42

NASA was using U.S customary units (inches, feet, nautical miles, pounds, tons, US gallons etc.) during the Mercury and Apollo programmes, and beyond.

There are significant disadvantages to using U.S customary units, the most obvious being that the ratios are not uniform or round numbers (e.g. 6076.12 feet in 1 nautical mile).

Considering that much of the maths involved was done by hand, this seems like an unnecessary complication. Why not just use metric? Metric was already widely used by scientists and some engineers anyway.

DrSheldon
  • 47,946
  • 12
  • 162
  • 329
user
  • 1,663
  • 2
  • 8
  • 18
  • 1
    Related: Did NASA use metric units for the Mercury missions? – DarkDust Jul 23 '19 at 10:19
  • 38
    Imperial units were used in aviation industry before and this industry was needed to built rocket parts. Astronauts were military and test pilots before, they were used to altimeters calibrated in feet. Aviation is still not full metric. – Uwe Jul 23 '19 at 10:54
  • 26
    @Uwe, I think the OP was saying that in metric the ratio between any two units is almost always a power of 10, while in American, the ratio between any two units is almost never anything obvious or memorable. Conversion within metric conversion is trivial, while conversion within American is complicated and potentially error-prone. ("How many litres in a cubic metre?": 1000. "How many cubic feet in a gallon?": "Is that a dry gallon or a fluid gallon?") – Ray Butterworth Jul 23 '19 at 12:47
  • 2
    Given the types of things being calculated, I hardly think being able to easily convert between meters and kilometers would be the biggest source of errors. Round constants of proportionality go out the window once you have start working with values like g, no matter what units you are using. – chepner Jul 23 '19 at 18:43
  • 1
    @Ray Butterworth: Metric has its own set of confusions. It's pretty darned easy to confuse a prefix (or abbreviation: for instance, I can never keep the abbreviations for milli and micro straight), or accidentally get your calculations off by a power of 10.,,, – jamesqf Jul 24 '19 at 03:31
  • 2
    I work at NASA. We still use all manner of units. It's relatively easy to convert, and so many things rely on historical data which is all recorded in imperial. Many US manufacturers (read: the contractors who build stuff for us) still mostly work in imperial. It would be great if we could only use metric, but we have to use imperial because we're in the USA. Any engineer in the world will almost be guaranteed to work with imperial units at least once in a while. – Aaron Jul 24 '19 at 21:16
  • Comments are not for extended discussion; this conversation has been moved to chat. – called2voyage Jul 29 '19 at 12:54

3 Answers3

81

The Apollo Guidance Computer did use metric/SI units internally for its calculations. But it converted to imperial/USC units when it displayed data on the DSKY. This is probably because the Apollo astronauts (mostly trained as test pilots) had an intuitive "feel" for imperial/USC units.

Although data was stored internally in metric units, they were displayed as United States customary units - Wikipedia

.

The computer display readouts were in units of feet, feet per second, and nautical miles – units that the Apollo astronauts, who had mostly trained as US Air Force pilots, would have been accustomed to using. Internally, however, the computer’s software used SI units for all powered-flight navigation and guidance calculations, and values such as altitude and altitude rate were only converted to imperial units when they needed to be shown on the computer’s display. - UK Metric Association

