Vibration Analysis and Condition Monitoring for Stone Machinery

Rotating equipment almost never fails without warning. Bearings roughen, shafts drift out of alignment, mounting bolts relax, and impellers lose material long before anything stops turning. All of that shows up as a change in how the machine shakes, and it shows up weeks or months before it shows up as a breakdown. The signal is there in every stone shop already. Almost nobody is collecting it.

Condition monitoring has a reputation for being an oil-refinery discipline that needs a specialist and a five-figure analyzer. That reputation is out of date. A maintenance tech with a handheld meter, a written route, and a spreadsheet can build a program that catches most of what matters in a fabrication shop. This guide covers baselines, measurement points, how to read a trend, and what vibration will never tell you.

Why Vibration Is the Earliest Signal

Every rotating machine has a vibration signature made of forces it generates by design and forces it generates by defect. Imbalance, misalignment, looseness, bearing damage, and hydraulic disturbance each add energy at characteristic frequencies. As a fault develops, the energy at its frequency grows. The machine keeps running and keeps producing, but the number you can measure at the bearing housing has already begun to climb.

The most important idea in the whole discipline is the baseline. A vibration reading with nothing to compare it against is close to meaningless, because acceptable levels differ enormously between a rigid-mounted spindle and a skid-mounted slurry pump. A reading taken on a healthy machine, at a known load and speed, from a marked spot, is the reference every later reading is judged against. Establish it while the equipment is in good order.

The standards work is consolidated in ISO 20816, which is the modern replacement for ISO 10816 and merges the older ISO 10816 approach for measurements on non-rotating parts with ISO 7919 for measurements on rotating shafts. It expresses severity using RMS vibration velocity in millimeters per second over a frequency range of 10 to 1000 hertz, which is the band where most general machinery faults live.

The standard sorts machines into four evaluation zones. Zone A covers newly commissioned or reconditioned machines. Zone B is acceptable for unrestricted long-term operation. Zone C is not suitable for long-term continuous operation and calls for remedial action. Zone D is dangerous, damage could occur, and immediate action is needed. Those four labels turn a raw number into a decision, which is what a maintenance program needs.

Commonly cited zone boundaries for the medium-machine class are 1.4 mm/s at the A/B line, 2.8 mm/s at B/C, and 7.1 mm/s at C/D, applying to equipment above 15 kW running between 120 and 30,000 revolutions per minute. Treat those as orientation, not as law. Zone boundaries vary by machine class and by support type, so a rigidly mounted machine and a flexibly supported one are not judged against the same lines.

This is why the trend usually beats the absolute value. A pump that has sat quietly at the same reading for two years is telling you something reassuring even if that reading is not impressive. A spindle whose reading has doubled in six weeks is telling you something urgent even if it still sits inside a comfortable zone. Direction and rate of change are the actionable information.

Building a Route a Tech Can Actually Run

A monitoring route is a fixed list of machines, in a fixed order, with fixed measurement points and a fixed frequency. The value of the fixed part cannot be overstated: comparability is the whole product. A route that takes a tech ninety minutes once a month and gets run every month beats an elaborate program that gets run twice and abandoned when a job ships late.

Where to Put the Sensor

Readings belong on the bearing housing, as close to the bearing as the casting allows, on solid metal rather than on a cover, a shroud, or a painted sheet panel. The path between the bearing and the sensor is what you are measuring through, and every gasket, guard, and air gap in that path attenuates the signal you came for. Bare, machined, load-zone metal is the target.

Take three directions where access permits: horizontal and vertical radial, plus axial. The comparison between them carries diagnostic information that no single reading holds. Radial energy dominating in one plane points toward imbalance or looseness. Strong axial energy usually points toward misalignment or a thrust bearing problem. Two directions consistently is better than three directions occasionally.

Mark the points physically. A paint dot, a stamped mark, or a permanently mounted mounting pad removes the largest source of scatter in hand-collected data, which is the tech placing the probe two inches from where it went last time. Magnetic mounts are fine on flat machined surfaces. Handheld stinger probes are acceptable for repeatable low-frequency work but demand consistent pressure and angle.

Saws, Spindles and Polishers

On a bridge saw, the blade spindle bearings are the priority: radial readings on the housing both sides of the blade if the design allows. Blade and flange condition shows up here as imbalance, and a bent flange or a warped blade is often visible in the trend before it is visible in the cut. Add the carriage drive gearbox and the travel motor as secondary points.

CNC spindles deserve the most attention and the tightest interval, because they are the most expensive rotating assembly in the shop and the most sensitive to bearing degradation. Read the front housing near the nose bearing and the rear housing, radial and axial, and always at the same programmed speed with no tool cutting. Toolholder imbalance and drawbar problems both surface in the same data set.

