Sync Time = The base time from which the "Interval" is calculated from. If you have "00:00:00" as your sync time and 1m as the interval, then the dataset will be evaluated at:

00:01:00
00:02:00
00:03:00
etc

This is what you want as you need to keep an eye on the values every minute (certainly for half hourly, maybe for daily you could space out the interval).

The Refresh Interval is how often the dataset will be refreshed and data updated on your display (e.g. on your trend). If you set it to 5 minutes then you will see data update on the display at:

00:05:00 (data from 00:00:00 -> 00:05:00 is updated on the display)
no dataset data updated until...
00:10:00 (data from 00:05:00 -> 00:10:00 is updated on the display)
no dataset data updated until...
00:15:00 (data from 00:10:00 -> 00:15:00 is updated on the display)
etc

Problem you will have is your operators will continually see 4 minute gaps in dataset data then on the 5th minute see the last 5 minutes of data appear, by which time your emissions could have sky rocketed.

My advice would be to move the processing of calculations to the PI server asap as opposed to the client machine.
Or
Implement a switch so you only show the halfhourly or the daily on the display at one time.

You will see a huge difference once the calculations are being performed by performance equations.

Hope this helps.