JFL
  • 109
  • 4
gandalf61
  • 1,421
  • 7
  • 12
  • 14
    That's fascinating. I wonder how much of its limited computing power and memory was wasted on those conversions? – user Jul 23 '19 at 12:51
  • 41
    @user Well, it just involves storing a handful of constant scale factors in ROM and a single memory lookup and multiply or divide each time a value is written to or read from the DSKY. Compared to the really complex stuff that the AGC had to handle (real-time interrupts, task scheduling, hot restarts, multiple 3D reference frames, Kalman filters ...) it's not a big overhead. – gandalf61 Jul 23 '19 at 13:58
  • 5
    WP says 96 cycles of a multiply instruction, not actually that bad. – user Jul 23 '19 at 14:35
  • 35
    Internal storage was almost certainly in some manner of small units unsuitable for human consumption. Given that computers are going to do math to display it for the humans no matter what (the only thing trivial for them is multiplying or dividing by 2), there is essentially no difference to the computers what multiplication factors are used. Powers of 10 are only easier for humans – T.E.D. Jul 23 '19 at 17:52
  • 13
    @user, almost none. Since the physics calculations were done 100 times per second, almost every rate was measured "per centisecond". Additionally, many numbers (eg. "distance to Earth") had binary scaling factors to keep the numbers within reasonable bounds. Because of this, even a metric display would require conversions. You can see this looking at the variable definitions in http://www.ibiblio.org/apollo/Documents/j2-80-MSC-70-FS-2_text.pdf – Mark Jul 23 '19 at 21:41
  • 2
    @gandalf61 Division on such a computer is a very expensive task. Interrupts on the other hand mostly trivial to implement and quick in execution. – asdfex Jul 24 '19 at 08:52
  • @T.E.D. While it's true that powers of 10 are not easier for computers than powers of 2, being able to say: $x\ m/s^2 \cdot y\ kg = xy\ N$ is much better than $x\ ft/s^2 \cdot y\ slug = xy \cdot ???\ lb$. The powers of 10 part is only relevant when using the prefixes, which are often only for display purposes. – tjjfvi Jul 24 '19 at 16:12
  • 2
    @tjjfvi It doesn't matter what units you're using there, as long as they're consistent. If you insist on ft/s^2 and slugs, then you're using ft-slug / s^2 for weight. Wikipedia quotes a 1928 college physics book as dismissing the need for a name of the unit of mass; it's just "in lbs. per ft./sec.2 or in grams per cm./sec.2." – prosfilaes Jul 24 '19 at 16:48
  • @prosfilaes True, but for those implementing the system it is easier to implement an existing standard and do sanity checks off of that, rather than choosing 7 base units and going from there. Since you are going to be converting it for display anyway, you might as well pick a system that's more standard and therefore less bug/miscalculation prone. – tjjfvi Jul 24 '19 at 16:54
  • 2
    @tjjfvi The existing standard in practice was some variant of the Imperial system. For NASA, that probably was more standard and less bug/miscalculation prone. – prosfilaes Jul 24 '19 at 17:03
  • @prosfilaes Given that it was stored internally as metric, it seems that they decided that metric was better. My point was just that the powers of 10 part is largely irrelevant for internal storage. – tjjfvi Jul 24 '19 at 22:53
  • @user it did not really cost them a lot of computing power .... but it did eventually cost them millions of $ (more than a hundred). Mars climate orbiter crashed because of wrong units conversions; the kind of thing that has so much more chances to happen when the unit system is not well standardized and when you try to use units the rest of the world forgot about. And this accident is not unique, it happened only a few years after NASA lost SOHO to .... wrong conversion factor. – m.raynal Jul 25 '19 at 13:29
  • @T.E.D. to bounce on what you said (only division by 2 is trivial for a cpu), actually dividing by small integer values can be done much, much faster than dividing by 'random' float numbers. I'd assume any decent compiler to have really heavy tricks to perform multiplication/division by 10. I'm dubious about the level of optimization of the multiplication/division by 32.174. – m.raynal Jul 25 '19 at 13:47
  • @m.raynal - Not a particularly useful point in this case though, as its a pretty safe bet all such quantities in the guidance computer were stored either in floating point or fixed-point format. – T.E.D. Jul 25 '19 at 13:53
  • 2
    @T.E.D.Multiplying by 10 is trivial if you are using BCD, which a lot of computers did well into the 1970s. I don't know if any of the onboard systems used BCD, although I doubt it -- it was more for accounting than engineering. – Mike Harris Jul 25 '19 at 18:17
  • USC/English units are not the same as Imperial. The quote from the UK Metric Association perpetuates the inaccuracy. See Organic Marble's answer. – Rich Jul 26 '19 at 16:24
37

NASA used English Engineering Units not Imperial units. (This phrasing is a reply to the original, un-edited question title)

They did this because the program was implemented by the US aerospace industry and that industry's industrial base was all in English units. Every manual, tool, data book, milling machine, and fastener used those units.

Conversion of the industry to metric would have taken time and NASA started out with a lot of schedule pressure.

Eventually they came around, the ISS is metric.

Organic Marble
  • 181,413
  • 9
  • 626
  • 815
  • When I earned my aerospace engineering degree in the late 1970s the industry was in transition, we had to deal with both sets of units. I would assume it's mostly metric now with some leftover English stuff in aviation. – Organic Marble Jul 23 '19 at 11:25
  • 1
    "Conversion of the industry to metric would have taken time and NASA started out with a lot of schedule pressure." Time pressure certainly was the case through Apollo. But the Shuttle had a generous development period. Was the factor there that the Shuttle was seen as an aircraft, and therefore be tied to standard aviation units? – DrSheldon Jul 23 '19 at 14:56
  • 4
    @DrSheldon there's a lot of Apollo DNA in Shuttle - look at the controls and displays alone. I don't think the real press to convert to metric in the US started till the Carter administration and Shuttle was approved under Nixon. – Organic Marble Jul 23 '19 at 15:10
  • 6
    Eventually they came around, the ISS is metric. The ISS is also part Russian, part European, part Japanese... the US section of the ISS mixes metric and English units, and NASA generally to this day continues to use a mix of units for other vehicles and missions. They're not done coming around just yet. – J... Jul 23 '19 at 19:14
  • 3
    @OrganicMarble I got my aerospace engineering degree within the last few years and we still had to use both sets of units. Despite what some condescending Europeans would have everyone believe, US Customary units are still widely used in industry. – zaen Jul 24 '19 at 17:17
  • Is the ISS fully metric? Metric fasteners and metric plumbing connections have been an exceeding tough nut to crack. As much as Elon Musk despises US customary units, there are still some aspects of SpaceX's rocketry that use customary fasteners, customary drill bits, customary plumbing connectors, customary milling machines, etc. – David Hammen Aug 04 '21 at 07:31
11

