A factory planning non-contact level detection on a plastic tank can easily buy a sensor that works on a demonstration container but behaves unpredictably after installation. Selecting and deploying a capacitive level sensor requires the liquid, vessel wall, mounting gap, deposits, wiring and control response to be evaluated together. This guide helps a Thai factory’s purchasing, engineering and maintenance teams compare quotations on the same basis and accept the installation with evidence.
Define the decision before selecting a level sensor
“Level” covers three different requirements: a point switch that reports whether liquid is present at a defined height, several points that indicate a band, and continuous measurement of level. A low-level pump inhibit may need a point signal. Replenishment forecasting may need a continuous value. Write the action first: “When tank A is below the low point, block the pump start request and show the cause on the HMI.” For overflow prevention, specify the high point and the valve response separately. Whether either signal is an independent protective layer requires a separate risk assessment; an ordinary process sensor is not automatically a personnel-safety device.
Autonics describes its CP series as cylindrical capacitive proximity switches suitable for presence and level-point detection. Its stated maximum nominal sensing distance of up to 25 mm is model dependent. It is not a guaranteed permitted tank-wall thickness, nor a verified sensing margin for a selected liquid. IO-Link and standard variants exist; not every CP variant has IO-Link. The CP is not a continuous level transmitter. In contrast, ifm markets KQ10 as a different capacitive sensor for continuous level measurement through non-metallic vessels. These products illustrate distinct functions, not an endorsement of either for a particular tank. See Autonics CP and ifm KQ10.
When non-contact capacitive detection can work
Capacitive sensing responds to a change in the dielectric environment near the sensor. Some configurations can detect material through a non-metallic wall without touching the liquid. “Non-contact” does not mean every tank and medium will work. Metal walls, excessive wall thickness, mounting air gaps, low-dielectric liquids, foam, residue and nearby metal can affect the result. Ifm’s installation guidance explicitly discusses dielectric contrast between material and vessel, wall thickness and low-dielectric media. Test the actual combination, rather than transferring a catalogue maximum to the site. See ifm installation guidance.
Balluff’s battery-production example describes point-level detection of raw-material and electrolyte containers through non-metallic walls, with stated conditions for conductive media and wall thickness. Its conditions apply to the specified product and application; do not copy a 10 mm vessel-wall figure into a different product’s specification. See Balluff’s application example.
Record the vessel and medium as one test object
Survey tank manufacturer and model, material, measured wall thickness and tolerance, curvature, ribs, covers and the gap between sensor and wall. For every medium, record product and formulation range, temperature, available dielectric or conductivity data, viscosity, foam, sediment, cleaning fluid and persistent film. A test with water does not validate oil, resin or solvent. If formulation varies by supplier or lot, acquire samples representing the difficult cases.
Overlay the proposed sensing point with pump suction, return pipe, agitator, foam layer and filling stream. A sensor beneath the return line may respond to droplets rather than bulk level. Liquid clinging to an emptied vessel may look like “present.” Photograph and record empty, just below threshold, just above threshold, normal operation, full, immediately after draining and immediately after cleaning. Agree which reading is correct in each state before tuning sensitivity.
Treat foam and deposit compensation as product-specific
Balluff describes compensation in particular Smart Level configurations; ifm describes wall-deposit monitoring on KQ10. Neither statement proves that every capacitive switch tolerates every foam, chemical film or crust. Higher sensitivity may detect the liquid but also a residual film. A setting that ignores buildup may miss a lower-dielectric medium. Test both error directions with the real cleaning and production cycle. Sources: Balluff technology overview, ifm KQ10 overview.
Build a selection matrix before asking for prices
Capacitive sensing is one candidate. Other options include top-mounted ultrasonic or optical instruments, immersed probes or floats, pressure and weighing. If the vessel is metallic, the liquid has insufficient dielectric contrast, or wall and residue conditions cannot be controlled, an external capacitive solution should not be assumed in the RFQ. Compare installation, hygiene, maintenance and shutdown cost for the alternatives.
| Decision | Buyer supplies | Supplier answers |
|---|---|---|
| Purpose | Low/high point, continuous inventory, alarm or interlock | Point switch or continuous value; exact output meaning |
| Medium | Every liquid, concentration, temperature, foam, residue, cleaner | Applicable and excluded conditions; actual-liquid tests |
| Vessel | Wall material and tolerance, curvature, ribs, metal nearby | Specific model, mounting and demonstrated margin |
| Electrical | Power, PLC inputs, PNP/NPN, cable length | Diagram, failure output, protection, IO-Link variant |
| Site | Washing, water, oil, vibration, maintenance access | Enclosure rating, materials, cable and replacement |
| Acceptance | Allowed false decisions and response time | FAT/SAT procedure, logs and responsibility |
A catalogue maximum is not a simple wall-thickness allowance. Attach the chosen model’s manual, installation requirements and test evidence to the contract. Ask the supplier to state assumptions and exclusions rather than simply marking “liquid compatible.”
