Material Testing Frequency Chart for Construction in India: IS Code Tables, Sample Sizes and Acceptance Limits

There is no single Indian Standard that lists "material testing frequency". The frequency comes from three different places: IS 456:2000 (for concrete sampling), the works specification you are building under (CPWD Specifications, MoRTH Specifications for Road and Bridge Works, or the client's own QAP), and the product standard's own lot-sampling clause (IS 1786 for steel, IS 5454 for bricks). The IS test-method codes (IS 4031, IS 2386, IS 516, IS 2720, IS 3495) tell you HOW to test and what the acceptance limit is. They almost never tell you HOW OFTEN. This guide puts both halves in one place, material by material, with the code edition against every row.

Print the tables. Fix the frequency column against your contract before the first truck arrives on site, because the argument you will have in month four is always about a test that was never done in month one.

How to read a testing frequency requirement#

Every line in a testing schedule has five parts, and site teams routinely record only two of them.

1. Test - what property you are measuring.
2. Method code and edition - IS 2386 (Part 4):1963 is not the same instruction as IS 2386 (Part 1):1963. Write the part number.
3. Frequency - per tonne, per cubic metre, per square metre, per lot, per source change, or per shift. Note which trigger applies; several run in parallel and the tightest one governs.
4. Sample size - how many specimens make one valid test. A "cube test" is three cubes, not one.
5. Acceptance limit and record - the number you compare against, and the register or format the result is entered in.

Two rules save more arguments than anything else. First, any change of source is a fresh full test, regardless of how recently you tested. New quarry, new brand of cement, new brick kiln, new borrow area - restart the schedule. Second, the tightest trigger wins. If your spec says one cube set per 50 cubic metres and also at least one per shift, a 12 cubic metre pour on a separate day still needs its own set.

For the underlying codes and where they sit in the hierarchy, see our complete guide to IS codes for Indian construction.

Cement: test frequency and acceptance#

The governing specifications are IS 269:2015 for Ordinary Portland Cement (the sixth revision, which absorbed the older IS 8112 for 43 grade and IS 12269 for 53 grade into one standard covering grades 33, 43, 43S, 53 and 53S), IS 1489 (Part 1):1991 for fly-ash-based Portland Pozzolana Cement, and IS 455:2015 for Portland Slag Cement. All the physical test methods sit in the IS 4031 series.

IS 269 itself does not impose a site retest frequency. It defines a lot for the manufacturer and refers sampling to IS 3535. The frequency you follow on site comes from your works specification or QAP. The column below shows what is commonly specified on Indian projects; confirm the number against your own contract.

Cement checks that need no laboratory#

Storage compliance is a daily check, and it fails more consignments than the lab does. Bags stacked not more than 10 high, off the floor on a raised platform, clear of external walls, oldest bag issued first. Cement older than three months from the date of manufacture should be retested for strength before use - that is a standard clause in most Indian works specifications, and the week-and-year code printed on the bag is what you check it against. Lumpy cement that does not crumble to powder under light finger pressure goes back on the truck.

Aggregates: test frequency and acceptance#

IS 383:2016 is the specification (what is acceptable). IS 2386, Parts 1 to 8, all dated 1963, are the test methods. IS 383:2016 also formally permits manufactured sand and, within limits, recycled aggregate, which the 1970 edition did not.

The bulking check nobody does, and why it costs you#

Damp sand swells. A few percent of moisture can bulk sand by 20 to 30 percent in loose volume. If you gauge by boxes and do not measure bulking every morning through the monsoon, you are putting far less sand into the mix than the design calls for, and the mix you are placing is not the one that was designed. The test is IS 2386 (Part 3):1963, it needs a measuring cylinder and ten minutes, and it is the highest-value daily test on a volume-batched site. Weigh-batched sites still deduct free moisture from the mixing water, but the bulking geometry stops mattering.

Water for concrete and curing#

IS 456:2000 Clause 5.4 governs, and the test methods sit in the IS 3025 series. Water that is fit to drink is generally fit for concrete, but "generally" is not "always" - borewell water in parts of Gujarat, Rajasthan and coastal Tamil Nadu will fail on chlorides or sulphates while being perfectly drinkable.

Curing water is held to the same standard, and this is where sites cut corners. Water that leaves stains or deposits on the finished surface should not be used for curing exposed concrete even if it passes the chemical limits. Sea water is not permitted for mixing or curing reinforced concrete.

Reinforcement steel: IS 1786 test frequency#

IS 1786:2008 covers high strength deformed bars and wires - Fe 415, Fe 415D, Fe 500, Fe 500D, Fe 550, Fe 550D and Fe 600. Our TMT steel bar grades guide covers the grade selection side in more detail.

Be careful here, because two different frequencies get confused with each other. IS 1786:2008 Clause 11.1 sets the manufacturer's sampling: test specimens for nominal mass, mechanical properties, bend and rebend are cut at random from each size of finished bar at a frequency of not less than 2 per cast for casts or heats below 100 tonnes, and 3 per cast for casts or heats of 100 tonnes or more. That is a mill-level rule. The frequency at which you independently test steel arriving at your site comes from your works specification or QAP, and is normally expressed in tonnes per diameter. Both apply. Neither replaces the other.

Three site rules around steel that no code writes down but every good QA/QC engineer enforces. Never cut a test sample from the end of a bar that has already been cropped - the cropped end may not represent the bar. Cut from a random bundle, at random. Never accept a lot where the heat number on the bundle tag does not appear on the MTC; if the tags are missing, the lot has no traceability and cannot be lot-tested. Never let the supplier choose the sample - the site engineer or the third-party inspector picks it, in front of the supplier's representative.

Concrete: sampling frequency under IS 456:2000#

This is the one frequency that is genuinely fixed by an IS code, and it is the one most often mis-stated on site. IS 456:2000 Clause 15.2.2 gives the minimum frequency of sampling for each grade of concrete:

Read that carefully: the count is per grade of concrete per day's work, at least one sample from each shift, and the sampling must be random across the whole period of concreting and across all mixing units. Three specimens form one sample for the 28-day acceptance test. Casting an extra set for 7-day information is normal practice but those cubes do not substitute for the 28-day sample.

Note the edition on the cube test. IS 516:1959 has been superseded by IS 516 (Part 1/Sec 1):2021 for compressive, flexural and split tensile strength. If your test reports still print "IS 516:1959" in 2026, your lab has not updated its formats, and that is worth raising - it is exactly the kind of detail an auditor picks up. We have a full walk-through in the concrete cube test guide.

Acceptance criteria, stated correctly#

IS 456:2000 acceptance for compressive strength works on two conditions, both of which must be satisfied:

• The mean strength of any group of four non-overlapping consecutive samples must be at least fck + 0.825 times the established standard deviation, or fck + 4 N/sq mm for M20 and above (fck + 3 N/sq mm for M15), whichever is greater.
• No individual sample result may fall below fck minus 4 N/sq mm for M20 and above (fck minus 3 N/sq mm for M15).

One cube below fck is not automatically a failure. One sample below fck minus 4 is. The distinction between a cube and a sample of three cubes is where most site arguments start.

Bricks, blocks and masonry units#

AAC blocks deserve one extra line of caution: drying shrinkage and moisture content at delivery are the two properties that cause plaster cracking, and they are the two most often skipped because the manufacturer's certificate is accepted at face value. Test the first lot from any new supplier independently.

Soil, subgrade and compaction#

Compaction frequency is where the specifications diverge most sharply, so this is one row you must copy from your own contract rather than from any generic chart. See our soil testing guide for the test procedures.

Bituminous works#

Bituminous testing on Indian projects is governed by the MoRTH Specifications for Road and Bridge Works, Section 900 (Quality Control for Road Works), rather than by a single IS code. The methods are drawn from IS 73 and IS 1203 to IS 1220 for bitumen properties, IS 2386 for aggregates, and ASTM/AASHTO methods for Marshall design.

The frequencies below are the ones commonly worked to on MoRTH and NHAI projects. Confirm every row against the revision of the MoRTH specification named in your contract and against the approved QAP before you use it - Section 900 has been amended across revisions, and bituminous layers are paid against tested lots, so an out-of-date frequency table is a direct payment risk.

Mix temperature is a continuous control, not a periodic test: record it at the plant, at delivery and immediately before rolling. A mix laid below the specified temperature will never reach density no matter how many passes the roller makes, and the density core taken afterwards is the expensive way to find that out.

