The pour finished at 4:40 PM. By 6:15 the finishers were still on the slab and someone noticed hairline cracks running in a loose herringbone across the middle bay: 300 to 600 mm long, spaced roughly half a metre apart, none of them reaching the edge forms. By the next morning there were forty of them. The batching plant got the call first. Something must be wrong with the cement, or the sand was dirty, or the mix was too rich.
None of that was it. The cylinders broke fine at 7 and 28 days. The slump was on ticket. What actually happened is that between 1:00 PM and 5:00 PM the slab surface lost water faster than the concrete could bleed it back. The top few millimetres went into tension while the paste still had essentially no tensile strength, and it tore. That is plastic shrinkage cracking, and the variable that drives it does not appear anywhere in the mix design.
Plastic shrinkage cracking is a race between two rates: the rate at which water evaporates from the surface, and the rate at which bleed water rises to replace it. When evaporation wins, the surface dries, capillary tension builds in the paste, and the slab cracks before it has any strength to resist. Everything on the left side of that race is weather: air temperature, relative humidity, wind speed, and the temperature of the concrete itself.
Three things make it invisible on a typical Philippine job. First, the four inputs are site conditions, and site conditions belong to nobody on the inspection and test plan. QA/QC measures slump, air content where required, and fresh concrete temperature (the standard test method for concrete temperature is well established and usually on the ITP), but air temperature, humidity and wind speed are almost never recorded on the pour card. Second, the check has no natural owner. The mix designer cannot fix it, the batching plant cannot fix it, and the finishers only find out when it is already too late. Third, the cracks appear before the first cylinder result, so by the time anyone argues about cause, the argument has already been reframed as a materials problem.
It is worth saying plainly: neither the concrete chapters of NSCP 2015 nor ACI 318 give you a numerical evaporation-rate limit. They require that concrete be protected and cured. The evaporation-rate relationship and its threshold numbers come from hot-weather concreting guidance, principally ACI 305R, and are reproduced in slab construction guidance. That is a guide, not a code mandate.
There is one exception that matters more than the rule, and it is the reason the guide is not as toothless as it sounds. ACI 305.1 is the companion specification to that guide, written in mandatory language rather than advisory language, and where a project specification incorporates it, or incorporates ACI 301, by reference, hot-weather protection stops being advice and becomes a contract obligation. Incorporation by reference is exactly how a guide becomes a requirement. So read your own specification's reference list before you decide which situation you are in. What you should still not do is write the evaporation-rate number itself into a specification as a pass/fail acceptance criterion without saying exactly what happens when it is exceeded.
The widely reproduced empirical expression, attributed to Menzel and normally presented as a four-quadrant nomograph, is:
E = evaporation rate in kg/m²/h; Tc = concrete surface temperature in °C; Ta = air temperature in °C; r = relative humidity as a decimal; V = wind speed in km/h. The caret means "raised to the power of", so [Tc + 18]^2.5 is (Tc + 18) squared, multiplied by √(Tc + 18).
This is an empirical fit, not a derivation. Its published origin is evaporation from a free water surface, not from concrete, and it was developed to bound the problem, not to predict it precisely. Its practical limits of use are worth stating: it assumes a bleed-water film is present, so it overestimates once bleeding stops; it carries solar radiation only indirectly, through the concrete temperature you feed it; and it takes no account of how fast a particular mix bleeds. A reviewer is entitled to point out that cracking has been reported below the guidance thresholds for some mixes and absent above them for others, and that simplified algebraic versions of the same nomograph have since been published giving broadly similar numbers. Treat E as a screening index that tells you how hard the environment is pulling, not as a prediction of crack width.
Two working figures are in general use, and both of them come from the same document, ACI 305R, the hot-weather concreting guide. Attribute them to it by name whenever you write them down, and record which edition you worked from, because the guide has been reissued and an unattributed threshold is the first thing a reviewer will strike. Around 1.0 kg/m²/h is the level at which plastic shrinkage cracking becomes likely and protective measures are normally taken. Around 0.5 kg/m²/h is the level to start watching for mixes that bleed slowly: low water-cement ratio, silica fume or other supplementary fines, high cement content, or any mix with a sticky, tight surface. Both are guidance values out of a guide, not mandates, and neither is a code limit.
Everything below is a screening calculation on a stated set of inputs, and on a calculation like this the inputs decide more than the arithmetic does. Here is the full list, including the ones that usually stay hidden inside the working.
Air 34 °C, concrete 31 °C, relative humidity 55 percent, wind 15 km/h.
That is below 1.0 and about 16 percent above the 0.5 watch level. Say it that way rather than calling it comfortable, because on an index this rough a 16 percent margin is no margin at all. For an ordinary bleeding mix it is a watch condition; for a low water-cement ratio podium mix with fines, it is a protect condition. Now note how fragile that reading is. If the concrete arrives at 35 °C instead of 31, because the aggregate stockpile sat in the sun and the cement came hot from the silo, then Tc + 18 = 53, and 53^2.5 = 2809 × 7.28011 = 20449.83. Subtracting the same 10724.35 gives 9725.48; 5 × 9725.48 = 48627.40; × 19 = 923920.60; E = 0.924 kg/m²/h. Four degrees of concrete temperature took the pour from 0.58 to 0.92.
