Density Control of Heavy-Medium Suspensions in Coal Washing

A Dense Medium Density Meter Without a Nucleonic Source? Suspension Density Control in Coal Preparation Plants

Closed-loop dense medium control using the ultrasonic acoustic-impedance method in place of a nucleonic (gamma-ray) density gauge

Scope: steam coal and coking coal preparation plants — the correct medium tank, the dense medium circulation main, the feed to three-product cyclones (or inclined-wheel bath separators), and the discharge lines from clean coal drain-and-rinse screens and centrifuges

▍Project Snapshot / Process Conditions

1 Process Background and Measurement Challenges

▍1.1 The suspension density is the separation density

Dense medium separation is the mainstream process in large Chinese coal preparation plants. Raw coal enters a three-product cyclone (or an inclined-wheel dense medium separator) and is mixed with a dense medium suspension made up of magnetite powder and water in a set ratio; the density differences between coal, middlings and refuse then separate the feed into clean coal, middlings and refuse. The suspension density is directly equal to the separation density, with a typical working band of 1.30 ~ 1.85 g/cm³ — it is not a quantity to be monitored, it is the variable that directly determines the separation result.

In more highly automated plants the density signal is the closed-loop input to the make-up water valve (to lower the density) and to the medium addition screw (to raise it), so the continuous availability of the density measuring point determines directly whether the separation system can run automatically over the long term:

▪Density in the correct medium tank and the dense medium circulation main: the first control point for separation density, determining clean coal ash and yield;

▪Make-up water valve and medium addition screw: both take the density signal as their closed-loop input; if the signal is lost or drifts, control reverts to manual adjustment by operator experience;

▪Magnetic separation recovery and dilute medium return: variation in magnetic separation recovery changes the medium concentration and the coal slime content of the suspension, and the density data must reflect this in real time;

▪Clean coal drain-and-rinse screen and centrifuge discharge: reflects medium removal performance and medium loss, and is the observation window for medium consumption management.

The experience widely quoted in the industry is that a deviation of the order of 0.02 g/cm³ from the working density setpoint of the dense medium suspension is enough to shift clean coal ash by a few tenths of a percentage point, which in turn affects the price grade of the clean coal product. This order of magnitude comes from published coal preparation literature and from field experience; it varies considerably with coal washability, size composition and separation equipment, and each mine should refer to its own washability curves and production test results.

The suspension density is continuously disturbed by several factors at once — changes in raw coal size composition, variation in the slime content of the feed, variation in magnetic separation recovery, and the make-up water and medium screw feed rates. Precisely because there are so many disturbance sources, the density measuring point has to be online and continuously available over the long term. Once the instrument is taken out of service for cleaning, the diaphragm drifts or the signal is distorted, the loop degrades into adding water and medium by experience, and the separation density goes out of control with it.

▍1.2 Limitations of Existing Measurement Methods

▪Nucleonic (gamma-ray) density gauges: long the default in the coal preparation industry, but they are radiation-based measuring instruments requiring a Radiation Safety Licence together with annual inspection, qualification training for radiation workers and source decommissioning and disposal. In recent years new coal preparation projects in a number of regions have run into restrictions at the radiation safety and environmental approval stage, and existing installations face pressure to be decommissioned.

▪Vibrating fork density meters: the fork tines sit directly in a suspension of magnetite powder and coal slime, where slime readily wraps around them and the tines wear. This plant reported that a shutdown for cleaning was needed every 1 ~ 2 months; the measurement was frequently interrupted and closed-loop control had to be switched out.

▪Differential pressure / diaphragm density meters: the diaphragm is in direct contact with the suspension, and magnetite particles are hard and heavy. Long-term abrasion thins and eventually perforates the diaphragm, zero drift is pronounced and repeated shutdowns for recalibration are required; the pressure tappings and their flushing water system are themselves a maintenance burden.

▪Issues common to all insertion-type instruments: the dense medium circulation line runs at high velocity with a heavy solid phase, so the leading edge of any part protruding into the flow path is continuously scoured, while at the same time creating a downstream flow disturbance and a point for material build-up. Wear and blockage are determined by the geometry; changing the material of construction delays them but does not remove them.

Figure 1 Schematic process layout for suspension density control in a dense medium coal preparation plant (PS7000 #1 dense medium circulation main — DCS closed-loop control of the make-up water valve and the medium addition screw; #2 slurry concentration downstream of clean coal medium removal)

2 The PS7000 Technical Approach

▍2.1 Measuring principle: the chirped (linear frequency modulation) acoustic-impedance method

When an ultrasonic wave travelling in a medium meets a solid-liquid interface, the echo amplitude is determined by the difference in acoustic impedance across that interface, and acoustic impedance Z = medium density ρ × sound velocity c is directly related to the density of the medium. The PS7000 uses a single self-transmitting, self-receiving sensor that continuously emits chirped (linear frequency modulation) ultrasonic pulses and acquires the echoes in real time. The wide bandwidth of the chirp signal helps the pulse penetrate bubble layers and suppresses multiple-reflection interference, and the transmitter then resolves the acoustic-impedance signature of the echo to recover the density of the medium.

