What a Flexible Mount Does and What It Cannot Do

Why this matters

Isolators get treated as a general-purpose cure for vibration, and they are not. They are a narrow-purpose device with one job, one operating condition, and three ways to defeat them, every one of which happens during a normal installation without anybody noticing. A shop that adds springs to a machine that is already shaking usually makes the complaint worse, then adds more springs. The value of understanding what a mount cannot do is that it stops you selling the wrong fix, and it tells you what to look at on a job where the isolators are already there and the customer still feels the machine two rooms away.

Before an isolator comes out from under anything

The machine's weight is on those mounts. Support the equipment on a lifting device or cribbing rated for the load before any mount is unloaded or removed, and never let piping, ductwork, conduit or a partially removed mount take that weight, because all four will let go without warning and none of them is a support. On a suspended unit, hangers are carrying the load in tension and the failure is instant and total, so the unit gets independently supported from below or rigged from above before a hanger is touched. Lock and tag the disconnect and confirm the rotor has stopped rather than slowed before working under a unit, per 29 CFR 1910.147 and its verification-of-isolation step, since a coasting wheel above your head is stored energy. Flexible conduit at a machine is a live electrical path even when the unit is off if it is fed from another source, so any work inside a connection box follows 29 CFR 1910.333(b)(2), with the conductors proven dead using the live-dead-live sequence in NFPA 70E-2021, 120.5.

The one thing a mount actually does

An isolator lowers the natural frequency of the machine-on-support system so that running speed sits far above it, which puts the machine in the region of the response curve where the support transmits less force to the building than a rigid connection would.

That is the entire mechanism. For a machine well above the mount's natural frequency, the fraction of force transmitted is roughly 1 divided by (ratio squared minus 1), where the ratio is running speed divided by the mount system's natural frequency. That expression is the undamped approximation and it holds only in the region above a ratio of about 1.41; below that it is not merely inaccurate, it has the wrong sign.

You estimate the mount system's natural frequency from its deflection: about 187.8 divided by the square root of the static deflection in inches, cycles per minute, which is derived for a mass on a linear spring and is a fair description of a machine on isolators. A mount deflecting 1.0 in sits near 188 cpm. Under a machine at 1,750 rpm that is a ratio of 9.3 and a transmitted fraction near 1.2 percent, which is what people mean when they say isolation works.

It cannot reduce the force inside the machine

The first and most consequential limit. The unbalance force in the rotor, the coupling reaction from a misalignment, the blade pulse against the housing: none of them change when you put the machine on springs. The bearings carry exactly what they carried before. A mount changes where the energy goes after the bearings, not how much of it exists.

This is why isolating a machine that is failing bearings does nothing for bearing life, and why a customer complaint of "it is louder since we put it on springs" is often literally true. A machine free to move sometimes moves more visibly and audibly at the unit while transmitting less into the structure. Two different measurements, two different answers, both correct.

If the goal is less force, the correction is on the rotor, the alignment or the drive. The companion articles on unbalance force and on telling imbalance from misalignment cover that side.

It cannot help below a ratio of about 1.41

Below the square root of 2, a flexible support transmits more force than a rigid one. The crossover point is exact and damping does not move it, which is unusually convenient for a rule of thumb: at a ratio of 1.41 the transmitted force equals the applied force, below it you are amplifying, above it you are isolating.

That is why design practice puts the mount system's natural frequency at a third of running speed or lower, giving a ratio of 3 or better and a transmitted fraction of about 12 percent. It is also why the slowest machine in a building is the hardest to isolate: a reciprocating compressor at 450 rpm needs a much softer mount to reach a ratio of 3 than a 3,500 rpm blower does, and soft mounts on a machine with large unbalanced forces bring their own stability and piping problems.

It cannot work if it is not actually deflecting

Here is the failure that happens most often and is spotted least often. A mount's rated deflection is the deflection at its rated load. Put a machine on mounts sized for twice the load it actually applies, and you get about half the deflection, a higher natural frequency, a lower ratio, and less isolation than the submittal promised.

Worse, the load is almost never distributed evenly. A machine whose center of gravity sits toward the motor end loads two mounts far harder than the other two. Suppose the heavy pair carries 35 percent of the total weight each and the light pair 15 percent each, and all four are identical mounts selected by dividing total weight by four (25 percent each). The heavy pair is loaded 40 percent over its selection and the light pair 40 percent under, so the four corners now sit at four different natural frequencies, the machine rocks rather than translating, and the two lightly loaded corners do most of the transmitting. Mounts get selected per corner from the actual load distribution, which the manufacturer publishes, not by dividing the total by the number of feet.

