Carnegie Mellon University

MS in Civil & Environmental Engineering - Research

Our Master of Science in Civil and Environmental Engineering with a Research Focus program is tailored for aspiring researchers, emphasizing research skills development alongside disciplinary knowledge. Through this program, you'll acquire competencies in critical analysis, literature review, research question formulation, validation planning, and effective communication. It's a pathway for those considering careers in academia or research-driven industries.

Required Units
  • Minimum of 180 units
  • Minimum of 60 units of Graduate Project (12-791)
  • Minimum of 84 units of coursework. At least 60 must be in CEE 48 (of the 60) at the 700-level or above and no more than 18 units of 300+ level courses. The restrictions are similar to our course-based MS.
  • 36 units of Summer Reading & Research. This research will take place in the summer after the first spring semester. Students will be in intensive summer research with their faculty advisor.
Length of Program 5 Semesters: 2 Fall, 2 Spring, 1 Summer
Required Courses
  • Graduate Project (12-791)
  • No core academic courses

Academic Outcomes

In the MS Research program, students embark on a dynamic academic and professional growth journey.

Beyond mastering the core concepts of your chosen concentration, you'll develop a keen ability to engage with civil and environmental engineering and science literature critically. Hands-on experience teaches you to conduct comprehensive literature reviews, formulate precise research questions, and design robust validation plans. This program equips you with the skills to craft compelling research reports, proposals, and manuscripts and confidently communicate your findings through oral and written formats.

It's an immersive experience that prepares you to excel in your field and thrive in research-oriented careers.

MS Research Projects

Listed below are examples of some of the research projects currently open for students applying for the Fall 2024 academic term. 

Integrating Physic-based Modeling and Sequential Learning Systems for Dynamic Systems

Many engineering systems can be modeled as dynamic systems, where some (possibly hidden) state variables evolved depending on some inputs. The state variable can be partially observed and inferred via probabilistic data analysis. This analysis is challenging when the evolution model is non-linear, as the classical framework of the Kalman Filter cannot be applied to those settings. This thesis project focuses on alternative approaches for sequential inference, such as particle filters and unscented Kalman Filter, and explores how to optimize these approaches in applications related to the sequential management of wind turbines, including bolt inspections and fixing.

Faculty: Matteo Pozzi

Incorporating Equity into Transportation Resilience Modeling

This project aims to assess how heavy rainfall affects the reliability of the transportation system across low, middle, and high-income communities. This study will also consider hazard mitigation measures' performance benefits and economic costs (e.g., drainage upgrades or roadway retrofits). We are seeking a student to assist with data collection, data analysis, and development of equity-focused modeling tools that could help policymakers and infrastructure owners and operators make more informed resilience infrastructure investment decisions. 

Faculty: Corey Harper

Equity-Focused Assessment of Shared Automated Vehicle Deployment

Shared automated vehicles (AVs) promise to improve transportation access in urban centers (due to improved operational efficiency) while drastically reducing transportation-related energy consumption and air pollution. This project develops a replicable, open, deployable model that can (1) identify areas with high concentrations of low-income or minority populations or both within a region and (2) conduct a scenario analysis to help public agencies assess how regulatory requirements for shared AV services (e.g., number of handicap accessible vehicles and fleet size) could impact energy use, travel demand (i.e., vehicle miles traveled), average wait times across different population groups, and AV operator profit.  

Faculty: Corey Harper

Global Glacier Projections

Global glacier mass loss considerably impacts sea-level rise, water resources, natural hazards, and culture that will affect communities worldwide. Improving projections of glaciers' response to climate change is thus critical for informing adaptation and mitigation strategies. This project will use the Python Glacier Evolution Model (PyGEM) to refine glacier mass change and runoff projections and quantify the resulting impacts. Project objectives may include a focus on model development (e.g., the application of data assimilation or machine learning algorithms to incorporate remote sensing data to refine projections) or quantifying impacts (e.g., using projections to quantify changes in runoff).

Faculty: David Rounce

Improving System-level Mobility and Accessibility Through Data-driven Network Flow Modeling

Network traffic flow modeling and simulation are central to transportation system analysis. Existing research has been primarily focused on cars as a single mode, while other modes of transportation, such as trucks, shared mobility, micro-transit, micro-mobility, for-hire vehicle trips, and parking, are overlooked or modeled separately from typical car trips. Unfortunately, characteristics of multi-modals of transportation, such as when and how passenger or freight travel and how that multi-modal flow impacts the mobility and accessibility of regional networks, particularly to essential resources, are unclear. This becomes the main hurdle for setting policies for improving mobility and accessibility.

