Cement Separator & Ball Mill Optimization

The separator and the ball mill are the two halves of the cement grinding circuit, and their optimization is one problem, not two: the mill grinds whatever the separator sends back, and the separator classifies whatever the mill produces, so the circuit’s performance — its capacity, its specific power and its product quality — is decided by the pair together. The plants that optimize the mill while neglecting the separator, or the separator while neglecting the mill, leave most of the potential on the table, because the two machines interact: a sharp separator raises the mill’s efficiency by removing the finished product immediately, and a correct charge raises the separator’s efficiency by presenting it a well-prepared feed. This article is a complete technical guide to the joint optimization of the cement separator and the ball mill: the separator’s design and its classification science, the mill’s internals and their behavior, the measurement of both through the Tromp curve and the audits, the operating strategies, the control, the aids, the common faults and the complete optimization roadmap.

1. The Circuit as One Machine

The closed-circuit ball mill operates as a loop: the feed enters the mill, the mill product goes to the separator, the separator splits it into the fines and the reject, and the reject returns to the mill with the fresh feed. The loop’s state variables are the circulating load — the ratio of the separator feed to the fresh feed, normally 150 to 300 percent — the product fineness and the mill’s power and temperature. The loop’s efficiency is the product of two efficiencies: the mill’s efficiency at converting the energy into new surface, and the separator’s efficiency at removing the finished surface before it is overground. The two efficiencies are linked by the loop: an inefficient separator sends coarse material back to be reground, wasting the mill’s energy; an inefficient mill sends an overground feed to the separator, wasting the separator’s sharpness.

The joint view has a concrete consequence: every optimization action must be evaluated on the loop, not on the machine. A separator speed change moves the fineness but also moves the circulating load, the mill load and the mill temperature; a charge change moves the mill’s output but also the separator’s feed size distribution. The optimization therefore starts with the loop’s measurements — the feed, the product, the reject, the power and the fineness — and every change is verified by the same measurements. The loop is the machine, and the separator and the mill are its organs.

2. The Separator: Design and Classification Science

The modern separator is a third-generation air classifier: a rotating cage inside a static body, with a primary air flow that disperses the feed and a secondary air flow that sharpens the classification. The physics of the separation is the balance of forces on each particle: the centrifugal force of the rotating cage throws the coarse particles outward, while the drag of the air flow carries the fine particles inward through the cage. The cage speed sets the cut size: faster rotation, stronger centrifugal force, finer cut. The air flows set the sharpness: the ratio of the primary to the secondary air, and the balance of the upward and the tangential velocities, decide how cleanly the size fractions are separated. And the dispersion — the uniform presentation of the feed around the cage — decides the efficiency at every size.

The classification is quantified by the Tromp curve: the probability that a particle of a given size reaches the fines, plotted against the size. The curve’s three features are its cut size x50 (the size at 50 percent probability), its sharpness (the steepness of the transition, characterized by the ratio of the sizes at 25 and 75 percent) and its bypass (the probability at the large sizes, which should approach zero). The ideal curve is a vertical step at the cut size; the real curve always has a finite sharpness and a nonzero bypass. The separator’s health is read from the curve: a rising bypass means worn vanes, leaking seals or poor dispersion; a falling sharpness means an unbalanced airflow or a choked secondary inlet; and a drifting cut size means the cage speed or the feed rate is wrong. The Tromp curve, measured by the simultaneous sampling of the feed, the fines and the reject, is the separator’s X-ray, and its measurement twice a year is the minimum standard of the optimizing plant.

3. The Mill: Internals and Their Interaction with the Separator

The mill’s side of the loop is its internals: the charge, the liners and the diaphragm, whose state decides the mill’s output size distribution and therefore the separator’s feed. The charge’s role in the loop is the preparation: the first compartment must reduce the feed to a size range that the second compartment can finish, and the second compartment must produce the fine fraction that the separator removes. A charge that is too coarse for the feed leaves the coarse particles to the separator, which returns them to a mill that cannot grind them, and the loop runs with a high reject and a low output. A charge that is too fine wastes the impact energy and produces the overground fines that the separator cannot save. The charge grading is therefore set against the feed size and the separator’s cut, and it is reviewed whenever either changes.

The diaphragm sets the mill’s internal staging: its slot size holds the media in the compartments and controls the maximum size passing to the next, and its condition — the wear of the slots and the free area — sets the mill’s throughput and its ventilation. The ventilation completes the loop’s material flow: the air that sweeps the mill carries the fines to the separator, and a ventilation deficiency leaves them in the mill, cushioning the charge and degrading both machines’ efficiency. The loop’s joint audit therefore measures the mill internals and the separator in one campaign: the charge grading, the liner profile, the diaphragm slots, the Tromp curve and the ventilation state, and the findings are interpreted jointly.

