Reconnaissance with the SR-71 Blackbird

Mach 3.2 at 85.000 ft – the SR-71 in Flightgear

Author: Thorsten Renk


A flight in the SR-71, or the ‘Habu’ as the crews call it, starts long before you enter the cockpit. With the aircraft making Mach 3, you can’t simply fly where you like or take a wrong turn. The Blackbird goes half a mile in the time it takes you to say ‘Oops’, it can be 20 miles in enemy territory by the time it takes you to check a map, and the turn radius is more than hundred miles. This means that basically anything you do needs to be planned in advance.

On longer recon flights, we would have tankers waiting for us in certain locations, but today is just a training flight. We will take off from Nellis AFB, Nevada, go north climbing, then turn around and overfly Nevada at 85.000 ft under mission conditions, then descend and head back to Nellis. All the waypoints for this flight have to be entered into the Astro-Inertial navigation system of the Blackbird in advance.

I am using Flightgear’s route manager to define the waypoints. As in reality, the plane is difficult to control at high altitudes manually, so the autopilot will have to take care of the climb to 85.000 ft. The waypoints need to be defined carefully such that the course is even possible to follow – at service ceiling, the plane is not very maneuverable. I could also, using the AI system of Flightgear, arrange for various tankers to meet me at certain points during my mission if I would want to fly a realistic long range mission profile for the SR-71.

When everything is ready, we finally enter the plane and taxi to the runway. The weather conditions are ideal for reconnaisance – it’s a very clear day with dry air and few clouds.

Many airports in Flightgear have a detailed network of taxiways and one can start the simulation on a specified parking position rather than ready on the runway, and an ever-increasing number of airports also is populated in full detail with 3d models showing not only the main buildings, but also other operations currently ongoing. Nellis AFB is one of the most detailed airports, where one can spend literally hours to explore every detail.

Takeoff and climb

With full afterburners, we gain speed and take off.

The two J58 engines with 34.000 pounts of thrust each sure look impressive with full AB thrust engaged – but the Habu is also a rather heavy bird. Moreover, the engines are designed for high altitude operations, so we just have a thrust/weight ratio of about 0.44, nowhere near to a fighter jet, and so even with full AB thrust, the climb is rather slow.

At about 25.000 ft, we go just a little supersonic for the first time. In this regime, wave drag is very high and the engines actually are not powerful enough to accelerate the aircraft any further. Also, in the thin air, the plane becomes increasingly difficult to handle precisely, and I transfer control to the autopilot.

In order to climb out to full altitude, we have to use gravity’s help and perform the so-called ‘dipsy’ maneuver – we climb to 33.000 ft, level off and let the plane go as fast as it can, then do a shallow dive to about 30.000 ft to let gravity accelerate us to Mach 1.25. Now we’re out of the wave drag region, i.e. drag is much reduced and we can climb further.

Flightgear handles the procedure rather accurately, It is not possible to simply hit the afterburners and fly to 85.000 ft, and if you do not reach sufficient speed at a given altitude, you can’t climb any further. The Blackbird reqires the pilot to adhere to the essential procedures. As in reality, in this altitude it is very difficult to control the plane manually with the precision required for the maneuver, but the autopilot can handle it well.

At the edge of space

Under the control of the autopilot, we continue to climb with a constant KEAS (equivalent airspeed) value of 450 kt all the way up to 70.000 ft, and then let the KEAS value drop to 400 kt while we reach 85.000 ft and Mach 3.2. At this altitude, we’re literally on the edge of space, and utterly alone – no other aircraft can reach this altitude.

The view from 85.000 ft is spectacular on a clear day, and at mission altitude the operator in the back seat becomes busy while the pilot can relax a little since the plane does little but fly straight under AP control.

Flightgear has an experimental skydome shader which tries to solve the physics of light scattering in the atmosphere in addition to the default skydome which handles both foggy and clear conditions reasonably well. The more detailed scattering solution is especially suitable for a thin atmosphere, such as at high altitude or on a very clear day, and it can give quite spectacular results under the right conditions.

At this altitude, the difference between indicated airspeed and the actual speed over ground is very pronounced: While we read just about 400 kt in the cockpit, we’re actually going more than 1900 kt groundspeed.

Flightgear has accurate models for the atmosphere at high altitude and effects like ram pressure taking the difference between true airspeed, indicated airspeed and equivalent airspeed, as well as Mach number to airspeed change with altitude into account. For most planes, these effects are not very prominent, but for the Blackbird they show up rather pronounced.

Returning to base

After completing the recon run, we slow down to 350 KEAS and descend again to 20.000 ft where I switch off the autopilot and resume manual control. In evening light, we head back to Nellis through a scattered cloud layer.

Some more dense clouds hang over Las Vegas as we merge into the approach pattern for Nellis AFB.

Cloud formation is tied to some degree to location: clouds are much more likely to form over the sun-warmed city than over cool open water. Also, terrain elevation plays some role.

The Habu is a supersonic bird – at low speeds it handles like a brick. One needs to be very careful not to lose too much airspeed when turning into the final approach. As compared with other planes, the approach is also really fast to retain enough lift – the Habu approaches with about 220 kt and touches down with litte under 200 kt – more than many propeller-driven aircraft will ever make. However, there remains the problem of deceleration… As we turn into final approach, I arm the drag chute, which is automatically deployed as we touch down.

The JSBSim Flight Dynamics Model handles object like the drag chute rather well as external forces. The drag chute has its own aerodynamical properties, it feels the wind and the drag effect is velocity dependent. As in reality, it takes quite a lot of space to decelerate a plane touching down with 200 kt, and in fact without the drag chute it would be a problem to slow down even given the long runway at Nellis.

After a successful training mission, we reach the temporary parking position of the Habu and head for debriefing, before we leave the base for a nice, cold beer in Las Vegas.

Carrier Ops (USS Carl Vinson)

Carrier Operations in Flightgear

Author: Thorsten Renk

Pre-flight preparation

The flight deck of the USS Carl Vinson, 8:30 am Pacific Daylight Time, off the US west coast: an F-14b is made ready for a flight. The weather is rough, 16 kt of winds coming from the open ocean, with gusts reaching up to 20 kt and changing directions. The Vinson has just crossed a patch of rain, but the clouds seem to be breaking up.

While the ground crew takes care of the plane, the pilot and the RIO go through mission briefing. Our flight this morning will be an intercept training – there is an intercept target north of us which we are to identify.

The scenario is set up using Flightgear’s AI system – both the carrier group and the intercept target are defined as AI scenarios which are defined before starting the simulation. Here I am using a simple setup placing a target on a predefined course – but using Flightgear’s scripting language, it would easily be possible to set up a situation completely unknown to me, or an unknown number of targets, or even a scenario which reacts to my presence in a certain way. AI scenarios can be quite complex – the Vinson scenario simulates the movement of a whole carrier group! The weather conditions can come from live weather reports, or be generated by a sophisticated offline weather system. Many planes in Flightgear (such as the F-14b) offer multi-crew support, i.e. in principle I could share this mission with a human as RIO – in this case however, I’m actually flying alone.

Ready to launch!

We enter the cockpit and close the canopy. While the crew arms the plane (we’ll be carrying a light air-superiority loadout), I am busy adjusting the plane for takeoff. Among other things, I adjust my altimeter to the current pressure and enter the TACAN channel of the Vinson into the left console. TACAN (TACtical Air Navigation) will be my guide back to the Vinson across a cloud-covered, featureless ocean. I also check the fuel loadout – due to the somewhat rough weather conditions and gusty winds, I prefer to take a lighter fuel load rather than launch with all tanks full.

After all preparations are done, I taxi the plane to the launch catapult and it is attached to the guiding rail. I set the throttle to full afterburner – we are good to go. Windgusts blow the catapult steam all over the deck.

Aircraft in Flightgear allow to customize fuel load, and quite often also the weight distribution of cargo, passengers, or in the case of the F-14, the armamant. All this influences the behaviour the plane will show later in the air, thus this is also an important part of pre-flight preparation. For western fighter jets such as the F-14b, radio navigation is done using the TACAN system. Flightgear has both ‘fixed’ TACAN installations (for instance at airbases) which are part of the scenery, as well as definable TACAN channels to be assigned to AI objects. 

In the air

The catapult launches us forward, and will full afterburners roaring our jet is in the air. For a moment the gusty winds shake us hard, but with rolling friction gone the plane accelerates quickly, and as I retract the gear we can climb steeply into the more quiet air above.

The weather simulation distinguishes between the (usually more gusty) boundary layer winds, and the stronger, but less gusty high altitude winds. The thickness of the boundary layer depends largely on terrain roughness, i.e. it is rather thin – as I pull the plane up, I can leave it quickly.

We keep climbing through scattered clouds into a brilliant morning sky.

At 25.000 ft, I level the plane and turn to the planned intercept course. I could use the autopilot for a while, but I enjoy actually flying myself too much.

Many planes in Flightgear have realistic autopilots. In the case of the F-14b, the AP is carefully limited to what functionality its real counterpart can provide – it is a simple system that can level wings, hold an altitude and hold a course, but it cannot by itself follow radio navigation as the more modern systems of other planes do.

As we go supersonic and race towards the intercept target, the wings automatically fold into their delta configuration to optimize for supersonic flight.

However, today we are in for a disappointment: We do not find the intercept target in time, and racing with full afterburner power, our fuel reserves are quite limited. I decide to abort the chase eventually. To be on the safe side, I ask the Vinson for a tanker.

Tankers could have set up in advance as AI scenario, but Flightgear also has the option to call a tanker for aerial refueling right to your current location – which is what I am using now. 

The KA6 used to refuel the F-14b is quite a small plane and difficult to detect visually, but as we ask for a tanker, we get its TACAN channel to guide us into position. However, I decide to track it on the radar instead (as I would for an intercept) and fly the approach by radar.

The F-14b has a fairly radar that is modelled in quite some detail – it has both a scanning and a tracking mode, it provides information about the target heading and groups targets into different types.

Aerial refueling

Getting fuel from a tanker requires some precision flying – the idea is to approach from behind just a bit faster than the tanker, and then to decelerate without dropping altitude just in the right spot. The trick is to gauge accurately how quickly the plane will slow down once the throttle is pulled back – a mistake there will inevitably lead to oscillations around the right position.

With the probe extended, we approach with just above 250 kt into the sweet spot of the KA6.

and finally start receiving fuel so that we can make it back to Vinson

Aerial refueling, both via probe (as demonstrated here) and boom is implemented in Flightgear. Although many aspects are easier than in real life (there is no turbulence induced by the tanker for instance), it is a tricky enough maneuver to master – especially since the AI tankers fly realistic racetrack patterns, i.e. at some point they start to turn!

Back to Vinson

TACAN guides us back to the Vinson. This time, I fly in the subsonic regime. Another 15 minutes later, we start to descend at the position of the Vinson. Here’s the view from the RIO position as we descend towards a cloud later at around 8000 ft.

We overfly the Vinson and its escort group to get into position for an approach.

Then I slow down the plane, extend flaps, the hook and gear and turn into my final approach. Carrier landings, especially in rough winds, are always more of a controlled crash than a proper landing… but TACAN and the Fresnel Lens Optical Landing System are there to help me align properly in difficult conditions.

However, in this case, the unpredictable crosswinds blows me off course.

Weather in Flightgear can change – gust speed and direction may vary on a short timescale, but winds may also change driven by a new weather report in the live weather system or by the dynamics of the offline weather system. 

At this point I decide to go around, so I switch afterburners back on and retract the gear, blast by the Vinson and come again for a second try. After contacting the Vinson, the carrier turns into a new recovery course.

AI control allows to modify the behaviour of Ai scenarios runtime. In this case, I direct the Vinson to a new course better suited for my landing while I go around.

Caught by the wire on the second attempt…

Missing the approach the first time is not too uncommon with the carrier – it’s always better to try again and hope that things go better than to try to force the aircraft down onto the deck when thing are not going right. Even when touching the deck, it’s not guaranteed that the wire catches, so one should always be prepared to yank the throttle forward.


We get out of the plane…

This time, the mission was a failure – we did not manage to reach the intercept target as planned. But this is as life goes – sometimes things do not work out as planned, sometimes something goes wrong with the plane, sometimes the weather does unpredictable things. The important thing is to be prepared to abort whatever you’re doing if it’s unsafe, and to react to the conditions. It’s always better to stay on the safe side than to end the day in flames.

Flightgear has the option to randomly fail systems with a certain probability. Had I wanted, I could have set up the simulation in such a way that my altimeter wouldn’t work. In several planes, even quite detailed emergency procedures are supported, such as extracting gear without pressure in the hydraulic system, or engine restart in the air after flameout.


Youtube video of Carl Vinson Ops.  (Best viewed by clicking “Watch on YouTube” and then going “Full Screen”)  Seriously FULL SCREEN and CRANK UP THE VOLUME!!!


Predator drone video footage circling the Carl Vinson …