SYSTEM PINPOINTS LEAKS ON POINT ARGUELLO OFFSHORE LINE

Jay L. Sperl
Chevron Pipe Line Co.
Gaviota, Calif.

Chevron Pipe Line Co. has installed a leak-detection system on its Point Arguello natural-gas pipeline that directly monitors leaks and does not alarm as a result of flow disruptions.

Because of its response time, however, Chevron sees the Leak Alarm System for Pollutants (LASP) as a supplement to more conventional systems, rather than a substitute.

LASP has an advantage over many other systems by being a "direct" system instead of inferential. That is, the leaking liquid or gas is detected, not a condition caused by the leak, such as pressure loss or volume imbalance.

POINT ARGUELLO LINE

LASP was installed on 10 miles of the 17-mile onshore segment of the Point Arguello natural-gas line during the fall and winter 1989. Built in 1987, this pipeline is part of the Point Arguello project (Fig. 1) on the California coastline near Santa Barbara.

The project consists of three Outer Continental Shelf platforms, parallel crude and gas pipelines, a gas plant, and a marine terminal. The facilities are each owned by a consortium of up to 18 petroleum companies.

Chevron operates two of the platforms, both of the pipelines, and the gas plant, while Texaco operates the other platform and the marine terminal. The gas pipeline was designed to carry produced natural gas from the platforms to the gas plant and was built with a pressure-point analysis (PPA) leak-detection system which uses pressure changes in the gas flow to detect leaks (OGJ, Dec. 19, 1988, p. 48).

The pipeline, however, has never been operated as it was intended because the facilities, finished in late 1987, have yet to go into production. Instead, Chevron currently buys gas from the utility company and transports it from the gas plant to the platforms to operate turbines.

Most recently this delay in production has resulted from a dispute over the method of crude oil transportation out of the County of Santa Barbara. But originally the delay was caused by an increased assessment of the potential hydrogen sulfide (H2S) concentration from the wells.

In August 1985, Chevron obtained the final development plan (FDP) from the County of Santa Barbara. Unfortunately, this plan and its corresponding Environmental Impact Report (EIR) were based on an assessed H2S concentration of 7,000 PPM.

In December 1987, the county received information that the H2S concentration could reach as high as 20,000 ppm. The county decided this higher concentration may not have been properly addressed or mitigated in the EIR and, therefore, required a Supplemental Environmental Impact Report (SEIR) to determine if the Point Arguello project was still in substantial conformity with the FDP.

The 1988 SEIR concluded that the gas pipeline had "no leak detection for holes smaller than 1/2 in." or when a pig was in the pipeline.

Further, the report recommended LASP as a viable mitigation measure, even though it had a response time of several hours.

Subsequently, Chevron agreed to install LASP or a similar system in the more densely populated areas. In so doing (along with other measures not in the scope of this article), Chevron gained a finding of substantial conformity.

As a precursor to design, Chevron briefly reviewed other underground leak-detection systems. Because LASP had been recommended in the SEIR, however, and in Chevron's brief 1988 review it appeared to be the state-of-the art underground leak-detection system, the decision to use LASP was straightforward.

SYSTEM'S EVOLUTION

LASP is an air-sampling system developed because conventional leak-detection systems did not offer satisfactory sensitivity of leak location.

It is able to provide direct leak detection instead of inferential. That is, the leaking liquid or gas itself is detected, not a condition caused by the leak, such as pressure loss or volume imbalance.

This offers increased sensitivity because the system is not affected by operational transients.

LASP development was initially supported by the German Ministry of Research and Technology. Its first installation was on an ethylene pipeline in West Germany in 1978. That system has operated since then with no failures, while detecting and locating numerous small leaks.

LASP is based on the principle of diffusion. The key component is a semipermeable sensor tube, which will be described later. This tube is characterized by a high permeability for gas and liquid vapors but is impervious to water.

