Another useful data source from the US DoT is named T2.  DoT describes it like this: “This table summarizes the T-100 traffic data reported by U.S. air carriers. The quarterly summary is compiled by aircraft types/configurations, carrier entities (geographical regions in which a carrier operates), and service classes, and includes available seat miles (ASMs), available ton miles (ATMs), revenue passenger miles (RPMs), revenue ton miles (RTMs), revenue air hours (RAHs), revenue miles flown (MILES), revenue departures performed (FLIGHTS), and aircraft fuels issued in gallons. T2 summary includes reported international flights and military service which may not be available in the T100 segment and T100 market data tables released in TranStats. There is no fuel data available in Schedule T-2 for some carriers because while 14 CFR Part 298 (298C) air carriers report T-1 traffic data, they do not report any fuel data.”

We make a lot of use of the T2 because there’s so much in it.  For example, we are able to calculate the average seat capacity by aircraft type because it has the parts necessary.  Another data point we derive from the data is seat miles/gallon.  This is a metric we use to show how fuel-efficient aircraft are.  From this metric, we are also able to build an ESG model showing the carbon emissions by aircraft type.  Suffice it to say the T2 is very handy.

Like the other DoT data sources, we like to model T2 in several ways. Fuel burn is one of those models.  Below is our T2 fuel burn model, where we show fuel burn per seat mile by aircraft types and airlines. 

This model has six pages.  We want you to go to page 6, please.  Here we show the percentage fuel burn improvement of the MAX vs NG and NEO vs CEO. You would imagine this is very interesting for pretty much the entire industry since fuel burn is the single largest input cost to airline operations.

Looking at page 6, you would also find it interesting to see that the newer models are, overall delivering far better fuel burn numbers than the engine and aircraft OEMs expected.  A positive number in the tables means an improvement and a negative number means a deterioration.  

MAX8


The Southwest numbers are suspect. There is no way the MAX8 has worse fuel burn than the 800NG.  We asked the airline about this and were delighted to be advised as follows: “..there’s been some turnover in the group that does the DoT reporting and I suspect they may have gotten some of the equipment codes confused when they reported. Needless to say, they’re straight now but I suspect they may have gotten the MAX ETOPS portion of the fleet jumbled.”  This is an example of how data errors or oddities can be corrected without rancor.  No finger-pointing, no judgment, no harm.  In short a positive outcome for all.

MAX9
The data is plausible, in our view. Alaska and United filings have not shown any “surprises” before.  The data also shows why the MAX9 is the most fuel-efficient single-aisle in US use.

A320neo


There are odd numbers for the first-year operations at both airlines. But all the numbers for Spirit are suspect as the A320neo should be showing double-digit improvement over the A320ceo.  As it does at Frontier. 

A321neo
Here we see consistent numbers which is a good outcome. But the order of magnitude in fuel burn improvement is significantly better than we would expect.  The American A321 fleet numbers and age is shown in the following chart. The A321ceo average is 11 years old and the A321neo average is 1.5 years.  Can a new A321neo be so much better than a decade-old A321ceo? Perhaps.


The situation at JetBlue is that its A321ceos average 6 years old and its A321neos average 1.5 years.  So a much younger fleet overall. But its A321neos are being used across the Atlantic serving JFK-LHR.  This means a quite different deployment than the A321ceos and the longer hauls favor the A321neo fuel burn. Are the numbers plausible? Probably.

The Problem
Once again we see certain data filings are likely to be faulty or almost certainly unaudited. The problems start at the airline level where they should be checked for plausibility.  When DoT processes the data internally they should also have checks.  The T2 data is derived from T-100 traffic data which is a major dataset.

We have been exchanging emails with DoT since 2019 about the data source items that looked odd. For example, we had a question about Spirit’s data dated 2019 and asked “Looking for Spirit Air Lines A320neo data and I see there is nothing for 2018 or 2017. Yet the airline took delivery of their first A320neo in 2016“. In April 2019 DoT advised us, “Spirit has recently filed revised data that includes the A320neo. The updated numbers will be available on the BTS website on April 15 at 11 a.m.” So there is history with Spirit’s data filing.

To their credit, communications with DoT have been unfailingly courteous. Several times we had to correct our analysis to report accurately. DoT interlocutors were helpful to enable this.

Bottom Line
The DoT data remains the most granular available and that makes it very useful. But there is a case for better data quality control improvements. As noted airlines have staff changes and the filing process likely is onerous. Pulling all the numbers together at an airline might be tedious – and analysts showing promise likely jump at the first opportunity to get into real operations.  Given the process, rules-based automation is probably the way to go. 

But automation makes “data creativity” much more difficult.  And there is data creativity at play as we showed in Part 1.

The airline industry and its supply chain are under rising pressure to show better green credentials. Improving fuel burn is a key metric.  The newer models from Airbus and Boeing are showing this clearly.  But we would all feel so much better about the numbers if there weren’t these niggling issues.

Please follow and like us:
Pin Share

Co-Founder AirInsight. My previous life includes stints at Shell South Africa, CIC Research, and PA Consulting. Got bitten by the aviation bug and ended up an Avgeek. Then the data bug got me, making me a curious Avgeek seeking data-driven logic. Also, I appreciate conversations with smart people from whom I learn so much. Summary: I am very fortunate to work with and converse with great people.

%d bloggers like this: