Graduate school

Coursework

USC

M.S. Aerospace Engineering

5

Graduate courses, 2025–2027

Combustion

AME 513a / 513b · Fall 2025 – Spring 2026 · University of Southern California

  • Python
  • Cantera
  • Flame structure
  • Chemical kinetics
  • Turbulent flow

Course description

In Combustion I spent the semester learning how flames work from the inside out, then applying that in practice with Cantera. I started with the thin structure of a freely propagating premixed flame, then added stretch and turbulence. I paired those simulations with classical analysis so I could go from a plot to a clear explanation of why a flame thickens or extinguishes.

Core competencies

  • Computed freely propagating C2H6\mathrm{C_2H_6}/air, C2H4\mathrm{C_2H_4}/air, and C2H2\mathrm{C_2H_2}/air flames across equivalence ratio and pressure, then extracted species and radical profiles, heat-release rate, key reaction rates, flame thickness versus reaction-zone thickness, SL0S_L^0, and m˙′′=ρuSL0\dot{m}''=\rho_u S_L^0
  • Ran sensitivity analysis to find the rate-controlling kinetic steps and mapped the conditions that increase NOx\mathrm{NO}_x and soot
  • Stretched twin premixed H2\mathrm{H_2}/air and C3H8\mathrm{C_3H_8}/air flames and non-premixed CH4/N2\mathrm{CH_4/N_2}–air flames in counterflow, tracked Tmax(K)T_{\mathrm{max}}(K), found KextK_{\mathrm{ext}}, and explained the trends with Lewis number and differential diffusion
  • Separated thermal versus prompt NOx\mathrm{NO}_x in the stretched non-premixed cases
  • Estimated turbulence Reynolds number, dissipation, Kolmogorov and Taylor scales, and Kolmogorov time for methane–air jet flames, then placed lab and practical setups on the Borghi–Peters regime diagram
Deep diveFlame structure, stretch, turbulenceOpen

Freely propagating · Cantera

Flame structure

I simulated freely propagating premixed flames using Cantera for ethane, ethylene, and acetylene in air, sweeping equivalence ratio and pressure. As shown below, the fuel and oxygen sit flat in the unburned gas, then drop through a thin consumption zone. Radicals and heat release peak within the consumption zone, and in my simulations I found that added pressure continuously compresses the consumption zone.

  • Acetylene is the fastest of the three. Ethane is the slowest. That ranking holds in SL0S_L^0 and in m˙′′=ρuSL0\dot{m}''=\rho_u S_L^0.
  • Ethane and ethylene peak near stoichiometric and fall off rich. Acetylene keeps climbing on the rich side.
  • Mass burning rate still rises with pressure even as flame speed falls, because unburned density wins.
  • Sensitivity is dominated by H+O2⇌OH+O\mathrm{H}+\mathrm{O_2}\rightleftharpoons\mathrm{OH}+\mathrm{O} and CO+OH⇌CO2+H\mathrm{CO}+\mathrm{OH}\rightleftharpoons\mathrm{CO_2}+\mathrm{H}. High pressure shifts weight toward termination.
  • Thermal NOx\mathrm{NO}_x tracks temperature and is strongest near stoichiometric. Soot wants rich, hot acetylene.
Species, radicals, heat release, and key reaction rates through a lean ethane–air flame
Laminar flame speed and mass burning rate versus equivalence ratio at 1, 10, and 20 atm
Normalized sensitivity of flame speed to the controlling kinetic steps
NOₓ mass fraction versus temperature across the three fuels
Soot precursor mass fraction versus temperature, peaking in rich acetylene

Twin counterflow

Stretch and Lewis number

I extended twin premixed flames in counterflow, followed by a methane–nitrogen versus air diffusion flame. Stretch represents the strain that opposing jets apply to the reaction zone. Its effect on peak temperature depends on which species is deficient and the rate of its diffusion. The form of TmaxT_{\mathrm{max}} versus stretch already indicates whether the Lewis number is aiding or hindering the process.

