The Challenge
In 2020, Round Mountain Water & Sanitation District completed a $4.9M USDA-funded upgrade from lagoon-based treatment to a modern mechanical wastewater plant. The new facility vastly improved treatment quality, but it created an unexpected challenge: the operators didn't know how to interpret the flood of data from the new SCADA system.
Field Operations Manager Steven Koch explained: "We went from checking lagoons twice a day to monitoring dozens of sensors continuously. BOD, TSS, dissolved oxygen, mixed liquor suspended solids, return activated sludge rates—we had all this data but no context for what was normal versus concerning."
The Near-Miss Violation
In June 2024, six months after the new plant came online, Round Mountain nearly violated their NPDES permit. The effluent total suspended solids (TSS) reading was 28 mg/L—comfortably below their 30 mg/L permit limit. But the operators had no way to know that the TSS analyzer had been gradually drifting upward, reading 3-4 mg/L lower than actual values.
"We trusted that 28 mg/L reading," Steven recalled. "Our permit limit is 30, so we thought we had headroom. But when state inspectors took a grab sample during their routine inspection, it came back at 31.5 mg/L—over our limit. We got a notice of violation."
The violation triggered a mandatory corrective action plan and six months of increased monitoring. More significantly, it shook the operators' confidence. "If we couldn't trust our TSS analyzer, what else couldn't we trust?" Steven asked. "We were second-guessing every number on our screens."
⚠️ The Compliance Context
NPDES permit violations for small utilities can result in fines from $10,000 to $50,000+ per event, mandatory corrective action plans, increased monitoring requirements, and potential EPA oversight. For Round Mountain, the violation also created board-level concerns about operational competence just 6 months after completing a major infrastructure investment.
The Solution: Upstream AI Trust Foundation + Compliance
In October 2024, Round Mountain deployed Upstream AI's Trust Foundation tier specifically to address their data quality concerns. Two months later, they added the Compliance-as-a-Service add-on to automate their increased monitoring reporting requirements.
What Changed Immediately
Every sensor got a trust score. The TSS analyzer that had caused their violation? Within 30 days, Upstream AI flagged it with a trust score of 52%—immediately alerting operators that the readings were questionable.
"The system cross-referenced the TSS reading against influent characteristics, settling time, and return activated sludge rate," Steven explained. "It calculated what TSS should be based on process parameters, then compared that to what the analyzer was reporting. When they didn't match, it flagged the analyzer as unreliable."
Transparent calculations provided context. Operators could click any treatment metric to see how it was calculated. For example, the food-to-microorganism ratio (F/M) showed the formula, source data (influent BOD, MLSS concentration, aeration volume), and whether the result was in the optimal range.
"Before Upstream, I'd see 'F/M: 0.28' and have no idea if that was good or bad," said Operator Jackson Malcolm. "Now I click it and see: 'Optimal range 0.2-0.4. Current: 0.28. Status: Normal. Calculated from: Influent BOD 180 mg/L, MLSS 2,400 mg/L, Aeration vol 45,000 gal.' I understand what I'm looking at."
"The trust scores changed our entire approach. We went from 'hope this reading is right' to 'know whether this reading is right.' That confidence lets us operate proactively instead of reactively."
The Compliance Violation That Didn't Happen
In February 2025, Upstream AI's data quality monitoring prevented what would have been Round Mountain's second NPDES violation in under a year.
The effluent phosphorus reading was 0.8 mg/L—well below their 1.0 mg/L permit limit. But Upstream AI flagged it with a 34% trust score and a critical alert: "Phosphorus analyzer reading inconsistent with chemical dose rates and historical performance. Recommend immediate calibration and grab sample verification."
Operators immediately collected a manual sample and sent it to the lab. The result: 1.12 mg/L—over their permit limit. The analyzer had been reading low due to a calibration drift that would have gone undetected until the next monthly state sample.
"Without that alert, we would have submitted 0.8 mg/L on our monthly report while our actual discharge was 1.12 mg/L," Steven said. "That's not just a violation—that's falsified reporting, which carries criminal penalties. Upstream AI literally saved us from a catastrophic compliance failure."
The Response
Round Mountain immediately increased polymer dosing to reduce effluent phosphorus, calibrated the analyzer, and implemented twice-weekly grab sampling to verify analyzer accuracy until trust scores stabilized above 85%. Within three weeks, effluent phosphorus was consistently below 0.7 mg/L with validated analyzer readings.
"We never reported a violation because we caught it before the reporting period ended," Steven noted. "That's the difference between proactive monitoring with trusted data versus reactive firefighting with questionable sensors."
Compliance Automation Results
Zero manual data entry
Upstream AI pulls all required data directly from SCADA and lab results. Operators review auto-generated reports but no longer spend hours transferring numbers to spreadsheets.
18 hours saved monthly
Operators previously spent ~22 hours/month on compliance reporting. Now it's ~4 hours for review and sign-off. Time saved is reallocated to preventive maintenance.
Proactive compliance alerts
System monitors all parameters against permit limits continuously. If anything trends toward a limit, operators get 5-7 days advance warning to take corrective action.
Board confidence restored
"I can show the board our compliance status in real-time: all parameters green, all reports submitted on time, zero violations. That wasn't possible before," Steven said.
The Audit That Proved It
In April 2025, Colorado Department of Public Health & Environment conducted Round Mountain's annual compliance audit. The audit included a detailed review of the previous year's monitoring data, reporting accuracy, and quality assurance procedures.
The inspector's finding: "Exemplary documentation and data quality controls for a utility of this size. The automated compliance system with transparent calculations and data quality scoring represents a best practice we rarely see in small facilities."
Round Mountain's corrective action plan from the June 2024 violation was formally closed with a commendation for implementing advanced quality controls.
"The best compliance strategy is not needing to explain violations—it's preventing them entirely. Upstream AI moved us from defensive compliance to confident compliance. We know our data is good, we know our reports are accurate, and we sleep better because of it."
ROI Analysis
Annual Cost: $9,000 ($600/month Trust Foundation + $150/month Compliance add-on)
Documented Value (First 8 Months):
• Avoided second NPDES violation (potential fine): $25,000-50,000
• Avoided corrective action plan extension: $5,000 in engineering/consultant fees
• Operator time saved (18 hrs/month × $40/hr × 8 months): $5,760
• Avoided grab sample verification costs (reduced from 2×/week to 1×/week): $2,400
• Early closure of corrective action plan (saved 6 months of increased monitoring): $3,200
Conservative documented value: $41,360
Return on Investment: 459% in first 8 months
"The $9,000 annual cost is nothing compared to what a second violation would have cost us in fines, consultant fees, and reputation damage," said District Manager Dave Schneider. "Plus, the board loves seeing zero violations and automated reporting. It's a governance win as much as an operational win."
Lessons Learned
New infrastructure needs new operational approaches. Round Mountain's $4.9M plant upgrade brought modern treatment capability, but the operators needed modern operational tools to match. Data quality monitoring isn't optional—it's essential when managing complex treatment processes.
Trust scores prevent violations. The value isn't just knowing that a sensor is bad—it's knowing BEFORE you make decisions based on bad data. Proactive data quality monitoring catches problems before they become compliance events.
Automation reduces human error. Manual compliance reporting creates opportunities for transcription errors, missed deadlines, and calculation mistakes. Automation eliminates these risks while freeing operators for higher-value work.
Transparency builds confidence. Operators, managers, and board members all benefit from seeing the logic behind numbers. When everyone understands how values are calculated and whether they're trustworthy, operational decisions improve at every level.