Seal Failures can Happen even on Idle
Stand-By Pumps
By
Sourav Kumar Chatterjee
Manager
Rotary Equipment
Many
different factors contribute towards the failure of a pump. Some factors that
did not seem to have contributed to earlier failures were not examined or
considered and may become the cause of failure under other situations. Failures do not occur suddenly. Usually there
are many symptoms, which signal a failure situation and are generally termed as
potential failure modes. The ignorance of such symptoms almost always leads to
failure.
Below are
two case studies pertaining to mechanical seal systems, a vital component of
pumps operating in running process plant.
Case A: Even a standby pump can have a seal failure, and a hazard
associated with it
We have come
across many instances of seal failure or hazard due to seal failure of running
pumps, especially in light hydrocarbon or high temperature hydrocarbon
services, but the fire hazard due to seal failure of idle pump is a rare event
and unpredictable.
Pump type and operating parameters: single-stage centrifugal back pullout
design, made by Thomson-Byrton. Flanges: 300# 4”
suction, 3” discharge. Impeller has balancing holes.
Service: Jute Blending (JPB), 2900 RPM, 310 0C, 89.4 m3/hr flow,
90.8 m head, 0.67 specific gravity. Suction pressure 1.5 Kg/cm2
(gage units), V.P. @ P.T. (vapour pressure at pumping temperature) is 1.0 Kg/cm2
. Minimum flow (MCSF) is specified by manufacturer as 22.7 m3/hr
Mechanical parameters: bearings 6311C3/7310; bellows seal with
flush plan 02 and 62; cooling plan G. Material of major parts: 410ss, 316ss,
CS. Lubrication type: oil lubrication with Turbinol-68
This pump did
not have on-line monitoring, and thus no alarm or trip signals available.

Mating Ring Seal Ring Bellows Sleeve
Purge bush![]()
![]()
![]()


Description
of the initial problem
Two pumps are
connected to the distillation tower. They are referred to as RCO (Reduced Crude
Oil) pump and a JBO (the one described here) pump. Due to process yield
requirements, the JBO pump operates at 17 m3/hr, which is below the
minimum flow requirement of 22 m3/hr. In case the tank level drops
below a certain value, the JBO draw-off control valve begins to close to raise
the fluid level which is critical to RCO pump. Unfortunately, this causes
problem for the JBO pump, which goes to cavitation leading to subsequent
component failures under cavitation.
Atmospheric JBO Pump Level Switch JBO Draw off RCO Pump
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()

![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()
![]()

The JBO pump is
fitted with single stationery bellow seal having face combination carbon vs.
silicon carbide and initially the flushing plan was 02/62. The seal used to fail
(“smoked”) very often indicating inadequate seal face cooling and twice seal
failed under situation of starvation. This became highly alarming as the seal
failure caused fire due to flashing of hot product in to atmosphere.
To curb the
recurrence of such incident two immediate measures were taken:
Installed
recirculation line
Provided
external flushing Plan 32 with TPA media fluid (V.P 1.3 kg/cm2
(gage) at 200 0C) at actual pressure of 4.5 kg/cm2 at 150 0C was provided. Seal box
pressure was 3.2 kg/cm2