When does an IO-Link level sensor make sense?
IO-Link is useful when the chosen variant actually supports it and the IO-Link master, PLC integration, IODD, parameter backup and maintenance process are available. The IO-Link Community explains that an IODD describes device identity, configurable parameters, process data, diagnostics and communication. The actual fields depend on the device and its IODD. IO-Link does not by itself turn a point switch into continuous measurement or guarantee fouling diagnostics or automatic field recalibration. See IO-Link IODD.
If a single low alarm and manual replacement setting suffice, a conventional output may be economical. For many tanks, variants and replacement operations, compare an IO-Link variant, including master and integration cost. Our IO-Link sensor integration guide covers the wider data architecture. The decision here is whether this vessel and this liquid can be detected and proved.
Write an RFP that yields comparable quotations
Provide tank drawings and photos, the medium list, control action, acceptance criteria, work scope and exclusions together. A sample clause follows; replace illustrative conditions with approved site values:
The scope is point detection at the low and high locations on plastic tank A. Test the normal medium, the medium with the smallest expected dielectric contrast and the cleaning fluid. For empty, immediately below and above each threshold, after draining and after cleaning, the supplier shall identify the device model, mounting, sensitivity, output, delay and test method. List any unsupported medium or wall location as an exclusion.
Request a wiring drawing, PLC I/O map, normal and broken-wire/power-loss responses, startup behavior, replacement and recalibration method, maintenance access, local spare stock and lead time, and warranty. If existing PLC logic must change, quote backup, change authority, code review, rollback and production stop separately. A sensor-only offer cannot be directly compared with one including controls, SAT and training.
Specify normal, fault and recovery behavior
Does low level stop a running pump or inhibit its next start? Does high level close a feed valve or ask for operator confirmation? Where are debounce and hysteresis implemented? What happens if the signal stays fixed after cable failure? These are system design decisions, not properties determined by a model number. Document them in a state table. A phrase such as “fail safe” may conceal conflicting overflow and dry-running priorities. Evaluate independent protection and functional safety separately where required.
Our water and wastewater automation guide addresses wider pump, valve, PLC and SCADA control. This article focuses on detecting a defined point through a vessel wall and proving that behavior.
Expose quotation exclusions
Compare candidate and alternative models, assumed liquid and wall, brackets, cables, PLC changes, IO-Link master, parameter storage, FAT sample count, SAT duration, training, spares and warranty in common columns. “Decide after site survey” is acceptable only if the survey fee, decision deadline and fallback for incompatibility are clear. Assign owners for medium changes, cleaning-recipe changes and sensor-setting changes. Record the reason, old and new setting, test and approver after every change.
Turn FAT and SAT into acceptance evidence
FAT tests the proposed model and settings on a supplier rig or real wall sample before site work. SAT tests the installed sensor against the actual tank, cable, liquid, temperatures and operating actions. A single LED demonstration is insufficient. Each test needs an ID, tank, medium and lot, temperature, wall and position, settings, independent reference level, sensor output, PLC/HMI response, timestamp, operator, photo or log, and result.
| State | Test stimulus | Acceptance record |
|---|---|---|
| Empty/full | Drain and fill with real medium | Sensor, PLC and HMI agree with reference and timestamp |
| Threshold | Cross both sides repeatedly | Decision repeats within agreed band without unwanted toggles |
| Difficult medium | Low dielectric contrast and temperature/concentration edges | Documented decision and margin |
| Foam/residue | Production-like foam and film after washing | False-decision count and recovery time |
| Failure | Controlled cable, power and communication faults | Defined PLC/HMI state |
| Replacement | Maintenance installs a spare | Configuration, field check and evidence complete |
An illustrative criterion could require 30 crossings near each point for every specified medium with zero false decisions and PLC transitions within specification. Thirty is a hypothetical test count, not a statistical reliability guarantee. Set actual sample counts, distributions and permitted error rates from the site’s risk and downtime cost. A continuous transmitter needs additional range, error, calibration, drift and update-rate criteria.
Include refilling, shift change, cleaning and vessel replacement in SAT. Carry any real-liquid condition not reproduced at FAT into SAT as unverified, and close it before sign-off. Tie operator alarm, maintenance signal and PLC state to the same test ID. When a test fails, investigate medium, wall, mounting, wiring and input logic before changing sensitivity.
Handover should include model and serial numbers, location photos, distance and bracket drawing, tank and medium specifications, wiring and I/O, threshold/hysteresis/delay, IODD and master settings where applicable, raw FAT/SAT data, fault and cleaning procedures, and spares. Automatic parameter restoration does not necessarily replace site calibration with the actual medium; the IO-Link system description explicitly notes that some field adaptations must be restored after replacement.
