ABSTRACT

Ultrasonic flowmeters have been the center of attention within the natural gas industry for the last several years. To date, current commercial devices have been developed using Gaussian or proprietary integration techniques to measure the velocity of the flowing stream to eliminate the sensitivity to piping induced installation effects. When these proprietary integration techniques are applied, the ultrasonic meter is required to determine the swirl and asymmetry of the flowing stream. Published research has indicated both integration techniques are limited in their sensitivities to installation effects.

This paper presents a novel ultrasonic flowmeter concept, proposed by the author, that combines0.25 percent under field piping configurations for natural gas applications. This performance has not been demonstrated by any commercial or scientific design to date. The novel concept accomplishes this performance with fewer transducers and chordal paths resulting in considerable savings in manufacturing costs.

Combining these technologies into a flowmeter has created a method to measure the 'real time' health of the flowmeter.

To determine the validity of the invention, experiments were conducted in natural gas using a two-path ultrasonic flowmeter. Perturbation tests were conducted under the following fluid dynamic conditions:

*
fully developed flow
*
swirling flow
*
non-symmetric, non-swirling flow

The results for the experiments demonstrate the validity of the concept and the performance of this novel approach.

FLOWMETERS

Flowmeters are generally classified as either energy additive or energy extractive. Energy additive meters introduce energy into the flowing stream to determine flowrate. Common examples of energy additive meters are magnetic meters and ultrasonic meters. Energy extractive meters require energy from the flowing stream, usually in the form of pressure drop, to determine the fluid's flowrate. Examples of energy extractive meters are PD meters, turbine meters, vortex meters and head meters (orifice, pitot, venturi, etc.).

Further subclasses of flowmeters are based on determining if the meter is discrete or inferential.

Discrete meters determine the flowrate by continuously separating a flow stream into discrete segments and counting them. Inference meters infer flowrate by measuring some dynamic property of the flowing stream.


ULTRASONIC FLOWMETERS

Ultrasonic flowmeters have been the center of attention within the natural gas industry for the last decade. To date, current commercial devices have been developed using Gaussian or proprietary integration techniques. When the proprietary integration technique is applied, the ultrasonic meter is required to determine the swirl and asymmetry of the flowing stream. Published research has indicated both integration techniques are limited in their sensitivities to installation effects and have demonstrated additional bias uncertainties due to piping configurations.

Acoustic flow measurements are well known. They involve determining the average chordal velocity of the flowing stream from the difference in transit time of acoustic pulses transmitted in the downstream and upstream directions respectively between acoustic transducers. These acoustic pulses are transmitted along a chordal path, and a measure of the chordal velocity is determined from the measured transit times. The fluid can be gas or liquid.

The transit times depend on the mean velocity of the chordal path, the flow profile and the turbulence structure of the flowing stream. The reliability of the measured chordal velocity depend on the path length, the configuration and radial position of the acoustic path, the transmitted acoustic pulse form, the electronic timing and gating performance and the calculations involved in reducing the measured parameters to the mean chordal velocity.

The acoustic transducers may be mounted in an invasive or non-invasive manner. An invasive mount invades the channel's containment structure through an aperture. An invasive mount does not transmit acoustic pulses through the containment structure, sometimes referred to as 'wetted' transducers. A non-invasive mount transmits the acoustic pulses through all or part of the channel's containment structure, sometimes referred to as 'non-wetted' transducers.

The invasive mount is further classified as intrusive or non-intrusive. The intrusive term relates to a part or all of the transducer intruding into the flowing medium. The nonintrusive term defines the transducer mounting as not intruding into the flowing stream.

The acoustic paths may be arranged in a reflective, non-reflective or hybrid geometry.

A reflective path is arranged in a geometric manner to reflect one or more times off the containment structure or a reflective body installed inside the channel.

A non-reflective path is arranged in a geometric manner that does not reflect off the containment structure or a reflective body inside the channel.

A hybrid design employs any combination of both reflective and non-reflective paths and/or invasive and non-invasive configurations.

