Austin Metcalf
Austin Metcalf
Engineer at LANL
Engineer at LANL
Graduate student at Johns Hopkins University (Whiting)
Graduate student at Johns Hopkins University (Whiting)
Embedded Files
Cogeneration Power Plant Optimization
Cogeneration Power Plant Optimization
This semester-long project involved the complete thermal systems engineering lifecycle: conceptualization, component modeling, system simulation, optimization formulation, and solution. The project analyzed a cogeneration (combined heat and power) plant inspired by the facility at Rowan University in Glassboro, New Jersey, where I completed my undergraduate education.
Cogeneration plants are a compelling application of thermodynamic principles because they extract useful work from energy that would otherwise be wasted. Rather than rejecting heat to the environment through a condenser (as in a conventional power plant), a cogeneration system redirects thermal energy to serve heating loads, in this case, campus housing and academic buildings during winter months.
The challenge was to design a system that could simultaneously satisfy two competing requirements: provide sufficient heating for approximately 6,500 students while generating a meaningful amount of electricity. This dual-objective creates inherent trade-offs that make the system an excellent candidate for optimization.
Machine Learning: optimization of concrete composites
Machine Learning: optimization of concrete composites
In this final project at Johns Hopkins University, our team achieved a remarkably accurate model by developing a multi-paradigm machine learning system that expanded a limited 425-datapoint concrete dataset into a robust 50,000-point repository through advanced data augmentation techniques. We successfully implemented and compared six different machine learning models, utilized Generative Adversarial Networks (GANs) to synthesize realistic concrete mixture data, and developed two sophisticated regressors that demonstrated exceptional predictive accuracy with minimal error rates. Our optimization algorithm successfully identified the ideal concrete mixture design that maximizes compressive strength-to-weight ratio while satisfying complex engineering constraints, ultimately achieving 38.75 MPa compressive strength with optimal material utilization. This project showcased our ability to tackle real-world engineering challenges by seamlessly integrating cutting-edge AI techniques including Random Forest regression, Principal Component Analysis (PCA) visualization, combinatory feature analysis, and constrained optimization, demonstrating how machine learning can transform sparse engineering data into actionable insights for material design optimization.
bearing prognostics NASA Ames Intelligent Systems Division
bearing prognostics NASA Ames Intelligent Systems Division
I developed and validated acoustic analysis methods to enable early detection of bearing faults in critical rotating machinery, addressing a major cause of 40-50% of motor failures in industrial settings. Using NASA vibration datasets, I compared five different signal processing techniques including time-domain analysis, frequency-domain analysis, spectrograms, characteristic frequency identification, and root mean square calculations. Through systematic analysis of bearing degradation data, I identified that characteristic frequency analysis provides the earliest and most reliable fault detection, enabling identification of outer race failures approximately halfway through the bearing's remaining useful life. This early warning capability allows operations teams to schedule maintenance proactively, preventing catastrophic failures that can damage motors, adjacent equipment, and cause costly production downtime. The methodology I developed directly supports maintenance strategies for turbomachinery systems where unplanned failures pose significant operational and safety risks.
design of Flat burner for Material testing
design of Flat burner for Material testing
My senior design project has revolved around the topic of combustion; we have created and designed multiple facilities that have unique purposes and solve different problems. The flat flame burner (also known as the McKenna burner) is a tool that can produce a flame that is completely flat and uniform. In the image above we use this heat source to test the thermal resilience of a sample. The motivation of developing the flat flame burner came from the need to test materials that could be used to protect the hull of an aircraft moving incredibly quickly. The air-friction would cause the aircraft to heat to extreme temperatures; ergo we provide a solution that is precise and repeatable for research labs to use for initial testing.
Design and construction of Bunsen Burner
Design and construction of Bunsen Burner
My senior project also involved creating a Bunsen burner to study flame speed. The flame speed is characterized by an intrinsic property of combustion called the laminar burning velocity. We aim to study the laminar burning velocity for esoteric fuels like nitrous oxide. The Bunsen burner design allows for multiple gas flows and includes a gas tank that allows for premixing to occur.
