A pump-motor set on a cooling tower skid was vibrating at 4.5 mm/s RMS, up from 2.1 mm/s six months earlier. The vibration was highest at 1x running speed, and the pump bearing ran warm. The plant called in a laser alignment contractor. The alignment was out by 0.25 mm parallel and 0.15 mm angular — roughly 0.5 mm/m angularity. The contractor corrected it. The vibration dropped to 1.8 mm/s within an hour. The story is common. What is less common is the opposite version: the alignment looked perfect with a laser, and the coupling still failed, because the thermal growth moved the machines after the alignment was done.
Misalignment is the most diagnosed mechanical problem in rotating machinery and the most misdiagnosed. It shows up as 1x running speed vibration (when it is angular), as 2x (when it is parallel), and as axial vibration. It heats the coupling, the seals, and the bearings. But the same symptoms come from a bent shaft, a soft foot, a resonance, or an unbalanced coupling half. The laser shows the truth at the moment you measure it. The machines are not at that position when they run.
What misalignment actually does to bearings
A misaligned coupling forces the shaft to bend elastically at the coupling. The bending pushes the bearing loads off-axis. For a 0.25 mm parallel offset on a standard flexible coupling, the induced load on the pump bearings is small — maybe 5% of the radial load. The bearings do not fail from that load alone. They fail because the seal runs eccentrically and leaks, because the coupling flex element heats and hardens, and because the pump shaft deflection changes the impeller clearance. The bearing damage is the end of a chain, not the cause.
The exception is a rigid coupling. A rigid coupling transmits misalignment directly to the bearings. A 0.1 mm offset on a rigidly coupled motor-pump set puts a bending moment on both bearing sets. The moment is roughly F = E × I × δ / L², which for a 40 mm shaft over 400 mm between bearings with 0.1 mm offset is in the order of hundreds of newtons. On a pump bearing rated for a few kilonewtons, that is not trivial, and it is continuous. Rigid couplings demand near-perfect alignment, within 0.02 mm. Flexible couplings absorb the rest.
The thermal growth problem
The cooling tower pump was aligned cold, at 20°C. When it ran, the motor reached 60°C and grew axially by about 0.4 mm over its length (steel grows 11 μm per meter per degree Celsius; a 600 mm motor foot spacing at 40°C rise grows 0.26 mm). The pump casing, full of 35°C water, grew less. The relative growth tilted the motor. The cold alignment was perfect; the hot alignment was off by 0.3 mm. The vibration showed up after the machine warmed up. The plant fixed it by aligning hot — running the machine, letting it reach thermal equilibrium, then shooting the laser. The vibration never came back.
The rule that works: align the machine at operating temperature, or calculate the thermal growth and shim for it. The calculation: ΔL = α × L × ΔT. For a motor with 600 mm foot spacing, α = 11 μm/m·°C, ΔT = 40°C: ΔL = 11 × 0.6 × 40 = 264 μm = 0.26 mm. Shim the drive-end feet by 0.26 mm so that when the motor grows, it comes into line. This is a five-minute calculation that prevents a recurring vibration complaint.
The vibration signature that tells you it is alignment
| Condition | Dominant frequency | Direction |
|---|---|---|
| Angular misalignment | 1x running speed | Axial, high |
| Parallel offset | 2x running speed | Radial |
| Unbalance | 1x running speed | Radial, steady |
| Bent shaft | 1x running speed | Axial, at one bearing |
| Soft foot | 1x and 2x | Radial, on one foot |
The 1x axial component with a 2x radial component is the classic alignment pair. Unbalance is pure 1x radial with no axial. The distinction matters because unbalance is fixed by adding weight and alignment is fixed by moving the machine. Treating one as the other wastes a day. The cooling tower pump showed both 1x axial and 2x radial. That pointed to alignment before the laser was even set up.
The soft foot that mimics alignment
Before aligning, check for soft foot. A soft foot is a machine foot that does not sit flat on the base — a burr, a chip, or an uneven shim pack. When the hold-down bolt is torqued, the frame twists. The shaft alignment shifts. The laser says “misaligned,” you shim it, and the reading does not change because the twist is in the frame, not the feet. The check: loosen each hold-down bolt in turn, measure the shaft movement with a dial indicator. If the shaft moves more than 0.05 mm when a foot is loosened, that foot is soft. Fix the soft foot first, then align. Doing it in the wrong order produces a frustrating hour of “the laser keeps changing.”
Shim packs are the other common source. Stacking more than four shims makes the pack springy. The alignment reads fine cold, then shifts as the machine warms and the shims compress. Use one or two precision shims per foot, not a stack of five random washers.
Acceptance criteria that actually hold
For a standard flexible coupling (elastomeric jaw or gear), alignment within 0.1 mm parallel and 0.1 mm/m angular is fine. For a rigid coupling or a high-speed shaft, tighten to 0.02 mm and 0.02 mm/m. For a cardan or a gear coupling, the spec depends on the operating angle. The cooling tower pump, at 1450 RPM with an elastomeric coupling, needed 0.1 mm and got it.
The mistake to avoid: chasing the last 0.01 mm. Below 0.05 mm, the measurement uncertainty of the laser itself starts to dominate. A coupling that runs at 0.04 mm will run exactly like one at 0.09 mm. The time spent on the last 0.05 mm is better spent checking thermal growth and soft foot. That is where the real errors hide.
The cooling tower pump was fixed in an afternoon. The alignment contractor charged $450. The vibration dropped by 60%. The pump bearing temperature dropped 8°C. Six months later the vibration is still at 1.9 mm/s. The failure mode is not mysterious — it is just that alignment must be measured hot, soft foot must be checked first, and the signature (1x axial plus 2x radial) must be read before the laser comes out.
Misalignment shows as 1x axial and 2x radial, heats the bearings, and returns if you align cold. Check soft foot first, align at operating temperature or shim for growth, and stop chasing microns below 0.05. The cooling tower fix was a hot alignment and one shim change. The vibration has stayed down for six months.