4. The Measurement Campaign

The optimization campaign is a structured measurement exercise, and its design determines the quality of the findings. The campaign has four parts. Part one is the loop baseline: two weeks of the plant’s data on the feed rate, the power, the temperature, the reject rate, the fineness and the quality, to establish the current performance and its variability. Part two is the sampling campaign: a full day of simultaneous sampling — the fresh feed, the mill feed, the mill product, the separator feed, the fines and the reject — with each stream weighed, sieved and, for the fine fractions, measured by the laser diffraction. The samples yield the circulating load by the ash or the fineness balance, the Tromp curve of the separator and the size distributions of every stream. Part three is the mill stop inspection: the charge grading and filling, the liner and the diaphragm condition, and the ventilation measurements. Part four is the data analysis: the loop’s mass balance, the efficiency calculation of each machine and the allocation of the losses.

The campaign’s outputs are the numbers that drive the optimization: the circulating load and its trend, the separator’s cut, sharpness and bypass, the mill’s charge state, and the specific power of the loop with its breakdown. The comparison of the numbers with the design values and the benchmarks locates the opportunities, and the opportunities are ranked by their economic size. The campaign is repeated after the changes to verify the gains, and the annual campaign is the loop’s health check.

5. The Separator Optimization in Practice

The separator optimization follows the measurement with the adjustments, in the order of their cost. The first adjustment is the mechanical restoration: the worn vanes and the seals, the leaking flaps and the plugged inlets are repaired, because no setting can compensate for the hardware. The second is the dispersion: the feed distribution around the cage is checked and corrected, because the bypass lives in the dispersion. The third is the airflow balance: the primary, the secondary and the tertiary air flows are set against the design ratios, and the fan and the damper positions are verified. The fourth is the cage speed: the cut size is set against the fineness target, and the speed-fineness curve is established for the operator. The fifth is the operating practice: the separator’s reject rate is established as the loop’s load indicator, and the fineness loop is tuned on the reject feedback. The measured effects of the completed separator optimization are a bypass reduction of 5 to 15 percentage points and a capacity gain of 5 to 15 percent, on the same mill, because every percentage point of bypass is a percentage point of the mill’s energy spent on regrinding.

6. The Mill Optimization in Practice

The mill’s side of the campaign is the charge and the internals, and its optimization follows the same order. The charge audit measures the grading against the design; the correction removes the undersized media, rebuilds the grading against the feed size and the separator’s cut, and restores the filling. The liner profile is reviewed against the compartment duty and the wear data, and the diaphragm and the grates are restored to their design free area. The ventilation is restored to the design flow, and the water injection is calibrated. The measured effects of the completed mill optimization are a charge-related capacity gain of 5 to 12 percent and a specific power reduction of 3 to 8 percent, and the effects interact with the separator’s: a corrected charge presents the separator a better feed, raising its sharpness, and a sharpened separator lets the mill run at a higher load without overgrinding.

The joint effect is the reason the two optimizations are one project: the individual gains are not additive but multiplicative, because each machine’s improvement amplifies the other’s. The completed joint campaign — the charge re-grade, the separator restoration and the loop rebalancing — typically delivers 15 to 25 percent more capacity and 10 to 15 percent lower specific power, and the individual measurements verify each step’s contribution.

7. The Operating Strategies of the Optimized Loop

The optimized loop is operated by a strategy that holds its optimum. The load strategy: the mill is run at its maximum stable load, indicated by the mill power or the differential pressure, with the feed rate as the manipulated variable and the reject rate as the alarm. The quality strategy: the fineness is held by the separator speed, with the reject rate as the secondary input and the quality data as the verification. The temperature strategy: the ventilation and the water injection hold the mill outlet in its band, because the temperature decides the gypsum and the separator’s dispersion. The disturbance strategy: the clinker grindability and the moisture changes are met by the feed-forward — the grindability result adjusts the load set-point before the batch reaches the mill. And the review strategy: the daily data review, the weekly quality review and the monthly performance review keep the loop’s state visible and its settings current.

The control implementation follows the strategies: the basic loops first — the load, the quality and the temperature — then the supervisory layer that coordinates them, and then, where justified, the model predictive control that optimizes the loop against the constraints. The documented gains of the control layer over the manual operation are 3 to 6 percent capacity and a measurable reduction of the fineness variability, and the gains are the same order as the mechanical gains, which is why the control optimization is part of the joint project rather than an optional extra.

8. The Grinding Aids in the Joint Optimization

The grinding aids act on both machines of the loop, which is why they belong in the joint optimization. In the mill, the aid reduces the surface energy of the fine particles, limiting the agglomeration and the ball coating, and improving the grinding efficiency; in the separator, the better dispersion of the feed raises the classification sharpness and lowers the bypass. The net effect is a loop that grinds more and classifies better at the same energy, and the documented joint effect of a well-chosen aid is a 5 to 15 percent capacity gain with an improved distribution. The aid selection is a plant trial: the candidates are tested on the plant’s clinker and additions, first in the laboratory and then in the mill, and the dose-response curve establishes the economic optimum — the dose where the value of the capacity and the energy gains exceeds the aid cost. The aid is then operated as a process variable: the dosing is proportioned to the feed rate, the delivery system is maintained and the effect is verified in the monthly review.

9. The Common Faults and Their Signatures

The joint loop has characteristic faults, each with its signature in the data. A rising reject rate at constant fineness and falling output points to a separator fault — the worn vanes, the leaking seals or the plugged secondary air — and the Tromp curve confirms it with a rising bypass. A falling output at constant reject and rising power points to the mill’s charge — the degraded grading, the worn liners or the choked grates — and the stop inspection confirms it. A rising fineness variability at constant settings points to the feed — the grindability, the moisture or the recipe changes — and the quality data confirms it. A rising temperature with a falling output points to the ventilation or the water injection. And a distribution that flattens while the Blaine holds points to the bypass — the coarse tail that the separator is leaking. Each signature is read from the daily data, and each diagnosis is confirmed by the specific measurement — the Tromp curve, the charge audit or the feed data — before any change is made.

10. The Economics of the Joint Optimization

The economics of the joint optimization are its justification, and the numbers are substantial. The capacity gain of 15 to 25 percent on a 150-tonne-per-hour mill is 22 to 37 tonnes per hour, or 180,000 to 300,000 tonnes per year, worth millions of dollars of contribution margin. The specific power reduction of 10 to 15 percent on a 32 kWh per tonne baseline is 3 to 5 kWh per tonne, worth 0.3 to 0.5 dollars per tonne, or 450,000 to 750,000 dollars per year on a 1.5-million-tonne plant. The cost of the campaign — the sampling, the audit and the corrections — is a fraction of the first month’s gains, and the recurring costs — the media, the aid and the maintenance — are covered by the recurring gains. The investment decisions that follow the campaign — the new separator internals, the new diaphragm, the pregrinder — are justified by the same data, and the joint project’s payback is measured in months, not years.

Optimization Element	Capacity Gain	Energy Gain	Investment
Separator mechanical restoration	5-10%	3-5%	Low
Ball charge re-grade	5-12%	3-8%	Low
Loop rebalancing and control	3-6%	2-4%	Software
Grinding aid optimization	5-15%	5-10%	Operating
Separator internals upgrade	5-10%	3-5%	Medium
Pregrinder installation	20-40%	10-20%	High

11. The Optimization Roadmap

The complete roadmap of the joint optimization is the sequence that respects the interactions. Step one is the baseline: the loop’s current performance and its variability, measured over two weeks. Step two is the campaign: the sampling, the Tromp curve and the stop inspection, with the findings ranked by their economic size. Step three is the mechanical restoration: the separator’s vanes, seals and dispersion, and the mill’s charge, liners and diaphragm, corrected in one planned stop. Step four is the loop rebalancing: the circulating load, the ventilation and the water injection set against the corrected machines, with the control loops retuned. Step five is the aid optimization: the trial, the dose-response and the economic optimum. Step six is the verification: the baseline measurements repeated, the gains confirmed and the new operating envelope documented. Step seven is the sustainment: the annual campaign, the monthly review and the change control that keep the loop at its optimum. The roadmap’s discipline — measure before and after every step — is what separates the successful projects from the failures.

12. The Pitfalls and the Success Factors

The pitfalls of the joint optimization are the errors of the partial view. Optimizing the mill without the separator leaves the bypass and the sharpness unaddressed; optimizing the separator without the mill leaves the charge and the ventilation unaddressed; and both leave the loop at a local optimum that the full view would pass. The second pitfall is the unmeasured campaign: making changes on impressions and verifying nothing, which produces the “improvement” that disappears in the next month’s data. The third is the unrepeated campaign: optimizing once and letting the loop decay, which wastes the gain over the years. The success factors are the mirror images: the joint view, the measurement discipline and the sustained program. The plants that succeed are the plants that treat the separator and the mill as one machine, measure it completely and optimize it continuously — and the loop rewards them with the capacity and the energy numbers that their competitors cannot match.

Frequently Asked Questions

Why must the separator and the mill be optimized together?

Because they interact in the loop: the separator sends the coarse material that the mill grinds, and the mill sends the feed that the separator classifies. Optimizing one alone leaves the other’s inefficiency in the loop, and the joint optimization multiplies the individual gains rather than adding them.

What does the Tromp curve reveal?

The Tromp curve shows the separator’s recovery of each particle size to the fines: its cut size, its sharpness and its bypass. The bypass — the coarse material leaking into the fines — is the direct measure of the wasted regrinding energy, and its reduction is the separator’s main optimization target.

