Concrete Cube Test: Compressive Strength Procedure and IS 456 Acceptance Criteria (India, 2026)

The concrete cube test measures the compressive strength of concrete by crushing 150 mm cubes in a compression testing machine at 28 days. In India the test method is IS 516 (Part 1/Sec 1):2021, the making and curing of the cubes is IS 1199 (Part 5):2018, the sampling of fresh concrete is IS 1199 (Part 1):2018, and acceptance or rejection is decided by IS 456:2000 Clause 16. Concrete passes only when the mean of any group of four non-overlapping consecutive test results is at least fck plus 0.825 times the established standard deviation, or fck plus 3 N/mm², whichever is greater, AND every individual test result is at least fck minus 3 N/mm².

That last sentence is where most site engineers get it wrong, and where almost every article on the Indian internet is out of date. Read on.

First, Fix Your Code References#

Before anything else: if your Quality Assurance Plan, your ITP, or your lab's report header still says "as per IS 516:1959", it is quoting a code that BIS has been dismantling since 2018. Here is the current picture, verified against the standards themselves.

What actually happened to IS 516:1959#

BIS split IS 516 into 12 parts. The foreword to IS 516 (Part 1/Sec 1):2021 is precise about the transition, and worth quoting exactly:

"These test methods shall be applicable in place of the corresponding provisions given in IS 516 : 1959 'Method of tests for strength of concrete' and IS 5816 : 1999 'Methods of test for splitting tensile strength of concrete (first revision)'. IS 516 : 1959 shall be superseded after the publication of all the parts of the standard. IS 5816 : 1999 shall stand withdrawn after the publication of this standard."

So the honest, accurate statement is this: for compressive strength testing, the 2021 Part 1/Section 1 provisions apply in place of the 1959 provisions right now. The 1959 document is formally superseded only once every part is published. Either way, there is no defensible reason to run a compressive strength test to the 1959 method in 2026.

The one big surprise: making and curing moved out of IS 516#

This trips up more engineers than anything else. IS 516 (Part 1/Sec 1):2021 tells you how to crush a cube. It does not tell you how to make one. Clause 3.1 hands that job to IS 1199 (Part 5). So if your method statement says "cubes shall be cast and cured as per IS 516", it is pointing at the wrong document.

The loading rate myth, settled#

You will find blogs claiming the new code changed the loading rate to 0.2 to 1.0 MPa/s. That range is from EN 12390-3 and ISO 1920-4, not from the Indian standard. IS 516 (Part 1/Sec 1):2021, Clause 3.5.2, says:

"The load shall be applied without shock and shall be increased continuously at a constant rate of 14 N/mm²/min until no greater load can be sustained."

Note what this means. The old IS 516:1959 rate of 140 kgf/cm²/min converts to about 13.7 N/mm²/min. BIS effectively kept the same rate and simply expressed it in SI units. The rate did not change; the units did. For a 150 mm cube the platen load should therefore rise at roughly 315 kN per minute, and for a 100 mm cube at roughly 140 kN per minute. A lab operator who slams the load through in 20 seconds is inflating your result; one who dribbles it in over five minutes is depressing it.

The Full Cube Test SOP, Step by Step#

Step 1: Take the sample correctly (IS 1199 Part 1:2018)#

This is the step that most often ruins a test before a single cube is cast.

IS 1199 (Part 1):2018 distinguishes a composite sample from a spot sample and is blunt about it: "Spot samples are not representative of the batch and should not be used to make strength specimens." If your engineer walks up to a truck, scoops one bucket from one place, and casts cubes from it, that is a spot sample and the code says it should not have been used.

A valid composite sample requires:

• A minimum of 0.02 m³ of concrete for strength specimens.
• Increments taken from at least four points, uniformly distributed through the batch.
• Discarding the first and last 10 to 15 percent of the discharge from a mixer or transit mixer.
• Increments taken across the whole width and thickness of a falling stream, never from a part of the concrete that looks segregated.
• Thorough remixing in a non-absorbent container before the moulds are filled.

IS 456:2000 Clause 15.2.1 layers a further requirement on top: the sampling must be random, spread over the entire period of concreting, and cover all mixing units. Casting all your cubes from the third truck of the morning because that is when the engineer was free is not random sampling.

If your maximum nominal aggregate size exceeds 40 mm, IS 1199 (Part 5) Annex B requires wet sieving before you cast.

Step 2: Run the slump test (IS 1199 Part 2:2018)#

The slump test is not a formality; it is your first line of defence, because it is the only strength-related check you get before the concrete is already in the shutter.

Key parameters from IS 1199 (Part 2):2018:

Only a true slump is a valid result. If the specimen shears or collapses, the code requires you to take another sample and repeat the entire procedure. Recording a collapsed slump as "220 mm" is not a test result.

For how slump ties back into the mix itself, see our guide to concrete mix design as per IS 10262.

