The paper discusses the need for and how to measure flare gas. More and more operators invest in flare gas measurement systems, and this is initiated not only by governmental legislation. Improved control of emissions and leakage rates implies reduced costs and hence increased profits. A brief overview of different flare gas measurement technologies is presented, with emphasis on what is regarded as the flare gas measurement technology of the future; ultrasonic time-of-flight meters. More specifically, a description of the Fluenta FGM 130 Flare Gas Meter is given. Finally, some views of the future trends within flare gas meters are presented.

1 INTRODUCTION

Flare systems at offshore production platforms, refineries and chemical plants are primarily installed for safety purposes. The flare systems are mainly activated due to an unexpected shut-down or when it becomes necessary to suddenly dispose rapidly large amounts of gas. Going 20 years back, a common sign of an offshore production platform or process plant was the ever-burning flare, to be seen from far distances. The burning flare was in a way the mark of the oil production age. At that time, few, if any, regarded the burning flare as an unwanted proof of unprofitable production and gas emissions.

This has now changed, and there is an entirely different awareness amongst operators and oil companies about the effect of gas emission as both an environmental and an economical issue. In addition to the obvious safety purposes of a flare, national legislation in more and more countries requires control of the emission, and in some countries operators have to pay taxes for their CO₂ emission. Since the Kyoto climate conference in December 1997 the focus on the global warming has increased, and the emission of CO₂ has become an international responsibility.

This yields also for the flare gas emission, and oil production platforms are nowadays both designed and rebuilt for zero-flare operation. This change in operation of the flare systems has also influenced on the requirements of the flare gas metering systems. From a continuous, more or less steady flowing amount of flare gas, today's picture is more binary in nature, wit the gas flow either to be approximately zero, or at the specified maximum rate.

2 WHY MEASURE FLARE GAS?

From an operator's point of view, there is no reason to measure the flare gas unless it is of economical benefit or it is required by e.g. the national government for tax payment purposes. In order to achieve economical benefit of a flare gas measurement, the purpose of the measurement could be to identify points of leakage, to obtain better control with emission rates or mass balancing. These application areas for ultrasonic gas flow meters have added metering requirements beyond the direct flare gas metering requirements. Also, this has opened a new market within refineries and onshore process plants.

Irrespective of the application, the operator would not invest in a flare gas metering system, or any other system for that matter, unless the investment would be economically beneficial in the long term. In that respect the choice of technology for the metering application is of most importance. An evaluation of cost versus benefit should be made, and the basis for the evaluation would be parameters as investment, installation and maintenance costs, measurement uncertainty, repeatability, measurement range, reliability and non-intrusiveness of the measurement technology. The ultimate flow meter would of course be the best sum of all these parameters. With a very low selling price. However, as they say, you get what you pay for, so the general rule still applies; quality costs. Irrespective of this, more and more operators world wide have experienced that measurement of flare gas implies control of emission. And control pays off, so the investment in measurement technology is really an investment in increased profit.

As earlier stated, more focus have been put on the environmental and economical aspects of the gas flaring, and in some countries the operators have to pay taxes for their CO₂ emissions. Accordingly, in order to fulfil the regularity requirements, the operators requirements regarding the flare gas metering systems have changed.

2.1 Governmental Legislation

In Norway, in 1993, regulations relating to measurement of fuel and flare gas for calculation of CO₂ tax in the petroleum activities were resolved (1) the regulation was stipulated by the Norwegian Petroleum Directorate (NPD) by virtue of Section 5 of Act of 21 December 1990 relating to CO₂ tax in the petroleum activities on the Norwegian continental shelf. The purpose of the regulation was to ensure that the calculation and reporting of CO₂ tax was based on accurate measurements.

Inevitably, oil companies operating on the Norwegian continental shelf had to relate to this regulation. However, also manufacturers of flare gas metering systems operating in this market were forced confirm that their instrumentation did comply with these regulations.

