Recent Experience on Implementation of New Techniques of Gas Metering
Andrew W. Jamieson, Shell U.K. Exploration and Production
Gas metering in the oil and gas industry is changing rapidly with the steady introduction of new metering techniques, which will supplant or co-exist alongside traditional methods. Gas metering fits readily into a concept in which any oil and gas industry metering application is a special application of multiphase metering. Recent experience in applying new metering techniques is discussed for ultrasonic meters onshore and offshore, Coriolis meters offshore and Venturi meters for wet gas service. The constraints affecting the various parties involved in implementing new metering techniques are discussed and conclusions drawn for the kind of guidelines that the industry requires.
As with all areas of metering, gas metering in the oil and gas industry is going through a revolution. Techniques developed over the last two decades are now being applied widely, supplanting traditional orifice metering. Old techniques, such as Venturi meters, are making a comeback. Application of powerful data processing techniques means that complex metering systems can be operated and interrogated remotely. However, with these significant advances comes the difficulty for metering specialists of being able to acquire a sufficiently wide and deep knowledge of the large variety of systems to be able to manage the metering systems of the future. It is self evident that without a sound technical foundation, the most elaborate metering management system is useless.
In this paper, I first set gas metering in an overall metering context using the ‘Multiphase triangle’. I then consider several recent gas metering applications using ultrasonic, Coriolis and Venturi meters. I discuss the role of standards and the positive and negative effects of the current drive to reduce operating costs, especially in the North Sea. This leads naturally to indications of how gas metering may develop over the next few years.
3. Gas metering and the multiphase triangle
Over the last two years I have strongly promoted the use of the ‘Multiphase triangle’ in discussing all measurement problems in the upstream oil and gas industry. The triangle (Fig 1) gives the composition of an unprocessed crude oil stream in terms of fractions of oil, gas and water. The vertices of the triangle represent 100% oil, gas and water; the sides represent two-phase oil-gas, water-gas and oil-water. (Strictly speaking, the oil and water ‘components’ together form the liquid ‘phase’, but in multiphase metering it is usual to blur the distinction between ‘component’ and ‘phase’ and consider oil and water as separate phases.) In the middle of the triangle we have the enormous variation of three-phase flows.
Already one can see that each application is unique. Are we processing the fluid onshore, in the jungle, in the desert, or in the Arctic? Are we processing the fluid offshore, on a platform or subsea? Are we considering processing downhole? Each of these locations brings its own design problems and implementation difficulties. Then we have to consider the fluid properties, namely flowrate, temperature, pressure, viscosity; then the economics of processing and how much “impurity” we can afford to leave in the processed streams. These latter considerations are especially important in sales gas metering.
Figure 1 Multiphase Triangle
The traditional view was that metering was essentially a matter of measuring single phase fluids, and that one had to adapt the methods to cope with the contaminants, such as liquids in gas streams. I consider a better view is to treat all oil and gas metering applications as multiphase metering, and that one chooses from the variety of components and techniques available to build a system to measure the composition and flowrate of that multiphase flow.
One might ask, “Is this really necessary for gas metering, and especially sales gas metering, where we are obviously dealing with a single phase fluid?”
If we look at typical sales gas metering system, we usually analyse the gas for about ten components, eight of which are hydrocarbon, plus nitrogen and carbon dioxide. The composition and flowrate are probably fed into an allocation system, which, if it is to work properly, will be based on component mass flowrates. Given that in multiphase metering we blur the distinction between ‘component’ and ‘phase’ it is evident that a sales gas metering system is nothing other than a multiphase metering system giving the mass flow of each component. Current multiphase metering systems give the volume flowrate for each phase; I believe that multiphase allocation systems will need to work in mass terms, and as such, the sales gas allocation systems provide the ideal model of what is required.
The gas metering applications I will discuss occupy the very tip of the multiphase triangle. In volume terms, the liquid content will be less than about 1-2%. In mass terms, however, the liquid may then make up some 25% of the mass, and as the price of oil and gas is roughly the same in mass terms, the liquid content will then be very valuable.
