Is sewer record-keeping adequate?

We were testing whether rainfall drives sewer pump-outs (it doesn’t — see the sludge cost overrun page). To do that we cross-checked the operator’s gauge-reading column on the DEC monthly forms against the NOAA volunteer-station record from a gauge 1.4 miles away. The operator series turned out to disagree with NOAA in ways that can’t be explained by gauge separation. Once we looked at why, the same issue showed up in other DMR columns and in sludge invoices.

Each point below is one day where both the operator and NOAA reported a value. Perfect agreement would put every point on the dashed y = x line. 85% of paired days agree within 0.10″ — but 25days disagree by more than half an inch, and the operator’s annual total is 11% lower than NOAA. The bigger issue is the days where one source recorded a major storm the other missed entirely: almost always because the operator did not read the gauge that day and either left it blank or wrote 0.00.

The November 2025 monthly DMR is the worst case: across every plant filing for that month, every day’s precipitation reads as 0.00″ or blank. NOAA recorded 2.10″ of rainfall over the month with rain on 13 of 30 days. There is no plausible reading of the data in which both records are accurate.

No. We applied a precision-and-clustering audit to every numeric column on the DMR forms across the three operator eras (prior contractor → in-house staff → H2O Innovations). Two flags were tracked:

• Quantized: the column’s last digit is statistically non-uniform — values are clumping at round numbers rather than reading at the precision the form requests. Tested via χ² of last-digit distribution at p < 0.05.
• Clustered: more than 20% of all reported values are the same single number. A flow meter that reads real values doesn’t produce that.

The roll-up across both plants:

The clearest case is daily flow at Village WWTP. In Era A, 150 days produced only 32 distinct flow values, every single one ending in zero in the fourth decimal — i.e., the operator recorded flow only to three decimals (0.001 MGD), where the form requests four (0.0001 MGD = 100 gallons/day). In Era B the same column has 149 distinct values across 150 days. The difference isn’t a meter change; it’s the operator actually reading a gauge that was previously being filled in by routine.

The same pattern shows up in temperature, settleable solids, and pH. The data has been getting cleaner over operator transitions, but Era C still has 4 quantized columns — notably precip_in, which is quantized in all three eras. Every operator we’ve seen so far has mishandled the rainfall column.

On a healthy activated-sludge plant, two operational variables drive everything else: MLSS (Mixed Liquor Suspended Solids — the standing biomass in the aeration tank, target ~3,000 mg/L) and WAS (Waste Activated Sludge — minutes per day the operator is wasting biomass to maintain that target). Without those two numbers, no one outside the plant can tell whether the operator is managing sludge inventory, whether biology is healthy, or whether emergency hauls are catching up on a wasting deficit.

Reporting of these fields tracks the operator transitions:

The DEC form template the village uses (92-15-7, dated 1995) does have MLSS, RAS, and WAS columns — they live on the supplementary Page 4 alongside loadings calculations. Under H2O Innovations those rows are filled in for D.O. (dissolved oxygen) and partially for ammonia, but the MLSS and WAS cells are left blank every month from September 2025 onward.

The four-month window when MLSS and WAS were reported daily (April–July 2025, in-house operator) is itself revealing. Over that window WAS = 0 minutes/day for nearly the entire quarter — biomass was being grown but never wasted. MLSS climbed from ~500 mg/L in April to ~3,000 mg/L by mid-June, then overshot to 4,200–5,080 mg/L in late July before reporting stopped. The first emergency sludge haul (June 26, 2025) falls just before the operator finally ramped wasting from 0 to ~780 min/day on July 1 — a step change consistent with recognizing biomass over-accumulation, not gradual maintenance. Per the May 2025 Delaware Engineering report (wwtp::2053), the Purestream treatment unit at Outfall 1A cannot retain biomass at influent flows above 35 gpm — storms or wet weather flush biology out, and the plant is structurally underaerated. That diagnosis was published; the operational response would have been visible in MLSS/WAS, but reporting stopped before any response could be measured.

