Barbara Jean Lence
Relevant Degree Programs
Graduate Student Supervision
Doctoral Student Supervision
Dissertations completed in 2010 or later are listed below. Please note that there is a 6-12 month delay to add the latest dissertations.
Generally, the use of the water distribution network (WDN) modeling is divided into two main categories, WDN design, and WDN hydraulic analysis, which itself is employed as part of WDN design. The former generally identifies components of the network, considering the system cost and the ability of the network to satisfy consumer demands for water availability, pressure, and quality. The latter evaluates the distribution of nodal pressures and pipe flows under a specified network design and known or estimated water consumption levels. WDN analysis is often embedded within design algorithms, where, for every potential design considered, the hydraulic energy and continuity equations that govern the system are solved. If water demands and the physical characteristics of the network design are known with certainty, deterministic approaches for solving these equations may be used. If some information is uncertain, non-deterministic approaches are used for identifying the Probability Density Function or the fuzzy membership functions of the pressure and flow conditions at all locations in the network. This dissertation is divided into three main parts, 1) the application and development of evolutionary algorithms for single- and multi-objective optimization of WDNs, 2) the introduction of new frameworks and performance surrogates as objectives in the optimization of WDNs, and 3) the advancement of an efficient gradient-based technique for fuzzy analysis of WDNs under uncertainty. First, efficient Evolutionary Algorithms (EAs) are compared and advanced for reducing the computational burden of single- and multi-objective design of WDNs, respectively. We investigate and identify the most appropriate operators and characteristics of EAs for optimization of realistic-sized networks. Based on the experiences and capabilities of EAs obtained, a new EA is introduced for single-objective optimization of WDNs which is faster and more reliable than other popular algorithms presented in the literature. Next, two different frameworks are introduced for implementing many objectives in the optimization of realistic-sized WDNs. Both approaches can distinguish appropriate design solutions with minimum cost and maximum hydraulic and mechanical reliability. Finally, a fuzzy method is introduced for analysis of WDNs under uncertainty. The proposed technique significantly reduces the CPU time of uncertainty analysis of large-scale networks.
Rapid-Onset, High-Intensity hazards such as dam failures, tsunami, flash floods, volcanic lahars, urban-wildland interface fires and industrial accidents can produce catastrophic mortality for Populations at Risk (PAR). Governments, local communities and other stakeholders can use risk management, sustainable hazards mitigation and emergency/disaster management processes before an event to establish a Community Protection System (CPS) to protect the PAR. A CPS is a system-of-systems that combines the capabilities of the natural, critical infrastructure and social infrastructure environments. Since a CPS can be expensive to establish and maintain, and since there can be many uncertainties associated with system performance, there is a need to develop reliability-based Life Safety Measures that can be used to analyse and rank alternatives, to optimize designs and to inform stakeholders. Forensic datasets that describe historic disaster outcomes generally cannot support the process of loss and survival estimation; therefore, analytical and simulation-based methods must be used to develop synthetic CPS performance data. Life Safety can be assessed using two limit state equations that assess the sufficiency of time and the sufficiency of protection offered to individuals in the hazard impact zone. These equations consider causal event chains, spatial pathways, network interdependencies, management decisions, differential vulnerabilities, individual decisions and emergent/non-linear systems behaviours. The performance estimates can be estimated and visualized using a Life Safety Performance Space and a time-dependent Life Safety State Space. A Systems Modelling Framework is developed to guide the integration of the analytical and systems simulation models used to estimate mortality and survival. The framework combines concepts from systems engineering, systems safety, Geographic Information Systems, systems simulation, critical infrastructure modelling, hazards research and disaster research. The resulting probabilistic-causal-quantitative framework provides a basis for developing estimates that are transparent and defensible. Detailed theoretical formulations of the Systems Modelling Framework and the Life Safety Measures are developed. A series of hypothetical examples are used to demonstrate the methods. Applications are developed for tsunami preparedness and dam safety at the macro-, meso- and micro-resolutions. The tsunami example considers the Cascadia Subduction Zone tsunami hazard. The dam safety examples consider the St. Francis and Malpasset Dam Failures.
