Detecting 0.6% leaks in 13 minutes on highly transient networks.
Detecting 0.6% leaks in 13 minutes on highly transient networks.
How Pipewise delivers faster leak detection with fewer false alarms, especially during pump starts/stops and other operational transients.
Minute-level detection of small leaks that are often below meter accuracy limits
Minute-level detection of small leaks that are often below meter accuracy limits
Minute-level detection of small leaks that are often below meter accuracy limits
More robust performance during transients and short data interruptions
More robust performance during transients and short data interruptions
More robust performance during transients and short data interruptions
Fewer false alarms during normal operations (pump starts/stops, network changes)
Fewer false alarms during normal operations (pump starts/stops, network changes)
Fewer false alarms during normal operations (pump starts/stops, network changes)
For operators managing dynamic, multi-inlet/outlet networks
For operators managing dynamic, multi-inlet/outlet networks
For operators managing dynamic, multi-inlet/outlet networks
Case study profile.
Case study profile.
Case study profile.
A large Canadian producer operates a highly transient condensate pipeline network with frequent operational changes. In this environment, traditional leak detection approaches can struggle, either detecting too slowly or generating alarm noise during normal operations.
A large Canadian producer operates a highly transient condensate pipeline network with frequent operational changes. In this environment, traditional leak detection approaches can struggle, either detecting too slowly or generating alarm noise during normal operations.
A large Canadian producer operates a highly transient condensate pipeline network with frequent operational changes. In this environment, traditional leak detection approaches can struggle, either detecting too slowly or generating alarm noise during normal operations.
A large Canadian producer operates a highly transient condensate pipeline network with frequent operational changes. In this environment, traditional leak detection approaches can struggle, either detecting too slowly or generating alarm noise during normal operations.
A large Canadian producer operates a highly transient condensate pipeline network with frequent operational changes. In this environment, traditional leak detection approaches can struggle, either detecting too slowly or generating alarm noise during normal operations.
A large Canadian producer operates a highly transient condensate pipeline network with frequent operational changes. In this environment, traditional leak detection approaches can struggle, either detecting too slowly or generating alarm noise during normal operations.
Midpoint risers introduce elevated leak risk due to exposure, pressure changes, and localized failure modes, making rapid detection essential to minimize volume loss and response time.
Asset type — condensate pipeline network
Configuration — 8" network with 17 inlets and 2 outlets.
Average throughput — 42,000 bbl/day (6,700 m³/day).
Operating context — highly transient (frequent flow changes from pumps turning on/off and shifting network conditions).
Midpoint risers introduce elevated leak risk due to exposure, pressure changes, and localized failure modes, making rapid detection essential to minimize volume loss and response time.
Midpoint risers introduce elevated leak risk due to exposure, pressure changes, and localized failure modes, making rapid detection essential to minimize volume loss and response time.
Asset type — condensate pipeline network
Configuration — 8" network with 17 inlets and 2 outlets.
Average throughput — 42,000 bbl/day (6,700 m³/day).
Operating context — highly transient (frequent flow changes from pumps turning on/off and shifting network conditions).
Midpoint risers introduce elevated leak risk due to exposure, pressure changes, and localized failure modes, making rapid detection essential to minimize volume loss and response time.
Midpoint risers introduce elevated leak risk due to exposure, pressure changes, and localized failure modes, making rapid detection essential to minimize volume loss and response time.
Asset type — condensate pipeline network
Configuration — 8" network with 17 inlets and 2 outlets.
Average throughput — 42,000 bbl/day (6,700 m³/day).
Operating context — highly transient (frequent flow changes from pumps turning on/off and shifting network conditions).
Midpoint risers introduce elevated leak risk due to exposure, pressure changes, and localized failure modes, making rapid detection essential to minimize volume loss and response time.