Polishing lines carry many small rotating assemblies, and the head bearings are the ones that fail messily. Read each head housing radially and log which head is which, because the pattern across heads is more informative than any one head's number. An uneven abrasive block set produces imbalance that looks like bearing wear until you compare it to the neighboring heads running the same grit.

Pumps, Compressors and Auxiliaries

Slurry pumps live hard and repay monitoring better than almost anything else in the shop. Read both bearing housings radially and the drive end axially. Impeller wear from abrasive solids, partial clogging, cavitation from a starved suction, and packing or seal trouble all move the numbers, and a pump that starts climbing usually gives several weeks of notice before it takes a production line down with it.

Air compressors are worth a monthly point on the motor bearing housings and on the pump or crankcase. Belt-driven units add drive misalignment and tension problems to the list of things the trend catches. A compressor failure stops every pneumatic clamp, every air-driven polisher, and every blow-off station at once, which makes it a higher-consequence machine than its purchase price suggests.

Do not skip the dust collection fan. Fan impellers accumulate material unevenly, and the resulting imbalance both destroys bearings and shakes the mounting structure. It is one of the few faults where a rising trend has an obvious, cheap remedy: clean the impeller. Catching that before the bearing is scrapped is the kind of small, unglamorous win that justifies the whole route.

Points are a starting framework. Manufacturer guidance and machine construction take precedence over any generic list.

Reading the Data Without a Full-Time Analyst

Start by setting alert levels from your own baseline rather than from a table. A common practical approach is to flag a machine for attention when its overall level has risen meaningfully above its own established healthy value, and to escalate when the rate of increase accelerates. That method works without any spectral analysis and without knowing which machine class the equipment technically belongs to.

Overall level tells you that something changed. It does not tell you what. Diagnosis needs spectral data and someone who can read it, which is the natural point to bring in an outside analyst. The economics are straightforward: your own route identifies the two or three machines worth paying an expert to look at, instead of paying an expert to survey the whole shop and find nothing.

Vibration misses entire categories of failure, and pretending otherwise builds false confidence. It will not catch electrical faults in a drive, a failing control board, coolant delivery problems, hydraulic leaks, valve wear, blade condition, or the slow loss of accuracy in a machine whose geometry has shifted. It also gives little useful warning on components that fail suddenly rather than progressively.

Temperature and current draw are the natural companions. An infrared reading at the same bearing housing takes seconds while the tech is already standing there, and a rising temperature trend alongside a rising vibration trend converts a maybe into a certainty. Amp readings on drive motors catch load-side problems that neither of the other two measurements sees clearly.

Set the response rules before the data arrives. Decide in writing what happens at an alert level, what happens at an alarm level, who has authority to pull a machine out of production, and what the escalation path looks like when a job is due. Those decisions are far easier to make on a quiet Tuesday than at four in the afternoon with a truck waiting at the dock.

Keeping the Program Alive

The main threat to condition monitoring is not technical, it is attrition. Programs die because the route slips a month, then a quarter, then permanently. Put the route on the maintenance schedule as a recurring work order with a named owner, give it a realistic duration, and review completion the same way you review any other planned maintenance task. Consistency generates the value; sophistication does not.

Use one instrument and, where possible, one tech. Different meters apply different filtering and different mounting behaves differently, and both introduce steps in the data that look like real changes. If a second tech has to run the route, walk it together the first few times so probe placement, pressure, and machine conditions transfer as habits rather than as written instructions.

Store the data somewhere durable and boring. A spreadsheet with machines down the side and dates across the top, backed up with the rest of the shop records, outperforms a proprietary system nobody renews. What you need years later is the ability to plot one point over time, and that requirement is modest enough that simple tools serve it better than complex ones.

Feed findings back into procurement and setup. If one pump position keeps climbing regardless of rebuilds, the problem is the installation, the piping, or the duty, not the pump. If spindle trends jump every time a particular toolholder family goes into service, that is a tooling decision the data just made for you. A monitoring program that never changes a purchasing habit is not being used fully.

Review the whole thing once a year. Add machines that have proven troublesome, drop points that have never moved, adjust intervals to match actual severity of service, and re-baseline anything that has been rebuilt or realigned. A rebuilt machine has a new signature, and comparing it against pre-rebuild history produces confusion instead of information.

Condition monitoring protects the machines; the consumables mounted on them still set the quality of the cut. Browse tooling and shop equipment at dynamicstonetools.com, and keep your finishing line supplied through the polishing pads collection so worn abrasives never become the reason a head runs rough.