Project

Ellerby House, Farringdon

An open-plan floor held to a lecture room’s reverberation band, and a rooftop plant deck that had to lose 19 dB.

Mid-frequency RT60

0.67 s

Brief: Lecture room, 0.6–0.8 s

An empty seating rake photographed before handover
An empty seating rake photographed before handover
Specification

The room, as recorded.

Every figure below is an input to the arithmetic further down this page. Anything we have not measured is simply absent.

Room schedule
Volume 2,693 m³
Total surface area 1,899 m²
Length × width × height 34.0 × 22.0 × 3.6 m
Client Ellerby & Vane LLP
Architect Ardley Mottram
Completed 2023
Surface schedule
Surface Area
Suspended acoustic tile 320.0 m²
Plaster on masonry 428.0 m²
Carpet on underlay 500.0 m²
Parquet on concrete 248.0 m²
Heavy plate glass 220.0 m²
Plasterboard on studs 183.2 m²
Reverberation

What we predicted, and what it did.

0.67 s at mid frequencies, against a Lecture room brief of 0.6 to 0.8 seconds.

Both equations are published. Where they part company the room has become absorptive enough that Sabine's assumption of continuous decay is straining, and the lower figure is the one to believe.

Band 125 250 500 1,000 2,000 4,000
Measured, occupied 1.02 s 0.88 s 0.71 s 0.62 s 0.58 s 0.52 s
Computed — Sabine 1.11 s 0.93 s 0.73 s 0.62 s 0.60 s 0.60 s
Computed — Eyring 0.99 s 0.81 s 0.61 s 0.49 s 0.48 s 0.47 s
0.0 0.5 1.0 1.5 125 250 500 1k 2k 4k seconds Hz
Reverberation time by octave band. At mid frequencies the room measures 0.67 seconds, against a Lecture room target of 0.6 to 0.8 seconds.
Modal behaviour

Below the crossover, a room has notes.

Schroeder frequency 31.4 Hz.

Above 31.4 Hz this room behaves statistically and a reverberation time means something. Below it the room does not reverberate at all — it resonates, at the discrete frequencies listed here, and where those bunch together is where the bass will be uneven.

Axial modes, shown warm, run between one opposed pair of surfaces and carry the most energy. They are the ones anybody in the room actually hears.

Room modes below 120 Hz
Hz Order Kind
5.0 1 0 0 Axial
7.8 0 1 0 Axial
9.3 1 1 0 Tangential
10.1 2 0 0 Axial
12.7 2 1 0 Tangential
15.1 3 0 0 Axial
15.6 0 2 0 Axial
16.4 1 2 0 Tangential
17.0 3 1 0 Tangential
18.6 2 2 0 Tangential
20.2 4 0 0 Axial
21.6 4 1 0 Tangential
21.7 3 2 0 Tangential
23.4 0 3 0 Axial
23.9 1 3 0 Tangential
25.2 5 0 0 Axial
25.5 2 3 0 Tangential
25.5 4 2 0 Tangential
26.4 5 1 0 Tangential
27.9 3 3 0 Tangential
29.7 5 2 0 Tangential
30.3 6 0 0 Axial
30.9 4 3 0 Tangential
31.2 0 4 0 Axial

122 further modes below 120 Hz are not listed.

Take it apart

The same room, in your hands.

These are this building's own dimensions and its own surface schedule. Change a material or move a wall and the reverberation time moves with it, by the same equation we used on the drawings.

The room

38 m
21 m
17 m

The surfaces

The fourth floor of Ellerby House is 34 m by 22 m by 3.6 m — 2,693 m³ of open plan for about 170 people — and we held it to the lecture-room band of 0.6 to 0.8 s. That will look like a filing error to anyone who reads the target label, so it is worth explaining: the problem in an open-plan floor is the same problem as in a teaching room. Somebody is speaking, somebody four metres away needs to understand them, and somebody twelve metres away needs not to be able to. Reverberation control is how you make the second of those true, and the band that works for a lecturer works for a desk. The computed figure is 0.73 s.

Getting there in a flat-ceilinged box is mostly a question of not overdoing it. A fully tiled ceiling with carpet throughout computes to 0.50 s here, and a floor that dry is a floor where every conversation carries with unnatural clarity because there is no diffuse field to bury it in. Open-plan offices are made unbearable by too much absorption at least as often as by too little. We tiled 320 m² of the 748 m² ceiling in rafts over the desk clusters, left the rest as exposed plaster on the slab for its thermal mass, carpeted 500 m² and left the circulation spine in hard parquet so that footfall reads as movement rather than as intrusion.

The 220 m² of heavy plate glazing is the one element working against us, at 0.04 across the speech range, and it is why the 125 Hz figure sits at 1.11 s. Nobody notices. Low-frequency reverberation in an office is inaudible against the ventilation, which is the point at which room acoustics stops and noise control begins.

Most of the fee went on the roof. The landlord’s chillers and two AHUs sit on a deck twelve metres from the nearest residential window across the mews, and the planning condition was 10 dB below the measured background of LA90 41 dB at night — which is to say 31 dB at the facade, from plant with an unattenuated sound power that put 50 dB there. We took 19 dB out with a 3.5 m acoustic screen on isolators, attenuators on both AHU discharges, and low-speed fan selections on the chillers that cost the mechanical budget about nine per cent in capital and gave most of it back in running cost. We monitored for a fortnight after commissioning across four positions rather than the one the condition required, because a single position and a single night is a number you can be unlucky with, and one we would rather find ourselves than have found for us.

Enquiries

Have you a room like this one?

A drawing, a volume, or a recording of the problem. Any of the three is enough to start, and the first conversation is not charged for.

Unit 4, Rope Walk
London SE16