Got a client in Cleveland on FirstEnergy Schedule GS-2 and their interval data is showing these crazy 15-minute demand spikes that don't match their actual usage patterns. We're talking 800 kW spikes when their normal peak is around 450 kW. The spikes always happen at exactly :15 and :45 intervals which seems fishy. Has anyone else noticed weird timing patterns in FirstEnergy interval data? Starting to wonder if there's a meter reading issue or if they're doing some kind of load estimation.
FirstEnergy interval data showing weird 15-min spikes - anyone else see this?
Frank, I've seen similar issues with FirstEnergy in Youngstown. The key is to cross-reference those spike intervals with their meter reading logs. Sometimes they use estimated reads when the AMI system has communication issues. Did you check if those exact time stamps correlate with any outages or maintenance windows? Also worth requesting the raw meter data files, not just the billing determinant reports. FirstEnergy has had some AMI deployment issues that create these phantom spikes.
This sounds like classic interval estimation gone wrong. Georgia Power had similar issues a few years back where the AMI system would insert estimated peaks when it couldn't get a clean read. The :15 and :45 timing is a dead giveaway - that's when many utilities run their automated meter reading cycles. I'd file a formal dispute and demand they provide proof those readings came directly from the meter, not from their load research database.
Jim and Derek, you guys nailed it. I got the raw meter logs and sure enough, those spike intervals are marked as "estimated" in the system but somehow that flag didn't carry over to the billing data. FirstEnergy admitted they had AMI communication issues during those periods. They're now recalculating three months of bills. This is going to save my client about $4,200 in demand charges. Thanks for pointing me in the right direction!
Great catch Frank! TVA has been pretty solid on interval data accuracy, but I always cross-check suspicious peaks against the customer's operational schedule. If you're seeing 800 kW at 2:15 AM on a Sunday, that should raise red flags immediately. The timing pattern analysis is crucial - real equipment doesn't magically turn on at exact quarter-hour marks unless it's programmed to do so.
Alabama Power uses a similar AMI system and we've caught them doing load profile estimation without proper documentation. The trick is to always request the "quality codes" or "status flags" that should accompany each 15-minute interval. If they can't provide those codes, that's another red flag. I've found that utilities are much more cooperative when you show you understand their data collection processes.
MidAmerican Energy in Iowa requires all estimated intervals to be clearly marked, but other utilities aren't as transparent. Frank's case shows why we need to be forensic accountants as much as utility auditors. The $4,200 savings is nice, but the real value is establishing that this utility has systematic data quality issues. I'd be checking all my other FirstEnergy clients for similar problems.