What a Vibration Isolating Connection Can and Cannot Do

Why this matters

An isolator is not a noise absorber and it is not soft-equals-better. It is a filter with a cut-off frequency, and on the wrong side of that cut-off it does not merely fail to help, it multiplies the force going into the building. A shop that installs springs because the tenant complained, without knowing the machine's forcing frequency and the isolator's own natural frequency, has about even odds of making the complaint worse and then having no way to explain why. The case below is one where the isolators were correct on the day they were installed and became amplifiers two years later without anybody touching them.

Before working on or under a mounted machine

A spring isolator under a load is a compressed spring, which is stored mechanical energy: take the machine's weight on rigging or jacks and set cribbing under it before releasing any isolator, and keep out of the plane a spring would travel in if it let go. Isolate and lock the machine under 29 CFR 1910.147 first.

Where the machine is driven by a variable frequency drive, the drive's internal bus holds a charge after the disconnect opens. De-energize, lock, tag and prove dead under 29 CFR 1910.333(b)(2) for general industry or 29 CFR 1926.417 for construction, wait the discharge time the drive's manual states, and then verify the bus is discharged with an instrument before touching anything, proving your instrument live-dead-live per NFPA 70E-2021, 120.5, which reaches you through your employer's electrical safety program in the edition it adopted.

Any measurement taken with the machine running is taken with all guards in place, as 29 CFR 1910.212 requires, from outside the rotating plane, with sleeves and lanyards secured.

The complaint

A rooftop air handler over an office suite. Two years after installation, occupants on the floor below started reporting a low rumble, worst in the mornings and worst on mild days. The unit had spring isolators under it from day one and nobody had reported anything for the first two seasons.

The obvious candidates got checked and cleared in the first visit. Fan wheel balance was within the shop's acceptance limit and the bearing signature was clean. Curb fasteners were tight. The unit was level. Nothing had been added to the roof. The complaint was not a squeal or a whine, which would point at bearings or belts, it was a low rumble, which points at something happening at or near the running speed itself.

The number nobody had

The visit that solved it started with a question nobody on the earlier visits had asked: what speed is the fan actually running when they complain?

The site had been retrofitted with a variable frequency drive fourteen months earlier, as part of an energy project. Design fan speed had always been about 900 rpm. The drive now turned the fan down to as low as about 300 rpm on mild days and mornings, which is exactly when the complaint appeared.

That is the whole diagnosis, and the rest is arithmetic.

The arithmetic

An isolated machine behaves, to a very good first approximation, as a mass on a spring. Its natural frequency in cycles per second is about 3.13 divided by the square root of the isolator's static deflection in inches. That expression assumes an undamped single-degree-of-freedom system, the deflection measured under the actual supported load, and a supporting structure much stiffer than the isolator. Hold on to that last condition; it comes back.

The isolators on this unit had 1 inch of static deflection, so the system's natural frequency is 3.13 divided by the square root of 1, which is 3.13 Hz.

For a mass on a spring, the fraction of the machine's force that reaches the structure depends only on the ratio of the forcing frequency to that natural frequency. Undamped, that fraction is 1 divided by the absolute value of the ratio squared minus 1. Three things fall straight out of it:

• The fraction is less than 1, meaning the isolator is actually isolating, only when the ratio is greater than the square root of 2, about 1.41.
• At a ratio of exactly 1 the expression blows up. That is resonance, and only damping keeps it finite.
• Below a ratio of 1.41 the fraction is greater than 1: the isolator transmits more force than a rigid mount would.

At the original design speed. 900 rpm is 15 Hz. The ratio is 15 divided by 3.13, which is 4.8. Transmission is 1 over 4.8 squared minus 1, which is 1 over about 22, or about 0.045. Roughly 95 percent of the force is stopped. The isolators were doing exactly what they were bought to do, which is why the first two seasons were quiet.

At turndown. 300 rpm is 5 Hz. The ratio is 5 divided by 3.13, which is about 1.60. Transmission is 1 over 1.60 squared minus 1, which is 1 over about 1.55, or about 0.64. Only about 36 percent of the force is stopped, so nearly two thirds of it now reaches the deck. Nothing broke. The machine simply walked down toward the cut-off.

Where it stops isolating at all. Break-even is a ratio of 1.41, which is 4.43 Hz, or about 266 rpm. Below that speed the isolators amplify, and the drive's low limit was set at 300 rpm, so on mild mornings the unit was operating within a few percent of the point where its isolation goes away entirely.

