Direct answer

No single technology measures the entire phenomenon. A manual or automated test characterises a sample under defined conditions. Online instruments observe process variables such as bed, inventory, density, flow or clarity. Models and soft sensors combine signals to estimate states that are not measured directly.

The phrase ‘real time’ is insufficient unless it states what is measured, where, how often, with what delay and under which validation. A frequent automated test is not the same as a sensor continuously installed in the thickener.

First define the object of measurement

Settling can refer to the trajectory of an interface in a vessel, the separation velocity of a sample, liquid clarity, solids inventory in a thickener or an estimate of future feed behaviour.

Each object answers a different question. Technology selection should begin with the decision: compare reagents, detect a material change, monitor equipment, verify clarity, estimate inventory or support a control action.

The time scale must also be stated. A camera recording several frames per second may track a twenty-minute test, but the result remains discrete relative to the plant if only one sample is taken per shift.

Manual settling test

  • What it observes Interface height and supernatant appearance in a prepared sample.
  • Strength Low cost, direct inspection and flexibility for exploring conditions.
  • Limitation Operator dependence, low frequency and difficulty maintaining identical reading and mixing criteria.
  • Appropriate use Exploratory comparison and verification of a protocol before automation.

A manual test can be technically valid when the method is documented and its uncertainty is acceptable. Automation does not correct an unrepresentative sample or a poorly defined protocol.

Automated test with interface tracking

  • What it observes Temporal evolution of an interface or optical pattern during a batch test.
  • Strength Greater data density, temporal traceability and consistent application of the reading algorithm.
  • Limitation It still depends on sampling, preparation, visibility and the relationship between the optical signal and the physical phenomenon.
  • Appropriate use Comparing curves, detecting changes between samples and retaining auditable evidence of the test.

This is the scope that should be stated for a platform such as GENKO while the available evidence concerns automated tracking of tests. It must not be described as an online thickener sensor without that installation and validation.

Online thickener instrumentation

  • Bed level or profile Seeks to locate an interface or internal distribution; performance depends on technology, geometry, bubbles, solids and maintenance.
  • Bed pressure or inventory Uses a measurement associated with hydrostatic load or accumulated mass; interpretation requires densities and location.
  • Feed or underflow density and flow Can produce mass-flow estimates when both signals have sufficient quality and synchronisation.
  • Overflow turbidity or solids Observes carry-over and clarity at the installed point, not the complete settleability of the feed.
  • Rake torque, position and pressure Describe mechanical response and equipment condition; by themselves they do not identify the cause of change.

These signals are continuous with respect to the process, but none replaces material characterisation. Resolution, maintenance, calibration and blind zones must be part of the design.

Historian analytics and soft sensors

A soft sensor estimates a variable from available measurements and a model. It may combine flow, density, dosage, bed, torque, turbidity and delays to classify regimes or anticipate a response.

Its strength is integrating plant context and frequency. Its limitation is that the estimate inherits biases, missing data, operator actions and regime changes. It must be validated on data not used for fitting and compared with an independent reference.

Calculation frequency does not turn an estimate into a direct measurement. The result must be identified as observed, derived or predicted.

Recommended architecture

A robust architecture is usually hybrid. Sampling and tests characterise material response. Online instrumentation observes thickener state and outputs. The historian retains context and actions. A model links those layers only after synchronisation, uncertainty and regimes have been reviewed.

Technology options should be compared by:

  • physical variable and location;
  • effective frequency from sample to result;
  • spatial and temporal representativeness;
  • resolution, bias, repeatability and uncertainty;
  • maintenance, cleaning and availability;
  • delay relative to the decision;
  • method used to validate the result;
  • consequence of an incorrect reading.

The recommendation is not to maximise sensor count. It is to close the observability gap that constrains a defined decision.

Sources and evidence status

  • Endress+Hauser, Thickener optimization lists industrial measurements of inventory, level, density, flow, dosage and clarity. It is used to identify instrument classes, not to validate universal performance.
  • Metso, Thickener optimizer describes a commercial architecture that uses process measurements and manipulates pumping and flocculant. It is cited as an integration example, not as an independent vendor comparison.
  • NIST/SEMATECH, Process Monitoring and Control provides the framework for distinguishing measurement, statistical monitoring and control action.
  • Editorial status PUBLISHED. The page compares categories; it does not certify instruments, vendors or an online GENKO installation.