The number of paths and their placement in the channel vary among commercial and scientific designs.

STATE-OF-THE-ART

The state-of-the-art for ultrasonic flowmeters employs one of three integration methods to determine the average flowing velocity in a circular duct. The first two methods are commercially available. The third method is under development by the scientific community.

The first commercial method, known as Gaussian integration, is based on a fixed number of paths whose fixed locations and weighting factors are based on the numerical Gaussian method selected by the designer. Several Gaussian methods are available from publications (Jacobi & Gauss, Pannell & Evans, etceteras). The advantages of this approach are clear. No additional information of the flow profile is required for calculating the average flowing velocity. The weighting factors are fixed in advance as a result of the number of paths and the Gaussian method selected by the designer. The minimum number of paths is four regardless of the Gaussian method selected.

The second commercial method, which is a proprietary method, determines the swirl and/or asymmetry of the flowing stream by transmitting acoustic pulses along two or more paths having different degrees of sensitivity to swirl and to symmetry. The proprietary method uses a 'trade secret' matrix to determine the weighting factors for the chordal velocities based on the measured swirl and asymmetry. The recommended number of paths is five for the proprietary method.

The third scientific method, now under development by the National Institute of Standards and Technology (NIST) is an eleven-path arrangement. The unit, termed the advanced ultrasonic flowmeter (AUFM), is based on computer modeling of pipe flow fields and simulations of their corresponding ultrasonic signatures. The sensor arrangement for the AUFM will have enhanced velocity profile diagnostic capabilities for deviations from non-ideal pipe flows. Interpreting the signals produced by the ultrasonic sensors will be a pattern recognition system capable of classifying the approaching unknown flow among one of a number of typical flows contained in its electronic onboard library. This library will be created using results from computational fluid dynamics simulations. Both commercial methods perform well in the laboratory environment of 'fully developed' pipe flow.

In the industrial environment, multiple piping configurations are assembled in series generating complex problems for flow metering engineers. The challenge is to minimize the difference between the actual or "real" flow conditions and the "fully developed" flow conditions in a pipe to maintain a minimum error associated with the selected metering device's performance. The two state-of-the-art commercial methods attempt to accomplish this objective.

With respect to installation effects and the near term flow field, the correlating parameters that impact similarity vary with meter type and design. However, it is generally accepted that the level of sensitivity to time-averaged velocity profile, turbulence structure, and bulk swirl is dependent on the metering technology and the specific design of that meter.

Significant research from the European Gas Research Group (GERG) and the Gas Research Institute (GRI) has attempted to quantify the additional uncertainties associated with installation effects. The first commercial method has an additional bias uncertainty of ±0.0 to 3.0 percent due to various piping configurations. The second commercial method has an additional bias uncertainty of ± 0.0 to 1.0 percent due to various piping configurations. Obviously both methods have clear disadvantages in 'real' performance to the user community.

THE ARTEFACT PACKAGE

A novel ultrasonic flowmeter concept combines the strengths of acoustic and isolating flow conditioner technologies to determine the flow velocity and/or throughput in a channel. The performance of this novel concept exceeds the current technology performance by an order of four to twelve times and has significant savings in manufacturing costs. The novel approach allows for creation of a method to measure the 'real time' health of the flowmeter.

The isolating flow conditioner eliminates swirl (less than 2° of swirl) and provides an axisymmetric velocity profile (±5 percent between parallel chords) upstream of the acoustic path(s). Acoustic pulses are transmitted along a chordal path, and a measure of the chordal velocity is determined from the measured transit times. An individual chordal weighting factor is applied to the chordal velocity to obtain the average flow velocity and/or throughput of the medium. An individual calibration factor for the chord, based on laboratory testing, may be applied in lieu of the weighting factor or in addition to the weighting factor.

The flowmeter uses a fixed weighting factor based on the position of the acoustic path(s) and the turbulence level of the flowing medium. For two-path or more designs, the weighting factor may be correlated on the chordal position, a relaxation term related to the profile development and the turbulence level of the flowing medium.

