When a factory in Thailand introduces a cooling tower chemical dosing pump monitoring system, a pump-run contact is only the beginning. A command can be present while an empty tank, suction air, a failed check valve, a blocked discharge line or an injection-point problem prevents the chemical from reaching the water. The reverse problem also matters: a wiring or meter configuration error can hide unintended dosing. This guide focuses on checking the command against evidence of actual delivery. It explains measurement points, alarms, an RFP and FAT/SAT tests. A water-treatment specialist and the facility’s water management program must determine the chemicals, dose, control limits and sampling procedure for the specific site.
Key decisions for a dosing pump monitoring project
- Distinguish the dosing command, pump operating status, delivery evidence and water-side result. They answer different questions.
- Select flow verification against the chemical, low flow rate, pulsation, back pressure, maintenance capacity and required detection time.
- Treat a missing sensor value as delivery unverified, not as normal operation.
- Specify fault behavior, responsibilities, logs and acceptance criteria in the RFP before selecting a dashboard.
- During FAT/SAT, simulate no delivery, an empty tank, wiring failure, communications loss, manual operation and restart, as well as normal running.
The CDC identifies sediment and biofilm, temperature, water age and disinfectant residual as important cooling tower factors, and recommends monitoring and automated adjustment of disinfectant residual within a water management program. The US EPA describes a broader treatment program covering pH, alkalinity, conductivity, hardness, microbial growth, biocide and corrosion inhibitor. These are design references, not Thai statutory limits. See the CDC cooling tower module and the EPA WaterSense cooling tower guidance.
How this differs from general cooling tower water-quality monitoring
The existing guide to cooling tower water quality monitoring covers make-up water, blowdown, conductivity, sampling and control as a whole. The guide to factory water usage monitoring explains how to place meters between incoming water and individual users. This article concentrates on the chemical feed train: tank, suction pipe, dosing pump, check and back-pressure devices, delivery verification, injection valve and the cooling-water connection. It asks whether a pump that was told to run actually delivered to its intended circuit.
Scale inhibitors, corrosion inhibitors and biocides may use different injection points, control methods and compatible wetted materials. Document each chemical separately. The supplier’s safety data sheet and site procedures govern handling and maintenance. Conductivity can help describe concentration of dissolved minerals; it does not demonstrate that a biocide was delivered or effective. Pump running time likewise cannot prove the resulting water chemistry. A water-treatment specialist must set the applicable targets from analysis and the treatment plan.

Why a run contact cannot prove delivery
A PLC output says that the control system requested dosing. A pump contact says that its motor or electronics report a state. Neither necessarily proves liquid passed the injection valve. Some digital pumps offer internal flow-management functions; Grundfos describes Flow Control Management for its DDA SMART Digital. External metering monitors are another option, as illustrated by ProMinent’s product range. Their detection mechanisms, operating envelopes and fault outputs differ. Do not assume an internal estimate or stroke confirmation is identical to a calibrated independent flow measurement in every pipe and chemical. Check the actual model documentation and test its behavior under site conditions. Grundfos DDA · ProMinent metering monitor.
Keep four layers of evidence separate: commanded dose, pump state, delivery indication and water-side response. The water-side sample can lag because of mixing, residence time, blowdown and other treatment activity. Relying on samples alone may delay fault discovery. Conversely, a flow indication near the pump cannot rule out every obstruction at a downstream injection nozzle. Matching equipment tags, time stamps and location lets operators narrow the fault without overstating what any single signal proves.
Low chemical flow can sit below the useful range of a conventional pipe meter. Consider a compatible low-flow meter, stroke monitor, pump-internal estimate or tank weight/level trend according to chemical properties, pulsation, pressure, detection time and serviceability. Label each displayed value as measured, estimated or commanded. Record units, scaling, totalizer reset behavior and last valid update. Confirm wetted materials and the method for cleaning and calibrating the device.
Measurement points and control boundaries
Start with tank low level, pump start/stop and fault, dose setting, a delivery indicator, discharge pressure, cooling circuit run state and the injection permit. Where continuous measurement is impractical, a phased design may combine low-level and stroke monitoring with daily independent checks of tank consumption and sampling. The RFP must state the faults that this arrangement cannot detect and the likely maximum discovery delay. A tank sensor requires attention to vessel geometry, foam, sediment, refill events and dead volume. A pressure switch by itself is not a universal diagnosis: suction air, a changing back pressure and valve position may all affect it.
For make-up-water-proportional feed, specify the make-up meter pulse value, missing-pulse behavior, totalizer resets, backflow and communications delay. For timed or residual-based biocide control, specify the sequence, hold time, measurement validity and manual override separately. In all cases, define what happens when the tower, make-up meter, delivery monitor or network stops. A site-approved local interlock should be reconciled with the monitoring software; the monitoring server must not accidentally become the only safeguard.
