Current Projects:
Coming soon!
Previous Projects:
Flapping Foil Trajectory Control
DwasForce control in flapping foils using in-line motion and passive pitch
MIT Tow Tank, Undergraduate Thesis
January - May 2017
Overview: The goal of this project was to control the trajectory of a flapping foil with in-line motion and a set-up that allows for feathering, or passive pitching of the foil. This effectively removes pitch angle control and reduces the system from 3 to 2 degrees of freedom. Controlling the trajectory of the foil allows for the control of the thrust and lift forces on the foil. If similar force control to a 3-DOF system can be achieved for a system where the foil is able to feather during part of its trajectory, the complexity of the system will be reduced. This should also eliminate the need to optimize the foil trajectory over multiple trials. I designed and built an adjustable stopper that limited the maximum pitch angle the foil. I ran experiments with this set-up and collected data on the hydrodynamic forces on the foil for two distinct trajectories.
Results: I found that force control of a flapping foil is possible for a passive foil system, but this force control is limited to forwards in-line motion trajectories, or lift force control. While the passive system can generate substantial thrust force for backwards in-line motion trajectories, the performance of the force control is poor in this regime compared to that of an optimized, fully actuated system. Therefore, while optimization is wholly unnecessary for lift force control, it may be required for thrust force control.
Lessons & Skills: Over the course of my thesis work, I gained experience integrating hardware and software, designing and machining custom parts, and designing and troubleshooting experiments. I also improved my technical writing skills, practiced making professional figures and charts, and improved my data analysis and communication skills.
MIT Tow Tank, Undergraduate Thesis
January - May 2017
Overview: The goal of this project was to control the trajectory of a flapping foil with in-line motion and a set-up that allows for feathering, or passive pitching of the foil. This effectively removes pitch angle control and reduces the system from 3 to 2 degrees of freedom. Controlling the trajectory of the foil allows for the control of the thrust and lift forces on the foil. If similar force control to a 3-DOF system can be achieved for a system where the foil is able to feather during part of its trajectory, the complexity of the system will be reduced. This should also eliminate the need to optimize the foil trajectory over multiple trials. I designed and built an adjustable stopper that limited the maximum pitch angle the foil. I ran experiments with this set-up and collected data on the hydrodynamic forces on the foil for two distinct trajectories.
Results: I found that force control of a flapping foil is possible for a passive foil system, but this force control is limited to forwards in-line motion trajectories, or lift force control. While the passive system can generate substantial thrust force for backwards in-line motion trajectories, the performance of the force control is poor in this regime compared to that of an optimized, fully actuated system. Therefore, while optimization is wholly unnecessary for lift force control, it may be required for thrust force control.
Lessons & Skills: Over the course of my thesis work, I gained experience integrating hardware and software, designing and machining custom parts, and designing and troubleshooting experiments. I also improved my technical writing skills, practiced making professional figures and charts, and improved my data analysis and communication skills.
BB8 Yo-Yo
Design and Small-scale Production of a Yo-Yo
2.008: Design and Manufacturing II, semester group project
September - December 2016
Overview: In 2.008, teams of 5 people were tasked with designing and manufacturing 50 yo-yos. Our group's yo-yo was based on the Star Wars character BB-8. The character’s head moves freely while the body rotates. This inspired us to design a yo-yo with a free spinning body and a head that independently hovers above. The body was divided into two hollow hemispheres, each with a press fit inner face. Each inner face piece had an insert molded hex nut and radial ribs to increase strength. The head was a smaller, solid hemisphere with a smooth hole through the center. The string was threaded through the head and wound in the gap between the two body pieces. In order to keep the head close to the body while the yo-yo spun, we insert molded a steel shim into the inner face and press fit a magnet into the head hemisphere. The entire yo-yo will sat on top of a thermoformed and spray painted “sand dune” base. We 3-D printed a prototype of our yo-yo to test our design, which worked surprisingly well. We injection molded the body hemisphere, inner face, and the head. The body and head were spray painted, using 3D printed parts as stencils. To incorporate for design for manufacturing principles, we reduced the number and complexity of our parts to just 3 unique Injection Molded parts and one thermoformed part.
Results: We designed the yo-yo as a team; we knew we wanted to have a yo-yo that moved like the BB8 character, so we brainstormed ideas for a "levitating" head and settled on the magnet-shim combination. I designed and made the molds for the inner face of the body, which included an insert molded steel shim. I ran the test and production runs for this part on the Engel injection molding machine. I also filmed, directed, and edited our promo video for the yo-yo. We worked together to design and execute our multi-stage assembly line (see our video) and put together the poster for the 2.008 final presentation and yo-yo expo.
