Safe and efficient operations depend on a comprehensive risk assessment method for pipelines that can give better, more correct and real time risk assessments, visualized in actual time. MARV™ performs pipeline risk assessment with uncertain and missing data by combining data and pipeline operator's knowledge and allows the user to easily visualize the cause-consequence relationships between various factors that impact a threat's likelihood in a layered manner.
The principal value of MARV™ is in helping to make quantitative, location-specific, risk management decisions, even if the data is sparse and uncertain. MARV™ helps to identify what data to collect that will have the most impact on risk reduction. Once mitigation actions are taken, MARV™ helps to quantify how they contributed to overall risk reduction.
The MARV™ method integrates under one framework all the available knowledge of an asset. This includes but is not limited to:
- Pipeline-related data from diverse sources (SCADA systems, data-bases, inspection data, sensors)
- Mathematical models (mechanistic, empirical, risk-based)
- Subject matter expertise (SME), that is expert's knowledge.
Bayesian networks handles uncertain data
MARV™ captures the various corrosion threats comprehensively through a Bayesian network method that allows using uncertain data and cases where data is missing. The Bayesian network methodology allows to integrate the operators data, existing corrosion models, and SME knowledge under one framework. The MARV™ method is consistent with direct assessment processes, but goes beyond them in considering uncertain data and knowledge.
Benefits and features:
- Prioritize data gathering
- Prioritize location-specific corrosion risks
- Optimize inspection intervals
- Prioritize use of mitigation actions
- Quantify the effects of mitigation actions for life time extension
- Conduct cost-benefit analyses for mitigation actions.
Extract value, even with limited data
The MARV™ methodology can run with limited data (missing data increases uncertainty). The first run can be performed with the readily available data, then more data and/or subject matter expert knowledge can be entered into the MARV™ framework until the uncertainties are reduced to the desired level. The quantitative nature of MARV™ output enables validation against inspection and failure data and the Bayesian methodology allows easy updates of the risk models if required.