Mills Circulating Load Limestone Blaine: Complete Guide & Do

Subtitle: Mass Balance of the Separator, the Tromp Curve, and the Blaine Calibration of Limestone Content — Cementequipment Package

Among the least glamorous and most profitable subjects in all of cement manufacturing is the closed-circuit grinding of the finish mill, the loop in which the ball mill grinds, the separator classifies, the finished material leaves for the silos, and the oversize returns to be ground again. The numbers that govern that loop, the circulating load, the separation efficiency, and the fineness of the product, decide how many kilowatt-hours each ton of cement costs, how much the grinding equipment produces, what the particle size distribution of the cement looks like, and therefore what the cement does in the concrete, and yet the full quantitative treatment of the loop, the mass balance around the separator, the Tromp curve, and the interplay of the fineness with the composition of the blend, especially the limestone content, is the professional territory of the grinding specialist. This article is written as the complete technical companion to the Mills circulating load Limestone blaine file (Mills circulating load Limestone blaine (CMB) – Cementequipment.org.rar) in the cementequipment.org package, and it develops exactly that territory: the mass balance of the closed circuit and the definition of the circulating load, the separator and its efficiency expressed through the Tromp curve, the Blaine fineness and its measurement, and the calibration of the Blaine against the limestone content of blended cement, which is one of the most practical and least documented problems of quality control in the modern cement plant.

The motivation for the article is the observation, sharpened by every mill audit, that most finish grinding circuits are running measurably below their potential because the operators and even the engineers are working from rules of thumb rather than from the balance. The circulating load is the most abused number in the circuit, computed differently by different handbooks, interpreted against different scoring, and rarely measured with the sampling rigor the balance demands. The separator is tuned by feel, by the residue reading and the tailings, without the Tromp curve that would separate the genuine classification sharpness from the by-pass and the short-circuit. And the Blaine, the universal daily fineness number, is interpreted against the limestone content of the blend as if limestone were inert filler, when in fact the limestone grinds differently, packs differently, and changes the Blaine and the behavior of the cement in ways that must be calibrated, not assumed. This article corrects all three habits with the worked arithmetic, the measurement discipline, and the operating craft that the package’s grinding files and mill audit material embody.

The Closed Circuit and Its Parties

The finish grinding circuit is a loop with three actors, and its description is best started from the flows. The mill, a ball mill in the classic circuit, receives the new feed, the mixture of clinker, gypsum, and the other constituents such as limestone, slag, or pozzolan that the cement recipe calls for, and it grinds everything it receives, its output being a mixture of finished material, fines, and unfinished material, coarse particles, all mixed together. The separator, the classifier, then receives the entire mill discharge and divides it: the fine fraction, the finished cement, leaves as the product; the coarse fraction, the tailings or reject, returns to the mill inlet as the circulating load. The two flows around the separator, the return and the product, and the internal flow that connects them, form the closed loop, and the entire behavior of the circuit is expressed by the ratios and the balances among these flow rates. The circuit is closed in the strict sense: over a steady period, what enters the loop as new feed must leave it as product, and everything the separator returns to the mill is ground again, so the mill sees not the new feed flow but the total feed flow, new feed plus return, and it is that total, higher than the new feed, that sizes the mill and sets its efficiency.

The purpose of the closed circuit is sharp classification, and its logic is the economics of grinding. Because the energy to grind a particle grows steeply as the particle shrinks, it is wasteful to keep grinding particles that are already fine enough: the separator allows the mill to stop grinding the fine material once it reaches the target, concentrating the mill’s energy on the unfinished coarse, and the result is a higher production rate at a given fineness, or a finer product at a given production, than an open circuit that grinds everything once to nothing. The closed circuit also gives the operator the control of the product, because the separator settings, the rotor speed, the vane angle, the airflow, directly set the fineness cut, so the operator changes the separator rather than the mill charge to steer the product. The circulating load is the traffic of that loop, and the next sections give it its exact definition and its arithmetic.

