Flight Test Notes: High-Altitude Landings and Wildfires
Earlier this month, our autonomy development Caravan was scheduled to get its “High-altitude” rating by performing automated take offs and landings at high altitude. The objectives of these flight tests were multiple: measure climb performance, validate automatic takeoffs and landings robustness at high altitude airports, validate system performance at “never seen before airports”, and look for system maturity via long flights within normal FAA airspace.
Density Altitude
While altitude (elevation) plays an important role for aircraft performance, density altitude is a concept that more completely predicts changes in performance. The lift, drag, and propulsion forces on an airplane are proportional to the air density, so density altitude is important because it’s the altitude at which the airplane feels it flying. Higher density altitude (DA) means lower air density, hence lower lift, drag, and propulsion. Density altitude is pressure altitude (altitude measured with a barometric altimeter) corrected for nonstandard temperature (Standard temperature is 15C at sea level). Since air density is affected by temperature, as temperature and altitude increase, air density decreases.
The airplane flies differently, typically with much reduced performance (longer take off roll, slower climb, longer landing distance, reduced control effectiveness, …). For landing, one of the significant changes is the True Airspeed (TAS) (speed of the airplane relative to the air) that is significantly higher so the speed across the ground is significantly higher (roughly 2% per 1000' increase in DA).
As the density of the air decreases (higher DA), a wing needs to fly at a higher TAS to maintain the same indicated airspeed. At high density altitude, therefore, a given indicated airspeed (airspeed measured by the air data computer, proportional to air density) equates to a faster ground speed than it does at sea level (assuming the same wind conditions). This means a lot more energy to account for and dissipate when landing.
High density altitude can be achieved with a combination of high elevation and/or high temperature. For our tests, we had selected a few airports, with a preference for lower elevation but higher temperature (in Nevada) as the airports at lower altitude presented less challenges due to surrounding terrain.
Always have backup plans
Our initial plan, was to fly to Minden-Tahoe Airport (KMEV) from our home base of Concord (KCCR). With an elevation of 4,707 ft and a temperature of 90F KMEV’s density altitude was 7272 ft. KMEV was selected to allow us to gradually build up to higher density altitudes without significant terrain features within the vicinity of the airport.
It was on Wednesday August 19, the second day of the major wildfires ravaging California. We climbed to an altitude of 11,500 ft, over the smoke in the central valley. When reaching the Sierras, the conditions worsened with the visibility quickly decreasing.
In order to remain VFR (currently required to keep our experimental system engaged), we decided to replan and navigate around the smoke via the North, heading to Truckee, then Carson City, then Minden-Tahoe (KMEV). As we approached Carson City, air traffic control informed us that a new Temporary Flight Restriction (TFR) due to a fire just popped up right next to our destination airport. A TFR is a part of the airspace where it is temporarily strictly prohibited to fly (Typically due to firefighting activity or when the President is around). Time for another change of plan, this time, change of destination: Reno-Stead (KRTS), just north of Reno. With a temperature of and an elevation of 5,050 ft and a temperature of 90F, the density altitude was 7,265 ft.
As we approached Reno-Stead, ATC informed us that the airport was very busy with 6 fire-fighting aircraft currently in the traffic pattern. We decided to go the next best option: South Lake Tahoe.
With an elevation of 6,269 ft, South Lake Tahoe (KTVL) and a temperature of 84F, the density altitude was 9,130 ft. Quite a difference from our typically hot, but sea level home airport.
KTVL presented the additional challenge to be surrounded by high and steep mountains. No better way to help define requirements for mountain flying than doing it.
We flew over beautiful Lake-Tahoe in a light smoke and approached from the west.
Weather was favoring runway 18, which has a nice straight in approach from over the lake… But we like a bit of challenge and decided to push our system. N101XW entered on crosswind a left pattern for runway 18 and executed a perfect automated landing on the first try (See animated gif).
We then taxied back to runway 18 for a fully automated pattern, from takeoff to landing. The second landing was also perfect, confirming that our controllers are robust to changes in density altitude. This opens up the door to a lot more airports!
We took a quick break, but didn’t stay too long as the smoke was getting more dense.
During the return flight, we noticed that the images from our cameras were getting darker and darker… they were collecting ash. Another example of a challenge encountered by camera based systems. Fortunately the Xwing system uses a fusion of multiple sensors in its detect and avoid package.
During the entire 2h flight from KCCR to South-Tahoe, Xwing’s autoflight system was in control of the airplane. The aircraft was either following waypoints, or being vectored from our control station. An autonomous system must be able to quickly adapt to fast changing environment. While avoiding the smoke was more of an R&D constraint, avoiding popping-up TFRs and replanning in real time is a necessity.