Where frequencies differ: CPWD vs MoRTH/NHAI vs private developer#

This is the part that generic frequency charts on the internet never address, and it is the reason those charts get people into trouble. The same M25 concrete, the same Fe 500D bar, tested at three different frequencies depending on who owns the job.

Three consequences worth internalising. First, a QA/QC engineer moving from a highway job to a high-rise cannot carry across an instinct for "one test per 500 square metres" - buildings are measured in cubic metres and structural members, not area of layer. Second, on MoRTH and NHAI work the frequency is a payment mechanism: an untested lot is an unpayable lot. On private work it is a certification mechanism, where the NCR blocks the certificate. On CPWD work it is a departmental record that follows the file. Third, private QAPs are the only ones you can negotiate - if the PMC's QAP demands a frequency the site lab cannot deliver, raise it in writing at the kick-off meeting, because nobody will relax it in month six.

Building the Inspection and Test Plan#

A frequency chart is a list. An Inspection and Test Plan is that list turned into a sequence with names, sign-offs and stop points against it. Every ITP row carries an intervention level:

• H, Hold point. Work stops. It cannot proceed until the named party has inspected and signed. Reinforcement inspection before shuttering closes is the classic hold point - once the shutter closes, the evidence is gone.
• W, Witness point. The named party is given notice and may attend. If they do not attend within the notice period, work proceeds and the record notes that notice was given. Cube casting is normally a witness point.
• R, Review point. No attendance; the party reviews the record afterwards. Mill test certificates are review points.
• S, Surveillance. Random monitoring of a repetitive activity, such as slump at the pour point.

A minimal ITP row for a slab pour looks like this:

The two failures that make an ITP worthless are: hold points that nobody actually holds for, and notice periods that are not written down. Fix the notice period in the ITP - "24 hours written notice" - and hold the line on it from the first pour.

When a test fails#

A failed test is a normal event on a working site. Handling it badly is what turns it into a dispute. The sequence is fixed.

Step 1: Do not retest first#

The instinct is to immediately cast fresh cubes or pull a fresh steel sample. That destroys the timeline. First quarantine: identify exactly what quantity of material or what portion of work the failed sample represents, and physically stop it being used or covered up. Tag the stack, mark the pour on the drawing, and record the affected quantity in the register the same day.

Step 2: Check the test before you check the material#

A significant share of failures are testing errors. Was the cube water-cured, or did it sit in the sun? Was the machine calibrated, and when? Was the loading rate correct, the identification correct, the age within tolerance? For steel, was the gauge length marked correctly and the right mandrel used for the bend? A test done outside the method code is not a test result at all.

Step 3: Retest under the code's own rules#

Different codes give different retest rights.

• Steel, IS 1786:2008. The retest rule is explicit. If any one of the test pieces first selected fails any test, two further samples are selected for testing in respect of each failure. If the test pieces from both additional samples pass, the material represented is deemed to comply for that particular test. If the test piece from either additional sample fails, the material represented does not comply. Note what that means in practice: two out of two, not one out of two. A supplier who tests three more samples and shows you the one that passed has not satisfied the standard.
• Concrete, IS 456:2000. There is no "retest the cube" provision, because the cube is gone. Clause 17.4 sets out what happens instead: the engineer may call for core tests under IS 516 (Part 4):2018, and may additionally call for load testing of the structure. Core acceptance is average equivalent cube strength at least 85 percent of fck with no single core below 75 percent.
• Bricks and blocks. Sampling standards generally allow a second, larger sample from the same lot. If the second sample fails, the lot is rejected.
• Cement. A failed strength test on a lot means the lot is not used pending a confirmatory test at an independent NABL lab, with the sample drawn jointly.

Step 4: Raise the non-conformance formally#

The NCR is not a punishment, it is a record. It must state: what failed, against which specification clause, the quantity and location affected, the immediate containment action, the disposition (use as is with justification, repair, rework, or reject), and the sign-off required to close it. A disposition of "use as is" always requires the designer's written concurrence - never the site engineer's alone.