  • Lean hydrogen: fuel is deficient and it is the fast diffuser (LeF<1\mathrm{Le}_F<1). Preferential diffusion feeds H2\mathrm{H_2} into the reaction zone, so TmaxT_{\mathrm{max}} rises at first. Then stretch wins and the flame dies.
  • Rich hydrogen: oxygen is deficient and it is the slow diffuser. TmaxT_{\mathrm{max}} falls from the first step, but the flame holds on longer.
  • Propane flips the story. Lean propane is a heavy, slow fuel (LeF>1\mathrm{Le}_F>1), so stretch only hurts. Rich propane lets the faster oxygen help, and the flame is much more stretch-resistant.
  • Non-premixed methane lives where the fluxes meet stoichiometry, on the air side of stagnation. Diluting the fuel drops TmaxT_{\mathrm{max}} and kills stretch resistance. Full NOx\mathrm{NO}_x chemistry puts a peak at the flame. Thermal-only NOx\mathrm{NO}_x is lower and falls off faster as residence time shrinks.
NOₓ versus stretch for full chemistry versus thermal-only, undiluted methane
Lean hydrogen/air: peak temperature versus stretch, rise then fall to extinction
Rich hydrogen/air: peak temperature falling with stretch from the first step
Lean propane/air: peak temperature decreasing to extinction
Rich propane/air: peak temperature holding high over a much wider stretch range
Non-premixed methane versus air: peak temperature versus stretch
Non-premixed 50/50 methane/nitrogen versus air: lower peak temperature and weaker stretch resistance

Estimated, then placed on Borghi–Peters

Turbulent scales

I estimated the turbulence Reynolds number, dissipation, Kolmogorov and Taylor scales, and Kolmogorov time for methane–air jet flames at the nozzle and within the hot gas. Viscosity increases by more than an order of magnitude once the gas reaches flame temperature, so a jet appearing turbulent at the exit may lie near the laminar edge where heat release occurs. I then positioned several laboratory and practical premixed configurations on the Borghi–Peters diagram.

  • A small, fast jet loses most of its turbulent Reynolds number in the hot gas because u′u' drops and ν\nu rises.
  • A larger, slower jet still looks turbulent at the flame. Its Kolmogorov length is centimeters, not microns.
  • A grid-stabilized flame sits in wrinkled flamelets. A bluff-body duct sits in corrugated flamelets. A small high-speed jet chamber can cross Ka=1Ka=1 into thin reaction zones.
A · Grid flameB · Bluff-body ductC · Open cloudD · Jet chamber10⁰10¹10²10³10⁴10⁵10⁶0.010.1110100Lᵢ / δₗu′ / SₗWrinkledCorrugatedThin / brokenu′ = SₗKa = 1
Borghi–Peters regime diagram. Ka = 1 separates flamelets from thin reaction zones. u′ = Sₗ separates wrinkled from corrugated flamelets.

Compressible Gas Dynamics

AME 511 · Spring 2026 · University of Southern California

  • Shock relations
  • Nozzle theory
  • CFD

Course description

In Compressible Gas Dynamics, I applied conservation laws and shock/expansion relations to study wave propagation and discontinuities. Starting with continuum acoustics, I progressed through finite-amplitude waves and the method of characteristics to shocks, expansions, and nozzles in propulsion and high-speed flight. Combining calculations with classical analysis, I used jump conditions and xx–tt diagrams to explain shock steepening, smooth expansion fans, and Mach-number control in nozzles.