According to (1), only three measurement methods were acknowledged for flare gas metering on the Norwegian continental shelf:

. Ultrasonic measurement,
. Insertion turbines with density measurement/density calculation,
. Thermal mass meters.

However, a new revision of (1) expected to be accepted this year, states an operational range of flare gas meters up to 100 m/s, which in terms only qualifies the ultrasonic time-of-flight flow meter technology. This is not only a clear indication of what the future flare gas metering technology is expected to be, but is states that it is, today, the only proven technology to be utilised for these demanding applications. The single requirement of ultrasonic flow meter for flare gas metering is also stated in NORSOK STANDARD I-104, Section 7.1.3.(2).

3 FLARE GAS METERING

With the change in the operation of the flare systems, an adaptation of the flare gas metering systems has been imperative. With flare systems being installed primarily for safety purposes, the flare gas metering systems must cope with dramatically changes in the flow velocity, gas composition and temperature over a very short time scale. Hence, the measurement challenges may vary a lot over a short time period.

Due to the nature of e.g. a process shut-down, when all of the process gas is flared, the flow velocity may exceed 100 m/s. As a result of this extremely high flow velocity, unwanted particles and components such as oil, water, salt and scale may be transported along the flare pipeline. Knowing this, it is quite evident that any instrumentation that intrudes into the flare pipeline might get influenced, or at the worst get damaged, during such a shut-down.
Accordingly, limitations of what metering systems that can be put in operation have arisen.

3.1 Flare Gas Measurement Methods

Traditionally, conventional metering systems were used for flare gas measurements. Typical meters that were utilised are:

. insertion turbines
. Thermal mass meters
. Annubars

A turbine meter utilises the principle that the gas is led through the meter rotor. The rotor is designed with a specific number of blades positioned at a precise angle to the flow stream.
The gas impinges on the rotor blades causing the rotor to rotate, with the angular velocity of the rotor being directly proportionally to the gas velocity. Clean fluids are required to prevent contamination of the bearings unless sealed bearings are used. Insertion type turbine meters cause negligible pressure drop, but due to the local velocity measurement, the measurement uncertainty is higher than for conventional full-bore turbine meters. Typical flow range for such meters is up to 30 m/s.

Thermal mass meters are typically based on two Thermowell-protected Resistance Temperature Detectors (RTDs). When placed in the process stream, one RTD is heated and the other is sensing the process temperature. The temperature difference between the two elements is related to the process flow as higher flow rates cause increased cooling of the heated RTD. Thus, the temperature difference between the two RTDs is reduced. As with the insertion turbine meter, the thermal mass meter causes negligible pressure drops. In addition, it has no mechanical parts, high temperature range and requires little installation space. Typical flow range for the thermal mass meters is 0.3 to 30 m/s.

Annubars have been used for years on flare applications. An annubar is a differential pressure device with the signal increasing proportional to the square of the flow. Annubars are good for high flow rate applications, but are not good for low flow applications due to the small pressure difference these flows represent. For mass flow applications, annubars require pressure and temperature compensation. The characteristics of the annubar are high measurement principle, it causes a pressure drop in the pipe as it intrudes with the process flow. This again implies potentially high maintenance costs. Typical turn down ratio is 10:1,
So that several annubars are required in order to cover a large flow range.
Other metering types, such as positive displacement meters, vortex meters, hot-wire anemometers, coriolis mass flow meters and sonic nozzles have too limited flow range to be considered for such metering applications. In addition, some of these metering types introduce an unwanted pressure drop in the pipe.

A metering technology that has gained more and more acceptance for flow measurements is the ultrasonic time-of-flight meters.

4 ULTRASONIC TIME-OF-FLIGHT FLOW METERS

The technique of transit-time flow metering is well known to the physicists dealing with flow-metering problems and was first used by the German engineer Rutten in measuring water and steam flows in large canals as found in power station practice (3).

