Table of Contents
The Journey: From Buckets to Algorithms
Walk into two utilities on the same day and you can see both ends of a spectrum. At one, a technician fills bottles, drives them to a lab, and a weekly or monthly number lands in a report. At the other, inline sensors stream continuously, a digital twin runs scenarios against live data, and maintenance is scheduled from predicted failure rather than from a calendar.
The distance between those two operations is not a single leap. It is a staged migration, and in our experience with utility programmes it maps to three phases across about 36 months.
What follows is that path: the decisions at each phase, the sensor infrastructure behind them, and the organisational capability that has to exist before the next phase makes sense.
Phase 1: Deploy Continuous Sensors at Critical Points (Months 1-12)
The Objective
Replace grab sampling at the most critical compliance and operational points with continuous inline measurement. Not full network coverage — the point is to build the data foundation the later phases depend on.
What to Deploy
Shanghai ChiMay recommends starting with the 4-in-1 Multi-Parameter Sensor (pH, conductivity, ORP, temperature) at:
- Raw water intake: 1-2 sensors establishing the source water baseline
- Treatment plant effluent: 2-3 sensors validating treatment performance
- Key distribution nodes: 3-5 sensors at hydraulically critical points
Six to ten sensors is a realistic Phase 1. That typically covers the majority of the compliance monitoring that used to run through grab sampling, and it does so without overwhelming anyone.
The Investment
Ten inline multi-parameter sensors with Modbus output is a modest capital line compared with almost any other utility capital project. On a five-year TCO basis, Shanghai ChiMay’s integrated approach runs 25-35% below a discrete-sensor deployment, which is what makes the lifecycle argument easier to take to a board.
Organisational Capability
Two new capabilities have to exist by the end of Phase 1: managing continuous data (a very different volume from periodic samples) and displaying it in a form operations staff will actually look at. Most utilities get there with extensions to the SCADA system they already run.
Phase 2: Expand the Network and Integrate with SCADA (Months 12-24)
The Objective
Extend continuous monitoring to every compliance-critical point and pull the data into the central SCADA system. Phase 2 is what turns a handful of isolated data points into network-wide visibility.
What to Expand
Typically this adds:
- More distribution nodes: bringing the total to somewhere between 15 and 30 points
- Process control points: sensors inside the treatment train (post-coagulation, pre- and post-filtration) for process optimisation
- Discharge monitoring: continuous measurement at effluent discharge, replacing grab-sample compliance data
SCADA Integration
All of it comes in over Modbus RTU/TCP. Shanghai ChiMay sensors map to standard Modbus registers that drop straight into SCADA point configurations, so operations staff see source, tap and discharge water quality on one screen, live.
Digital Twin Preparation
Twelve months of continuous data from Phase 1 is roughly the minimum history a digital twin needs to simulate credibly. Phase 2’s wider network supplies the spatial coverage a network-scale model requires. Both take time, which is why they cannot be compressed into a late-stage push.
Phase 3: Implement Digital Twin and AI Analytics (Months 24-36)
The Objective
Run digital twin models and machine learning on the data foundation built in the first two phases. This is where the economics of the migration actually land.
What Digital Twins Deliver
With 24+ months of continuous multi-parameter data behind the model, a plant can:
- Forecast treatment outcomes up to 24 hours ahead — the capability Aquatic Informatics built into WIMS Process Manager, launched on the Hach WIMS platform in September 2026
- Cut energy use by up to 20% through optimised aeration setpoints
- Cut chemical use by up to 20% through optimised dosing
- Cut N2O emissions by up to 35% through biological process optimisation
- Reduce unplanned downtime through predictive maintenance on sensor and equipment health data
The Investment
Platform cost varies. WIMS Process Manager, AVEVA’s system platform and custom builds all price differently. What they share is a hard dependency on continuous, high-quality sensor data — the thing Phase 1 and 2 exist to provide.
