PSI - Issue 44

Federico Germano et al. / Procedia Structural Integrity 44 (2023) 902–909 F. Germano et al./ Structural Integrity Procedia 00 (2022) 000 – 000

908

7

It will be very helpful also to have certified proof of this performance for each physical sample of sensor. A calibration certificate, better if traceable, could ensure certainty on the acquired data where physical sample distribution of properties may vary. This is especially relevant when automated analysis is applied upon the acquired data, so the result of the monitoring is often seen only after a number of analysis passages that can mask or make impossible to recognize an error in the initial time history (for instance, a variation in amplitude over frequency due to sensor performance). 1.5. Data acquisition system architecture Great attention must be taken for the Data Acquisition System performance. The requirements mentioned in the beforehand sections have to match an artifact who can extend itself for kilometers and needs that every part of the structure is giving synchronous data simultaneously. As said, there is need for synchronize all the sensors. For quasi-static phenomena, where phase errors are not particularly relevant as 0.1s time error will not change much the physical quantity measured, a wifi connection will still be a very nice compromise to have a good measurement scenario, a simple installation, a reduction of cabling cost. Their digital output can be connected through API or MQTT later to the rest of the acquisition set. Choices have to be made about the architecture of a net of dynamic sensors. We can recognize at this level a compromise between two extreme situations, none of them possible by itself, but giving enough hints about the right balance. The best possible scenario to be absolutely certain about synchronization would be to connect physically through cable each sensor to a central acquisition system, compensate if needed differences due to cable length, and acquire everything at the same time by the same unit. Digitalized, slow-changing quantities can be synchronized afterwards through API or other methods. This will be a fail-proof method to ensure synchronization, but of course it is not practical as cable quantities will explode where principal dimension of the structure and number of sensors grows. On the other extreme, digitalizing every single node/sensor directly where acquired and connect them through digital network by various wifi protocol will simplify installation cost, eliminate most of the cabling, but synchronization will never be ensured. We will be limited to monitor consistently low-changing quantities, but we will never be able to perform an OMA or describe the shape of a deformed beam when passage of a heavy vehicle occurs. The good news is that a nice compromise exists. By selecting a limited number of acquisition nodes, who can collect information from cabled sensor only in their surroundings, and connecting properly (for instance, with optical fiber) the acquisition nodes each other in a daisy chain, it is possible to maintain an indisputable synchronization even for dynamic sensors, preserving: Reduced cable length Of course this strategy will work if the system is also able to be flexible enough to collect sensors of different type, customizing each acquisition node with the need of the surrounding sensors. For instance, in one node acceleration sensors with voltage output have to be acquired together with strain gauges, current output displacement sensors and weight in motion; each node have to maintain a complete freedom to choose what sensors have to be conditioned independently. The data can be afterwards stored in a cloud-based architecture, where an interface can be prepared for first evaluation and assessment of the data acquired. • • • Synchronization up to 1 microsecond Simplified installation