Estimation of flexible riser curvature distribution and bend stiffener polyurethane behavior using the Levenberg-Marquardt algorithm in full scale bending-tension tests

Abstract

Long term full-scale bending-tension tests are traditionally performed to verify the numerically estimated flexible riser lifetime. For a proper tensile armour stress calculation, the riser curvature distribution has to be accurately determined, being highly affected by the bend stiffener polyurethane response. The actual material response may be significantly influenced by the loading rate, environmental temperature and humidity, ageing and self-heating phenomenon, requiring an extensive experimental campaign in addition to advanced constitutive models. In this work, an inverse problem methodology is proposed to decrease riser curvature distribution estimation uncertainties by combining optical configuration measurements, a direct finite element model and the Levenberg-Marquardt algorithm to estimate a representative polyurethane response. A full-scale riser/bend stiffener bending-tension test is conducted and an optical monitoring system is employed to track photoluminescence targets along the system length to estimate its deformed configuration. Five tests are performed and eight targets close to the bend stiffener tip selected for the inverse calculation of a representative bend stiffener hyperelastic response and riser top tension. The remaining target displacements and the tension measured with a load cell are employed for validation. The case study shows an excellent correlation between the numerically calculated deformed configuration and experimental measurements with the verification targets.