How realistic is it? We measured.

Thermiker, a glider simulator that runs in the browser, is checked against real data in two ways. First, each of its five gliders is flown by a scripted test pilot and its polar, stall, roll rate and spin are set against published figures: the 18 m glider's min sink matches its published 0.48 m/s exactly, and the ASK 21's best glide is 34.2 against 34 from the maker. Second, a realism bench scores the 15 m glider 84.1 of 100 against published polars, textbook handling targets and 344 real flight logs. Below are the numbers, how we got them, and where the sim is still off.

The same comparison, with charts, is in the game: Learn tab, "How realistic is it?" (open it).

How we measure

Every sim number here comes from flying the flight model, not from a table. In calm air over flat ground:

  • Polar: a steady straight glide every 2.5 km/h from just above the stall to near Vne. Each point is the mean speed and sink over 12 s, after 8 s to settle.
  • Stall: the speed bleeds off at about 1 km/h per second with the wings level. The stall speed is the lowest speed at 1 g.
  • Roll: from 45 degrees of bank one way to 45 the other at 100 km/h, full aileron and coordinated rudder.
  • Spin: from the stall warning, full back stick and full rudder for 12 s, then the standard recovery. Flown at the standard loading and with the centre of gravity aft.

The published side is a three-point polar from the XCSoar polar store at a reference mass, scaled to the sim's flying mass, plus the maker's own figures where they exist. Our test suite flies all of this again on every run and fails if a result no longer matches the published numbers.

Results per glider

15 m class ASK 21 18 m class K 8 B G 103 Twin II
Flying mass 400 kg 500 kg 380 kg 280 kg 520 kg
Reference polar LS 8 (15 m), 325 kg ASK 21, 468 kg ASW 28-18, 345 kg Ka 8, 290 kg G 103 Twin 2, 580 kg
Best glide, sim 42.2 at 108 km/h 34.2 at 90 km/h 48.2 at 95 km/h 27.5 at 73 km/h 37.4 at 95 km/h
Best glide, published polar 41.6 at 99 km/h 32.8 at 92 km/h 48.9 at 93 km/h 27.2 at 75 km/h 36.5 at 109 km/h
Maker's figures none (a class, not one type) L/D 34, 0.65 m/s L/D 48, 0.48 m/s L/D 25, 0.67 m/s L/D 37, 0.70 m/s (flight manual, 580 kg)
Min sink, sim 0.61 m/s at 83 km/h 0.65 m/s at 73 km/h 0.48 m/s at 75 km/h 0.66 m/s at 60 km/h 0.65 m/s at 78 km/h
Min sink, published polar 0.55 m/s at 68 km/h (extrapolated) 0.68 m/s at 70 km/h (extrapolated) 0.48 m/s at 76 km/h 0.70 m/s at 63 km/h (extrapolated) 0.75 m/s at 89 km/h (extrapolated)
Largest polar gap +0.36 m/s (17 %) at 190 km/h +0.33 m/s (12 %) at 170 km/h -0.02 m/s (2 %) at 123 km/h -0.09 m/s (5 %) at 116 km/h +0.83 m/s (30 %) at 193 km/h
Stall at 1 g, sim 72.5 km/h 64.8 km/h 67.7 km/h 53.3 km/h 69.7 km/h
Stall, reference 69 to 75 km/h (design band) about 67 km/h (flight manual, scaled) not published 52 to 55 km/h 69 to 71 km/h (flight manual, scaled)
Roll 45 to 45 at 100 km/h, sim 3.8 s 4.5 s 4.8 s 3.4 s 4.8 s
Roll, reference 3 to 5 s (design band) 4.0 to 5.5 s (our estimate) 4.2 to 5.5 s (our estimate) not published 4.0 to 5.5 s (our estimate)

"Extrapolated" means the published value lies below the slowest published polar point, so it is a guess too.