The ultrasonic time-of-flight gas flow meter is based on measurement of contra propagating ultrasonic pulses, in which the transit time of the sonic signal is measured along one or more diagonal paths in both the upstream and downstream directions. The flow of gas causes the time for the pulse traveling in the downstream direction to be shorter than for the upstream direction, and this time difference is a measure for the rate of the gas flow, see Fig. 4.1. By utilizing Equations (4.1) - (4.3), the gas volume flow rate can be calculated. In Equation (4.1), the axial flow velocity along the acoustic cord is calculated, and (4.2) gives the average flow velocity along the pipe axis. A single path ultrasonic flow meter measures the axial flow velocity along a single cord. Dependent on the flow velocity, the flow profile of the flowing gas will be some degree of a fully developed turbulent profile. In order to compensate for the flow profile to

Figure 4.1 Basic time-of-flight measurement principle.

Obtain the average axial flow velocity, some order of correction to the measured flow velocity is required. One way to utilize this correction, K, is to use the Reynold's number as a measure of the flow profile, and adjust the measured axial flow velocity according to a function based on the Reynold's number estimated. The volume flow rate at reference conditions is calculated from Equation (4.3), where the input of line pressure and temperature are required.

As can be seen from these equations, the flow velocity measured along the ultrasonic cord does not depend on pressure, temperature or any other process parameter. This is a very important characteristic of an ultrasonic flow meter, as it implies that no adjustment due to changes in e.g. gas composition is required. Accordingly, an ultrasonic flow meter should present valid measurements independent of the process conditions. Thai is, within the flow, pressure and temperature range specified for the meter in question.

In addition to the axial flow velocity, also the velocity of sound can be calculates on the basis of the time-of-flight measurements, see Equation (4.4). Once the velocity of sound, c, is known, the isentropic index, can be found using known equations from thermodynamics relating isentropic index and the density, of the gas with the state variables. Empirical formulae have been developed for finding the molecular weight and the density of the gas from the transit times (t₁₂and t₂₁). Accordingly, in addition to the volume flow rate, the Fluenta FGM 130 can also present the mass flow rate of the flare gas.

For more and more flare gas metering challenges, only ultrasonic transit-time meters are regarded to be applicable. This is, today, the only proven technology that can meet the extremely high turn-down ratios required for these applications. The NORSOK standard (2) claims an operating velocity range of 0.2 -100 m/s, giving a turn-down ratio of 500:1. One could of course discuss this requirement dependent on the installation, e.g. an LP (Low pressure) or Vent flare will generally not represent measurement challenges in this flow range. However, a flare system does usually comprise of a HP (High Pressure) flare, and potentially of an LP and/or a Vent flare. Experience has shown that the maximum specified flow range for these HP pipelines are likely to be in the 100 m/s area. From the operators'view, it is unquestionable preferable to utilize one metering instrument for all these applications, if only ultrasonic flare gas meter on the market able to operate up to three flare pipelines simultaneously with one Flow computer, see Fig. 4.2. Accordingly, one Fluenta FGM 130 can Fluenta FGM 130 with three pairs of sensors can be utilised for leak detection by measuring on different pipelines in e.g. a process plant or refinery.

Figure 4.2 The Fluenta FGM 130 Flow Computer can perform measurements
on up to three pipes simultaneously

Putting the specified upper flow velocity for flare gas meters in perspective; a hurricane is defined as wind speed of approximately 30-32. m/s. Thus, the required measurement range is more than three times the wind speed of a hurricane. Anyone having experienced a hurricane knows that both the carry along effect of voices in the wind and the general noise level is dramatic. Bearing this in mind, it is obvious that both the ultrasonic signal propagating along the measurement card and the signal processing system must be very robust in order to extract time-of-flight information under such conditions. At the same time, the meter must perform accurate measurements at the lowest flow velocities. Both the Norwegian CO₂ tax regulations (1) and the NORSOK standard (2) state measurement uncertainty limits of ±5% of measured volume flow rate for flare gas meters. This applies for the entire measurement range. The measurement uncertainty of a meter for flare gas applications shall be verified by an uncertainty analysis within a 95% confidence level.