Do not read 0.92 as still safely inside the protect band. The 1.0 trigger is a screening figure with scatter of its own, this same article notes that cracking has been reported below the thresholds, and the workflow above puts anything above 1.0 at reschedule. A computed 0.92 is inside that scatter. Treat it as a reschedule candidate, not as a number you place against with the standard package.
Air 27 °C, concrete 29 °C, relative humidity 80 percent, wind 6 km/h.
The afternoon pour evaporates 0.578 ÷ 0.214 = 2.70 times as fast as the evening pour. Put the other way round, the evening pour runs at 37 percent of the afternoon rate, a 63 percent reduction, and it sits below both guidance figures. Now put it in litres, with the caveat attached: what follows is an upper-bound screening figure and not a measured loss, because it holds an instantaneous E constant for four hours and treats the whole 220 m² as exposed the entire time, when placement is progressive and the expression overestimates once bleeding stops. On that basis, over a four-hour window, the afternoon pour is pulling 0.578 × 220 = 127.16 kg/h, so 127.16 × 4 = 508.6 kg off the surface. The evening pour is pulling 0.214 × 220 = 47.08 kg/h, so 47.08 × 4 = 188.3 kg. The afternoon slab is the one that has to replace 508.6 kg, roughly half a cubic metre, by bleeding, and cannot. Of that, 508.6 - 188.3 = 320.3 kg is the penalty for pouring at 1:00 PM instead of 6:00 PM.
One more lever from the same arithmetic. Keep the 1:00 PM conditions but assume a hoarding and shade cloth drop the wind from 15 to 5 km/h. The (V + 4) factor goes from 19 to 9, so E = 30413.25 × 9 × 10^-6 = 0.274 kg/m²/h. That is a reduction of 10 ÷ 19 = 52.6 percent from a windbreak alone, on an assumed wind reduction you should verify with the anemometer inside the enclosure.
If you must pour in the afternoon, the term you can still buy down is Tc. Assume this batch, per cubic metre, with the masses as batched and aggregate free moisture neglected, as stated in the assumptions above:
Specific heats: solids 0.92 kJ/kg per °C, water 4.19 kJ/kg per °C, latent heat of fusion of ice 334 kJ/kg.
Heat capacity of the batch. Solids = 1050 + 750 + 380 = 2180 kg; 2180 × 0.92 = 2005.6 kJ per °C. Water = 190 × 4.19 = 796.1 kJ per °C. Total = 2005.6 + 796.1 = 2801.7 kJ per °C.
Heat content above 0 °C. Solids: 1050 × 34 = 35700; 750 × 34 = 25500; 380 × 45 = 17100; sum = 78300; and 0.92 × 78300 = 72036 kJ. Water: 796.1 × 28 = 22290.8 kJ. Total = 72036 + 22290.8 = 94326.8 kJ.
Baseline mixture temperature = 94326.8 ÷ 2801.7 = 33.67 °C. That is a hot mix before it even leaves the plant, and it is the number to argue from for this pour, not the 31 °C of Case 1. Add the neglected aggregate free moisture and it goes higher still.
Now replace 60 kg of the 190 kg of mixing water with 60 kg of flake ice at 0 °C. The ice is part of the water, not additional to it, so the water-cement ratio does not change. Remaining water = 190 - 60 = 130 kg, carrying 4.19 × 130 × 28 = 544.7 × 28 = 15251.6 kJ. The ice contributes no sensible heat at 0 °C and absorbs 60 × 334 = 20040 kJ of latent heat as it melts. Since the melted ice ends up as water in the mix, the denominator is unchanged at 2801.7 kJ per °C.
Feed 24 °C back into the afternoon evaporation case: Tc + 18 = 42; 42^2.5 = 1764 × 6.48074 = 11432.03; less 10724.35 = 707.68; 5 × 707.68 = 3538.40; × 19 = 67229.60; E = 0.067 kg/m²/h. Do not take that at face value. The nomograph wants the concrete surface temperature, and a slab discharged at 24 °C under an open sky will drift upward within the hour. Assume the surface reaches 28 °C by the time you are finishing, an assumed drift and not a derived one: 46^2.5 = 2116 × 6.78233 = 14351.41; less 10724.35 = 3627.06; 5 × 3627.06 = 18135.30; × 19 = 344570.70; E = 0.345 kg/m²/h. Still below the 0.5 watch level, which is the honest answer for that assumed drift, and the reason to state the assumption in the open is that the whole result hangs on it.
The practical ceiling on ice is real. You cannot replace more water than the mix contains. Take all 190 kg as ice and the numerator becomes 72036 - (190 × 334) = 72036 - 63460 = 8576 kJ, giving 8576 ÷ 2801.7 = 3.06 °C, which is nonsense for placing and would never fully melt in the drum anyway. Unmelted ice at discharge leaves voids and local wet pockets. Use crushed or flake ice, extend mixing time, and inspect the discharge.