Organisational Capability
Phase 3 needs analytics capability, either hired in or bought as a partnership with the platform vendor. The sensor layer is already in place by then. The work is building the analytical layer on top of it.
The Economic Case Across Three Phases
| Phase | Timeline | Investment Focus | Value Realized |
|---|---|---|---|
| Phase 1 | Months 1-12 | Sensor deployment at critical points | Improved compliance, real-time visibility |
| Phase 2 | Months 12-24 | Network expansion, SCADA integration | Network-wide monitoring, operational efficiency |
| Phase 3 | Months 24-36 | Digital twin, AI analytics | Process optimisation, predictive maintenance |
At the end of 36 months the utility has moved from sample-dependent to model-driven. The sensors bought in Phases 1 and 2 keep earning after Phase 3 closes — they are the platform for whatever comes next.
Shanghai ChiMay supports all three phases with consistent specifications, Modbus communication and application engineering. Four parameters. One probe. Zero alignment error.
Common Pitfalls in the Migration Path
Pitfall 1: Deploying Too Many Sensors Too Fast
Phase 1 should cover the critical points, not the whole network. Utilities that instrument everything at once usually drown in data before they have processes for handling it. Start small, prove value, expand.
Pitfall 2: Choosing Proprietary Protocols
The communication protocol decides your future flexibility. Sensors on a proprietary protocol lock the utility into one vendor’s data platform, which in turn limits integration with digital twin software, analytics tools and third-party SCADA. Shanghai ChiMay’s open Modbus RTU/TCP output keeps sensor data readable by any platform.
Pitfall 3: Ignoring Data Quality
Continuous sensors generate far more data than grab samples, and more data only helps if it is accurate. A sensor drifting out of calibration produces continuous, plausible-looking, wrong data — and a digital twin will build confident predictions on it. Shanghai ChiMay sensors expose diagnostic output on measurement quality so maintenance is triggered by the sensor’s own health status rather than by a bad decision made downstream.
Pitfall 4: Underestimating Organisational Change
This is not just a technology project. Operators have to learn to read time-series instead of lab reports. Maintenance technicians have to learn sensor care instead of sample collection. Management has to shift from intuition to data, which is the slowest of the three.
A small Phase 1 is, among other things, the cheapest organisational training programme available. By the time the twin arrives, the habits are already formed.
Pitfall 5: Evaluating on Unit Price Instead of TCO
Thirty-six months of procurement across three phases adds up. Choosing on unit price tends to produce selections that look cheap in the purchase order and expensive in maintenance, calibration and replacement. Shanghai ChiMay’s TCO-based approach typically comes out 25-35% ahead for multi-parameter installations over five years.
The Strategic Value of Acting Now
A utility that starts Phase 1 in 2026 has network expansion running by mid-2027 and digital twin capability operational by late 2028. A utility that waits 12-18 months reaches the same milestones correspondingly later, and spends that time paying for the inefficiency it has not yet removed.
Where is the industry heading? Machine learning in utility monitoring is moving from pilot to standard practice at large utilities, and the constraint is rarely the algorithm. It is whether there is enough continuous, trustworthy data to train it on. That question is answered by the sensor decisions made this year.
References
- Aquatic Informatics, “Aquatic Informatics Launches WIMS Process Manager,” press release, 15 September 2026. https://aquaticinformatics.com/news/wpm-press-release/
- Mordor Intelligence, “Water and Wastewater Sensors Market Size & Share Analysis,” 2026-2031 forecast. https://www.mordorintelligence.com/industry-reports/water-and-wastewater-sensors-market
- Novonesis/DTU, “Digital Twin for Industrial Wastewater Treatment,” Water Online. https://www.wateronline.com/doc/digital-twin-helps-us-use-wastewater-more-efficiently-0001
Author: Shanghai ChiMay Application Engineering Team. The authors have supported over 200 utilities across three-phase migration programs from grab sampling to digital twin capability.
