Use of Digital Image Correlation for monitoring of concrete buttress dam failure model tests

Abstract: The monitoring of dam structures is of great importance for the safety of the dam.A dam failure can cause disastrous consequences for people and infrastructure downstream.Dams are monitored by periodic inspections and monitoring systems. Inspectionsare time consuming, limited to the accessible parts of a dam and theresult can vary depending on the experience of the inspector. Instrumentation formonitoring is often installed on the dam surface and inside the dam. This instrumentationfills the function of measuring important parameters for the safety likedisplacements, piezometric level, seepage, temperature, ice loads and crack widths.At the Vattenfall Research and Development facility in Älvkarleby, a series of experimentalconcrete buttress dam failures have been performed. Three of these testtrials were analyzed in this report. The aim of this master thesis was to investigateif Digital Image Correlation (DIC), an application of photogrammetry, can beused effectively to monitor displacement of a dam model in laboratory setting, andto assess the usability of the method on a large scale dams. The experiment wasperformed on a dam model with a scale of 1:15 that was subjected to an increasingreservoir pressure until the dam failed. The dam was instrumented with wiregauges and pressure sensors to study the dam behaviour. The displacements were,additionally to the wire gauges, also monitored by cameras used for DIC. The measurementsbetween these two monitoring methods were later evaluated. The resultsshowed that the DIC was viable for the tracking of early displacements and also thedisplacements during the failure of the dam. The result from the early displacementsshowed a great similarity between the three cameras used for the image correlation,both in terms of when the early displacements occurred and their magnitude. Themeasured displacement from the wire gauge occurred approximately at the sametime but deviated slightly from the measured displacements from DIC. The differencewas within 2 to 3 mm for the first and second trial. The parameter study ofthe early displacements showed great robustness of the user defined input parameters.The result from the measurements during failure showed a different character.The program showed complications in tracking the pixels, possibly due to largerdisplacements but also an increased number of obstacles and external factors affectingthe appearance of the pixels. However with an expense of an increased amountof processing, the behaviour of the dam monoliths could be captured for the entiredisplacement during the failure, but with larger uncertainties. The study shows thatthe program had difficulties in tracking templates with challenging conditions andtherefore, must development must be performed before a commercial application ofDIC as a global monitoring system can be implemented.