Utilizing a Load Cell Sensor to Collect Data on Slope Failure from Small-scale Physical Modeling Experiments

Authors: Heribeto Patino Luna, Corban Larson

Mentors: Ozgur Yavuzcetin, Juk Bhattacharyya

Abstract

As weather conditions become more extreme due to climate change, the triggers of slope failure, such as rainfall and the freeze-thaw cycle, increase in intensity. Slope failure can be costly, both financially and in human lives. This is why we are motivated to design a device that can accurately record field measurements and use these to predict when a failure might occur.

Our experimental setup consists of a plastic container that holds 30 kg of sand in total: 20 kg of 0.5 mm sand grains and 10 kg of 0.25 mm sand grains. Both layers are pre-saturated with deionized (DI) water using a ratio of 100 cc/kg for the bottom layer and 80 cc/kg for the top layer so we can create a slope of 45°. A buried sandbag (0.05 kg), attached to the load cell sensor by a fishing line, communicates measurements, collected at a frequency of 33.33 Hz, to a Raspberry Pi platform.

We simulated rainfall conditions by adding 60 cc of DI water to the system for 10 seconds over 30-second intervals until slope failure occurred. In these experiments, two patterns were observed in the plotted data. The first of these patterns shows negative readings in the first half of the plot, before rapidly rising when the failure is approaching. The second pattern indicates negative readings and a rapid decrease in reading values right before failure. Understanding how to interpret the observed data patterns is necessary to make accurate slope stability assessments.

To interpret the observed data patterns, in terms of slope stability, we have calibrated the values recorded by the load cell sensor by varying the angle made by the fishing line with the vertical Z-axis. The baggie was replaced with known weights of different amounts (10, 200, 300, and 400 g) in every experiment.

Additionally, we explored the effects the frequency and volume of water being added had on the data. We observed that these variables influence the data pattern to a minor extent, not influencing the overall pattern of the data.

Although slope failure is almost inevitable, the consequences can be lessened by identifying early warning signs. We will present and discuss the data of our experiments aimed at understanding the data patterns, and how this knowledge could be applied to our small-scale physical modeling experiments. Ultimately, this work will lead to the development of a monitoring device that collects field measurements in real time.

Geological Society of America Abstracts with Programs. Vol. 56, No. 5, 2024 doi: 10.1130/abs/2024AM-403912