At first, this
modification of system appeared to have solved the problem, but
again there was a seal blow-off event after three months.
Surprisingly, this
time it occurred to hot standby pump. Hot stand by means primed idle pump with
discharge valve partially open condition and the fluid recycling through warm
up line by thermal convection, to keep the pump ready for start in case failure
of running one.
On
investigation it was noticed that external flush to seal box of stand by pump
was stopped earlier to facilitate adequate flush to running pump seal following
a pressure drop in external flush media system. The seal blown off while the
operator re-opened the seal flush line after two days to put the subject pump
back in service as per routine changeover schedule.
Analysis
During
suspension of external flush TPA flow, the line content got cooled off due to
stagnation within the bare pipeline. The seal box pressure of hot stand by pump
was nearly equal to suction pressure, i.e. 1.5 kg/cm2 (gage).
Subsequently while recommissioning the external flush flow again to the seal
box containing service liquid at above 300 0C , the cold TPA has a
fast temperature rise, with a problem at a pressure much lower than its
corresponding vapour pressure (V.P of TPA at 300 0C is 7 kg/cm2
gage).
This had
resulted rapid vaporization at seal box and on seal faces as well causing
opening of seal faces due to sudden pressure rise and the hot JBO flashed in to
atmosphere causing minor fire.
Action
Taken
It was decided
and advised to operation crews that the external flush flow must be kept ON
even the pump is standby. In case an emergency requires to stop the external
flush flow, the pump must be isolated from the system closing its suction,
discharge, warm-up lines etc.
The external
flush media source changed to a dedicated system having separate pumps and
source for flush. While recommissioning, the external flush to be established
before opening the suction valve for priming.
Case 2: Seal Failure of RCO Pump
Service – Reduced Crude
Oil
Unit - Fuels Refinery
Expansion
Make- M/S Arai JAPAN
Year of Commission - 1999
Type - Two stage. Reverse
suction, centerline-mounted between-bearing centrifugal pump.
Process
Parameters:
Flow: 127 m3/hr
Sp.Gr: 0.7
Process Temperature:
352 Deg C
V.P at P.T: 1.5 kg/cm2
gage
Minimum Flow: 4 m3/hr
Suction Pressure: 1.8
kg/cm2 gage
Discharge Pressure:
29.5kg/cm2 gage
NPSHa: 2 meters
NPSHr: 1.8 meters
Mechanical
Parameters:
Radial Bearing- 6311C3
(DE)
Thrust Bearing -
6309C3 (double row) (NDE)
Shaft sealing –
Mechanical seal rotary bellows, Carbon/SC, cartridge type
Flushing Plan –32
(external seal oil)
Cooling Plan –G
Stuffing box Pressure:
drive end 1.8 kg/cm2 gage, opposite end 13.0 kg/cm2 gage
Materials of
construction (MOC) of major parts: shaft SS410, impeller CF8M, wear
parts-SS316, casing and bearing housing- cast steel.
Previous failure
history (before introduction of
plan-32):
The dedicated external
seal oil system for the RCO pumps
was commissioned in
January, 2003. Initially this pump
was having flushing
plan 22/62 (flushing from pump
discharge through
strainer, cooler and orifice to seal) with
steam quenching.
MTBF of seal prior to
introduction of Plan 3 was 8 months.
The initial flush plan
22 was not effective due to plugging up of cooler and associated piping,
especially while pump used to be standby leading to non-availability of pump as
hot standby. Subsequent to this plan 02/62 was also tried but it did not
improve the situation significantly: seal continued to “smoke” and covered with coke formation. Ultimately it was decided to provide external
flush (virgin gas oil) through a dedicated seal oil system to avoid smoking of seal
and adequate seal face cooling, to prevent coke formation and seal leakage.
The
external flush was introduced through orifice of size calculated based on the
differential pressure between the flush media pressure (16 kg/cm2 gage) and
individual seal box pressure with due consideration to required flow (5 to 6
lit/min).
Few days
after commissioning the system the seal leak reported again, although not as a
fire hazard due to presence of the external flush as barrier fluid. It was
observed the bellows got punctured and uneven wear track on seal faces. The
runout of shaft portion exposed for seal and the gland plate pilot surface
trueness were checked with no noticeable deviations observed. Following this
seal assembled with new bellows and the pump was put back to service.
The seal leakage started within few hours of operation of the pump.
Observations after dismantling the seal:
1 - Bellows found
ruptured.
2 -The seal mating
face wear pattern was not concentric.
3 - Both the faces
were good.
4 - Thrust bearing was
good and intact.
5 - Secondary packing was
good and intact.
Suspected
causes:
Shaft
deflection due to piping stress, unbalance rotor, hydraulic unbalance
Out-of-perpendicularity
of shaft with respect to stuffing box face
Low
pressure limit of bellows.
Analysis:
As all process parameters and rotodynamics
symptoms were normal and both the bearings were good and intact, the
possibility of shaft deflection due to hydraulics / rotor dynamics were ruled
out.
As a standard procedure, stuffing box, casing & bearing housing bore
and faces are machined & maintained concentric and perpendicular with
respect to shaft by the pump manufacture and after machining; all these parts
are assembled with proper trueness with help of piloting steps or dowel pin
each to avoid the problem of eccentricity and out of perpendicularity. During
subsequent dismantling/assembly by user at time of repair/overhauling the
factory set dowels/piloting steps act as reference guides to ensure
concentric/perpendicular assembly of stationery parts with respect to rotor
assembly.
The maintenance practice does not specify the checking of
concentricity/perpendicularity, as a regular activity which is only done in
case any symptom of eccentricity is noticed. It is to be noted that same seal
never displayed this type of failure mode prior to introduction of high
pressure external seal flush.
Parameters
changed after introduction of plan 32 were:
1. Stuffing box
pressure at NDE increased form 14 bar to 17 bar.
As per the seal vendor spec, the max pressure limit for bellows is 15 bar,
which is the border line case and could likely be the cause of bellow failure.

Seal: 2.750” Single PBR Cartridge Seal Make: FlowServe Sanmar
Applicable
pressure range
![]()
![]()

2. The pressure limit of PBR seal (4mils bellows AM350 ply thickness) as
per the PV graph is 15 bar. The bellows did not develop this type of failure
when the flushing Plans were 02/62 and the stuffing box pressures were around
14 bar. When the API Plan was changed to 32, the stuffing box pressures
increased to 17-18 bar. As the Plan 32 pressure is more than the seal pressure
allowable limit, it could cause over pressurization of the seal and eventually
the seal will fail.
Immediate actions: Recommended to change the bellows assembly material to Inconel with a bellows thickness of 8 mils to withstand higher pressure. The bellows mounting changed to stationery type from rotary type. Thus
the rotodynamics (due to torsion) loads and fatigue
loads even in presence of any concentricity problem, will not affect the bellow
stability.
For comments to the author and feedback on this
Field Case, please e-mail to: