• No results found

to optimize compressor operations

In document Hydrocarbon Processing May 2015 (Page 48-50)

Degradation of labyrinth seals in centrifugal impellers can have a significant impact on not only compressor power con- sumption but also on plant production rate. Upgrading to rub- tolerant seals can be beneficial in not only making improve- ments to the present efficiency and capacity, but, even more importantly, maintaining new operating levels. With damaged labyrinth seals, the head and efficiency are reduced. The per- formance curve is shifted down and to the left, and the capacity of the compressor is reduced, thus limiting plant output.

As shown in FIG. 1, such damage can occur in an instant as

the compressor goes through the critical speed on startup. This then results in the remaining run of the compressor (two to four or more years) with increased clearances and resulting ef- ficiency losses.

With rub-tolerant seals installed, the shaft may deflect dur- ing a vibration excursion, such as passing through a critical speed, and contacting the seal. The seal teeth deflect and then return back to as-installed when the vibration level settles back to normal levels. Seal clearances remain at design clearances, thus maintaining compressor performance.

For a closed impeller with a labyrinth seal at the impeller eye, the leakage through the impeller eye seal is recirculated through the impeller. This leakage is recompressed, and the compressor efficiency is affected in proportion to the amount of flow being recycled. The same is true for the impeller shaft labyrinth and the balance piston labyrinth.

Example. Assume the compressor is designed with four impel- lers. The compressor main inlet pressure is 200 psia (13.8 bara), and the discharge pressure is 315 psia (21.7 bara). For simplic- ity, the pressure ratio for each impeller is the same:

Rpx = P2/P1

or:

Rp = (P2/P1)1/x

where:

Rp = Pressure ratio across each compressor stage.

X = Number of impellers P1 = Inlet pressure

P2 = Discharge pressure

For this example: Rp = (315/200)1/4

Rp = 1.12

The pressure ratio is approximately 1.12, so the pressure fol- lowing the first wheel is 224 psia (15.4 bara). Assume that the

FIG. 1. Damaged labyrinth seal.

Δ H HBase 1 1 3 3 Base (clean) Damaged labyrinth seals

Flow 2 Head Efficienc y 2

FIG. 2. Compressor performance with new and damaged labyrinth seals.

46MAY 2015 | HydrocarbonProcessing.com

Maintenance and Reliability

pressure development through the impeller is approximately 60% of the stage pressure rise (the rest of the pressure rise is in the dif- fuser, as shown in FIG. 5), thus, a pressure of 214.4 psig (14.8 bara)

is possible at the impeller tip and at the impeller eye seal area.

Design condition.FIG. 2 shows the operating conditions for a

clean compressor with the labyrinth seals per design specs. With damaged labyrinth seals, the head and efficiency are reduced. The performance curve shifts down and to the left. Result: The capacity of the compressor is reduced, thus limiting plant out- put. Rub-tolerant seals can be installed with clearances that are smaller than the original aluminum labyrinths, as shown in FIG. 3.

Both the head (pressure rise), efficiency (power) and compres- sor through flow will be improved over the original equipment.

The shaft may deflect during a vibration excursion, such as passing through a critical speed, and rub the seal. Since the seal material is flexible, the angled seal teeth deflect and then return back to normal when the vibration level settles back to normal levels, as shown in FIG. 3.

Efficiency issues. Leakage through the impeller eye seal, QL, is recirculated through the impeller. The compressor efficiency

is affected in proportion to the amount of flow leaking through the impeller eye labyrinth seal, QL, relative to the main gas flow,

Qi. The same is true for the impeller shaft labyrinth and the

balance piston labyrinth. For a closed centrifugal impeller, the efficiency loss for the increased eye labyrinth seal clearances is about one percentage point for each percent increase in Qi (as

a result of QL increasing and the inlet flow to the compressor

constant). For closed centrifugal impellers with labyrinth seals at the eye, the following rule of thumb may be used:

%Δη α – %ΔQi

where:

Qi = Impeller flow

Using a software program based on the Adolf Egli procedure (free download), the seal leakage for the impeller eye seal for the standard seal and proposed rub-tolerant seal can be calculated.

Clearance issues. From the original equipment manufacturer (OEM) manual, the radial seal clearance for the first-stage im- peller and the eye seal diameter can be found. This information and the gas analysis can be entered into the labyrinth seal leak- age program. The approximate value for the radial seal clear- ance for the new rub-tolerant seals is:

CLRT = CLbrg + DEye/8,000

where:

CLRT = Radial clearance for the new rub-tolerant seals, in.

CLbrg = Radial clearance for the compressor journal

bearings, in.

DEye = Impeller eye diameter at the eye seal, in.

For this example:

CLRT = 0.0025 in. + 14/8,000

CLRT = 0.004 in.

A rule of thumb used by one OEM is that approximately 60% of the pressure is developed in the impeller. The rest of the pres- sure rise for a stage is developed in the diffuser, as noted previ- ously. This is a crude rule of thumb, and it is not a precise calcu- lation. This value can vary depending on the stage design and operating conditions. Accurate values can be obtained from soft- ware prediction programs that model the compressor in detail. 30°

typical

Deflected shaft

G.E.B.

Centerline Shaft

FIG. 3. Rub-tolerant seal cross-sectional view.

Qi

QL

Cl

FIG. 4. Closed impeller with labyrinth seal at impeller eye.

P P Velocity Diffuser Impeller Inlet Pressure V V

FIG. 5. Velocity/pressure development through a centrifugal compressor stage.

Hydrocarbon Processing | MAY 201547

Maintenance and Reliability

If the compressor through flow is 2,500 lb/min, then the ef- ficiency improvement is: (48–11)/2,500 = > 1.5% if the impel- ler eye seals are changed out to rub-tolerant seals. Note: This is relative to aluminum seals in the as-new condition. Often, alu- minum seals may be worn out, and they may be more than two or three times the normal spec clearances.

More improvements. Changing out the impeller shaft seals and balance piston seals will result in even greater improvement. For an accurate assessment, use a software program that prop- erly models the compressor aerodynamically to confirm both the efficiency and power improvements and the capacity enhance- ment. Use a software program to continuously monitor the com- pressor before and after the changeout of the labyrinth seals to demonstrate improvements in compressor performance.

FIG. 6 shows that the leakage rate for the first-stage impeller

with standard aluminum labyrinth seals is 48 lb/min, as calcu- lated based on the Egli procedure. Assume that the other three stages are similar. Recalculating the leakage rate for the first- stage impeller with rub-tolerant labyrinth seals gives 11 lb/min. If you assume that the aluminum labyrinth seals will eventually open to two times the normal clearance (leakage = 71 lb/min), then the efficiency improvement for the changeout to rub-tol- erant seals will be: (71–11)/2,500 = > 2.5%. This calculation is based on the changeout of all of the impeller eye seals. Chang- ing out the impeller shaft seals and balance piston seals will re- sult in even greater efficiency improvement, as shown in FIG. 7.

Monitoring the compressor before and after changeout of the labyrinth seals can demonstrate the improvement in com- pressor performance (FIGS. 7 and 8). With the changeout, a 5%

FIG. 6. The calculated leakage rate for the first-stage impeller with standard aluminum labyrinth seals is 48 lb/min. The leakage rate for the first-stage impeller with rub-tolerant seals (0.004 in. clearance) gives 11 lb/min.

BARTEC ORB 4724 South Christiana Chicago, IL 60632/USA Phone: + (1) 773 927-8600 [email protected] www.bartec-orb.com

In document Hydrocarbon Processing May 2015 (Page 48-50)