In order to meet these requirements, the Fluenta FGM 130 utilises broadband ultrasonic transducers and Chirp excitation to obtain a Signal-to-Noise ratio requisite at the higher flow velocities. By utilizing this t4echnique, the Fluenta FGM 130 has demonstrated flow measurements up to 120 m/s in a wind tunnel test (4).

4.2 CW-Chirp Measurements

Using continuous-wave (CW) signals and phase detection at the receiver, the transit time of ultrasonic signals can be determined in order to calculate the axial flow velocity. However, at higher flow velocities, the time-of-flight measurements may become ambiguous as the

Associated change in time-of-flight become larger than one CW signal period. To eliminate this ambiguity, a Chirp signal is utilised.The whole number of CW periods, n _chirp' is measured using the Chirp signal, and added to the time obtained from phase measurements using the CW signal, tcw. Thus, the time measured using Chirp and CW is:

This technique ensures time-of-flight measurements with 100 ns resolution, giving a flow velocity measurement resolution of 0.01 m/s (36" pipe). With the changing operation of flare stacks, with zero-flaring becoming a common situation, it is very important to continuously present accurate measurement also in the lowest flow range.

Figure 4.3 The resulting signal peak of a correlation between a transmitted
signal and a pulse-compressed reference signal at the receiver.
By means of peak-detection, the time-of-flight is determined.

In order to overcome high noise levels, above 15 m/s, solely the Chirp signals are used with a correlation detection to determine the transit times, see Fig. 4.3. the important feature of the Chirp signalling method is the ability to transmit detectable signals at very low power. For hazardous area Zone 0 transducers, the level of excitation on each transducer is limited to only 1.94 W. The FGM 130 ultrasonic transducers, complying with the Ex Zone 0 requirements, are specially designed to generate a Chirp signal that is detectable at the receiving transducers (both transducers act both as transmitter and receiver) at flow velocities from 0.05 to 100 m/s.

Figure 4.4 Measurement cycles at low flow velocities, utilising both CW and Chirp signals. At flow velocities above 15 m/s, only Chirp measurements are utilised, reducing the measurement cycle to 20 ms

The details of the measurement cycle for low flow velocities are shown in Fig. 4.4. As can be seen, a train of Chirp and CW signals are transmitted every 10 ms, implying 100 measurements every second. In one measurement period, 100 measurements are carried out for Chirp and CW, both upstream and downstream. This results in an updated flow measurement value every 4-5 seconds in the low flow range. At the higher flow rates, only Chirp measurements are utilised, giving an updated flow measurement value every 2-3 seconds. The measurement rate of 100 per second ensures a fast dynamic response to rapid changes in the flare stack. This is a situation likely to occur where zero-flaring is the general picture, followed by a process shut-down or a sudden disposal of large amounts of flare gas.

4.3 Features of Importance

What is a very important feature of a flare gas meter at these conditions, with flow velocities up to and above 100 m/s, is that no meter parts intrude into the pipe cross-section. If this were the case, particles and droplets may affect the metering performance not only at the time of depositing, but also on a permanent basis if the deposits are not removed from the metering parts. At worst, the metering parts can be damaged, resulting in malfunction and erroneous readings. Generally, ultrasonic flare gas meters utilize transducers are mounted with the front centre point flush with the inner pipe wall for all pipe sizes from 6" to 72", see Fig. 4.1. However, some ultrasonic meters utilize other transducer mounting configurations, with the transducer intruding up to ? D into the pipe cross section. Depending on the application and the pipe size, this may be favourable, but the sensors will be exposed to process debris.