Step 5: Close the loop with corrective action#

Containment fixes this occurrence. Corrective and preventive action fixes the cause. If cubes failed because the sand was bulking uncorrected, the CAPA is a daily bulking measurement with a signed record, not "counselled the batching operator". Verify the CAPA's effectiveness after a defined period, and record that verification. An NCR closed without effectiveness verification will reopen as the same NCR in three months.

Test register format#

The register is the legal record. Keep one register per material, in a fixed format, with no blank fields. Here is a workable format with a completed row.

Six fields decide whether the register survives an audit: quantity represented (without it you cannot prove coverage, which is the whole point of a frequency chart), lot or batch reference, date sampled and date tested as separate fields (a 28-day cube tested on day 41 is a finding), the edition of the code, the named sampler, and the linked NCR number, which must never be blank when the result column says Fail.

Registers kept as loose paper get reconstructed the night before an audit, and it shows. Wherever the register is maintained, generate it from the actual test records rather than typing it in parallel, so frequency compliance can be computed rather than claimed.

Eight site malpractices and how the supervisor catches each#

1. Cubes cast from a hand-picked "good" batch. The gang waits for a well-mixed load before filling the moulds. Catch it: the sampling must be random and spread across the pour, per IS 456:2000 Cl. 15.2.2. Fix the sampling times before the pour starts, write them on the pour card, and have the PMC nominate at least one of the batches.

2. Cubes cured in a corner instead of a curing tank. Air-cured cubes still often pass at M25, which is exactly why the habit survives - until the day they do not. Catch it: the tank must have a thermometer and a daily temperature log. Cubes with dry, dusty faces or unstained surfaces were never in water.

3. Slump measured only when the PMC is watching. Catch it: slump readings that cluster suspiciously at exactly the design value across every mixer are fabricated. Real slump scatters. Look at the distribution, not the individual readings.

4. Steel samples cut from a bar chosen by the supplier. Catch it: the sampler is named in the register and must be site or PMC personnel. Cut from a bundle picked at random, record the bundle tag and heat number, and photograph the tag next to the cut sample.

5. Sand accepted without a silt or bulking check during the monsoon. Catch it: the daily bulking record is a signed sheet. If the sheet has identical readings for two weeks including three rainy days, it was filled in once. Spot-check by doing the test yourself with the supervisor watching.

6. Cement retested only when a bag looks bad. Catch it: check the week-and-year code on random bags at the gate and against the stack. Cement more than three months old, or stacked more than 10 bags high, or touching an external wall, is a finding regardless of test results.

7. One test result photocopied across several lots. The same lab report number appears against three different GRNs. Catch it: the register's "quantity represented" and "lab report no." columns together make this visible in seconds. Any report number appearing twice must be investigated.

8. Backdated test entries filled in before an audit. Catch it: compare date sampled, date tested and date entered. If the register was written in one pen, one handwriting, one sitting, for three months of work, it is reconstructed. Digital records with entry timestamps remove the possibility entirely.

9. Testing only the first consignment from a source, then never again. The first load from a new quarry is always the good one. Catch it: the frequency chart's per-quantity trigger, tracked against actual quantities received from the GRN totals. Compute tests-due against quantity-received monthly.

10. Failed results quietly discarded and a fresh sample tested. Catch it: lab report numbers are sequential per lab. Gaps in the sequence for your project, or reports missing from the file, indicate suppressed results. Ask the lab directly for a list of all reports issued against your project code.

FAQ#

Is there a single IS code that gives the frequency of testing of construction materials?#

No. This is the most common misconception. IS 456:2000 Clause 15.2.2 fixes the sampling frequency for concrete, and product standards such as IS 1786:2008 for steel and IS 5454:1978 for bricks define lot sizes and sample sizes for acceptance sampling. Everything else - cement, aggregates, water, soil, bituminous mixes - takes its frequency from the works specification: CPWD Specifications, MoRTH Specifications for Road and Bridge Works, or the project's own Quality Assurance Plan. The IS test-method codes give you the method and the acceptance limit, not the interval.

How many cubes should be cast for 50 cubic metres of M25 concrete?#

Under IS 456:2000 Clause 15.2.2, 50 cubic metres of a single grade in one day's work requires 4 samples. A sample is 3 cubes for the 28-day acceptance test, so 12 cubes minimum. Most sites cast an additional 3 cubes per sample for a 7-day indicative result, giving 24 cubes. The 7-day cubes are for information and early warning only; acceptance is decided on the 28-day results.

