Most plants discover cement strength variation the same way: a customer complains, or a 28-day result lands lower than the one before it with no obvious explanation. The mix didn’t change. The kiln ran normally. The mill was on the same setpoint. And yet the number moved.
The frustrating part is that strength variation almost never has one cause. It is the accumulated effect of several small shifts, each individually within tolerance, that happen to line up in the same direction. Chasing it one parameter at a time rarely works, because the parameter you check first is usually not the one carrying the variation.
This article sets out where batch-to-batch variation actually comes from, roughly how much each cause is worth in MPa, and the order of investigation that resolves it fastest.
Before You Chase the Plant, Check the Test
This is the step that gets skipped, and skipping it wastes weeks. A meaningful share of apparent strength variation is measurement variation, not production variation. The standard test method itself is built around the assumption that individual results will scatter.
Under EN 196-1, a single strength result is the arithmetic mean of six compressive determinations taken from three prisms. The method then applies an outlier rule: if one of those six deviates by more than ±10% from the mean of the six, it is discarded and the remaining five are averaged. If a second result among those five again deviates by more than ±10%, the entire set is thrown out. A standard deliberately written to accommodate that much internal spread is telling you something important about how much confidence a single result deserves.
The conditions around the test matter just as much as the cement in it. Curing water has to sit at 20 ± 1 °C, specimens come out of the water no more than 15 minutes before testing, and the standard sand has to be certified, at least 98% silica, below 0.2% moisture, and graded within defined sieve limits. A curing tank drifting two degrees, sand from a new batch that was never checked, or prisms left on the bench for half an hour on a busy morning will all move results without anything changing in the plant.
So the first question is not what changed in the kiln. It is this: if you re-tested the same retained sample today, in your own lab, would you get the same number? Until you know the answer, you are debugging two systems at once.
What Actually Causes Cement Strength Variation Between Batches
Once the test is ruled out, the real causes sort into a rough order of magnitude. The sensitivities quoted below come from published cement-industry literature on strength optimisation and are useful for ranking causes, not for predicting your plant’s exact numbers. The response depends on your clinker, your SCM and your grinding circuit.
Clinker Source Mixing and Silo Segregation
This is the cause most often missed, because nothing in the process data shows it. A plant drawing from more than one clinker stockpile, running imported clinker alongside its own, or pulling from a silo that has stratified during a long storage period is feeding the mill a different material from one day to the next. Clinker from two sources can meet the same chemistry specification and still hydrate differently, because mineralogy and crystal size depend on burning and cooling history, not just oxide composition.
If your strength variation correlates with nothing in the kiln data, check what the mill was actually fed rather than what the clinker analysis said.
LSF Drift and Free Lime
Lime saturation factor controls how much C3S forms, and C3S governs strength development out to 28 days. Published sensitivities put a ±1% change in LSF at roughly ±5% on 1-day strength and ±1.5% on 28-day strength. That looks modest until you notice how easily LSF drifts: a change of quarry bench, a blending pile built in the rain, a corrective-material feeder running out of calibration.
There is a ceiling, though, and it catches people out. Pushing LSF higher does not keep buying strength. Raising it from 96 to 98 has been reported to add only 1 to 2 MPa, and beyond the point where the kiln can no longer combine the extra lime, free lime starts appearing instead. Free lime of +1% costs around 2% of 28-day strength, and a swing of +0.3% is worth roughly −0.6 MPa at 28 days and −1.2 MPa at one day. A kiln that is hunting, cycling between underburning and overburning, produces exactly this pattern, and the early-age result moves more than the 28-day result, which is a useful diagnostic signature in itself.
If LSF and free lime both look unstable, the root cause is usually upstream in raw mix control rather than in the kiln. Raw material investigation and characterisation is where that gets resolved, particularly where deposit variability was never fully mapped.
Clinker Cooling Rate
Two clinkers with identical chemistry can give different strength purely because of how fast they were cooled. Rapid cooling locks in small, reactive alite crystals. Slow cooling, especially lingering below roughly 1250 °C, allows C3S to revert towards C2S and free lime, which costs 28-day strength directly.
Cooling rate is not a setpoint you can read off the DCS, which is why it is easy to overlook. It is a consequence of cooler condition: grate plate wear, air beam condition, bed depth control and fan performance. A cooler that has degraded since the last shutdown is quietly changing clinker reactivity, and strength variation that tracks cooler performance is often blamed on grinding instead. Our note on the clinker cooler’s role in the pyro section covers what to inspect.
Sulphate Optimisation and Gypsum Dehydration
SO3 content has a clear optimum. Strength rises as sulphate increases, peaks, and then falls. Overshooting carries a risk of delayed expansion, not just lost strength. Published figures put +0.5% SO3 at around +1.0 MPa at 28 days within the rising range, which makes gypsum one of the more responsive levers available.
