Q: We have a problem of production mismatch on the custodian metering system in our natural gas station (Figure 1). There are two trains with two different runs. It is a gas treatment and export station. We export to the LNG station and also to some domestic users. The details:
- Total production: 12-13.5 million standard cubic feet per day (MMSCFD), confirmed.
- Trains: A and B.
- Runs: Train A has two runs; train B has two runs.
- Export parameters: -5 to 5 °C
- Flow computer: Emerson FloBoss S600+ Series
- Primary element: Daniel Senior orifice
The issues: We always observe a loss or mismatch of 0.3 MMSCFD whenever train A is not available and we flow through train B. Also, could you tell me the full meaning of AR and how can we get the AR reference?
Alabi Femi, lead, control room
Total E&P, Nigeria
A: My wife's scale reads two pounds below mine. Whenever I want to feel good, I measure my weight on her scale. After the two measurements, I still don’t know what my weight is because either or both scales are probably in error. In other words, whenever you’re selling gas, use train A and whenever you're buying it, use train B. But joking aside, if I understand you correctly, it seems you have a lousy metering system in your station.
You're measuring natural gas flow at 12 MMSCFD (8,333 SCFM) through two alternate trains, using orifice flow elements, and you observe that train B always reads lower by 0.3 MMSCFD—a difference of 2.5% between the two readings. Assuming the flowmeter in train A is correct (probably a false assumption, as we have no idea what the actual flow is), this would mean you have a 2.5% full scale (FS) error. This error is a fixed quantity (0.3 MMSCFD), and therefore if flow drops, this quantity becomes a larger percentage of the actual reading (AR) of the actual flow (AF).
In an ideal installation, if the orifice bore diameter is correctly calculated (and the calculation includes the bleed hole flow), the edges of the orifice are sharp, and the plates are correctly installed, the error contribution of the plate itself within a 3:1 flow range should be about 0.1% AR. These days, the smart transmitters have high rangeability and very good accuracy (about 0.1% FS), which, at a minimum flow on a 3:1 installation, corresponds to 0.3% AR. Now, if the station was calibrated against a 0.1% AR reference, the total error should be about 0.5% AR.
Therefore, if at a full flow of 12 MMSCFD your train B reads 0.3 MMSCFD below train A, that error is 2.5% FS, and at the minimum flow on a 3:1 installation, it is 7.5% AR. This is obviously unacceptable and you must recalibrate the stations.
A: We can't assume the error is 2.5% because the error or errors are not defined by a difference. (The equivalent old statement is, if you have two watches, then you don’t know what time it is.) Either or both are probably in error.
Achieving good accuracy is neither cheap nor easy. The ideal solution is a proper, in-situ flow calibration for both meter runs. Next best is a detailed inspection/review of the installations, all the sensors, flow computer, and anything that can affect the reported flow. Since you have orifice fittings, it is reasonably easy to inspect and/or replace both orifice plates—new orifice plates will eliminate one set of possible defects. Any errors in the temperature, gas density or differential pressure (DP) are flow errors, of course.
If you have access to the ISO flow standards you can see what they say about installations and sources of error.
I like to have the "value of the measurement" defined to provide justification for the costs involved in reducing uncertainties or improving reliability. It should be possible to learn the potential annual cost for an actual 1% error. This might provide justification for better calibration equipment or field instruments.
I get upset when someone tries to save money where it should be spent. None of the salesmen I’ve met go out of their way to explain that the listed accuracy of an instrument is only under laboratory conditions and may be worse at a low percent of scale, or that errors can increase for weather conditions in the real world. Probably the most accurate part of a flow loop today is the DP transmitter; specifications there are greatly improved over the past few decades.
A: To further Béla Lipták's comment, there are three more issues that may arise to give you the error you mention:
- During start up and shutdown, there may be measurement errors since the orifice plate pressure drops will not be exactly the same for the orifice in full-volume mode.
- Even though you didn’t mention it, the majority of orifice plates have either bleed or vent weep holes. Again, if this small orifice is not calculated into the overall orifice plate calculations, it may cause measurement errors.
- Finally, the orifice plate, over time, will see erosion/abrasion/corrosion impacts in the orifice.
Alex (Alejandro) Varga
Q: Is the Instrument and Automation Engineers' Handbook, 5th edition, in two volumes or three volumes? Does is it have updated control methods and control systems sections, as the 4th edition has?
A: The status of my handbook is as follows:
- Volume 1 on Measurement and Safety is in its 5th edition (2,199 pages).
- Volume 2 on Analyzers and Analysis is in its 5th edition (1,314 pages).
- Volume 2 on Industrial Control and Optimization is in its 4th edition (2,387 pages)
- Volume 3 on Software and Networks is in its 4th edition (874 pages).
In the 5th edition, I split the Measurement and Analysis volume into two volumes because with all the new material, it would have ended up a 3,500 page volume, which would have be hard to handle. I left the Industrial Control and Optimization volume in its 4th edition because the control strategies of traditional industries have not changed.
I'm now working on a new volume, probably titled "Controlling the Future," in which I'm trying to show that analyzing and controlling new processes (like climate change or artificial intelligence AI development, etc.) requires application of the same principles that we've developed for analysis of the dynamics of industrial processes—by determining their time constants, dead times, gains, manipulated variables, interactions, etc.