Step 3: Prepare the moulds#

Moulds must conform to IS 10086. Before use:

• Clean every face and the base plate. Old hardened concrete on a face makes the cube out of square and out of tolerance.
• Apply a thin film of mould oil or release agent. Thin. A thick smear of burnt oil migrates into the cube face and softens it.
• Bolt the mould up tight. A mould that leaks grout at the joints loses fines and water and produces a honeycombed, low-strength cube.
• Sit the mould on a rigid, level, vibration-free surface.

Step 4: Cast the cube (IS 1199 Part 5:2018)#

Which size?

IS 1199 (Part 5):2018 Clause 4.1.1 is clear: the test specimens shall be 150 mm cubes. If the largest nominal size of aggregate does not exceed 20 mm, 100 mm cubes may be used as an alternative. So the 100 mm cube is permitted, conditionally.

An important honesty note: neither IS 1199 (Part 5):2018 nor IS 516 (Part 1/Sec 1):2021 prescribes any conversion or correction factor between a 100 mm cube result and a 150 mm cube result. The 100 mm cube result is the result. Any "0.95 factor" you have seen quoted is from international literature, not from the Indian standard, and you should not apply it to an IS 456 acceptance calculation without a written instruction from the engineer-in-charge.

Filling and compaction:

• Place concrete with a scoop in layers of approximately equal depth, each layer not more than 50 mm thick. A 150 mm cube therefore takes three layers. A 100 mm cube takes two.
• Move the scoop around the top edge of the mould as the concrete slides off it, so the concrete distributes symmetrically.
• Compact immediately after each layer is placed.

The code now selects the compaction method by slump, which is a genuinely useful addition:

Number of strokes when rodding, from Clause 6.3.1:

Write that down, because "25 blows per layer" is repeated on Indian sites constantly and it is the wrong number for a 150 mm cube. The 25-stroke figure belongs to the 100 mm cube and to the slump test.

Also from Clause 6.3.1: distribute strokes uniformly over the cross-section, do not let the rod significantly penetrate the previous layer, do not let it strike the bottom of the mould on the first layer, and after compacting each layer tap the sides of the mould with a mallet until large air bubbles stop appearing and the rod depressions close. The number of strokes shall be recorded.

Tamping rod specification: 16 ± 1 mm diameter, 600 ± 5 mm long, steel, with rounded roughly hemispherical ends. A piece of 12 mm rebar cut in the bar-bending yard is not a tamping rod.

Step 5: Finish and mark#

Clause 6.4: after the top layer is compacted, strike off the excess with a trowel or float, level flush with the top of the mould, and cover the surface with suitable material to prevent evaporation.

Clause 6.5: identify the specimens with a clear and durable marking that does not damage the specimen, and keep records so that the specimen identity is known from sampling all the way to testing.

In practice the marking should carry the sample number, date of casting, grade, and the pour location. The CPWD Quality Assurance Manual 2022 goes further and requires that cubes bear the name of work, mix, date, work location, and the signature of the engineer taking the sample.

Do not scratch the identity into the top face with a nail, and do not rely on a chalk mark that will wash off in the curing tank within a day. That is how cubes get mixed up, and a mixed-up cube is an untraceable cube.

Step 6: Demould at the right time#

Here is the second big change nobody has caught up with. IS 1199 (Part 5):2018 Clause 7 says:

"Leave the test specimens in the mould for at least 16 h, but not longer than three days, from the time of addition of water to the dry ingredients. Protect the specimens from shock, vibration and water evaporation. Store the specimens at a temperature of 27 ± 3 °C."

That is a window of 16 hours to 3 days, not the "24 ± ½ hours" that everyone still quotes from IS 516:1959. The wider window is a practical concession, particularly useful for low-early-strength mixes with high fly ash or slag content that would be damaged by stripping at 24 hours. It also means a night pour whose cubes get demoulded at 18 hours is compliant.

Note the reference datum: the clock starts from the addition of water to the dry ingredients, not from when the cube was cast.

Step 7: Cure properly#

Immediately after demoulding, submerge the specimens in clean, fresh water. Store them in water at 27 ± 2 °C until just before testing. The permitted alternative is a chamber at 27 ± 2 °C and relative humidity of at least 95 percent.

Three things about that 27 ± 2 °C that Indian sites routinely ignore:

• It is a requirement, not a target. An uncovered curing drum in Nagpur in May will sit well above 29 °C and will over-accelerate early hydration and depress 28-day strength. A tank in Shimla in January will sit below 25 °C and will under-report strength. Either way, your cube is no longer telling you about your concrete; it is telling you about your tank.
• Water level must stay above the top of the cubes at all times. Cubes with a dry top face for two days of a long weekend are not IS-compliant specimens.
• The water should be changed periodically. Stagnant tank water leaches lime and grows algae.

Buy a thermometer, leave it in the tank, and log the temperature daily. This costs about the price of one bag of cement and removes the single most common technical objection to a low cube result.

If cubes must travel to an external laboratory, Clause 7 requires them to be covered with wet cloth, wet sand or sawdust, or sealed in plastic bags containing water, and to reach the laboratory in a damp condition not less than 24 hours before the time of testing. At the lab they go back into water at 27 ± 2 °C until just before testing. Cubes that arrive dry in the back of a pickup at 9 a.m. on test day have already failed the method.

