KartaSoft at IPCE 2026

Professor Imre Bartos of the University of Florida will present the research behind KartaSoft's anomaly detection approach and explain why it works.

The KartaSoft team will be available throughout the conference.

Meet KartaSoft at IPCE

21–25 September, 2026 - BMO Centre at Stampede Park, Calgary

Asset 14@4x

Close the transient blind spot

 

Operational transitions change what normal pipeline behaviour looks like. 

 

Asset 15@4x

Compressor reconfigurations, maintenance, demand shifts, and equipment changes can cause conventional monitoring systems to raise thresholds or suppress alarms. These responses reduce sensitivity precisely when an integrity threat may be developing alongside a legitimate operational event.

KartaSoft’s IPCE 2026 paper presents Operational Event Governance, a structured method for adapting expected detection baselines during operational change while preserving sensitivity to concurrent integrity threats. 

Join the waiting list to gain access to the scientific paper, available after the conference

 

The reported results come from a purpose-built simulation of a 2,500 km gas transmission network. 

Presented at IPCE 2026 

Operational Event Governance for Pipeline Integrity Monitoring:

Closing the Transient Blind Spot

Asset 16@4x

Prof. Dr. Imre Bartos
University of Florida
Chief Scientist, KartaSoft


Session:
9/24/2026 
10:30 AM - 12:00 PM 
Digital Twins, AI/ML & Advanced Analytics in Operation - Part B

Add the session to your calendar

The paper examines 

→ Why monitoring sensitivity degrades during operational transitions
→ The limits of static thresholds and alarm suppression 
→ A two-stage, topology-aware monitoring architecture 
→ How event records can govern changes to expected behaviour 
→ The use of physics residuals for diagnostic interpretation 
→ Detection and precision results from five years of simulated operation 
→ Operational requirements, limitations, and the path to operator-data validation

When operations change, the baseline can become unreliable

 

Pipeline monitoring depends on a useful definition of expected behaviour.

During stable operation, pressure, flow, compressor performance, cathodic protection, and related channels may follow patterns that a monitoring system can learn.

An operational event changes those patterns.

The resulting deviations can resemble an integrity problem even when the system is operating as intended.

Conventional monitoring commonly responds in one of two ways: 

1

Raise the alarm threshold

This reduces nuisance alarms, while smaller developing threats become harder to detect. 

2

Suppress alarms during the event

This removes operational noise from the queue, while also creating a period with reduced monitoring coverage.

Both approaches create a transient blind spot. 

The operator still needs to answer a critical question: 

Is this an expected operational transition, or is an integrity threat developing at the same time? 

Asset 1-2Validation environment 

2,500 km

Asset 18@4x

Gas transmission network

150

Asset 18@4x

Connected network nodes 

Five years

Asset 18@4x

Of simulated operating data

600

Asset 18@4x

Operational and integrity events

Asset 1-2Performance during operational transients

At the 95th percentile anomaly-score threshold: 

Monitoring approach Integrity threats detected during transients Precision
Governed model 100% 14%
Ungoverned model 33% 8%
Conventional monitoring with alarm suppression  Effectively 0% Less than 1%

 

Asset 1-2The governed model maintained detection sensitivity during operational transitions

 

 The simulation also showed:
No loss of sensitivity during baseline operating periods
Fewer false alarms than the ungoverned approach
Strongest event-type performance for compressor degradation
Lower detection performance for some slow leaks and gradual instrument drift
Diagnostic value from physics residuals that helped indicate likely underlying mechanisms

What this could change for pipeline teams

 

Maintain monitoring coverage during operational events 

Asset 39

Operational transitions no longer have to create automatic periods of reduced sensitivity. 
Event context can be incorporated directly into expected behaviour, allowing unexplained deviations to remain visible.

 

Improve anomaly triage

Asset 39

An alert can include the operating event, affected measurements, network relationships, and relevant physics residuals.
This gives integrity and operations teams a clearer starting point for review. 

 

Reduce avoidable alarm noise 

Asset 39

Legitimate operating changes can be accounted for through governed corrections rather than broad alarm suppression.

The objective is a smaller queue with a higher concentration of conditions that warrant attention.

 

Preserve a reviewable baseline history

Asset 39

Teams can record which event changed the expected baseline, when the correction applied, and which measurements were affected.

This creates a clearer evidence trail for engineering review, model governance, and management of change.

 

Support tiered alarm handling

Asset 39

The paper proposes a tiered structure that separates high-confidence alarms from conditions requiring continued monitoring. 

Operational-event context can help teams decide which deviations require immediate escalation and which warrant observation as the transition develops. 

This research advances KartaSoft’s Baseline Governance foundation

Asset 17@4x

Operational Event Governance is the method described in the 'Operational Event Governance for Pipeline Integrity Monitoring:  Closing the Transient Blind Spot' paper. 

Baseline Governance is the wider operational capability it supports. 

It determines when and how expected behaviour should change as pipeline operations move between configurations, maintenance states, demand regimes, and other legitimate conditions. 

Baseline Governance is one of four operational foundations within KartaSoft’s broader consequence-aware pipeline risk intelligence approach. 

Built on Physics-Informed AI

KartaSoft’s Physics-Informed Artificial Intelligence, PIAI, combines engineering context, network relationships, operating configurations, observed behaviour, and machine learning. 

PIAI supports four connected operational foundations. 

Asset 1-2Baseline Governance

Interpret a changing condition against the correct operating state. 

Asset 1-2Consequence Intelligence

Determine what is exposed and how much the condition could matter operationally. 

Asset 1-2Risk Islands™ 

Identify connected concentrations of elevated segment risk that should be managed as one operational concern. 

Asset 1-2Explainability  

Show the signals, baseline context, consequence factors, likely mechanism, confidence, and evidence behind each priority.  

Together, these foundations help pipeline teams move from changing signals to a defensible order of action. 

Detect → Score → Risk Islands → Rank → Prioritize 

Know where pipeline risk matters most, what deserves action first, and why.

Asset 17@4x

Hear about the research at IPCE 2026

Operational Event Governance for Pipeline Integrity Monitoring:

Closing the Transient Blind Spot

Session: 

9/24/2026 
10:30 AM - 12:00 PM 
Digital Twins, AI/ML & Advanced Analytics in Operation - Part B 

The presentation will cover the monitoring problem, proposed architecture, simulation methodology, detection results, diagnostic approach, operational requirements, and remaining validation work.

Book a 1:1 time with the KartaSoft team in Calgary

The KartaSoft team will attend throughout the conference and will be available for scheduled technical and commercial conversations.

Book a meeting to discuss:

→ The paper’s methodology and results
→ Baseline Governance during compressor and maintenance events
→ Event-catalogue requirements
→ SCADA and topology inputs
→ Physics-informed diagnostics
→ Alarm and anomaly-triage workflows
→ Operator-data validation
→ Proof-of-value design
→ Consequence Intelligence and Risk Islands
→ Broader pipeline risk prioritization
 
 

 

KartaSoft
Consequence-Aware Pipeline Risk Intelligence 

Built on PIAI: 

Baseline Governance → Consequence Intelligence → Risk Islands → Explainability 

Asset 17@4x

Know where pipeline risk matters most