Lessons & Skills: I gained experience in SolidWorks while designing the molds and learned to use MasterCAM to code tool paths. I also learned to use and Engel injection molding machine and modify parameters to improve part production. Finally, I learned a bit about filming and video editing in Adobe Creative Cloud.
2.008: Design and Manufacturing II, semester group project
September - December 2016
Overview: In 2.008, teams of 5 people were tasked with designing and manufacturing 50 yo-yos. Our group's yo-yo was based on the Star Wars character BB-8. The character’s head moves freely while the body rotates. This inspired us to design a yo-yo with a free spinning body and a head that independently hovers above. The body was divided into two hollow hemispheres, each with a press fit inner face. Each inner face piece had an insert molded hex nut and radial ribs to increase strength. The head was a smaller, solid hemisphere with a smooth hole through the center. The string was threaded through the head and wound in the gap between the two body pieces. In order to keep the head close to the body while the yo-yo spun, we insert molded a steel shim into the inner face and press fit a magnet into the head hemisphere. The entire yo-yo will sat on top of a thermoformed and spray painted “sand dune” base. We 3-D printed a prototype of our yo-yo to test our design, which worked surprisingly well. We injection molded the body hemisphere, inner face, and the head. The body and head were spray painted, using 3D printed parts as stencils. To incorporate for design for manufacturing principles, we reduced the number and complexity of our parts to just 3 unique Injection Molded parts and one thermoformed part.
Results: We designed the yo-yo as a team; we knew we wanted to have a yo-yo that moved like the BB8 character, so we brainstormed ideas for a "levitating" head and settled on the magnet-shim combination. I designed and made the molds for the inner face of the body, which included an insert molded steel shim. I ran the test and production runs for this part on the Engel injection molding machine. I also filmed, directed, and edited our promo video for the yo-yo. We worked together to design and execute our multi-stage assembly line (see our video) and put together the poster for the 2.008 final presentation and yo-yo expo.
Lessons & Skills: I gained experience in SolidWorks while designing the molds and learned to use MasterCAM to code tool paths. I also learned to use and Engel injection molding machine and modify parameters to improve part production. Finally, I learned a bit about filming and video editing in Adobe Creative Cloud.
Flapping Foil Propulsion
Dynamic modeling and control of flapping foil propulsion
Fluid Structure Interactions Group at the University of Southampton, MISTI-UK summer internship
June - August 2016
Overview:
This summer I studied the behavior of flapping foil systems. Flapping foils are devices modeled on the swimming mechanism for cetaceans and many fish. These devices have been widely researched, and it has been shown through experiments and simulation that flapping foils can be used to both generate power and produce thrust. The relationship between the magnitude and phase of the foil heave and pitch motions defines these two modes. Previous research suggests that the angle of attack and wake profiles are key to the propulsive power generated by the foil. Flapping foils have been successfully implemented in several full scale applications, including ships and wave energy converters (WECs). Currently, the optimal control and stability of flapping foil systems is being investigated, and more research is needed to solve this problem. The system for this project was a model vessel with two pairs of flapping foils attached at the bow and stern. The frequency response of the foils was being investigated for mechanical and wave actuation of the foil heave. The goal for this project was to gain an understanding of the dynamic behavior of the system and design a velocity controller for the vessel.
Results: I modeled the dynamic behavior of the system in MATLAB for a simplified case. This model output the frequency response of the vessel velocity as well as the angle of attack over a range of frequencies. I configured the linear foil actuators, through the CME2 software, and figured out how to control the actuators through a CompactRIO. I ran several open loop tests with the model vessel, FLEUR (left), and determined the useful range of flapping amplitudes and frequencies for this vessel.
Future Work: The ultimate goal is to use the open loop data along with the dynamic model to design a feedback controller for the system that controls the velocity of the vessel by adjusting the heave of the foil.
Lessons & Skills: I was unable to work on the controller design for this project because I ran out of time, but I completed a dynamic model of the foil in MATLAB, studied hydrofoil theory, learned basic LabVIEW code, and learned how to use a CompactRIO to interface with on-board sensors. After this experience, I have a better understanding of data collection hardware and software, and I feel better equipped to execute my own experiments.
