How to make scramjets to work

Recently, as a passion project, I've been playing around with one of aerospace engineering's most fun puzzles: how to make scramjets work. And I think I found something - though not the thing I set out to find. But before we pop the champagne, let's explore what scramjets are, why they're awesome, and why they did not go mainstream for decades.

What Are Scramjets?

Scramjets (supersonic combustion ramjets) are air-breathing engines built for hypersonic speeds, taking over somewhere around Mach 5. Unlike conventional jet engines that rely on compressors, scramjets do not have any moving parts, and incoming air is compressed simply by the geometry of the inlet. They are super simple devices capable of moving vehicles of the air (opens in a new tab) at absurd speeds.

Scramjets are awesome because they can unlock things like:

• Rapid global travel: think SF to Tokyo within 1.5 hours vs. 11 hours.
• Cost-effective space access.
• Real-world Darkstar.

And so on.

Why Scramjets Aren't Everywhere

The big boss problem in this story is the flameholding problem - an issue best described as trying to keep a match lit in a hurricane (a line that has been floating around the hypersonics community for decades; I couldn't pin down the original author).

Here's the basic description of the problem: the time between air entering through the inlet and exiting through the nozzle (known as residence time) is measured in milliseconds, and within that timeframe, we need to:

a) Inject the fuel. b) Mix it with the air. c) Ignite and combust it. d) Expand the hot gas to generate thrust. e) Keep the flow supersonic and attached the whole way through - no unstart, no blowout.

Pure madness. We'd need God himself to carefully mix the air and fuel atoms together in the allotted time.

In addition, there are challenges with heat management and cooling, materials, and testing limitations. All of these are challenging yet ultimately solvable. The big boss problem here is flameholding. We need a solution that avoids the usual pitfalls (shock-induced quenching, sensitivity to flow uniformity, etc.) - a solution that could be the key to unlocking hypersonic flight.

So What Do We Do?

Being a good engineer is less about solving problems at face value and more about asking the right questions that stem from fundamental truths rooted in physics (yes, that overused first-principle thinking).

Reframing the Problem

Instead of asking, "How do we solve the flameholding problem?" we can ask, "How else can we add enthalpy to compressed air?" After all, the exhaust velocity in a scramjet is determined by the energy added to the airflow, and nowhere is it written that energy must come from combustion of carbon-based fuels.

Don't get me wrong, combustibles are cool. I also read the Ignition! book by John D. Clark and have a soft spot for the decorated community of, as Isaac Asimov eloquently put it:

Now it is clear that anyone working with rocket fuels is outstandingly mad. I don't mean garden-variety crazy or a merely raving lunatic. I mean a record-shattering exponent of far-out insanity.

But we need to get the job done, so let's put combustibles aside for now and see how far we get.

So, What If We Bypass Combustion?

The idea is very simple: instead of injecting fuel into a combustion chamber, place a heat exchanger (say, an array of tungsten rods) into the airflow path where the combustion chamber used to be. Tungsten melts at about 3422 °C (roughly 3700 K), so it can tolerate some of the extreme temperatures we'd encounter at hypersonic speeds.

That raises the question: where does the energy come from to heat those rods, and in turn, heat the air? Two candidates:

1. Li-ion batteries.
2. Harnessing an "external" energy source (more on this later, including why the quotes are doing a lot of work).

A Quick Thermodynamic Refresher

Before diving into energy sources, we need a sense of how much enthalpy must be added to the airflow. This is the benchmark for everything that follows.

Let's pick:

• Initial speed: Mach 6
• Target speed: Mach 8
• Altitude: ~30 km (where the ambient temperature $T_{\infty} \approx 226.5\,\text{K}$)

Using standard compressible-flow formulas, the total (stagnation) temperature at Mach $M$ is:

where $\gamma \approx 1.4$ for air.

At Mach 6:

At Mach 8:

So to go from Mach 6 to Mach 8 at the same altitude, the flow's total temperature must rise by about 1272 K. With $c_p \approx 1100\,\text{J/(kg.K)}$ for air at these temperatures:

For each kilogram of air flowing through our "heater," we must supply about 1.4 MJ. At a modest mass flow of 10 kg/s, that's 14 MW, continuously. Supplying 1.4 MW is like lifting 140 one-ton trucks 1 m off the ground every second; we need ten times that.

Approach 1: Li-ion Batteries. Wall Number One.

Li-ion batteries store around 0.7 to 1.0 MJ per kg of battery mass (about 200 to 300 Wh/kg). Running 14 MW means eating through roughly 15 to 20 kg of battery per second of flight. A minute of acceleration costs you a tonne of dead battery mass. You could carry more batteries, of course, but then your vehicle mass skyrockets and the whole exercise eats itself.