The fundamental difference from the gamma-ray method is that the PS7000 contains no radioactive source of any kind. It is a non-nuclear instrument: no radiation safety licence is required, and there is no annual source inspection, no radiation-worker qualification and no source decommissioning and disposal.

▍2.2 Non-intrusive measurement: the sensor is wetted but does not protrude into the flow path

The PS7000 measuring spool is flanged into a straight run of the process line and the ultrasonic sensor is mounted on the spool. This is the key structural difference from conventional insertion-type instruments, and the point most often misunderstood during selection, so it is set out here exactly as it is worded in the Operating Manual:

▪Non-intrusive (non-intrusive is not the same as non-contact): the sensor must be wetted by the medium in order to work, but it does not protrude into the flow path — there is nothing projecting into the bore. The manual is explicit on this point: for horizontal installation with the sensor on top of the pipe, a full pipe must be guaranteed, otherwise the sensor loses contact with the liquid and the measurement fails; for remote-mount horizontal installation the sensor should be placed on the underside of the pipe so that it stays wetted.

▪There is therefore none of the leading-edge erosion, build-up at the probe root or downstream flow disturbance associated with insertion probes, and no risk of blocking the line.

▪The acoustic window of the sensor is sapphire: a fine surface finish, wear-resistant and corrosion-resistant. The measuring spool can be supplied with a lining (PTFE / rubber / ceramic) to suit the duty.

▪Heavily scaling media should still be included in routine inspection — the manual lists "scaling on the sensor face" as one of the items to check when calibration fails, the remedy being to clean the sensor and recalibrate.

▪Applicability limits: bubbles in the medium are permitted; the medium should not, however, contain large quantities of suspended or settled coarse particles.

Figure 2 Sectional view of the PS7000 installation — the structural difference from conventional insertion types (schematic drawing of the principle, not a record of measured data)

▍2.3 Key Technical Specifications

※ In dense medium plants the suspension is conventionally described in density units (g/cm³), which is the same measured quantity expressed differently from the "% concentration by weight" used in the table; a magnetite suspension of 1.30 ~ 1.85 g/cm³ converts to a value inside the 0 ~ 80 % range. The units and range scaling of the local display and of the analogue output should be confirmed at the selection stage.

▍2.4 Selection and Installation Requirements (important)

The following conditions directly determine the stability of the measurement in the field. We recommend confirming them while the measuring point is still being designed; doing so avoids the great majority of post-commissioning disputes over deviation.

▪Vertical installation with upward flow is preferred. For horizontal installation, confirm that the medium is not stratified and that the measuring point is representative.

▪Flow velocity > 1 m/s; the higher the density of the solid phase, the higher the velocity required. Avoid low-velocity or stagnant duties, in which particle settling disturbs accuracy and stability.

▪Straight run requirement: the installation position should be well clear of pumps, valves, bends and other local resistances; for horizontal installation, 10D upstream / 5D downstream of straight run must be provided (5D upstream / 3D downstream as an absolute minimum).

▪Keep the sensor wetted: for integral horizontal installation with the sensor on top of the pipe, the pipe must run full; for remote-mount horizontal installation the sensor should be placed on the underside of the pipe. Where the pipe does not run full, the remote-mount version is preferred.

▪Avoid points where free gas bubbles accumulate. Aeration, twin-pump suction, negative-pressure inlets and similar conditions introduce large quantities of free gas bubbles; the bubbles raise the echo energy and bias the reading high. This is an inherent limitation of the acoustic method and must be handled by the choice of measuring point, not by compensating in calibration.

▪The comparison sampling point should be as close as possible to the measuring point. A sampling point far from the measuring point introduces transport lag, so the laboratory reference value and the displayed reading no longer refer to the same moment, and the difference is easily mistaken for instrument error.

▪On the dense medium circulation line, choose a stable section downstream of the pump: pressure pulsation and entrained bubbles are strongest immediately at the pump discharge, so allow enough distance downstream to satisfy the straight run requirement in 2.4, and preferably choose the vertical upward-flow section ahead of the cyclone feed.

▪Magnetite powder is heavy and settles quickly: medium settles readily in the circulation line when the plant is stopped or lightly loaded, and readings taken shortly after restart do not represent the steady-state density. We recommend using the circulation pump running signal on the DCS side to qualify the data, so that readings from start-up and shutdown transitions are kept out of the control loop.