Is one cube below the design strength a failure?#

Not necessarily. IS 456:2000 acceptance works on samples, not individual cubes. A sample is the average of the specimens tested at 28 days. The two conditions are that the mean of any four non-overlapping consecutive samples is at least fck plus 4 N/sq mm (for M20 and above), and that no individual sample falls below fck minus 4 N/sq mm. A single low cube within a sample whose average is satisfactory does not, by itself, fail the concrete - but it should trigger a look at the sampling and curing.

Which edition of IS 516 should test reports quote?#

IS 516:1959 has been superseded. Compressive, flexural and split tensile strength testing of hardened concrete is now covered by IS 516 (Part 1/Sec 1):2021. Core testing sits in IS 516 (Part 4). If a laboratory's report format still cites IS 516:1959, ask them to update it, and check whether the rest of their formats are equally out of date.

How often must sand be tested for bulking?#

Whenever the concrete or mortar is batched by volume, bulking must be measured daily and re-measured after any rain, because moisture content changes the loose volume of sand by 20 to 30 percent. The method is IS 2386 (Part 3):1963. On weigh-batched sites the bulking geometry stops mattering, but the free moisture content still has to be measured and deducted from the mixing water. Skipping this during the monsoon is the single most common cause of a concrete mix drifting away from its design.

What is the sampling frequency for TMT bars under IS 1786?#

There are two different answers and they are often confused. IS 1786:2008 Clause 11.1 sets the manufacturer's frequency: not less than 2 test specimens per cast for casts below 100 tonnes, and 3 per cast for casts of 100 tonnes or more, cut at random from each size of finished bar. That governs what the mill does and what the MTC reports. The tonnage-based frequency widely quoted on Indian sites - one sample per 25 tonnes for bars under 10 mm, per 35 tonnes for 10 to 16 mm, per 45 tonnes above 16 mm - is a works-specification and QAP convention for independent site testing, not a clause of IS 1786. Take the site frequency from your contract, and do not cite IS 1786 as its authority.

Does a manufacturer's test certificate replace independent testing?#

No. An MTC is a review-point record: you check that it exists, that the heat or batch number matches the delivered material, and that every parameter is within the standard. Independent testing at the specified frequency runs in parallel. The MTC certifies what left the plant; the independent test certifies what arrived on your site, which is not always the same material.

How many bricks are tested from a delivered lot?#

Under IS 5454:1978, for a lot of up to 50,000 bricks, 20 bricks are drawn at random from different parts of the stack. From that sample, 5 are tested for compressive strength under IS 3495 (Part 1):1992, 5 for water absorption under Part 2, 5 for efflorescence under Part 3, with the balance used for dimensional and warpage checks. Larger lots take larger samples per the standard's table.

Which lab should the tests be done in?#

Where the contract specifies NABL accreditation, only a NABL-accredited laboratory's report is acceptable for the parameters listed in that lab's scope of accreditation - and the scope matters, because a lab accredited for concrete may not be accredited for chemical analysis of steel. Site field labs are used for routine control tests such as slump, moisture and field density; confirmatory and dispute-resolution testing goes to the independent accredited lab. Check the lab's certificate number and the exact scope before you appoint it, not after the first dispute.

What happens to the work if concrete cubes fail?#

Under IS 456:2000 Clause 17.4, the engineer decides. The options run from accepting the work with a reduced load rating, through core testing under IS 516 (Part 4) and load testing of the member, to demolition and rebuilding. Core acceptance is an average equivalent cube strength of at least 85 percent of fck with no individual core below 75 percent. Every step is recorded through the non-conformance process, and any decision to accept sub-standard work must come from the designer, in writing, not from the site.

How do I prove that the required test frequency was actually met?#

You need the quantity of material received or work executed, and the number of valid tests covering it, in the same record. That is why the "quantity represented" column in the test register matters more than any other field. At the end of each month, compute tests-due from the quantities in the GRNs and the measurement records, compare to tests-done, and record the reconciliation. If the two do not match, catch it that month, not at handover.