This research aims to develop a holistic framework for mesoscopic traffic simulation that mixes multi-modal transportation of all modes, for passengers and freight, by considering their interrelations simultaneously. The result includes predicting travel time, travel delay, vehicle-mile-traveled, energy use, and emissions for each mode at each road segment and intersection by time of day. More importantly, system-level metrics on accessibility, such as Mobility Energy Productivity, and accessibility measures on essential resources accessible by any community of interest, can be quantified and modeled under this general framework. Thus, potential management strategies and policies for passenger cars and freight transportation can be evaluated and deployed.

Faculty: Sean Qian

Computational Modeling of Environmental Systems

Professor Dayal's research group collaborates with faculty in the department (Fakhreddine, Pozzi, Rounce) on the computational modeling of environmental systems in which chemical reactions and phase changes play an essential role. Specific topics include reactive transport modeling of contaminants, mechanics of glaciers accounting for melt and crevasse propagation, and CO2 sequestration and H2 storage in geological formations. Other topics of interest include applying similar computational techniques to advanced manufacturing, nuclear and H2 energy generation, and robotic materials.

Faculty: Kaushik Dayal

Situationally-Aware Autonomously Controllable Streetlights

Many cities are converting their existing streetlights to LED-source luminaires due to anticipated energy and maintenance savings. Addressable electronic lighting controls and sensors are now available that can transform basic streetlights into intelligent, smart city devices with numerous public safety and community benefits. However, several issues persist. For example, glare remains a serious problem not addressed in cost-competitive devices on the market. Additionally, while progress has been made in providing color temperature choices, the uniform installation of a single color temperature for all occasions is contextually limited. The inability to provide high color rendition decreases the quality of walkability for pedestrians and renders colors inconsistent with how a person perceives color during daylight hours. Selecting and specifying street lighting also remains an educated guess, requiring weighing multiple variables that are not fully understood. The objective of this project is to create a universal streetlight sensing system that can autonomously adjust its light source and light quality (e.g., temperature, color, acuity) to meet optimal location and operating conditions for drivers and pedestrians without the need for operator management.

This interdisciplinary and integrated research project will design an LED streetlight sensing system to control lighting in response to continuously monitored atmospheric, environmental, and roadway surface conditions. The sensing system will be capable of communicating with adjacent streetlights to maintain lighting compatibility within a localized setting. In situ, continuous self-monitoring will accommodate uneven luminaire spacing, mounting height, fixture head location, road surface widths, and surface reflective qualities, including deterioration, tree canopy interference, and ambient lighting. Localized preferences, whether for psychological, safety, or health concerns, can be considered by engaging the public in the municipal decision-making process.

Faculty: Katherine Flanigan

The Impact of Generative AI Tools in Civil and Environmental Engineering Classrooms on Student Learning and Equity

The rapid advancements in artificial intelligence (AI) and the development of generative AI tools, such as ChatGPT, can potentially revolutionize various aspects of education within civil and environmental engineering. Our curriculum strongly emphasizes problem-solving in complex, real-life scenarios, encouraging students to cultivate innovative solutions. Integrating generative AI tools can potentially influence student learning outcomes and foster equity in education significantly. This research is dedicated to examining the impact of these generative AI tools on student learning and equity within specific teaching contexts and learning objectives by employing a combination of experimental studies, surveys, and interviews to provide valuable insights and recommendations for the effective integration of AI in civil and environmental engineering classrooms.

Faculty: Fethiye Ozis

Plant Nanobiotechnolgy for Sustainable Agriculture

Society cannot be sustainable without a tractable path towards sustainable agriculture. This environmental chemistry project will transform agriculture by developing novel nano-enabled delivery vehicles for the highly efficient delivery of nutrients and other active ingredients into plants. This precision delivery method minimizes crop inputs while maximizing crop yields, simultaneously mitigating the environmental impacts of agrochemical runoff. The student working on this interdisciplinary and multi-institution project will conduct hands-on laboratory-based experiments with PhD students and postdocs to synthesize nanocarriers for testing, assess their uptake into plants using state-of-the-art characterization tools, develop novel plant digital twins to enable the design of efficient nanocarriers and assess the overall benefits of this approach over the status quo.

Faculty: Greg Lowry

Reactive Sorbents for In Situ Degradation of Per- and Polyfluorinated (PFAS) Contaminants in Groundwater

Per- and polyfluorinated (PFAS) compounds are ubiquitous groundwater contaminants. There is a great need to find tractable and sustainable solutions to control the migration of PFAS in groundwater aquifers. In situ sorption barriers with carbon are a leading approach to manage PFAS-contaminated groundwater, but over time, they become a long-term PFAS source. The student working on this project will work with PhD students to develop and test novel sorbent materials that are injectable into an aquifer but that can also degrade adsorbed PFAS compounds with an added amendment. The student will identify novel materials with the required properties, synthesize them, and test their efficacy under various groundwater chemical conditions.

