Apr 22, 2003

Intermittent Seal Leak in LPG Pump

Case Study

In a process plant industry, the reliability and effective
availability of equipment are the prime requirements to having high
productivity and for rendering faithful service to the consumers. The
reliability of equipment must be ensured and probably could be estimated as

*               By
proper equipment selection: 60 percent

*               By
proper installation: 15 percent

*               By
proper operation and maintenance: 25 percent

The following case study describes a reliability problem
with a Liquified Petroleum Gas (LPG) handling pump in a field and the way it
was troubleshooted and fixed.


These pumps were installed and commissioned in March 1999.
The mechanical seal started to leak intermittently at the interval of 45-60
seconds. This problem was persisting even after increasing the
restrictive/regulating orifice (RO) size to 5 mm from 3 mm. (See Figure 1.)

Observations at the site included the following.

*               Seal
was leaking intermittently

*               Suction
pressure: 7.5 kg/cm2(g)

*               Discharge
pressure: 16 kg/cm2(g) (steady)

*               Motor
load: 55 amps (steady)

*               Vibrations:
6.2 mm/sec max

Some of the suspected probable causes of intermittent seal
leak included the following.

*               Hung
up rotary head assembly.

*               Distortion
in seal faces.

*               Axial
float in rotor.

*               Seal
chamber pressure below vapor pressure and liquid flashing into vapor.

After dismantling, the following observations were made.

*               Shining
marks observed on carbon face. Rotary face found good and intact.

*               Elastomers
found in good condition.

*               Both
bearings found good and intact. No axial float observed in the rotor.

*               Impeller
found intact.

*               Axial
movement of rotary unit found unrestrained and free of sticky deposits.

*               Seal
faces checked for flatness and found to be ok.

*               Pump
impeller was found to have a back wear ring, but no balancing hole was

Analysis <.h2>

*               The
rotary head assembly was found free on the sleeve. Probable cause #1 was ruled

*               Seal
faces flatness found within two bands. Probable cause #2 was ruled out.

*               No
axial float observed in the rotor. Probable cause #3 also was ruled out.

*               The
shining marks observed on carbon face revealed that there was a loss of
lubricating film on the mating faces, which could be due to the formation of
vapors in the seal chamber that were not getting out from the chamber due to
the too close clearance in the throat bush. This accumulation of vapors may be
due to the heat generated at mating faces and dead-ended sealing chamber. The
seal chamber bush clearance appeared to be insufficient to flush out the liquid
vaporization due to heat generated by the seal faces, especially due to the
fact that cooling water initially was not supplied.

The fact that the seal was
failing intermittently with the periodic opening of seal faces and release of
LPG into the atmosphere seems significant. This could be happening because the
accumulated vapors (due to phase change from liquid LPG to gas) gradually would
increase in volume filling out the seal chamber. Then, when the pressure would
build up sufficiently, the faces would open up, causing the vapor to be
released and the cycle to repeat.

Applied Solution and Recommendations

At first, changing to Plan 13
was considered. For LPG / Propane services, which have a narrow margin between
suction pressure and vapor pressure at operating temperature, seal flushing
Plan 13 was thought to take the excessive built-up vapors from the sealing
chamber back to suction. This plan would consist of flush line from the seal
chamber through flow regulating orifice (RO) to suction. However, it was
decided that this would not solve the problem of vaporization in the seal chamber,
because the seal box pressure then would be even lower than when using Plan 11,
and there would be even less margin between the box pressure and vapor
pressure. Thus, the Plan 13 idea was rejected.

The history of attempted
modifications includes the following steps.

Step 1. Since the LPG service
(the seal chamber pressure and heat generated by the rotating seal faces) is so
close to vapor pressure, it is important to dissipate the heat generated at
seal faces to avoid rapid vapor formation at the seal area. Initially, it was
assumed that this was due to vaporization inside the seal box. The orifice size
was increased in order to increase the seal box pressure. Unfortunately, this
did not solve the problem.

Step 2. The API 610 8th Edition
Cooling Water Plan 61 specifies the "tapped connection for purchasers use.
Typically used when the purchaser provides fluid (steam, gas, water, etc.) to
an auxiliary sealing device."

This initially was not connected
as the location did not have cooling water available. This was discussed with
the manufacturer, pointing to the fact that the LPG service may be sensitive to
heat generation in the seal chamber. The manufacturer felt, however, that the
pump may not require additional cooling, and so the provisions for the cooling
water availability were not made. With a problem persisting, this needed to be
addressed, and cooling water circulation through the sealing chamber jacket
then was provided by hooking up the inlet and outlet lines to distanced headers
of the neighboring unit without the need to dismantle the pump. Unfortunately,
this did not solve the problem either.

Step 3. A 5-mm hole was drilled
in the upper portion of the throat bushing. At the same time in the impeller,
four balancing holes also were added. It was discovered upon disassembly and
examination that they were missing. This modification worked, and the pump now
is running satisfactorily without any leaks.

The seal box pressure for this
type of a pump (after modification with back wear ring and balancing holes) is
a suction pressure of more than 35 percent of discharge pressure as per
manufacturer’s rule of thumb for the light hydrocarbons.

8.0 kg/cm2 + 0.35% ¥ 16
kg/cm2 = 13.6 kg/cm2 (g)

Interestingly, this almost is
equal to box pressure before modification (as used by the "90 percent
rule" calculation above). Theoretically, this modification does not change
the seal box pressure substantially. What changes, however, is the amount of
liquid from the discharge connection to the box and then through the opened-up
bushing. In other words, the restrictive factor was the bushing clearance and
not the orifice in the Plan 11 piping.

Hence, by making the hole in the
throat bush and adding balancing holes, the passage for circulation of fluid
carrying out frictional heat became less restrictive and solved the problem
without excessive downtime and cost. The light hydrocarbon service pumps with a
narrow margin between vapor pressure and suction pressure should be provided
with a 3-5 mm drilled hole in the top portion of the throat bushing to allow
the vapors to vent away from the seal chamber.

About the author

Sourav Kumar Chatterjee is the manager of rotary equipment for the HPCL Plant, Mumbai, India.