What exactly constitutes a machine learning (ML) model? Correspondingly, what constitutes a Data-Driven model, and what is the difference? Why do we need to assess credibility of ML models for applications of high consequence (i.e., involving human health)? Given the principles of verification, validation (V&V), and uncertainty quantification (UQ) customarily applied to structural mechanics and fluid dynamics, what’s the extent to which we can adopt those for ML and data-driven models? When we know the physics involved, how does incorporation of governing equations benefit the ML modeling (or not)? We here pose a candidate dataset by which one can develop ML models of differing forms, recognizing that no single ML model perfectly characterizes the information. In light of such model imperfections, how might we effectively develop frameworks whereby one might meaningfully assess the credibility of ML models, given the opaque (i.e., “black box” nature) of them, and non-uniqueness challenges.

Your work would involve use of such computational tools as Matlab and Python exercised on datasets of reasonably representative size for real-world applications, but small enough to handle on desktop computers. We can use the computational studies to draw conclusions, write reports or presentations, and contribute to the American Society of Mechanical Engineers. If interested, please contact Prof. Banyay (gab5631@psu.edu).

4D-Flow magnetic resonance imaging (MRI) is one of the few technologies that can be used to obtain velocity and flow information within a brain aneurysm in a patient. This flow information is critical to help physicians determine if a brain aneurysm is likely to rupture and requires immediate surgical intervention. However, current 4D-Flow MRI capabilities are limited; it can only provide flow fields with very low spatial and temporal resolution which results in over a 40% error in flow metrics. In vitro (benchtop) 4D-Flow MRI flow data is needed in order to develop reconstruction algorithms to improve 4D-Flow MRI velocity fields. This requires the development of a portable flow loop, whose components are all MRI-compatible (i.e., non-magnetic). For this project, the student will build and optimize a flow loop design, considering things such as different pump options and the possible incorporation of pneumatic pressure ports to generate repeatable unsteady flow waveforms. The designed flow loop will be tested using the 4D-Flow MRI scanner available here at Penn State’s University Park campus. The flow loop will be optimized for portability as well as it will need to be regularly transported from Dr. Brindise’s lab in Reber Building to the campus MRI scanner (SLEIC) and back for future experiments.

Microplastics (MPs) are a broad and ubiquitous class of pollutants, which may be detrimental to the health of both humans and larger-scale ecosystems [1], [2]. We are learning that MPs can be found almost everywhere, but we don’t usually know how they got there—the physical mechanisms of MP transport are largely unstudied. Current research focuses largely on the presence of MPs, rather than the pathways (i.e. origins and subsequent transport). This is reflected in a body of literature that skews toward marine MPs, even though the majority of MP sources are terrestrial [3]. Furthermore, MPs are often treated with a lack of specificity: though MPs can be composed of very different materials and have dramatic variation in shape [4], they are frequently viewed as a single monolithic category. This approach may handicap analysis and muddy potential mitigation strategies and/or policy recommendations.

For this project, the honors student will measure the quiescent settling/rising velocities of several common microplastic shapes including fragments, films, fibers, pellets, et al. The student will then compare the measured velocities to those observed in laboratory-generated homogeneous, isotropic turbulence.  The project will involve working with high-speed imaging, with a possible extension to 3D kinematic tracking and/or laser-based velocimetry. If interested, contact Dr. Margaret Byron at mzb5025 [at] psu [dot] edu. 

[1]      S. L. Wright, R. C. Thompson, and T. S. Galloway, “The physical impacts of microplastics on marine organisms: A review,” Environ. Pollut., vol. 178, pp. 483–492, Jul. 2013.

[2]      L. G. A. Barboza, A. Dick Vethaak, B. R. B. O. Lavorante, A. K. Lundebye, and L. Guilhermino, “Marine microplastic debris: An emerging issue for food security, food safety and human health,” Marine Pollution Bulletin, vol. 133. Elsevier Ltd, pp. 336–348, 01-Aug-2018.

[3]      J. R. Jambeck et al., “Plastic waste inputs from land into the ocean,” Science, vol. 347, no. 6223, pp. 768–71, Feb. 2015.

[4]      I. Chubarenko, A. Bagaev, M. Zobkov, and E. Esiukova, “On some physical and dynamical properties of microplastic particles in marine environment,” Mar. Pollut. Bull., vol. 108, no. 1–2, pp. 105–112, Jul. 2016.