This project is laboratory-based with a strong environmental chemistry focus. As a result, the student will learn how to use a number of different analytical techniques used for PFAS and fluoride analysis, as well as methods for synthesizing materials with controllable surface properties.

Faculty: Greg Lowry

Robotics and AI for Autonomous Characterization and Remediation of Contaminated Sites

Contaminated industrial and brownfield sites can cover vast areas. One of the most dangerous and expensive phases of remediation projects is site characterization to determine the extent of contamination. In this project, advances in robotic mobility platforms (e.g. legged robots), sensors, and AI/ML are leveraged to develop robots that can autonomously navigate to desired sampling locations, collect and prepare samples for analysis, accurately measure and analyze with an onboard sensor, and then use that information to decide where to sample next to reduce uncertainty in the contaminant distribution at the site.

Students will work with an interdisciplinary team of students and postdocs from environmental chemistry and robotics backgrounds to engineer autonomy into selected sensing platforms and adaptive sampling algorithms and field test those platforms to quantify performance.

Faculty: Greg Lowry

Automated 3D and 2D Image-Based Interpretation of Fire Patterns with Spatial Relationship Analyses

Fire pattern investigation is systemic to determining the cause and origin of fire. However, fire is a highly three-dimensional, time-variant process with variant boundary conditions. A fire investigator must know the physics and variables that influence a fire’s development. Reliable fire scene investigation depends on human inspectors. If investigators cannot interpret fire-pattern data correctly, their analyses may lead to errors and legal consequences. Interpretation is implicit and subject to perception bias, with the assignment of interpretation to patterns largely dependent on the investigator’s knowledge, experience, training, and skill, without the benefit of a structured framework to help guide the investigator through the process.

This problem could be solved by developing data-driven tools to analyze fire patterns automatically. Image-based techniques are emerging to support field judgments. Many fire patterns are spatially distributed, and their causal relationship to fire dynamics usually cannot be identified purely through their attributes; e.g., the spatial relationships of openings with fire patterns are necessary to identify whether fire patterns are ventilation-generated. This paper proposes an automated image-based interpretation system of fire patterns with spatial relationships based on 2D and 3D image analysis to understand the spatial relationships on fire scenes to help the investigator quantitatively interpret.

Faculty: Pingbo Tang

Human and Drone Interaction Through Process Mining and Reinforcement and Imitation Learning during Bridge Inspections

Bridge inspection is a labor-intensive, time-consuming, and sometimes dangerous process that needs the bridge engineer to visually inspect and analyze the observed bridge with the help of mechanical equipment. The bridge inspection industry has adopted unmanned aerial vehicles (or drones) to improve safety, efficiency, and cost-effectiveness. Due to the complexity of the on-site environments, such as the illumination, traffic, weather conditions, and different structural types of bridges, it is critical to keep the inspector in the control loop.

The drone-based inspection strategies about how experienced inspectors control the drone to find the defects could explain their knowledge and understanding of the structure. Capturing the inspection strategies is critical for transferring inspectors’ experiences to the autonomous drone for more efficient defect detection. This study examines a framework for capturing the inspectors’ inspection strategies for identifying cracks on the bridge by controlling drones in a simulated bridge inspection in Unity.

The comparison of different human-machine interactions: (1) the inspector controls a drone manually; (2) the autonomous drone learns to find the cracks by reinforcement learning; (3) the autonomous drone learns from the inspectors’ inspection strategies by imitation learning. The framework has built a platform for exploring advanced solutions to the human–drone cooperative inspection of the bridge.

Faculty: Pingbo Tang

Self-diagnosis and Predictive Control of Human-in-the-Loop Water Treatment Processes with Virtual Reality and Augmented Reality

Operating water treatment plants in an open and uncertain environment (changing demands and changing water properties) is challenging for humans and machine-learning-based control algorithms. Anomaly handling in such operation processes needs collaboration between humans and control algorithms to reduce safety, health, and efficiency concerns. This project aims to create self-evolving intelligent control systems that can continuously learn from control system sensor logs and operators’ actions while handling off-nominal scenarios.

