Assessing Glass Cockpit Safety
The expected operational and attendant safety improvement from the glass cockpit has not materialized as expected. What’s missing?
On November 8, 2007, a G1000 equipped Cessna T182T collided with terrain in the vicinity of Mount Potosi (8,514 feet MSL) on a clear but dark night, 21 miles southwest of the departure airport, North Las Vegas Airport (KVGT), Nevada. The Cessna was flying on a VFR flight plan to L00 (Rosamond Skypark Airport in California), approximately 168 miles southwest of KVGT.
At the time of this now well publicized accident, the crew was in radio and radar contact with Las Vegas Departure Control and was instructed to stay clear of the Class B airspace. There are many lessons to be learned from this and other accidents that have involved glass displays.
While the weather was clear, the surrounding terrain had no lighted roads or structures that could have provided some ground reference. While legally, VFR is defined by visibility of three miles or greater; the flight in such conditions was in fact, flying in IMC given the lack of outside visual cues.
The G1000 can display topographic data on its MFD (multi-function display), but due to extensive thermal damage resulting from the crash, the NTSB was not able to determine if the pilots were seeing this information. Nevertheless, Garmin’s Pilot Guide states: “CAUTION: Use of terrain proximity information for primary terrain avoidance is prohibited. It is the pilot’s responsibility to provide terrain avoidance at all times.”
The NTSB’s probable cause of this CFIT accident was not surprising: “...failure to maintain an adequate terrain clearance/altitude during climb to cruise. Contributing to the accident were rising mountainous terrain, the dark nighttime lighting condition, the pilot’s loss of situational awareness...”
Pilot Experience Factors
There were two highly experienced pilots on board. The left seat pilot held ATP with several type ratings and CFI certificates with over 25,000 hours. The right seat pilot also held ATP and CFI certificates and had amassed over 28,000 hours. It was not clear who was acting as the PIC. Both pilots were experienced with the G1000 and the Cessna T182T.
There might have been other factors such as having two highly experienced pilots with no clear authority of responsibility for flight safety. Or could the G1000 have played a role? This naturally leads to the question: do glass cockpits increase safety?
TAA vs. Glass Cockpits
Some pilots equate technically advanced aircraft (TAA) with glass cockpits. However, not all TAA have glass cockpits. A TAA (generation I) needs to have a moving map display, an IFR approved GPS and integrated autopilot. These aircraft could have round gauges. TAA generation II have glass cockpits such as the G1000. For simplicity, the term “glass cockpit” will be used here rather than TAA generation II.
Correlation vs. Causation
Before addressing the question of the impact of glass cockpits on safety, let’s review the concepts of correlation and causation. This is not just applicable to aviation but also to other disciplines: medicine, economics, marketing research, law, etc.
Correlation simply means that there is a mathematical relationship between two variables or factors. Both factors move in the same direction or in opposite directions. The statistical measure of correlation is called Pearson Correlation Coefficient.
The first issue with correlation is spurious correlation—mathematically it exists but just by chance.
The second issue is that we might have correlation between two variables but no cause and effect between the variables. However, there might be cause and effect if a third variable exists. For example, hypothetically, the tire pressure of an airplane is positively correlated to the number of seats, no cause and effect relationship. Both are related to aircraft gross weight. There are cause and effect relationships between number of seats and GW and tire pressure and GW.”
Let’s suppose we can be definitive that correlation exists; then the question becomes: is there a cause and effect relationship? Incidentally, one cannot have causation without correlation. And if causation does exist, what is the cause and what is the effect?
Causation is much harder to prove as it cannot be proven mathematically but through a profound understanding of the situation. All this preamble is to state up front that to have a definitive conclusion that glass cockpits are safer, we need to prove both correlation and causation. A difficult goal.
The NTSB conducted a rather extensive study on the “Introduction of Glass Cockpit Avionics in Light Aircraft” and issued a lengthy 91-page report in 2010.
Obviously with the passing of time, the statistics may have changed, but given that this is only extensive available study, let’s start here. The question is: Have glass cockpits improved safety?
The study used a three-prong approach. 1) Review of accident statistics comparing conventionally equipped airplanes to those with glass cockpits; 2) Current glass cockpit training to identify potential areas of improvement in training and; 3) Identification of emerging issues.
The overall conclusion based on the accident analysis is: “The results of this study suggest that, for the aircraft and time period studied, the introduction of glass cockpit PFDs has not yet resulted in the anticipated improvement in safety when compared to similar aircraft with conventional instruments.”
A key set of data seems to indicate that when a fatal accident occurs, glass cockpits have a higher incidence. The conclusion appears to contradict the data—glass cockpits have worsened safety levels. This can be in part explained that neither correlation nor causation have been proven.
Additionally, when compared to VMC and IMC, the glass cockpit equipped airplanes when flown in IMC, the incidence of fatal accidents is also higher.
Where do we go from here? The bottom line is that the study could not determine correlation and much less causation. This is not negative comment regarding the study but rather a common situation whenever we are trying to prove cause and effect especially when we have multiple factors involved. The best we can do is to state a broad conclusion that can tangentially be supported by the data: “has not yet resulted in the anticipated improvement in safety.” We can go into all the details of the study. We could also discuss weaknesses in the methodology that have been pointed out by aviation safety experts. But given the available data, no matter how artful we are in slicing and dicing the data, we probably will not come out with other conclusions that would withstand the sniff test.
Based on my experience and what I have observed, my conclusion regarding glass cockpit safety is rather simplistic: an airplane is as safe as the pilot. However, the equipment might also enhance safety, but we cannot simply conclude that advanced avionics automatically result in more safety.
Some other points to consider. Glass cockpit airplanes are typically flown with the autopilot on. This allows the hands and mind to focus on other (hopefully) higher-level attentions. Perhaps they are flown more in IMC and for longer distances. We can also identify other potentially contributing factors that could contribute to a higher perceived risk.
While we may or may not agree with the NTSB conclusions, if you accept my simplistic explanation related to “it’s the pilot” not the technology that is at the crux of the matter, the six NTSB recommendations allude to training and additional information that pilots need.
Based on the study findings, the NTSB Press Office summarized the six recommendations in the table to the right (source NTSB press office).
I have gone a step further and summarized my thoughts in the table below on how to increase the safety and value of glass cockpits. As usual, I look forward to comments from fellow subscribers.
Additionally, I have been flying a C182T with the new G1000 NXi. It is a great piece of avionics but it still has many buttons, knobs of different sizes and shapes and soft keys.
Some G1000 NXi installations like in the Mooney Ovation Ultra also have a mechanical keyboard located below the throttle quadrant that requires substantial heads down.
In G1000 installations, I tend to look down more often because of the ergonomics of the controls. I’m going out on a limb, I predict that the next generation of the G1000 will be touchscreen like the GTNs and the new G500/600 TXis. Touch screens require less heads down and more intuitive controls vs. scroll/twist/push knobs and assorted buttons in G1000s.
Luca Bencini-Tibo, ATP/CFII, is a FAASTeam Lead Rep, aircraft owner and is a graduate of MIT and Harvard Business School.
For more great content like this, subscribe to IFR Refresher!