Extreme floods pose a significant risk to communities and environments in river systems throughout the world. In many cases, sensitivity of this issue is heightened for regulated rivers where downstream impacts of reservoirs are directly affected by operational decisions. Therefore, many North American jurisdictions require asset owners to assess downstream effects. Loss of life and economic impacts have been widely addressed in the literature. Nevertheless, immediate and long-term environmental impacts of such extreme events have not been holistically addressed. This work develops a framework for quantitatively estimating immediate and long-term fisheries impacts of extreme floods. The framework may also be generalized to other environmental systems. Several models are developed to support it. The immediate effects of extreme events are assessed with three models. These include: a probabilistic individual-based model that employs the results of a transient hydrodynamic model to estimate fish loss during extreme floods; a sampling simulation model that utilizes the results of a transient morphodynamic model and derives a probabilistic relationship between egg loss and flood intensity; and a habitat change estimation model that evaluates the available habitat difference before and after extreme events, given the results of hydrodynamic and morphodynamic models. A fish population recovery model is also developed and employed to estimate long-term impacts of extreme events, given the results of the immediate impact estimation models. An approach for estimating a number of risk-based performance measures that characterize the impacts and recovery from extreme events is also developed. These performance measures include existing formulations for vulnerability, engineering resilience, and ecological resilience, as well as a new measure which is introduced in this work, as vulnerability divided by engineering resilience. This new performance measure is designed to characterize both short- and long-term performance of the environmental system. Planning, design, and real-time operation of reservoirs, participatory water use planning, and licensing and relicensing decisions for proposed and existing water resource projects are cases in which such estimates may be useful. Applicability of this framework is demonstrated for the case study of the Lower Campbell River in British Columbia, Canada.
Corrosion of cast iron pipes in distribution systems can lead to the development of corrosion pits that may reduce the resistance capacity of the pipe segment, resulting in mechanical failure. These pipes have a tendency to corrode externally and internally under aggressive environmental conditions. The mechanical failure of pipes is mostly the result of this structural weakening coupled with externally, environmental, and internally, operational, imposed stresses. While external corrosion has been shown to significantly affect the likelihood of mechanical failure, the risk of failure may be further heightened if internal corrosion is occurring. This thesis develops a methodology for estimating the probability of mechanical failure of cast iron pipes due to internal corrosion that incorporates the relationship between chlorine consumption and the rate of internal corrosion in a cast iron pipe. A probability analysis is developed that incorporates the internal corrosion model as well as the Two-phase nonlinear external corrosion model to calculate the overall probability of mechanical failure. Monte Carlo Simulation (MCS), First Order Reliability Method (FORM)-, and Second Order Reliability Method (SORM)-based approaches are used to estimate the probability of mechanical failure. Next, a methodology is developed for analyzing pipe condition based on the data resulting from the probability of mechanical failure analysis incorporating internal and external corrosion. A modeling strategy inspired by survival analysis is used to obtain the predicted number of pipe breaks for a given exposure time. The likelihood of failure at a given residual pipe wall thickness is estimated and coupled with the predicted number of pipe breaks as surrogates for pipe condition. These condition indices may support decisions regarding replacement planning and can be coupled with economic assessment models in the development of future asset management strategies.
Master's Student Supervision
Theses completed in 2010 or later are listed below. Please note that there is a 6-12 month delay to add the latest theses.