▍2.5 Field calibration: the gravimetric sampling method, performed online

Calibration brings the displayed value into agreement with the true density of the medium: several pairs of "instrument reading + corresponding laboratory reference value" are collected and the instrument automatically regresses the correction coefficients. The whole procedure requires no shutdown and can be carried out online provided the process is steady and the sensor is clean.

Process conditions to be confirmed, item by item, before calibration:

The most common situation in the field is that only the process medium is available and clean water cannot be introduced temporarily. In that case two consecutive acquisitions are made on the same medium (about 2 minutes each, about 4 minutes in total): the two measured points plus the virtual zero point added automatically by the system form three regression points, and the instrument uses the goodness of fit r² to cross-check the consistency of the two signals. If bubbles, a step change in flow or an abnormal valve movement during acquisition make the two data sets inconsistent, the system rejects the calibration outright rather than returning an incorrect coefficient. This is the key mechanism that prevents a calibration that looks successful but is in fact wrong.

Recommended calibration interval: calibrate a newly installed instrument once within the first 12 ~ 24 hours of operation; once operation is stable, verify every 6 ~ 12 months; verify every 1 ~ 3 months at measuring points with demanding accuracy requirements; recalibrate immediately after changing the sensor or the medium.

3 Before / After Comparison

4 Field Verification and Operating Record

In a retrofit project at a large steam coal preparation plant in Shanxi (about 6 million t/a of raw coal feed), three PS7000 units replaced the existing nucleonic density gauges on two dense medium circulation loops and on the line downstream of clean coal medium removal. According to the plant's own figures over six months of operation, the working density band narrowed from about ± 0.03 g/cm³ to about ± 0.008 g/cm³, and over the same period the average clean coal ash fell by about 0.4 percentage points.

It should be stated plainly that clean coal ash is the combined result of separation density, raw coal washability, feed slime content, the condition of the separation equipment and other factors; better density control is only one of them, and the change in ash should not be attributed simply to a single instrument. We recommend that acceptance be based on directly measurable process quantities such as the density variation band and the sampling comparison deviation.

Calibration checks during operation were all carried out online by the gravimetric sampling method, with no shutdown scheduled for instrument calibration; once the original nucleonic density gauges had been decommissioned, the plant was correspondingly relieved of the annual inspection of its radiation safety licence and of maintaining radiation-worker qualifications.

▍Verifiable Benefits for the Customer

▪No licence application, annual inspection, personnel qualification training or source decommissioning and disposal for radiation-based instruments, which simplifies the approval route for both new and retrofit projects.

▪The monthly shutdowns previously needed to clean the fork tines and calibrate the differential pressure diaphragms have been eliminated, and the density signal in the dense medium loop is now continuously available over the long term.

▪With the density signal brought into the DCS, the make-up water valve and the medium addition screw have a stable closed-loop input, and the working density band has narrowed noticeably (plant figures over six months).

▪Adding a measuring point on the clean coal drain-and-rinse screen and centrifuge discharge line has given an online means of observing medium removal performance and medium loss, providing data for medium consumption management.

5 Frequently Asked Questions

Q1 Dense medium working density has to be controlled to the order of 0.01 g/cm³. Is a ± 1 % specification good enough?

A To be clear at the outset: the manual gives one figure, ± 1 % FS over the full range (the published datasheet figure is ± 1 % concentration by weight), and does not give a separate short-term repeatability or resolution figure, so we do not promise ± 0.01 g/cm³ absolute accuracy. The reality of dense medium duty is that the instrument only ever works within the narrow band of 1.30 ~ 1.85 g/cm³, and what the loop uses is the relative change and the trend of the density. For a control loop, long-term zero stability and signal continuity are often more critical than full-range absolute accuracy — this is where insertion instruments lose out, not on their nominal accuracy. The correct engineering approach is to calibrate with the plant's own magnetite suspension within the actual working density band, and to set out the sampling comparison rules and the acceptance deviation in an annex to the contract, with acceptance based on the mean deviation and the trend consistency across several consecutive comparison sets. If absolute density values are required for custody-transfer metering, a different solution should be selected.

Q2 Do coal slime, bubbles and magnetite settling affect the measurement?

