M25 Raft: Cube Results That Fail the Acceptance Rule
M25 grade concrete for the 900 mm thick raft of a 33/11 kV substation control building was produced from a site batching plant with 53-grade OPC, 20 mm crushed aggregate and river sand, at a designed water–cement ratio of 0.50 and a target slump of 75 mm. The approved mix had a target mean strength of 31.6 N/mm². Sixty cubes were cast over three weeks; the 28-day results of the last four consecutive sets were 23.5, 22.1, 26.8 and 19.8 N/mm², against an overall standard deviation of 4.6 N/mm². The contractor's records show that on two occasions the sand had a silt content near 8%, that 40% of the cement bags arrived three weeks late and were stored in the open, and that the batching plant's water meter was last calibrated two years ago. The contractor argues the results are 'close enough' because no cube fell below 20 N/mm². The raft has already been cast and is to be loaded with 11 kV switchgear next month.
1. Apply the standard acceptance criteria to these results and state whether the concrete can be accepted. Address the contractor's argument directly.
Read the worked answer
Acceptance test (IS 456, cl. 16.4, read with IS 516). Two conditions must both hold for a group of results:
- The mean of any four consecutive results ≥ fck + 0.825σ.
- No individual result < fck − 3 N/mm².
Arithmetic. Mean of the four results = (23.5 + 22.1 + 26.8 + 19.8)/4 = 23.05 N/mm². With the permissible assumed σ = 4.0 N/mm² for M25 under good control, the required mean is 25 + 0.825×4.0 = 28.3 N/mm²; using the observed σ of 4.6 N/mm² it is 28.8 N/mm². The individual limit is 25 − 3 = 22.0 N/mm², and one cube gave 19.8 N/mm².
Verdict. Both conditions fail. The concrete is non-conforming and cannot be accepted on cube evidence; the mean criterion alone is decisive, so the conclusion does not depend on how the individual limit is interpreted. Quote the specific clause and the observed figures in the non-conformance report rather than arguing from impression.
On the contractor's argument. 'No cube below 20' is not an acceptance criterion in the code. Also note the shape of the data: the target mean was 31.6 N/mm² and the achieved mean is 23.05 N/mm² - a systematic shortfall of about 8.5 N/mm², not just scatter. That points to a materials or water-batching bias, not random testing noise, which is exactly why the acceptance rule keys on both mean and variability.
2. Rank the probable causes using only the evidence in the data, and say what each would look like in the results.
Read the worked answer
1. Sand silt near 8% (most likely dominant). IS 383 limits fines for this use to about 3%. At 8% the silt coats aggregate, raises water demand and, in practice, the plant adds water to hold the 75 mm slump - so the effective w/c climbs above the designed 0.50 and strength drops sharply. Signature: a uniform downward shift across all cubes, which is what the data show.
2. Cement stored in the open, delivered late. Bagged cement exposed to humidity for weeks loses strength through lumping and aeration, and the degree of loss varies from bag to bag. Signature: increased scatter (observed σ of 4.6 against an assumed 4.0) superimposed on the mean shift.
3. Water meter uncalibrated for two years. An over-reading meter silently raises w/c; an under-reading meter makes the mix harsh and the crew adds water on the platform. Signature: variation between batches, worse where the slump was 'adjusted' on site.
4. Curing and cube handling. A 900 mm raft sheds heat slowly and is hard to cure; poor initial curing lowers the 28-day result by a few N/mm² but would not by itself produce an 8.5 N/mm² mean deficit.
5. Testing error. Least likely - the low values are consistent across four sets and 60 cubes, which is not a pattern of random cube-testing mistakes. Do still verify cube identity, curing tank records and the calibration of the compression machine before serving notice.
3. Set out the investigation and disposition plan for the already-cast raft. State the acceptance values you would use for core results and flag them as code items to be confirmed against the contract.
Read the worked answer
Step 1 – stop the next pour and issue a non-conformance. Preserve cube records, batch sheets, and cement/sand test reports; quarantine the switchgear erection date.
Step 2 – map uniformity by NDT. Rebound hammer (IS 13311 Part 2) and ultrasonic pulse velocity (Part 1) on a 1 m grid over the raft, with a few readings at depth on the side faces. NDT alone cannot be the acceptance basis, but it shows whether the low strength is local or general and guides core locations.
Step 3 – cores. Extract at least three 100 mm diameter cores from locations away from reinforcement and from the heaviest loaded areas, per IS 516. The criteria commonly applied under IS 516 read with IS 456 are: average core strength ≥ 0.85 fck and no individual core < 0.75 fck. For M25 that is 21.25 and 18.75 N/mm² - confirm these against the contract specification before using them, since some specifications impose different core limits.
Step 4 – disposition. If cores pass, accept the member with a record note and proceed, tightening mix control and testing frequency. If cores fail marginally, commission a structural check of the raft under the switchgear and cable-trench loads and consider strengthening (increased screed/raft overlay, jacketing). If cores are far below, or the concrete is porous and carbonated, plan partial demolition and re-casting of the affected panels - a 900 mm raft with switchgear on it is not a place to trade strength for schedule.
Step 5 – prevention. Wash the sand and test silt daily, reject silt above 3%, calibrate the water meter, store cement on dunnage under cover, and revert to the designed w/c with a plasticiser rather than adding water.