Combining these technologies into a flowmeter has created a method to measure the 'real time' health of the flowmeter. A one-path design provides a low-level 'real time' health of the flowmeter. A two or more path design provides a high-level 'real time' health of the flowmeter.

In the industrial environment, a flowmeter with these built-in diagnostic capabilities is referred to as a 'smart' or 'intelligent' flowmeter.

EXPERIMENTAL APPARATUS

To determine the validity of the single and multi-path ultrasonic designs, experiments were conducted in natural gas at the Gas Research Institute's Meter Research Facility under the auspices of Southwest Research Institute. Independent research has been conducted extensively on 200mm meters with both single path and multi-path ultrasonic designs. The pipe velocity was varied from 1.5 to 21.3 mps (5 to 70 fps) resulting in pipe Reynolds numbers from approximately 600,000 to 7,500,000.

Perturbation tests were conducted under the following fluid dynamic conditions:

* fully developed flow
* swirling flow
* non-symmetric, non-swirling flow

Fully developed flow was established with the use of an isolating flow conditioner, a minimum of forty diameters (40D) of straight pipe, a tee mounted in the same plane and approximately eighty diameters (80D) of straight pipe prior to the test section.

Non-symmetric, non-swirling flow was established with the use of an isolating flow conditioner, a minimum of forty diameters (40D) of straight pipe, and a tee mounted in the same plane prior to the test section. Swirling flow was established with the use of an isolating flow conditioner, a minimum of forty diameters (40D) of straight pipe, followed by a ninety-degree (90°) elbow and a tee out of plane prior to the test section. This combination has been known to generate swirl angles of fifteen to twenty degrees (15° to 20°)

SINGLE PATH OR MORE RESULTS The experiments demonstrated the validity of the novel concept. The single-path approach demonstrated an uncertainty of ±0.50 percent of actual flowrate in both perturbed and 'good' flow conditions for velocities greater than 3 mps (10 fps). While this performance equals the five-path proprietary design discussed previously, it achieves this performance with one-fifth of the transducers and chordal paths.

To determine the validity of the two-path or more invention, experiments were conducted in natural gas using a two-path and a threepath invention.

Again, the experiments demonstrated the validity of the novel concept. The two-path designs demonstrated an uncertainty of ±0.25 percent or better of actual flowrate in both perturbed and fully developed flow conditions for velocities greater than 3 mps (10 fps). While this performance exceeds the five-path proprietary design or the four-path Gaussian designs by an order of two to six times, it accomplishes this performance with at least one-half of the transducers and chordal paths.

The three-path design demonstrated a performance of ±0.15 percent of actual flowrate in both perturbed and fully developed flow conditions for velocities greater than 3 mps (10 fps). This performance exceeds the five-path proprietary design or the four-path Gaussian designs by an order of four to twelve times. This performance has not been demonstrated by any commercial or scientific design to date. The novel concept accomplishes this performance with fewer transducers and chordal paths resulting in considerable savings in manufacturing costs.

A four or more path design is predicted to have a performance of ±0.10 percent or better of actual flowrate in both perturbed and fully developed flow conditions for velocities greater than 3 mps (10 fps). Of course, it is important to note that the claimed uncertainty for state-of-the-art worldclass flow laboratories is approximately onefourth of one percent (±0.25%) using natural gas or air as the flowing medium.


SMART' OR 'INTELLIGENT' FLOWMETER

Combining these technologies into a flowmeter has created a method to measure the 'real time' health of the flowmeter. A onepath design provides a low-level 'real time' health of the flowmeter. A two or more path design provides a high-level 'real time' health of the flowmeter. In the industrial environment, a flowmeter with these built-in diagnostic capabilities is referred to as a 'smart' or 'intelligent' flowmeter.

Due to the brevity of the paper, it is not possible to explore the 'real time' health monitoring of the flowmeter.

However, the following is the VOS residual analysis for all meter designs for the complete experimental pattern.