Coursework
Vehicle Performance Analysis in Python
A modular program that models wide open throttle performance across every gear, driven by real dynamometer data.
Goal
Build a modular Python program that analyzes vehicle performance at wide open throttle by modeling engine output, drivetrain behavior, road load, acceleration, and axle loads across every gear ratio, using real dynamometer data as the input.
The challenge
The hard part was connecting systems that are usually studied separately. Engine torque curves, transmission and final drive ratios, aerodynamic drag, rolling resistance, and axle load transfer all had to end up inside one working model where changing any input propagates correctly through the rest.
The second challenge was making it reliable rather than just correct once. The program has to import and validate spreadsheet data, fit real dyno measurements, catch engine speeds that fall outside the tested range, and organize repeated runs into outputs someone can actually read.
How it works
Input and validation
The program opens with a menu. You pick a vehicle, it loads the matching sheet out of the Excel workbook, and it validates the selection before continuing.
From there it pulls three separate blocks out of the sheet: the six gear ratios, the vehicle parameters, and the raw dynamometer points. Horsepower is computed for each dyno point as it is read.
The physics
The dyno points get fit with a polynomial, which turns a handful of measured torque readings into a continuous torque curve the model can evaluate at any engine speed. Road load comes from rolling resistance, grade, and aerodynamic drag combined. Traction comes from engine torque through the gear and final drive ratios divided by wheel radius. Acceleration falls out of the difference between the two.
Axle loads are handled separately, static first and then with the load transfer that happens under acceleration on a grade.
Output
Results are written out to multi sheet Excel files so runs can be compared side by side, and any engine speed that falls outside the range covered by the dyno data is flagged automatically rather than silently extrapolated.
Results
I ran the model on four vehicles with very different engines: a Corvette, a Honda Civic, a Mustang, and an Impala. For each one the program produces the fitted dyno curve and a set of performance sweeps across all gears.
Chevrolet Corvette
Honda Civic
Ford Mustang
Chevrolet Impala
Left in each pair is the fitted dynamometer data, with measured torque and horsepower plotted against engine speed and the polynomial fit drawn through it. Right is the sweep set: front axle load against wheel radius, acceleration against vehicle weight, and engine speed against road speed, with one curve per gear.
Putting the four side by side makes the differences obvious. The Corvette and the Mustang carry torque high into the rev range. The Civic makes far less absolute torque but needs much less of it to move the car, which shows up immediately in the acceleration against weight sweep.
Takeaways
This project pushed my experience in vehicle dynamics, numerical modeling, and data validation, but the part that stuck was the translation itself. Turning equations from a textbook into a program that takes messy real measurements, checks them, and produces trustworthy output is a different skill than solving the equations, and it is the one that actually gets used.