Honeycombed Column: Core Results Pass but the Sampling Does Not
At the first-floor level of a G+6 office building, de-shuttering of a 400 mm × 400 mm column of M30 concrete (Fe500, 8–20 mm bars, 40 mm clear cover) reveals a honeycombed patch about 350 mm × 250 mm and 60 mm deep on one face, 1.2 m above the floor slab, with two links exposed. The contractor proposes to plaster the patch and proceed with the upper floors. The owner's engineer orders three 75 mm diameter cores, which give 26.5, 25.2 and 28.0 MPa. Two cores were drilled in sound concrete immediately above the honeycombed zone and one in a sound column cast from the same pour. Six cubes from that pour averaged 34.0 MPa with a minimum of 32.0 MPa. The column carries a factored axial load of 1850 kN from the floors above and was designed for 2600 kN. Cores were tested 35 days after casting. The site is coastal, and the beam–column junction 600 mm above the defect is heavily congested with reinforcement.
1. Apply the standard core acceptance criteria to these results and state whether the concrete in the member can be accepted.
Read the worked answer
Criterion. Concrete in a member is accepted on cores if the average of three cores is at least 0.85 fck and no individual core falls below 0.75 fck.
- 0.85 × 30 = 25.5 MPa (average requirement)
- 0.75 × 30 = 22.5 MPa (individual minimum)
Check. Average = (26.5 + 25.2 + 28.0)/3 = 79.7/3 = 26.57 MPa ≥ 25.5 MPa - satisfied. Minimum = 25.2 MPa ≥ 22.5 MPa - satisfied.
So, read literally, the member passes. Two caveats must be stated rather than glossed over. First, the average clears the threshold by only 1.07 MPa (about 4%), which is inside the scatter of core testing; a single low core would have failed the member. Second, and decisively, the criterion accepts the concrete in the member, and the member includes the defective zone - which was never sampled. The 35-day age is acceptable (cores are normally taken at 28 days or later), but cores should be in the moist condition and should not be drilled where reinforcement is cut.
The correct answer is therefore: the numbers satisfy the acceptance rule, but they do not close out the defect, because the tested material is not the material in doubt. The investigation, not the concrete, is deficient.
2. Critique the sampling plan and explain how it changes the conclusion you would report.
Read the worked answer
What the plan actually proves. All three cores represent sound concrete: two from immediately above the honeycomb and one from a different column. The result establishes that the batching, mixing and cube compliance were fine - which the 34.0 MPa cube average already showed. It says nothing about the 60 mm deep, 350 × 250 mm void-riddled zone, which is the only region in question.
Why the location matters here. Honeycombing of that size indicates locally inadequate vibration, blockage of aggregate by congested steel, or a leak in the formwork. Such zones have a weak, porous skin with poor bond to the reinforcement; cover over the links may be effectively zero. In a coastal environment, that is a durability problem before it is a strength problem, because chlorides reach the steel through the honeycombed face.
What should have been done.
- Map the extent first by rebound hammer or ultrasonic pulse velocity to find the boundary of the defective concrete.
- Take at least one core straddling the defect–sound interface, so that the core log records the depth and character of the porous zone.
- Record core orientation, presence of reinforcement, and the density/voidage of the core face.
Revised conclusion. Reporting 'cores pass, hence column accepted' would be technically correct and practically misleading. The honest statement is that the surrounding concrete is compliant and the defect must be assessed as a localised cover and section-loss issue, repaired, and its cause traced through the placing records of that pour.
3. Recommend the remedial action and check whether the section loss is structurally significant.
Read the worked answer
Section loss check. Plan area removed from one face = 350 × 250 = 87,500 mm²... but only if the full face depth were lost. The patch is 60 mm deep into a 400 mm face, so the plan area of concrete affected is 87,500 mm² of a 160,000 mm² gross section, i.e. 55% of the face area, but the material lost is 60/400 = 15% of the depth on one face.
- Gross area = 160,000 mm²; lost area ≈ 350 × 60 = 21,000 mm² (13% of the section, eccentric to one face).
- Approximate axial capacity reduction, treating the loss as an eccentric area removal about the centroid, is of the order of 13–15%: 0.87 × 2600 ≈ 2260 kN.
- Applied factored load = 1850 kN, so Pu/φPn ≈ 0.82 even after the loss. Strength is adequate with a small margin; the defect is not a demolition trigger.
But the defect sits 1.2 m above the floor - outside the confined joint region - and the upper floors are not yet cast, so the load will increase. Re-check with the final Pu.
Repair sequence.
- Chip back honeycomb to sound, dense concrete with a square-edged boundary; do not feather the edges.
- Clean and pre-soak; apply a bonding agent and fill with polymer-modified mortar or micro-concrete of at least M35, in lifts, with proper curing.
- Verify cover over the two exposed links after repair; if less than 40 mm, add a protective coating (polymer-modified cementitious or epoxy) and consider a local jacket.
- If the defect encroaches on the junction, provide a reinforced jacket rather than a patch.
- Inspect all columns of that pour; review vibration practice and aggregate size against the 40 mm cover and bar congestion.