Traditional volume balance approaches can be compromised by normal operating events (pump starts/stops, network changes, pigging), which can distort signals and contribute to false alarms. In this case, the leak was also below typical meter accuracy thresholds, making detection even harder for conventional methods. The operator needed leak detection that could:
- Detect small leaks quickly (minutes, not hours)
- Remain reliable during transient operations
- Reduce false alarms and “lost leak signals” mid-event
Traditional volume balance approaches can be compromised by normal operating events (pump starts/stops, network changes, pigging), which can distort signals and contribute to false alarms. In this case, the leak was also below typical meter accuracy thresholds, making detection even harder for conventional methods. The operator needed leak detection that could:
Traditional volume balance approaches can be compromised by normal operating events (pump starts/stops, network changes, pigging), which can distort signals and contribute to false alarms. In this case, the leak was also below typical meter accuracy thresholds, making detection even harder for conventional methods. The operator needed leak detection that could:
- Detect small leaks quickly (minutes, not hours)
- Remain reliable during transient operations
- Reduce false alarms and “lost leak signals” mid-event
Traditional volume balance approaches can be compromised by normal operating events (pump starts/stops, network changes, pigging), which can distort signals and contribute to false alarms. In this case, the leak was also below typical meter accuracy thresholds, making detection even harder for conventional methods. The operator needed leak detection that could:
Traditional volume balance approaches can be compromised by normal operating events (pump starts/stops, network changes, pigging), which can distort signals and contribute to false alarms. In this case, the leak was also below typical meter accuracy thresholds, making detection even harder for conventional methods. The operator needed leak detection that could:
- Detect small leaks quickly (minutes, not hours)
- Remain reliable during transient operations
- Reduce false alarms and “lost leak signals” mid-event
Traditional volume balance approaches can be compromised by normal operating events (pump starts/stops, network changes, pigging), which can distort signals and contribute to false alarms. In this case, the leak was also below typical meter accuracy thresholds, making detection even harder for conventional methods. The operator needed leak detection that could:
Pipewise applied their Sequential Probability Ratio Test (SPRT) for high-sensitivity leak detection, combined with a probabilistic filter that identifies operational changes and filters their impact on the leak detection signal. SPRT is designed to improve key leak detection performance indicators, particularly reliability, sensitivity, and robustness under changing conditions and data interruptions.
1) SPRT: High-sensitivity evidence accumulation
SPRT accumulates evidence of leak conditions from flow discrepancy, using multiple “detection speed” filters (slow/medium/fast).
Pipewise computes both an instant SPRT score (fast response) and an aggregate score (robust alarming) to reduce false alarms while maintaining rapid detection.
Instant response can react within ~3–4 minutes depending on instrumentation and leak size
Aggregate alarming uses a longer integration period (example 15 minutes) to avoid transient noise dominating the alarm decision
2) Probabilistic filter: Detect operational changes and “freeze” the SPRT term
The probabilistic value spikes when the underlying measurement distribution changes, often indicating operational events (pump starts/stops, switching flows, transients).
When this occurs, Pipewise can freeze SPRT calculations temporarily to avoid spurious transient data from producing false signals. This helps distinguish:
Steady operation (low PV, low SPRT)
Operational changes (high PV)
Potential leak conditions (low PV, high SPRT)
Pipewise applied their Sequential Probability Ratio Test (SPRT) for high-sensitivity leak detection, combined with a probabilistic filter that identifies operational changes and filters their impact on the leak detection signal. SPRT is designed to improve key leak detection performance indicators, particularly reliability, sensitivity, and robustness under changing conditions and data interruptions.
Pipewise applied their Sequential Probability Ratio Test (SPRT) for high-sensitivity leak detection, combined with a probabilistic filter that identifies operational changes and filters their impact on the leak detection signal. SPRT is designed to improve key leak detection performance indicators, particularly reliability, sensitivity, and robustness under changing conditions and data interruptions.
1) SPRT: High-sensitivity evidence accumulation
SPRT accumulates evidence of leak conditions from flow discrepancy, using multiple “detection speed” filters (slow/medium/fast).
Pipewise computes both an instant SPRT score (fast response) and an aggregate score (robust alarming) to reduce false alarms while maintaining rapid detection.
Instant response can react within ~3–4 minutes depending on instrumentation and leak size
Aggregate alarming uses a longer integration period (example 15 minutes) to avoid transient noise dominating the alarm decision
2) Probabilistic filter: Detect operational changes and “freeze” the SPRT term
The probabilistic value spikes when the underlying measurement distribution changes, often indicating operational events (pump starts/stops, switching flows, transients).
When this occurs, Pipewise can freeze SPRT calculations temporarily to avoid spurious transient data from producing false signals. This helps distinguish:
Steady operation (low PV, low SPRT)
Operational changes (high PV)
Potential leak conditions (low PV, high SPRT)
Pipewise applied their Sequential Probability Ratio Test (SPRT) for high-sensitivity leak detection, combined with a probabilistic filter that identifies operational changes and filters their impact on the leak detection signal. SPRT is designed to improve key leak detection performance indicators, particularly reliability, sensitivity, and robustness under changing conditions and data interruptions.
Pipewise applied their Sequential Probability Ratio Test (SPRT) for high-sensitivity leak detection, combined with a probabilistic filter that identifies operational changes and filters their impact on the leak detection signal. SPRT is designed to improve key leak detection performance indicators, particularly reliability, sensitivity, and robustness under changing conditions and data interruptions.
1) SPRT: High-sensitivity evidence accumulation
SPRT accumulates evidence of leak conditions from flow discrepancy, using multiple “detection speed” filters (slow/medium/fast).
Pipewise computes both an instant SPRT score (fast response) and an aggregate score (robust alarming) to reduce false alarms while maintaining rapid detection.
