High-strength mixes are commonly proportioned to meet their design strengths at 56 or 91 days instead of 28 days, since the full strength is ordinarily not needed as early as 28 days. Indeed, after 28 days, it was noticed that the compressive strength keep increasing. So at 91 days for example, the increase is up to 20%. The splitting tensile strength is even higher. Often we test at 28, 56, 91 (multiples of 7).
Please check the pdf you sent before. I think it is not the file you mentioned. It happened all the times if you are using Windows. Sometimes the O/S changes the filename when you copy it. :)
In my oppinion the real and neccesary age for any concrete is 28 days, I thuk that to stablish the compression resistance to 56 or 91 days is only to quarantir this result, But the normal consideration is to 28 days.
It was not mentioned here that the period after which we should test the strength of concrete depends on the type of cement used. Some of them, for example those with significant share of ground granulated blast-furnace slag, develop their strength slower. In those cases the 28-days strength is not representative.
You may use 90 day strength, but be aware that the whole age adjustment is assumed to be for "Normal" cements (Old Type A).. The use of slags & etc changes the rates of strength gain. But so does the use of accelerators, higher or lower temperature curing, retarders and so forth. To do justice to your mix, you should adjust curing and age of test depending on all these parameters. To get an earlier idea of the strength at any given age, you can plot 1/strength against 1/time for constant temperature curing. Ages of 3, 7 and 14 days will provide a very good estimate right our to 1 year - but at least 3 samples for each age should be used.
If you are using high temperature acceleration (e.g. steam curing), then you should use Arrhenius Equation and add a temperature factor into it. For ease, you can use maturity meters to check the in place curing, or use in place temperature gauges and a plotter (digital or analogue) to get the appropriate real curing.
There are plenty of papers on this, including C Rajesh & NJ Carino, Malhotra M, N J Carino, and NJ Carino & RC Tank.