Does AF Analyses 2016R2 have limit on nested formulas? For complex formula, it showed Calc Failed. When I changed it to something simpler, it worked.
Hi Jeff,When looking through cases experiencing this known issue, others also had success in seeing the result in the Analyses tab when clicking evaluate, but then the mapped output tags displayed "calc failed".
When clicking evaluate, you're really doing an ad-hoc, client side calculation of the expression. KB01335 might help understanding factors that come into play when the PI Analysis Service is actually evaluating the expression.
It looks like we only have an integrated install kit for PI AF Services 2017, which includes Asset Analytics 2017 in addition to PI AF Server 2017. Therefore, you will be forced to upgrade your AF server if the location you're running the install kit (where analytics is currently installed) is also the same as where the AF Server lives. However, if they're on separate machines, then there shouldn't be a compatibility issue when just upgrading Analytics.
If you need assistance with an upgrade, please contact OSIsoft Tech Support.
It happened when the analyses tried to output formula result to a PI tag. The PI tag showed Calc Failed.
Is it showing the proper value when you evaluate the analysis in the Analyses tab? What is the complex formula that's failing?
Yes, it showed the proper value in the evaluation. The analyses uses the value of an Attribute to write to a PI tag for another attribute. Very basic. The attribute looks like this: F=Fk;P=qYes;Q=qNo;X=Xtp;Y=X;[if Y>= F*X then Q else P]
Each of the parameters above is the attribute that has formula data references. In turn, there are nested formulas in other attributes. One solution is to output the attribute value to a PI tag but that may be the last resort we will go because maintaining PI tags takes a lot of effort.
Based on the information you've provided, I believe you may be experiencing known issue 104269 and you can upgrade to Asset Analytics 2017 to solve the issue. If you do not wish to upgrade, then I would recommend what you've described by outputting the attribute value to a PI tag as an attempt to solve the issue.I hope this helps.
I have seen this KB. Does it imply the attribute value will be fine but time series view will show error?
In our case, the value could not be calculated from the analyses. The Evaluate shows the result.
Do we upgrade AF server only or do we need to upgrade both server and AF client on users' PC? It won't be an easy task.
Thanks so much for your help! We will evaluate the options and choose the one that works best for us.
Just a quick question, all we want to do is to save the value of an AF attribute to a PI tag every 30 seconds, are there any other ways than using Analyses?
Hi Jeff,I'm glad the info helped!I think in this case you do need to use Analyses as you need to use the PI Analysis Service to write back to the PI Data Archive.Best,Emily
I agree with Emily that everything points at that know issue. You would need to upgrade both the PI Analysis Service and the AF Server and the PI Analysis Service requires the AF Server to run the same or newer version.
Thanks to both. I ended up combining formulas into a long one and I was able to output to the tag. Upgrade is a must and will probably be done in the future as we just upgrade to 2016 R2 early this year. Wondering how architecture change is made in 2017 to handle this and if there is ever a limit?
Hi Jeff, re PI Analyses show Calc Failed, I had the same problem myself yesterday and resolved this without an update to software. I was running an analysis that calculated outputs to a pi tag that used another pi tag and formula based attributes in the output expression.
The tag always seemed to backfill successfully but could not update on a live basis despite changing to a natural trigger from a 5min freq.
Instead, I found that switching the inputs of the expression to static values, giving the analysis service a chance to re-cache the new values and reverting to the original referenced attributes forced the analysis to reference the values correctly. Maybe this has been resolved in the updated analysis service software version. Changing formula as you mentioned in the above message may have had the same effect.
Retrieving data ...