The importance of correct selection of turbo-compressor system control elements (such as the anti-surge valve selection, valve opening and closing times and others) should always be emphasised.
The dynamic simulation should be employed as a tool to achieve the optimum sizing and selection for control components and parameters in each stage of the project.
In other words, dynamic simulations should be performed at different stages from basic design, to detailed design, final verification, and during operation.
Usually in early stages of a project, some assumptions should be made for the first dynamic simulation.
While these assumptions should be on a correct basis, they may not be the best options for the final turbo-compressor arrangement. The design should not be limited to those initial assumptions.
In case of a "surge" the whole rotor system will vibrate axially which could damage the axial bearings or the seals.
An axial bearing (thrust bearing) is shown alongside.
The origin of many parameters particularly control valve or on/off valve opening/closing times are initial assumptions made in an early stage of design.
This is done to complete the first dynamic simulation for finalisation of some critical issues such as the anti-surge valve sizing, the hot-gas-bypass requirements, the basic anti-surge arrangements.
In a case study for a centrifugal compressor, for the dynamic simulation at the basic design stage, an assumption was made to use 0.8 second per 25 mm of valve sizes for opening and closing time for all suction/discharge valves. This was at the early stage of the project and valve data was not available at that time.
This assumption could reflect the maximum actuation time that may be expected for some large on/off valves (ordinary on/off actuated valves can achieve around 0.5-0.8 second per 25 mm).
These assumptions were used for the basic design dynamic simulation which resulted in the anti-surge arrangement freeze, anti-surge valve sizes and anti-surge valve opening/closing time (in this case 2 second opening time and 4 second closing time).
However, the engineering contractor used basic design dynamic simulation assumptions as a basis for specifying all actuated valves in the suction and discharge.
The engineering contractor hence specified 0.75-0.85 second per 25 mm for all main suction/discharge valves opening/closing time, considering if initial assumptions of the first dynamic simulation kept within certain limits, the first dynamic simulation could remain valid and there is no need for other dynamic simulations.
This is a poor design decision.
The specified values (0.75-0.85 second per 25 mm for opening/closing time) may not be wrong based on the initially assumed timing value (0.8 second per 25 mm), but there could be some difficulties to supply or verify actuated valves with such tight actuation timing and also the final result could be a sub-optimum arrangement.
Optimum valve characteristics should be obtained based on a new dynamic simulation. In this case, the second dynamic simulation is performed (considering the anti-surge arrangement, anti-surge valve sizing and anti-surge valve opening/closing time as final) and optimum reaction times are obtained as 0.71 and 0.64 second opening/closing time per 25 mm for the suction valve and the discharge valve, respectively.
To accurately and reliably predict the dynamic behaviour of turbo-compressor systems, the dynamic model has to be supplemented with accurate input data based on as-built equipment performance.
For example, volume of various vessels and piping systems should be accurately modelled based on fabricated (or isolated) pieces of equipment.
The controllers used in the dynamic simulation models are often simplified based on basic control strategies. In this way, the simulated control systems do not represent the functionality of actual field controllers, which could have an effect on the results of the dynamic simulation.
A recommended solution for inaccuracy of simulated controllers could be using a "direct control-hardware linked simulation" approach instead of conventional software emulation. This method can guarantee the simulation accuracy and the functionality of field-installed controllers.
During the dynamic study, an integrated software-hardware solution should be developed by linking a rigorous plant dynamic model to a vendor supplied controller emulator (based on the actual vendor controller).
The requirement should be discussed with the compressor vendor (or the supplier of the anti-surge system) in the bidding stage, before the order placement. Once a dynamic model is developed, it should be validated against the design and the actual operating data to ensure the accuracy of the modelling.
Valve selection
Anti-surge valve requirements depend to a large degree on the turbo-compressor details, and the turbo-compressor system arrangement. The different aspects of the anti-surge valve sizing and selection are described.
Figure alongside is an example of an anti-surge valve.
The larger the anti-surge valve, the more flow that can be moved from the discharge side of the turbo-compressor to the suction side.
The speed of valve opening is also important for a turbo-compressor anti-surge application.
Generally, the larger the valve, the slower its opening time. Also, the larger the valve, the poorer its controllability at a partial recycle.
The situation could be improved by using a large valve that is boosted to open, thus combining a high opening speed with a high-flow capability. An optimum anti-surge valve should be selected for any turbo-compressor.