If vapor is present outside the sensor tubing, it will diffuse into the tubing because of the concentration gradient. Air will then carry the vapor down the tube into a detector-pump unit where the hydrocarbon detector will sense the vapor concentration.

A typical monitoring section for this original system is shown in Fig. 2. This original LASP was built to detect extremely small leaks (0.4 l./day on a liquid pipeline) over a long period of time.

To do this it evacuated (pulled the air out of) the sensor tubing just once every day. This allowed the leaking fluid's vapor to form quite a large concentration slug in the sensor tube, large enough for the detector to sense an extremely small leak.

The original system (Pig. 2) had three components:

A sensor tube with a semipermeable ethylene vinyl acetate (EVA) membrane which allowed any fugitive vapors to permeate into it
A dryer unit (silica-gel dryer/carbon filter) at one end of the tube to ensure the air pulled into the end of the sensor tube was dry and hydrocarbon free
A detector-pump unit at the other end of the tube which provided a small vacuum on the tube to pull the sample past a hydrocarbon detector.

Teledyne Geotech, Garland, Tex., purchased the worldwide patent license to LASP in 1983. Teledyne distributes LASP worldwide but has licensed Siemens to distribute LASP in Europe for 5 years.

This original system provided for leak detection alarming once every 24 hr. Chevron, however, required a continuous monitoring and alarming leak-detection system which would alarm on any "dangerous" leak within 30 min, not 24 hr.

Fortunately, by 1988 the needs of another client had encouraged Teledyne to experiment with changes which would shorten the response time of LASP.

Teledyne had developed a version of the sensor tube with the EVA membrane thickness reduced to 0.015 in. This reduced to approximately 10 min the time it took vapor to pass from the outside of the sensor tube to the inside.

Teledyne found that further reduction in the EVA thickness made the tubing too delicate to work with efficiently. In addition, Teledyne set up the detector-pump unit to evacuate the tube continuously.

They called this continuous operating mode the "Emergency Response Mode." This was the LASP Chevron reviewed as it started design.

TWO-PART DESIGN

Design consisted of two parts: Component design and component location selection. The LASP had to satisfy the following criteria:

Continuous leak-detection monitoring and alarming
Response to any leak within 30 min
All installations less than 18 in. abovegrade
Length of monitored right-of-way more than 10 miles
Accessible
Not an obstacle or an eyesore.

In the spring of 1988, Chevron began working with Teledyne to design a system that would satisfy these criteria.

When Chevron began estimating the cost of the system, it estimated that installation was going to be the largest cost item.

Consequently, Chevron decided to install two sensor tubes to decrease the likelihood of ever having to replace any tubing. Discussions with Teledyne led to the idea of building a dual system.

This system would operate in two different modes: The low-level leak detection mode similar to the original LASP mode, and the emergency-response mode which was renamed the high-level leak detection mode.

The basic control logic of this dual system is that at any given time, one tube is in the low-level mode and the other tube is in the high-level mode. The tubes are switched between the modes at a user-defined interval, usually 12 hr, by a microprocessor-controlled solenoid valve.

This enables the system to detect extremely small leaks over an extended detection period like the original system, yet also provide continuous monitoring and alarming for relatively larger leaks.

(Extremely small leaks cannot be detected in the high-level mode because of the dilution caused by the moving air in the tubing.)

Both modes detect leak levels well below more conventional systems.

Low-level leak detection requires a long diffusion time. This is accomplished by letting the sensor tube sit dormant with no air flow for a user-defined length of time, usually 12 hr.
After 12 hr, the vapor contents of the entire length of sensor tube are moved past the detector.
As each analog sample is received, it is compared to an historical standard sample that defines the characteristics of the sample vapor over that length of the tube. If the new sample deviates from the historical sample by more than a user-defined amount, a leak alarm is generated in the form of a digital output.
Additionally, if a low air velocity of approximately 3 fps is maintained and the length of the buried sensor tube is known, small leaks can be detected and located.
Once the air is evacuated from the length of the tube, the control logic switches to the high-level leak detection mode as the pump continues to operate on the same tube.