Step 8: Capping — usually not needed#

IS 1199 (Part 5) Clause 6.6 requires capping only for cylindrical specimens for compressive strength testing. Cubes are not capped, because a cube is loaded on its two moulded side faces, which are cast against machined steel and are already plane. Cores are a separate matter and are capped per IS 516 (Part 4).

This is exactly why Clause 3.5.1 of IS 516 (Part 1/Sec 1):2021 insists that cube specimens be compressed perpendicular to the direction of casting, and why the rough trowelled face must never end up against a platen.

Step 9: Test the cube (IS 516 Part 1/Sec 1:2021)#

Specimens that are damaged shall not be tested. Specimens that are badly honeycombed shall not be regarded as representative, and if they are tested anyway, the report must state that the specimen was honeycombed.

Clause 3.5.3 requires the failure pattern to be assessed and recorded as satisfactory or unsatisfactory, with the unsatisfactory pattern identified against the figures in the standard. This is not busywork. An unsatisfactory failure pattern is often the only visible evidence that the cube was badly made or badly centred, and it is your defence when a number comes out low.

Step 10: Calculate the test result#

Two rules operate together and both must be checked.

IS 456:2000 Clause 15.4: "The test results of the sample shall be the average of the strength of three specimens. The individual variation should not be more than ±15 percent of the average. If more, the test results of the sample are invalid."

IS 516 (Part 1/Sec 1):2021 Clause 3.6 adds the remedy: average of three, provided individual variation is not more than ±15 percent of the average; otherwise a repeat test shall be made, and if there is no further sample, the average of the two closest values may be taken.

Worked check. Three cubes give 32.0, 24.0 and 31.0 N/mm².

• Average = (32.0 + 24.0 + 31.0) / 3 = 29.0 N/mm²
• ±15 percent band = 24.65 to 33.35 N/mm²
• 24.0 falls below 24.65, so the sample is invalid.

An invalid sample is not a failed sample. It is a sample that tells you nothing, and it usually points at bad casting rather than bad concrete. What it does mean is that you now have a gap in your group of four, which matters enormously in the next section.

Sampling Frequency: IS 456 Clause 15.2.2#

This is the minimum frequency of sampling of concrete of each grade, taken verbatim from IS 456:2000.

Two notes from the code carry real weight on site:

• At least one sample shall be taken from each shift. A single 40 m³ pour that runs across two shifts needs samples in both, regardless of what the table alone would give you.
• Where concrete is produced at a continuous production unit such as a ready-mixed concrete plant, the frequency of sampling may be agreed upon mutually by supplier and purchaser. Get that agreement in writing in the RMC contract, before the first truck arrives, not after the first failure.

Remember each sample means three cubes at 28 days under Clause 15.3. If you also want 7-day cubes, that is three more per sample, and those extra cubes are optional under Clause 15.1.1, not mandatory.

Government work is frequently stricter than IS 456. The CPWD Quality Assurance Manual 2022, for example, uses bands of 1 to 10 m³ for one sample and 11 to 30 m³ for two on design mix RCC, and specifies 1 test per 25 m³ for ready-mixed concrete up to 100 m³. Always check your contract specification against IS 456 and adopt whichever is more onerous.

IS 456 Acceptance Criteria: The Section That Actually Decides Pass or Fail#

This is where the money is, and where the Indian internet is comprehensively out of date.

The current Table 11#

IS 456:2000 Clause 16.1 says concrete complies with the strength requirements when both of the following conditions are met: the mean strength determined from any group of four non-overlapping consecutive test results complies with column 2 of Table 11, and any individual test result complies with column 3 of Table 11.

Table 11 was replaced in full by Amendment No. 4, May 2013. The current table reads:

Note 1: In the absence of an established value of standard deviation, the values given in Table 8 may be assumed, and an attempt should be made to obtain results of 30 samples as early as possible to establish the value of standard deviation.

Note 2: For concrete of quantity up to 30 m³, where the number of samples to be taken is less than four as per the frequency of sampling given in 15.2.2, the mean of test results of all such samples shall be fck + 4 N/mm² minimum, and the requirement of minimum individual test results shall be fck − 2 N/mm² minimum. However, when the number of samples is only one as per 15.2.2, the requirement shall be fck + 4 N/mm² minimum.

Read that table again, because three things changed in 2013#

1. There is now a single row. The old split between "M15" and "M20 or above" is gone. Every grade from M15 upwards uses the same rule.
2. The old "fck + 4" alternative for M20 and above is gone from the mean criterion. It is fck + 3 for all grades now.
3. The individual test result minimum is fck − 3 for all grades. The old "fck − 4" for M20 and above no longer exists.

If a lab report, a consultant's letter, or an article tells you that your M30 individual cube only has to reach fck − 4 = 26 N/mm², it is quoting a table that was superseded in May 2013. The correct figure is 27 N/mm². That one number has decided real disputes.