Using illustrative rates only, not quotations, and with no date, region or scope behind them: crushed ice delivered at ₱6.00/kg, evaporation-retarder film applied at ₱45/m² per application, a fogging rig plus operator at ₱3,400 per shift, and crack routing and sealing at ₱900 per linear metre.
Slab volume = 220 × 0.150 = 33.0 m³. Ice at 60 kg/m³ = 60 × 33.0 = 1980 kg, costing 1980 × 6.00 = ₱11,880, which is 11880 ÷ 33 = ₱360 per m³ or 11880 ÷ 220 = ₱54.00 per m². Retarder applied twice = 220 × 45 × 2 = ₱19,800. Fogging one shift = ₱3,400. Prevention total = 11880 + 19800 + 3400 = ₱35,080, or 35080 ÷ 220 = ₱159.45 per m². Sealing 60 linear metres of cracking afterwards = 60 × 900 = ₱54,000, or ₱245.45 per m² — and that only restores durability, not appearance and not the QA record.
Then say the uncomfortable part out loud, because that 60 linear metres is assumed rather than derived and it is the single figure deciding which side wins. The break-even is 35080 ÷ 900 = 38.98 linear metres. Below about 39 metres of cracking, the prevention package costs more than the repair. The forty cracks in the opening story, at 300 to 600 mm each, come to only about 18 metres, well under break-even, so on a narrow cash comparison that slab would have been cheaper to repair than to protect. What the cash comparison leaves out is that repair does not restore appearance, does not restore the QA record, and does not end the argument about whether the concrete was defective. Make the decision with the break-even in front of you rather than behind you.
Each of these acts on a different place, and only two of them touch the equation at all.
This is also why laying plastic sheeting over the slab once cracks have appeared achieves nothing structural. The tear happened while the paste was still plastic. Once it has passed final set, the crack faces cannot reknit, and sheeting only limits further loss from concrete that has already cracked. The only window to close a plastic shrinkage crack is before final set, by re-vibrating or re-floating the surface while the concrete still responds to a float. Miss it and the crack is permanent. After final set the remedy is durability repair, the routing and sealing priced above, not closure.
Time is the strongest discriminator. Plastic shrinkage cracks appear within roughly 30 minutes to 6 hours of placing, often while finishers are still on the slab. Drying shrinkage cracks show up days to weeks later.
Geometry is the second. Plastic shrinkage cracks are short, typically 50 to 1000 mm, roughly parallel or in a loose herringbone, often at an angle to the long axis of the bay or across the wind direction, spaced a few hundred millimetres apart, with rounded edges because the paste was still soft. They rarely run out to a free edge. Drying shrinkage cracks run between points of restraint, are sharp-edged, often go through or near through the depth, and start at re-entrant corners, penetrations, column heads and missed or late-cut control joints. If a crack runs the full bay width from a wall to a column, that is restraint, not evaporation.
Depth is the third discriminator and the weakest of the three, so use it last. Plastic shrinkage cracks are commonly reported dying out within the top 20 to 50 mm of a core, but that is a reported practice range and not a rule: deeper cracks are reported too, and in a thin slab they occasionally run the full depth. Do not treat shallowness you have not cored for as evidence that the crack is cosmetic.
None of the arithmetic above is hard, but doing it four times on a pour day with a phone calculator, in the sun, while trucks are queueing, is how it stops getting done. We keep the evaporation-rate calculation among the free field calculators at RHCES Web Tools so a site engineer can punch in four readings and get the number before the first truck backs up.
Yes, but stop it on the right grounds, and start by reading your own specification instead of a generalisation about specifications. Most incorporate a general obligation on the contractor to protect fresh concrete and to place only under conditions that permit proper finishing and curing, and many incorporate a hot-weather specification by reference, which is what converts guidance into a requirement. The evaporation rate is your evidence that those conditions do not exist, not the obligation itself. Record the four readings, the computed E, the mix type and its bleeding behaviour, and the protection available on site, then raise it as a request for instruction. Rewriting the spec mid-project to insert a numerical limit invites a claim; documenting an engineering judgment does not.
Fibres, particularly fine synthetic microfibres at typical dosages, are well established as a way to reduce plastic shrinkage cracking, and there is a recognised laboratory test method for comparing mixes on exactly that basis. What fibres do is reduce the crack area that forms under a given drying condition. They do not lower the evaporation rate, and they do not substitute for fogging, windbreaks or timing. Use them as one layer of the package, and still compute E.
It is realistic at discharge if the mix has been chilled, the aggregate is shaded and the trucks are not queued in the sun, and that chilled and shaded condition is an assumption of Case 1, not a general expectation. Without it the batch balance in the same example says the plant hands you 33.67 °C. It is also often not realistic an hour later. The surface warms toward the ambient and solar condition, and because Tc enters the expression raised to the power 2.5, that drift matters more than any other single input. This is the argument for taking the reading again during finishing rather than only at first discharge, and for treating the highest computed value of the day as the one that governs the decision.