Tiegel home Talk to a metallurgist

Flotation circuit surveys: planning, sampling, mass balancing and size-by-size recovery

Planning and running a flotation circuit survey: steady-state checks, sample points, two-product and least-squares mass balancing, and size-by-size recovery analysis.

· 5 minute read · By

Routine plant figures give the overall recovery but do not show where metal is being lost. A circuit survey provides a snapshot of the flotation circuit: samples of every stream entering and leaving each bank of cells, taken over the same period while the plant operates at steady state.

A mass balance calculated from the assays gives the flow of solids and metal in each stream. It identifies the bank that is losing metal and, with size analysis, the particle sizes in which the losses occur, so that a general concern about recovery becomes a specific target for improvement.

This article covers survey planning, sampling, mass balancing and analysis. The mass balance method is described in more depth in metallurgical mass balance and reconciliation, and bank sizing in flotation cell sizing.

Survey planning

The survey plan should begin with a written statement of the question the survey must answer, for example "where are coarse copper losses occurring?". The scope of sampling and analysis follows from that question.

Throughput, grind, reagent additions and cell levels should be held constant for at least an hour before and throughout the survey, and recorded every few minutes. Sample points are required on every stream entering and leaving each bank, so that every node can be balanced, and each point must allow the full stream to be cut safely.

Several cuts are taken from each stream over the survey period, which is often one to two hours, and combined into one sample per stream. The analyses normally include assays for the valuable metal and at least one gangue element, percent solids, and size-by-size assays on the key streams.

Flotation survey sample pointsA rougher bank feeds a scavenger bank; rougher concentrate goes to a cleaner bank. Sample points are on the rougher feed, rougher concentrate, scavenger concentrate, scavenger tail (final tail), cleaner concentrate and cleaner tail.RougherScavengerCleanerfinal tailfinal concentratescavenger concentrate returns to the rougher feedcleaner tail to scavengerrougher concsample point: several cuts over the survey period, combined per stream
Sample every stream that enters or leaves each bank, so that every node can be balanced. Circuit layouts vary; this one recycles scavenger concentrate and cleaner tail. Illustrative figures.

Bank recoveries from the two-product formula

For a single bank, the two-product formula gives the concentrate mass split and recovery from three assays; see plant recovery formulas.

Worked example: rougher feed 1.20% Cu, rougher concentrate 18.0% Cu, rougher tail 0.25% Cu. The mass split is (1.20 − 0.25) ÷ (18.0 − 0.25) = 5.35% and the copper recovery 18.0 × 0.95 ÷ (1.20 × 17.75) = 80.3%. The same streams assay 8.0%, 28.0% and 6.8% Fe, which gives a mass split of (8.0 − 6.8) ÷ (28.0 − 6.8) = 5.66%. The two elements give different mass splits, as real assays always do, and a method is needed to reconcile them.

Least-squares mass balancing

minimiseS=∑(xmeasured−xadjustedσ)2

where

xmeasured
each measured assay or flow
xadjusted
the adjusted value, constrained so that every node balances
σ
the standard deviation expected for that measurement; precise measurements move least

Least-squares balancing adjusts every measurement by a small amount, weighted by its precision, until all nodes balance for all elements and for solids simultaneously. The most precise measurements are adjusted least. The adjusted values are the best estimate of the circuit, and the size of each adjustment is a check on the data: a measurement that requires a large adjustment relative to its expected error is suspect and should be investigated rather than accepted.

The standard deviations should reflect sampling error as well as assay error. A stream that is difficult to sample, such as a coarse tail, warrants a larger error than a clean concentrate. The calculation is performed with a spreadsheet solver or dedicated mass balancing software.

Size-by-size recovery

Ri=C×ci×wC,iF×fi×wF,i

where

Ri
recovery of the valuable metal in size class i
C, F
concentrate and feed solids flows
ci, fi
metal grades of size class i in concentrate and feed
wC,i, wF,i
mass fractions of the concentrate and feed in size class i
Rougher copper recovery by size classRecovery is 55% above 150 micrometres, 82% at 106 to 150, 89% at 75 to 106, 90% at 38 to 75 and 77% below 38 micrometres.020406080100+150 µm: 5555+150 µm106–150 µm: 8282106–150 µm75–106 µm: 898975–106 µm38–75 µm: 909038–75 µm−38 µm: 7777−38 µmCopper recovery (%)
Size-by-size recovery shows where the losses are. Coarse losses point to grind or liberation; fine losses to slimes, entrainment or reagent problems. Illustrative figures.

In the illustrative results, recovery peaks at 90% in the 38 to 75 µm class and falls to 55% above 150 µm and 77% below 38 µm. Coarse losses usually indicate poor liberation or particles too heavy for the bubbles to lift, and point to grind, classification and cell conditions. Fine losses suggest slow flotation of fines, slime coatings or reagent problems, and point to reagent addition points, residence time and froth conditions. A size-by-size balance requires size distributions for each stream and assays of each size class. It is the most expensive part of a survey and also the most informative.

Survey report

  • A balanced flowsheet with solids, water and metal flows for each stream.
  • Bank and stage recoveries, residence times from measured flows, and comparison with design.
  • Size-by-size recovery for key banks.
  • Plant conditions during the survey, and any unusual events.
  • Recommendations, each tied to a finding, and a plan to test them.

Glossary

Survey
A planned sampling campaign around a circuit at steady state.
Node
A point in a flowsheet where streams join or split, such as a bank of cells.
Mass balance
Calculation of solids, water and metal flows that satisfy conservation at every node.
Least-squares balancing
Adjusting all measurements, weighted by their precision, so every node balances.
Mass split
Concentrate mass as a fraction of feed mass for a bank.
Size-by-size recovery
Recovery calculated separately for each particle size class.
Steady state
Plant conditions held constant so that flows and grades do not drift during sampling.
Scavenger
A bank that recovers remaining valuable mineral from rougher tailings.

Frequently asked questions

How long should a flotation survey run?

Long enough to take several cuts of each stream at steady state, often one to two hours, after the plant has been held steady beforehand.

Why do copper and iron assays give different mass splits?

Every assay carries sampling and analytical error. Least-squares balancing combines all of the measurements into one consistent set of flows.

Are size-by-size assays necessary?

For diagnosing losses, yes. They show whether the lost metal is coarse, fine or in the intermediate sizes, and each of these points to a different remedy.

How is the quality of survey data assessed?

The balance should require only small adjustments relative to the expected errors, the plant should have remained steady, and repeat cuts should agree.

Discuss your plant with a metallurgist

Tiegel is a Perth metallurgy company. We help with testwork, plant optimisation, metal accounting and investigations, and we build the Tiegel Suite software.

Talk to a metallurgist

References and further reading

  1. Wills, B.A. and Finch, J.A. (2016). Wills' Mineral Processing Technology, 8th edition. Butterworth-Heinemann. Chapter 3, Metallurgical accounting, control and simulation.
  2. Wills, B.A. and Finch, J.A. (2016). Wills' Mineral Processing Technology, 8th edition. Butterworth-Heinemann. Chapter 12, Froth flotation.
  3. Lynch, A.J., Johnson, N.W., Manlapig, E.V. and Thorne, C.G. (1981). Mineral and Coal Flotation Circuits: Their Simulation and Control. Elsevier.

Worked examples and charts marked illustrative use invented figures, not data from any operation.