Models that estimate the likelihood of failure of water mains are widely used to support the repair and replacement strategies of water utilities. Advances in the fields of statistics and machine learning have introduced a wide range of models and improvements to data management have made increasingly complex models more feasible. The datasets that are used to develop these models are frequently subject to change as strategies for the operation and renewal of water distribution systems evolve. This issue is potentially exacerbated by the nonstationary processes impacting these systems. For water main failure prediction models to be useful in this dynamic context, it may be necessary for utilities to periodically evaluate several models for their dataset or for researchers to examine the performance of one or more models across multiple datasets.This work presents a framework for the selection and analysis of water main failure prediction models that is intended to enable efficient development of a range of models for a single dataset or investigation of the performance of models across several datasets. Each step of the framework is described and recommendations are given for researchers and asset managers attempting to implement the processes defined herein. The framework is investigated using data from four different utilities, where each dataset is highly censored. Through the application of the framework, four models are selected and refined: Cox Proportional Hazards Model, Neural Multi-Task Logistic Regression Model, XGBoost Survival Embeddings Model, and Random Survival Forests Model. These models are trained on each of the utility datasets and the outputs are compared to assess the efficacy of the framework.Results show that the framework may be used to identify models that are sufficiently robust to achieve high performance using datasets from four different utilities. Of the final selection of models developed through the framework, the lowest performance among all four datasets is a C-index of 0.780. Additionally, the framework is able to establish at least one model for each utility that performs very well. The C-index values range from 0.880 to 0.913 for the best model developed for each utility.
Natural gas is one of the cleanest fossil fuels and most reliable energy source readily available for everyday use. Due to changes in weather conditions and in economic growth and market conditions, demand for natural gas by municipal, as well as commercial and industrial consumers, has greatly increased, and delivering gas to these varied consumers has become more complex. The problem of optimally operating gas transmission networks is generally formulated to minimize the total supply cost of the network while satisfying the demand of consumers at different delivery points, at minimal guaranteed pressures. If necessary, compressors are installed and operated to supply more energy to the network and increase gas pressures, so as to compensate for pressure losses in the pipe network. The nonlinear relationships between system discharges and headlosses due to friction and mechanical devices in pipes form a complex set of nonlinear constraints in optimization models. In order to solve this complex problem, researchers have used approaches that use linear approximation in classical optimization techniques. Recently, heuristic algorithms have been applied, as these techniques can be used to find approximate solutions to complex optimization problems more quickly than classical optimization methods, and can often obtain an acceptable solution when classical methods fail. This thesis compares the performance of four heuristic algorithms for optimizing operations of a gas transmission network, namely, the: Genetic Algorithm (GA), Differential Evolution Algorithm (DE), Artificial Bee Colony Algorithm (ABC), and Covariance Matrix Adaptation Evolution Strategy (CMA-ES). The algorithms are compared in terms of their computational burden, and the quality of the obtained solutions with respect to practical management concerns. Among the four algorithms, CMA-ES is found to achieve the global optimum, compared with that generated with classical optimization, most consistent results in solutions that meet pressure and flow requirements, and conserve compressor power.
Sediment transportation occurs during high flow events in gravel bed rivers resulting in a change in bed elevations. Some areas of the river experience a net degradation (scour) and others net aggradation (fill). During these events, incubating salmon eggs can be scoured from their pockets or sediment may be deposited above them, preventing intergravel flow and the emergence of fry. The purpose of this thesis is to develop a framework for estimating the probability of egg loss due to scour and fill for a range of possible high flow events in a river.The developed framework consists of four steps. Steps one and two are the application of 2-dimensional hydrodynamic and morphodynamic models. The hydrodynamic model provides outputs of velocity, depth and shear stress at specified locations within the river. In the second step, these results are input into a morphodynamic model that simulates bed elevation changes during a transient simulation of the event. In the third step for a range of events, pre and post-event bed elevations are compared and the values of scour and fill depth are described by probabilistic distributions. For a specific high flow event, given a specific egg burial depth, a relationship between the proportion of egg loss due to scour and fill may be determined based on these distributions. In the final step, uncertainty in the depth of egg burial is accounted for by developing an egg loss model using reliability analysis that determines the probability of not meeting a target egg survival rate. The developed methodology can be applied to any gravel river and is applicable to any salmon species. A case study of the Campbell River, British Columbia using the 2D hydrodynamic and morphodynamic models, River 2D and R2DM, is developed to demonstrate the methodology. For the case study, the Generalized Pareto Distribution is recommended to describe scour and fill in high flow events in spawning areas.