A The three act by different mechanisms, so take them separately. (1) A change in slime content changes the composition of the solid phase; the acoustic-impedance method measures the overall density of the suspension, so if the slime proportion changes substantially we recommend recalibrating, or covering the typical range of slime variation during calibration. (2) Bubbles — the manual explicitly permits bubbles in the liquid, and the wideband chirp signal penetrates fine dispersed bubbles well; but the negative-pressure region at a pump inlet, venting operations and points where the medium falls and entrains air all generate large quantities of free gas bubbles, which raise the echo energy and bias the reading high. This is an inherent limitation of the acoustic method: it can only be avoided by the choice of measuring point, not compensated by calibration coefficients. (3) Magnetite is heavy and settles quickly, so a horizontal pipe stratifies very readily when the plant is stopped or the velocity is low, and the suspension at the measuring point is then no longer representative — a vertical upward-flow section should be chosen in preference, with a velocity above 1 m/s. Note also that the medium should not contain large quantities of suspended or settled coarse particles, and that the measuring point should not be on a section through which lump coal or pieces of refuse may pass.

Q3 Can the mounting position of the existing nucleonic density gauge simply be reused?

A Usually not directly. A nucleonic gauge is generally an external arrangement with source and detector clamped around the pipe, which places few demands on straight run; the PS7000 measuring spool has to be flanged into the line, so the straight run conditions of section 2.4 (10D upstream / 5D downstream for horizontal installation, 5D / 3D as a minimum) together with the full-pipe and velocity conditions must be confirmed, and both the pipe length needed to insert the spool and a shutdown window must be allowed for. Before a retrofit we recommend supplying site photographs or an isometric drawing of that section of pipework so that Pisonics can work out a feasible position for the spool. If the pipework cannot be cut at all, or the shutdown window is extremely short, the PS7010 clamp-on solution (acoustic attenuation method, no tapping required) can be evaluated, but its applicable duties and accuracy positioning differ from those of the PS7000 and must be confirmed separately against the requirements of the measuring point.

Q4 Does the sensor actually touch the medium? Will it be worn away?

A It does touch it. The manual's formal wording is "non-intrusive", not "non-contact" — the sensor has to be wetted by the medium to work (with horizontal installation and the sensor on top, the pipe must run full or the measurement fails), but it does not protrude into the flow path and there is nothing projecting into the bore, so the main wear mechanism of insertion probes, the leading edge being continuously scoured by fast-moving solids, does not exist. Please note that some material in circulation describes this arrangement as "non-contact", which is not accurate. Within the Pisonics range, the products that genuinely do not touch the medium are the PS7010 clamp-on type (acoustic attenuation method) and the nuclear gamma-ray density gauge.

Q5 How is it calibrated in the field? Is a shutdown needed? How often?

A By the gravimetric sampling method, carried out from the handheld remote menu, with no shutdown — calibration can be done online provided the pipe is full, flow variation is within ± 5 %, no venting is in progress, the temperature is steady and the sensor is clean. Where only one medium is available on site, two consecutive acquisitions are made on that medium (about 4 minutes in total) and the system cross-checks the consistency of the two signals with r², rejecting the calibration outright if they disagree. Interval: first calibration within 12 ~ 24 hours of installation, verification every 6 ~ 12 months once stable, every 1 ~ 3 months for high-accuracy points, and immediate recalibration after changing the sensor or the medium.

Q6 Should the PS7000 or another model be selected for a given measuring point?

A The PS7000 (ultrasonic acoustic impedance) is aimed at the density of bubble-laden, high-solids slurries. If the pipe cannot be tapped, or an ageing line is being retrofitted, choose the PS7010 clamp-on type (acoustic attenuation); for concentration analysis of clean liquids choose the PS7020 (sound velocity); for scaling-prone acids and alkalis in chemical service choose the PS7400 (vibrating fork); for large-bore mains above DN200 the PS7300 (differential pressure) is an option; for custody-transfer accuracy choose the PS7200 (Coriolis); for multi-component concentration analysis choose the PS7100 / 7110 (spectral / refractometric); for aqueous suspensions the PS7600 (microwave) is available. If the selection is not clear-cut, send us the process data and we will work it out for you.

About Pisonics

Xi'an Pisonics Information Technology Co., Ltd. (PISONICS) specialises in ultrasonic inline measurement technology, with products covering slurry density, solution concentration, spectral concentration, level, flow and other industrial process parameters. The PS7000 Series Ultrasonic Slurry Density Meter is the company's own non-nuclear inline slurry density monitor, and is in service in power plant desulphurisation, non-ferrous and ferrous mineral processing, coal preparation, salt lake chemicals, dredging, municipal sludge and other industries.

Ultrasonic · Density · Concentration · Inline Measurement Specialists

※The process data and operating records in this case study are compiled from an actual project; Figures 1 and 2 are schematic diagrams of the principle and the layout, not curves recorded from measurement. The technical specifications of the instrument and its installation and calibration requirements are as stated in the Pisonics Ultrasonic Density Meter Operating Manual supplied with the goods, and are subject to change without notice.