Problem in Velocity Sensor in Gas Turbine


Thread Starter


In our Captive power plant Frame-5 Machine Velocity sensor of Matrix 5485C installed in Turbine Compartment, Load Gear and generator compartment. In Turbine Compartment One vibration sometimes showing erratic value and cross the alarm point 12.7mm. At the same time rest reading are constant value at below 2 mm/sec. We have tighten the probe and not solved. Please suggest where we have to checked.

The first question to ask when a problem like this occurs is:

When did this problem start?

Was it after a maintenance outage, when the sensor(s) were removed? Was it after a maintenance outage, when the sensor(s) weren't removed but work took place in the area?

Has it been abnormally hot in the Load Tunnel where the #2 Turbine Bearing is located?

How long has this sensor been installed and operating? (Sensors do fail over time, and while this sensor should be rated for high temperature (I haven't looked at the part number on the manufacturer's website to be certain) it may not be, and even when sensors are rated for high temperature there's a limit. The #2 Brg. area is VERY hot and prone to many exhaust leaks, which may raise the temperature in the area of the vibration sensors more than other areas.

Was any work done recently in the sub-junction box where the sensor wires are terminated to the interconnecting wiring that sends the signal to the turbine control panel? Have you checked the terminal for tightness in every junction box from the one closest to the sensor all the way to the turbine control panel? (Turbines running at more at approximately 5100 RPM have a lot of vibration, and screws do come loose over time. The terminal screws don't have to be gorilla tight, but they do need to be secure. It is--or should be--a common practice to periodically go through every junction box during a maintenance outage to check the terminal screws for tightness, look for moisture, bad crimped-on terminal lugs, etc. It prevents a LOT of nuisance problems!

Also, crimp-on terminal lugs also can become loose over time, especially if they weren't properly crimped on to begin with (both the sensor end of the wiring and any interconnecting wiring). Many times a slight tug can cause the wires to come out of improperly crimped terminal lugs.

Many times the screw-on connector at the sensor is not properly tightened, and or is cross-threaded, when it is installed or re-installed. While the screw-on connector should be safety-wired to the sensor, that doesn't always mean it was fully and firmly screwed in place before the safety wire was attached.

The bolts that secure the sensor to the bearing cap should be installed with at least medium-strength thread-locking compound, and some are also safety-wired to prevent loosening (there are sometimes very small holes drilled in the cap-heads of the screws through which retaining wire can be threaded to secure them screws from loosening).

Some sensors were also installed on thick plates where were bolted to the bearing cap to provide a flat surface on which to mount the sensors. If a bolted plate is used, you should be sure the bolts are also securely torqued into place and that at least a medium strength thread-locking compound is used on the threads.

The shield drain wire of the sensor cable must be properly grounded in the sub-junction box closest to the sensor. And, sometimes the wires used in the sensor cable are high temperature woven material, and they can get "loose" and allow shorting of the two conductors and/or the shield/drain wire which can cause problems. Care should be taken to ensure the sensor cable terminations are done properly and cannot easily develop shorts or grounds.

The stainless steel braided sheath of the sensor cable can also get "dinged" or nicked when mechanics are working in the area, which can cause the conductors to be shorted or grounded intermittently. You should inspect the entire length of the sensor cable for dings or nicks, and replace it if you find an suspect area(s).

These are all the least-cost and easiest things to do--if all of this checks out and is okay, then you need to simply replace the sensor and see if that solves the problem. If it doesn't, then try replacing the cable. Troubleshooting is very often a process of elimination, and it's best to eliminate the simplest and easiest things first, because if you replace the sensor first and that's not the problem then you've wasted the time and cost of the sensor, and still have to go back and check all of these potential causes. You may want to replace the sensor cable first (it usually costs less than the sensor) if you check all the other things first, and if that doesn't solve the problem then replace the sensor. But, by eliminating all of the other possible causes first, many of which can be done in a single outage, you've increased the chances of finding and solving the problem without having to replace the cable and/or sensor.

We don't know what turbine control system is in use on the turbine, and it is possible that the input channel is failed or failing. Newer digital Speedtronic turbine control systems (Mark IV, Mark V, Mark VI, and Mark VIe) will all annunciate Diagnostic Alarms to indicate a possible problem with the sensor input channel--you haven't told us if there are any related Diagnostic Alarms....

If the turbine control system is a GE-design Speedtronic system, it's normal for redundant vibration sensors to be used to prevent tripping the turbine and to extend the possible time after one sensor fails before the turbine needs to be shut down to replace the failed sensor. A failed, or failing, sensor should be troubleshot and replaced/resolved as soon as possible, but having redundant sensors can help with keeping the unit running with a good degree of safety until a more convenient time to shut down to replace the sensor can be arranged. Of course, if the redundant sensor fails during this time the unit will trip, but if there was only a single sensor and it failed, the unit would trip anyway.

Hope this helps! Please write back to let us know what you find and how you resolve the problem. A lot of people read these posts, and many learn from them as well--especially when someone provides to the effect that they found the information useful.