Bombing Your Own Engines: How NASA Solved the F-1 Combustion Instability Problem

Between 1962 and 1965, the most powerful rocket engine ever built kept trying to destroy itself. The Rocketdyne F-1, the engine that would power the first stage of the Saturn V, suffered from a problem that engineers could observe but not fully understand: combustion instability. And the solution they eventually found was one of the most unhinged approaches in engineering history.

They started bombing their own engines.

The Problem Nobody Could Model

Combustion instability is exactly what it sounds like: the combustion process inside a rocket engine becomes unstable. But the reality is far more violent than the name suggests.

Inside the F-1’s combustion chamber, liquid oxygen and RP-1 kerosene were being burned at a rate of nearly three tons per second. The chamber pressure was around 1,000 psi. Under these conditions, the combustion process could spontaneously develop oscillation patterns—pressure waves bouncing back and forth across the combustion chamber at thousands of cycles per second.

These oscillations weren’t gentle vibrations. They were acoustic resonance modes of the combustion chamber, and they could amplify themselves. A small perturbation in the combustion process would create a pressure wave. That wave would reflect off the chamber walls and interact with the incoming propellant flow, creating a feedback loop. Within milliseconds, the oscillations could grow large enough to burn through the chamber walls or blow the injector apart.

Rocketdyne engineers watched multiple F-1 injectors destroy themselves during test firings at Edwards Air Force Base. The engine would run normally for a few seconds, then pressure readings would spike chaotically, and the test would end with the engine in pieces.

The fundamental problem was that nobody could mathematically model what was happening. Combustion instability involves the intersection of fluid dynamics, thermodynamics, acoustics, and chemical kinetics—all operating at extreme temperatures and pressures with turbulent mixing of two different propellants. Even today, with supercomputers, this remains one of the hardest problems in aerospace engineering. In 1962, with slide rules and room-sized mainframes, it was essentially unsolvable from first principles.

The Injector: 6,300 Holes of Chaos

The F-1’s injector plate was a copper disc about 40 inches in diameter, drilled with approximately 6,300 holes arranged in a specific pattern. Fuel and oxidizer were injected through alternating rings of holes, creating a spray pattern designed to promote mixing and even combustion.

The injector was where instability either started or could be controlled. The pattern, size, and arrangement of the injection holes determined how propellants mixed, how combustion began, and critically, how the combustion process responded to perturbations.

Rocketdyne engineer Jerry Thomson led the effort to redesign the injector. His team went through injector design after injector design, testing each one on full-scale engines. Each redesign was essentially an educated guess—change the hole pattern, change the hole sizes, add baffles, modify the spacing—then fire the engine and see what happened.

Over the course of the program, Rocketdyne tested more than 40 different injector designs. The development consumed over 2,000 full-scale engine tests. Each test that ended in instability meant a destroyed injector and sometimes a destroyed engine. The cost was staggering, but the alternative was a Saturn V that couldn’t fly.

Bomb-in-the-Chamber Testing

Here’s where it gets interesting.

Early in the development, engineers realized they had a fundamental testing problem. Combustion instability was intermittent. An engine might run perfectly for dozens of tests and then suddenly go unstable. You couldn’t just fire the engine and wait for it to fail—that was too slow, too expensive, and didn’t give you useful data about how close to the edge you were.

The solution: deliberately trigger instability and see if the engine could recover.

Engineers developed small explosive charges—literally bombs—that could be detonated inside the combustion chamber while the engine was running. These charges were mounted on the injector face and were designed to create a sudden, violent pressure perturbation: exactly the kind of disturbance that could trigger combustion instability.

The test protocol was straightforward in concept and terrifying in practice:

Start the F-1 engine and bring it to full thrust
Detonate a bomb inside the combustion chamber
Observe whether the resulting pressure disturbance grows (unstable) or damps out (stable)
If the engine survives, you have a good design. If it doesn’t, go back to step zero with a new injector.

The bombs were small—just enough to create a pressure spike that mimicked a natural instability trigger. But they were detonated inside a chamber already running at 1,000 psi with 15 tons of thrust being produced. The fact that this was considered a reasonable test methodology tells you everything about the desperation of the engineering situation.

The criterion for success was that any bomb-induced instability had to damp out within 400 milliseconds. If the pressure oscillations died away within that window, the design was considered dynamically stable. If they grew or persisted longer, the design failed.

Baffles: The Solution That Worked Without Being Understood

The breakthrough came with the addition of baffles to the injector face. These were copper fins that protruded from the injector plate into the combustion chamber, dividing the chamber face into compartments.

The baffles worked by disrupting the acoustic modes that drove instability. The pressure waves that bounced across the chamber face were interrupted by the physical barriers. Instead of building into a coherent, growing oscillation, the waves were broken up into smaller, incoherent disturbances that couldn’t sustain themselves.

The final F-1 injector design featured a hub-and-spoke baffle pattern with radial fins extending from a central hub. This configuration was arrived at through exhaustive trial and error—not through theoretical prediction.

This is the part that bothers engineers to this day. The baffled injector design that flew on the Saturn V was empirically validated but not theoretically understood in complete detail. The team knew it worked because they had detonated bombs inside hundreds of engines and watched the oscillations die. But they couldn’t write down the equations that fully explained why one baffle pattern worked and another didn’t.

Paul Castenholz, Rocketdyne’s F-1 program manager, later described the approach as “cut and try”—an unusually blunt admission for a field that prides itself on analytical rigor. They had a problem they couldn’t model, so they tested their way to a solution.

The Numbers

The F-1 development program consumed extraordinary resources:

Over 2,000 full-scale engine tests during development
More than 40 injector designs tested
65 bomb tests on the final configuration to validate stability
Total development time: approximately 7 years from contract to flight qualification
Total development cost: roughly $500 million in 1960s dollars (over $4 billion today)

The final flight-rated F-1 engine produced 1,522,000 pounds of thrust. Five of them powered the Saturn V’s first stage, generating a combined 7.6 million pounds of thrust at liftoff. It remains the most powerful single-chamber liquid-fueled rocket engine ever flown.

Why This Matters Beyond Apollo

The F-1 combustion instability story is uncomfortable for engineers because it violates the narrative we like to tell ourselves: that engineering is the application of understood science to practical problems. Sometimes it isn’t. Sometimes the science isn’t understood well enough, and you have to resort to systematic empiricism—sophisticated trial and error.

When NASA and SpaceX developed the Merlin engine, and when Blue Origin developed the BE-4, combustion instability was still a concern. Modern computational fluid dynamics can model some aspects of the problem better than slide rules could, but the phenomenon is still not fully predictable from first principles. Engine developers still rely heavily on testing.

The F-1 team’s willingness to admit they couldn’t solve the problem analytically—and their pivot to bombing their own engines as a validation methodology—represents a kind of engineering courage. They didn’t pretend to understand more than they did. They found a way to verify their solutions empirically when theory failed them.

Sixty years later, the baffled injector design they arrived at through cut-and-try remains a standard approach in rocket engine design. The specific baffle configuration on the F-1 was reverse-engineered by NASA’s Marshall Space Flight Center in 2013 when they 3D-scanned original hardware. Even with modern analysis tools, the engineers who studied it found it remarkably well-optimized.

They may not have known exactly why it worked. But it worked.