Rylan
Runske

The Competition

• Competing in the ASNE/ONR Promoting Electric Propulsion (PEP) competition's Uncrewed Open division — an electric boat racing and design competition with a ≤55.5V battery limit, a required kill switch, and a 60 lb payload requirement.

The Platform

• Team is working with an existing Systemware-branded planing hull and electric drivetrain (Tenshock TS-5380-SLT water-cooled inrunner motor, Seaking 300A HV ESC) rather than building a hull from scratch, with the goal of a competition-ready boat plus documentation suitable for a white paper submission.

Current Focus — Failure Diagnosis

• Leading root-cause diagnosis on a motor that smoked during last season's race while the ESC survived — pointing toward thermal overload rather than an electrical fault, with prop mismatch and a possible cooling-loop failure as the leading candidate causes.
• Directing the team through a full documentation pass on the electrical system (battery specs, fuse ratings, ESC settings, prop/gearing) to close out unknowns before committing to repairs or a motor replacement.

Also Underway

• Redesigning the steering setup and moving to adjustable, actuator-driven trim tabs to replace the current fixed tabs and underperforming rudder.

Step 1 — Modeling

• Derived nonlinear equations of motion for the two-DOF system (arm angle φ, pendulum angle θ) via Euler-Lagrange, then linearized about the unstable upright equilibrium.
• Identified friction and effective arm inertia empirically against the plant's own p-code, correcting the textbook rod-inertia estimate by 2×.
• Validated the linear model against real plant response: 7.7° RMS pendulum error under a 1V step input.

Step 2 — Balance Controller

• Designed a state-feedback regulator via pole placement to stabilize the pendulum upright, targeting settle time < 0.5s, overshoot < 5%, and peak voltage < 10V.
• Linear model met spec: 0.43s settle time, 2.46% overshoot, 5.15V peak.

Step 3 — Real-Hardware Retuning

• The linear-model gains oscillated badly on the real plant (measured overshoot ~10× the prediction), so poles were re-placed with more conservative damping.
• Final gains passed all specs on hardware: 0.442s settle time, 0% overshoot, 3.03V peak.

Step 4 — Cascade Tracking + Swing-Up

• Built a two-loop cascade controller (fast inner loop balances pendulum tilt, slow outer loop commands that tilt to move the arm) to track a nonzero arm-position reference — achieved 1° steady-state arm error tracking to 30°.
• Implemented a hysteretic bang-bang swing-up controller to pump energy into the pendulum from hanging-down, with automatic handoff to the cascade balance controller once near upright (handoff at 3.86s in testing).

Base Plant

• Modeled the reference plant as a single-reheat Rankine cycle (2400 psi boiler, 800 psi reheat) using enthalpy states from steam tables, establishing baseline efficiency (33.63%), mass flow rate, and heat rejected for comparison.

Design 1 — Lower Reheat Pressure

• Reduced reheat pressure to 600 psi (25% of boiler pressure, per literature guidance) to isolate its effect on cycle performance.
• Result: efficiency +0.213%, essentially free to implement, but only a marginal efficiency gain.

Design 2 — Open Feedwater Heater

• Added an open feedwater heater at 200 psi to preheat feedwater before the boiler, re-deriving the cycle's efficiency and energy-balance equations for the added extraction/mixing stages.
• Result: efficiency +2.65% (to 34.52%), mass flow rate stayed within the 15% piping constraint, and heat rejected to the river dropped 3.9% — but required ~$225,000 in new equipment and labor versus ~$300 for the reheat-pressure change.

Cost, Emissions & Decision Matrix

• Converted efficiency gains to coal savings and daily fuel cost, then to CO₂ emissions, for both design options.
• Weighted a decision matrix (cost 40%, efficiency 40%, environmental effect 20%) across all three configurations; the open feedwater heater scored highest overall (3.2 vs. 3.0 and 2.6) despite its higher upfront cost, on the strength of its efficiency and emissions gains.

Gyroscope & Accelerometer — ADIS16470 IMU

• Specified the triaxial MEMS gyroscope (Coriolis-effect sensing, ±2000°/sec range, 8°/hr bias stability) and accelerometer (±40g range, 13 µg bias stability) for attitude and velocity/position estimation via integration.
• Addressed vibration rectification error and mounting near the vehicle's center of gravity to minimize drift.

Radar Altimeter — Garmin GRA 5500

• Specified a frequency-modulated continuous-wave radar altimeter for terrain-relative altitude, immune to the atmospheric-pressure errors of a barometric altimeter.
• Planned digital interfacing with the flight computer for altitude-hold corrective pitch commands, and filtering for low-altitude signal-to-noise degradation.

Doppler Radar Velocity Sensor — OmniPreSense OPS243-A

• Specified a 24 GHz continuous-wave Doppler sensor for direct, non-contact ground-speed measurement, derived the Doppler-shift-to-velocity relationship, and accounted for the mounting angle correction needed against the true velocity vector.

System Integration

• Defined the flight computer as the central hub, communicating with the IMU over SPI, the Doppler sensor over UART, and receiving digital altitude data from the radar altimeter, running navigation/control at a 50–200 Hz update rate.

Requirements & Concept Generation

• Set engineering requirements/limits from the challenge rules — 20 steel balls in continuous circulation, ≤5-minute reset on a single mechanical input, size and weight limits, and no electronics, adhesives, bands, or threading allowed in the build.
• Used a morphological matrix to break the machine into core functions (dose ball, launch ball, interact with obstacles, update score, capture balls, reset) and generated independent playfield concepts, one per team member, evaluated against a shared selection chart.

Design Selection

• Narrowed five candidate playfields to three finalists using a pass/fail selection chart on realism, cost, and safety, then scored the finalists in a weighted decision matrix (launcher 30%, static obstacles 20%, dynamic obstacles 20%, jackpot 10%, scoring 20%).
• Selected the highest-scoring concept — a peg-and-roller playfield with a jackpot button — for its strong, repeatable launcher and dense static-obstacle field.

Reliability — FMEA

• Ran a failure modes and effects analysis on the launcher and structure, identifying gear shearing as the highest-severity failure mode and designing the gear teeth and print orientation specifically to keep its risk priority number low.
• Addressed lower-severity risks (spring wear, launcher jams, bolt shearing) with tighter tolerances and printed/mechanical safeguards rather than added parts.

Fabrication

• Led final assembly and authored the SolidWorks 2D drawing package for all components, defining tolerances and fastener callouts to support fabrication and quality checks.
• Built the frame from MDF (laser-cut and band-sawed, drill-pressed and countersunk for clean fastening), the playfield obstacles from 3D-printed PLA, and a small aluminum counterweight (waterjet-cut) to correct front-heavy balance — assembled in layers with M3 hardware throughout.
• Used transparent acrylic for the front and rear playfield covers, chosen for visibility into the game and clean laser-cut tolerances.