In a case study for a turbo-compressor, two anti-surge valves with the same opening time (around 1 second) were evaluated.
The small anti-surge valve resulted in the shutdown surge at a moderately high pressure. The large anti-surge valve (with around two times the Cv compared to the small one) resulted in a major reduction in the head. The shutdown surge occurred at a pressure just above the suction pressure. Cv is the valve flow coefficient.
Most information required for the sizing of the anti-surge valve is available on the turbo-compressor map. Commonly used margins are that the anti-surge valve should be capable of passing 100 percent of the surge flow-rate at around 50 percent of the valve opening.
In other words, the anti-surge valve Cv is selected from approximate range of two times of the required Cv based on the surge flow on the curve of the highest turbo-compressor speed (on the turbo-compressor map).
Other requirements of anti-surge control valves are:
- Reduction of the stroke time.
- A stable response.
- Minimised overshooting during valve adjustment steps.
Too fast a response could result in an excessive overshoot and a poor accuracy.
However, too slow a response may result in sluggish opening of the anti-surge valve. Correct size and configuration of the required actuators, instrumentations and accessories could guarantee an anti-surge valve response time of less than 2 seconds (to fully open).
For a large anti-surge valve (as pictured alongside), the anticipated noise level (before an external attenuation) should never exceed 100-110 dBA with fluid velocities below 0.3 Mach.
The inline and symmetrical flow path eliminates indirect flows and unnecessary changes in flow directions through an anti-surge valve. An axial-flow anti-surge valve is a well-known option.
The "breaker vanes" are often used in the downstream section of the valve body, which cut and streamline any flow turbulence (significant reductions in the noise, the turbulence, and the vibration).
High range-ability (the ratio between the rated Cv with completely open valve and the minimum Cv that the valve can control) is required (typical range-ability "150:1"), which means a successful control even with a high deltaP and a low flow. To increase the range-ability, special trims should be used.
In order to obtain service reliability, following considerations should be respected:
- Anti-surge valves usually fitted with pressure balanced pistons (the thrust should be independent of differential pressure across the valve).
- Bushings are anti-seize and self-lubricated.
- Proper packing should be used.
- The trim materials are carefully chosen. The material should be corrosion-proof and erosion-proof. Proper stainless steel alloys for usual services, or sintered tungsten carbides for special cases are specified.
Rapid changes in the differential pressure across the anti-surge valve should have no effect on the stability of the valve position.
In modern designs, the sealing is achieved usually by the position (and not by the torque). Leakages across the anti-surge valve will influence the efficiency of any turbo-compressor system.
In a case study for a medium size high pressure centrifugal compressor, with an 8" class 900# anti-surge valve and 70 bar differential pressure, the valve leaked approximately 150 Nm3/h of the process gas.
A typical "one anti-surge valve, one turbo-compressor casing" arrangement is always recommended. More complex systems of cascaded valves or valves around multiple compressors require a more detailed analysis and complex provisions.
Generally two types of anti-surge valves are used:
- Globe valve.
- Noise-attenuating ball valve.
The globe valve's capacity (Cv) approximately varies with the square of the percentage travel. The noise-attenuating ball valve's capacity (Cv) varies roughly with the cube of the percentage travel:
- Cv~(travel)2 for a globe valve
- Cv~(travel)3 for a ball valve
The noise-attenuating ball valve will have more capacity to depressurise the discharge volume compared to the globe valve with the same size.
In a case study for a 150 mm (6") size anti-surge valve, the Cv of a selected ball valve was more than 2.5 times of the Cv of selected globe valve (the same size). At 2/3 of valve travel, the selected ball valve flow was more than 50 percent higher than the same size globe valve flow.
This additional flow capacity sometimes makes the noise-attenuating ball valve theoretically a better choice in an anti-surge installation (it is only a theoretical and textbook priority).On the other hand, the globe valve behaviour is more predictable and more control-able (less nonlinear). A ball valve usually offers a highly-nonlinear behaviour (Cv~(travel)3).
Both globe valve and ball valve are used in modern anti-surge systems. Practically the globe valve is more common in turbo-compressor anti-surge systems.
Amin Almasi is lead rotating equipment engineer at WorleyParsons Services in Brisbane.
He specialises in rotating machines including centrifugal, screw and reciprocating compressors, gas and steam turbines, pumps, condition monitoring and reliability.