Posts Tagged ‘simulator’

We Don’t Train For That

Monday, July 7th, 2014

The tragic Gulfstream IV accident in Boston has been on my mind lately, partly because I fly that aircraft, but also because the facts of the case are disquieting.

While I’m not interested in speculating about the cause, I don’t mind discussing factual information that the NTSB has already released to the public. And one of the initial details they provided was that the airplane reached takeoff speed but the pilot flying was not able to raise the nose (or “rotate,” in jet parlance).

My first thought after hearing this? “We don’t train for that.” Every scenario covered during initial and recurrent training—whether in the simulator or the classroom—is based on one of two sequences: a malfunction prior to V1, in which case we stop, or a malfunction after V1, in which case we continue the takeoff and deal with the problem in the air. As far as I know, every multi-engine jet is operated the same way.

But nowhere is there any discussion or training on what to do if you reach the takeoff decision speed (V1), elect to continue, reach Vr, and are then unable to make the airplane fly. You’re forced into doing something that years of training has taught you to never do: blow past V1, Vr, V2, and then attempt an abort.

In this case, the airplane reached 165 knots—about 45 knots beyond the takeoff/abort decision speed. To call that uncharted territory would be generous. Meanwhile, thirty tons of metal and fuel is hurtling down the runway at nearly a football field per second.

We just don’t train for it. But maybe we should. Perhaps instead of focusing on simple engine failures we ought to look at the things that are causing accidents and add them to a database of training scenarios which can be enacted in the simulator without prior notice. Of course, this would have to be a no-jeopardy situation for the pilots. This wouldn’t be a test, it would be a learning experience based on real-world situations encountered by pilots flying actual airplanes. In some cases there’s no good solution, but even then I believe there are valuable things to be learned.

In the case of the Gulfstream IV, there have been four fatal accidents since the aircraft went into service more than a quarter of a century ago. As many news publications have noted, that’s not a bad record. But all four have something in common: each occurred on the ground.

  • October 30, 1996: a Gulfstream IV crashed during takeoff after the pilots lose control during a gusting crosswind.
  • February 12, 2012: a Gulfstream IV overran the 2,000 meter long runway at Bukavu-Kamenbe
  • July 13, 2012: a G-IV on a repositioning flight in southern France departs the runway during landing and broke apart after hitting a stand of trees.
  • May 31, 2014: the Gulfstream accident in Boston

In the few years that I’ve been flying this outstanding aircraft, I’ve seen a variety of odd things happen, from preflight brake system anomalies to flaps that wouldn’t deploy when the airplane was cold-soaked to a “main entry door” annunciation at 45,000 feet (believe me, that gets your attention!).

This isn’t to say the G-IV is an unsafe airplane. Far from it. But like most aircraft, it’s a highly complex piece of machinery with tens of thousands of individual parts. All sorts of tribal knowledge comes from instructors and line pilots during recurrent training. With each anomaly related to us in class, I always end up thinking to myself “we should run that scenario in the simulator.”

Cases like United 232, Apollo 13, Air France 447, and US Air 1549 prove time and time again that not every failure is covered by training or checklists. Corporate/charter aviation is already pretty safe… but perhaps we can do even better.

The missing link in simulation

Thursday, October 31st, 2013

Several months ago I mused about the how ever-advancing computer technology has led to a marked improvement in simulators for the light GA market. After my post was published, reader Keith Smith alerted me to a corresponding service he had developed called PilotEdge. His company’s mission is to add a level of realism to the general aviation FTD that not even the multi-million dollar Level D boxes have thus far been able to offer.

I was intrigued. What could possible transform an inexpensive Flight Training Device that way? In a word: radios. As Keith said, “People use [simulators] for things they can’t easily do in the airplane because they lack real ATC and real traffic. If you had those elements, an ordinary end-to-end flight would now be beneficial in the sim, because it would more accurately model the workload associated with conducting the flight.”

That’s when it hit me: I’ve been training regularly in a full-motion Level D Gulfstream IV-SP simulator for a few years now, and despite the accuracy with which the cockpit, visuals, and motion are replicated, it’s never been exactly like flying the actual jet. I never spent much time thinking about why. Adding live air traffic control and filling the skies with actual traffic, operated by humans who spoke on the radio would completely revolutionize the experience, because for better or worse, pilots invest tremendous energy and attention on those two elements. We have to listen for our call sign, respond to queries, and interact with other people on a continual basis.

This isn’t about radio skills (although the service would definitely be useful for that purpose), it’s about workload. Keith related the story of a sim pilot who was so busy in the traffic pattern dealing with a Skyhawk ahead of him and a King Air on a three-mile straight-in for another runway that he failed to notice that he only had two green “gear down” lights.