The analog inputs are averaged for a user-defined time period, usually 1 min, and compared for a deviation from the previous time period average.

If the new value deviates from the previous average by more than the alarm limit and continues to ramp up at this rate for a selected consecutive number of times, usually five, a leak alarm is generated in the form of a digital output.

This high-level mode runs continuously for the selected diffusion time (12 hr) and then switches to the other tube and the sequence begins again.

Once Chevron agreed to the operation of the system, the component design was quickly completed. The original components' functionality remained fundamentally the same except for three key changes: The sensor tube has the thinner EVA membrane, the sensor tubing is buried in pairs, and the detector-pump unit (renamed "monitoring station") evacuates a tube continuously.

Additionally, another component was added, a test point to be installed near the center of each monitoring section.

COMPONENTS

Fig. 3 shows the sensor-tube construction. The outer layer of the tube is a protective sheath. Its sole purpose is to protect the EVA membrane.

The EVA membrane allows vapors to pass through but blocks liquids. The perforated support tube transports the vapors to the monitoring station once they pass through the EVA membrane.

Often called "End Point," the dual dryer unit ensures that the air entering the end of the sensor tubes is moisture and hydrocarbon free. It contains two 35-lb, silica-gel dryers, two carbon filters, and two flow switches.

The flow switches serve as an alarm should either of the buried sensor tubes become damaged. This unit has no power requirements but does require two alarm pairs for the flow switches.

These pairs were directly buried with the sensor tube for all monitoring sections, even though the cable is not shown in the trench sections.

The monitoring station provides the vacuum on the sensor tubes and contains the detectors which check the incoming sample for atypical vapor concentrations.

The monitoring station parts, housed in a stainless steel NEMA 4 enclosure, include a system controller, a vacuum pump, a flow controller, an H2S detector, a hydrocarbon detector, three flash arrestors, a water filter, and three solenoid valves.

A block diagram of the monitoring station is shown in Fig. 4. This station requires 240-v ac power and an alarm cable for the three digital output alarms.

The three alarms are a low-level hydrocarbon alarm (the H2S detector does not react quickly enough for the spiked shape of this input), a high-level H2S and hydrocarbon alarm (outputs are hardwired together), and a system alarm (any failure of the system outside of a leak alarm).

These alarms are transmitted to the supervisory control and data acquisition (scada) system either by multi-pair cable hardwired from the monitoring station to existing scada remote terminal units (RTUs) or by radio to the RTUS.

The local interface to the LASP monitoring station is a laptop computer.

Many parameters are user-definable so that the system can be used in a variety of installations without modification to the basic logic. The laptop computer is also used to retrieve low-level leak data for further analysis.

The test point is simply a three-valve manifold (one for each tube) near the center of each monitoring section.

This valve provides a port for gas injection for testing, as well as marker-gas injection to aid in the location of any anomaly on the monitoring section.

COMPONENT LOCATION

Concurrent to the design and manufacture of the components, the field location of each component had to be selected to satisfy the following criteria: 30-min response to any leak; accessible; not an obstacle or an eyesore.

The vacuum pump moves the air in the sensor tubing most effectively at approximately 3 fps (6 l./min).

Given the 1 0 min it takes vapor to pass through the EVA membrane, Chevron limited the transport time to 20 min, which computes to a maximum tubing length of 3,600 ft. To be safe, Chevron kept all monitoring section lengths to less than 3,200 ft, some as short as 1,600 ft.

As a result, Chevron had to install 21 independent monitoring sections to monitor more than 10 miles of right-of-way. A typical monitoring section is shown in Fig. 5.