The Mass Balance and the Definition of Circulating Load

The circulating load is derived from the mass balance around the separator, and the derivation is a model of how the discipline works. Let the new feed to the circuit be F, in tons per hour, let the mill discharge be the separator feed S, let the product, the finished cement, be B, and let the reject returned to the mill be R. The balance of the material entering and leaving the separator is S equals B plus R, and the balance of the whole circuit is that the new feed F equals the product B at steady state, since the material is neither created nor destroyed and the loop stores no net material. The circulating load is conventionally defined as the ratio of the reject flow R to the new feed F, or, equivalently, as the ratio of the total feed to the mill to the new feed, and it is usually stated as a number greater than one, a percent, or a multiplier: a circulating load of 2, for example, meaning that the reject returning to the mill is twice the new feed, which is an active but by no means unusual figure, with many circuits running between 1.5 and 3 or more depending on the design and the fineness.

The measurement of the three flows is the practical discipline. The product flow B and the new feed F are routinely measured by the belt weighers and the inventory, but the reject R is the flow that must be sampled and weighed, or derived from the fineness. The classical derivations use the fineness or the residue of the three streams: if the separator feed S, the product B, and the reject R have known residues or known fineness, then the reject can be obtained from the balance in fineness, R equals B times (feed residue minus product residue) divided by (reject residue minus feed residue), a formula that is exact when the residues are consistent and that is the workhorse of the balance. The sampling itself, the grab at the reject outlet, the divider, the careful timing so that the grab represents the steady flow, is where the balance is won or lost, and the package’s audit files drill the reader in the discipline that a good circulating load number comes from a good sample, not from a good formula. Once R is known, the circulating load follows, and with it the total mill feed, S equals B plus R, and the specific energy of grinding, the total feed divided by the power, which is the true measure of how hard the mill is working.

What the Circulating Load Tells the Engineer

The number itself is diagnostic, and the engineer who reads the ratio correctly learns more about the circuit than the ratio alone suggests. A low circulating load, other things equal, means the separator is sending little back, which for a given mill output means the product is coarser, because in a closed circuit the mill power is shared by less material and the mean size of the product shifts; a high circulating load means the separator is returning a lot, which generally allows a finer product and a higher mill output at the same fineness, but only up to the limit where the mill and the separator and the elevator become the bottleneck, at which point oversize metal, plugging, and power waste take over. The optimum circulating load is therefore an economic optimum, found at the point where the mill, loaded with the total feed, works at its most efficient filling and where the separator, stressed by the return flow, still classifies sharply, and the experienced grinding engineer is the one who knows, for each circuit, where that optimum lies and how to detect it: in the mill power draw, in the mill outlet temperature, in the product residue and the tailings residue, and in the Tromp curve, which the next section develops.

The relationship of the circulating load to the mill operation is close and mutual. As the separator return grows, the mill receives more feed and its load, its filling, its power, and its internal classification all respond, and the mill can be run harder, near its rated power, which is the efficient way to run a ball mill, because the mill’s power is best utilized when it is grinding a fully loaded bed. The danger is the runaway: if the fineness demand rises and the operator responds only by throttling the new feed, the circulating load can climb without the mill having the internal capacity, and the mill floods, the outlet temperature rises, the separator chokes, and the product quality degrades even as the load indicators rise. The tuning of the circuit, therefore, is a balancing of the three dials, the new feed, the separator rotor, and the mill charge, held so that the mill runs in its efficient window and the separator runs below its choking point, and the circulating load is the scoreboard of that balance. The article returns to that tuning after the Tromp curve, because the curve is the instrument that makes the tuning evidence-based.

The Tromp Curve: Reading the Separator’s Truth

The Tromp curve, named after the Dutch professor who introduced it, is the separator’s report card, and it is the single most important analysis the grinding engineer can perform. The curve plots, for every particle size, the percentage of the particles of that size that leave the separator with the coarse reject, and its shape narrates the whole behavior of the classification. A perfect separator would return every particle above the cut size and send every particle below it to the product: the curve would be a vertical step at the cut. No real separator achieves that, and the Tromp curve of a real machine rises from a finite base, the by-pass, the percentage of even the finest particles that ride with the coarse stream, up through the cut region to 100 percent at the coarse end. Two numbers and two features of the curve carry the diagnosis: the cut size, the diameter at which half the particles go to the reject, which is the separator’s set point, and the sharpness, the steepness of the curve through the cut, which measures the quality of the classification. A steep curve is sharp, close to the ideal; a flat, gentle curve is dull, mixing the fine and the coarse.