Instant response can react within ~3–4 minutes depending on instrumentation and leak size
Aggregate alarming uses a longer integration period (example 15 minutes) to avoid transient noise dominating the alarm decision
2) Probabilistic filter: Detect operational changes and “freeze” the SPRT term
The probabilistic value spikes when the underlying measurement distribution changes, often indicating operational events (pump starts/stops, switching flows, transients).
When this occurs, Pipewise can freeze SPRT calculations temporarily to avoid spurious transient data from producing false signals. This helps distinguish:
Steady operation (low PV, low SPRT)
Operational changes (high PV)
Potential leak conditions (low PV, high SPRT)
Pipewise applied their Sequential Probability Ratio Test (SPRT) for high-sensitivity leak detection, combined with a probabilistic filter that identifies operational changes and filters their impact on the leak detection signal. SPRT is designed to improve key leak detection performance indicators, particularly reliability, sensitivity, and robustness under changing conditions and data interruptions.
* Leak size: 40 m³/day, approximately 0.6% of average flow * Traditional VB detection time: ~3 hours (using a 6-hour VB window) * SPRT detection time: 13 minutes
Most importantly, SPRT held the leak signal steady throughout the leak until it was repaired, rather than losing the signal during operational transients.
* Leak size: 40 m³/day, approximately 0.6% of average flow * Traditional VB detection time: ~3 hours (using a 6-hour VB window) * SPRT detection time: 13 minutes
* Leak size: 40 m³/day, approximately 0.6% of average flow * Traditional VB detection time: ~3 hours (using a 6-hour VB window) * SPRT detection time: 13 minutes
Most importantly, SPRT held the leak signal steady throughout the leak until it was repaired, rather than losing the signal during operational transients.
* Leak size: 40 m³/day, approximately 0.6% of average flow * Traditional VB detection time: ~3 hours (using a 6-hour VB window) * SPRT detection time: 13 minutes
* Leak size: 40 m³/day, approximately 0.6% of average flow * Traditional VB detection time: ~3 hours (using a 6-hour VB window) * SPRT detection time: 13 minutes
Most importantly, SPRT held the leak signal steady throughout the leak until it was repaired, rather than losing the signal during operational transients.
* Leak size: 40 m³/day, approximately 0.6% of average flow * Traditional VB detection time: ~3 hours (using a 6-hour VB window) * SPRT detection time: 13 minutes
Why it matters.
Why it matters.
Why it matters.
Because the probabilistic filter removes day-to-day transient noise from the leak decision process, false alarms are described as “very rare” for this highly transient network once the combined method is applied.
Because the probabilistic filter removes day-to-day transient noise from the leak decision process, false alarms are described as “very rare” for this highly transient network once the combined method is applied.
Because the probabilistic filter removes day-to-day transient noise from the leak decision process, false alarms are described as “very rare” for this highly transient network once the combined method is applied.
Because the probabilistic filter removes day-to-day transient noise from the leak decision process, false alarms are described as “very rare” for this highly transient network once the combined method is applied.
Because the probabilistic filter removes day-to-day transient noise from the leak decision process, false alarms are described as “very rare” for this highly transient network once the combined method is applied.
Because the probabilistic filter removes day-to-day transient noise from the leak decision process, false alarms are described as “very rare” for this highly transient network once the combined method is applied.
For operators managing dynamic, multi-inlet/outlet networks, Pipewise’s SPRT, included with every VBx system, provides a practical path to:
* Minute-level detection of small leaks that are often below meter accuracy limits
* Fewer false alarms during normal operations (pump starts/stops, network changes)
* More robust performance during transients and short data interruptions
For operators managing dynamic, multi-inlet/outlet networks, Pipewise’s SPRT, included with every VBx system, provides a practical path to:
For operators managing dynamic, multi-inlet/outlet networks, Pipewise’s SPRT, included with every VBx system, provides a practical path to:
* Minute-level detection of small leaks that are often below meter accuracy limits
* Fewer false alarms during normal operations (pump starts/stops, network changes)
* More robust performance during transients and short data interruptions
For operators managing dynamic, multi-inlet/outlet networks, Pipewise’s SPRT, included with every VBx system, provides a practical path to:
For operators managing dynamic, multi-inlet/outlet networks, Pipewise’s SPRT, included with every VBx system, provides a practical path to:
* Minute-level detection of small leaks that are often below meter accuracy limits
* Fewer false alarms during normal operations (pump starts/stops, network changes)
* More robust performance during transients and short data interruptions
For operators managing dynamic, multi-inlet/outlet networks, Pipewise’s SPRT, included with every VBx system, provides a practical path to:
Want faster leak detection without alarm noise, even on transient networks? Ask us about the VBx system.
