Legacy Turbine Doctor, No. 2 in a series
By Luke Williams, PE, Consultant
www.geLegacyGasTurbineSupport.com
An MS7001B engine with Mark I Speedtronic™ controls failed to start reliably for about a year. Failure-to-start causes included loss of flame, trip at Complete Sequence to full-speed/no-load (FSNL), trip on 95% speed (14HS), TNH (high-pressure shaft speed) bog-down, and unknown. The troubleshooting effort was complicated by inconsistency in the different results recorded between starts. Because of the inconsistency of results on each start, the troubleshooting effort was directed at turbine hardware, the Mark I controls, and control wiring.
Using legacy troubleshooting techniques, the following hardware and systems were eliminated as the cause of the issue:
- Mark I trip logic, L4EC, and L4, which were forced to run for the start.
- Atomizing-air and compressor-bleed-valve function and potential leakage.
- The bleed valves were disabled and kept open during the start.
- Inlet-guide-vane actuation at 9.7 VCE (variable control electronics). The guide-vane actuator was disabled to keep the IGV closed. Note that VCE is an indicator of fuel-valve position, which determines fuel flow. This fuel valve is full-open at a VCE of 20 and closed at 0. Thus, a 9.7 VCE means the valve is about half-open.
- Fuel forwarding and fuel-pump-output pressure were monitored during the start.
- All of the Mark I cards were pulled and reinserted to eliminate pin corrosion.
- Control wiring from the Mark I to the fuel pump servo was replaced.
- Mark I cards from an operating sister unit were exchanged.
- The servo was not exchanged because the sister unit’s servo was the only reliable servo available.
The following results were observed when implementing the various troubleshooting procedures:
- Operation is very sensitive to VCE changes: Startup to speed control caused a flame-out.
- Raise/lower results in immediate loss-of-flame trip.
- One-third of start attempts achieved FSNL.
- Raise/lower at FSNL resulted in a slow deceleration to 3500 rpm and then recovery to 3600 rpm.
- During one deceleration, the fuel valve was gagged at 6.7 VCE; however, the pump stroke continued to decline resulting in loss of flame.
- As the number of start attempts increased, the number of trips at 95% speed decreased.
The majority of trips occurred at 95% speed; 14HS and 3 Complete Sequence trips were identified as the consistent problem. Flameout on unit deceleration had not caused a recent trip. Troubleshooting focused on the 14HS relays and wiring. There are eight 14HS relays and four 3 relays.
A set of “dummy” relays was made from three spare relays which were visually inspected and energized to confirm their electrical conformance to contact resistance and coil amperage. The dummies were used to replace the 14HS and 3 relays, one at a time, to check the effect of the relay coil and contacts on the decrease in turbine speed.
The Complete Sequence 3 relays were replaced with the dummy relay. It was noted that as a relay was energized, the speed would decrease from 3600 rpm and then recover to 3600 rpm. The turbine speed, servo voltage, RVDT, and VCE were monitored for changes during relay installation and raise/lower. The four 3 relays were replaced.
The 14HS, 14HS1, 14HS2, 14HX1, 2, 3, 4, and 5 relays were replaced one at a time, with the same drop in speed and recovery as with the 3 relays. The turbine speed, servo voltage, RVDT, and VCE were monitored for changes during relay installation and raise/lower. The turbine responded to raise/lower without a drop in speed.
Technicians checked and replaced 14HSY1 and 2 as well and restarted the unit. Response to raise/lower was better, 10 rpm compared to 30 rpm. The remaining seven 14HS relays were replaced.
At this point, reliable starts to Complete Sequence were being achieved; however, raise/lower was still a problem. The troubleshooting focus shifted from a problem in the relay logic causing a trip to a problem with energizing relays affecting the fuel-pump response.
An attempt to auto synchronize was not successful because turbine speed decreased and did not recover. There are a total of five load-selection 83LA relays (LA is the acronym for “load application.”) The 83LA2, 3, 4, and 83LAX relays were replaced, one at a time, starting with 83LA2. The turbine speed, servo voltage, RVDT, and VCE were monitored for changes during relay installation and raise/lower. The stability and raise/lower response checked. Results were stable.
Auto synch was attempted. Setpoint was reduced and at 3600 rpm the unit synchronized and a speed increase initiated. Load then was increased to 8 MW. Next, baseload was selected and the unit was ramped to 44 MW at 925F exhaust temperature. IGV activated at 9.9 VCE. A normal stop to shutdown followed.
A restart was initiated after a period on turning gear to demonstrate repeatability. After the unit achieved FSNL, auto synch was initiated and unit output increased to 6 MW, then to baseload (44 MW).
The unit was shut down and removed from forced-outage status.
Conclusions. As mentioned in the first paragraph, the subject unit had experienced starting problems for more than a year. The first troubleshooting effort determined that the fuel pump was not following the VCE signal and it was removed for overhaul, resulting in another three-month outage.
The major problem faced was the inconsistency of results on starts: Each start was different. The consistency of results improved as the number of starts increased.
The “breakthrough” was determining that the energizing of individual relays affect unit stability, usually a deceleration to loss of flame. The second conclusion was that the more relays energized, the more often the unit would trip on loss of flame. Troubleshooting shifted from looking for a relay or wiring trip to relays affecting the servo current.
Reliable starts were achieved by replacing individual relays in the 3, 14HS, and 83LA circuits. In the case of the 3 relays, four individual relays were involved. For the 14HS, there were eight and for the 82LA five. The initial troubleshooting effort focused on the 14HS relays because the most common trip was at 95% speed.
Root cause. The relays had become “sticky” during the year of inactivity in a salt-air environment and required an increase in amperage to pick them up. The increase in amperage affected the P28-Vdc power supply to the servo drive card and the milliamp output to the servo. The reduction in milliamps to the servo caused turbine deceleration and loss-of-flame trip. One of the best indications of this problem was that during one deceleration, VCE was gagged at 6.7 but the pump stroke continued to decline, resulting in loss of flame.
The P 28-Vdc power supply has both over-voltage and over-current “crowbar” protection. If amperage exceeds the limit, the current output is limited. The P28-Vdc voltage will remain at 28 Vdc because the current output is limited. In hindsight, the current output of the P28-Vdc power supply should have been monitored.
The unit started reliably after the mentioned adjustments were made. The corrective action recommended was to exercise the unit at least every two weeks. CCJ
Editor’s note: For more insight on troubleshooting early Speedtronic™ control systems, access the following CCJ articles:
