2025-06-05

Introduction

  • SpaceX’s Starship Super Heavy booster uses 33 Raptor engines.
  • These engines burn massive amounts of fuel per second.
  • We’ll explore how the variation in fuel consumption across launches can be modeled with a normal distribution.
  • The Starship is the focus here.

Simulating Starship Fuel Flow

set.seed(42)
fuel_starship <- rnorm(100, mean = 3300, sd = 150)
head(fuel_starship)
summary(fuel_starship)
## [1] 3505.644 3215.295 3354.469 3394.929 3360.640 3284.081
##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##    2851    3207    3313    3305    3399    3643

Here we’re randomly simulating 100 fuel flow rates for the Starship rocket, and here we show 6 of the results, along with a quick summary.

ggplot2 Histogram

This histogram shows the simulated fuel flow rates for Starship across 100 launches. The data is centered around an average assumed fuel burn rate of 3300 kg/s. The shape is roughly bell-shaped, like a normal distribution.

ggplot2 Boxplot

Boxplot shows distribution of simulated Starship fuel flow rates. Median fuel use is about 3300 kg/s, with most values between ~3200 and ~3400 kg/s. Outliers appear below 3100 kg/s, possibly under-thrust.

Interactive Plotly Histogram

Similar to the previous ggplot2 Histogram. Shape again looks normal, which reinforces our model.

Mathematical Interpretation

  • The mean \(\mu = 3300\) kg/s reflects average fuel burn rate.
  • The standard deviation \(\sigma = 150\) captures variability.

\[ P(3150 < X < 3450) \approx 68\% \text{ of launches} \]

Calculating a Probability with the Normal Model

Suppose we want to know the probability that Starship fuel flow is between
3200 kg/s and 3400 kg/s on a launch.

Assuming a normal distribution: \[ X \sim N(3300, 150^2) \] We can standardize:

\[ \begin{align*} P(3200 < X < 3400) &= P\left( \frac{3200 - 3300}{150} < Z < \frac{3400 - 3300}{150} \right) \\ &= P(-0.67 < Z < 0.67) \end{align*} \]

Conclusion

  • Starship’s fuel flow is variable but centered; it varies due to many small influences.
  • A normal model allows engineers to predict it; because despite the variation, it follows a predictable normal distribution.
  • This statistical model can help the rocket engineers estimate performance and design safer launches.