The by-pass deserves its own explanation because it is the most influential property of real separators. The by-pass is the fraction of the smallest particles that go directly to the reject without ever being separated, and its cause is physical: a share of the feed, the fines that are trapped in the coarse flow, the material that enters in the wrong stream, the dust that follows the returning air, rides through the separator without being presented to the classifying field. A high by-pass, of the order of 30 to 50 percent in poor classifiers and much less in the good high-efficiency machines, means that a large share of the finished material is being returned to the mill to be ground again, inflating the circulating load, wasting the power, and coarsening the product’s effective distribution, and the reduction of the by-pass is the central achievement of the third-generation high-efficiency separators and the most valuable tuning target of the older machines. The construction of the Tromp curve requires the fineness analysis, typically by sieve or by laser, of the separator feed, the product, and the reject, and the plant laboratory that performs this analysis on a schedule, and the engineer who interprets it, hold the diagnostic key to the circuit that the daily residue reading cannot provide.

Classification Efficiency and the Sharpness of the Cut

From the Tromp curve, the engineer extracts the quantitative measures of classification efficiency that let circuits be compared and tuned. The sharpness index, or imperfection, is computed from the curve: the sharpness is the ratio of the cut size to the size at which 25 percent of the particles report to the reject, or, equivalently, the imperfection is the difference between the 75 percent and the 25 percent sizes divided by twice the median, and a good high-efficiency separator achieves a sharpness of 0.8 or more and an imperfection of the order of 0.1 to 0.3, where the older mechanical separators were notably duller. The mass balance itself provides another check: the efficiency of the separation can be computed as the ratio of the effective to the theoretical separating area, or simply verified by whether the mass balance closes, whether the feed computed from the product and the reject matches the measured feed, and the failure of the balance to close within a few percent is itself a diagnostic, pointing to an unmeasured flow, a sample that was not representative, or an internal accumulation in the circuit.

The practical reading of the sharpness is direct. A dull classification returns a significant share of finished, fine material in the reject, inflating the circulating load, wasting the grinding energy on re-grinding, and coarsening the factory’s perception of the product, all of which the sharp separator avoids. The tuning knobs of the separator, in the third-generation machines, are the rotor speed and the guide vane angle, which set the cut and the sharpness together, and the airflow, which interacts with both, and the engineer who reads the Tromp curve knows whether a given set of settings is genuinely sharp or merely fast, whether the by-pass is low or the air is short-circuiting, and whether the machine is at its design point or being throttled by a plugged vane or a blocked inlet. The curve, in short, turns the separator from a black box into an instrument, and the reading of its shape is the skill that separates the grinding specialist from the operator who only watches the residue.

Blaine: The Universal Fineness Number

The fineness of the product is measured daily by the Blaine number, the air-permeability specific surface area in square meters per kilogram, and the Blaine deserves its own careful treatment because it is the most-read fineness statistic in the industry and the most misinterpreted. The Blaine method, standardized in the air-permeability test, measures the time a fixed volume of air takes to pass through a compacted bed of cement of known volume and porosity: the finer the cement, the larger its surface area, the more it resists the air flow, and the longer the time, from which the specific surface is computed by the standard formula. The method is fast, cheap, repeatable to about plus or minus a few percent in the hands of a trained operator, and entirely suitable for a daily control loop, and its value is that it responds to the total surface area, so it is a genuine index of the fineness. Its limitations are equally real: it is a one-number, average quantity, blind to the shape of the particle size distribution, so two cements with the same Blaine can have different distributions and different performance, which is exactly why the modern practice reads the Blaine together with the 45 micron residue and the laser-derived distribution, as Part 1 of the Innovations series described.