At its core, thermodynamics is about the study of energy and energy transfer and thermodynamics expertise is crucial given contemporary human reliance on energy. Unfortunately, for many students this foundational engineering course is overwhelming and often associated with poor performance. While thermodynamics, like all engineering, involves quantitative problem-solving, the real challenges for learners lie in its conceptual and integrated nature that requires students to formulate problems and identify proper assumptions before doing the more mathematical portion of problem-solving, with which they are typically more comfortable. In teaching complex engineering curriculum like thermodynamics, both active learning teaching approaches and strategies that encourage students’ metacognition have been shown to contribute to meaningful learning gains. Using active learning and metacognition together has also received attention as potentially more impactful than either strategy alone. 

Sometimes colloquially described as simply “thinking about thinking,” metacognition involves both knowledge of and beliefs about one’s cognition and the strategies employed to regulate one’s learning. It is generally thought that the more students are aware of their own cognitive processes and also able to regulate them, the more effective they will be at learning. While there are multiple ways of conceptualizing metacognition, a common framework distinguishes metacognitive knowledge (MK), which includes knowledge of persons, tasks, and strategies, from metacognitive regulation (MR), which includes monitoring, planning, evaluation, and control. 

The goal of this project is to use data collected from several asynchronous online and synchronous resident offerings of ME 300: Engineering Thermodynamics to identify ways that students are using metacognitive regulation to help enhance their learning. Several course artifacts, including exam wrappers, feedback surveys, and reflective writings will be analyzed for examples of MR and the “level” of MR, which is an indicator of the depth of their regulation abilities. The results of these analyses, in conjunction with findings from the literature, will be incorporated into new teaching strategies for improving student MR and hopefully their metacognitive knowledge as well. These new strategies will be tested and analyzed for their efficacy, then improved for future offerings. 

Project requirements: student researcher on this project will be required to take ethics training to work with human subjects data and will be expected to handle data and study details appropriately by following study protocols. The student researcher will be conducting qualitative coding and analysis as well as descriptive and basic inferential statistical analyses. The project timeline is as follows: 

• Fall 2022: analysis of previous data and determination of course improvements 
• Spring 2023: implementation of improvements in ME 300 resident offering; take thesis-writing course (ME 397) to write literature review; determine course improvements for online offering 
• Summer 2023: implementation of improvements for online offering (student does not need to be present for summer research) 
• Fall 2023: analysis of Sp23 and Su23 data to determine the efficacy of improvements; begin writing results in both ASEE paper as well as thesis 
• Spring 2024: completion of thesis and ASEE paper; presentation of ASEE paper 

Musculoskeletal injuries are a major cause of morbidity and disability. The standards of care for bone fracture are evolving, and it is now understood that controlled mechanical loading can improve the quality and speed of fracture repair. Additive manufacturing techniques may vastly improve mechano-therapeutics by allowing for the development of biodegradable trauma implants with site-specific surface morphologies. These devices can be tuned to deliver therapeutic mechanical loads and eventually degrade entirely. Zinc and zinc alloys are ideal materials for this purpose, but little is known about the mechanical behavior of AM zinc constructs and effects of AM surface morphology. From a biological standpoint, zinc is known to stimulate new bone formation, preserve bone mass, and regulate cellular apoptosis. The purpose of this study is to develop a series of cohesive in vivo and in vitro experiments that can elucidate relationships between mechanical load transfer and biological responses elicited by AM zinc implants in bone fracture healing.

Ref: nature.com/articles/s41596-019-0271-2

Animals move with maneuverability and agility that is unmatched by current robots. As engineers, we seek inspiration from biology to inspire the next generation of agile robots. For instance, flying insects can sustain damage to a wing and readily compensate whereas the best flying robots would certainly crash in an instant. The goal of this project is to reveal Nature’s secrets behind the unmatched robustness of insect flight. In particular, the student will study how sensing of visual and mechanical origin is coupled with rapid wing movement for effective flight control. Tasks for this project will include quantifying insect flight behavior in a virtual reality flight simulator (Figure 1), reconstructing 3D motion of the wings from high-speed camera using machine vision algorithms and modeling the insect’s response using mathematical methods. Students will work with an experienced graduate student and gain experience in computer vision, control theory, modeling of dynamic systems, and coding in Matlab/Python. The results from this study will yield fundamental new insights into the mechanisms that enable effective flight control and inspire the development of agile aerial vehicles.  Past Schreyer students in our lab have been authors on conference abstracts presented at international conferences and peer-reviewed journal publications.

