Which fractile set to choose, is entirely up to you. Assume e.g., that you choose the 10%-, 50% and 90%-fractiles as the low, medium and high estimates respectively. For a specific 窶彭ata node窶, say 窶廴yNode窶 you can then read its actual fractile values. E.g.,:
窶「 Lo.fract. = 0.8
窶「 Md.fract. = 1.01
窶「 Hi.fract. = 1.29
This means that (based on the simulation results) that the probability that the value of 窶廴yNode窶 is below 0.8, 1.01 and 1.29 is 10%, 50% and 90% respectively.
The interval between the Lo.fract. value and the Hi.fract. value is often referred to as an 窶彿nterval estimate窶 of the 窶彭ata node窶. With the fractile set used above, this interval estimate has a probability of (90% - 10%) = 80% of containing the value of 窶廴yNode窶. If you want a higher probability, you must choose another fractile set. In particular, if you want an interval estimate with say a 90% chance of containing the value of 窶廴yNode窶, you can change the fractile set so that Lo.fract and Hi.fract. are respectively the 5%- and the 95%-fractiles.