Note 2 is also new in 2013 and is quietly one of the most useful clauses in the code, because small pours are where sites get stuck: you cannot form a group of four from three samples. Note 2 tells you exactly what to do instead.

Table 8: assumed standard deviation#

Most sites do not have 30 test results when the first pour happens, so they use the assumed values from IS 456 Table 8.

These correspond to good site control. IS 10262:2019 uses the same values in its Table 2 and adds the important caveat that where site control is only fair, you add 1 N/mm².

Establishing your own standard deviation requires a minimum of 30 test results, and IS 10262:2019 requires it to be re-checked monthly and updated after every change of mix proportioning.

Worked example 1: M20 with assumed standard deviation#

fck = 20 N/mm². No established standard deviation yet, so Table 8 gives 4.0 N/mm².

Mean-of-4 limit:

• 0.825 × 4.0 = 3.3, rounded to nearest 0.5 = 3.5. So fck + 3.5 = 23.5 N/mm².
• Alternative: fck + 3 = 23.0 N/mm².
• Whichever is greater: 23.5 N/mm².

Individual limit: fck − 3 = 17.0 N/mm².

PASS case. Four consecutive test results: 24.5, 22.0, 25.5, 23.0 N/mm².

• Mean = (24.5 + 22.0 + 25.5 + 23.0) / 4 = 95.0 / 4 = 23.75 N/mm². 23.75 ≥ 23.5, so the mean criterion is satisfied.
• Lowest individual = 22.0 ≥ 17.0, satisfied.
• Result: ACCEPTED.

FAIL case, and this is the one that starts arguments. Four consecutive test results: 23.0, 22.5, 23.5, 22.0 N/mm².

• Mean = 91.0 / 4 = 22.75 N/mm². 22.75 is less than 23.5, so the mean criterion FAILS.
• Every individual result is above 17.0, and in fact every single cube exceeded the specified grade of 20.
• Result: REJECTED under Clause 16.1(a).

Every cube beat M20 and the concrete still fails. That is not a quirk; that is the whole point of the code. fck is a characteristic strength, defined in Clause 6.1.1 as the strength below which not more than 5 percent of results are expected to fall. To deliver that statistically, your average has to sit meaningfully above fck. A batching plant that consistently produces 22 to 23 N/mm² for M20 is producing concrete with almost no margin, and IS 456 catches it.

Worked example 2: M25, where the individual result is the one that bites#

fck = 25 N/mm². Table 8 gives 4.0 N/mm².

Mean-of-4 limit: 0.825 × 4.0 = 3.3, rounds to 3.5, so 28.5 N/mm². Against fck + 3 = 28.0. Governing value 28.5 N/mm². Individual limit: fck − 3 = 22.0 N/mm².

PASS case: 29.0, 30.5, 28.0, 29.5 N/mm².

• Mean = 117.0 / 4 = 29.25 ≥ 28.5, satisfied.
• Lowest individual = 28.0 ≥ 22.0, satisfied.
• ACCEPTED.

FAIL case: 30.0, 31.0, 21.5, 32.0 N/mm².

• Mean = 114.5 / 4 = 28.625 ≥ 28.5, so the mean criterion just scrapes through.
• But the third result, 21.5, is below 22.0. The individual criterion FAILS.
• REJECTED under Clause 16.1(b).

Now note the crucial consequence, from Clause 16.3: for the individual test result requirement, "only the particular batch from which the sample was taken shall be at risk." You do not condemn the whole four-sample block. You investigate the specific pour that the 21.5 sample represented. That distinction can be the difference between chipping out one column and chipping out a floor, and it is why your register must record exactly which pour location each sample came from.

Worked example 3: M30, and why a good plant does not always get an easier bar#

fck = 30 N/mm². Here the site has 30-plus results and has established its own standard deviation.

Case A, an excellent RMC plant with established standard deviation of 3.2 N/mm²:

• 0.825 × 3.2 = 2.64, rounds to 2.5. So fck + 2.5 = 32.5 N/mm².
• Alternative fck + 3 = 33.0 N/mm².
• Whichever is greater: 33.0 N/mm².

Case B, a plant with poorer consistency, established standard deviation 5.2 N/mm²:

• 0.825 × 5.2 = 4.29, rounds to 4.5. So fck + 4.5 = 34.5 N/mm².
• Alternative fck + 3 = 33.0 N/mm².
• Whichever is greater: 34.5 N/mm².

Individual limit in both cases: fck − 3 = 27.0 N/mm².

Two lessons fall out of this. First, the fck + 3 term is a floor. However tight your quality control, your mean-of-4 can never be allowed below fck + 3, so there is a hard limit to how much a low standard deviation can help you. Second, establishing your own standard deviation cuts both ways: if your real scatter is worse than the Table 8 assumption, the code holds you to a tougher number, not an easier one.

FAIL case for Case B: 34.0, 33.5, 35.0, 32.5 N/mm².

• Mean = 135.0 / 4 = 33.75. Against the required 34.5, the mean criterion FAILS.
• All individuals clear 27.0.
• REJECTED on the mean.

Had this plant been running Case A control, the same four results would have passed comfortably. Consistency is worth money.

