June 2017, Vol. 244, No. 6

Features

Measuring Through Valve Gas Losses Using Acoustic Emission

The detection of leaks by acoustic methods was used well before the technology of acoustic emission (AE) was applied elsewhere. However, the equipment available was susceptible to noise from sources other than leaks, such as that from normal plant operation and the surrounding environment.

As the AE industry grew in other areas, technology became available that largely overcame the problems of environmental noise. Soon, a trials program carried out by BP from 1982-84 deemed the AE method the most promising way to quantify leakage through valves.

The field development program lasted six years and involved removing over 800 leaking valves from service and retesting them in the laboratory in order to show a relationship between physical loss and acoustic signal level.

Following this, a “best fit” correlation was developed so the technology could be easily applied. Factors having a significant effect on the acoustic signal level include valve type, size and differential pressure. This development was followed by the commissioning of a new instrument that could record field measurements for use in the database.

The new instrument, the VPAC II, from Mistras Group, is intrinsically safe, portable and simple to use because of all the measurement functions being automatic.

The instrument, together with its valve-leak technology package, has become widely used in the oil and gas industry to identify and estimate through-valve gas losses, enabling quick cost-effective operational and maintenance decisions without plant disruption. Personnel can be trained to use the equipment in one hour.

Detection Principle

The source of acoustic emissions should be considered initially. These are generated by a fluctuating pressure field associated with turbulent flow of the fluid at the leak site. The conditions for turbulent flow are met when the inertia effect of the fluid flow overcomes the viscous drag; the ratio of the former to the latter is defined as the Reynolds number. Turbulence has been found when the number is between 103-104. Acoustic emission, therefore, is an effective method to detect through-valve leaks, where the velocity across the leak is sufficiently high with respect to the size of the leak orifice to produce a Reynolds number in this region.

Quantification

Calculating the flow rate in a cylindrical orifice was simple. However, this was far removed from the real situation in which the leak is likely to be other than a cylindrical orifice and could be caused by smaller leaks around the entire valve seat.

As a result, an alternative method of correlating the AE received and the flow rate through the leak was required. This was achieved empirically by testing 800 valves in the field and repeating the tests with the valves removed to a flow rig. Valves from 1-18 inches and covering a range of types were used in this exercise. A database of results was compiled from which a predictive equation was derived. It is this predictive equation that allows the quantification of through valve leaks in the field.

Test Method

The operation of the instrument is simple. The sensor is held in contact with a flat surface (Figure 2), using a suitable acoustic couplant, such as grease, on the valve to be tested.

The current value of the signal level (dB) is noted. This may be stored with a single key-press in one of 300 memory locations. If a leak is indicated by a reading greater than normal background (12-6 dB), readings are taken on the pipework upstream and downstream of the valve.

As the signal level will be highest close to the leak and lower the farther it is away, these upstream and downstream figures will be lower if the valve is truly the source of the acoustic emission. The reading is then inserted into a PC spreadsheet along with the other relevant information, such as valve inlet size, differential pressure across the valve and valve type.

This information is used in the spreadsheet by the predictive equation to calculate the loss rates. The spreadsheet is often modified to present the loss rate in convenient units such as tons/year, m3/day or even product value/period.

Figure 1: Suggested test points

Current Experience

The leak-detection system has been licensed for use on over 200 sites and has proved capable of quickly surveying large numbers of valves and estimating losses from the leaking valves. In one offshore survey, 20 valves were tested and the results recorded in just over an hour. In this one small survey, leaks totaling 5,100 liters/min were identified, equating to 4,500 tons/year.

Other successes have been commonplace with the largest leak found totaling over 3,800 liters/minute from a single 24-inch valve. In one facility, a 4-inch PRV with a signal level of 85db was found, equating to 1,100 tons/year. In another, four 24-inch control valves were tested. Two were shown to leak over 2,500 tons/year. One oil company identified losses of $14 million from four facilities.

Not all large-scale losses are caused by damaged valves; leaking control valves are a common problem and these often require only a minor adjustment. One particular 1-inch valve, shown to lose the company $20,000 a year, was fixed on the spot by simply adjusting the stop. An offshore PCV was discovered losing 500,000 Mscf/d, equating to $300,000 per year.

Troubleshooting

The system is also being widely used for operations troubleshooting and maintenance where the value of the losses may not be the most significant feature. This is particularly true for offshore projects where the value of the gas is not high. Its use is now written into several maintenance procedures in the North Sea.

Initial surveys will quickly highlight the large-scale losses which can then be dealt with. This is where the sensitivity of the system offers a significant advantage. Losses of as little as 1 liter/minute are detectable in the field. A facility suffering virtually no background losses can now quickly track down leaks as they occur, achieving in a couple of hours what used to take two to three days.

Even in apparently noisy environments it is possible to detect extremely small leaks. This is due to a sensor design that effectively rejects vibrational noise and advanced signal processing in the instrument. A leak of only 1.5 liters/minute was detected in a relief valve in an offshore gas production platform compressor module. The background signal level on the 5131 sensor was no higher than normal even in this high-noise environment.

Losses also have secondary effects. One user reported hydrogen as its production bottleneck. Each ton of hydrogen was used to make 30 tons of product. A control valve was identified as losing 770 tons of hydrogen a year to flare. This equates to a loss of production of 23,100 tons a year. These valves are now checked on a regular basis.

Many governments are concerned by environmental damage caused by excessive release of hydrocarbons into the atmosphere. Operators working in countries that lack strict environmental policies know it is only a matter of time before they too will face strictest emission regulations.

Further Developments

The detection of through-valve leakage is not confined to gas systems. Where there is sufficient differential pressure to satisfy the conditions for turbulent flow, liquid leakage can also be readily detected. The database of results on liquids has been built up, allowing quantification of through-valve liquid losses. Work is ongoing to expand this database which will further improve the liquid leak correlation to larger valve sizes.

A program run offshore and in the workshop extended the procedure and correlation to large offshore emergency shutdown valves of up to 48 inches in diameter, in both gas and liquid service, with soft and hard seats. The purpose was to replace the statutory SI1029 test requirement, requiring long platform shutdowns, with a test that could be applied rapidly during any temporary production stop, saving $400,000 per year in the Forties field.

In addition to the detection of liquid through valve leaks, the system is being used to detect the leaks causing sand erosion of valve bodies. This method can also detect sand erosion at vulnerable areas in pipework.

Handheld AE leak-detection systems are also identifying and quantifying through-valve losses of normally closed steam valves for power generation. Identifying the valves allowing the greatest loss is a critical tool in the financial management of these plants.

Conclusions

The measurement of through-valve losses helps identify significant cost savings through loss control and allows for quantitative prioritization of maintenance-based information. This leads to speedy, effective operations and maintenance troubleshooting, along with control over emissions.