Fluid Structure Interactions Group at the University of Southampton, MISTI-UK summer internship
June - August 2016
Overview:
This summer I studied the behavior of flapping foil systems. Flapping foils are devices modeled on the swimming mechanism for cetaceans and many fish. These devices have been widely researched, and it has been shown through experiments and simulation that flapping foils can be used to both generate power and produce thrust. The relationship between the magnitude and phase of the foil heave and pitch motions defines these two modes. Previous research suggests that the angle of attack and wake profiles are key to the propulsive power generated by the foil. Flapping foils have been successfully implemented in several full scale applications, including ships and wave energy converters (WECs). Currently, the optimal control and stability of flapping foil systems is being investigated, and more research is needed to solve this problem. The system for this project was a model vessel with two pairs of flapping foils attached at the bow and stern. The frequency response of the foils was being investigated for mechanical and wave actuation of the foil heave. The goal for this project was to gain an understanding of the dynamic behavior of the system and design a velocity controller for the vessel.
Results: I modeled the dynamic behavior of the system in MATLAB for a simplified case. This model output the frequency response of the vessel velocity as well as the angle of attack over a range of frequencies. I configured the linear foil actuators, through the CME2 software, and figured out how to control the actuators through a CompactRIO. I ran several open loop tests with the model vessel, FLEUR (left), and determined the useful range of flapping amplitudes and frequencies for this vessel.
Future Work: The ultimate goal is to use the open loop data along with the dynamic model to design a feedback controller for the system that controls the velocity of the vessel by adjusting the heave of the foil.
Lessons & Skills: I was unable to work on the controller design for this project because I ran out of time, but I completed a dynamic model of the foil in MATLAB, studied hydrofoil theory, learned basic LabVIEW code, and learned how to use a CompactRIO to interface with on-board sensors. After this experience, I have a better understanding of data collection hardware and software, and I feel better equipped to execute my own experiments.
The WallKnocker
Development of an accurate stud finder using system identification techniques
2.131: Advanced Instrumentation and Measurement (graduate level), semester project
February - May 2016
Overview: Through the semester, students worked in groups to design a portable measurement device that employed the principles of system identification. The device had to actively probe the system it was trying to measure, collect data and analyze the system response, and then communicate information about the system to the user. Essentially, we were packaging the scientific method into a hand-held product. My group chose to design a stud-finder that input a "knock" to a wall (the system), measured the frequency response of the wall with four piezo sensor "feet", and analyzed this information to determine the presence or absence of a stud. The project tasks can be broadly broken into five parts: actuator and sensor selection, circuit design, design of chassis for actuator and sensors, experiment design, and data analysis. I mainly focused on the circuit design, experiment design, and data analysis, but I learned a great deal about rapid prototyping and circuit manufacturing from observing my fellow group members.
Results: We were consistently able to observe a difference between the oscilloscope output signal when the device was over a stud and when it was over a cavity. However this signal was too rich in information and our ability to filter noise was not good enough to extract this information in the data processing phase. The deliverables for the class were a presentation, demonstration of the device, and a short paper detailing our design process (please see below).
Lessons & Skills: Although we were not able to arrive at a concrete demonstration of the device, I learned a great deal about system identification, methods for designing experiments, and the power and pitfalls of data analysis techniques. I also enjoyed the hands-on nature of the project and the lectures. I have expanded my engineering "toolbox", and am now more comfortable working in the lab and designing my own experiments.
2.131: Advanced Instrumentation and Measurement (graduate level), semester project
February - May 2016
Overview: Through the semester, students worked in groups to design a portable measurement device that employed the principles of system identification. The device had to actively probe the system it was trying to measure, collect data and analyze the system response, and then communicate information about the system to the user. Essentially, we were packaging the scientific method into a hand-held product. My group chose to design a stud-finder that input a "knock" to a wall (the system), measured the frequency response of the wall with four piezo sensor "feet", and analyzed this information to determine the presence or absence of a stud. The project tasks can be broadly broken into five parts: actuator and sensor selection, circuit design, design of chassis for actuator and sensors, experiment design, and data analysis. I mainly focused on the circuit design, experiment design, and data analysis, but I learned a great deal about rapid prototyping and circuit manufacturing from observing my fellow group members.
Results: We were consistently able to observe a difference between the oscilloscope output signal when the device was over a stud and when it was over a cavity. However this signal was too rich in information and our ability to filter noise was not good enough to extract this information in the data processing phase. The deliverables for the class were a presentation, demonstration of the device, and a short paper detailing our design process (please see below).
