Smart condition assessment, surveillance and management of critical bridges

Bridges with known deficiencies often have to be kept in service as they constitute links in highly stressed transport infrastructures. Disruptions in the traffic flow are often considered unacceptable. The current project was aimed at bridges with known deficiencies in particular, with the purpose of securing the serviceability and the safety using an innovative smart system for surveillance and condition assessment. Related to measurements and collection of data, the project has focused on wireless sensor networks. In comparison to traditional wired systems, the installation demands less resources and the locations of the sensor nodes are not restricted by cabling issues. The research activities regarding sensors have been aimed at energy harvesting and energy efficient scheduling of operations. A completely wireless installation requires local power sources as batteries with limited lifetime, which restricts the time for measurements. With the purpose of extending the lifetime, systems for energy harvesting have been evaluated as, e.g., units based on vibrations and electromagnetic waves. Another aspect is when and for how long measurements and communication should be performed. By scheduling the activities of the sensor node, the results of the project shows a potential to save energy. The purpose of a monitoring system is to provide data for condition assessment and damage detection. A condition assessment involves the estimation of the present safety level and a prediction of the remaining service life. Within this field, degradation models for the interaction between corrosion and fatigue have been studied, and a probabilistic model have been developed facilitating considerations of uncertainties in the measured response and in the results from inspections. The purpose of damage detection is to find anomalies in the behaviour of the structure that might be caused by damages or a change in structural behaviour. It can involve unexpected events difficult to predict in advance, but detectable by thorough investigation of measured data. A method based on machine learning and statistical inference have been developed in the project and tested on fictitious damage cases. If the results form a condition assessment or damage detection indicates insufficient safety, a decision has to be made on interventions. A decision support model based on Bayesian decision theory have been implemented in the project and tested on data from the Old Lidingö Bridge. Cloud based services have been developed to transfer data from the sensors, to the analysis parts, and on to the decision agent. The purpose was to simplify the management of large data volumes from a monitoring system and to visualize the results in a comprehensible format. Solutions for collecting, storing and sharing of data have been developed, together with a mobile app for presentation of the results. The objective to develop an integrated system for surveillance, sharing of data, condition assessment, and decision support is partly completed. Some parts of the integration remains to be solved before a complete solution can be launched.QC 20190515