Time delays of gravitationally lensed sources can be used to constrain the mass model of a deflector and determine cosmological parameters . We here present an analysis of the time-delay distribution of multiply imaged sources behind 17 strong lensing galaxy clusters with well-calibrated mass models . We find that for time delays less than 1000 days , at z = 3.0 , their logarithmic probability distribution functions are well represented by P ( \log \Delta t ) = 5.3 \times 10 ^ { -4 } \Delta t ^ { \widetilde { \beta } } / M _ { 250 } ^ { 2 % \widetilde { \beta } } , with \widetilde { \beta } = 0.77 , where M _ { 250 } is the projected cluster mass inside 250 kpc ( in 10 ^ { 14 } \textrm { M } _ { \odot } ) , and \widetilde { \beta } is the power-law slope of the distribution . The resultant probability distribution function enables us to estimate the time-delay distribution in a lensing cluster of known mass . For a cluster with M _ { 250 } = 2 \times 10 ^ { 14 } \textrm { M } _ { \odot } , the fraction of time delays less than 1000 days is approximately 3 \% . Taking Abell 1689 as an example , its dark halo and brightest galaxies , with central velocity dispersions \sigma \geqslant 500 \textrm { km } \textrm { s } ^ { -1 } , mainly produce large time delays , while galaxy-scale mass clumps are responsible for generating smaller time delays . We estimate the probability of observing multiple images of a supernova in the known images of Abell 1689 . A two-component model of estimating the supernova rate is applied in this work . For a magnitude threshold of m _ { \textrm { AB } } = 26.5 , the yearly rate of Type Ia ( core-collapse ) supernovae with time delays less than 1000 days is 0.004 \pm 0.002 ( 0.029 \pm 0.001 ) . If the magnitude threshold is lowered to m _ { \textrm { AB } } \sim 27.0 , the rate of core-collapse supernovae suitable for time delay observation is 0.044 \pm 0.015 per year .