Lessons & Skills: Although we were not able to arrive at a concrete demonstration of the device, I learned a great deal about system identification, methods for designing experiments, and the power and pitfalls of data analysis techniques. I also enjoyed the hands-on nature of the project and the lectures. I have expanded my engineering "toolbox", and am now more comfortable working in the lab and designing my own experiments.
Fast Tool Servo Controller Design
Design and implementation of a PID controller for a FTS system
2.14: Analysis and Design of Feedback Control Systems, final project
February - May 2016
Overview: This design problem focused the control of a fast tool servo (FTS), and was based upon the thesis done by one of the professor's PhD students. The design problem involved high-bandwidth current controller for driving the actuator, identification of the electro-mechanical plant dynamics from a measured Bode plot, and design of a controller to minimize following error for a sinusoidal reference trajectory in the presence of sensor noise. Exact servo performance metrics were not specified; instead students were expected to determine the standards and strive for good performance. The measure of performance was the reduction of the following error due to the reference trajectory and sensor noise. We began by designing the controller for the current loop, and then designed the FTS position controller.
Deliverable: The report for this project was a detailed explanation of our controller design process. This included tuning the current control loop, fitting a transfer function to the experimental mechanical plant BODE plot, methodology for designing the controller, and analyzing stability with a Nyquist test. Please see below for an excerpt from the final report.
Lessons & Skills: This project was an excellent opportunity to apply all of the techniques and concepts in linear control theory that we studied during the semester to a real system. I gained experience using MATLAB for control problems, learned how to analyze and build circuits for a variety of systems and how to build them, and deepened my knowledge of control theory. This class, and the final project in particular, inspired me to continue my study of control systems in graduate school.
2.14: Analysis and Design of Feedback Control Systems, final project
February - May 2016
Overview: This design problem focused the control of a fast tool servo (FTS), and was based upon the thesis done by one of the professor's PhD students. The design problem involved high-bandwidth current controller for driving the actuator, identification of the electro-mechanical plant dynamics from a measured Bode plot, and design of a controller to minimize following error for a sinusoidal reference trajectory in the presence of sensor noise. Exact servo performance metrics were not specified; instead students were expected to determine the standards and strive for good performance. The measure of performance was the reduction of the following error due to the reference trajectory and sensor noise. We began by designing the controller for the current loop, and then designed the FTS position controller.
Deliverable: The report for this project was a detailed explanation of our controller design process. This included tuning the current control loop, fitting a transfer function to the experimental mechanical plant BODE plot, methodology for designing the controller, and analyzing stability with a Nyquist test. Please see below for an excerpt from the final report.
Lessons & Skills: This project was an excellent opportunity to apply all of the techniques and concepts in linear control theory that we studied during the semester to a real system. I gained experience using MATLAB for control problems, learned how to analyze and build circuits for a variety of systems and how to build them, and deepened my knowledge of control theory. This class, and the final project in particular, inspired me to continue my study of control systems in graduate school.
Village-Scale Water
Fluid network analysis of a solar-powered reverse osmosis water desalination system
UROP at the MIT Global Engineering and Research (GEAR) Lab, undergraduate research position
September 2015 - May 2016
Overview: I worked on the Village-Scale Water Project as a research assistant with Natasha Wright, a PhD candidate in the MIT GEAR Lab. During my UROP, I collected data on the pressure drop across various valves, compared the cost of these potential components, and wrote a MATLAB script that performs a fluid network analysis of the current system. The goal of this project is to develop a cost-effective, low maintenance water desalination system that can be used in rural villages in India. In India, 73% of villages use groundwater as their source of drinking water, and brackish ground water with a salinity level greater than 480 g/L underlies 60% of the land area. It was found that for this level of salinity, electrodialysis reversal is a more cost-effective desalination method than reverse osmosis. Furthermore, the system was designed to be solar powered as many of the villages are off grid and the average annual solar irradiance received in India is 4-6 kWh/m^2.
Results: During the first semester of my UROP, I worked with another UROP student to build a testing set up for potential components. We collected data and plotted the pressure curves for three low-cost valves. I developed a MATLAB script that used this data and system parameters to analyze the system fluid network and output the percentage pressure drop of each component in the system. These results can be used to conduct a cost benefit analysis of the valves, filters, and other components.
Lessons & Skills: I practiced using MATLAB for data collection, data analysis, and fluid network analysis. This was the first time I could directly apply the concepts I learned in class, specifically 2.006 (Thermal-Fluids Engineering II), to my research. I also learned about the practical design process through meeting with my supervisor and discussing how she had gone about investigating the problem and iteratively designing a solution.