The chart on the sewer-flow page (Sludge tank operations & emergency hauls) shows this directly — the MLSS line goes dark on August 1, 2025, and 14 of the 16 emergency hauls in the QuickBooks ledger fall in the dark period.

The QuickBooks sludge-removal ledger covers 29 transactions totaling $53,484 over June 2025 – March 2026. We audited each memo. Findings:

The rate-determination matters: emergency pump-outs are billed at $299/kgal in this dataset; routine pumps at $213/kgal. The classification depends on memo text alone. A memo like “sludge tank — 4,000 gallons” is ambiguous, but if it’s an emergency the village pays 40% more. With 28% of transactions unclassifiable, the village has no audit trail justifying the rate paid.

Across April 2025 – February 2026, NYSDEC issued two consent orders, multiple Notices of Violation (April 2025, August 2025, February 2026), and ran inspections in July and December 2025. The village filed responses to most. Both the NOV cadence and the village’s response cadence accelerated after the April 2025 operator transition — likely because the in-house staff began reporting actual numbers that revealed exceedances the prior records had obscured.

The 4/14/2025 NOV cited exceedances of UOD, ammonia, TSS, DO, settleable solids, and fecal coliform without restating the numeric permit limits. A ratepayer reading the NOV can’t check current readings against permit without consulting the SPDES permit text directly. We surface those limits on the sewer-flow page where the comparison can be made — the village does not publish them in a single accessible place.

A working standard for the records the village should be publishing each month — not as an aspirational target but as a baseline that operators in other jurisdictions already meet:

• DMR fields filled to the precision the column requires. Flow to 4 decimals (0.0001 MGD), pH to 2 decimals, temperature in 0.5°F. No rounded defaults; no blank columns where the operator forgot to read.
• Sludge-process variables reported daily. MLSS (mg/L), WAS (min/day or lbs/day), and RAS belong on every daily form. The columns exist on the village’s current DEC form template; the practice of filling them in should resume. Without those numbers, ratepayers and trustees have no instrument by which to ask “is the biology healthy?” or “is the wasting schedule keeping up with biomass growth?”
• Per-pump-out cause record. One row per event with: which plant was pumped, what triggered the call (process upset / equipment failure / hydraulic load / scheduled maintenance), what was drawn down, EQ tank level before and after, and the actual measured gallons.
• Independent rainfall reconciliation.The operator’s gauge readings should be reported alongside a regional reference (NOAA, the village’s own volunteer station, or a mutual-aid station). Differences larger than 0.5″ should require explanation.
• NOV-to-resolution timeline published.Each NOV should be tracked publicly with the cited limits, the measured exceedances, the response filed, and the resolution date. The village’s WWTP-operations page currently has the documents but not the timeline view a trustee would use to oversee compliance.

None of this requires special equipment or specialist staff. It requires the operating agreement to specify what the operator must record, and the village to publish what it receives.

On DMR data quality, yes. Era A had 10 of 12 numeric columns flagged as quantized; Era C has 4 of 19. The trajectory is in the right direction across each operator transition. But the quality is still meaningfully imperfect — the precipitation column has been wrong in every era, and several columns are sparse where they should be dense (BOD5, SS, dissolved oxygen).

On sludge-process visibility, no. The four-month in-house window (Apr–Jul 2025) was the only period MLSS and WAS were reported daily. Reporting stopped under H2O Innovations even though the same form template carries the columns, and the village lost its operational view into sludge inventory at exactly the moment emergency haul-outs accelerated. This is a regression, not progress.

On sludge invoices, no improvement is visible. The most recent transactions in 2026 still lack plant attribution and cause records, including a $1,280 invoice in February 2026 with an entirely empty memo.

On NOV documentation, the village publishes the documents on the WWTP-operations page but does not maintain a tracking view linking each NOV to its response, resolution, and the permit limits cited. Trustees who want to track the village’s compliance trajectory have to assemble that themselves.