Sensor Placement to Monitor Launching Girder Operations in Segmental Construction

Launching Girder is an equipment extensively used in prefabricated segmental construction of viaducts. It is used to lift, assemble, post-tension and load the spans of bridges constructed using the incremental launching method. Currently, there is limited automation in the operation of launching girders worldwide. Construction efficiency can be improved significantly and risks of accidents could be reduced through automation. To automate the construction process, real time data about the state of the launching girder such as the current launch position, pressure in hydraulic pumps, internal forces, etc. is required. It is also necessary to ensure that assumptions made during the design phase are valid during the operation phase. Even though conservative assumptions are usually made, the impact of these assumptions are rarely studied. Measuring strains and stresses during operations will allow checking whether the design is conservative and in that case whether it can be optimized further. Manual monitoring cannot be adopted due to the high cost, human errors associated with it and the response time. Therefore automated monitoring using wireless sensors is proposed. Wired network is not found to be practical since interference from construction operations would induce a high risk of failure. Major challenges in designing a wireless sensor network for outdoor applications include reliable communication and power requirements. Fundamental risk in implementing a wireless sensor network in a construction site is that radio frequency environment is characterized by limited coverage, interference from electromagnetic fields generated by operating engines, multipath signal fading and non-line of sight conditions which severely impact wireless signal propagation. Apart from these technological challenges, there are computational challenges in determining the optimal sensor configuration. Determining the sensor locations such that the information content is the maximum is a combinatorial optimization problem reported to be NP-Complete. The optimal sensor placement strategy aims to correctly identify the current state of the launching girder with the minimum number of sensors. It should be able to distinguish between the many potential scenarios that are possible during the operation. Uncertainties in environmental conditions, quality of work, human errors as well as the actual activities performed create many scenarios which have to be correctly identified and evaluated for safety and other aspects. Correct identification of the state will also help to verify design assumptions and evaluate the potential for optimization. A methodology based on Shannon's entropy is proposed in this research. This involves the random generation of a population of candidate models which represents different operational states. These models are analyzed using methods based on physical principles and responses at possible sensor locations are computed. Shannon's entropy is calculated using the distribution of the predicted responses at all the possible sensor locations. Position with the highest entropy provides maximum separation between candidate models. Details of the sensor placement strategy is discussed in the paper. The methodology will be validated using data from the implementation on a full-scale segmental construction scenario and design assumptions will be verified. The paper also discusses the challenges in the implementation of the wireless sensor network, the network architecture and the preliminary results from the implementation on a segmental construction of a metro rail viaduct.