InfrastructureIssue 044 records
Utility Management
A small edition. Ohio's electric co-ops have organized decades of outage data into a dashboard and say AI may come next, with engineers and lineworkers still making the decisions. Stadtwerke Offenbach in Germany switched on an AI voicebot for routine calls that hands complex cases to staff and warns of early errors. K-Water reports large digital-twin and autonomous-treatment programmes, but its savings figures conflict and do little for cost recovery. A quasi-experimental study of 164 French wastewater plants puts AI aeration savings at 5.4% of electricity, a conservative benchmark. US gas, billing and OT security evidence remained thin for the date.
- Evidence records
- 4
- Cross-source patterns
- 2
- Evidence classes
- 1 public-sector association guidance1 independent reporting1 vendor claim1 academic research
- Outcomes
- 2 emerging1 mixed1 effective
- Source freshness
- 3 new this fortnight1 older, newly relevant
- Research completed
- 2026-11-03
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Utilities are scoping AI to routine or advisory work and keeping people as the decision point
Ohio co-ops say AI may inform but will not decide line rebuilds, equipment replacement or worksite safety, and their first step is an outage-history dashboard. Stadtwerke Offenbach limits its voicebot to standard questions, hands complex cases to staff and says humans remain clearly better on them. Neither reports measured results yet.
Operating questionFor each proposed AI use, which decisions or call types are in scope, who takes over when the AI is out of scope, and how will that hand-off be measured?
Supporting evidenceOhio Rural Electric Cooperatives (Ohio Cooperative Living); Lorain-Medina Rural Electric Cooperative, North Central Electric Cooperative, Mid-Ohio Energy CooperativeStadtwerke Offenbach (municipal utility holding), Stadtservice business unit
Treatment-plant AI savings look smaller under causal measurement than in operator-reported pilots
K-Water's Hwaseong pilot is reported to have cut power 10% and chemicals 14%, with network-wide savings figures that differ by source and no method given. A quasi-experimental study of 164 French plants estimates a 5.4% electricity reduction from AI aeration control, using roll-out timing and AI interruptions as controls. The two are different systems and settings, but together they suggest planning on single-digit savings and measuring them.
Operating questionBefore committing to AI process control, what baseline, control periods or comparison plants will the utility use to verify savings, and who signs off on the result?
Supporting evidenceKorea Water Resources Corporation (K-Water), as analysed by Our Future Water IntelligenceCollège de France, ENS-PSL, Sciences Po (OFCE), ETH Zurich, INSEAD; data from Veolia Water France and Purecontrol
Evidence ledger
Full record · every source keeps its link and limitations
Co-ops in the age of AI: Ohio cooperatives build an outage-history dashboard and set human-judgment limits before AI
Lorain-Medina Rural Electric and North Central Electric have built a dashboard that organizes decades of historical outage data so employees can spot trouble areas and choose between capital rebuilds and cheaper fixes such as rescheduling vegetation work. The article presents the dashboard as a data foundation and AI as a possible next step for finding patterns across records and regions. Co-op leaders say AI will not decide which line to rebuild or whether conditions are safe; engineers, safety staff and lineworkers keep those decisions. Cybersecurity, data governance and human review are named as essential. Safety observations and near-miss records are named as a possible future AI use. Ohio's co-ops have also put data centers in a separate rate class so their grid costs are not shared with existing members.
Why it matters, evidence and limitations
- Why it matters
- A current, plain-language account of how small distribution co-ops (Mid-Ohio serves about 8,000 meters) are sequencing AI: organize the data first with a dashboard, keep decisions with people, and set data governance before tools. This is useful for co-op and small municipal utility conversations because it shows where they actually are, which is before AI, not in production.
- Evidence and measured results
- Member-magazine feature by the statewide association, quoting co-op CEOs. No AI system is described as deployed; the dashboard is presented as data organization that AI could later build on. No metrics, baseline or evaluation are given. The data-center rate treatment is a policy statement, not an outcome.
- Limitations and uncertainty
- Advocacy-toned member publication; no deployed AI, no measured results. The case for including it is that it was published on the edition date and shows where small co-ops really are on AI. It should not be cited as evidence that AI improves co-op reliability.
