Spec:
Product version:
OPC 10000-13 - UA Specification Part 13 - Aggregates Draft 1.05.2

Accepted fix in 1.05 Draft -- waiting for 1.04 Errata to close.

0005145 negates the need for this fix. Change resolution to no change required. Backed out of 1.05 and removed from 1.04.11 Errata

Current Draft for WorstQuality2 reads
If multiple values exist with the worst quality but different StatusCodes then the StatusCode of the first value is returned. If multiple values exist with the worst quality the MultipleValues bit is
set.

With Raw data as mentioned previously:
[08] 01:17.000Z, 70, Uncertain (0x40000000)
[09] 01:23.000Z, 70, Good (0x00000000)
[10] 01:26.000Z, 80, Good (0x00000000)
[11] 01:30.000Z, 90, Good (0x00000000)

TreatUncertainAsBad = true, and Simple Bounding resulting in data for the interval being

[0] 01:20.000Z, Null, BadNoData (0x809b0000)
[1] 01:23.000Z, 70, Good (0x00000000)
[2] 01:26.000Z, 80, Good (0x00000000)
[3] 01:30.000Z, 90, Good (0x00000000)
[4] 01:36.000Z, Null, BadNoData (0x809b0000)

Worst quality is Bad multiple times

Interval should be

Reopened as per note on January 4 2022

Nothing to change. The suggest value and status for Historian 3 is already the value and status.
12:01:20.000 BadNoData Good, Calculated, Partial, MultipleValues

This Mantis is still open to make sure the CSV export contains the row:
12:01:20.000 BadNoData Good, Calculated, Partial, MultipleValues

See the Note: https://mantis.opcfoundation.org/view.php?id=6160#c15668
TreatUncertainAsBad = true, and Simple Bounding resulting in data for the interval being??
[0] 01:20.000Z, Null, BadNoData (0x809b0000)

According to the specification - 5.4.3.36 WorstQuality2:
The last statement in the first paragraph: The server shall ignore the TreatUncertainAsBad for this aggregate.
Therefore, the Raw data with Uncertain StatusCode is NOT Bad.

If you consider the above statement,

if the value before timestamp is non-Bad then it is the bounding value → YES (70 - Uncertain)
If the value before the timestamp is Uncertain the StatusCode is Uncertain_DataSubNormal. → YES (Uncertain)
The Simple Bounding resulting in data for the interval
01:20.000Z, 70, UncertainDataSubNormal (0x40a40402)

V1.05 Draft 7:
5.4.3.35 WorstQuality
The WorstQuality Aggregate defined in Table 46 returns the worst status of the raw values in the interval where a Bad status is worse than Uncertain, which is worse than Good. No distinction is made between the specific reasons for the status. The server shall ignore the TreatUncertainAsBad for this aggregate.

V1.04:
https://reference.opcfoundation.org/Core/docs/Part13/5.4.3/
5.4.3.35 WorstQuality
The WorstQuality Aggregate defined in Table 46 returns the worst status of the raw values in the interval where a Bad status is worse than Uncertain, which is worse than Good. No distinction is made between the specific reasons for the status.

The text: "The server shall ignore the TreatUncertainAsBad for this aggregate" was added in V1.05.
If we don't consider the text, the current values are correct --> Remove the text from Draft, then the TreatUncertainAsBad is to be respected for the calculations.
If we need to consider the text, then should revise the calculations as mentioned in https://mantis.opcfoundation.org/view.php?id=6160#c16416

Changes in Part 13 1.05 draft 8

Changes are made as per the comments and recalculating the values.

Errata added to 1.04 draft 11

Agreed to changes in telecon. Errata entries will appear in 1.04.12 (not 11).