Do windows on the side of the fuselage increase drag?

I wonder this because I'm not too sure glass is easier for air to travel over than metal.

The long length of an airliner fuselage ensures turbulent flow at the station of the cabin windows, so their only contribution to friction drag are the gaps between metal and polycarbonate. The transition from a laminar to a turbulent boundary layer happens before the windows start. In Sighard Hoerner's classic Fluid Dynamic Drag we find in Chapter V these figures on the drag of surface imperfections:

The size of the gaps and the dynamic pressure at their location drives their drag contribution. Let's assume the windows are 0.4 m high, the grooves 2.5 mm (0.1") wide and shallow enough to have a drag coefficient of 0.008. Let's further assume we talk of a Boeing 747-200 with 100 such windows per side and restrict the drag calculation to the crossflow over the vertical part of the gaps. The full opening area of all gaps is $2\cdot100\cdot2\cdot0.4\cdot0.0025$ = 0.4 m² or 0.078% of the reference area of the 747. Wikipedia gives the parasitic drag coefficient of the Boeing 747 as 0.022, so the windows contribute one threethousandfivehundreth of that drag. Negligible, in other words.

But there is another source of drag: The windows are heavier than the metal which they replace, and to carry them around will increase induced drag. If we again assume 200 windows for a Boeing 747-200 at 7 kg each (these are from pre-polycarbonate times!), they weigh a total of 1.4 tons. This is 0.8% of its empty mass and equivalent to 12-14 passengers. To find the impact of this mass on a long-range flight over 6000 NM, we can use the Breguet equation; once with and once without the 1.4 tons of additional mass. The empty mass of the Boeing 747-200 plus 50 tons of payload is 228,100 kg, its L/D is 16, the specific fuel burn of its JT-9D-7 engines is $b_f$ = 0.0000184 kg/Ns and the cruise speed is Mach 0.84, which equates to $v$ = 250 m/s in FL 350:

$$m_1 = m_2\cdot e^{\frac{R\cdot g\cdot b_f}{v\cdot L/D}}$$

When we use $R$ = 11,112 km and run this once with landing masses $m_2$ = 228,100 kg and again with $m_2$ = 226,700 kg, the difference in $m_1$ is bigger by a factor of 1.65 (the take-off masses $m_1$ are 376,584 kg with and 374,237 kg without windows), so carrying the windows around costs the Boeing 747-200 911 kg of fuel or 0.61% of the total fuel mass per flight. This is one onehundredsixtythird of the total drag - still not much, but fourty times as much as the absolute contribution of the windows to friction drag (if we assume that the 747 flies at the polar point for optimum range).

This explains why conversions of airliners to freighters almost always replace the windows by metal plates but don't care that much about the gaps they leave. In the case of the 747 pictured below they are certainly big enough to be visible from afar.

A Japan Airlines Boeing 747 passenger-to-freighter conversion featuring aluminum window plugs. (Photo credit: Flickr)

• Nice answer including the relative influences Aug 27, 2015 at 14:44
• It might be better to say NM for nautical miles instead of the symbol for nanometers. Dec 4, 2017 at 14:25
• @Ruslan: Good idea, have updated the answer. Dec 4, 2017 at 15:55

The simple answer is yes, but it is not the material that makes the difference in drag production. The difference between painted aluminum or composite surfaces and glass is negligible.

Where there are windows, there are likely to be seams, transitions and other variations in the skin of the aircraft. These variations can cause turbulence in the air flowing over the skin which can induce drag.

This is why gliders have such smooth surfaces and contours. The airflow remains attached to the surface in laminar flow which minimizes drag. Even bugs stuck to the leading edge of the wing can degrade performance.

• It's worth noting that with modern composites the exterior pane of the windows could theoretically be bonded into the structure, all but eliminating the seams and variations described above. Doing so would however be a major sacrifice in the ease of maintenance, which is why it's typically not done. Aug 26, 2015 at 2:49
• With the size and speed of airliners, laminar flow is restricted to the first few percent of its surfaces. All passenger windows are well within the turbulent boundary layer, and their aerodynamic drag is negligible. Aug 27, 2015 at 15:34