K-Water's digital twin now covers 56 dams and weirs; autonomous operation at 43 treatment plants reported to save about KRW 11 billion
By 2025 K-Water's digital twin covered 56 dams and weirs and could test up to 48 release scenarios at once (8 rainfall cases by 6 release options). Before the 2025 flood season K-Water cleared more than three times its 2.18 billion cubic metres of design flood-control space at 20 multipurpose dams, and when up to 519 mm of rain fell on 17 to 19 July 2025, 18 dams held the inflow without gate releases. Autonomous operation runs at 43 multi-regional treatment plants, with about KRW 11 billion of reported operating savings in 2025; the Hwaseong pilot (260,000 cubic metres a day) cut power use 10% and chemicals 14%. The analysis notes that savings figures differ by source (the CEO cited about KRW 9.4 billion a year) and that the savings are small against KRW 4,451.5 billion revenue, with bulk water recovering only 78.1% of cost in 2025.
Why it matters, evidence and limitations
- Why it matters
- K-Water has just signed an AI cooperation MOU with Austin Water (archived in this stream), so independent context on what K-Water's systems have and have not delivered is directly relevant to US utilities considering similar digital twins or autonomous treatment. It also shows that operational savings do not fix a utility's cost-recovery gap.
- Evidence and measured results
- Independent analyst summary of figures reported by K-Water and Korean government rankings; the full analysis is a paid report that was not opened. No method is given for the savings, and the analyst itself flags inconsistent figures. The flood outcome is an operational account, not a controlled comparison, and is partly attributable to pre-season drawdown.
- Limitations and uncertainty
- Self-reported savings relayed by an analyst; inconsistent figures; national utility with a different regulatory and scale context from US municipal systems. The full report ($499) was not accessed. Published 31 Oct, inside the window.
Stadtwerke Offenbach's city service launches an AI voicebot for routine customer calls, with hand-off to staff
From 2 November 2026 an AI voicebot answers calls to the Stadtservice of Stadtwerke Offenbach. It understands speech and answers frequent standard questions in German and English, drawing on the service's website content and information it was specifically given. Complex questions, technical problems and cases the bot cannot handle are passed to staff during opening hours; outside hours or when all agents are busy the bot records the request for a callback. The managing director says human staff remain clearly superior on complex requests. The service starts with a learning phase, and callers are warned that not all requests may be answered reliably at first.
Why it matters, evidence and limitations
- Why it matters
- Utility billing and customer-service AI has been a recorded gap in this stream. This is a municipal utility putting a voicebot in front of callers with a defined scope (FAQs, two languages), a hand-off rule and an open admission of early unreliability, which is the pattern US municipal utilities are weighing for billing and service lines.
- Evidence and measured results
- Operator press release; no vendor, call volumes, containment rate, accuracy or satisfaction data are given. Classified as a claim because it is the utility's own announcement without evaluation. Event date is the go-live (2 Nov), after the edition date.
- Limitations and uncertainty
- German municipal utility; no outcome data; announced before go-live. Stadtservice covers city services broadly, so not all calls concern utility billing. Published 29 Oct and not in the archive; included because it is the newest municipal utility customer-service AI deployment found and fills the billing/service gap.
Green AI in Industry: quasi-experimental evidence that AI aeration control cut wastewater plant electricity use 5.4%
Using staggered roll-out across more than 200 plants and unplanned interruptions of AI control, the authors estimate that full-time AI aeration control reduced electricity use 5.4%, carbon emissions 6% and energy spending 8.2%, while effluent quality improved. Electricity used by the AI computing represents less than 1% of the savings. AI-equipped plants were more resilient under extreme weather and pollutant peaks and shifted load from peak to off-peak hours. The analytical sample is 164 plants (139 AI-managed by the end) observed January 2022 to July 2025. Eligible plants were 10,000 to 50,000 population equivalent using activated sludge; plants volunteered and early ones needed no IT changes.
Why it matters, evidence and limitations
- Why it matters
- Why now: K-Water's autonomous treatment results were reported on 31 Oct (10% power cut at one pilot, self-reported). This paper is the strongest available independent, causal estimate of the same class of AI control, and it gives US wastewater utilities a more conservative planning figure. It is not in the archive.
- Evidence and measured results
- Quasi-experimental panel study (staggered adoption and within-plant interruptions) with operator monitoring data; conference paper, not yet peer reviewed. Data and support came from Veolia and the technology provider. The authors note that external validity beyond these sectors and electricity systems still needs testing; broader DICE welfare figures are extrapolations.
- Limitations and uncertainty
- Published July 2026, so older than the window; included only as the independent counterpoint to K-Water's 31 Oct claims (see sledRelevance). French plants of a specific size and process, run by a private operator under delegated contracts, so transfer to US municipal plants is not guaranteed. Not peer reviewed.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Public-sector association guidance
- Practitioner guidance or an association-supplied case; not independent outcome evidence.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Vendor claim
- A supplier-provided assertion that has not been upgraded to independent evidence.
- Academic research
- Research produced through an academic institution or peer-reviewed venue.