Understanding leak detection capabilities is an important element in assessing pipeline risk. Leak detection is a potential consequence mitigation measure, as detailed in Part 1 of the Leak Detection Capability Analyses discussion. The extreme variability in potential scenarios warrants a specific analysis for each portion of each pipeline, if an objective valuation is sought.
Every pipeline has multiple opportunities for leak detection, even when an operator has only identified one as their ‘official’ LDS. For instance, ROW patrol, happenstance detection by staff, landowner reporting, and passerby reporting are often not made a part of an evaluation of an LDS. While such systems should not be relied upon as a primary means of detection, they nonetheless play a role. Historically, over 60% of pipeline leaks have been detected by these opportunities rather than by a dedicated LDS. Their impact will vary depending on location-specific characteristics.
No single LDS should purport to solve all leak detection challenges. Rather, additional systems may potentially fill gaps in current LDS capabilities. In order to identify the potential gaps, LDS curves (or equivalent) must be generated.
Beyond the investigation of potential new LDS subsystems, LDS capabilities should be understood and quantified. For certain pipeline assets, this understanding must be documented, per regulatory requirement, along with a decision-process that determines sufficiency of these capabilities.
Table of Contents
Capability Curves
Point estimates for LDS capability are incomplete measures. Since most LDS sensitivities are related to leak rate or leaked volume, their ability to detect (time to detect) must be paired with either a leak rate or leaked volume. Therefore, each LDS can be characterized by a curve relating leak rate with detection time. The area under the curve is the amount of product leaked before detection.
Using the tool described below, the analyses of capabilities is demonstrated by the generation of the curves. A composite curve can be created that combines the most effective LDS available for each leak rate or volume–ie, the composite curve combines the individual LDS curves.
The sometimes-required sufficiency decision-process can then be demonstrated by adding a cost benefit analysis for each potential LDS subsystem. Employing the well-known ALARP process of risk management, all options with cost/benefit ratios up to the point of ‘grossly disproportionate’ should be undertaken in the interest of public safety. US IMP regulations mandate such decision processes for assets under the jurisdiction of that regulation.
LDS Types
While it is beyond the scope of this discussion to detail every type of pipeline LDS, it is useful to at least recognize the myriad different system types that are available.
Ref PLRMM 3rd Ed says:
Pipeline leak detection can take a variety of forms, several of which have been previously discussed. Common methods used for pipeline leak detection include
- Direct observation by patrol
- Happenstance direct observation (by public or pipeline operator)
- SCADA-based computational methods
- SCADA-based alarms for unusual pressure, flows, temperatures, pump/compressor status, etc.
- Flow balancing
- Direct burial detection systems
- Odorization
- Acoustical methods
- Pressure point analysis (negative pressure wave detection)
- Pig-based monitoring
Each has its strengths and weaknesses and an associated spectrum of capabilities.
Despite advances in sophisticated pipeline leak detection technologies, the most common detection method might still be direct observation. Leak sightings by pipeline employees, neighbors, and the general public as well as sightings while patrolling or surveying the pipeline are examples of direct observation leak detection. Overline leak detection by handheld instrumentation (sniffers) or even by trained dogs (which reportedly have detection thresholds far below instrument capabilities) is a technique used for distribution systems. Pipeline patrolling or surveying can be made more sensitive by adjusting observer training, speed of survey or patrol, equipment carried (may include gas detectors, infrared sensors, etc.), altitude/speed of air patrol, topography, ROW conditions, product characteristics, etc. Although direct observation techniques are sometimes inexact, experience shows them to be rather consistent leak detection methods.
More sophisticated leak detection methods require more instrumentation and computer analysis. See more detailed discussion of LDS techniques.
PHMSA LDS Study DTPH56-11-D-000001
This study categorizes LDS’s into ‘internal’ and ‘external’ systems, as described below:
Internal Systems
API 1130 is devoted to Internal systems, and further subdivides them into these groups:
- Regular or Periodic Monitoring of Operational Data by Controllers:
- a. Volume balance (over/short comparison)
- b. Rate of pressure / flow change
- c. Pressure point analysis
- d. Negative pressure wave method
- Computational Pipeline Monitoring (CPM)
- a. Mass balance with line pack correction
- b. Real time transient modeling
- c. Statistical pattern recognition
- d. Pressure / flow pattern recognition
- e. Negative pressure wave modeling / signature recognition
- Data Analysis Methods
- a. Statistical methods
- b. Digital signal analysis
External Systems
External leak detection is both very simple – relying upon routinely installed external sensors that rely upon at most seven physical principles – and also confusing, since there is a wide range of packaging, installation options, and operational choices to be considered. Whereas there is at least a set of API recommended practices to follow and to cite when recommending an Internal system, there is no such guideline with External systems. This often requires the engineer to make original design decisions, without the support of an engineering standard to quote. External leak detection sensors depend critically on the engineering design of their deployment and their installation. A sensor placed in the wrong location can quite easily miss an escaping plume of hydrocarbons. The number and density of placement of sensors needs to be weighed against requirements. Poorly installed sensors can perform orders of magnitude worse than
laboratory specifications.
It is useful to categorize these systems by three dimensions:
1. The physical principle that is used
2. How the sensors are packaged and deployed
3. How the system is utilized for leak detection
As we remarked earlier, there are relatively few physical principles that are actively and commonly used for hydrocarbon leak detection on pipelines:
- Sensing of the acoustic emissions of a leak
- Sensing lost product with a fiber optic cable, specially treated to change refractive index when wet with hydrocarbons
- Sensing strain and/or temperature change due to a leak with a fiber optic cable
- Utilizing conductive cables whose resistance and/or AC impedance change when wet with hydrocarbons
- Sensing hydrocarbons using permeable tubes that are swept with gas that is tested chemically for traces of contamination
- Detecting hydrocarbon vapors with chemical testers
- Detecting hydrocarbon vapors via optical methods
Most of these physical principles can be deployed using sensors in at least a couple of these packages:
a) Instrumentation attached to the pipeline
b) Point sensors
c) Continuous sensors, typically in the form of a cable
d) Hand or vehicle carried tools
e) Tools launched internally to the pipeline
Similarly, they can be utilized in several operational modes:
- i. Permanent installation with continual sampling
- ii. Permanent installation with periodic / intermittent sampling
- iii. Periodic or on-demand deployment, typically as part of a manual inspection program
An example using this scheme would be an airborne LIDAR camera that is used in monthly gas pipeline leak survey patrols would be 7.d.iii – an optical method, in a vehicle carried camera, deployed periodically. Similarly, an on-line acoustic sensor array for an oil terminal would be
1.a.i – an acoustic method, permanently attached to the pipeline, with continual sampling.
Leak Detection Capability & Valuation Analyses
The challenge in valuing the risk reduction benefit of leak detection is the extreme variability not only in risk but also in leak detection capability. The variability is driven by pipeline characteristics (diameter, wall, SMYS, etc.), operations (product, pressure, flowrates, etc.), location (topography, receptors, etc.), leak detection systems in place, and many other factors, some of which change foot-by-foot along a pipeline.
Some LDS methods rely on leak rate, some only on leaked volume, and some can find leaks based on either. This makes calculations more challenging. For example, a person on-site at a pump station may detect a leak based on sound (leak rate), smell (leak volume), sight (rate or volume), or even vibration (rate). Logically, more person-hrs on-site should result in greater likelihood of finding a leak quickly as would more instrumented capabilities.
For liquid spills, the time for a leak to develop a surface pool is an important aspect for below-grade leaks.
Risk management will sometimes point to the benefit of increased LDS capabilities. Beyond the investigation of potential new LDS subsystems, existing LDS capabilities should also be understood and quantified. For certain pipeline assets, this understanding must be documented, per regulatory requirement, along with a decision-process that determines sufficiency of these capabilities.
Spreadsheet Tool
To at least somewhat manage this variability, spreadsheet tools have been created. These are intended to assist in understanding potential application scenarios in order to facilitate discussions of valuation for specific owner/operators on specific applications.
No single LDS can replace all LDS’s. Rather, a new capability can play a valuable role for a certain set of leak scenarios. Those scenarios must be understood in terms of their potential consequences and frequency of occurrence as well as the potential benefits of earlier detection specific to those certain scenarios.
Basic Analysis
As the name implies, this is a simple tool requiring only a few inputs. Depending on those inputs, the benefit of various LDS additions can be quantified. Only the consequence-reduction aspect of risk reduction is shown in this tool. If PoF-reduction is also possible, that aspect can be added and should show additional value-added
Detailed Analysis
As with the basic analysis, the benefits of additional capability is an output of this detailed analysis, based on user-specified inputs. An exhaustive listing of factors potentially impacting the valuation is not an objective of this tool. Rather, this tool contains a higher level, but still specific listing of the most often critical variables.
For instance, the user can specify the leak rate at the pipeline (where failure has occurred, often underground) and also the effective leak rate that occurs at the surface. The vapor generation and stability at the surface is key to detection by certain LDS techniques and may differ dramatically from the subsurface release rate.
LDS Capability Curve
As a way to compile the multiple leak detection capabilities that pipeline operators generally possess, and to test the potential benefits of an addition of an LDS service, this spreadsheet tool has been created. It seeks to demonstrate where and how in the possible realm of leak scenarios, another LDS may provide added value. This tool, being less self-explanatory and also having value as part of a standalone service, is discussed below and in the following section.
In this prototype, the user approximates four detection times corresponding to leak rates, for each ‘system’ at a given location along a pipeline. The tool then fills in some intermediate points and generates a leak-rate vs time curve. The area under the curve represents the volume spilled prior to detection. curves closer to both axes have the best leak detection capabilities—ie, the smaller spill sizes prior to detection. A composite curve is also calculated and represents the total LDS capabilities, considering the contributions from each subsystem.
Curves representing subsystems can be turned on/off to perform what-if analyses.