Turn command-versus-delivery differences into useful alarms
An example no-delivery candidate is: dosing permitted, command present, start-up grace time elapsed, yet no delivery indication. Another is delivery indicated with no command. Both require investigation. A remaining pulse after shutdown, siphoning or maintenance pumping can explain an apparent mismatch, so manual work and alarm suppression must be logged with a start and end time. Whether a fault stops the pump or allows supervised manual operation depends on the chemical and site risk assessment; the facility and treatment specialist must approve it.
An alarm should identify tower and chemical, command and verification values, event time, data quality, first checks and recipient. “Pump abnormal” gives a technician little direction. For example, distinguish an electrical fault from an empty tank and from an unverified sensor. Display the age of the last value. During a communications outage, do not silently carry forward an old reading as current normal flow. Avoid alarm delays that are too short for priming or pulsation, but set the final delay by measured startup and update behavior rather than convenience.

Assumed calculation: checking a daily dosing discrepancy
The following figures are illustrative assumptions, not recommended chemical settings. Suppose the pump is set to 2.0 L/h and is commanded for 6 hours. The theoretical commanded volume is 12 L. An independent delivery totalizer records 8 L over the same time window. The difference is 4 L, or 33.3% of the commanded amount: (12−8)/12×100. Do not immediately call this 4 L of actual underdosing. Check pump calibration, meter range, air in the line, the time base and refill records.
Suppose tank level suggests 9 L of consumption, assuming no refill, leak or temperature-related level shift and a valid tank-volume conversion. It differs from the meter’s 8 L by 1 L. This triangulation points to what should be inspected; it cannot itself determine the true dose or water-quality impact. A water-treatment specialist needs the prescribed water sampling and analysis. If the same 4 L command-to-verification gap occurred on 30 days, it would total 120 L of discrepancy, not automatically 120 L of chemical loss or a monetary saving. A sizing or calibration error could produce the same pattern.
Veolia’s chemical-feed guidance discusses checking actual pump output with a calibration cylinder because output changes with operating conditions such as discharge pressure. Include an installed-condition calibration and repeat it after meaningful changes. Tank level is useful corroboration, not a substitute for this check. Veolia Water Technologies.
A fault table for cooling tower dosing alarms
| Observed condition | Possible explanations | First verification |
|---|---|---|
| Command on, run contact off | Power, interlock, pump fault | Local display, breaker and permit |
| Run contact on, no delivery | Empty tank, suction air, valve or blocked pipe | Level, pressure and safe line check |
| Command off, delivery on | Manual operation, siphon or wiring | Local switch, valves and signal history |
| Delivery on, sample outside target | Mixing, chemical selection, injection point or sample conditions | Treatment specialist review |
| Reading frozen | Sensor or network failure, held last value | Last update, calibration and wiring |
This table is a triage tool, not a remote diagnosis. A common tank feeding several cells creates a further mapping risk: the dashboard might show a healthy flow for a different cell. Walk down the piping and compare equipment tags and drawings. Record planned draining and pump tests so that suppressing alarms does not erase the history needed for a later investigation.
What to put in the RFP
Identify the tower and cells, existing treatment arrangement, each chemical and the reason for the project. State what is reused and who verifies compatibility. Ask for a P&ID, instrument tags, I/O list and a signal register with range, units, update interval, normal and failure logic, and data-quality status. Specify the delivery verification method and its detection limits. List alarms for low level, pump failure, no delivery, unexpected delivery, pressure, manual mode, failed instruments and lost communications. The HMI should show command and delivery on the same time axis, with make-up flow, conductivity and tower status available for context.
Separate the responsibilities. The water-treatment vendor proposes chemistry and operating limits; the facility approves interlocks and fault action; the system integrator implements signal processing, permissions, logging and reporting. For every setpoint change, log old and new values, reason, actor and approval. Include recipients for day, night and holiday alarms, and escalation after an unanswered notification. Request updated drawings, wiring diagrams, calibration records, manuals, a screen list, alarm matrix, I/O test sheets, operator instructions and backup/restore instructions.
Compare quotes by pipework, electrical and network modifications, commissioning, calibration, training, spares and maintenance, as well as pump and sensor price. Product data such as ProMinent gamma/X illustrates why rated capacity and interfaces must be checked against the installed chemical and back pressure. A dashboard cannot replace chemical compatibility, leak control or safe isolation for maintenance.
Site survey before detailed design
Collect a period of operating data agreed with the site—three to four weeks is only an example—covering cooling load, make-up and blowdown, tank refills, apparent consumption, pump run time, faults and water samples. Include night, weekend, startup and cleaning conditions if they matter. Record tank shape, residual volume, refill practice, suction and discharge pipe length, check valves, back-pressure devices and injection point. For duty/standby pumps, check whether the shared delivery meter can identify which machine delivered. Walk the line to confirm that the P&ID matches reality and that the sample location is meaningful after mixing.
