Friday, November 6, 2015

PIA: Safer than most airlines at any altitude

The only thing worse than PIA’s safety record is the research that went into this article titled ‘Pakistani airlines: Unsafe at any altitude’, written by Murtaza Haider. Despite being a professional analyst, Mr Haider’s article is flawed in several ways, both statistically and analytically.

Right at the onset, the author is unable to comprehend/relay to readers the difference between an ‘incident’ and an ‘accident’, and proceeds to invent his own terminology of “fatal incident”. By definition, any aircraft incident which results in fatalities is an accident.

For the readers’ information, the definition of an accident and incident from International Civil Aviation Organisation (ICAO) and Federal Aviation Administration (FAA) are similar and state:

Accident: Means an occurrence associated with the operation of an aircraft which takes place between the time any person boards the aircraft with the intention of flight and all such persons have disembarked, and in which any person suffers death or serious injury, or in which the aircraft receives substantial damage.

Incident: Means an occurrence other than an accident with the operation of an aircraft, which affects or could affect the safety of operation

Airline Safety records are anything but open ‘secrets’ as the author so eloquently puts it. There is nothing secret about them and data is openly available on the internet from both regulatory and private/commercial websites.

The “fatalities” numbers mentioned in the comments by the author below the article are incorrect and display the authors misunderstanding of the article from which he derived all his data. The “famous blogger/analsyst/author” of the sourced article in fact has quoted these figures after normalising them. Understanding or doing one’s own research would show PIA actually had far fewer fatalities than the author quoted and significantly fewer than other airlines in Pakistan.

During 1985-99, PIA had three fatal accidents – Two Fokker F27s, one being the infamous lost F27 in the Himalayas with 54 souls on board (widely acknowledged as being shot down) and one that crashed on approach to Peshawar with 14 fatalities. The third accident was the airline’s worst in Kathmandu with 167 fatalities, bringing the total for this period to 234 fatalities. The author in his comments claims 861 fatalities, almost four times as many.

During 2000-2014, PIA only had one fatal accident of an F27 out of Multan with a loss of 45 souls and one passenger died when shots were fired at an aircraft landing in Peshawar. The author in his comments claims 175 fatalities, again almost four times as many.

This brings us to the “fatal” flaw in Nate Silvers blog analysis that the author has carried over. Factoring or “normalising” by Available Seat Kilometres (ASKs) to determine accident rates is like comparing car accident rates to the number of seats within the car and how far the car drives. For example, if a small two-seater car has an accident on its first journey or a large seven-seater car had the same accident on the same single journey, the accident rate would be different due to the difference in number of seats. Same goes for the distance and type of driving. A car driven from point A to B at rush hour is far more likely to have an accident than a car on the same journey at 3am. But these sorts of factors are ignored in this analysis. By only incorporating ASKs into his analysis, he has made a carrier that operates purely jet aircraft on mostly long haul routes for example a 14-hour 7000 kilometres journey with one take-off and landing equal to a turboprop that may do 20 to 30 take-offs and landings in the same distance/time.

It is common knowledge supported by data that proves turboprop accident rates are significantly higher than that of jets. IATA data shows that turboprops have an accident rate, depending on the global region, of anywhere from six to 20 times higher per flight as compared to jet aircraft. Hence, accident analysis figures are not based on ASKs, which is purely for airline marketing functions, but on number of flights or cycles (one cycle being a take-off and landing).

PIA incorporates accident figures from turboprop aircraft in its accident reports. Most of the airlines that have been compared do not even operate turboprop/regional aircraft let alone 40 and even 50 years old ones. If you remove the turboprop data from PIA’s accident/incident data, then you are left with only one fatal accident (Kathmandu crash) in the first period and none in the next. For this very reason, accident rate is usually normalised with some form of number of take-offs and landings or number of cycles incorporated.

Finally, Mr Murtaza Haider justifies normalising incidents/accidents versus ASKs for airlines so as to offset any difference in the size of their fleet size/operation, however, when it comes to private Pakistani airlines, they are conveniently too small to show up on the radar. Yet, the referenced article shows PIA with a fleet of now around 25 aircraft compared to airlines such as American, with the largest fleet in the world of almost 1,000 aeroplanes. In fact, Shaheen Airlines last year had, and possibly still does, a larger operational fleet than PIA but without any turboprop aircraft. But wasn’t that the point of normalising in the first place?

So as to eliminate any difference between a smaller and larger airline? Quite contradictory.

Shaheen Airlines aircraft has a runway excursion for the third time in recent years and an analyst writes an article about PIA’s supposedly abysmal safety record. Bhoja Airlines in its short operation span of a few months lost 127 souls in one accident. Normalise that versus PIA’s 14 years (2000-2014) of 45 (all turboprop aircraft) fatalities? In fact, Bhoja would be quite literally off the chart. In the same period, Airblue lost one aircraft with 152 fatalities.

And since in his article he prefers to concentrate on the incidents and accidents, let’s consider the following:

Accident: Shaheen B734 at Lahore on November 3, 2015 – runway excursion, both main gears collapsed
Incident: Shaheen B734 at Quetta on July 16, 2015 – burst tyre on landing
Incident: Shaheen B734 at Islamabad on April 21, 2015 – engine shut down in flight
Incident: Jordan/Shaheen B762 at Islamabad on February 15, 2015 – engine shut down in flight
Incident: Shaheen A320 near Lahore on April 5, 2015 – cabin pressurisation problems
Accident: Shaheen B734 at Lahore on December 30, 2014 – runway excursion
Incident: Shaheen A333 near Lahore on September 6, 2014 – engine failure
AccidentShaheen B734 at Karachi on April 22, 2012 – left main gear collapse
AccidentShaheen B732 at Peshawar on February 8, 2010 – went off runway
Incident: Shaheen B732 at Dubai on March 2, 2008 – engine failure en route

To sum up, Mr Murtaza Haider and his famous blogger’s analysis is skewed and does not factor in basic accident reporting normalisation techniques commonly used. Incorporating ASKs is just a method of portraying data to give the required results and is not the commonly followed norm. The ease with which the author dispenses other operators within the country would indicate that he favours private airlines as most of the mainstream media seems to do. This is rather worrying as his primary role is that of a data analyst and in fact the data clearly indicates that now more than ever, PIA is in fact the safest airline in Pakistan and should be the top choice of travellers. But then I’m not a data analyst.


from The Express Tribune Blog http://blogs.tribune.com.pk/story/30161/pia-safer-than-most-airlines-at-any-altitude/

No comments:

Post a Comment