New technology, more powerful signal-and microprocessors have increased the availability of measurement data from an ultrasonic gas flow meter. As more and more of the signal processing can be implemented in software, both raw data and processed data can be acquired and analysed. By having this information on a digital form, e.g. trough a serial communication line, no information is lost due to non-linearities, bit resolution and offset and gain errors found in digital-to-analogue and analogue-to-digital converters. By using e.g. RS-422 or RS-485 serial communication, data can be transmitted over long distances. Combined with the Modbus protocal, data can be transferred to and from a supervisory system with high data integrity. From the Fluenta FGM 130 Flow Computer over 100 parameters are continuously available for e.g. a supervisory system. This information can be utilised for monitoring the meter performance and trending over longer time periods.

In addition, the flare gas metering system itself, being "intelligent", can utilise the internal trend information for self-diagnostics, to improve the quality of the meter performance. All measured parameters, e.g. transit times up-and downstream, pressure and temperature, reflect the process condition in the flare stack. Disproportionate change in one of these parameters in proportion to the other measured parameters could indicate an erroneous measurement condition.

4.4 Automated Condition based Maintenance

Automated condition based maintenance implies that regular service intervals on e.g. a transmitter are omitted at the expense of service only demand. This maintenance scheme requires direct information of the transmitter status, so that an evaluation of the transmitter can be carried out. If the transmitter status itself is not sufficient to give information of the transmitter condition, a duplicated transmitter solution might represent the required solution.

The Fluenta FGM 130 has implemented an interface enabling up to twelve HART transmitters to be connected to one Flow Computer. For each of the maximum three measurement systems, up to two pressure and two temperature transmitters can be configured. The Flow Computer will continuously present both the measurement values and the communication status for each of the HART transmitters to the supervisory system through the modbus serial communication link. By utilizing duplicated transmitters, the supervisory system can compare the measurement values for each transmitter and give a warning if the measurement values are adrift or the transmitter status indicates an error.

Incorporating the possibility for automated condition based maintenance, the Fluenta FGM 130 is aligned to the NORSOK standard (2), which states normative requirements in this respect.

5 CONCLUSION AND FUTURE TRENDS

As of today, the proven technology for flare gas measurement is ultrasonic time-of-flight meters. The high turn-down ratio, the fast dynamic response, the non-intrusive design and the low maintenance costs presented by this technology substantiate this. Operating in the market including the Norwegian continental shelf, operator and governmental requirements puts high demands on the metering systems in this respect. Being a world-wide supplier, Roxar Flow Measurement AS, manufacturing the Fluenta FGM 130 Flare Gas Meter, must at all times be updated on the legislation regarding flare gas measurements. Also, it is important to keep updated on the operators' requirements beyond what is given by the regulatory.

Future trends point in the direction of more intelligent, "Smart" systems, that con communicate with a supervisory system through well-defined digital protocols. Further, remote diagnostics via e.g. internet or satellite communication are areas to be looked into. A future scenario might be that the manufacturer himself, given the permission by the operator, could sit at his desk and remotely monitor a single installation for status check, possibly saving a costly service trip for doing the same job. This feature could be possible with "intelligent" systems provided with self-diagnostics capabilities, given by the powerful hardware and sophisticated software available today.

6 NOTATION

7 REFERENCES

(1) NORWEGIAN PETROLEUM DIRECTORATE. Regulations to measurement of fuel and flare gas for calculation of CO2 tax in the petroleum activities, August 1993. ISBN 82-7257-395-4.
(2) NORSOK STANDARD. Fiscal Measurement Systems for Hydrocarbon Gas. I-104. Rev. 2, June 1998. Norwegian Technology Standards Institution.
(3) RUTTEN, O. VERFAREN UND Vorrichtung zum messen von stromenden Flussigkeits-, Gas-, oder Dampfmengen. Deutches Patent No. 520484, 19 Sept. 1928.
(4) MYLVAGANAM, K.S. Ultrasonic gas flowmeters-Novel techniques of transducer orientation asnd signal processing make high-rangeability possible. Measurements & Control, Dec. 1989, pp. 122-127.