EcoStruxure IT forum
A support forum for Data Center Operation, Data Center Expert, and EcoStruxure IT product users to share knowledge on installation, configuration, and general product use.
Posted: 2020-09-08 09:51 AM . Last Modified: 2020-09-08 10:21 AM
Dear Community,
Good day!
Recently, we noticed there is a huge difference of the Usage (kWh) from Energy Cost report generated (July vs August). It was confirmed by user that there were no load removed from existing racks during the period. Please find below the attached Energy Cost report snapshot for three (3) racks usage (kWh) for the month of July and August.
Month of July (01 - 31 July)
Month of August (01 - 31 August)
With reference to the above Energy Cost report, the usage (kWh) for the month of August is significantly lower as compared to the month of July (highlighted in red). The only difference we observed from the report was the indication of asterisk sign * vs ** next to the rack identification (highlighted in orange). Referring to the description in the Energy Cost report, it describes as below.
For the month of August, it seems that measured data was not available for the full time period for the racks (**). However, based on the description, we should expect a higher data values when no measured data is available as the Energy Cost report will be based on estimated load. We have cross-checked the estimated load of the racks (based on adjusted nameplates) is significantly higher than the measured peak data. Thus, would like to check whether anyone have any idea with regards to the findings above?
Additional findings:
When we generated the Energy Cost report for the period of 01 July to 31 August (two months), the usage (kWh) was not compounded but instead showing a low value close to the readings observed for month of August only.
Month of July and August (01 July - 31 August)
To summarize, below are my inquiries:
1. In what situation would the system categorize "measured data not available for a full period" for the racks (e.g. no data from DCE for a prolonged period)?
2. Shouldn't the Energy Cost calculation for usage (kWh) expected to be higher when measured data is not available, where the estimated load is used instead?
3. What could be the reason for the usage (kWh) for the period of 01 July to 31 August not compounded or is lower when Energy Cost report generated for two months instead of individual month?
4. Is there a more detailed description or clarification on Energy Cost calculation than the information that is readily available on our DCIM support page (as per attached link below)?
Referred references:
https://dcimsupport.ecostruxureit.com/hc/en-us/articles/360038075894-Energy-Cost-Report
System | Version |
Data Center Expert (DCE) | 7.6.0 |
IT Advisor (ITA) | 9.0.7 |
Appreciate any help and input with the above.
Cheers,
Nicholas
Posted: 2020-09-08 02:24 PM
Hello Nicholas.
Would your customer be willing to share a backup so we can more closely look at the data its using in each case? It can be difficult to be sure otherwise. But here are a couple of general comments based on what I see below.
As I understand the energy cost report it looks for measured data, then uses estimated load if not present. Based on which it finds it then looks historically based on the time range.
So based on the charts you provided it would appear DCO/ITA didn't have measured data for these racks in July but started to have measured data in August? In most cases I'd expect estimated load values to be higher than measured as most customers do not derate their racked assets in a manner which reflects actual power draw.
One difference to note when measured power is present is that the energycost report is using average power values for a day, not peaks.
So based on the data, its possible the July and August readings could be quite different assuming my comment regarding when measurements started is accurate.
Now, in regards to the July through August chart. Hopefully someone else from the team will comment on my below response. As I know that report it looks for measured, if its not found then it uses estimated load. With August 31st being the end date, this means it would have found measured and then used measured for the entire timeframe as it doesn't mix estimated load and measured data in its report (as far as I know), so I believe the reason the numbers from July through August are so similar to just August is because the report decided to use measurements pulled the data for august and historically used 0's for July since there were no measurements then.
This is admittedly a guess as of now without being able to see the actual data.
Regards
Greg Sterling
Posted: 2020-09-09 05:50 AM
Hi Greg,
Good day! Thank you for the prompt reply.
Well noted on the above and appreciate your detailed explanations. Just to clarify the below if I understand the idea/concept correctly.
Ref | Description | Remarks |
[1] | without asterisk | Indicates that measured data is available throughout the past month and usage (kWh) was calculated based on the measured data (average across the period). |
[2] | with TWO asterisk ** | Indicates that measured data is not available for the full time period, however, usage (kWh) was still calculated based on the measured data available (average across the period). |
[3] | with ONE asterisk * | Indicates that usage (kWh) was calculated solely based on estimated load as there is possibly no measured data available during the period. |
Your explanation for the scenario observed on the Energy Cost report generated from July to August seems logical, assuming ITA uses only either the measured data or estimated load across the period for the calculations.
My team and I managed to retrieve the latest backup for the IT Advisor (ITA) from our customer site today. And YES, we can share the backup files. I have uploaded the backup onto Box and will share the hyperlink to the files with you separately via email. We can discuss further on this once we have gotten a chance to have a closer look at the data as well.
Cheers,
Nicholas
Create your free account or log in to subscribe to the forum - and gain access to more than 10,000+ support articles along with insights from experts and peers.