margin.
The controller is configured such that, when S = 1, the compressor speed control is overruled by the compressor antisurge surge control and the antisurge valves are opened.
In a well designed system, the safety margin is an optimum between keeping the compressor out of surge (this requiring a large safety margin) and operating with minimum energy consumption, which means minimizing recycle operation (this requiring a small safety margin).
For each of the four configurations, this optimum was established by tuning the system.
EVALUATING THE RUNS
Run A: ESD. The surge behavior of the different configurations is shown dynamically in Fig. 4 (20568 bytes).
Note that the upset condition of ESD is introduced after 10 sec of stable control. The horizontal line in the figures represents Ss = 1. When this value is exceeded, the compressor suffers from surge.
The results of the ESD run are summarized in Table 1 (5802 bytes). The table shows that for all configurations, the settle-out pressure of the first stage is less than 110 bar, which is the design pressure of the suction system.
For all four configurations during ESD rundown, surge starts at an absorbed compressor power of 25 mw. (The compressor runs down in 35 sec to 25% of its operating speed.)
In Configurations A and B, both compressor stages are surging for an extended time (Table 1) (5802 bytes). The reason for surge is that, in an ESD situation, the compressor cannot develop enough head to supply positive flow to the compressor's first and second-stage suction.
The large dead time is caused by the coolers which are used during antisurge protection. Figs. 4a and 4b (20568 bytes) show that the surging is very hectic, which is assumed to be violent for the system. The system experiences difficulties in recovering stable control.
The fast hot gas antisurge lines in Option C almost instantaneously supply positive flow to the compressor suction. This configuration experiences a very short time of surge in the compressor's first stage (Fig. 4c) (20568 bytes) and (Table 1) (5802 bytes).
After surge, the system recovers and shows stable behavior.
ln Configuration D, the first stage is surging violently for 7 sec (Fig. 4d) (20568 bytes) and (Table 1) (5802 bytes), and the second stage has no surge. The surge in the first stage is due to the layout of the system.
The pressure differential is measured across both stages (Fig. 3d) (9953 bytes), so that where on the operating curve each stage is operating cannot be pinpointed. In this way, no effective surge control is possible.
A problem in the first stage is discovered too late due to considerable dead time.
Run B: Blocked discharge caused by a hipps valve. The four configurations show comparable surge behavior: the first stage surges for approximately 1 see and the second stage has no surge.
Since closure of the hipps valve is as fast as opening the antisurge valve and faster than opening of the recycle valve, it is very difficult to escape a short period of surge. An example of this typical dynamic surge behavior is shown in Fig. 5 (6484 bytes).
Note that the upset condition of closing the hipps valve is introduced after 10 sec of stable control.
ln Table 2 (7089 bytes), the results of Run B are summarized.
All systems, except Configuration B, return to stable control.
At a sudden blocked discharge without stopping the motor, the compressor motor moves to its maximum speed in an attempt to fulfill the flow setpoint. This results in a very high discharge pressure.
For none of the configurations, however, is the design pressure of 390 barg reached (Table 2) (7089 bytes). For Configuration B, the margin with the design pressure is small.
Note that in this simulation the pressure-protection instruments were disabled in order to simulate a worst case.
How the system recovers to stable control after surge shows something about the energy consumption in general. The configuration which shows the largest value (< l) of Ss after a period of surge is able to work with the smallest safety margin, thus reducing the need of recycle operation.
Table 2 (7089 bytes) shows that Configuration B is unable to recover. Configuration C appears to be the most energy efficient with Ss = 0.75.
EVALUATING LAYOUT PARAMETERS
A qualitative cost comparison of the four different configurations can easily be deduced from the number of coolers and valves (recycle, antisurge, and hipps).
Configurations A and D show both an interstage cooler and a second-stage recycle cooler.
Configuration B shows an interstage cooler also used during antisurge protection and a cooler in the common recycle line over both compressor trains.
Configuration C shows only one cooler fulfilling both services. Configuration C, however, shows the largest amount of two recycle valves and two antisurge valves for each compressor train.
The other extreme is System D, which has only one combined antisurge/recycle valve for each compressor train. Configuration D is the only one without a second stage knockout drum.
These comparisons lead to the conclusion that Configurations C and D are the least expensive.
For safety reasons, System B, having a common recycle for both trains over two stages, will need extra safeguarding and hipps valves. By sharing the recycle line, the two compressor trains are no longer independent.
Towards safety, Configurations A, C, and D are comparable and all three safer than Configuration B.
SELECTING, OPTIMIZING
From the evaluation summarized in Table 3 (6059 bytes), Configuration C shows the best test results.
For the study team, the dynamic simulation furnished clear proof that there is no better alternative than selecting the proposed antisurge and recycle system for the storage installation.
After the configuration proposed by the study team was selected as the preferred antisurge and recycle system, the system was optimized (Fig. 6) (6495 bytes) by deletion of the second-stage knockout drum and combination of the second-stage hot gas antisurge and recycle function. This change does not worsen the surge behavior.
The hipps closure and the ESD simulation runs were repeated to review the effect of the optimizations. They revealed that compressor surge could now be completely avoided in case of ESD (Fig. 7) (5861 bytes).
SIMULATION RESULTS
The dynamic simulation of the four possible antisurge and recycle layouts confirmed that the system proposed during the study phase, without recycle cooler, is technically feasible and is the optimum system. Dynamic simulation of the selected system showed it could be optimized to achieve a system with the following:
- A minimum occurrence of surging.
- A minimum number of valves by combining the second stage antisurge and recycle valve into one valve with double function, without compromising surge protection.
- The largest possible turndown without recycling compared to the three other alternative systems
- Optimum use of cooling capacity by combining the interstage and the recycle cooler into one cooler.
The proposed antisurge and recycle layout proved to be the best system for the storage gas-injection compressors, especially in view of the anticipated 250 starts and stops a year.
Moreover, by performing this dynamic simulation and selecting the optimum design during the preparation of the project specification, the system is defined before detailed engineering starts.
This avoids delay and costly change orders in the detailed phase of the project.
THE AUTHORS
Copyright 1995 Oil & Gas Journal. All Rights Reserved.