4. Recent applications
I will discuss my involvement over the last few years with new technology gas metering systems: gas ultrasonic meters offshore on separators and onshore for sales gas, Coriolis meters used for metering unusual fluids, and Venturi meters used on wet gas service. A recurrent theme will be the effort needed to get such systems working, the effort required to keep them working and deal with the novel issues that seem to turn up with each new application, and then how the hard gained information is fed back within one’s own company, to partners and other operators, to designers, system builders, and manufacturers.
4.1 Ultrasonic meters on separators
I have been involved with a number of projects where ultrasonic meters have been installed to meter gas off separators. On many of these applications the ultrasonic meters have not worked, and often have not even had a chance of working. It is too simplistic to blame the meters; I have also seen many conventional orifice meters on separators that could never work, but still produce “metering” information that is used by optimists in hydrocarbon management systems.
Let us consider what happens to a multipath ultrasonic meter from when it leaves the factory, with glistening paint outside and shiny smooth inside, until it goes into service perhaps eighteen months later. First, there is the trip to the calibration station. The transducers may be installed on the meter, or they may be kept separate from the body. The transducers comprise relatively light sensor parts mounted on heavy flanges, especially those intended for high pressure service. They tend to be fragile, and the first unhappy event in the meter’s life may be a delay in calibration because a transducer has broken in transit. The calibration itself offers opportunities for interesting things to happen. The manufacturer may have coated the internal surface with a waxy preservative; unless this is removed a significant calibration error will result. Upstream and downstream pipework may not match the meter bore; again a significant calibration error may occur.
Once calibrated, the meter has to be cared for. If another transducer gets damaged, and there may be up to twelve of these for a meter, one has to consider what to do to maintain the calibration. In principle, all one should have to do is replace the transducer with another as it should be possible to manufacture transducers to be nearly identical. In practice this is not yet so. Transducers tend to come as matched pairs, so is it possible just to replace a pair with another pair, without having to recalibrate the meter? Some manufacturers claim this can be done very repeatably, but as there are many variants of transducers it is difficult for the user to be sure that this is indeed the case for the meter and transducers for his application. Should one then hold a complete set of transducers, calibrated for the meter? But then one is faced with the problem of how to look after these as well as the meter.
Let us suppose the meter has arrived on site for installation. It may be that the construction engineers have decided to keep it in safe, dry storage until the metering station has been fabricated, cleaned and about to be put into service. In this case, the meter will have a dull life for several months, but don’t feel too sorry for it, it may yet have fun during its commissioning, as we shall see shortly. But instead the meter may be destined to see everything that goes on at the fabrication yard. It will be left out in all weathers, be used as a pup-piece in the construction, will have the fabrication debris flushed through it and will be pressure tested with the associated pipework. Its local electronics box will be opened several times, and get filled with rain on snow.
Remember that the meter was for service on a separator on an offshore platform. We have now got offshore, and are about to put the meter into service. The separator is brought to pressure; the precommissioned meter is switched on and shortly after stops working. Everything about the meter is suspect; either the transducers or the electronics may have failed, or both. Working on the electronics is difficult; the platform is of the open module construction and it is a rare day that one can open electronic boxes without rain or salt spray getting inside. Because ultrasonic meters are considered to be low maintenance items, isolation valves were not installed to permit them to be removed relatively easily. We have to wait until there is a scheduled production shutdown before the meter can be removed.
Not unreasonably, the operating company wants to be sure that whatever meter goes back in, it will work. No-one can give such an assurance for the ultrasonic meter, but in reality they cannot give that assurance for any other meter, as the complex mixture of problems revealed when the ultrasonic meter is removed defies easy explanation. It is impossible to say whether the failure resulted from flaws in design or manufacture, or from mishandling during the time in the fabrication yards, or from failure in the short period in actual service.