What is the normal circulating load of a finish mill?

150 to 300 percent of the fresh feed. The optimum is set by the circuit’s bottleneck — the elevator, the separator or the mill itself — and is found by the incremental test rather than by a fixed rule. The reject rate is the loop’s best load indicator.

How much can the joint optimization gain?

A complete joint campaign typically delivers 15 to 25 percent more capacity and 10 to 15 percent lower specific power on the same mill, with the operational steps — the charge re-grade, the separator restoration and the control tuning — delivering most of the gain at low investment.

How often must the optimization be repeated?

The loop decays continuously — the media wears, the vanes erode and the settings drift — so the campaign is an annual cycle, with the monthly KPI review and the change control sustaining the state between the campaigns.

Summary

The cement separator and the ball mill are one machine, and their joint optimization is one project: the loop of the mill, the separator, the reject and the product is measured completely, corrected on both sides and operated as a system. The separator’s classification science — the cut, the sharpness and the bypass, read from the Tromp curve — and the mill’s internal engineering — the charge, the liners and the diaphragm — are the two halves of the same measurement campaign, and the operating strategies, the aids, the control and the economics bind them together. The joint project delivers 15 to 25 percent more capacity and 10 to 15 percent lower specific power, and its discipline — measure before and after, optimize the loop not the machine, sustain the program — is the difference between the plants that collect the gain and the plants that leave it on the table. The loop is the machine, and the plant that masters the pair masters the economics of its grinding.

13. The Separator Types and Their Performance Characteristics

The separators of the cement grinding circuits belong to the three generations: the first-generation static separators (the gravity and the centrifugal units with the fixed geometry and the limited efficiency), the second-generation classifiers (the cyclone-type units with the air circulation and the moderate efficiency), and the third-generation dynamic separators (the rotor-type classifiers with the guide vanes, the variable speed and the controlled air flows). The third-generation separators deliver the classification efficiency of the 70-90% for the cement size ranges, the sharper cut sizes with the bypass of the 5-15%, and the adjustable fineness over the wide range: the separator performance is quantified by the Tromp curve (the partition curve of the classifier) with the cut size (the d50), the bypass and the sharpness index (the d25/d75 ratio), and the efficient classification is the key to the mill capacity and the product quality.

14. The Efficiency Measurement and the Tromp Curve Analysis

The separator efficiency measurement follows the sampling and the mass balance protocol: the samples of the feed, the fine product and the rejects are taken simultaneously with the air flow measurements, the size distributions of the three streams are determined, the internal circulating load is computed from the mass balance, and the Tromp curve is constructed from the partition factors of the size classes: the curve parameters (the cut size, the bypass, the sharpness) quantify the classification performance, and the comparison with the design values identifies the problems (the bypass increase from the seal air leakage, the cut size drift from the rotor wear, the sharpness loss from the vane settings). The complete efficiency test is the standard diagnostic of the separator, and the optimized separator performance (the minimal bypass, the target cut size, the maximum sharpness) directly improves the grinding circuit capacity and the product quality.

15. The Optimization Cases of the Separator and the Mill Combination

The optimization cases of the separator-mill combination demonstrate the achievable gains: the plant that replaced its second-generation separator with the third-generation unit gained the 15-25% capacity at the same Blaine with the improved PSD steepness; the plant that corrected the separator bypass (the seal air reduction from the 25% to the 8%) gained the 5-8% capacity and the 2-4 kWh/t energy reduction; the plant that tuned the rotor speed and the vane angles to the target d50 improved the strength stability and reduced the over-grinding; and the plant that balanced the mill ventilation with the separator air flows stabilized the outlet temperature and the filter load. The common pattern of the successful cases is the complete circuit view: the separator optimization is always the separator-and-mill optimization, and the measurement-driven approach delivers the documented results.

16. The Air Flows and the Sealing of the Separator System

The air flows and the sealing of the separator system determine its classification efficiency: the primary air enters through the guide vanes with the tangential velocity that distributes the feed, the secondary and the tertiary airs adjust the classification zone conditions, the fines leave with the air to the filter, and the rejects fall back to the mill: the air balance of the separator is the mass balance of the classification, and the seal air prevents the short-circuiting of the feed into the fines stream (the bypass): the internal bypass of the third-generation separators is typically the 5-15%, and each percentage of the bypass carries the fine material directly into the product without the classification. The optimization of the air flows includes the vane angle adjustments (the guide vane settings change the classification intensity), the rotor speed tuning, the damper balancing and the seal air pressure control: the complete air system of the separator is measured and adjusted together, because the classification efficiency is the result of the integrated aerodynamic design.

This file is part of the Complete Cement Technical Package (931 files) available from cementequipment.org. Respective rights holders; library copy for the licensed single user.