I'M IN the middle of producing a technical report for the institute of measurement and control to gain my mature charter status. The paper I'm writing concerns an existing gas compressor station operating at 85 bar using a 36 elbow flowmeter as the compressor station's means of flow measurement. I'd appreciate any guidance you could provide me with.
THE TWO critical components of a surge control loop are the flow sensor and the surge valve. Both must be fast, accurate and reliable. Flow oscillations under surge conditions occur on a cycle of little more than a second. The flow transmitter should be fast enough to detect these. The time constants of various transmitter designs are as follows:
- Pneumatic with damping: Up to 16 seconds
- Electronic d/p: 0.2 to 1.7 seconds
- Diffused silicone d/p: Down to 0.005 seconds
Only the diffused silicone type sensor design is fast enough to follow the precipitous flow drop that occurs at the beginning of surge or the oscillations during surge. Measurement noise is another serious concern, because it necessitates a greater margin between the surge and the control lines. Noise can be minimized by the use of 20 pipe diameters of upstream and 5 diameters of downstream straight runs around the streamlined flow tube type sensor. Noise will also be reduced if the low-pressure tap of the d/p cell is connected to a piezometric ring in the venturi type flow tube. The addition of straightening vanes will also contribute to the reduction of noise. For details on the installation of Herschel venturi with annular pressure chamber, refer to Section 2.29 in the 4th edition of Volume 1 of the Instrument Engineer’s Handhook.)
The error in elbow flowmeter type measurements is usually too high (5%) for consideration, it is also sensitive to material buildup on the inside wall of the elbow and to variations in the pressure and SpG of the gas. For details on the capabilities, installation, etc. of elbow flow sensors, refer to Section 2.6 in the 4th edition of Volume 1 of the Instrument Engineer’s Handhook.)
Anti-surge control usually requires a flow sensor on the suction side of the compressor. If good, noise-free flow measurement cannot be obtained on that side, a corrected discharge side differential pressure reading (hd) can be substituted. Equation 8.15(3) can be used to obtain the suction side differential pressure (hs) from readings of (hd) and the suction plus discharge pressures and temperatures (Ps, Pd,Ts and Td).
IT SEEMS that the gas compression station is already existing and they use a 36-in. elbow to measure the flow. It is not clear if it is used on the suction for antisurge control purposes or in the discharge line downstream the branch for antisurge recicling to measure the gas sold to a client. The reason for the choice could have been the minimal pressure drop required by the elbow that could be less expensive than having a much better meter but requiring a large amount of energy to be wasted. If it is for antisurge control and the compressor very rarely works close to the surge protection line, I think that they can keep the surge protection line far away from the true surge line and still save money.
If the elbow is used to prepare the bill, if I was the customer, I would ask something better. Apart from these considerations, your answer is exhaustive and should allow Steve to proceed with his report.
I WOULD separate surge detection from the flowmeter. Valves of the required size will most probably require booster relays. The surge detector should be fast and used either for alarm or shutdown. A vibration monitor is commonly used. I once installed a DP switch as a surge detector with one port connected directly to the pipe and the other port connected through a small volume with a needle valve to lag and average the pressure. The thought was to have an adjustable and sensitive detecter to discover a slow transition into incipient surge early enough to take action and not shut sown.
The elbow meter should be reviewed to be sure that there is an adequate signal. Without a flow rate given it cannot be checked now.
I must take issue with 17 seconds for the pneumatic transmitter. We never damped signals at the transmitter. Flow valve response and noise suppression was smoothed out by using a wide controller PB and fairly fast reset. A damped recorder kept the operators happy.
One very difficult compressor with very little space between surge and stonewall was brought under control only after adding nonlinear gain and signal modification.
THESE DAYS, I would look to using a V-cone in such service – easier to install in existing pipe than a venturi and more forgiving of swirl. Permanent loss is not that much worse than a venturi. And the price is dropping to ‘reasonable.’ Accuracy is not so much a requirement as reproducibility – the surge line will need to be determined experimentally anyway. Note also the recent work on venturi meters which indicates that flow into the Hershel design with a 15 degree inlet cone tends to break away at high Re, and NEL tests make a 10.5 deg cone look better (See NEL Flow website).
The elbow meter is a low differential device, and is certainly less accurately predictable than any of the other devices. Assuming that it is not a reversible flow problem which seems likely, then the 22.5 degree tap position is superior to the 45 degree even though it generates less head (45 degree is close to a flow breakaway location); short radius bend is superior to long radius in generating much more head. See IEH Section of course.
When it comes to surge, prediction of incipient surge is MUCH better than getting the machine out of surge. The old Solar approach using microphones to detect the noise of incipient surge, and tripping the machine after a very few cycles had much to recommend it on pipeline compressors – if the machine was in surge, then its contribution to developing head was probably unnecessary.
Ian H. Gibson, Process, Control and Safety Engineering Consultant, Melbourne, Australia
|Need an Expert Opinion?|