Have a look at Littlefuse's Fuseology application guide where much of this is discussed.
The graph Y-axis is in seconds. 10,000 s is < 3 hours. A year - and I would be selecting fuses for life measured in years rather than hours - is about 32 Ms so if you extend the Y-axis by three decades (100k, 1M, 10M) and extrapolate the curves you will be close to the rated current.
Relevant highlights from the Littlefuse application guide:
CURRENT RATING: The nominal amperage value of the
fuse. It is established by the manufacturer as a value of
current which the fuse can carry, based on a controlled set
of test conditions (See RERATING).
Under the heading of Nuisance Tripping we can read:
For example, one prevalent cause of nuisance opening in
conventional power supplies is the failure to adequately
consider the fuse’s nominal melting I2t rating. The fuse
cannot be selected solely on the basis of normal operating
current and ambient temperature. In this application, the
fuse’s nominal melting I2t rating must also meet the inrush
current requirements created by the input capacitor of the
power supply’s smoothing filter.
The procedure for converting various waveforms into I2t
circuit demand is given in the FUSE SELECTION GUIDE.
For trouble-free, long-life fuse protection, it is good design
practice to select a fuse such that the I2t of the waveform
is no more than 20% of the nominal melting I2t rating of
the fuse. Refer to the section on PULSES in the FUSE
In addition when selecting a fuse you should note:
RERATING: For 25ºC ambient temperatures, it is
recommended that fuses be operated at no more than
75% of the nominal current rating established using the
controlled test conditions. These test conditions are part of
UL/CSA/ANCE (Mexico) 248-14 “Fuses for Supplementary
Overcurrent Protection,” whose primary objective is
to specify common test standards necessary for the
continued control of manufactured items intended for
protection against fire, etc.
It's an interesting subject but a read of the linked article will be worthwhile.