The shower of sparks was impressive — but nothing compared to the look of horror on his face. He was sure he had confirmed the landing gear position. In fact, he heard the gear coming down and felt the vibration, but a badly timed call from the controller asking him to widen out on downwind distracted him and he never finished the checks. His radio work was perfect, but he failed to prioritize the necessary tasks. You couldn’t duplicate that without PilotEdge.

Bringing the workload closer to real world levels reveals chinks in the student’s armor; in fact, it’s exactly what instructors do with their students in real life: give them a heavy workload to see how they deal with the stress.

Imagine running an emergency in the simulator — say, an engine failure or depressurization scenario — and how much better it would be with a controller on the other end of the radio. You declare an emergency, and they start asking you about fuel remaining, souls on board, what are your intentions, do you need assistance, etc. That’s realism. It’s also a great opportunity to learn things a simulator normally never teaches you, like the fact that ignoring ATC is sometimes the best and safest option when you need to fully focus on flying the airplane. Imagine a copilot trying to read a challenge-response checklist to you in one ear while ATC is yammering away in the other.

Instructors using the PilotEdge service have a textual “back channel” to the controllers and can request scenarios like lost comm, a late go-around, poor vectoring, holds, and literally anything else a real controller would throw at you.

How It Works

The goal is 100% fidelity. ATC services are as realistic as PilotEdge can make them. They used the Freedom of Information Act to obtain SOPs for Southern California towers, approach control, and Center sectors. They also familiarize themselves with local airport customs by listening to The sim controllers are paid by PilotEdge and use the same phraseology and procedures utilized by FAA-certified ATC specialists.

But “live” ATC is not very realistic if you’re the only one in the sky. So PilotEdge uses what they call “traffic shaping”. Rather than merely hoping for traffic, they coordinate actual pilots with simulators in remote locations to be at the right place at the right time flying a specified route to create that traffic. And they’re on the frequency as well. Listening for your call sign is something you have to do as much or more in the simulator than you’d be doing in real life. You’ll wait for departure, get stepped on during transmissions, and do all the other things that would happen in a real airplane.

PilotEdge’s service area covers Southern California. Some of their traffic is live, while the rest is computer-generated. PilotEdge has 400 drones flying around the area at all times in Echo and Golf airspace, squawking 1200 and not talking to anyone. They’re programmed to fly exactly as real-world “non-participating” targets do. They’re in the VFR practice areas, the Palos Verdes aerobatic area, and so on. They have military aircraft flying at high speed on military training routes, light GA aircraft on multi-hour cross-countries, gliders (again, without a transponder) flying ridge lift off of Warner Springs and around Mojave, etc.

Here’s a three minute overview of the PilotEdge service:

The Genesis

I’d never heard of a service like PilotEdge before, but Keith said they are not the only one providing ATC services for simulators. The difference is, the “other guys” are using voice-recognition software limited to prepackaged scenarios rather than a room full of human controllers who can deal with — and dish out — anything you can dream up.

Keith Smith started with an early internet-based attempt at simulating air traffic control called VATSIM, which began by using text and later went to Voice-Over-IP.

“That’s where the idea came about; I was a controller there for seven years or so. It’s got lots of flaws for commercial use, but it was the genesis. I couldn’t convince other pilots to use VATSIM due to technical difficulty, so I built PilotEdge from the ground up, licensed the radar scope technology, and off we went.

The radio source code is fairly complicated, but beyond that the service is more evolutionary than revolutionary. Technology is not the key. The secret is our operating model: ATC services provided fifteen hours a day, no requirement for scheduling in advance, and it’s just like the real ATC system.

Also, VATSIM strictly prohibits commercial use, whereas we are built for that purpose. Once a fee is charged, a volunteer service like VATSIM gets complicated. Who gets paid and who does not?”

I asked him how the reception has been for PilotEdge. “It’s a tricky question to answer. It depends on the market. Right now we’re sitting at around 400 users and we’ve been there for 3-4 months. We bring some flight schools on, others drop out. The middle of the market has not been strong, but relationships on the upper-end have made up for it. But we’re a small company, only two years old and definitely still a start-up as far as funding goes.”

A PilotEdge air traffic controller working the "virtual" tower cab at Long Beach (LGB) Airport.

A PilotEdge air traffic controller working the “virtual” tower cab at Long Beach (LGB) Airport.

On the light GA side, PilotEdge is about building radio skills and proficiency at a low cost. With the price of flying spiraling upward at an alarming rate, it’s getting too expensive to operate a real airplane just to build mastery of radio communication.

Even so, it’s been hard for PilotEdge to get much traction with the prototypical flight school. These FBOs tend to be run by people who are overworked. Changes to their programs — especially if it’s an FAA-approved Part 141 syllabus — are difficult to make, and the main emphasis for these companies is keeping the leaseback airplanes flying. Likewise, instructors need to build time, so they want to fly, not sit in a simulator.