The goal is to create a computational framework that can keep learning from operators to handle new configurations or scenarios and keep operators informed about emerging risks.  Such a framework will be adaptive to different water treatment plants and systems. More specifically, the primary goal of this study is to enable "proactive" computer systems to learn from operational histories and conduct self-diagnosis of real-time human-in-the-loop processes to suggest improved control strategies, which include 1) improving the mechanical system safety to ensure sustained water production without interruptions of water service; 2) enable high efficient performances of water treatments by achieving sufficient filtering and identifying possible ways of improving the filtering process designs; 3) make tradeoffs between the operating costs, downtime, performances, and water quality risks of the water system.

Faculty: Pingbo Tang

Robust Framework for Joint Ventilation System Component Production and Maintenance Scheduling in Dynamic Manufacturing Environments

This project will study a robust joint production and maintenance scheduling framework that ensures desired production performance in uncertain manufacturing environments. The project will involve investigating various factors that influence production performance, creating discrete event simulations to predict the impacts of these factors on production variations, and developing robust reinforcement learning-based methods that guarantee desired production performance despite uncertainties.

The proposed project will involve 1) investigating the factors that influence production performance through expert interviews and production records; 2) building discrete event simulations to predict the impacts of influential factors on production variations; 3) developing robust reinforcement learning-based methods that guarantee desired production performance considering the potential impacts of these factors.

Faculty: Pingbo Tang

Proactive Performance Deterioration Assessment in Heavy-Duty Vehicles for Safety and Efficiency of Commercial Fleets

Many truck accidents and crashes are highly influenced by defective truck equipment. The practical problem of conducting safety inspection programs while minimizing the impact on fleet mobility for commercial vehicles is ensuring fleet safety and effectiveness. Both safety and energy efficiency belong to a broader category called fleet performance. Efficiency is related to fuel or electricity consumption considered by an electric vehicle. They are analyzing the historical data related to safety and efficiency results in detecting the factors and vehicle attributes that mostly contribute to crashes.

Consequently, by knowing the types of those vehicles, we can derive their problems and predict which vehicles are more likely to be involved in crashes and consume more energy. This will help the drivers and fleet managers avoid repetitive inspection violations and improve fleet operational strategies. The project consists of connecting to large historical datasets, data cleaning, and applying natural language processing to text documents to make them understandable for the machine.

Faculty: Pingbo Tang

State-of-the-Art Review and Investigation of Workflow Stability in Modular Construction

This project aims to review and investigate the factors influencing workflow stability in modular construction. Modular housing, a type of off-site construction, provides opportunities for massive housing production with lower costs, a safer work environment, and higher productivity than on-site construction. Modular housing takes advantage of production in manufacturing, which uses assembly lines to provide large quantities of standardized products.  However, the current manufacturing approach adopted in modular housing constrains the diversity of house designs. Increasing demand for customized house design challenges the risk management of costs and schedules in modular housing. A lack of systematic investigation of the relationship between different factors and manufacturing workflow stability impedes the development of flexible modular housing.

This research first investigates the relationship between the attributes of modular housing design, work packages, and manufacturing productivity, including the elements, workflow, and uncertainties in production. Then, this research constructs an agent-based simulation to identify the relationship between multiple design components and workflow stability in modular construction using the attributes specified in the investigation. The outcome of the research will contribute to relieving the workload of the modular housing designer and support the factory manager in assisting with flexible modular housing designs.

Faculty: Pingbo Tang

Investigating water quality changes in coastal aquifers due to sea-level rise

Sea-level rise combined with over-exploitation of groundwater resources threatens the water quality of coastal aquifers. Degradation of groundwater quality can include contamination ranging from increased salinity to more complex geochemical perturbations impacting the fate and transport of a variety of contaminants.

This project seeks to decipher geochemical shifts occurring in coastal aquifers due to sea-level rise by applying data analytics to large datasets for groundwater levels, water quality and observed sea-level rise. Results will be used to inform potential future shifts in coastal groundwater quality and the development of effective strategies to protect freshwater resources for human and ecosystem health.

Faculty: Sarah Fakhreddine and David Rounce

Evaluating opportunities and constraints on future freshwater management

Managed aquifer recharge (MAR) is increasingly used to enhance local water supplies and build freshwater resilience. MAR involves capturing excess water supplies in wet periods for subsurface storage in depleted aquifers. Water can then be recovered for later use in dry periods or times of high demand. While MAR adoption is rapidly increasing, current planning for MAR heavily relies on historic data of hydrologic conditions rather than future projections that account for population growth and climate change. This projects involves a comprehensive evaluation of future opportunities and limitations of MAR, including quantifying excess water availability, identifying potential locations for subsurface storage, and simulating downstream impacts to community water rights, environmental flows, and water quality. This study uses hydrologic models coupled with water accounting tools to better understand and inform the viability of future freshwater management strategies.

Faculty: Sarah Fakhreddine