Bently Nevada systems at Ukrainian plants: what we find on site and how to plan a modernisation

Many Ukrainian plants protect their most critical machines with Bently Nevada systems: turbocompressor trains in nitrogen and ammonia production, steam turbines, gas compressor units and large pumps. A large part of this installed base is 20–30 years old. The systems still trip the machine when they should, but they are getting harder to maintain and give little diagnostic insight.

These notes come from our work on turbocompressor trains in nitrogen production and similar machines. They cover what we typically find, what causes problems, and what to check before planning an upgrade.

1. Several generations of equipment on one machine

It is common to find a 3300 rack, 3300 XL probes on some bearings, older 7200-series probes on others, and extension cables of unknown length. Each generation works on its own, but mixing them causes problems:

• Probe, cable and proximitor must match. System length (5 m or 9 m) and probe type must be consistent across the chain. A probe on the wrong cable length gives a wrong scale factor, so vibration is under- or over-reported without any alarm.
• Spare parts drift. After years of repairs, the spares cabinet often contains components that are not interchangeable with what is installed.

What to check: record the part numbers of the probe, cable and proximitor for every channel, and compare them with the rack configuration.

2. Gap voltages nobody has looked at for years

The DC gap voltage tells you where the probe sits relative to the shaft. On older installations we regularly see:

• gaps drifted towards the edge of the linear range after bearing replacement or probe reinstallation;
• axial position probes with zero references that no longer match the actual thrust bearing position;
• channels left in bypass after a past fault and never returned to service.

What to check: a gap voltage survey on all channels with the machine stopped and running, compared with installation records. For axial position, confirm the zero against the actual float of the thrust collar.

3. Calibration and periodic testing are often skipped

A protection system that has never been tested since commissioning is an assumption, not protection. The 3500 rack’s self-diagnostics confirm that the electronics and the transducer circuit are working. They do not confirm that the reading is correct. A probe with a drifted scale factor or a wrong cable length will report a plausible but wrong vibration value, with the OK status still green.

On older installations we often find:

• no records of transducer calibration since installation;
• relay outputs and trip circuits never functionally tested after DCS or wiring changes;
• channels whose alarm behaviour nobody has verified against the configured setpoints.

A periodic test programme usually covers three levels:

1. Transducers. Verify the scale factor and linearity of each proximity probe system (probe, extension cable, proximitor) with a calibration kit such as the TK-3, against the nominal sensitivity (7.87 V/mm, 200 mV/mil for 3300 XL). Check accelerometers and velocity sensors on a reference shaker where possible.
2. Monitor channels. Inject a known signal at the channel input and confirm the displayed value, Alert and Danger levels, and the 4–20 mA output to the DCS.
3. Protection function. Test the trip path end to end: rack relay logic, voting, wiring to the shutdown system and the final element, with the machine in a safe state. If the machine protection is part of a safety instrumented system, the test scope and interval follow the plant’s SIL proof-test requirements (IEC 61511).

How often: most plants tie testing to planned outages. Transducer and channel checks at every major outage, and a full trip test whenever the protection logic or wiring changes. The exact interval should be defined in the plant’s maintenance programme, not left to memory.

What to check: find the date of the last calibration and the last trip test for each machine. If there is no record, treat the system as unverified and plan the tests for the next outage.

4. Setpoints without a source

Alarm and danger setpoints are often set to values whose origin nobody remembers. They may have been copied from another machine, changed after a nuisance trip, or taken from the original OEM documentation for a machine that has since been rebuilt.

What to check: compare the current setpoints with OEM recommendations and API 670 guidance, and document the reason for each value. This is also a good moment to review trip logic and voting (e.g. dual voting on radial vibration), because nuisance trips and missed trips both come from here.

5. System 1 Classic on ageing servers

System 1 Classic is still in use at many plants. Typical issues:

• servers running old Windows versions that IT no longer wants on the network;
• OPC DA connections to the DCS that depend on fragile DCOM configuration;
• historical data that nobody exports before hardware fails.

System 1 Evolution is the current platform. A migration is easiest to plan together with a server replacement or a 3500 upgrade, not as an emergency after a disk failure.

What to check: back up the System 1 configuration and database now, list all OPC and Modbus connections to the DCS, and note which plots and alarms the diagnostic engineers actually use.

6. Upgrading 3300 to 3500: what can stay

A 3300 to 3500 upgrade does not always mean replacing everything:

• Probes and cabling in good condition can often stay, if the chain is consistent (see point 1).
• Rack replacement can usually be done within a planned outage if configuration and testing are prepared in advance.
• The DCS interface needs attention. Hardwired 4–20 mA outputs, relay contacts and communication protocols change, so update the I/O list and test the signals.

The main risk is not the hardware but missing information: unknown setpoints, undocumented wiring and unclear trip logic. A site survey before ordering equipment saves time during the outage.

7. Planning checklist

Before a modernisation project on a Ukrainian plant, it helps to prepare:

1. A machine list with train diagrams and measurement points.
2. Part numbers of the existing rack modules, probes, cables and proximitors.
3. The current configuration file (3300 or 3500) and setpoint list.
4. A gap voltage survey and records of recent alarms and trips.
5. Calibration and trip test records: when each transducer and protection channel was last verified.
6. The DCS interface: hardwired signals, Modbus or OPC tags.
7. Outage dates and how long the machine can be stopped.
8. Delivery logistics and import documentation for equipment from the EU.

With this information, an upgrade can be specified accurately, equipment can be ordered once, and site work fits into the planned outage.

FAQ

Can existing 3300 proximity probes be used with a 3500 system?
Often yes, if the probe, cable and proximitor form a consistent system and are in good condition. This has to be verified channel by channel.

How often should a Bently Nevada protection system be calibrated?
There is no single interval for every plant. A common practice is to verify transducers and monitor channels at each major outage, and to test the full trip path after any change to protection logic or wiring. If the system is part of a SIS, the proof-test interval comes from the SIL assessment.

Is System 1 Classic still worth keeping?
It continues to work, but new development is focused on System 1 Evolution. Plan the migration together with server or rack upgrades.

How is Bently Nevada equipment delivered to Ukrainian plants?
Equipment is usually supplied from the EU, and lead times depend on manufacturer stock. When planning a modernisation, place the equipment request early, before outage dates are fixed, so that delivery does not become the bottleneck of the project.

We work on Bently Nevada systems in Central and Eastern Europe, including Ukraine. If you are planning a modernisation, send us your machine list and we can help with the survey and specification.