My view is that metering on separator offtakes is far more difficult than is generally believed. This type of metering is becoming steadily more important in allocation schemes and consequently it is important to have good gas meters. I believe that ultrasonic metering is the best technique for this application, but current meters, essentially the same as those for the stable conditions of onshore sales gas metering, need to be better adapted to the offshore environment.
4.2 Ultrasonic meters onshore
Let us now consider a meter that was looked after carefully before it was put into operation. Whereas the meter described in the separator application did not really get a chance to give useful information, this meter from the same manufacturer has strongly vindicated the view that ultrasonic meters give a new way of looking at gas flows, and give much useful diagnostic information.
The meter was intended for an important sales gas metering station, which comprised two streams fed from an inlet header. The system vendors had been asked to design the inlet pipework configuration to minimise the transmission of ultrasonic noise from upstream and to ensure that the swirl at the meters would not degrade the accuracy of the meters significantly. During commissioning, the meter vendors noticed that the flow profile at the meter in the right hand stream was dramatically different from that at the meter in the left hand stream. (Right and left are defined with respect to the direction of flow of the gas.) The right hand 4-path meter was showing velocities on the two inner chords some 10-15% lower than the two outer chords over the whole flow range tested, whereas the left hand one showed the normal pattern with the velocities on the two inner chords higher than on the outer chords. Note that 4-path meters measure the average velocity over the whole length of the chords, thus a wide range of flow profiles can give the same chord velocities.
Initially swirl was discounted, as the manufacturer usually associated swirl with unstable velocity readings. The meter readings were very stable on all chords, and the velocity pattern indicated very symmetrical flow conditions in the pipe. The two outer chords showed virtually identical velocities, as did the two inner chords. We were forced to consider that there might be something in the line, something that would have to be large and fairly close to the meter to give the observed velocity pattern, but had difficulty in believing this given the care that had been taken during construction to keep the pipework clear of debris. The upstream pipework was opened up and the pipework was clear of obstructions to beyond the inlet header.
An intense discussion with NEL gave the almost certain cause of the problem. The major difficulty was in explaining the difference between the right and left hand streams. Our metering station was mirror-symmetrical for two valves and four bends upstream of the metering. The fifth bend was to the underground feed line to the station. It appeared evident that the combinations of out-of-plane bends were generating swirl that was increased in the right hand stream and partially cancelled in the left hand stream. There was a relatively long length (22D) of upstream straight pipework which would allow any contra-rotating vortices to dissipate and leave a strong steady swirl in the right hand stream. As the only mechanism for removing this swirl is friction with the pipewall, it can persist for very long distances.
The vendors had quoted work done by Grimley at SwRI in the USA  as justification for saying that the swirl levels induced by the header configuration would not lead to significant degradation in accuracy of the meters. Unfortunately, Grimley’s work only considered the effect of two bends upstream of the meter, as does ISO 5167, the main guide in gas metering to requirements for upstream pipework. It is evident that the ultrasonic meter has shown that for our installation at least it is not sufficient to consider only the first two upstream bends or fittings when designing swirl free piping configurations.
We had already considered the possible need to install a thick plate flow conditioner if swirl levels were found to be higher than those described by Grimley, and are now making the necessary modifications for its installation. Work carried out by NEL in a JIP on header configurations upstream of an orifice several years ago gives us the supporting evidence that thick plate conditioners will remove the levels of swirl we believe to be present in the right hand stream and make a flow profile acceptable for the 4-path meter.
4.3 Coriolis metering of “funny” gas
I was asked to advise on a proposal by a Shell operating company to meter “gas” and “condensate” on an offshore facility for export into a gas pipeline. The reason for using inverted commas is that the “gas” has a density of about 300 kg/m3 and the “condensate” has a density of about 500 kg/m3 at metering conditions. The wellstream is fed into a 3-phase separator to remove water from the crude stream. The condensate and gas streams are then compressed to give the two supercritical (neither gas nor liquid) fluids described above. These are metered separately into the gas export line where they remain in the gas phase provided the other input streams to the export line have a sufficiently low average molecular weight.