One material note: steel coil springs are close to linear across their working range, so proportional reasoning about deflection holds. Elastomer mounts stiffen as they compress and soften with age and temperature, so an under-loaded elastomer deflects even less than proportion predicts, and the natural frequency of a ten-year-old elastomer mount is not the number on the original submittal.

It cannot survive a rigid path around it

An isolator only isolates the path it is in. Every other connection between the machine and the structure is a parallel path, and because those paths are stiff, they carry most of the force even when they look incidental:

• Hard-piped connections without a flexible connector, or a flexible connector installed under tension so it acts rigid
• Electrical conduit run straight and tight from the unit to the structure rather than with a deliberate slack loop
• A condensate drain hard-piped and clamped to the deck
• Duct connected without a flexible collar, or with the collar's fabric pulled taut
• Seismic snubbers or restraint bolts torqued down against the machine instead of set with their design clearance
• Grout, debris, a dropped shim or a wedge left under a rail during installation

The last one is worth its own sentence because it is so ordinary. A single hard shim left under one corner short-circuits four correctly selected isolators, and it is invisible from standing height.

Working an actual selection through

A machine on mounts rated for 1.0 in of deflection at rated load, but loaded to only 40 percent of that rating, so its actual deflection is about 0.4 in. Natural frequency: 187.8 divided by the square root of 0.4 = 297 cpm, against the 188 cpm the submittal assumed.

Put that machine at 1,750 rpm. Ratio with the design mount: 9.3, transmitted about 1.2 percent. Ratio as actually installed: 1,750 / 297 = 5.9, transmitted 1 / (5.9 x 5.9 - 1) = about 2.9 percent. The installation error more than doubled the transmitted force, and nobody will ever notice, because both numbers are small and the machine is quiet. On a fast machine, under-loading a mount is a paper problem.

Now put the identical mounts under a machine at 450 rpm. Ratio as designed: 450 / 188 = 2.40, transmitted 1 / (5.76 - 1) = about 21 percent. Ratio as installed: 450 / 297 = 1.52, transmitted 1 / (2.31 - 1) = about 77 percent. The same installation error took a marginal but functioning isolation job and made it almost useless.

And push it one step further. Suppose the mounts are stiffer still, or an elastomer has aged and hardened, so actual deflection is 0.15 in. Natural frequency: 187.8 divided by the square root of 0.15 = 485 cpm. Ratio: 450 / 485 = 0.93, which is below the crossover and nearly on the peak. Transmitted force at 5 percent damping works out above 600 percent. That machine now delivers roughly six times more force into the building than it would bolted solid to the floor, and every instinct in the room says to add more isolation.

That progression, from a harmless error at 1,750 rpm to a six-fold amplification at 450 rpm with the identical hardware, is the reason mount selection is a calculation rather than a catalog pick.

What a mount is genuinely good at

Given the ratio, the loading and the parallel paths are all right, isolation is excellent at what it does: keeping a machine's normal, unavoidable, correctly-tolerated residual forces out of an occupied structure. It is the correct answer for a rooftop unit over a conference room, a pump in a mechanical room sharing a wall with an office, and any equipment on a long-span floor. It is the wrong answer for a machine with a growing mechanical fault, for a machine below the crossover, and for any complaint measured at the machine rather than at the structure.

How to verify you got this right

Measure the deflection, do not read the submittal. On a spring mount, the free height and the installed height are both physically measurable and their difference is the deflection that sets everything above. Check all four corners, because unequal deflection tells you the load distribution is not what the selection assumed, and that is diagnostic in itself.

Then walk the machine's perimeter and find every path to the structure that is not a mount. Pipe, duct, conduit, drain, restraint. Each one should have visible slack, a real flexible element, or a design gap. If you cannot rock the machine slightly by hand with the disconnect locked and tagged and the rotor confirmed stopped, something rigid is holding it, and that something is carrying the force the springs were bought to stop.

References

• 29 CFR 1910.147 for isolation, lock and tag, and stopped-rotor verification before working under or unloading a mounted machine
• 29 CFR 1910.333(b)(2) and NFPA 70E-2021, 120.5 for the electrical fork when a connection box or flexible conduit is opened
• Isolator manufacturer documentation for rated load, rated deflection, per-corner selection data and elastomer temperature and aging limits
• See related: Why a Machine Is Worse at One Speed Than at a Higher One; How a Base and a Foundation Change the Machine on Them; How a Small Imbalance Becomes a Large Force