Scott McDougall

Tailings Dam Breach Analyses and various modelling software are used to estimate consequences of a hypothetical failure at a Tailings Storage Facility. The results of such forward-analyses are used in risk assessments and emergency planning. Many model inputs and approaches are currently based on expert judgment or adapted from other fields. There is some existing research and guidance for modelling approaches applicable for Tailings Dam Breach Analyses, however, there are few resources available that assess the hydrotechnical and geotechnical characteristics for dam breach and runout modelling from multiple diverse events. This presents a challenge for the experts who must make specific judgements for forward-analysis with little available hindsight into previous events.To address this knowledge gap, detailed investigations into 12 historical tailings dam breach events across a range of Tailings Storage Facility arrangements and taxonomies were completed. The investigations considered the outflow volumes, breach characteristics, and observations of downstream impacts. Some of these values or observations were previously reported by others for some events, but never previously compiled for the specific purpose of breach and runout modelling. These previous reports were critically assessed and included where relevant. Several original interpretations were made, and novel data were uncovered for many of the events evaluated in this thesis. Misconceptions regarding the events were frequently encountered, which likely contribute to the existing dearth of hindsight.Using the compiled information, a back-analysis model was developed to simultaneously simulate the breach and the tailings runout for the 12 events using HEC-RAS. HEC-RAS is a publicly available software for water resources and geohazards modelling. The modelling used the parametric breach method and non-Newtonian flow capabilities within HEC-RAS. The non-Newtonian flow parameters were determined through a comprehensive calibration process. The quality of the terrain data and misconceptions were found to be the most influential on model fit to any observed impact.The investigations and 12 models form the most comprehensive, diverse, and detailed tailings dam breach database to date. The insights from the investigations and modelling are applicable to forward-analysis Tailings Dam Breach Analyses, and the database serves as a springboard for multiple avenues of future research.

View record

The Site C Clean Energy Project is a new dam and hydroelectric generating station that is currently being constructed by BC Hydro on the Peace River in northeastern British Columbia, near the city of Fort St. John. The Site C dam will be approximately 60 m high and will create a reservoir approximately 83 km long. A detailed geotechnical assessment was undertaken in 2012 to generate preliminary impact lines that delineate potential stability, erosion, and flood hazard areas around the future reservoir. A construction headpond currently formed by river diversion provides site-specific data to potentially improve upon initial shoreline erosion predictions. Since the impoundment of the construction headpond in September 2020, a variety of data has been collected that contributes to observations of shoreline erosion. Collected data includes headpond elevation data, wind speed and direction data, multiple aerial lidar datasets, wave characteristic data at two sites on the construction headpond, and grain size curves of soil samples take at the headpond shoreline. This data was analyzed using two quantitative methods: the digital shoreline analysis system (DSAS), and change detection analysis. The key takeaways from the DSAS tool measurements and the change detection analysis are:1. All sites show the most material volume change in the first year of headpond impoundment, compared to the subsequent year and a half. 2. Most sites developed a notch at approximately the 416 m elevation, which corresponds with the elevation where the headpond spent the most time. The observations and analysis during the headpond impoundment led to lessons learned around monitoring, modelling, and communication. The research objectives of collecting shoreline erosion data and observations, analyzing the data with new tools and discussing the lessons learned from the headpond stage were met.The construction headpond provided a unique situation to be observed for approximately 2.5 years, but more research can be completed on the subsequent reservoir stage, as well as future reservoirs, to improve shoreline erosion prediction and monitoring methods.

View record

Wet muck (also known as mud rush) can be described as the sudden flow of fragmented rock into a drawpoint or other underground mine opening, exposing the mine to safety and operational risks. This hazard is analogous to an underground debris flow and is most commonly encountered in cave mines. Numerous fatalities, infrastructure damage, loss of reserves, and operational delays, have been reported in various caving operations. To better understand and manage this hazard, this thesis uses data and experiences from the PT Freeport Indonesia, Deep Ore Zone (DOZ) block cave mine in Indonesia where the ground conditions and operational factors that both increase the susceptibility of a drawpoint and act to trigger a wet muck event. Spatio-temporal relationships are drawn from this data, recognizing that the probability of wet muck events tends to increase as a cave matures, with increasing draw column heights contributing to increase secondary fragmentation and the generation of fines. Other contributing factors included in the analysis are extraction rate, uniformity of draw, Height of Draw (HoD), and drawpoint condition.Univariate and multivariate logistic regression models are developed, with the goal of improving prediction and mitigation of these events to improve safety and productivity in caving operations. Although the consequences of wet muck spill events are high, they are still relatively rare, resulting in an imbalanced dataset. Cost-sensitive learning is incorporated into the logistic regression models to address this technical challenge. These methods are used in this thesis to develop a spreadsheet-based wet muck susceptibility tool, which includes implementation guidelines and Python scripts. The concepts, methodologies and tools developed from this research are not restricted to the DOZ but can also be implemented in other caving operations that are susceptible to wet muck spills.

View record

Forecasting the spatial impact of debris flows is challenging due to complex runout behaviour, such as variable mobility and channel avulsions. Practitioners often base the probability of runout exceedance on a fan, or define avulsion scenarios, on judgement. To support decision making, spatial impact trends were studied at thirty active debris flow fans in southwestern British Columbia (SWBC), Canada. 176 debris flow impact areas covering an average observation period of 74 years were mapped using orthorectified historical airphotos, satellite imagery, topographic basemaps, lidar, and field observations. A graphical plotting method was developed that converts geospatial mapping to spatial impact heatmaps normalized by the fan boundary, allowing for comparison of runout trends across fans in the dataset. Probability of spatial impact was analyzed in two components: runout down-fan (i.e., how far debris flows tend to travel past the apex toward the fan toe) and runout cross-fan (i.e., how far debris flows tend to deviate from the previous flow path). For fans in SWBC, there is a characteristic decay in spatial impact probability from the fan apex and the previous flow path, represented by a normal and log-normal distribution for normalized runout in the down-fan and cross-fan components, respectively. Differences in spatial impact trends can be explained, in part, by event volume, Melton ratio, fan truncation, and fan activity, however not by fan morphometrics, such as the slope or the point at which channelization is lost. A tool was created that transposes the empirical runout distributions onto a fan to assist in risk-based decision making. Future work may involve fitting functions to the spatial impact data for a more robust and adaptable forecasting tool.

View record