That's no good - we hit the wall. The good news is that's what engineering is about: trying ideas, hitting dead ends, and learning from them. Pure electric heating from an onboard battery is not the path. Next idea.

Approach 2: Harvesting Aerodynamic Heating. Wall Number Two.

Remember the videos of large objects falling from the sky - rockets and shuttles - wrapped in that glowing plasma that looks pretty hot? Could we use that heat instead of just surviving it? Rocket engines have long used regenerative cooling, where fuel circulates through channels around the nozzle to absorb heat and prevent meltdown. So flip it: embed a closed-loop working fluid (say, helium) under the hottest skin - leading edges, nose tip - and route the captured heat into the flowpath.

The tempting way to see this is as a self-reinforcing cycle: the faster you go, the hotter the skin, the more energy you can feed back into the flow. I sat with that framing for a while, because it feels great. Two things are wrong with it, and they're worth taking in order, because each one teaches something different.

Wall 2a: The Books Don't Balance

Where did I think the energy was coming from? An "external source." It isn't external. Aerodynamic heating is drag doing its work: the air takes kinetic energy from the vehicle and hands part of it back as heat. Capturing a slice of that and returning it to the flow is recovering your own losses - useful, but it cannot be a net source of propulsive energy, any more than a car can accelerate by harvesting its own brake heat. There is no loop to ride upward. The books have to balance.

Loss recovery is still worth something - we'll get to exactly what. But first, the arithmetic wall: how much heat is actually on offer?

Wall 2b: How Much Heat Is Actually There?

Recovery only matters if there's meaningful heat to recover, so let's put real numbers on the leading edges instead of vibes.

The standard order-of-magnitude tool here is the Sutton-Graves correlation for stagnation-point heating:

with freestream density $\rho_{\infty}$ in kg/m^3, leading-edge radius $R_n$ in m, and velocity $V_{\infty}$ in m/s. At 30 km, $\rho_{\infty} \approx 0.0184\,\text{kg/m}^3$ and the speed of sound is about 302 m/s. For a sharp 5 mm leading edge:

The harvest column assumes 4 m of total leading edge, a 5 cm high-flux strip (stagnation heating falls off fast as you move around the edge), an average flux of half the stagnation value, and 60% capture. A blunter 20 mm edge halves the flux. Hot acreage on the inlet ramp and underside can add a few hundred kW more at Mach 6+.

For calibration against real hardware: the Shuttle's nose cap saw on the order of 0.5 MW/m^2 during reentry. Apollo lunar returns peaked around 5 MW/m^2. Fluxes in the tens of MW/m^2 belong to sub-millimeter leading edges at Mach 10 (X-43 territory), and the true triple-digit club is the Galileo probe hitting Jupiter's atmosphere at 47 km/s. If you've seen much scarier numbers quoted, they belong to reentry at orbital speeds and above - it's easy to borrow figures from the wrong regime here, since $V^3$ makes the two worlds differ by orders of magnitude.

So the honest harvest at Mach 6 to 8 is roughly 0.1 to 0.5 MW against a 14 MW requirement - short by a factor of 30 to 100. Combined with Wall 2a, the verdict is clean: harvested aerodynamic heating cannot replace combustion. Not thermodynamically, and not even arithmetically.

The Door

Here's what the failed numbers quietly revealed. The harvest is about 30 times too small to replace combustion - and about 10 times bigger than what ignition needs. Wrong size for an engine. Exactly the right size for a match.

So combustion comes back, and it should: at 10 kg/s of air, a near-stoichiometric 0.68 kg/s of kerosene-class fuel carries about 29 MW of chemical power - double the 14 MW we need - at 43 MJ/kg, with the oxidizer taken free from the sky. No battery or heat loop competes with that. The real question was never how to replace combustion. It was how to make combustion light and stay lit in a millisecond-residence-time hurricane. And for that job, a few hundred kilowatts of free heat, delivered at 1600+ K, is a gift.

The Pilot Ring, or: What the SR-71 Already Knew

The J58 engines on the SR-71 burned JP-7, a fuel so deliberately hard to ignite that a spark plug couldn't light it. Every start and afterburner light used a shot of triethylborane (TEB), a chemical that ignites on contact with air - and each engine carried enough for only 16 shots, which actually limited mission profiles. But there was a backup: six catalytic igniters inserted into three rings of the afterburner flameholders. If TEB ran out, the crew could over-trim the engine, superheat those igniters in the exhaust, and get a harsh but functional light off a glowing surface. A hot ring held ignition authority on the fastest air-breathing aircraft ever flown.

That's the door. Take the tungsten out of the middle of the flowpath (where it was pretending to be an engine) and shrink it into a pilot ring in the combustor - a permanently glowing surface held at 1300 to 1800 K, providing continuous ignition and a standing pool of hot radicals for flame anchoring. The energy budget for this is tiny: a ring of ~0.02 m^2 losing heat to the flow at combustor conditions needs on the order of 20 kW to stay hot. At Mach 6 the leading edges are offering 120+ kW. Covered six times over.

Better still, the temperature ladder means no pumps are required. At Mach 6, the leading-edge recovery temperature is about 1610 K; the ring wants 1300 to 1800 K; the air entering the combustor sits around 825 K (static, at combustor Mach ~2.5). Heat flows downhill the entire way. High-temperature heat pipes - sodium works across roughly 1100 to 2000 K - can move it passively from edge to ring: no helium loop, no pump power, no moving parts, which is very much in the spirit of the scramjet itself. And unlike TEB, the edges never run out of shots.

Where the Surplus Goes: Into the Fuel

The remaining captured heat routes into the fuel, not the air - and this is where the "loss recovery" from Wall 2a earns its keep. Endothermic fuels like JP-7's descendants can absorb roughly 2.5 MJ/kg through preheating and catalytic cracking; at 0.68 kg/s that's a 1.7 MW heat sink, comfortably swallowing everything the edges can deliver. Three things happen at once:

1. The recovered heat re-enters the cycle through the fuel - a genuine regenerative gain of a few percent of chemical power. Small, real, and honest this time.
2. Hot, cracked fuel ignites dramatically faster than cold liquid spray. Ignition delay is the enemy in a milliseconds-long combustor, and preheating attacks it directly.
3. The leading edges get actively cooled as a side effect, which lets them be sharper - and sharper edges mean lower wave drag.

The Elegant Part

Look at that table again. At Mach 3 the edges sit near 570 K - useless for ignition, and it doesn't matter, because at Mach 3 you're a ramjet with subsonic combustion and ordinary flameholders (getting to Mach 4 to 5 is the boost propulsion system's job - turbojets and ramjet mode, a story for another post). The concept switches on around Mach 5, as flux and edge temperature climb with $V^3$ - which is precisely the regime where flameholding becomes the big boss problem. The resource scales with the problem. You could not order that from a catalog.

Grain of Salt

Back-of-the-napkin says the architecture closes. Turning it into hardware is the actual work, and every assumption above needs to survive ground test. Open questions I'd attack first:

1. Heat pipes at the edge. Startup from cold (sodium is frozen on the runway), transport limits at high heat flux, redundancy if one pipe dries out, and the joint between a razor-sharp edge and a working-fluid channel.
2. Ring survival. Refractory metals oxidize enthusiastically in hot, oxygen-rich supersonic flow; the ring needs coatings (iridium, silicon carbide) or engineered ceramics, and it must take thermal cycling and vibration for thousands of cycles, not sixteen.
3. Flowpath integration. A ring in a scramjet duct is drag and shock structure. Does it anchor combustion across the whole duct, or only in its own wake? Likely answer: pair it with cavity flameholders and let the ring be the ignition authority, not the whole strategy.
4. Capture reality. The 60% capture and half-stagnation strip average are assumptions. Conduction losses, radiation from the hot skin, and installation details will tax them.
5. Mode transition. Ramjet-to-scramjet handoff around Mach 5, with the pilot ring coming alive exactly as the conventional flameholders give up.

Feasibility check is just the beginning. Turning raw theory into working hardware demands iterative design, systematic testing, and a healthy dose of reality check - this is where the fun is.

What's Next?

The next step is to build the thing, test it to the limit, break it, iterate, and keep going until we have a working scramjet.

For testing purposes, once we get closer to an integrated scramjet-powered prototype, I'll need to channel my inner Kelly Johnson and reach out to a DoD general, asking to donate a few missiles to get the prototype up to speed and test it in flight.

But that's a story for another time.

Final Thoughts

The 'c' in scramjet is for combustion, but if there's no combustion, what do we call it then?

That was the ending I planned to write. I don't get to. The 'c' stays. Combustion keeps its job as the energy source, because 43 MJ/kg with free oxidizer is not an offer you refuse. The heat I tried to replace it with gets a better job instead: lighting the fire, holding it steady, and paying back a few percent on the way through.

I set out to delete the flameholding problem and ended up with two walls and a door. The walls cost me a champagne cork. The door - a self-scaling, consumable-free ignition system that gets stronger exactly when the problem gets harder - was worth it.