Combining physics-based modelling with data-driven machine learning, DNV’s SAMBA (Safety through Agile and Module-based Assurance) methodology will reduce uncertainty related to risks in complex, high-risk systems and industries.
New advances in data-driven models and deep learning offer unprecedented insight into data and their correlations. Yet, observations and correlations alone can only inform based on what is in the data and not the causal effects in the systems producing the data.
Indeed, research reveals conventional methods based on simulations are too conservative; but sensor data cannot capture safety-critical behavior which we have not yet experienced.
While digital tools have the potential to unlock hidden value, physical assets create the ultimate value. There is a need for methodology that combines both elements, offering simplicity by connecting only relevant information to the decision context.
“Conventional finite-element simulations appear to be conservative compared with what we observe from sensor data. However, models based on sensor data do not capture safety-critical behaviour which we have not yet experienced. In effect, neither model is able to give proper decision support and we are left with conservatism to offset the uncertainty.”
Marie Lindmark Sandøy
Lundin-Energy Norway AS
Combining what-has-been and what-could-be for safety assurance
With the SAMBA tool, DNV has set out to combine casual knowledge with all relevant data to make sure that the best available, and most accurate, information is readily at hand for decision makers operating safety-critical systems.
The work has resulted in the new research project RaPID (Reciprocal Physics and Data-Driven models) funded by the Norwegian Research Council. Building on the basic methodology used in SAMBA, the RaPID project aims to provide more specific, accurate and timely decision support in operation of safety-critical systems.
This is achieved by increasing computational efficiency of advanced modelling tools by reduced-order modelling; developing hybrid analysis and modelling, combining physical models with data-driven ones; reducing computational demand and increasing safety by effective selection of relevant simulation scenarios based on probabilistic machine learning and risk-aware objective functions; and tailoring and demonstrating the integrated modelling approach to selected user cases.
The rationale that there is a need to combine causal knowledge with all relevant data runs throughout all the research within this project and its case applications.
The benefits
Combining casual knowledge with relevant data potentially brings huge benefits, including live, reliable, up-to-date safety assurance.
Potential case applications are vast and diverse, spanning high-risk systems and sectors. For example, weather scenario simulations for offshore drilling do not capture scenarios that have not yet been experienced; by better use of the combination of observed experience and physics-based models, potential down time can be reduced significantly, reducing unproductive rig time by millions of USD per year.
DNV will develop and document the methodologies and technologies needed to consistently combine physics-based and data-driven models to alleviate the deficiencies of both by capturing their complementary advantages.
DNV invites the industry to challenge the project with interesting case contributions.