Spins

Glider and loading Centre of gravity Result
15 m, 90 kg pilot 270 mm Spins; recovered in 0.16 turn, 154 m lost
15 m, 70 kg pilot 316 mm Spins; 0.10 turn, 142 m
ASK 21, two 70 kg pilots 380 mm No spin
ASK 21, two 70 kg pilots, 2 kg spin ballast 400 mm Spins; 0.45 turn, 139 m
18 m, 110 kg pilot 300 mm Spins; 0.02 turn, 104 m
18 m, 70 kg pilot 405 mm Spins; 0.13 turn, 104 m
K 8 B, 90 kg pilot 264 mm No spin
K 8 B, 70 kg pilot 350 mm Steep spin; 0.15 turn, 84 m
Twin II, two 70 kg pilots 391 mm No spin
Twin II, 70 kg pilot solo 446 mm Spins, 69 degrees per second at 87 km/h; 0.05 turn, 80 m

The ASK 21 flight manual says it spins only with the centre of gravity at 400 mm or further aft, and recovers within one turn. The sim matches that limit. The Twin II flight manual says it spins only with the centre of gravity aft, a turn every 3 s; two-up it stops by itself. For the other types we found no published spin figures. The certification standard for gliders (CS-22) asks for recovery within one more turn, which every case meets.

Where the sim differs most

  • 15 m class, fast end: above about 160 km/h the sim sinks more than the LS 8 polar, 17 % more at 190 km/h. It is also above the other 15 m polars we compare with, so the fast end is too draggy.
  • 15 m class, slow end: min sink is 0.61 m/s at 83 km/h, against 0.55 m/s at 68 km/h on the polar, which is itself extrapolated there.
  • ASK 21, fast end: 12 % more sink at 170 km/h. The two published ASK 21 polars disagree at the fast end too.
  • ASK 21 stall: 64.8 km/h against about 67 km/h from the manual at 500 kg.
  • Twin II, fast end: the sim sinks more than the XCSoar polar, 30 % more at 193 km/h. But R. H. Johnson's flight test measured 1.78 m/s at 148 km/h with two on board, against the sim's 1.50 m/s at 580 kg. So the sim sits between the maker's polar and the measured one.
  • Twin II spin: about 5 s per turn in the sim, against 3 s in the flight manual.
  • Roll rates: no published reference for any type, so the bands are our estimates.

The realism bench

The bench is a fixed set of 41 checks, run on request. Its last run scores the 15 m glider 84.1 of 100.

Part Weight Score Compared against
Physics 30 % 85.7 Published polars of the LS 8, Discus 2 and ASW 28, and the stall speed (10 checks)
Handling 25 % 94.4 Target bands from textbook values, not measured flights (17 checks)
Soaring 25 % 89.2 How the sim's soaring pilot circles, cruises and climbs, against 272 real flights of 15 m, Standard and Club class single-seaters (9 checks)
Tasks 20 % 62.5 Design targets for a competent pilot; no public reference (5 checks)

The real flights are 344 public IGC logs of sailplane flights, most of them from SkyLines; 272 of them are the single-seater classes the 15 m glider stands for.

The weakest checks are honest work to do: the size of turbulence gusts (12.8), tasks finished by the sim's own soaring pilot (18.7), its rate of field landings (22.3), and the glide ratio it achieves between thermals (74.1). The bench flies only the 15 m glider so far.

What we do not measure yet

  • The other four gliders on the bench. They are checked against their polars, stall, roll and spin, but not against real flight logs.
  • The weather against measured soundings day by day. The regions' cloud base and thermal spacing are set from ten summers of ERA5 data and published measurements, not checked flight by flight.
  • Handling against real flight test data. The handling bands are textbook values.

Use the numbers

You may quote these figures with a link to this page. Last checked in October 2026.

Keep learning

The DAeC (Deutscher Aero Club) publishes a free course, the basic training for glider pilots, in German: Segelfliegen Grundausbildung.

Send feedback

Send feedback

What did you notice? A bug, an idea, praise or a question: we read every message.

What is it about?

For a bug: what did you do, and what happened?

Only if you want a reply.

We send your message, your e-mail address (if you give one) and a few technical details (page, language, window size, browser) to Linear, the tool we track our work in. No cookies. More in the privacy policy.