A note on the rounding: the code says "0.825 × established standard deviation (rounded off to nearest 0.5 N/mm²)". The most common reading, and the one used above, is that the product 0.825 × s is rounded to the nearest 0.5 before adding fck. In the great majority of practical cases you get the same answer either way. Where a project sits on the boundary, get the engineer-in-charge to state the convention in the QAP rather than argue it after a result.

Worked example 4: a small pour, under Note 2#

A 300 mm thick raft patch of M25 consumes 18 m³. Per Clause 15.2.2, the 16 to 30 m³ band requires 3 samples. Three is fewer than four, so you cannot form a group of four, and Note 2 governs.

• Mean of all such samples required: fck + 4 = 29.0 N/mm² minimum.
• Individual required: fck − 2 = 23.0 N/mm² minimum.

Results: 29.5, 28.0, 30.0 N/mm².

• Mean = 87.5 / 3 = 29.17 ≥ 29.0, satisfied.
• Lowest individual = 28.0 ≥ 23.0, satisfied.
• ACCEPTED.

Watch the individual limit in Note 2. It is fck − 2, not fck − 3. Small pours are held to a tighter individual bar precisely because there are too few results to average out the noise. And if the pour is 5 m³ or less, so only one sample is taken, that single sample must reach fck + 4 on its own.

Three more acceptance rules that get forgotten#

Clause 16.3: the quantity of concrete represented by a group of four consecutive test results includes the batches from which the first and last samples were taken, together with all intervening batches. Where the mean rate of sampling is not specified, the maximum quantity that four consecutive test results represent is limited to 60 m³. This is the clause that defines how much concrete a failure actually condemns.

Clause 16.5: concrete of each grade shall be assessed separately. You cannot mix M25 and M30 results into a single group of four to make the numbers work.

Clause 16.6: concrete is liable to be rejected if it is porous or honeycombed, if placing was interrupted without a proper construction joint, if reinforcement has been displaced beyond tolerance, or if construction tolerances have not been met. Strength is not the only acceptance test. Hardened concrete may still be accepted after suitable remedial measures to the satisfaction of the engineer-in-charge.

7-Day and 28-Day Strength: What Is Code and What Is Not#

Be very clear on this, because it is quoted as gospel and it is not in any Indian code.

IS 456:2000 Clause 15.1.1 permits optional 7-day compressive strength tests in addition to the 28-day test, and then says: "For this purpose the values should be arrived at based on actual testing. In all cases the 28 days compressive strength specified in Table 2 shall alone be the criterion for acceptance or rejection of the concrete."

No Indian standard gives a fixed percentage relationship between 7-day and 28-day strength. Not IS 456. Not IS 516 (Part 1/Sec 1):2021. Not IS 10262:2019. The code deliberately declines to give a number and instructs you to establish it for your own materials and mix.

So the table below is a site rule of thumb for early warning only. It is not a code requirement, it cannot be used to accept or reject concrete, and it will shift with cement type, blend, admixture, and ambient temperature.

Two practical cautions. Blended cements, PPC and slag cements in particular, gain strength more slowly and will often sit at the bottom of or below these bands at 7 days while comfortably meeting 28-day requirements. And the correct way to use 7-day results is exactly what IS 456 says: build your own correlation from your own historical pairs of 7-day and 28-day results for that plant and that mix, then use it as a trigger to investigate early, not as a basis to accept or condemn.

A useful discipline here comes from the CPWD Quality Assurance Manual 2022, which requires that the 28-day test be done at the same laboratory where the 7-day test was done. Splitting the pair across two labs introduces machine-to-machine variation into the one comparison you were relying on.

When a Cube Fails: The Decision Tree#

A failed cube is not a failed structure. IS 456 Clause 16.4 says that where concrete is deemed not to comply, the structural adequacy of the parts affected shall be investigated per Clause 17, and any consequential action taken. Investigate first. Demolish last.

Work through this sequence in order. Do not skip to coring because someone is shouting.

Stage 1: Validate the test before you doubt the concrete#

Before accepting that the concrete is bad, prove the test was good. Check, in this order:

• Was the sample a composite sample from at least four increments, or was it a spot sample?
• Did the three cubes of that sample pass the ±15 percent individual variation check? If not, the sample is invalid under Clause 15.4 and there is nothing to investigate yet.
• Curing tank temperature log for the full 28 days. Any period outside 27 ± 2 °C?
• Was the water level ever below the top of the cubes?
• Demoulding time within 16 hours to 3 days?
• Was the cube compressed perpendicular to the direction of casting, with the trowelled face not against a platen?
• Machine calibration certificate: current, within the last 12 months, and covering the load range used?
• Loading rate: does the machine's printout or the operator's timing correspond to about 315 kN/min for a 150 mm cube?
• Failure pattern recorded as satisfactory or unsatisfactory?
• Batching records for that pour: cement content, water-cement ratio, admixture dosage, slump at site.

A surprising share of "failures" die here. This is also the reason a disciplined, tamper-evident record beats a paper register, which is where construction quality management software earns its keep.

