Load Environment Modeling and Forecasting

Integrity management of offshore structures constitutes a pressing societal challenge for Denmark and other countries relying on production of offshore oil and gas resources. Most offshore structures presently in operation were built during the period 1970-1980 and typically for design service lives around 20-30 years. Most of these structures are still in operation despite exceeding their originally intended service lives. The justification of service life extensions usually takes basis in structural performance assessments involving a more detailed modeling and analysis of their performance compared to the original design verification. This in turn facilitates that the various uncertainties affecting their life cycle performance can be quantified and that the reliability and risks associated with the structures can be updated based on observations, inspections and repair actions and documented transparently to the responsible authorities.

The Danish oil and gas reserves are by now rapidly decreasing why the efficiency in recovery of the remaining oil and gas resources plays a big role for the extent to which the Danish state can benefit from these – and how long. In appreciation of this the Danish Underground Consortium (DUC) partners lead by Maersk Oil and Gas have established the Danish Hydrocarbon Research and Technology Centre (DHRTC) at DTU in collaboration with other universities and industrial partners in Denmark and abroad. The main objective of this 1billion DKK investment over 10 years is to identify possible technical means to increase the recovery efficiency of the remaining oil and gas resources, to facilitate efficient and prolonged exploitation activities and at the same time ensure that the safety to personnel environment and assets is kept at world leading levels. This constitutes a major challenge and calls among others for the development of new methods and technology in support of assets integrity management.

Environmental loads such as waves, wind and current play key roles in the design and assessment of offshore structures and facilities. Traditionally, models of the offshore environmental loads are established in a marginal sense where each load component is addressed and modeled individually. Considering wave loads, basis is taken in a hierarchical approach where the starting point is a statistical characterization of (typically 1 or 3 hour) sea-state statistics (i.e. significant wave height, zero-crossing period and directional spreading). Conditional on the sea-state events, statistical characterizations of short term wave loading characteristics of relevance for design and assessment are then provided; such as crest height and shape as well as relevant time partial derivatives of the water particle positions over the water column. This approach has been significantly facilitated by wave tank experiments representing the sea surface elevation for give sea-states in conjunction with theoretical models of the sea surface elevation within sea states.

During the last decade it is, however, increasingly appreciated in the offshore engineering community that traditional offshore environmental load models have significant potential for improvements. This concerns not least the inclusion of joint (and consistent) consideration of waves, wind and current – but also with respect to facilitation of use of information of different type and origin such as laboratory experiments, on-site observations, hind-cast data and weather forecasts. The project will address such improvements utilizing Big Data techniques such as Bayesian Probabilistic Nets in conjunction with physical and phenomenological models to establish joint dependent environmental load models for waves, wind and current events. Moreover, for the purpose of supporting decision making with respect to operations shutdown and evacuation of personnel the developed models shall be coupled with real time environmental loading observations and weather prediction models to forecast potentially critical environmental load events in probabilistic terms. Finally, the developed models shall facilitate that the value of additional measurements and model improvements can be assessed.