The Blaine is tied to the separator in a practical way: in a sharp closed circuit, the Blaine is largely set by the separator’s cut, because the product leaving a sharp separator carries the fines and the reject carries the coarse, and the operator steers the Blaine with the rotor and the vane. The relationship, however, is not one-to-one, because the mill condition, the charge, the media, the ventilation, and the temperature change the shape of the distribution that the separator receives, and the same separator setting can deliver a different Blaine on different days. The daily discipline, therefore, is to hold the Blaine by adjusting the separator against the residue and the distribution, to spot the drift before the quality department flags it, and to appreciate what the Blaine is buying: higher Blaine, more surface, faster hydration, higher early strength, but also higher water demand and a higher grinding cost, so the Blaine target is an economic optimum set by the market, the clinker burnability, and the grinding energy, and the plant that understands that optimum does not chase the Blaine upward beyond its value.

Limestone in the Finish Mill: A Different Grind, a Different Role

The modern finish mill blends limestone into the cement, and the limestone is far from an innocent filler, so its behavior in the circuit must be understood before its calibration can be discussed. The limestone for the cement, a finely ground calcium carbonate with a high carbonate content and a low loss, is added as a clinker replacement in the composite cements, and its roles are physical and chemical. Physically, the fine limestone contributes to the particle packing of the paste, fills the voids, and, through its presence in the fine fractions, modifies the water demand and the rheology of the concrete. Chemically, the fine limestone participates in the hydration: it reacts with the aluminate phases, forming carbo-aluminate hydrates, it accelerates the early hydration of the alite by providing nucleation sites, and it acts as a micro-filler that densifies the paste, so that a limestone-bearing blend can match or improve the early strength and the durability of the plain clinker at a lower cost and a lower carbon footprint. The reason the limestone content cannot be treated as an inert dilution, and the reason it clouds the Blaine reading, is precisely this participation: the limestone grinds much more easily than the clinker, it concentrates in the fine fraction of the product, and it changes the surface area and the water demand out of proportion to its mass.

Because the limestone grinds faster than the clinker, the inter-ground product is not a uniform blend but a distribution in which the limestone is over-represented in the fines and the clinker in the coarse, and this segregation, which the separator reinforces, means that the same Blaine result corresponds to different distributions as the limestone content changes. The practical consequences are everywhere: a cement with a higher limestone content at the same Blaine has a different water demand, a different particle packing, and a different strength development than the same Blaine pure-clinker cement, and a control system that reads only the Blaine can be fooled into thinking the fineness is stable while the true distribution, and the product behavior, have drifted. The modern practice therefore controls the limestone blend against the full distribution, or at minimum reads the Blaine and the residue together with the actual limestone content, and it is at exactly this point that the calibration problem of the next section becomes urgent.

Blaine Calibration of Limestone Content

The calibration of the Blaine against the limestone content is one of the most practical and least documented problems in cement quality control, and the file in the package is devoted to exactly this work. The problem is stated plainly: in a plant that adds limestone to the blend, the laboratory’s Blaine result is a function of two variables at once, the true fineness of the grinding and the limestone content of the blend, and the operator who adjusts the separator to hold the Blaine can be chasing a shadow, because a rise in the limestone content, which grinds fine easily, will raise the Blaine without any change in the grinding, and a fall will lower it. The standard industrial practice is a calibration curve or, better, a calibration model: sample the product over the practical range of the limestone content, measure the actual oxide or carbonate content, the limestone by XRF or by the loss on ignition or by the carbonate titration, and the fineness of the distribution, and fit the relationship, so that a Blaine reading can be corrected to the standard limestone content or interpreted for what it really means. The correction is not cosmetic; it is the difference between controlling the product and controlling the illusion.

The engineering of the calibration deserves the detail the file gives it. The samples must cover the corners of the operating space, the low and the high limestone and the low and the high fineness, because a linear fit through a narrow range will not extrapolate to the edges; the limestone content must be measured by a method, the XRF calcium plus the carbonate-carbon or the loss on ignition, whose own error is smaller than the effect being calibrated; and the reference fineness should include the distribution parameters, not the Blaine alone, because the limestone, being soft, changes the shape of the distribution as well as its surface. The resulting calibration is then used two ways: the plant can read, from a new Blaine and a new limestone content, whether the grinding has truly drifted, and it can predict what Blaine the separator should be set to deliver at the current limestone content to hold the true fineness on target. The file presents the arithmetic, the sampling plan, and the reconciliation, and the article’s message is that this calibration, done once and reviewed, converts the Blaine from a number that fools the plant into a number that serves it.