Next, Chevron tried to locate the monitoring stations in areas easily accessible for both maintenance personnel and to utility termination, both power drops and alarm cabling. Part of this included locating the monitoring stations back-to-back because this was the only component which required power. This reduced the number of power drops required. It also allowed use of a single multipair alarm cable for the two monitoring stations.

Finally, Chevron examined each site and tried to place the monitoring stations, test points, and end points at the edges of fields so each was out of the way of local residents and in the most inconspicuous places possible, while still satisfying the other criteria.

INSTALLATION

Finally, in the fall of 1989, installation of LASP with conventional methods started. The system was installed in eight steps:

Trenching
Laying sand cushion in bottom of trench
Laying sensor tubing on sand
Pouring sand over tubing
Laying any required cables
Backfilling in two lifts
Restoring the area
Installing the dual dryer units and the monitoring stations (delayed until all sensor tubing was installed).

A 3-ft deep trench was used as shown in Fig. 6. This meant the sensor tubing was from 2 to 15 ft from the pipeline.

Preferably the sensor tube would be buried with the pipe, but because LASP was installed after the pipeline, this would have required a huge increase in the work and cost.

In reviewing options, Chevron found information which indicated that any gas leaking at the pipeline pressure (900 psi) would permeate the soil instantly. Therefore, 3 ft was determined an adequate trench depth.

The contractor used one crew for the entire installation. Work commenced in early August 1989, and the sensor tube installation was finished by December 1989.

The dual dryer units and the monitoring stations were then installed in January and February 1990. The original installation plan had called for multiple crews, but one crew was actually more efficient, especially when it came to handling the sensor tubing carefully.

Because the semipermeable sensor tubing is the key component to the entire system, proper handling of the tubing is critical. The 0.015 in. EVA membrane is delicate, and if it is punctured, water could enter the tubing and block the transportation of vapor samples to the monitoring station.

One installation challenge was posed by gophers gnawing through the original polyethylene protective sheath and the EVA membrane after a short period of construction (8,000 ft). The original protective sheath was made of a "gopher-resistant" polyethylene braid, which obviously did not work.

This forced rebraiding of the tubing with stainless-steel (over the original sheath). Not only did this eliminate the gopher problem, it also improved the durability of the sensor tubing, making the sensor tubing much easier to install.

Following at least five puncture leaks during the installation of the first 8,000 ft of nonstainless-steel braided tubing (later replaced), Chevron had only one leak during the installation of more than 100,000 ft of stainless-steel braided tubing.

Still, Chevron handled the sensor tubing with kid gloves. It was laid by hand and pressure tested (7 psi) following installation to ensure its integrity. Moreover, Chevron required each person who handled the tubing to read tubing handling instructions before starting work.

The initial system costs were overestimated, as shown in Fig. 7. Note that construction support and field design were not estimated costs in the original estimate.

The cost overestimation resulted from a lack of prior experience with LASP. Chevron expected that peculiar construction methods might be required to install LASP, which proved not to be the case.

Furthermore, the scope was not ensured until late in the design process. This prompted larger contingencies than were necessary. The relative breakdown for each cost category is shown in Fig. 8.

LESSONS

Several changes would be made where possible if the project were to start over today:

Avoid retrofit installations. Construction was 51% of the total LASP cost at Point Arguello (Fig. 8). This installation was done after pipeline construction, thus termed retrofit installation.
It may be impossible to avoid retrofit installations at times, but if it appears such a leak detection system may be required at a later date, put it in with the pipeline. Chevron estimates 80% of the installation costs could have been saved if this system had it been installed during the initial pipeline construction.
Use a vibratory hoe instead of trenching. If retrofit construction cannot be avoided, LASP may be able to be installed with a vibratory hoe.
Essentially, this is a deep (3-4 ft) plow that vibrates slightly as it is pulled in the ground behind a tractor. The sensor tubing is fed down a tube to a depth of 3-4 ft.
Earth falls in on the tubing as it is laid in place, allowing installation in one pass.
Teledyne has run preliminary tests on this type of installation and it has worked well on a small scale. Not only would this reduce right-of-way damage, but installation would be much more rapid and less expensive.
Ensure protective sheath is sufficient for environment. There is a large gopher population at Point Arguello. The sensor tubing, therefore, had to be gopher-resistant. Eventually, Chevron installed stainless-steel braided tubing in lieu of the polyethylene braided tubing.
Teledyne has recently revised this protective sheath with a new polyethylene braid which is quite an improvement over the original.
However, if gophers are a problem in your area, make sure this sheath can withstand their gnawing. We were quite pleased with the stainless-steel braid, but if this new sheath tests satisfactorily against gophers, it will be less expensive than the stainless-steel braid.
Use perforated tubing more frequently. When Chevron had the gopher problem, it also experimented with installing the sensor tubing inside a 4-in. perforated tube (Fig. 6).
Although the stainless-steel braid was eventually chosen, Chevron found that installing this perforated tube and then pulling the sensor tubing through this perforated tube eliminated the need for the protective bedding of sand.
This kept large sand-hauling equipment off steep terrain, reducing the revegetation effort in these areas. And, if it should ever be necessary, the sensor tubing can be easily replaced in these areas.
Install all components in underground vaults. Although the only abovegrade component installed was the monitoring station, Chevron has received complaints that these are eyesores. In the future, these enclosures might be installed underground as well.

It would cost only a little extra but at the same time eliminate the complaints and reduce vandalism.

OPERATING EXPERIENCE

Although the gas pipeline is not operating as it was intended, Chevron has been operating in a "gas buyback" mode. That is, the company has been buying gas from the utility and transporting it to the platforms to operate turbines.

This has provided 11 months' operating experience with LASP which has proven very sensitive. Most alarms have been a result of settings being too low. Therefore, adjustments to these values are continuing to be made to avoid overly sensitive alarms.

Teledyne continues to research alternate detectors (infrared, for example) which have greater resolution to specific gases in a given vapor.

Interestingly, others' "leaks" have been discovered.

The County of Santa Barbara has a fire station near one LASP section. Occasionally they use a fire-fighting foam which contains small amounts of alcohol.

When they wash this foam out of the trucks and let the residue run down the storm drain, it soaks into the soil over the LASP section. Within a day, an alarm due to this residue is recorded.

Additionally, a water line ruptured near another LASP section and washed out the soil around the sensor tube. As the water association was repairing its pipe, the PVC glue set off an alarm.

Many operators have inquired about the quantitative capabilities of LASP. Although this was not a concern when Chevron installed LASP, the company has found that the low-level mode allows concentration analysis because the concentrations actually equalize across the EVA membrane.

However, the high-level mode does not allow concentration analysis because the moving air dilutes the sample before the concentration can equalize across the EVA membrane.

Unfortunately, fears of natural vapors have been realized.

Three sections have continued to alarm, up to 60 ppm, on naturally occurring vapors. This problem continues to be addressed.

Alternate detectors are available that are insensitive to many natural vapors. In Chevron's case, however, the natural vapors being detected are too similar to the natural-gas spectrum to allow a change in detectors.

During commissioning, Chevron had to replace the vacuum pumps, the flow controllers, numerous controller boards, and the system software ROM.

The original pump was of adequate quality but oversized. This was not a problem when the pump ran infrequently, but when running continuously it tended to overheat.

It was replaced with a smaller pump.

The flow controllers were recalled by their manufacturer because they were part of a defective batch. The replacement flow controllers, the same model, are operating correctly.

Unfortunately, no cause for controller-board failure has been found.

But Teledyne has continued to repair all that have been sent back.

Teledyne has updated the ROM-resident software numerous times to correct minor bugs discovered. In addition, Teledyne continues to improve the user-friendliness of the software.

The only regular maintenance required is changing the filters (84 dryer filters every 3 months), inspecting the equipment for loose connections, and calibrating the detectors annually.