An acoustic detective story: A study in mysteries hums

A problem that AI cannot address

While carrying out acoustic consulting tasks, I often consider the possibility that this work could be completed by AI and to what extent. Acoustic design reports for buildings, especially offices, designed primarily to meet certain standards, likely fall into the category where well-trained AI can complete 80% of the work in 10% of the time. In contrast, AI cannot contribute much to onsite problem-solving, as the story below.

Mysterious hums coming from nowhere.

One day, I received a request for acoustic consultation from a property management company. A resident in one of the apartments they manage had complained about an unbearable mysterious noise in their unit.

The resident described the noise as follows: "The noise is a persistent, low-pitched hum with vibrations. It occurs at irregular times, with varying loudness, but it tends to be louder at night and late at night, seriously disrupting sleep. The noise can be heard in all rooms, but it's unclear where it's coming from. Some areas are louder than others. In the same spot, the noise gets louder when sitting down or lying in bed..."

Staff from the property management company reported hearing the noise during their investigation in the resident's apartment but couldn't identify its source.

Sherlock Holmes' science of deduction

Although this project was small, I was intrigued by the "black box" nature of deducing the cause from the phenomenon. As a fan of Sherlock Holmes, it reminds me of his science of deduction, adding a touch of investigative fun to the routine design work. The occurrence of the noise is like a case, and the mysterious source is the elusive culprit that must be tracked down and apprehended through keen deduction.

From the noise description, it seems to be a low-frequency noise caused by some vibration. Such noise is often hard to pinpoint its location, and due to room modes, it distributes unevenly in the room, typically louder near walls and floors, consistent with the resident's description.

A colleague suggested that there might be a quarry nearby, with vibrations transmitting far and causing secondary structure noise. Another colleague suggested it might be the kitchen's exhaust fan.

"'You don’t seem to give much thought to the matter in hand,' I said at last, interrupting Holmes’ musical disquisition. 'No data yet,' he answered. 'It is a capital mistake to theorize before you have all the evidence. It biases the judgment.'"— A Study in Scarlet

Recalling Holmes' method of avoiding bias in deductions, I decided not to make any assumptions about the noise source before collecting evidence at the "crime scene."

Questionnaire as the first step for the investigation

I called the complaining resident and the property management company, scheduling a visit to the "crime scene" on Friday noon. The resident worriedly asked, "What if the noise doesn't occur when you come?" I assured the resident, "Don't worry. We'll use the process of elimination, turning potential noise sources on and off until we find the culprit."

The management company sent over the apartment building's blueprints. It is a seven-storey building with two units per floor. The first step was to understand the extent and severity of the noise problem. For this, I prepared a questionnaire about the residents' health and their observations of the noise, asking the property management company to distribute it to all residents.

The Site - Island of Bute

From Glasgow, a 40-minute ferry ride took me to the Island of Bute, where the apartment building is located.

I walked from the dock up a path to the hill, and soon, a white cylindrical tower gleaming in the sunlight appeared amidst the green trees. The apartment building had a quite stylish design.

The surroundings of the apartment were very quiet. Besides the path leading to the building, there were no other public facilities nearby that could produce noise.

Within the apartment building, the only noise-producing facilities were the lift and the exhaust fans on the roof serving the bathrooms on each floor. There was no water pump, and with Scottish summers only around 20 degrees Celsius, there was no air conditioning. The apartments had only electric radiators, which the residents rarely needed to use even in winter. I later learned that the building was awarded as one of the most energy-efficient residential buildings in Europe. Its walls were made of double-layer insulation bricks with 300mm of insulation in between, making the total wall thickness 600mm. The windows had triple glazing with excellent airtightness. Despite having soundproof vents, the facade's U-value was only 0.1W/m²K. This also indicate high sound insulation of facades.

Out of the 14 households, 11 responded to the survey or verbally answered the key questions. Apart from the complaining household, the other residents did not notice the described "mysterious" noise. A resident on the top floor complained about the loudness of the roof exhaust fan, and another mentioned being disturbed by the neighbour's washing machine noise.

The complaining household was on the sixth floor, occupied by a mother and her daughter, approximately in their fifties and twenties, respectively. Neighbours mentioned that the daughter had poor eyesight and might be particularly sensitive to noise. However, it was the mother who complained vehemently, saying the "unknown" noise severely affected her sleep.

"That's the noise, that's the noise!"

The mysterious noise did not occur when I entered the flat. I first had the management company turn off the roof fan and measured the background noise with no known sources. Then, the roof fan was turned back on. I realised that the apartment was very quiet. With the fan off, the background noise was 20dBA (NR18), and with the fan on, it was only 24dBA (NR19), lower than in many recording studios. Given the low environmental noise, the fan noise was still audible. The complaining mother found the fan noise acceptable, indicating that this was not the problematic noise.

I had the management staff visit the neighbouring apartments, turning on devices like water pipes and household appliances one by one. I remained in the complaining apartment. Each time a device was turned on, they sent a text message, and if no noise was heard in the complaining apartment, the device was turned off before moving on to the next one.

Suddenly, a low, continuous rumbling noise filled the room. The mother and daughter exclaimed, "That's the noise, that's the noise!" The culprit was located downstairs, and upon inspection, it turned out to be a dryer. The periodic rotation of a washing machine makes its noise identifiable, but the dryer's smooth rotation produced no periodic noise, making it hard to pinpoint the source. Interestingly, the structure-borne noise from the dryer affected the upstairs unit.

The culprit apprehended - the turbo dryer

The noise level from the dryer in the complaining apartment was not high, only 25dBA (NR23). The local authority's night-time noise limit is typically 30dBA, with a night-time equipment noise limit of NR25. The dryer noise was below the Scotland's usual noise limits. The A-weighted noise level of the dryer was only 1dB higher than the roof fan's noise, but the resident found the dryer noise more disturbing. This was because the structure-borne noise from the dryer was concentrated at a low frequency of 100Hz, and the uncertainty about the noise source added to the psychological discomfort.

Noise measurement results

The noise measurement results showed that at the 100Hz one-third octave band, the dryer noise exceeded the NANR standard by 4dB.

The turbo dryer noise level against NANR criteria

Are the background noise levels the lower, the better?

Tsinghua University once conducted a study on the impact of extremely quiet environments on sleep quality. The conclusion was that in an anechoic chamber (0dB), compared to an ordinary residential room, the overall sleep quality of the subjects improved. While high environmental noise can affect sleep, in a real apartment building, if the environmental noise is too low, the sound insulation requirements for floors and walls become very high. Otherwise, some minor everyday noises from neighbours, which would otherwise be insignificant, will be amplified, making them more unbearable than broadband, informationless environmental noise.

Solution

I suggested the management company install Mason ND-A-Green rubber vibration isolators on the dryer. After installing the isolators, I revisited the apartment building. This time, the resident could no longer tell whether the dryer downstairs was on or off, and the measurement results were almost identical with the dryer on and off.

Rubber isolators fixed under the turbo dryer

The project was with CSP Acoustics