A Worked Balance and Calibration in Numbers

To make the arithmetic tangible, follow one realistic circuit in round numbers. Suppose the new feed F to the finish mill is 100 tons per hour of a blend of 92 percent clinker and 8 percent limestone, so the limestone feed is 8 tons per hour. The separator feed S is found by sampling to include a reject R of 200 tons per hour, giving a circulating load of 2.0, and the total feed to the mill is therefore 300 tons per hour, the mill’s true traffic. The mill draws, say, 4,500 kilowatts at the separator feed rate, giving a specific energy at the mill inlet of 4,500 divided by 300, 15 kilowatt-hours per ton of mill feed, while the circuit’s specific energy counting the product is 4,500 divided by 100, 45 kilowatt-hours per ton of new feed; the difference between the two is the essence of the closed loop, the mill grinds everything, so the product’s energy is paid on everything. The Tromp curve of the separator, built from the fineness of the feed, product, and reject, shows a cut size of about 25 microns and a sharpness of 0.85, with a by-pass of about 12 percent, and the engineer reads that by-pass and knows there is reform to be had in the separator before the mill is blamed.

For the Blaine calibration, suppose the plant’s calibration shows that a one percentage point increase in the limestone content raises the Blaine by about 4 square meters per kilogram at the same grinding, while a one point increase in the limestone lowers the residue negligibly in the coarse range because the clinker dominates it. If the limestone content rises from 8 to 10 percent and the Blaine reads the same, the plant must infer that the true fineness of the clinker has actually fallen, and the separator must be tightened to recover it; and if the Blaine rises by 8 while the limestone rose by 2 points and the separator was untouched, the plant knows the rise is the limestone’s signature, not a real gain in grinding, and no action is owed. These small reconciliations, repeated daily, are the working life of the calibration, and the rounding of the numbers in this example, the exact sample values in the file and the plant’s own data, do not change the method: measure the flows, build the curve, calibrate the fineness, and reconcile the number against the composition, shift after shift after shift. That is the entire professional discipline of the circuit, and it is the discipline these sections have, in each of its parts, been teaching.

Operating Practice and the Mill Audit

The article closes its technical development with the operating practice that holds the circuit at its optimum, and with the audit that re-finds the optimum when it has drifted. The daily practice is the closed loop of control: set the new feed and the separator by the Blaine and the residue, watch the mill power at its full-load window, watch the mill outlet temperature against the evaporation and the ventilation, and watch the reject flow and the product as the signatures of the balance. The weekly practice is the sampling: the Tyler bank or the Alpine or the laser along the three streams, the rebuild of the Tromp curve, the recomputation of the circulating load, and the reading of the by-pass, so that the drift of the separator, the blade wear, the vane creep, the air leak, is caught in the curve before it is felt in the product. The monthly practice is the audit: the full balance, the mill internals inspection, the charge level and the media grading by the dented ball and the charge percentage, the liner profile, and the mill mezzanine and the ventilation, all reconciled against the power and the output, and the conclusions fed back as the new operating target for the next month.

The audit deserves its own last word because it is the discipline that makes all the instruments of this article pay. A well-instrumented plant is, every day, holding its Blaine within a band and its circulating load within its window, and the audit is what asks the deeper question, whether that band and that window are the right ones, whether the separator by-pass is being tolerated instead of reduced, whether the circulating load has been allowed to creep up in the defense of a fineness that the product never required, and whether the mill charge, the media, and the lining are still earning their power. The audit compares the circuit against the mass balance, the ball charge against the power draw and the Bond work index of the feed, the separator against its design Tromp, and the product against the market’s requirement, and the result is the correction that saves the plant its kilowatt-hours. This is the exact purpose of the Mill audit file in the package, and the lesson is the same as in every file of the library: the plant that measures its circuit, reads its curves, and audits its balance is the plant that grinds its cement at the true minimum cost, and the circulating load, the Tromp, and the Blaine are the instruments of that profit.

Reference Data Table

The following table collects the definitions and order-of-magnitude values used throughout this article, so the reader can hold the subject’s numbers in one view.