The problem was to decide how best to meter these supercritical fluids. The metering specialist on the project had sought advice widely and had decided that the best way was to use Coriolis meters. The quantities were relatively low, falling within the range of 1½” or 3” meters. A master meter would be used to check both the “gas” and “condensate” meters periodically.
At first I was sceptical. I had no difficulty in accepting Coriolis meters for metering real liquids, and was in the process of being convinced that they were also suitable for metering real gas. I had shortly before agreed to using one for metering fuel gas at one of our installations, but was not so happy for their use in a more critical gas application. My concerns went back some years when we at the Shell E&P Research laboratory tried to develop a multiphase meter based on the Coriolis metering principle. We thought that a Coriolis meter could already handle two-phase oil and water mixtures, and that the way ahead would be to make a Coriolis meter that could tolerate increasing quantities of gas. However, as is now well known, Coriolis meters cannot tolerate more than a few percent of gas unless it is completely entrained in the liquid. Essentially the coupling of the vibrating tube to the fluid ceases and the meter stops working. For single phase gas, the Coriolis principle should work, especially at high pressures. Evidently it had taken some years for manufacturers to develop reliable products for real gas, but how would these perform on these supercritical fluids? I suspected that the coupling between the fluid and the vibrating tube would be degraded, and the calibrations performed with real gas and real liquid might not be meaningful.
I thought that the most comparable fluid would be supercritical ethylene, and learned that there was only one very special rig in the world for calibrating the turbine meters used for that application. I also learned that a JIP was testing Coriolis meters for that service and were showing good results.
By this time I had realised that there was not a more practical approach for offshore application. My remaining reservations now centred on the more practical aspects of ensuring the mounting of the meters followed the manufacturer’s best practices, and whether it would be practical to carry out an in-situ calibration of the master meter using a compact prover and a turbine meter as a transfer standard. From data inspected at a factory acceptance test, it was evident that the “proving” of a Coriolis meter by this method is still in development. To get a mass flowrate, the volume flowrate of the turbine meter must be multiplied by the density of the calibration fluid, and master meter rigs do not usually have on-line density measurement. Another complication is that the turbine meter may be calibrated accurately with the compact prover, but that free gas may appear when the turbine meter is used to calibrate the Coriolis meter. In that case the turbine meter will over-read, leading to an under-calibration of the Coriolis meter.
By the end of summer, we should know whether this novel approach to metering “funny” gas works as well as it deserves to.
4.4 Venturi wet gas metering
I give a short discussion on Venturi wet gas metering to emphasise the point that in general we do not get enough feedback from operational facilities to feed forward into new developments. Shell Expro now has two facilities using wet gas metering essentially for sales gas metering. The drive to use wet gas metering was to eliminate the bulk processing facilities that would have been necessary if conventional metering techniques were used. At the time when the development decisions were being made, the cost of these production facilities would have made the projects uneconomic.
On both facilities, the liquid content is less than 1%. The over-reading of the Venturi meters because of the entrained liquid is corrected using a variant of Murdock's equation. The liquid content is determined (in principle) from periodic well tests. On one facility the ultrasonic meter in the test separator gas outlet has several failed transducers which cannot be replaced as they are deemed obsolete by the manufacturer.
At the design stage, we had hoped that the data from the well testing would build up a picture of how the wells were changing with time, and that this could be used to guide the development of other fields where it would not be practical to have a test separator. Currently we have a project where the fluids are slightly wetter than those described above, and wet gas metering is an attractive option. Even more critical is the need to determine the onset of water breakthrough, and although there are several ways in principle of doing that, none have been fully demonstrated to date. For the project to move forward, the risk of choosing a low cost option using techniques that are promising but not fully demonstrated must be judged against a high cost traditional option using techniques that are known to give problems.
Good implementation of new technology gas metering systems requires favourable circumstances. Firstly, the technology must be sufficiently mature that it can be considered for application. This means that the measurement principles are theoretically sound, and that they have been incorporated in a practical manner into working, commercially available equipment. This in turn means that there is a manufacturer keen to support the new metering approach, and who can foresee an adequate commercial return for his efforts in developing and marketing the new equipment.