UROP at the MIT Global Engineering and Research (GEAR) Lab, undergraduate research position
September 2015 - May 2016
Overview: I worked on the Village-Scale Water Project as a research assistant with Natasha Wright, a PhD candidate in the MIT GEAR Lab. During my UROP, I collected data on the pressure drop across various valves, compared the cost of these potential components, and wrote a MATLAB script that performs a fluid network analysis of the current system. The goal of this project is to develop a cost-effective, low maintenance water desalination system that can be used in rural villages in India. In India, 73% of villages use groundwater as their source of drinking water, and brackish ground water with a salinity level greater than 480 g/L underlies 60% of the land area. It was found that for this level of salinity, electrodialysis reversal is a more cost-effective desalination method than reverse osmosis. Furthermore, the system was designed to be solar powered as many of the villages are off grid and the average annual solar irradiance received in India is 4-6 kWh/m^2.
Results: During the first semester of my UROP, I worked with another UROP student to build a testing set up for potential components. We collected data and plotted the pressure curves for three low-cost valves. I developed a MATLAB script that used this data and system parameters to analyze the system fluid network and output the percentage pressure drop of each component in the system. These results can be used to conduct a cost benefit analysis of the valves, filters, and other components.
Lessons & Skills: I practiced using MATLAB for data collection, data analysis, and fluid network analysis. This was the first time I could directly apply the concepts I learned in class, specifically 2.006 (Thermal-Fluids Engineering II), to my research. I also learned about the practical design process through meeting with my supervisor and discussing how she had gone about investigating the problem and iteratively designing a solution.
STEM for Students
Developing STEM workshops for students of all ages in the Cambridge-Boston Area and at the USA Science and Engineering Festival (USASEF) in Washington, D.C.
MIT Society of Women Engineers (SWE) Education Outreach, Off-campus Outreach co-chair
February 2014 - January 2016
Overview: MIT SWE is one of the most active student groups on campus. We are dedicated to connecting members with MIT faculty, local entrepreneurs, other students, and the national Society of Women Engineers. In addition, MIT SWE provides a number of exceptional STEM education outreach programs for local students and participates in larger STEM education events around country. I was one of the co-chairs for SWE Outreach for nearly two years, starting the second semester of my freshman year. My co-chairs and I planned educational STEM workshops for students of all ages in the Boston/Cambridge area. We organized special events on-campus, visited Mason Elementary School on a weekly basis to work with their after school program, and began a new weekly program at Joseph M. Browne Middle School. We also connected with several high school teachers in the area and held a high school engineering workshop called "Engineering Chats", to discuss engineering, MIT, and college life in general with high school students. During my freshman spring, a group of board members went to the USA Science and Engineering Festival (USASEF) to run an MIT SWE booth.
Lessons & Skills:
STEM education, especially for young girls and students in under served communities, is one of my priorities. The women of MIT SWE are some of the most caring, dedicated, and impressive people I have met at MIT, and I loved working with and learning from them. Working as an outreach chair I gained organization, lesson planning, and communication skills. I also learned how important teaching is to learning the material yourself, and I discovered that I love to teach. Regardless of my profession, I plan to make STEM education outreach part of my career.
MIT Society of Women Engineers (SWE) Education Outreach, Off-campus Outreach co-chair
February 2014 - January 2016
Overview: MIT SWE is one of the most active student groups on campus. We are dedicated to connecting members with MIT faculty, local entrepreneurs, other students, and the national Society of Women Engineers. In addition, MIT SWE provides a number of exceptional STEM education outreach programs for local students and participates in larger STEM education events around country. I was one of the co-chairs for SWE Outreach for nearly two years, starting the second semester of my freshman year. My co-chairs and I planned educational STEM workshops for students of all ages in the Boston/Cambridge area. We organized special events on-campus, visited Mason Elementary School on a weekly basis to work with their after school program, and began a new weekly program at Joseph M. Browne Middle School. We also connected with several high school teachers in the area and held a high school engineering workshop called "Engineering Chats", to discuss engineering, MIT, and college life in general with high school students. During my freshman spring, a group of board members went to the USA Science and Engineering Festival (USASEF) to run an MIT SWE booth.
Lessons & Skills:
STEM education, especially for young girls and students in under served communities, is one of my priorities. The women of MIT SWE are some of the most caring, dedicated, and impressive people I have met at MIT, and I loved working with and learning from them. Working as an outreach chair I gained organization, lesson planning, and communication skills. I also learned how important teaching is to learning the material yourself, and I discovered that I love to teach. Regardless of my profession, I plan to make STEM education outreach part of my career.