FAT and SAT: test failures deliberately
Test three layers of acceptance. At the device layer, confirm pump settings, low-level and fault contacts, and the delivery monitor’s response within its rated range. At the integration layer, trace the same tag and time stamp from PLC command through wiring, network, HMI, notification and event log. At the operational layer, have the responsible person receive an alarm, identify the correct tank and injection point, contact the treatment vendor and record the corrective action. A successful screen demonstration alone does not prove that operators can respond. Each test sheet should include the input, expected and observed behavior, evidence, result and open items. Remove every forced or simulated signal before handover and have two people confirm the live inputs. When relating samples to dosing logs, let the specialist define the sampling time window from pipe volume, circulation and mixing rather than assuming immediate response.
FAT verifies the panel, PLC, HMI and simulated I/O before shipment. Check normal permits, command, run, delivery, totals, time stamps and display. Simulate empty tank, no delivery, unexpected delivery, open-circuit sensor, pump fault, communications failure and power return. Confirm notification, latching, acknowledgement, history and manual-mode recording, not just a colored alarm icon.
SAT checks the installed pump, wiring, pipe, meter and injection point under site safety procedures. If testing with the actual chemical is unsafe, use an approved substitute or simulated inputs and record precisely what remains unverified. Check zero and totalization, tank conversion, installed-condition pump calibration, manual mode and power return. A specialist must define sampling location and timing to judge whether the chemical reached the intended water. FAT/SAT demonstrates compliance with specified equipment and logic requirements; one sample cannot guarantee long-term water quality.
Acceptance limits must come from the project specification, not from generic numbers in an article. Define allowed command-to-confirmation time, missing-data display, notification delivery and log retention in the contract. Record tester, witness, date, tag and configuration version. Keep failed attempts, corrections and retest evidence, with owners and due dates for remaining items.

Daily operation and change control
The opening screen should put tower run state first, then dosing permit, command, delivery indication, recent samples and outstanding corrective actions. Avoid plotting flow and cumulative volume on an unlabeled common axis. Do not interpolate a communications gap into an apparently continuous run. Daily checks compare the command and delivered volume with tank consumption and open alarms; scheduled checks inspect tubing, valves, injection point and calibration status according to the facility’s plan.
A change in chemical, concentration, pump, injection point, sensor, make-up source or tower operating mode triggers a review of prior settings. Link chemical batch, pump calibration, instrument scale, PLC version and water sample to a change record. When the delivery sensor is unavailable, define temporary manual checks and a repair deadline with the treatment specialist and facility manager. A recurring alarm should be grouped for diagnosis without silently deleting or indefinitely suppressing it.
Frequently asked questions
Is a run contact enough for cooling tower chemical dosing pump monitoring?
No. It reports an electrical or equipment state, not necessarily delivery. Consider at least tank level and a suitable delivery indication, and specify how the injection line and water-side sample are checked. State explicitly what a selected monitor cannot detect.
What flow threshold should a dosing pump monitor use?
There is no universal value. Verify the installed pump range, chemical, pulsation, back pressure, sensor minimum flow and acceptable detection delay. The treatment specialist sets dose and water-quality limits from site analysis and the treatment plan.
Which dosing faults belong in FAT/SAT?
Test command without delivery, empty tank, suction air where safely possible, sensor disconnection, unexpected delivery, communications loss, manual mode and power return. Record the boundaries of simulated tests and any unverified portion of the installed system.
What is essential in a chemical dosing system RFP?
List tower and chemical scope, P&ID, I/O, verification method, alarm and notification rules, permissions, fault action, logs, responsibilities and FAT/SAT acceptance criteria. Specify re-evaluation after a chemical or equipment change.
Does a successful FAT/SAT guarantee acceptable water quality?
It verifies the specified equipment, signals, logic and site connection. Continuing water quality needs sampling, analysis, maintenance and corrective action under the water management program. A single commissioning sample cannot guarantee long-term control.
Summary
Cooling tower chemical dosing pump monitoring works when the command, equipment state, delivered volume and water-side result are interpreted together. Select a compatible verification method, make “cannot verify” visible, assign alarm response, and test failure modes during FAT/SAT. The project then gives operators evidence they can act on instead of a reassuring pump-run light.
For an assessment of an existing dosing line, its measurement points, RFP and FAT/SAT in a Thailand factory, contact TOMAS TECH. We can define the investigation scope from your drawings and current treatment procedure.
Primary references
- CDC: Controlling Legionella in Cooling Towers.
- US EPA: WaterSense at Work, Cooling Towers.
- Grundfos: DDA SMART Digital.
- ProMinent: gamma/X metering pump.
- ProMinent: Metering Monitor.
- ASHRAE Handbook: Water Treatment.
- Veolia Water Technologies: Chemical Feed Systems.
These are international technical references. Site chemistry, limits and legal duties require local confirmation.