When I was trained in Aerospace Engineering at Texas A&M in the 1980s, we were taught to use what has been referred to as the Gravitational FPS system, where distance is measured in feet, force in pounds, and mass in slugs. This was a big disappointment to me at the time, since I had been quite comfortable with metric units in high school physics. However, at the time FPS still dominated the US engineering literature.

I remember that one semester we had a visiting lecturer in dynamics who was also continuing his work as an engineer at NASA (Johnson Space Center.) He announced to us early in his classes that he would only accept work in FPS, and not "those damned communist units."

When I got to NASA/JSC in 1984 I found that FPS was quite common in most of the Shuttle simulation code that I worked on (I never saw any actual flight code,) but metric units were also used by some teams, and if anyone harbored any political biases around systems of measurement, they kept it to themselves. In almost all cases, however, when data were presented to crew or (especially senior) engineers, they were converted from consistent units (whether SI or FPS) to feet, pounds, pounds-mass, nautical miles, knots, degrees (both Fahrenheit and angular,) and other "traditional" units.

Michael Matthews
  • 119
  • 4
  • 5
    I was a student at Texas A&M in the early 2000s. ENGR 111/112 still heavily used English units (with both pound-mass and pound-force variants), although by then the faculty had at least stopped referring to SI as "damned communist units". – dan04 Jul 23 '19 at 22:16
  • 5
    This is interesting, but it doesn't answer the question "Why did NASA use Imperial units"? Stick around, spend some time on the site, and you will soon be able to comment. – Organic Marble Jul 23 '19 at 23:04
  • 4
    It does, however, give some indication of the culture @OrganicMarble and as such is supplemental information. –  Jul 24 '19 at 06:15
  • 1
    The point being made (I think) is that mostly: groups used consistent base units and stuck with them. The conversion to the most natural or common 'old units' was only an outer wrapper to the heavy lifting. The heavy lifting (code/design etc) might not be too heavily impacted by the full weight of the crazy of the imperial system. (voting to keep). – ANone Jul 24 '19 at 10:04
  • 5
    "those damned communist units."? When the metre convention was signed in Paris on 20 May 1875 by representatives of 17 nations (Argentina, Austria-Hungary, Belgium, Brazil, Denmark, France, Germany, Italy, Peru, Portugal, Russia, Spain, Sweden and Norway, Switzerland, Ottoman Empire, United States of America, and Venezuela) there was no communist system, not in France and not in Russia. French revolution was over, Napoleon dictatorship too, France was a republic and Russia was reigned by the Zar. – Uwe Jul 24 '19 at 10:17
  • 4
    @Uwe: To a certain part of the US population, everything originating in Europe is communist, or at the very least socialist. Their political scale of reference is somewhat skewed that way. – DevSolar Jul 24 '19 at 10:46
  • @JCRM the accepted answer doesn't answer the question either...such is the way of stack exchange :) – Organic Marble Jul 24 '19 at 11:33
  • 2
    The accepted answer is a "frame challenge" -- which is an acceptable form of answer @OrganicMarble, and while it doesn't answer the question title, it does speak to the "why add this complexity to the calculations" from the question itself. –  Jul 24 '19 at 11:39
  • 2
    @DevSolar How often do you visit the US? You're talking about 1% of people... it's negligible. Every country has stupid people who don't understand basic things. – user91988 Jul 24 '19 at 16:48
  • 3
    @Uwe to be clear, I am in no way endorsing the "communist units" view (which I took as hyperbole,) but it was a reminder that for many in that era the moon race was thought of as a proxy battle in the Cold War. – Michael Matthews Jul 24 '19 at 17:06
  • 1
    @only_pro I've got several friends in the states... and we see your election results too. ;-) – DevSolar Jul 24 '19 at 17:19
  • 2
    @DevSolar I don't see what election results have to do with your claims, and I guarantee I have more friends in the States than you. :) – user91988 Jul 24 '19 at 17:20
  • @MichaelMatthews It was clear to me that you are in no way endorsing the "communist units" view. – Uwe Jul 24 '19 at 17:33