Keith feels he’ll be most successful with home users and corporate training centers, because all they do is simulation. The center of market is going to be soft because simulation is not as mature there (although that’s starting to change due to the Redbird Effect).

Expansion on the Horizon

Chicago Jet Group recently obtained an STC to put CPDLC (Controller-Pilot Data Link Communication — basically ATC via text) into Falcons and Gulfstreams, and they contacted PilotEdge to help provide training. VATSIM started with text-only, so it’s an easy transition. Keith said anyone who worked with VATSIM would feel right at home.

I wondered if PilotEdge would ever expand their service area beyond SoCal, and he responded by saying that airspace is airspace, but if the need arose, sure. They picked ZLA because there are simple, moderate, and highly complex areas around SoCal. Keeping the service area restricted increases density of traffic and that congestion helps training and realism. Having said that, there is a company looking to provide PilotEdge service for the New York area because they have a commercial contract to fulfill for that region.

The brass ring for a company like PilotEdge is, of course, the major training centers like Simuflite, FSI, and Simcom. Even NASA has shown an interest.

They’re already making some inroads there via a partnership with ProFlight LLC, a Part 142 training facility in Carlsbad, CA. Founder Caleb Taylor has deployed PilotEdge in their simulators and is basing their business model on that service. Their goal is not just recurrent training, but continual training where pilots can come in any time at no cost and use the device, solo. Well, if it’s used solo, there’s no instructor pretending to deliver ATC (badly, in most cases). So, enter PilotEdge.

Additionally, during ground training, where simulators are not generally used until after classroom training is complete, they want to use their $6 million sim as a training aid. Students will jump in the cockpit and practice using all the systems, including the FMS. There, too, ATC has a role. Lastly, students enter the flight training portion of the formal initial or recurrent program and log their sessions with an instructor. But they will be encouraged to follow up with a bunch of solo sessions, again, with PilotEdge.

All Roads Lead to Savannah

The PilotEdge virtual air traffic control center set up at the 2011 Airventure show in Oshkosh.

The PilotEdge virtual air traffic control center set up at the 2011 Airventure show in Oshkosh.

Keith knew that I fly Gulfstreams for a living and mentioned that they’re working with the folks in Savannah as well. Of course, that piqued my curiosity pretty quickly. He said that Gulfstream is using PilotEdge to save on certification costs related to the avionics in the G650. They’re developing the first FMS update for that airplane, and traditionally the human factors certification takes place in the actual jet. That’s expensive. Operating a G650 costs thousands of dollars per hour. PilotEdge allowed them to move that work into a simulator with full FAA blessing.

“We’re a small company nobody’s heard of, but the Gulfstream project got us in the door at FlightSafety. But even then, they were under the impression that it was voice recognition software, a synthetic product using rigid scenarios.”

It’s Not Just for Pilots

PilotEdge can work in reverse, too. Sacramento City College trains controllers before they go to Oklahoma City for formal coursework with the FAA. They setup a lab with simulators and use PilotEdge to get trainees a leg up on the intricacies of keeping a flurry of flying aluminum sequenced and separated.

Keith said they just put together a proposal for the Mexican Navy as well. Again, competitors use voice recognition software, but that technology doesn’t scale easily when the language in question is Spanish rather than English. He said PilotEdge’s pricing is also superior.

Speaking of English, no matter where you go — and I’ve been on virtually every continent — controllers and pilots are supposed to be capable of communicating in English. There’s no other way to ensure a pilot whose native language is Portuguese can talk to a controller in China who’s primary tongue is Mandarin. So a huge aspect of the international training market is dictated by the ICAO Level 6 English requirements. That regulation has teeth to it, and everyone’s struggling to get their people up to speed. Guess who can help with that?

The Bottom Line

I’m frankly a little surprised that nobody’s come up with a service like PilotEdge before Keith Smith and his team made it happen. As previously noted, the requisite technology has been with us for many years. In some ways PilotEdge is almost anachronistic. From manufacturing to fast food, industries are moving toward greater automation and a lower employee count. PilotEdge is doing the exact opposite, supplanting automated ATC simulation with live humans. Not that I’m complaining, mind you. I’ve had the misfortune to interact with a couple of these computerized programs in the past and always come away wishing I could get the last two hours of my life back.

The combination of a new generation of simulators and PilotEdge’s addition of air traffic and ATC has the potential to vastly improve the way pilots train while simultaneously reducing the cost of obtaining everything from a sport pilot certificate to a turbojet type rating. I can see this powerful duo creating an aviation equivalent of the smartphone explosion and helping turn the tide toward a more prosperous future.

Perhaps evolutionary is revolutionary after all.