Design and modeling of constant volume combustion chamber
Design and modeling of constant volume combustion chamber
Another outcome of my senior project was the design of a pressure vessel; a partner and myself have completely designed a combustion chamber to study the laminar burning velocity under high-pressure conditions. The vessel is approximately 6 inches in length and 6 inches in diameter. We have used simulation results to confirm the structural integrity of the chamber, bolts, windows, and flanges. We have collaborated with machinists and technicians to ensure our design is robust and machinable. [Read the paper here]
Ball and beam w/ pid control
Ball and beam w/ pid control
For this project, I built a beam that can be controlled by a servo using a four-bar linkage. I have used a soft-potentiometer, an Arduino, and a dial to communicate where I want the ball to be balanced (steady-state). The code used integrates a PID control library. I tuned the PID controller to reach steady-state under 20 seconds.
ASME/RAS Trebuchet Team Pumpkin chunkin 2022
ASME/RAS Trebuchet Team Pumpkin chunkin 2022
This is a snapshot of Rowan's ASME team competing in the Fall 2022 Pumpkin Chunkin event. We designed and built this traditional trebuchet to compete against other local schools and universities. With our design we managed to place in the finals after the first elimination round.
MATLAB: Conduction Heat Transfer Analytical Solutions
MATLAB: Conduction Heat Transfer Analytical Solutions
Using MATLAB to visualize how heat propagates through a material given certain boundary conditions, I employ the method of relaxation to estimate how temperatures change over time within each unit cell. My MATLAB code is verified by performing hand calculations, and using SolidWorks.
gear design for compressor drive train
gear design for compressor drive train
Using AGMA standards, I performed a rigorous and robust design of spur gears used for a compressor drive train. Provided to me was the torque-time function and certain properties to help determine design factors. The analysis showed my individually designed gears would withstand the operating conditions of a 10 year shift cycle.
Robotic Claw Design
Robotic Claw Design
Four students including myself worked together to build a robotic arm from some basic electrical components, 3D printed, laser cut parts and screws. This project requires coordination between members, and heavily incentivizes members to delegate responsibilities based on skill-set. The robotic arm successfully transferred a payload 120° from a starting position.
Manufacturing a Digital Scale
Manufacturing a Digital Scale
For this project I designed and manufactured my own scale. The outer casing, the wiring, the soldering, and the coding all came with their own design considerations. This project showed me how to go from product idea to prototype. The scale successfully passed all calibration tests.
Authenticate DAQ Functionality
Authenticate DAQ Functionality
Rowan University often will create tools that expedites certain cumbersome process in Engineering Education. This data acquisition board is meant to use different electrical components to take temperature readings. My task was to review the code the board used, and to ensure the board performed it's task before being introduced to students. I created a manual that would onboard a student quickly, to start taking measurements.
Modeling Basic Plane Flows with MATLAB
Modeling Basic Plane Flows with MATLAB
Using MATLAB, I created a script that when given some inputs that represented the shape of an object, I could see how the object would disturb a uniform flow, which could be air or some other fluid. The image above represents the contour lines of the stream function, which provides a clear visual of the behavior of a fluid close to our object.
Analysis of Thermo-Fluid Cycles
Analysis of Thermo-Fluid Cycles
Using Google Colab, I worked with a partner to solve real-world design problems that involve thermodynamic cycles like a Rankine, or Brayton Cycle. We used Python and PYroMat to import thermodynamic properties to perform our analysis.
Differential Equations Group Solution
Differential Equations Group Solution
The final project for my differential equations course included a group project where we would solve a difficult real-world problem, and present our solution in an entertaining yet informative way to a diverse audience. Clarity and conciseness were top priorities, we included just enough mathematics that any STEM student could independently verify our result.
Page updated
Google Sites
Report abuse