The complication is that the optimum is not fixed. It moves with clinker C3A, with alkali content and with fineness, so the gypsum setting that was right two years ago may not be right for the clinker you are producing now. Two further variables sit on top of that. Gypsum feed rate accuracy is one, since the sulphate actually entering the mill is often less stable than the recipe suggests. The second is the dehydration state of the gypsum itself: mill exit temperatures in the region of 90-115 °C produce partial dehydration to hemihydrate, which is what gives controlled setting. A mill running cold in winter and hot in summer is delivering a different sulphate species in each season, from the same gypsum.
Particle Size Distribution, Not Just Blaine
Blaine is the number most plants control to, and it is an incomplete proxy. What actually governs strength is where the mass sits in the distribution. Particles below about 3 µm drive one-day strength. The 3-25 µm band governs 28-day strength. Material coarser than about 25 µm contributes very little to strength at any age. It is mostly filler.
This means two cements at identical Blaine can give different 28-day strength if one has more of its mass in the strength-contributing band and the other has it split between ultrafines and coarse tail. Separator performance is the main control, and separator efficiency drifts as wear develops without changing the Blaine target at all. The relationship breaks down further when limestone or other SCM fines are present, because those fines raise measured surface area without adding reactive clinker surface. A rising Blaine with flat strength is a classic sign of exactly this.
If PSD is the suspect, the classifier and separator circuit is where to look first, and the grinding circuit configuration sets the limits on what distribution you can achieve at all.
SCM Variability
For blended cements, the supplementary material is frequently the least controlled input in the whole process, because it arrives from outside. Fly ash varies in fineness, loss on ignition, glass content and reactivity between sources and even between consignments from the same source, depending on what the power station was burning and how it was running. Slag varies in glass content and grinding behaviour. Limestone used as a filler contributes no reactivity at all beyond a small nucleation effect.
Two consequences follow. A change of SCM source will move strength even if the replacement percentage is identical, and a plant that increased SCM content to reduce clinker factor may have moved its strength distribution without realising how much of the margin it was spending. The relevant background sits in our pieces on SCMs, calcined clay and slag and on the clinker factor, and the applicable requirements are covered in BIS standards for blended cement.
Prehydration During Storage and Handling
This one deserves more attention than it usually receives, because the effect per unit is disproportionate. General loss on ignition of +1% is worth roughly −1.0 MPa at 28 days. But LOI arising specifically from prehydration, meaning cement that has picked up moisture and partially reacted in storage or transit, has been reported at around −1.7 MPa at 28 days for just +0.1% LOI.
The comparison is worth pausing on. Per unit of LOI, prehydration damage is an order of magnitude more severe than ordinary LOI, because the moisture has already consumed reactive surface before the cement ever reaches a mixer. Humid monsoon conditions, a silo with a compromised aeration system, long storage before dispatch, or bagged product held in a damp godown will all produce the pattern plants find most confusing: cement that tested fine at production and tests low weeks later.
Alkali Content
Alkalis behave in a way that catches out anyone optimising for a single age. Published sensitivities show +0.1% Na2O equivalent producing around +1.2 MPa at one day, +0.8 MPa at three days, and +0.4 MPa at seven days, and then −0.8 MPa at 28 days. The sign flips.
So a plant that has been chasing better early strength, perhaps for a precast customer, may have been eroding its 28-day performance at the same time. Alkali input comes from the raw materials and from the fuel, which is why increasing alternative fuel substitution can shift the alkali circuit and therefore the shape of the strength curve. If your early-age results have improved while 28-day results have slipped, alkalis belong near the top of the suspect list.
How Much Variation Is Normal
There is no universal figure, and anyone quoting one without knowing your plant is guessing. What matters is that you establish your own baseline and then judge against it.
Practically, that means computing the standard deviation of 28-day strength from your own production history over a meaningful period, plotting results on a control chart rather than comparing one batch to the last one, and separating within-batch scatter from between-batch scatter so you know which you are dealing with. A single low result inside your established spread is noise. A run of results trending in one direction is a cause. Reacting to the first is how plants end up chasing variation they created themselves by adjusting too often.
Two further points are worth building into the way results are read. Your dispatch strength needs enough margin above the grade requirement to absorb normal variation, and that margin is a commercial decision about risk, not a technical constant. And the compliance figure that matters is the one at the point of use, not the one at production, which is why the prehydration question above is a quality-control issue rather than a logistics footnote.
A Diagnostic Order That Works
- Re-test retained samples. Confirm the variation is real before investigating the plant. Check curing water temperature, sand certification and test timing at the same time.
