On Friday 13 November 2026 the US Air Force announced that a B-21 Raider flying from Edwards Air Force Base, California, had released a conventional weapon the week before. The headline was the weapon. The interesting part was a sentence further down: the test, the service said, had been built on thousands of hours of virtual modelling inside the bomber’s “digital twin”.
That sounds like a press-office flourish. It is actually the latest chapter in one of the oldest and least glamorous problems in military flight test: how do you prove that a weapon will leave an aircraft cleanly, without letting it hit the aircraft first? Engineers call it store separation, and for most of aviation history the honest answer was “carefully, and with a lot of nerve.”
Kurzinfo
- Store separation: Proving a bomb, missile, tank or pod leaves an aircraft safely across its release envelope
- Early method: “Hit or miss”: drops at gradually increasing speeds
- Wind tunnel: Captive Trajectory System developed in the 1960s
- Key US tunnel: AEDC 4T, Arnold AFB, Tennessee: below Mach 0.2 to Mach 2.5
- CFD: Able to model a store in an aircraft flow field by the late 1970s and early 1980s
- USAF authority: Air Force SEEK EAGLE Office, chartered December 1987
- Example: JSOW on the F/A-18C, 1994: 24 planned flights cut to 15
- Latest: B-21 conventional weapon release announced 13 November 2026, built on digital twin modelling
Why letting go is hard
A bomb, missile or fuel tank hanging under a jet is a “store”. Releasing one sounds simple: open the hook, let gravity work. But the air around a fighter at several hundred knots is anything but calm. It is bent by the fuselage, the wings, the pylons and every other store nearby, and the moment a weapon is released it is flying through that disturbed flow, a few feet from the aircraft that just carried it.
NAVAIR engineer A. Cenko, reviewing 30 years of the discipline in a 2010 paper, singled out the forces and moments acting on a store while it is still attached, the “carriage moments”, as the most critical feature in deciding where it goes next. A weapon that leaves with its nose already being pushed up can swing straight back into the wing.
Stealth makes it harder. A NASA Langley study of stores leaving a box-shaped cavity listed the attractions of internal carriage, including lower drag and a smaller radar cross-section, and the drawbacks: large dynamic loads on the weapons bay when it is open and, for some bay shapes, real difficulty getting the store out cleanly. Afterburner covered exactly that problem when an F-35B first jettisoned a Meteor from its bay this autumn.

The hit-or-miss years
The original method was brutally empirical. You dropped the weapon at a modest speed, watched what happened, then went a bit faster next time. Cenko described how it ended.
He added that, in some cases, this led to the loss of aircraft, and had made test pilots reluctant to take part in store separation flight testing. Nobody needed a second explanation of why the industry wanted a better way.
Into the wind tunnel
The better way arrived in the 1960s with the Captive Trajectory System, or CTS. A scale model of the aircraft sits in a wind tunnel next to a separate model of the store on its own motorised support. The aerodynamic loads on the store model are used to simulate its path, and a computer drives the model to each newly computed position and attitude, point by point, so the trajectory is traced without anything ever being dropped.
The workhorse in the United States is the Arnold Engineering Development Complex in Tennessee. Its 4-foot Transonic Wind Tunnel, 4T, runs from below Mach 0.2 to Mach 2.5, simulates altitudes from sea level to 98,000 feet, and has been used for weapons integration testing on the F-35, F-22, F/A-18C Hornet, F-14, F-15 Und F-16.
Store separation testing in the wind tunnel, from the British Aircraft Research Association.
In one test there, a one-fifteenth scale F-35A was fitted with GBU-54/38 Laser JDAM bombs and AIM-120 and AIM-9X missiles. The first phase measured how carried stores changed the jet’s own stability and control; the second used captive trajectory and grid survey methods to map how the weapons would separate.
Wind tunnels had a catch, though. Models had to be small, and Cenko notes that the tunnel predictions often failed to match what happened in flight. Worse, the tunnel engineers and the flight testers usually worked for different organisations in different places, with no mechanism for settling who was right.

Then came the computers
By the late 1970s and early 1980s, computational fluid dynamics had matured enough to solve for a store inside an aircraft’s flow field. In Cenko’s telling, it did not spark a renaissance. It started a three-way argument. The CFD specialists claimed they could replace the wind tunnel, the tunnel engineers said the CFD crowd did not grasp the complexity of the problem, and the flight testers said neither group gave them the data they actually needed.
A computational fluid dynamics simulation of a store separation case, published by CFD software developer Metacomp Technologies.
The fix was to stop arguing and chain the tools together. The US Navy’s integrated approach uses CFD to design the wind tunnel test, and the tunnel results to design the flight test matrix. The flight results then check both the tunnel data and the original CFD, so the models get better with every programme.
The payoff shows up in flight hours. When the AGM-154 Joint Standoff Weapon was cleared on the F/A-18C in 1994, the plan called for 24 separate flights. Because the pre-flight trajectory simulations matched so well, the team needed only 15. Better telemetry and dropping stores from both wings on one sortie squeezed things further: in September 2015 an F-35C test team at Patuxent River released four inert GBU-12s on a single flight, four flights’ worth of work in one.

Where the digital twin comes in
In the strict sense, a digital twin is a model that keeps updating itself with real data from its physical counterpart; looser usage covers any detailed virtual replica. For weapons integration, the useful idea is the loop: predict the separation in software, fly a small number of carefully chosen test points, feed the measured results back, and trust the model a little further next time.
In the US Air Force, every weapon, pylon, tank and pod carried by an aircraft has to be certified for safe carriage, release and jettison. That job belongs to the Air Force SEEK EAGLE Office, chartered in December 1987, whose process combines digital modelling, simulation and analysis with ground and flight tests. What has changed is the balance between the first part and the last.
That is the claim behind the B-21 test flown from Edwards: most of the separation work was done before the bomber took off, and the drop confirmed it. The Air Force has not said which weapon fell, nor published the data, so how close the prediction came remains its own business. You can read what it did confirm in our B-21 weapons release report.
Lockheed Martin footage of the first AIM-120 AMRAAM release from an F-35, the flight-test end of the separation process.
The principle is the one Cenko spelled out in 2010. Looking back at the 24-flight JSOW plan, he reckoned that with modern methods the same clearance would probably take one or two flights and no more than four weapons. Every new aircraft still has to drop something for real. The trick is making sure that, when it does, nobody is surprised.
Sources: A. Cenko (NAVAIR), “Store Separation Lessons Learned During the Last 30 Years”, ICAS 2010; NASA Technical Paper 3110 (Stallings, Wilcox and Forrest, 1991); US Air Force / Arnold Engineering Development Complex (F-35A store separation test release, via DefenceTalk); US Department of the Air Force B-21 release, 13 November 2026 (via Mirage News and Air & Space Forces Magazine); Lockheed Martin; US Navy
Header photo: Lockheed Martin photo by Andy Wolfe / Wikimedia Commons, public domain




0 Comments