Second, the ideas underlying the new technology have to have circulated long enough in the metering community that there is a general expectation that the new technique can work in a oil and gas environment. Often the ideas will have been generated in oil and gas company laboratories or in Universities, and before practical equipment can be developed there has to be close interaction between inventor and manufacturer.
Third, suitable pilot applications are required, and these must be at locations where the potential user of the new metering technique sees clear commercial advantages from its use. This means it can give significantly enhanced performance for the same cost, or, more usually, the same performance, or even a lower performance, at a significantly lower cost. This means, inevitably, that the new technology system will be installed on a “difficult” application, and those responsible for the installation cannot expect things to be easy.
Fourth, a reasonably tolerant user is essential. Neither the inventor nor the manufacturer will have been able to anticipate all of the circumstances of the installation; often the new technology metering will be the first means of making an assessment. Traditional views may have to make way for ones based on more accurate and reliable measurements, and that process does not happen easily.
Fifth, even after successful pilot applications, there comes the slow, difficult process of establishing the new metering technique as a general method to be applied by projects through design houses, main contractors and metering system suppliers. Current practice is to use a single contractor on a lump sum basis. Sub-contractors make their bids on fixed prices. This approach does not, in my experience, lead directly to successful implementation of new technology systems. In practice, each application will be somewhat different to the pilot schemes; precautions considered essential by those carrying out pilot schemes may appear unnecessary to project teams; nagging doubts over the actual flow conditions may suddenly crystallise into unwelcome hard facts that cannot be ignored. At the end of the day the new technology metering system may have been installed for a much lower cost than a traditional system, but the metering specialist is unlikely to receive much praise from a project manager who had reckoned on getting all of the savings, and finds he is over-budget because of the work required to get the system working properly.
Sixth, operators and users of the new systems need to learn how they work, and indeed discover what extra information the new systems can give them that will enhance their operations. In these days where staff numbers are being reduced dramatically, it is unlikely that the learning process will be easy. One must rather expect that in the immediate period after a new technology system is put into operation that more resources will be required, rather than less, to ensure that any problems are dealt with promptly and the associated lessons learnt and circulated to other users.
Seventh, a balance has to be struck between the view that development work cannot be carried out on operational facilities and opposing view that unless new systems can be developed in real conditions, there is no chance of realising the benefits that can come from their use. How such a balance is struck depends on the value the industry puts on the availability of the new techniques.
Implementation of new techniques for gas metering has clearly led to improvements in gas metering performance, lower installed costs, and better understanding of what actually happens in gas flows. This better understanding raises questions on installation requirements in current metering standards, and suggests that it would be unwise for the industry to draw up rigid standards that could be shown to be invalid in a relatively short time. I consider it would be preferable to have guidelines for a general approach to metering, with each specific application considered on its own merits and the proposed solution submitted for some sort of peer review to judge its suitability. I would suggest that this is more or less what happens in practice.
Gas metering clearly fits into a general treatment of any oil and gas metering application as being a special case of multiphase flow metering. Indeed, as the industry blurs the distinction between “phases” and “components”, sales gas metering systems which measure the flowrates of perhaps ten components form an ideal model for future multiphase allocation metering systems.
The major difficulty with new technology gas metering is still in the quality of the implementation. Too much is expected from the new systems too soon in terms of reduced operating and maintenance costs. All sections of the industry, developers, manufacturers, project teams, design houses, contractors, government agencies, gas shippers, facility operators and gas transportation controllers all require an appreciation of the other parties’ problems. For the increasingly sophisticated gas measurement systems required in the 21st century, a sound technical basis, reliable equipment, sound project management, workable contracts, practical operating procedures, sensible regulation are all essential.
1. GRIMLEY, T.A. The influence of velocity profile on ultrasonic meter performance,
A.G.A. 1998 Operations Conference, Seattle, Washington.