- Establish what the mill was actually fed. Clinker source, stockpile, silo drawdown. Not the clinker specification, but the physical material.
- Plot free lime and LSF against the strength results. If early-age strength moves more than 28-day strength, kiln stability is implicated.
- Check cooler condition. Grate plates, air beams, bed depth, secondary air temperature trend since the last shutdown.
- Verify sulphate, not just gypsum feed rate. Measured SO3 in the cement, plus mill exit temperature as a proxy for dehydration state.
- Run a full PSD, not a Blaine. Look at the 3-25 µm fraction specifically, and at separator performance.
- Audit the SCM. Source, consignment-level LOI and fineness, actual replacement percentage versus nominal.
- Check storage and dispatch. Silo aeration, storage duration, moisture exposure, and strength at point of use versus at production.
Most plants find the cause within the first four steps. The reason the order matters is that steps 5 to 8 are where the expensive interventions sit, and arriving there without ruling out the earlier causes usually means solving the wrong problem well.
Where variation persists across several of these at once, it is normally a symptom of process control capability rather than a single defect, and it is addressed through a structured review of the raw mix, pyro and grinding sections together rather than parameter by parameter. That is the work covered by a cement plant technical and energy audit, and the process engineering side is where the corrected operating basis gets established.
FAQs:-
How much batch-to-batch variation in 28-day strength is normal?
There is no single correct figure, because it depends on your clinker consistency, SCM control and grinding circuit. The useful approach is to calculate the standard deviation of 28-day strength from your own production history and treat that as your baseline. A result inside your established spread is noise; a sustained trend in one direction is a cause worth investigating. Plants that react to every individual low result usually end up adding variation rather than removing it.
Can the strength test itself cause the variation I am seeing?
Yes, and it should be ruled out first. Under EN 196-1 a result is the mean of six determinations from three prisms, with an outlier rule that discards any determination deviating more than ±10% from the mean of the six. Curing water must be held at 20 ± 1 °C, specimens must be tested within 15 minutes of removal from water, and the standard sand must be certified and below 0.2% moisture. A drifting curing tank, uncertified sand, or prisms left out too long will all move results with nothing wrong in the plant.
Why does our cement meet 28-day strength but fail on early strength?
Early and late strength are controlled by different things, so they can move independently. One-day strength is driven mainly by the fraction below about 3 µm, by C3A, by alkali content and by sulphate availability. Twenty-eight-day strength is governed by C3S and by the 3-25 µm fraction. Free lime and LSF swings also hit early strength harder than 28-day strength, which makes a divergence between the two ages a useful clue that kiln stability rather than grinding is the problem.
Does a higher Blaine always mean higher strength?
No, and assuming it does is a common source of confusion. Strength comes from the mass sitting in the 3-25 µm band. Material coarser than about 25 µm contributes very little, and ultrafines below 3 µm mainly affect early strength. Blaine can also rise without any strength benefit when limestone or SCM fines increase measured surface area without adding reactive clinker surface. Rising Blaine with flat strength usually points to separator performance or to the filler content rather than to the mill.
Why did our strength drop after we increased fly ash content?
Partly because clinker was replaced with a less reactive material, which is expected and should have been quantified in advance. But the variation that follows is often caused by something else: fly ash quality varies between sources and between consignments in fineness, loss on ignition and glass content. A plant that increased replacement percentage and also changed source has made two changes at once. Consignment-level testing of the SCM is what separates the two effects.
How does clinker cooling affect cement strength?
Cooling rate determines the crystal size and reactivity of the alite formed in the kiln. Rapid cooling preserves small, reactive crystals. Slow cooling, particularly dwelling below roughly 1250 °C, allows C3S to revert towards C2S and free lime, which reduces 28-day strength. Because cooling rate is a consequence of cooler mechanical condition rather than a setpoint, deteriorating grate plates or air beams can change clinker reactivity gradually without appearing anywhere in the operating data.
Can cement lose strength during storage?
Yes, through prehydration, and the effect is larger than most expect. Cement that picks up moisture in storage or transit partially reacts before use. Published figures put just +0.1% loss on ignition arising from prehydration at around −1.7 MPa at 28 days, compared with roughly −1.0 MPa for a full +1% of ordinary LOI. This is why cement can test correctly at production and test low weeks later, and why silo aeration, storage duration and humidity exposure belong in quality control rather than only in logistics.
Need Your Strength Variation Diagnosed Properly?
TECHCEM carries out technical and energy audits for cement plants across India, covering raw mix control, pyro-section stability, grinding circuit performance and quality control practice together, because strength variation rarely sits in one of them alone. See how our cement plant technical and maintenance audit services work, or call +91 81046 42385 to discuss your plant.