A 90-day proof of concept with explicit gates
The following is an illustrative plan; adapt it to production stops and material availability. During days 1–30, survey vessels, media and failure modes, compare sensing methods, and run FAT with a real wall sample. The gate is a selection matrix with unsupported and untested conditions. During days 31–60, instrument one tank, log real media across shifts, washing and refilling, connect PLC/HMI, and inject controlled faults. The gate is an acceptance table, false-decision log and corrective-action history. During days 61–90, replace the device with a spare, retest different lots, temperatures and maintainers, and calculate rollout cost. The gate is approved SAT, usable maintenance instructions and a go/no-go decision.
Schedule difficult conditions deliberately: cold morning, hot afternoon, new liquid lot, fresh foam, long-idle wall film and post-cleaning state. Do not mark an untested condition as passed. If two nominally identical tanks differ in wall thickness, ribs or mounting gap, define what “same design” means before copying settings. Retest when vessel supplier or formulation changes.
Separate false presence from false absence at both low and high points. They can cause different harms: dry running, overflow, needless stoppage or unnecessary replenishment. Record duration and downstream control action, not just “accuracy.” If a delay filters chattering, verify that the resulting reaction time is still acceptable for the pump or valve.
An independent reference is essential. If the sensor’s own output is called the truth, its errors disappear from the evaluation. Use a sight scale, scale weight or reconciled fill and discharge record, with the reference uncertainty noted. Log every setting change and repeat relevant tests. Before closing the PoC, have the factory’s maintenance team replace the sensor and prove the empty/full check themselves. A supplier-led demonstration alone does not establish recoverability.
The go/no-go conditions should include all acceptance tests passed, no unresolved critical anomaly, a maintenance-led replacement completed, a stated range of valid vessels/media, and an accepted lifetime-cost estimate. A negative PoC can save a larger failed rollout: change technology, mounting or operating scope rather than hiding nonconformities.
Model the cost beyond sensor unit price
The following numbers are hypothetical illustrations, not Thai market prices, a quotation or guaranteed savings. Suppose ten plastic tanks need two points each: 20 sensors at THB 4,000, brackets and wiring at THB 2,000 per point, THB 90,000 for PLC/HMI changes and testing, and THB 50,000 for FAT/SAT and training. Initial cost is 20 × (4,000 + 2,000) + 90,000 + 50,000 = THB 260,000. Two spares at THB 4,000 make it THB 268,000. Actual products, import costs, downtime, control changes and test coverage will alter every item.
For an illustrative inspection cost, two visits a year × two hours × two people × ten tanks × THB 300 per person-hour equals THB 24,000 per year. This excludes replacement, cleaning and unplanned stops. A five-year simple total is initial cost plus five times inspection, replacement, downtime and recalibration. Establish pre-installation counts and costs for dry running, spills, scrap and rework. Compare afterward using the same definitions.
If annual baseline loss is hypothetically THB 200,000, residual loss THB 80,000 and operating cost THB 30,000, the hypothetical annual net benefit is THB 90,000 and simple payback is 268,000 / 90,000 ≈ 3.0 years. Other process improvements may also explain a loss reduction, so this is not a causal claim. Show a zero-benefit or medium-change scenario alongside the central case. Decision makers need to know which assumptions can break the economics.
FAQ: capacitive level sensor procurement
Can every plastic tank be measured without touching the liquid?
No. Material, wall thickness, curvature, mounting gap, liquid dielectric behavior, nearby metal and deposits matter. Validate the exact model with the actual vessel and difficult liquids. Do not assume the same approach works through a metal wall.
Does a capacitive sensor provide a continuous level value?
A point proximity switch normally reports presence at its installed position. Choose a device explicitly specified for continuous level if that is the requirement. The CP and KQ10 examples represent different functional classes.
Does IO-Link eliminate PLC changes?
Not necessarily. A master, cable and power arrangements, IODD, PLC mapping, alarms and maintenance procedures may change. Include these costs when comparing IO-Link and conventional outputs.
Will foam and residue cause false decisions?
It depends on the medium, model, compensation feature and settings. Test drained-wall film, cleaner and realistic foam against numerical false-decision and recovery criteria.
What differs between FAT and SAT?
FAT tests the proposed model and settings in a supplier environment. SAT verifies the actual tank, wiring, liquid and operation in the factory. Carry anything not reproduced at FAT explicitly into SAT.
Should the RFP name one model?
Existing maintenance standards may justify a candidate model. To judge an equivalent, keep the medium, wall, output, fault behavior and acceptance evidence identical. If applicability is unknown, require sample testing before fixing a model.
Conclusion: procure a repeatable decision
External capacitive point detection is a useful candidate for selected non-metallic vessels. Suitability belongs to the combination of sensor model, medium, wall, mounting, fouling and control logic. Distinguish point detection from continuous measurement, test the actual boundary conditions, and write RFP, FAT/SAT and handover requirements that make quotations and results comparable.
If you are still checking whether a plastic tank and its liquids are suitable for external point detection in a Thai factory, we can help organize the medium list, vessel information and acceptance plan at the planning stage. Contact TOMAS TECH.
