A Two Group Experiment to Measure Simulator-Based Upset Recovery Training Transfer

This research by Rogers, Boquet, Howell & DeJohn (2010) was conducted to see if upset-recovery training in a simulator was effective in real life upset-recovery. The 'human factor' that they have noted as being the cause of accidents is the 'surprise' or 'startle' factor. Their research hypothesizes is that “upset-recovery training in a low-cost desktop flight simulator develops flying skills that improve a pilots ability to recover from a serious upset in a real airplane”. This article summarizes the results from their research in a more user friendly format.

Rogers, Boquet and Howell compare each of the independent variables and dependent variables against each other separately. To understand whether the trained participants performed better we can summarize the results for all maneuvers as shown below in Table 1. The differences between the Trained and Un-Trained participants clearly showed that the trained participants on average performed better in each of the recovery categories than the untrained participants.

##### Table 1. Dependent Measure Means and Difference Between Trained and Untrained Groups
 Altitude Loss (feet) Min Unload G in Rolls Max G in Dive Pullout Seconds to First Throttle Seconds to First Roll Seconds to recover Trained 556.86 0.17 3.43 2.35 2.07 9.25 Un-trained 650.72 0.31 2.76 4.05 4.01 10.75 Difference 93.86 0.14 0.67 1.7 1.94 1.50

They then conducted one-way Analysis of Variance (ANOVA) using the results with significant differences between trained and untrained groups for each of the maneuvers as shown in Table 2. The 'Min Unload G in Rolls' data was excluded because the trained pilots had been taught to use rolling pullouts from upright dives. Table 2 statistically confirms with significant F- values that the research hypothesis is in fact true, that simulator training in a low-cost desktop flight simulator develops flying skills that improve a pilots ability to recover from a serious upset in a real airplane.

##### Table 2. One-Way Analysis of Variance (ANOVA) of Significant Differences Between Groups
 Altitude Loss Max G in Dive Pullout Seconds to First Throttle Seconds to First Roll Seconds to Recover Nose-Low Upright F(1,49) = 19.48 p = 0.0001 F(1,49) = 25.52 p = 0.0001 F(1,49) = 14.02 p = 0.0001 F(1,49) = 12.19 p = 0.001 F(1,49) = 14.82 p = 0.0001 Nose-High Upright F(1,49) = 10.11 p = 0.003 F(1,49) = 7.18 p = 0.01 F(1,49) = 6.83 p = 0.012 Nose-Low Inverted F(1,34) = 5.45 p = 0.03 F(1,34) = 16.02 p = 0.0001 F(1,34) = 4.38 p = 0.04 F(1,34) = 17.61 p = 0.0001 Nose-High Inverted F(1,48) = 8.912 p = 0.004 F(1,48) = 7.46 p = 0.009 F(1,48) = 22.29 p = 0.0001 F(1,48) = 10.90 p = 0.002

# Methods

### Research approach

A study to see in low-cost simulator upset recovery training improves a flight students ability to recover from a serious upset in a real airplane.

### Sample

This study used students from Embry-Riddle Aeronautical University who had their Private Pilots Licence with a current Instrument Rating. They had all studied basic aerodynamics as part of the Private Pilots Licence training. None of the participants had aerobatic experience or training further than required for their Private Pilots Licence.

### Design

One group of students had classroom training and simulator training, using Microsoft Flight Simulator (MFS), in upset-recovery prior to conducting the same maneuvers in a real aircraft. The control group of students had no MFS or classroom upset-recovery training prior being subjected to the upsets in the real aircraft.

### Variables

Independent Variables
They used a 2x4 Repeated measure factorial design. The first independent variable is training and the two categories are trained and un-trained. The second independent variable is the upset attitude and has four categories which are Nose-high Upright, Nose-low Upright, Nose-high Inverted and Nose-Low Inverted.

Dependent Variables
Rogers, Boquet, Howell & DeJohn (2010) considered a good upset recovery where a pilot returns the aircraft to straight and level flight with the least amount of altitude lost. There are five dependent variable measures which contribute to a good upset-recovery. They are Altitude Loss (feet), Maximum G-force in dive pullout, minimum G-force unloading during rolls, time to first throttle response (seconds), time to first roll response (seconds) and time to recover (Seconds).

The criteria for a prompt and correct recovery is high G (while remaining within limits) in dive pullout, least amount of G-unloading during rolls, least amount of time to throttle response and roll response which all help in reducing altitude loss as reducing time to recover.

### Procedure

The students who received the training received ten hours’ classroom training and ten hours MFS training. Both trained and untrained students were then subjected to the same maneuvers in a real aerobatic aircraft and recovery parameters were measured and compared between the two groups.

### Data analysis

The data was collected using a video camera which recorded the flight instruments. An Appareo GAU-1000 AHARS data recorder that used GPS to calculate various parameters was also used but G-Force data was the only reliable data during aerobatic maneuvers.

### Generalization potential

This study proved that flight students with recent aeronautical study regarding aerodynamics performed better in upsets in a real airplane with low-cost simulator upset recovery training. It would take further study to determine if experienced airline pilots would react the same as their aerodynamic theory study would not be as current and many would not have flown a light aircraft for many years.

References
: Rodgers, R.O., Boquet, A., Howell, C. and Dejohn, C. (2010). A Two Group Experiment to Measure Simulator-Based Upset Recovery Training Transfer.International Journal of Applied Aviation Studies, FAA Academy, August 2010, Volume 10, Number 1. ISSN: 1546-3214 +++ Notes +++
1. ###

# Want to know more?

item
explanation

 Other interesting sites
page revision: 24, last edited: 25 Sep 2013 03:43