How can the date of lunar eclipses be calculated? Especially without the aid of a computer.
Lunar eclipses are much easier to calculate than solar ones, and it has been done over the ages via a variety of methods. E.g. there is evidence that the design of Stonehenge embodied some aspects of calculating eclipses.
If we can assume that you have information on previous lunar eclipses at hand, you'll observe patterns. One of the most useful for this purpose is the pattern of near-repetition of eclipses after a Saros (Wikipedia), which is a period of approximately 6585.3213 days, or nearly 18 years 11 days. One saros after an eclipse, the Sun, Earth, and Moon return to approximately the same relative geometry, and a nearly identical eclipse will occur, in what is referred to as an eclipse cycle. That is because a saros is a nearly whole number of repeats of each of three cycles of the lunar orbit: the synodic month, the draconic month, and the anomalistic month. See the article for details.
After decades of observation (or via access to tables like the List of Saros series for lunar eclipses or the Five Millennium Canon of Lunar Eclipses: -1999 to +3000), you'll have a list of the currently active saros series. It is then easy to extend any particular saros series by hand, using the information in the saros article. If there is a particular date you're interested in, we could work through an example here.
Perhaps this pdf will help. Which starts:
The end of the pdf warns: