I'm reviewing TVA interval data for a manufacturing client and seeing demand spikes that are physically impossible. The 15-minute intervals show jumps from 850 kW to 3,200 kW and back down to 900 kW in consecutive readings. This facility doesn't have equipment capable of those loads - their main transformer is only rated for 2,500 kVA. Anyone else seen AMI systems report bogus demand readings like this? We're talking about $15,000 in demand charges that shouldn't exist.
AMI interval data showing impossible demand spikes - software glitch?
Gary, I've seen similar issues with Westar Energy's Landis+Gyr AMI system here in Kansas. The meters would occasionally report phantom demand spikes during communication errors or when the cellular modem was resetting. Westar eventually admitted it was a firmware bug that caused the meter to report stored maximum demand values as current readings during specific error conditions. Have you checked if the impossible spikes correlate with communication outages?
Rachel's onto something with the communication errors. I had a similar case with Santee Cooper where their Itron AMI meters would "hiccup" during firmware updates and report random high demand values. The utility's billing system wasn't programmed to validate readings against historical patterns or facility capacity. We ended up getting $23,000 in demand charges reversed after proving the spikes exceeded the customer's connected load by 400%.
Marcus and Rachel, you're both right about the communication correlation. I pulled the AMI system logs and the impossible spikes all occurred during "communication retry" events. TVA admits their Silver Spring Networks system has a known issue where meters report cached maximum values during communication handshake failures. They're working on a firmware fix but it won't be deployed until Q2 2017. Meanwhile, they agreed to manually review and adjust all affected intervals.
This is exactly why I always request both AMI data AND the traditional monthly register readings for comparison. Wisconsin Public Service had a similar firmware bug where their Sensus AMI meters would double-count demand during daylight saving time transitions. The monthly register readings were accurate, but the interval data was wrong. Always cross-check AMI against the mechanical registers when possible - it's saved me thousands in bogus charges.
Dan, that's a great point about DST transitions. I've seen Dominion Energy's AMI system in South Carolina create phantom 25-hour billing days during spring forward, resulting in inflated monthly demand charges. The worst part is most customers never catch it because they only look at kWh usage, not demand patterns. Always audit the number of intervals per month - should be exactly 2,976 intervals for 31-day months with 15-minute data.
George, excellent point about interval counts. I've started building spreadsheet templates that automatically flag months with incorrect interval counts or impossible demand changes between consecutive readings. TVA finally deployed their firmware fix and provided adjusted billing for the past 8 months. Total recovery was $31,400 in bogus demand charges. The key was having detailed documentation of the impossible readings and facility load limitations.
Gary, would you mind sharing that spreadsheet template? I'm dealing with similar AMI data validation issues with Chugach Electric here in Alaska. Their AMI system occasionally reports demand readings in MW instead of kW due to a decimal point error in the data export. A customer with 500 kW demand gets billed as if they used 500 MW - the bills are astronomical until someone catches the error.
Lisa, I'll email you the template. That MW vs kW error is brutal - I can't imagine the bill shock. For everyone else, the template flags: 1) Demand changes >50% between intervals, 2) Intervals >150% of historical peak, 3) Monthly interval counts ≠ expected, 4) Negative power factor readings, 5) Demand >110% of connected load. It's caught dozens of AMI errors over the past year.
This thread is gold! I wish I'd had these validation checks when CPS Energy rolled out AMI in San Antonio. We missed several months of inflated demand charges because I trusted the AMI data was accurate. Now I know better - AMI systems are powerful but the data quality can be inconsistent, especially during the first year after installation. Always validate, especially on high-demand accounts where errors cost thousands.
Excellent discussion everyone. This is exactly why we need to evolve our audit methodologies for the AMI era. The old approach of just checking rate application and demand ratchets isn't sufficient anymore. We need to validate the data integrity itself. I'm working with several utilities to establish AMI data quality standards and validation protocols. Will share updates as they develop.
Randy, that would be incredibly valuable. Having industry-standard AMI validation protocols would help us all catch these issues faster and give us more credibility when challenging bogus charges. Too many utilities still treat AMI data as infallible when we're proving it's not. The technology is great but the implementation and data validation processes need improvement across the board.