Stage 2: Raise the non-conformance formally#

If the test survives Stage 1, raise an NCR against the identified pour with the exact quantity of concrete at risk per Clause 16.3, and start a formal investigation. Use NCR management to hold the record and drive it to closure, and open a CAPA in parallel so that the plant-side root cause gets fixed rather than just the affected element.

Stage 3: Rebound hammer, for mapping only#

Use the rebound hammer, now covered by IS 516 (Part 5/Sec 4):2020, to compare the suspect element against known-good elements of the same age, mix and finish. Take six readings around each point of observation and average them after deleting outliers per IS/ISO 16269 (Part 4). Keep the point of impact at least 25 mm from any edge.

What the rebound hammer will not do is give you a strength you can use for acceptance. It reads surface hardness in roughly the outer 30 mm, and the standard is explicit that carbonation of the surface has a very significant influence, up to 50 percent in extreme cases. Treat it as a screening and mapping tool that tells you where to look, never as a number that settles a dispute.

Stage 4: Ultrasonic pulse velocity, for integrity#

IS 516 (Part 5/Sec 1):2018 covers UPV. Its quality grading table is:

The standard attaches a warning that most site engineers have never read, and it matters: "This table is only for concrete quality grading and shall not be used for estimating the concrete grades from ultrasonic pulse velocity values." UPV tells you about homogeneity, voids, cracks and segregation. It does not tell you the grade. A "Doubtful" result means further tests are needed, and the code says so.

Stage 5: Core test, the first result that can actually accept the concrete#

This is the decisive step, and it is the one governed by real acceptance numbers.

Procedure per IS 516 (Part 4):2018:

• In no case shall fewer than three cores be tested. Locations at the discretion of the engineer-in-charge, representative of the whole of the concrete concerned.
• Core diameter generally 100 to 150 mm (± 10 mm), with 100 mm preferred for nominal maximum aggregate size up to 20 mm. The ratio of core diameter to nominal maximum aggregate size must be greater than 3.
• Preferred length-to-diameter ratio is 2. Values of l/d from 1 to 2 are permitted, the length including capping material.

Corrections, applied in order:

1. Diameter correction, for cores under 100 mm diameter: multiply by 1.03 for 75 ± 5 mm diameter, or 1.06 for under 70 mm.
2. Length-to-diameter correction: F = 0.11 N + 0.78, where N is the l/d ratio. Sanity-check it: at N = 2, F = 1.00, as it must be.
3. The product gives the corrected cylinder strength, equivalent to a cylinder of h/d = 2.
4. Equivalent cube strength = corrected cylinder strength × 5/4.

Worked core example. A 100 mm diameter core is trimmed to 130 mm length, so l/d = 1.3. It fails at 240 kN.

• Cross-sectional area = π × 50² = 7854 mm².
• Measured strength = 240000 / 7854 = 30.6 N/mm².
• No diameter correction, since the core is 100 mm.
• F = 0.11 × 1.3 + 0.78 = 0.923.
• Corrected cylinder strength = 30.6 × 0.923 = 28.2 N/mm².
• Equivalent cube strength = 28.2 × 5/4 = 35.3 N/mm².

Acceptance, from IS 456 Clause 17.4.3: concrete in the member represented by a core test is acceptable if the average equivalent cube strength of the cores is at least 85 percent of the cube strength of the grade of concrete specified for the corresponding age, and no individual core has a strength less than 75 percent.

So for M30:

• Required average equivalent cube strength = 0.85 × 30 = 25.5 N/mm².
• Required minimum for any single core = 0.75 × 30 = 22.5 N/mm².

The 35.3 N/mm² core above clears both comfortably. Cores frequently rescue elements that cube results condemned, because a cube tells you about the concrete that was delivered while a core tells you about the concrete that is actually in the member.

Stage 6: Load test, for flexural members#

If core results do not satisfy Clause 17.4.3, or cores were not taken, Clause 17.5 permits a load test.

Per Clause 17.6, load tests should be carried out as soon as possible after the expiry of 28 days from the time of placing the concrete. The structure is subjected to full dead load plus 1.25 times the imposed load for a period of 24 hours, and then the imposed load is removed. Deflection due to the imposed load only is recorded. If within 24 hours of removal the structure does not recover at least 75 percent of the deflection under superimposed load, the test may be repeated after a lapse of 72 hours.

Clause 17.7 is important and often missed: members other than flexural members should preferably be investigated by analysis, not by load test. You do not load-test a column.

Stage 7: Decide and remediate#

The realistic outcomes, in descending order of preference:

Whatever the outcome, the decision belongs to the engineer-in-charge in writing, with the core report, NDT results and analysis attached. If the affected element is significant, this is exactly the trigger for a formal structural audit.

Cube Test Register Format#

Every site needs one register, maintained continuously, with rows added at casting and completed at testing. The format below combines the report fields required by IS 516 (Part 1/Sec 1):2021 Clause 3.7, the sample record fields of IS 1199 (Part 5) Clause 8, and the register structure of CPWD Annexure-22.