And every start. Resonance sits at 3.13 Hz, which is 188 rpm. Every ramp up and every coast down passes straight through it. Occupants described the rumble as worst in the mornings, which is both the mildest load and the time the unit starts.

Mechanism and symptom point the same way in both directions here, which is the check worth doing on any claim like this: at high speed the ratio is large and transmission is small, so the complaint should be absent at design load, and it was. At low speed the ratio approaches 1.41 and transmission approaches 1, so the complaint should appear at turndown and worsen as the drive slows, and it did.

What the isolators could never have done

The second half of the visit was walking every other path from the machine into the building, because an isolator only controls the path it is in.

• Rigid conduit landed from the unit to a curb-mounted disconnect, which was a solid steel bridge across the isolators.
• The condensate drain, hard-piped and clamped to the curb.
• A duct connection with a flexible section that had been installed pulled taut, which makes it a stiff link rather than a soft one. A flexible connector only works slack.
• The curb seal itself, packed hard enough to bear.

Each of those is a flanking path, and a flanking path in parallel with a good isolator sets the floor on what the isolator can achieve. Getting from 36 percent isolation to 95 percent means nothing if a piece of rigid conduit is carrying force around it.

Two more limits worth stating plainly, because they are what customers assume isolators do:

• Isolators do nothing about airborne noise. They address force transmitted through the mounts into structure. A machine that is loud in the air is still loud in the air.
• Isolators do not reduce the machine's own vibration, they let it move more. Softly mounted equipment displaces more at the machine, which puts more demand on every connection at the machine: flexible pipe connectors, electrical whips, duct connections. Fasteners at a softly mounted machine also see more relative motion, which is its own failure mode with its own article in this library.

And the condition set aside earlier: the natural frequency expression assumes the structure under the isolator is much stiffer than the isolator. A long-span roof deck has its own natural frequency, and where that lands near the machine's forcing frequency the deck amplifies whatever reaches it. Isolators cannot fix a structural resonance, and the tell is that the complaint is worse at one location on the floor below rather than directly under the unit.

The fix and the confirmation

They changed the isolators to a 3 inch static deflection type and corrected the flanking paths in the same visit. That costs you the attribution, since afterwards you cannot say which one did the work, and it was the right call anyway, because either alone would have left the tenant still complaining and the shop back on the roof.

New natural frequency: 3.13 divided by the square root of 3, which is about 1.81 Hz.

• At 900 rpm, 15 Hz, the ratio is about 8.3 and transmission is 1 over about 68, or roughly 1.5 percent. Better than before at design speed, though nobody was complaining there.
• At 300 rpm, 5 Hz, the ratio is about 2.77 and transmission is 1 over about 6.7, or about 15 percent. That is 85 percent isolation at the speed that had been running at 36 percent.
• Resonance moved to 1.81 Hz, which is about 108 rpm, below anything the drive commands in normal operation. The unit still passes through it on coast-down, briefly, which is unavoidable and acceptable.
• Break-even moved to about 2.56 Hz, which is about 153 rpm, so the drive's 300 rpm low limit now sits at nearly twice the frequency where isolation would disappear.

They also set the drive's minimum speed and, where the drive supported it, a skip band around the old resonance, so a commissioning change cannot walk the unit back down into it.

Confirmation was the same measurement in the same place: a reading taken on the structure below at the complaint location, at design speed and at minimum speed, compared against the readings taken before the change and against a reading with the unit off. The unit-off reading is the one people skip and it is the one that tells you how much of what you are measuring was never the unit.

References

• 29 CFR 1910.147 for isolating and controlling stored mechanical energy in a compressed isolator before releasing it; 29 CFR 1910.212 for guarding during running measurements.
• 29 CFR 1910.333(b)(2), general industry, and 29 CFR 1926.417, construction, for de-energizing and verifying, including the drive's stated bus discharge time; NFPA 70E-2021, 120.5, for live-dead-live, binding through the employer's electrical safety program in the edition adopted.
• Isolator manufacturer selection data, which owns the rated static deflection under load, the damping characteristics and the seismic or wind restraint options for a given mount.
• See related: Why Things Vibrate Loose; What a Support Has to Do Besides Hold Weight; The Vibration Signature Reference.