Finding Peace in Shared Spaces: Tackling Apartment Noise in Korea — Epoche B2
The Sound from Upstairs The Korean term cheunggan soeum — inter-floor noise, the sound that arrives from the flat above — is the subject of this essay, and the fact that the language has a settled compound for it tells you how ordinary the problem is. Almost everyone I know in Seoul lives in an apartment, and almost all of them have a story about the floor above. Ours is a child who begins running at about seven in the morning and stops, roughly, at nine at night. Not stamping. Running, in a straight line, the length of the flat and back. What follows is in three parts: why the sound behaves the way it does, why the reaction to it is so much stronger than the measured level would suggest, and which of the available remedies actually worked in our building. The physics comes first because both of the other parts depend on it. Why it is in the ceiling and not merely above it The first thing to understand is that there are two quite different ways sound gets from one flat to another, and inter-floor noise is almost entirely the second. The first path is airborne. A television upstairs sets the air moving; that air pushes on the underside of the slab; the slab flexes very slightly and radiates into your room. This path is extraordinarily inefficient, and the reason is a mismatch in what acousticians call characteristic impedance — the product of a material's density $\rho$ and the speed of sound $c$ within it, written $z = \rho c$, which measures how hard it is to get a material moving with a sound wave. For air, $\rho \approx 1.2\ \mathrm{kg\,m^{-3}}$ and $c \approx 343\ \mathrm{m\,s^{-1}}$, so $z \approx 410\ \mathrm{Pa\,s\,m^{-1}}$. For concrete, $\rho \approx 2300\ \mathrm{kg\,m^{-3}}$ and the wave speed is of the order of $3400\ \mathrm{m\,s^{-1}}$, giving $z \approx 7.8 \times 10^{6}$ in the same units — a ratio of about nineteen thousand. When a wave meets a boundary between two materials of impedance $z_1$ and $z_2$, the fraction of its power that crosses rather than reflecting is $$T = \frac{4 z_1 z_2}{(z_1 + z_2)^2},$$ which, when the second impedance is very much the larger, reduces to approximately $4z_1/z_2$. Putting the numbers in: $$T \approx \frac{4 \times 410}{7.8\times 10^{6}} \approx 2.1 \times 10^{-4},$$ about two parts in ten thousand, or a loss of roughly 37 decibels at the air-to-concrete interface alone — and the wave has to make the crossing twice, once going in and once coming out. This is why you rarely hear your upstairs neighbour's conversation. The air can barely get hold of the floor. The second path evades that obstacle entirely. A heel striking the floor is not air pushing on concrete; it is a solid object delivering an impulse directly into the slab, with no interface to cross. The energy travels through the structure as a bending wave and is then radiated by the whole underside of your ceiling, which is a large and efficient loudspeaker. Footsteps, a dropped chair, a suitcase dragged across a room: these do not arrive downstairs as noise from somewhere else. They arrive as noise inside your own ceiling, from directly overhead, with no wall or door to place them behind — and that difference in apparent location, rather than the level, is a large part of why they are so hard to ignore. What 120 millimetres means Part of the explanation is the building itself, which went up in 1998. Slabs in Korean apartments built before the 2005 revision of the Regulations on Standards for Housing Construction are often only around 120 millimetres thick, against the 210 millimetres the revision requires for the wall-bearing construction that dominates Korean housing. It is worth being precise about what that extra thickness does, because the intuitive answer — nearly twice as thick, therefore vastly quieter — is wrong, and the regulation is built around the fact that it is wrong. For airborne sound through a simple partition, transmission loss follows what is called the mass law: it improves by about six decibels for each doubling of mass per unit area. A 120 mm slab at $2300\ \mathrm{kg\,m^{-3}}$ weighs $276\ \mathrm{kg\,m^{-2}}$; a 210 mm slab weighs $483\ \mathrm{kg\,m^{-2}}$. The ratio is 1.75, so the mass-law improvement is $$20\log_{10}(1.75) \approx 4.9\ \mathrm{dB},$$ about five decibels — audible, but nothing like a solution, and impact sound does not obey the mass law cleanly in any case, since a thicker slab also changes the bending stiffness and shifts which frequencies radiate. This is why the 2005 rules do not stop at a thickness. They also set measured performance limits on floor impact sound, tested with standardised sources: a tapping machine of small hammers for light impacts, and a heavy source for the low-frequency thud of a footfall, with the ratings derived by the method set out in the international standard ISO 717-2. A builder may satisfy the requirement either by using an approved standard floor construction or by demonstrating that a particular assembly meets the measured limits. In practice most of the improvement comes not from the slab but from what sits on it: a resilient layer and a floating screed above the concrete, which decouple the surface you walk on from the structure that carries the sound. Our 1998 building has none of that. Why the reaction is out of proportion to the reading Here the essay turns from acoustics to people, and the turn matters, because everything in the next section depends on it. If annoyance were a function of sound level, the remedy would be engineering and nothing else. It is not. The decibel itself is a logarithmic measure: a sound pressure $p$ is expressed as $L = 20\log_{10}(p/p_0)$ against a reference $p_0 = 20\ \mu\mathrm{Pa}$, roughly the quietest sound a young ear detects at a thousand cycles per second. On that scale five decibels is a factor of about $10^{5/10} \approx 3.2$ in acoustic power, which is a real difference and not a rounding. But R. F. S. Job, reviewing several decades of community noise surve