Normal distribution with known
population mean,
mu = 10.9
population standard deviation
sigma = 2.4
Normal variate z has a value of
z = (x - mu)/sigma
x = measurement
We seek the z value for which only 2.8% = 0.028 exceeds the values from this distribution.
From a table of cumulative normal distribution values F(z)
(areas under the frequency distribution curve) we want the fraction
0.028 to be in the tail, where
z = 1.91 = (x - 10.9)/2.4
x = (1.91)(2.4) + 10.9 = 15.5 years
for the warranty.