Core competencies

  • Derived the thermodynamic closures and conservation statements for a calorically perfect gas, then used them on a piston-driven isentropic compression and on isothermal versus isentropic discharge from a rigid tank
  • Computed Pitot-tube readings for subsonic and supersonic flight, derived the Hugoniot equation from the Rankine–Hugoniot relations, and found the maximum isentropic velocity from a reservoir
  • Solved the β\beta–θ\theta–MM relation for oblique shocks, wall reflections, opposite-family shock intersections, and a symmetric expansion–compression dip
  • Calculated rocket-nozzle exit Mach number, temperature, density, and velocity, and the design numbers for a supersonic wind tunnel
  • Tracked a moving normal shock in laboratory and shock-fixed frames, then reconstructed the full shock-tube wave system (incident shock, contact surface, expansion fan, reflected shock) including the density ratios η\eta and ζ\zeta
  • Applied linearized theory to a flat plate at angle of attack, compared lift, drag, and moment coefficients with exact shock-expansion theory, and mapped the relative error with angle of attack and Mach number
  • Ran a 3-D rotating-frame CFD study of a high-speed propeller in Earth, Mars, and Titan atmospheres, extracting relative Mach and pressure at 50%50\%, 75%75\%, and 100%100\% span and showing how a=γRTa=\sqrt{\gamma RT} and density send the same blade supersonic on Mars and subsonic on Titan
Deep diveHigh-speed propeller in Earth, Mars, and Titan atmospheresOpen

AME 511 · Group 1 · Spring 2026

High-speed propeller

I worked with a four-person team on a 3-D rotating-frame CFD study of a high-speed propeller. Local speed is V(r)=ωrV(r)=\omega r, so the hub can sit in weakly compressible flow while the tip goes through Mach 1. We ran the same geometry at 35,00035{,}000 RPM in Earth, Mars, and Titan atmospheres. The only things that changed were TT, pp, ρ\rho, γ\gamma, and RR.

35,000 RPM
Operating speed
50 / 75 / 100%
Spanwise slices
3 atmospheres
Earth, Mars, Titan
  • Orthogonal slices at 50%50\%, 75%75\%, and 100%100\% span are how I tracked Mach number and static pressure from hub to tip
  • NASA used the same rotating-frame approach for Ingenuity’s rotors in thin Martian air
Propeller geometry with orthogonal slices at 50%, 75%, and 100% span. Local speed is ωr.

Atmosphere properties

Earth, Mars, and Titan

Identical ω\omega means identical tip speed. Mach number is that tip speed over the local speed of sound, a=γRTa=\sqrt{\gamma RT}. I used each atmosphere’s TT, pp, ρ\rho, γ\gamma, and gas constant so the Mach field and the pressure loading could move independently.

  • Mars is cold CO2\mathrm{CO_2}, so aa is low and the blade goes more supersonic, but pp and ρ\rho are so low that the loading stays tiny
  • Titan is a dense, cold N2\mathrm{N_2} column. Mach stays subsonic and the pressure force is the largest of the three
  • Earth is the baseline: weakly supersonic at the tip, moderate loading
Earth

Earth

Air · sea level

TT
288 K
pp
101.3 kPa
ρ\rho
1.23 kg/m³
γ\gamma
1.4
MW
28.97

M_tip ≈ 1.03

Weakly supersonic tip. Baseline loading.

Mars

Mars

CO₂ · near-surface

TT
210 K
pp
0.60 kPa
ρ\rho
0.015 kg/m³
γ\gamma
1.3
MW
41.01

M ≈ 1.05–1.12

Highest Mach, weakest pressure loading.

Titan

Titan

Dense N₂ · near-surface

TT
94 K
pp
146 kPa
ρ\rho
5.3 kg/m³
γ\gamma
1.4
MW
28.01

M ≈ 0.82–0.91

Entire blade subsonic. Strongest loading.

Mesh and solver

Moving reference frame

The production setup is a two-cylinder moving-reference-frame mesh: an inner rotating zone around the blade and a stationary outer cylinder. About 491k elements, max skewness 0.8450.845. We ramped 2,5002{,}500 RPM at a time from rest to 35,00035{,}000, first-order, then finished second-order.

491k
Production mesh
1.9%
Mass-flux error, 2-cyl
2.85M
Refinement check
  • A single cylindrical domain with a rotating wall would not converge. Residuals jumped at every RPM step, mass flux diverged, and a negative-pressure pocket formed at the outlet
  • Interface mass-flux error on the two-cylinder mesh was about 1.9%1.9\%
  • A 2.852.85 M refined mesh showed the same Mach pattern. The student-license cap kept us on the coarse mesh for the atmosphere sweep
  • The gas is calorically perfect. The flow is inviscid, steady in the rotating frame, and isentropic except across shocks
Density-based residuals while RPM is ramped 20k to 35k. Spikes are the RPM steps.
Single cylindrical domain
Two-cylinder MRF mesh: inner rotating zone and outer stationary zone

Mach and pressure

Earth, hub to tip

On Earth I extracted relative Mach and static pressure at three spanwise stations.

  • Mid-span is high-subsonic. Peak relative Mach about 0.980.98, with mild pressure change across the blade
  • At 75%75\% span the flow is transonic, M≈1.00M\approx 1.00, and the pressure gradients get sharper
  • At the tip the relative flow is weakly supersonic, M≈1.03M\approx 1.03, and the pressure jump is the strongest of the three stations
  • Compressibility grows from hub to tip. Shocks show up first at the outermost station
Earth, 50% span, relative Mach, peak ≈ 0.98
Earth, 50% span, static pressure
Earth, 75% span, relative Mach, peak ≈ 1.00
Earth, 75% span, static pressure, stronger gradients
Earth, 100% span, relative Mach, peak ≈ 1.03
Earth, 100% span, static pressure, strongest jump

3-D field

Pathlines and tip vortices

The spanwise slices are 2-D cuts of a 3-D flow. Relative-velocity pathlines show the tip vortex and the radial pumping you expect from a rotating blade.

  • Treating the propeller as an inviscid wall forces the flow to run parallel to the surface, so the solver over-speeds around curvature and the suction peak on the convex face drops out
  • That was an acceptable cost to isolate shock and Mach-number trends
  • Centrifugal three-dimensionality and the trailing vortex mean a given slice is never a pure ωr\omega r problem
Relative-velocity pathlines off the propeller, with a tip vortex
Relative velocity in a spanwise plane, showing the rotating-frame field

Earth, Mars, Titan

Results across atmospheres

I ran the same blade and the same rotation rate in all three atmospheres so the only change was the speed of sound and the density.

  • Earth: MM walks from 0.980.98 at mid-span to 1.031.03 at the tip
  • Mars: colder, heavier CO2\mathrm{CO_2} puts every slice higher, about 1.051.05 to 1.121.12. Highest Mach, weakest pressure loading
  • Titan: entirely subsonic, M≈0.82M\approx 0.82–0.910.91, and still produces the largest pressure force because the atmosphere is dense
  • Atmosphere changes the Mach field through a=γRTa=\sqrt{\gamma RT} and the loading through ρ\rho and pp
  • A calorically perfect, inviscid gas is enough to see that split. Real CO2\mathrm{CO_2} and N2\mathrm{N_2} thermodynamics, and viscosity, would likely stretch the differences further
  • Pick tip speed for the local speed of sound, then check that the loading you need still exists in that atmosphere

Entry and Landing Systems for Planetary Exploration

ASTE 577 · Fall 2026 · University of Southern California

  • MATLAB
  • Decelerators
  • Flight computer
  • Systems engineering

Course description

In the Entry and Landing Systems class, I learned about the process a vehicle undergoes from approach to landing on a safe surface. I applied this knowledge by using MATLAB and Excel for sizing and a Raspberry Pi as the flight computer. The course began with an exploration of historical EDL missions and the causes of TPS failures. I then delved into atmospheric flight mechanics, aerodynamic decelerators, and landing attenuation, analyzing case studies from Mars, Venus, and Titan. Additionally, I led a group of six people in designing a mission, where we transitioned from creating heat-load or timeline plots to providing clear explanations of why specific landing methods, such as heat shields, parachutes, or propulsive landings, were appropriate for each celestial body.

Core competencies

  • Walked heritage EDL architectures and mapped physical versus organizational failure onto entry-system requirements
  • Sized the stack across aerothermodynamics, decelerators, landing attenuation, and telecom / C&DH, not just the heat shield
  • Built an EDL sequence with timings, comms, and fault response on a Raspberry Pi (temperature, accel, landing trigger, transmit)
  • Wrote Level 1 requirements and a science traceability matrix for a unique surface measurement, then closed the architecture at the end of the semester
  • Coordinated with Team Discovery to vet a landing system, from research briefs through the term project
Deep diveIn developmentOpen

Human Spaceflight

ASTE 524 · Fall 2026 · University of Southern California

  • ConOps
  • Design Reference Mission
  • Human rating
  • Life support / safe haven
  • Crew vehicles

Course description

In Human Spaceflight, I gained a comprehensive understanding of the engineering, technologies, and systems essential for human spaceflight. I applied this knowledge to lead a group project focused on designing a habitat and crew transfer to Mars. I started by exploring space environments and crew accommodations. Then, I expanded to include life support systems, launch and entry procedures, rendezvous and docking operations, and extra-vehicular activities. By integrating these systems with mission operations and safety protocols, I successfully transitioned from establishing a Need/Goals/Objectives statement and a Design Reference Mission to developing a Concept of Operations that could be implemented by a viable transportation architecture.

Core competencies

  • Wrote Need, Goals, and Objectives for a human Mars program that stay short at the top and measurable in the objectives, without locking a vehicle solution early
  • Turned the Ares Program Design Reference Mission into capacity, cadence, transit-time, landing, and Earth-return criteria a Crew and Cargo Transportation System had to meet
  • Mapped NASA Procedural Requirements 8705.2C human-rating tenets (accommodate, utilize, protect) into abort windows, failure tolerance, onboard autonomy under light-time delay, safe haven, and solar particle event shelter requirements
  • Built a phase-by-phase Concept of Operations (cargo ahead of crew, Earth parking-orbit checkout, Trans-Mars Injection, cruise, Mars capture, landing within 3 km of the base, surface transfer, ascent, and CONUS return) with labeled contingencies from pad escape through abort land landing
  • Kept the design as one crewed transportation system: habitat stayed out of scope, architecture choices deferred to ConOps rather than the Need
Deep diveIn developmentOpen

Spacecraft System Design

ASTE 520 · Fall 2026 · University of Southern California

  • Requirements
  • ConOps
  • Mass & power

Course description

In Spacecraft System Design, I developed and applied the skills needed to create a complete lunar communications architecture from mission concept through first-order subsystem design to a closed vehicle. I translated mission needs into clear requirements and a concept of operations, then progressed through payload definition, orbit selection, propulsion, thermal and power management, structural design, avionics, and attitude control. I evaluated cost, risk, and design feasibility against the mission objectives, ensuring the proposed architecture met the original RFP. I also performed system-level trades to understand how design decisions—such as mass budgets and Δv requirements—drove changes across subsystems and validated the overall vehicle solution.

Core competencies

  • Translated an RFP into testable mission requirements and a ConOps with phases, modes, and a first-cut verification path
  • Built and kept living first-order mass, power, Δv, data, and pointing budgets with named margins
  • Sized the major bus subsystems (payload, propulsion, thermal, EPS, structures and mechanisms, C&DH / telecom / ground, ADCS) and tracked the interfaces between them
  • Ran architecture trades on orbit, propulsion type, configuration, and pointing, and revised earlier choices when later analysis said so
  • Closed a conceptual design against cost, risk, I&T / V&V, and RFP compliance, and said whether it was credible enough to proceed
Deep diveIn developmentOpen