Dust in the Desert
Optimizing solar plant performance in the Atacama Desert
Enel Green Power, MISTI-Chile summer internship
June - August 2015
Overview: The goal of this project was to develop a theoretical model to predict the effect of soiling on production of solar plants in the Atacama desert. Over 10 weeks, I studied the current methods of measuring and analyzing soiling, identified and defined key environmental variables, developed a theoretical model of the effect of soiling based on the literature, and tested this model using data from an existing solar plant. After several weeks of literature investigation, my supervisor and I decided to structure the theoretical model in two steps in order to account for the numerous variables involved in the phenomenon of soiling. I first obtained dust deposition data from an external dust model and then used this data as the input for empirical equations from the literature that calculate the percentage irradiance loss due to dust accumulation. This structure allowed us to account for nearly all of the relevant variables.
Results: The end product of this internship was a theoretical soiling model, written in MATLAB, that predicts the percentage of irradiance lost due to PV soiling on an hourly resolution using regional, hourly dust deposition data and empirical relations between dust deposition density and transmittance loss. I successfully completed the first round of tests on the model, but several assumptions were made in the processes. First, it was assumed that the resolution and scale of the dust deposition data provided from all three sources are sufficient for our purposes. All three sources had a scale of about 100 km and their resolutions range from yearly to 3-hourly values. For future calculations, the best data set would be the most recent on the smallest scale and at an hourly or 3-hourly resolution. The second assumption was that the empirical equation, used to related dust deposition density and transmittance loss, accounts for the combined effect of angle of incidence, time of day, and dust deposition. This meant the effect of time of day, and more specifically the changing ratio of direct to diffuse irradiance, was not considered. In future trials, a simple factor based on the hour of the day may be sufficient to account for this ratio. Ideally, a more theoretical, optics-based equation will be used to calculate transmittance loss due to soiling. Several papers mentioned simulation techniques, such as ray tracing, to model this phenomenon and it may be worthwhile to do further research into the physics of dust particle light scattering to improve the second half of the theoretical model.
Lessons & Skills: I greatly improved my ability to code in MATLAB throughout this internship, and I conducted a literature search for the first time independently. I also learned a great deal about power plant operation and working in a company, as opposed to a university research lab.
Enel Green Power, MISTI-Chile summer internship
June - August 2015
Overview: The goal of this project was to develop a theoretical model to predict the effect of soiling on production of solar plants in the Atacama desert. Over 10 weeks, I studied the current methods of measuring and analyzing soiling, identified and defined key environmental variables, developed a theoretical model of the effect of soiling based on the literature, and tested this model using data from an existing solar plant. After several weeks of literature investigation, my supervisor and I decided to structure the theoretical model in two steps in order to account for the numerous variables involved in the phenomenon of soiling. I first obtained dust deposition data from an external dust model and then used this data as the input for empirical equations from the literature that calculate the percentage irradiance loss due to dust accumulation. This structure allowed us to account for nearly all of the relevant variables.
Results: The end product of this internship was a theoretical soiling model, written in MATLAB, that predicts the percentage of irradiance lost due to PV soiling on an hourly resolution using regional, hourly dust deposition data and empirical relations between dust deposition density and transmittance loss. I successfully completed the first round of tests on the model, but several assumptions were made in the processes. First, it was assumed that the resolution and scale of the dust deposition data provided from all three sources are sufficient for our purposes. All three sources had a scale of about 100 km and their resolutions range from yearly to 3-hourly values. For future calculations, the best data set would be the most recent on the smallest scale and at an hourly or 3-hourly resolution. The second assumption was that the empirical equation, used to related dust deposition density and transmittance loss, accounts for the combined effect of angle of incidence, time of day, and dust deposition. This meant the effect of time of day, and more specifically the changing ratio of direct to diffuse irradiance, was not considered. In future trials, a simple factor based on the hour of the day may be sufficient to account for this ratio. Ideally, a more theoretical, optics-based equation will be used to calculate transmittance loss due to soiling. Several papers mentioned simulation techniques, such as ray tracing, to model this phenomenon and it may be worthwhile to do further research into the physics of dust particle light scattering to improve the second half of the theoretical model.
Lessons & Skills: I greatly improved my ability to code in MATLAB throughout this internship, and I conducted a literature search for the first time independently. I also learned a great deal about power plant operation and working in a company, as opposed to a university research lab.