How do iHistorian and OSI-PI UDS use percent good quality data parameter?

[Former Member]

How do the each of the following UDS interfaces use the percent good quality data value in Statistics mode for the Total calculation?

- GE iHistorian versions 3.0

- GE iHistorian versions 3.5

- OSI-PI versions 3.5

If the time window from which we ask for a statistics TOT value contains bad quality data records, the TOT is adjusted down by the percent of time between the bad quality timestamp and the next good quality timestamp historian records.

[Former Member]

Hi,

We cannot tell exactly the how much percent of good quality data these three present.But we can say as below:

Percent Good Quality Data depends on the performance of the OPC server/OPC Client which serves the data to the historian.

Suggested that the use the OPC server of the Which Product is running in plant instead of other third party OPC server available in the market.This also matters the performance of the OPC server pertaining the Good Quality Data.

Again the collection performance of the historian (from collector file) also matters the data quality. GE iHistorian versions 3.5 presents more good quality data due to its good collection performance compared to other you have mentioned.

Hope this will give you some idea regarding the data quality.

-Suresh

[Former Member]

UDS list should read:

- GE iHistorian versions 3.0

- GE iHistorian versions 3.1

- OSI-PI versions 3.5

[jcgood25]

Both the iHistorian UDS and the OSISoft PI UDS work quite similar when it comes to statistics. Both vendor API's support native statistics, but TOT is a derived number with Time Weighted Average as the baseline.

The conceptual formula is:

TOT = TWA * (ED - SD) * TotalizerFactor

TWA

iHistorian calculationmode = Average

PI stattype = ARCAVERAGE

By default in an MII TagQuery the TotalizerFactor is 1 so obviously does not impact this formula, but the ED and SD are in relation to the actual StartDate and EndDate of the query in seconds, so equivalently a simpler formula version would be:

TOT = TWA * (Overall query range in seconds)

So in both cases, the TOT method is mathematical over the whole range of the query request, and from an accuracy standpoint depends upon how the corresponding native TWA method is tabulated by the historian when queried. The UDS gets a single number for TWA and multiplies it by the overall time range (which assumes the whole span).