Preview only show first 10 pages with watermark. For full document please download

Using Pv Curves To Diagnose A Recip Valve Problem

   EMBED


Share

Transcript

CASE HISTORY UCASE S I HISTORY NG PV CURVES TO D I A G N O S E A R E C I P VA LV E P R O B L E M Condition Monitoring Prevents Two Days of 20% Reduction in Total Refinery Output Greg Davis Diagnostic Specialist Bently Nevada email: [email protected] Recips Critical to Plant Production Capacity THE LOSS OF ONE RECIPROCATING COMPRESSOR RESULTS IN A 20% LOSS IN PRODUCTION 48 ORBIT 1Q04 Three reciprocating compressors provide compressed hydrogen for the catalytic cracker in a major US refinery. In order for the unit to run at full output, all three compressors must run simultaneously. The loss of one machine results in a 20% loss in production. The three compressors (six-throw, horizontal, balanced-opposed) operate in parallel. Capacity control is provided through suction valve, plug-type unloaders on the head end of each cylinder and a variable pocket clearance unloader on the head end of the first stage. Each compressor has two services – low purity (LP) and high purity (HP) – and each service has three stages of compression. Pressure and Inertia Loads, Not Unbalance, Principal Stress in Recips Unlike rotating equipment, where large stresses to rotating parts are often caused by rotating unbalance, misalignment, rubs, and other malfunctions, large stresses on reciprocating compressors are caused by a CASE HISTORY Monitoring System As part of the plant’s reciprocating compressor management program, various machine parameters are monitored and trended with online instrumentation. A 3500 Series monitoring system provides alarms on high vibration and transmits all the parameters to System 1™ software for storage and display. Parameters monitored include cylinder pressure (displayed as pressure vs. volume, or PV), cylinder valve temperature, main bearing temperature, suction and discharge temperatures, frame velocity, crosshead acceleration, piston rod displacement vibration, and rod position. Important parameters derived from these measurements include rod load (compression and tension), degrees of reversal, peak cylinder pressure, discharge pressure, minimum cylinder pressure, suction pressure, and compression ratio. 4 ACCELERATION: 0.2g pk/div combination of the pressures acting on the piston and the inertial loads from starting and stopping the reciprocating masses. The combined load on the machine is the sum of the gas pressure and the inertial load. These forces act on the stationary parts of the compressor as well as the reciprocating parts. Experience has shown that the crosshead pin and bushing are the weakest links in this chain. The crosshead pin and bushing transmit the piston rod load from the sliding elements to the connecting rod. Since there is relative motion between the crosshead pin and the crosshead bushing, they require lubrication. 3 VIBRATION LEVEL BEFORE REPAIRS 2 ALERT LEVEL EXCEEDED 1 0 11:49 01 MAY 2002 11:49 02 MAY 2002 11:49 03 MAY 2002 TIME: 8 hours/div LP stage 3 crosshead acceleration vibration before repair. | FIG. 1 The Event At midnight on May 1, 2002, the crosshead vibration (overall acceleration) on the low-purity, third-stage cylinder changed from approximately 1.3 g’s to over 2.5 g’s (Figure 1). The 3500 monitor activated the alert relay, which caught the attention of the Rotating Equipment Engineering group. THE 3500 MONITOR ACTIVATED THE ALERT RELAY, WHICH CAUGHT THE ATTENTION OF THE ROTATING EQUIPMENT ENGINEERING GROUP 1Q04 ORBIT 49 CASE HISTORY Load curves for crank end, before repairs. | FIG. 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . INERTIAL LOAD . . . 12,000 . . Tension TDC . . . . . . . . . . . . . . . . GAS LOAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Compression . . . . . . . . . . . . . . . . . . . . 200 . . 100 . . 0 . . -92,000 COMBINED LOAD . . -50,000 55˚ REVERSAL . . STRESS: 5000 psi/div 0 300 CRANK ANGLE: 20 degrees/div ANALYSIS OF THE DATA SHOWED THAT THE RUNNING GEAR WAS STRESSED Load and Pressure Curves Key to Diagnosing Valve Failure Because the crosshead accelerometer showed such a dramatic increase in vibration, the engineers plotted the combined load curve for this throw using the online data in System 1™ (Figure 2). The combined load must reverse from compression to tension during crankshaft rotation for the crosshead pin and bushing to get proper lubrication. The closer the reversal is to 180°, the better the lubrication. Also, in normal operation, the magnitudes of the peak compression and tension values would be more nearly equal. 50 ORBIT 1Q04 In this case, the load curve showed only 55 degrees of reversal, and there was a large difference between the magnitudes of the peak tension and the peak compression,12,000 psi versus -92,000 psi. As the degrees of reversal decrease and the difference between tension and compression increases, the life of the crosshead pin bushing is reduced. Failure of the crosshead pin bushing can result in damage to the crosshead, crosshead pin, and connecting rod, so it is very important to identify and correct the problems as soon as possible. Because the rod load curve showed such a low value of degrees of CASE HISTORY PV curves for crank end and head end, before repairs. | FIG. 3 TDC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . HEAD END ACTUAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CRANK END ACTUAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 . . . 20 CRANK END THEORETICAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0 . 1500 . PRESSURE: 50 psig/div ROUNDED CORNER HEAD END THEORETICAL 2000 60 80 100 DISPLACED VOLUME: 5 %/div reversal, the cylinder pressure curves in PV format were then plotted (Figure 3). The head end (HE) curve (orange), with sharp corners on the PV curve, indicates a cylinder in good condition. In contrast, the crank end (CE) curve (blue) shows a rounded heel near bottom dead center (BDC). As the piston nears BDC, the velocity of the piston slows and the velocity of the gas slows. In cases of severely leaking suction valves or pressure packing cases, the pressure developed by the piston cannot keep up with the leak, resulting in the rounded corner of the PV curve. The theoretical reference curves, green and red, are generated from the suction and discharge pressures measured on the actual PV curve. They represent the perfect compression process for that one crankshaft revolution, and usually closely follow the actual PV curves. The significant disparity between the crank end theoretical and actual curves supports the same diagnosis as the rounded corner on the actual curve. Suspecting either a leaky suction valve or leaky pressure packing case, engineering then examined the temperature trends from the cylinder valves and pressure packing cases. At approximately midnight on May 1st, the temperature trend SUSPECTING EITHER A LEAKY SUCTION VALVE OR LEAKY PRESSURE PACKING CASE, ENGINEERING THEN EXAMINED THE TEMPERATURE TRENDS 1Q04 ORBIT 51 CASE HISTORY Temperature for suction valves on LP stage 3. | FIG. 4 CRANK END SUCTION VALVE SUCTION GAS TEMPERATURE 50 0 02:07 02 MAY 2002 13:23 26 APR 2002 Damaged ring from concentric valve. | FIG. 5 13:23 06 MAY 2002 TIME: 24 hours/div 4 BEFORE REPAIR Crosshead acceleration vibration before and after repair. | FIG. 6 WHEN MAINTENANCE REPLACED THE DAMAGED CONCENTRIC VALVE, THEY DISCOVERED THAT ACCELERATION: 0.2 g pk/div TEMPERATURE: 5 deg F/div 100 3 AFTER REPAIR 2 1 ONE OF THE RINGS IN THE VALVE HAD FAILED 0 11:49 03 MAY 2002 11:49 04 MAY 2002 11:49 05 MAY 2002 TIME: 8 hours/div 52 ORBIT 1Q04 11:49 06 MAY 2002 CASE HISTORY TDC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . CRANK END ACTUAL AND THEORETICAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PRESSURE: 50 psig/div . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 . . . . . 40 . . . . 20 . . . . . . . . . . . . . . . . . . . . . . . . . HEAD END ACTUAL AND THEORETICAL . . . . . . . . . . 0 . . 1500 . . 2000 100 80 DISPLACED VOLUME: 5 %/div PV curves for crank end and head end, after repairs. Compare with Figure 3. | FIG. 7 for the crank end suction valve showed a sharp rise, indicating a malfunctioning valve (Figure 4). This timing corresponded with the high crosshead vibration noted in Figure 1. Analysis of the rod load, cylinder pressure data, and vibration data showed that, in this condition, the running gear (crosshead, connecting rod, and crankshaft) of the compressor was stressed. Since action needed to be taken to correct the valve problem before further damage occurred in the running gear, maintenance scheduled the machine for shutdown for repair. Damaged Ring Caused Valve Failure The outage occurred during an offpeak period and required approximately four hours of repair time. When maintenance replaced the damaged concentric valve, they discovered that one of the rings in the valve had failed (Figure 5). With the new valve installed, the vibration on the crosshead dropped below the level before the malfunction (Figure 6), and, as the piston moved towards the crank end, the pressure built as expected (Figure 7). A LARGE PART OF THE SAVINGS CAN BE ATTRIBUTED TO THE DECREASE IN UNSCHEDULED MAINTENANCE DOWNTIME 1Q04 ORBIT 53 CASE HISTORY Load curves for crank end, after repairs. Note increase in tension and degrees of reversal compared to Figure 2. | FIG. 8 . . . . . . . . . . . . . . . . . . . . . . . 0 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Compression . . . . . . . . . -50,000 85˚ REVERSAL . . STRESS: 5000 psi/div . . . GAS LOAD INERTIAL LOAD . . . . . . . COMBINED LOAD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28,000 . Tension TDC -93,000 0 100 200 300 CRANK ANGLE: 20 degrees/div The combined rod load curve (Figure 8) shows that the reversal improved from 55 degrees to 85 degrees and the magnitude of the tension increased. The improvement in reversal duration and better balance between compression and tension loads will lengthen the service life of the crosshead pin and crosshead pin bushing. System 1™ Returned Major Savings to Customer Through the cylinder pressure information provided online by System 1™, engineering was able to assess the stresses that the malfunction induced upon the reciprocating compressor and, based upon 54 ORBIT 1Q04 these stresses, decide when to shut down the machine so that further damage did not occur. This resulted in a shorter downtime, lower maintenance cost, and the prevention of a major failure. The leaking suction valve decreased the degrees of rod reversal, which in turn decreased the lubrication between the crosshead pin and its bushing. This situation decreases the service life of the crosshead bushing and possibly the crosshead pin. The diagnosis of the leaking suction valve allowed the maintenance personnel to fix the leaking valve before it caused a crosshead pin bushing failure. While it typically takes four hours to replace a valve, especially if the leaking valve can be narrowed down to suction or discharge valves on one end of the cylinder, it can take up to two days to replace a crosshead pin bushing. Because a 20% reduction in production results from having one machine down for maintenance, a large part of the savings can be attributed to the decrease in unscheduled maintenance downtime. Also, if the leaking valve had not been replaced and the crosshead bushing eventually failed, the valves might not have been checked for damage or replaced during the outage. This could have resulted in another premature failure of the crosshead pin bushing.