Figure 1. Fly magnetically levitated in a virtual reality flight simulator.

As part of a National Science Foundation-sponsored project, we are seeking up to two undergraduate researchers to participate in research focused on the design, fabrication and characterization of magnetic field actuated polymer composites. The objective is to investigate the fabrication and application of magneto-mechanical (smart) composite actuators. The fabrication process will focus on field-assisted 3D printing of multimaterials. The carefully assessed magneto-mechanical responses will be leveraged to develop complex actuations, as shown in the figures.

Responsibilities include fabricating (3D printing) responsive polymer composites under the guidance of a graduate student; characterizing the mechanical and magnetic properties of the 3D printed composites; working closely with a graduate student to demonstrate repeatable magneto-mechanical response; and attending regular research meetings to discuss progress.

Interested students should email Dr. Ounaies at zxo100@psu.edu. Applications from those who are traditionally underrepresented in engineering are particularly encouraged.

Objective: The objective of this project is to conduct extensive testing on the thermal and hydraulic performance of liquid-cooled heat sinks under consideration for patenting and licensing.

Activities: (1) Characterization of the heating load delivered by electrical heaters at different operating conditions. Fourier’s law and a three-thermocouple measurement will be used for this. (2) Characterization of the flow rate from a pump under different valve opening conditions. (3) Pressure drop measurements at different flow rates. (4) Determination of the effective thermal resistance of different heat sinks as a function of flow rate. (Optional) If time allows it and the student shows proficiency at computational fluid dynamics modeling, novel heat sinks can be designed, built, and tested.

Deliverables: A report (thesis) of the experimental data for the pre-built heat sinks and comparison with commercially available models.

Project timeframe: 1 year.

Infectious diseases resulting from antibiotic-resistant bacterial pathogens, or superbugs, represent a major global healthcare challenge. The Centers for Disease Control and Prevention estimates that at least two million illnesses and 23,000 deaths are caused by antibiotic-resistant bacteria in the United States each year. However, existing microbiological diagnostic techniques require at least 3-5 days. The significant delay in diagnosis drives the empirical use of antibiotics, which results in poor patient outcomes and accelerates the emergence of multidrug-resistant bacteria. This project aims to address this global health challenge by developing a microfluidic system for rapid antimicrobial susceptibility testing (AST) to improve antimicrobial stewardship. The project will involve designing and implementing a microfluidic device for single cell AST. Students will gain experience in handling biological samples, performing optical microscopy, and fabricating medical devices. Students will also learn to apply machine learning techniques to improve the accuracy and speed of diagnostic devices. Students will participate in meetings with our clinical and industrial collaborators. If interested, please contact Prof. Pak Kin Wong (pxw28@psu.edu).

Work is being conducted in the Combustion Research Laboratories under Prof. Rich Yetter to investigate the use of new metal hydride fuels as additives in hybrid rocket motors. These new fuels are first being analyzed using an Opposed Flow Burner, as seen in Figure 1. This test apparatus flows oxidizer perpendicular to the burning fuel, forming a flame at the stagnation plane between the fuel rod and oxidizer nozzle, as seen in Figure 2. A spring keeps the fuel surface at a constant location and an LVDT measures the rate at which the fuel is consumed.

The regression rate (i.e. how quickly the fuel burns) is a key quantity for the design and application of hybrid rocket motors. The greater the regression rate, the higher the thrust of the hybrid motor, historically the limiting factor in the successful implementation of hybrid motors in space launch applications. The Opposed Flow Burner is able to semi-quantitatively analyze the difference in regression rates of various fuel/oxidizer combinations and visualize the surface of the regressing fuel, all at atmospheric pressure. This allows for rapid analysis of numerous test articles of varying compositions.

The Opposed Flow Burner has been used previously in the lab, but work is required to refurbish and improve upon the system. The objective for the Honors student research will be to refurbish the Burner, test fuel samples prepared by a PhD student under various conditions, and devise a way of heating the oxidizer prior to injection into the burner through the use of a fluidized heated bed or other heat exchange method. Using a heated oxidizer will also require the fabrication of new hardware for the burner that can withstand the higher temperatures.

Figure 1: Opposed Flow Burner

Figure 2: Example of combustion experiment in Opposed Flow Burner