Three columns in that register do more work than all the others: quantity represented, mean of last 4 consecutive results, and acceptance status. They are also the three that handwritten registers almost never carry, because the mean-of-4 has to be recomputed on a rolling basis every time a new result lands. That is arithmetic a spreadsheet or a quality management system should do automatically, not something an engineer should recalculate by hand at month end.

Twelve Site Malpractices That Corrupt Cube Results#

Each of these is common, each changes the number, and each has a specific way a supervisor catches it.

Two more that are not malpractice so much as recordkeeping failure, but which cost just as much in a dispute: cubes marked with chalk that washes off in the tank, and a register in which nobody recorded which pour location a sample represented. When a result comes in low, Clause 16.3 lets you limit the risk to a specific batch, but only if you can prove which batch it was. A structured inspection checklist at the point of casting closes both gaps.

Who Should Do the Testing#

Site laboratory, client laboratory, or independent third party: all three are used in India, and the choice is contractual rather than statutory.

NABL, the National Accreditation Board for Testing and Calibration Laboratories, accredits testing laboratories against ISO/IEC 17025:2017. Two points site engineers should be clear on:

• NABL accreditation is voluntary. NABL is a constituent board of the Quality Council of India and states plainly in its own FAQ that it has no regulatory powers. There is no Indian statute that makes NABL-accredited third-party cube testing compulsory on all construction.
• It becomes binding through the contract. CPWD, for instance, permits testing to be outsourced to NABL-accredited laboratories or to laboratories accredited to ISO/IEC 17025 by a body operating under ISO/IEC 17011, alongside government and academic institutes. Many private developer contracts simply specify NABL. Read your specification.

NABL also publishes NABL 224:2024, a specific criteria document for the recognition of temporary site laboratories testing aggregates and concrete on building projects, aimed at projects with constructed area of 50,000 sq ft and above. Its scope covers compressive strength of hardened concrete to IS 516 (Part 1/Sec 1) and slump to IS 1199 (Part 2). It is worth knowing that this site-lab recognition is explicitly a voluntary scheme and is not covered under the APAC and ILAC mutual recognition arrangements, so it is not the same thing as full NABL accreditation.

Whatever the arrangement, insist on three things: the laboratory's compression testing machine calibration certificate dated within the last 12 months, the same laboratory for both the 7-day and 28-day tests of a given sample, and the engineer-in-charge or their representative present at casting. CPWD requires witnessing of testing for 50 percent of samples and sample collection by the Engineer-in-Charge personally for 10 percent of mandatory tests. Those are sensible benchmarks even on private work.

Quick Reference Card#

For the wider code landscape these standards sit within, see our complete guide to IS codes in Indian construction, and for the reinforcement side of the same acceptance discipline, TMT steel bar grades under IS 1786.

FAQs#

Is IS 516:1959 still valid for the concrete cube test?#

No, not for compressive strength testing. IS 516 (Part 1/Sec 1):2021 states that its test methods "shall be applicable in place of the corresponding provisions given in IS 516 : 1959". BIS split IS 516 into 12 parts, and the 1959 document will be formally superseded once all parts are published. In practical terms, from 2021 onwards a compressive strength test should be run to IS 516 (Part 1/Sec 1):2021, and the making and curing of the cubes should follow IS 1199 (Part 5):2018, not IS 516:1959. If your ITP or lab report still cites the 1959 code, update it.

What is the exact IS 456 acceptance criteria for concrete cubes?#

Both conditions in Clause 16.1 must be satisfied. First, the mean strength from any group of four non-overlapping consecutive test results must be at least fck plus 0.825 times the established standard deviation, rounded to the nearest 0.5 N/mm², or fck plus 3 N/mm², whichever is greater. Second, every individual test result must be at least fck minus 3 N/mm². This single-row table applies to M15 and above and comes from Amendment No. 4 of May 2013, which removed the old separate treatment of M20 and above and eliminated the fck plus 4 and fck minus 4 values that most references still quote.

Why do cubes fail at 28 days when they passed at 7 days?#

Because the 7-day result was never a promise. IS 456 Clause 15.1.1 permits 7-day tests only as an optional early indicator and requires the correlation to be established by actual testing on your own materials. A 7-day pass followed by a 28-day failure usually points at something that happened after day 7, most often curing: tank water dropping below the cubes, tank temperature drifting outside 27 ± 2 °C, or cubes taken out and left dry. It can also mean the 7-day number was inflated by the concrete having been hot at casting, which accelerates early strength but depresses ultimate strength. Check your curing tank log first.

Can 7-day strength predict 28-day strength?#

Only through a correlation you build yourself. No Indian standard, including IS 456, IS 516 (Part 1/Sec 1):2021 and IS 10262:2019, gives a fixed percentage. The widely quoted "65 to 70 percent at 7 days" is an industry rule of thumb, not a code requirement, and it is unreliable for blended cements such as PPC and slag cement, which gain strength more slowly. IS 456 Clause 15.1.1 is explicit that 7-day values "should be arrived at based on actual testing", and that 28-day strength "shall alone be the criterion for acceptance or rejection of the concrete".