Quantity	Symbol / Definition	Typical Value / Range
New feed	F (t/h)	Plant rating
Separator feed (mill discharge)	S = B + R (t/h)	New feed times (1 + circulating load)
Product	B = F (t/h, steady state)	Equals new feed
Reject	R = S – B (t/h)	Derived from fineness balance
Circulating load	CL = R / F	~1.5 – 3 (and higher)
Circulating load from residues	R = B · (feed% – prod%) / (rej% – feed%)	Dash when residues consistent
Tromp cut size	d50 at 50% to reject	~15-40 μm typical
Sharpness of cut	d25 / d50, or imperfection	Good: >0.8 sharpness; imperfection ~0.1-0.3
By-pass	% fines to reject	Good high-efficiency: <15-20%
Blaine specific surface	Air permeability, m²/kg	~280-450 m²/kg
Limestone Blaine shift	ΔBlaine per % CaCO3 at same grind	Plant-calibrated, e.g. ~4 m²/kg per %

Frequently Asked Questions

What exactly is the circulating load and why does it matter?

It is the ratio of the separator reject, the material returned to the mill, to the new feed, and it is the direct expression of the closed-circuit mass balance. It matters because it decides how much material the mill must grind per ton of product, how the mill power is utilized, and how fine the product can be made, and because the difference between the circuit’s energy per ton of mill feed and per ton of product is the whole point of closed-circuit grinding.

How do I measure the circulating load without a scale on the reject?

From the fineness balance around the separator. With the residues, or better the fineness distributions, of the feed, product, and reject, the reject flow is derived exactly, R equals B times (feed residue minus product residue) divided by (reject residue minus feed residue). The accuracy of the result is set by the quality of the three samples, so the sampling discipline is the real measurement.

What does the Tromp curve tell me that the residue does not?

The residue tells you the state of one sieve class; the Tromp curve tells you how the separator behaves at every size, showing the cut size, the sharpness of the classification, and the by-pass, the share of fines that ride the reject. It distinguishes a dull, by-passing classifier, which inflates the circulating load and wastes power, from a genuinely sharp one, and it shows which setting to change and why.

Why can’t I just hold the Blaine and trust it?

Because the Blaine is a one-number average that is blind to the distribution shape and, in blended cement, to the composition. The limestone grinds fine and easily and raises the Blaine without any real change in the clinker fineness, so a Blaine-based control held alone can chase the composition’s shadow. Read the Blaine with the residue, the distribution, and the calibrated composition, and you control the reality, not the mirage.

What is the first step to improve a dull-looking grinding circuit?

Measure before changing anything: sample the three streams, build the Tromp curve, close the mass balance, and read the by-pass and the circulating load. In most real circuits the first reform is in the separator, the by-pass and the sharpness, not in the mill charge, and the evidence of the curve shows it. Change one thing, remeasure, and let the curve decide whether the change helped.

Summary

This article has developed the complete technical treatment of the closed-circuit finish mill as delivered in the Mills circulating load Limestone blaine file of the cementequipment.org package: the mass balance of the circuit and the exact definition of the circulating load, the separator and its performance read through the Tromp curve, the sharpness and the by-pass that decide the true quality of the classification, the Blaine fineness and its measurement and its limitations, the limestone and its different grind and different role in the blend, and the calibration of the Blaine against the limestone content that has been the article’s central practical problem. It worked the marriage of the balance and the calibration in concrete numbers, with a realistic circuit showing the circulating load, the Tromp curve, and the limestone-shifted Blaine reconciled to the true grinding, and it closed with the operating practice and the audit that hold the circuit at its optimum and re-find it when it drifts. Throughout, the discipline was the library’s own: measure the flows, build the curve, calibrate the fineness, and reconcile the reading against the composition, shift after shift, until the circuit tells the truth.

The enduring lesson of the circuit is that its profit is made in the small, well-measured numbers: the percentage points of by-pass reduced, the points of sharpness gained, the kilowatt-hours that the balance and the calibration expose as waste, and the Blaine that is finally read for what it means rather than what it shows. The engineer who masters the mass balance, honors the Tromp curve, and calibrates the Blaine against the limestone is the engineer who grinds the cement at its true cost and guards its true quality, and that is the standard this library teaches, with the circulating load as its arithmetic and the calibration as its honesty.

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.