Elastic Modulus from Compressive Strength

Every concrete code carries some version of this: an estimate of stiffness from compressive strength, with a square root in it. It is genuinely useful, because compressive strength is the one property that always gets measured and elastic modulus is one that usually does not, and a designer needs a modulus for every deflection calculation, every camber, every dynamic analysis and every seismic model.

The square root is the physically interesting part. Stiffness rises much more slowly than strength. Increase the compressive strength by a quarter and the modulus rises by only about twelve percent, because a square root halves any proportional change. This has a direct consequence for design: specifying a higher strength class is a poor and expensive way to buy stiffness. Deflection is governed by $EI$, and section depth enters $I$ as a cube — so fifteen percent more depth beats a whole strength class, every time, and costs less. Anyone who has tried to solve a deflection problem by upgrading the concrete has discovered this the slow way.

Now the honesty, and it is the reason $k$ is an input on this site rather than a constant. These are regression lines through somebody's test population. ACI drew one, CSA drew another, Eurocode drew a third, and they used different populations, different specimen conditions and different definitions of what $E_c$ even means — a secant modulus taken to some fraction of the peak stress, and the fraction is not the same everywhere. The coefficients are not interchangeable. Using a coefficient from one code inside another code's framework of load factors, resistance factors and deflection limits is not conservative in either direction; it is simply outside both calibrations, and nothing in the arithmetic will tell you. Some codes also carry a density term for lightweight concrete, or a modification factor, or an alternative expression for high-strength mixes — and if yours does, a plain square root is not the equation you want at all.

And the coefficient is unit-bound, which is the trap that catches people who move between markets. In $E_c = k\sqrt{f'_c}$, $k$ carries the dimensions of a square root of pressure, so its numerical value depends entirely on the unit the strength was written in. The same ACI relation appears as roughly 4700 in megapascals and roughly 57 000 in psi. Those are one line, not two. The conversion is a factor of about 12.04 — the square root of the number of psi in a megapascal — and a coefficient carried across without it is wrong by that factor while printing a perfectly plausible-looking modulus. This site evaluates the root with the strength in megapascals and asks for $k$ in its MPa form, and says so on the page every single time, because a silent convention here is worth an order of magnitude.

The scatter is the real story, though. Concrete's modulus is dominated by its aggregate, which occupies most of the volume and is far stiffer than the paste. Two mixes of identical compressive strength made with a hard trap rock and a soft limestone genuinely differ in stiffness by tens of percent, and no function of $f'_c$ can see that — the strength was set by the paste and the interfacial zone, and the stiffness is being set by the rock. Every code that publishes one of these correlations says in its commentary that the band around it is wide.

So the practical rule is this. For ordinary design, use your code's coefficient inside your code's framework; that is what it was calibrated for and the calibration includes the scatter. But where stiffness genuinely governs — a long-span deflection check, a prestressed member's camber, a vibration-sensitive floor, anything seismic — measure the modulus on the mix you are actually using. It is a real test on a real cylinder and it settles in an afternoon what no correlation can settle at all.