What if only one cube out of three fails?#

First check whether the sample is even valid. Under IS 456 Clause 15.4, if any individual specimen varies by more than ±15 percent from the average of the three, the test results of the sample are invalid, and IS 516 (Part 1/Sec 1):2021 Clause 3.6 requires a repeat test, or if no further sample exists, the average of the two closest values. If the sample is valid, remember that the "individual test result" in Table 11 means the average of the three specimens in that sample, not one cube. A single low cube within an otherwise consistent set is far more often a casting or handling defect than a concrete defect, so examine the recorded failure pattern and the cube's dimensions and weight.

All my cubes are above the grade strength but the concrete was rejected. Why?#

Because fck is a characteristic strength, not a minimum. IS 456 Clause 6.1.1 defines it as the strength below which not more than 5 percent of results are expected to fall, which means your average has to sit well above fck to deliver it. For M20 with an assumed standard deviation of 4.0, the mean of four consecutive results must reach 23.5 N/mm². A set of 23.0, 22.5, 23.5 and 22.0 averages 22.75 and fails, even though every cube beat 20. It is a signal that the plant is running with almost no margin, and the code is designed to catch exactly that.

How much concrete does one failed cube result condemn?#

Under IS 456 Clause 16.3, the quantity represented by a group of four consecutive test results includes the batches from which the first and last samples were taken plus all intervening batches, and where the mean sampling rate is not specified, this is capped at 60 m³. Critically, for a failure of the individual test result criterion only, the clause states that "only the particular batch from which the sample was taken shall be at risk". So an individual failure implicates one pour, while a mean-of-four failure implicates the whole block. This is why your register must record the pour location and quantity represented for every single sample.

Is a 100 mm cube allowed instead of a 150 mm cube in India?#

Yes, conditionally. IS 1199 (Part 5):2018 Clause 4.1.1 permits 100 mm cubes as an alternative when the largest nominal size of aggregate does not exceed 20 mm. Note two things. The compaction requirement changes to a minimum of 25 strokes per layer instead of 35. And no Indian standard prescribes a correction or conversion factor between 100 mm and 150 mm cube results, so the 100 mm result is used directly in the IS 456 acceptance calculation. Do not apply size-conversion factors from international literature without written instruction from the engineer-in-charge.

What is the correct loading rate for the concrete cube test?#

IS 516 (Part 1/Sec 1):2021 Clause 3.5.2 requires the load to be applied without shock and increased continuously at a constant rate of 14 N/mm²/min until no greater load can be sustained. For a 150 mm cube that is roughly 315 kN per minute, and for a 100 mm cube roughly 140 kN per minute. This is essentially the same rate as the old IS 516:1959 figure of 140 kgf/cm²/min, restated in SI units. The 0.2 to 1.0 MPa/s range quoted in some references comes from EN 12390-3 and ISO 1920-4, not from the Indian standard.

What happens after a cube fails: what is the correct sequence?#

Validate the test first, then investigate the structure. Check sampling method, the ±15 percent validity rule, curing temperature and water level logs, demoulding time, machine calibration and loading rate. If the test holds up, IS 456 Clause 16.4 sends you to Clause 17. Use rebound hammer per IS 516 (Part 5/Sec 4):2020 and UPV per IS 516 (Part 5/Sec 1):2018 for mapping and homogeneity, but neither can accept the concrete. Core testing per IS 516 (Part 4):2018 is the first result that can: under Clause 17.4.3 the concrete is acceptable if the average equivalent cube strength of at least three cores is at least 85 percent of the specified grade and no individual core falls below 75 percent. If cores fail, a load test under Clause 17.6 applies full dead load plus 1.25 times imposed load for 24 hours, with at least 75 percent deflection recovery required.

Does IS 456 still refer to IS 13311 for non-destructive testing?#

Yes, and that reference is out of date. IS 456:2000 Clause 17.8 cites IS 13311 (Part 1) for ultrasonic pulse velocity and IS 13311 (Part 2) for rebound hammer. BIS has since withdrawn both: IS 13311 (Part 1):1992 is recorded as withdrawn and superseded by IS 516 (Part 5/Sec 1):2018, and IS 516 (Part 5/Sec 4):2020 states that IS 13311 (Part 2):1992 "shall stand withdrawn after the publication of this standard". Specify the IS 516 Part 5 sections in your NDT scope of work; a lab quoting IS 13311 is working to a withdrawn standard.

Is NABL-accredited testing mandatory for concrete cubes in India?#

Not by law. NABL accredits laboratories against ISO/IEC 17025:2017 and describes its own accreditation as voluntary, stating that it has no regulatory powers. There is no Indian statute making third-party NABL testing compulsory across all construction. It becomes mandatory contractually: CPWD, for example, allows outsourced testing to NABL-accredited laboratories or laboratories accredited to ISO/IEC 17025 under an ISO/IEC 17011 accreditation body, and many private contracts specify NABL directly. Check your own specification, and regardless of accreditation, always verify that the compression testing machine has a calibration certificate less than 12 months old.