sample of OSA was determined by the change in weight of the
bubbler system after each pass.
The first series of six passes were performed at a N
2
flowrate of 250 ft
3
/hr. This rate was chosen as laboratory
calibration of the bubbler system showed that this rate of
N
2
flow produced an evaporation rate of OSA close to the
estimated minimum detectable evaporation rate, 40 g/min.
The intention with this series of tests was to see if our
estimate of the minimum OSA evaporation rate detectable
was correct.
The second series of six passes were performed at a N
2
flowrate of 500 ft
3
/hr. No lab calibrations at the rate were
performed previous to the in-field tests, but the evaporation
rate was expected to range from 60 - 80 g/min. The intention
with this series of tests was to demonstrate conclusively that
the instrument could detect plumes of evaporated OSA.
Table 3 summarises the results of the flight test, while
Figure 4 shows an image of one of the detected plumes of
evaporated OSA vapour. In the first series of tests, the average
measured evaporation rate of OSA from the evaporator system
was 30 g/min., close to the value achieved in lab calibrations
of the evaporator system. A plume of OSA vapours was
detected in three of five passes of the target (a 60% detection
rate). The pilot missed the target on one of the six passes. In
the second series of tests, the average measured evaporation
rate of OSA from the evaporator system was nearly double,
at 55 g/min. In this series of overpasses, OSA vapours were
detected in four of five passes of the target (an 80% detection
rate). On one of the passes, there was a weak signature. Finally,
the pilot missed the target on the second pass.
These tests showed that with an evaporation rate of
30 g/min. of OSA, the tested GFCR system was able to detect
the plumes roughly 60% of the time. With an evaporation rate
of 55 g/min. of OSA, the detection rate was higher at 80%.
With the lower end sensitivity threshold for this
technology demonstrated at about 30 g/min. of evaporated
OSA, the next step was to determine what size pipeline leak
this translates into. Due to the highly variable environment
that these leaks can occur under, the problem was
constrained by assuming that a small pool of the product
would be exposed to air and then the following model was
developed: a leak produces a pool of liquids on the surface
which is ‘fed’ by the leak, and both evaporates and ‘drains’
away.
Figure 5 shows the modelled steady-state evaporation
rates for OSA as a function of the leak rate (into the pool) and
wind speed (2.5 - 10 km/hr). Also shown on this figure (dashed
line) is the field measured instrument minimum sensitivity
to OSA of 0.035 lpm (or 30 g/min.). The model shows that
the evaporation of OSA from a 4 m
2
pool of leaked OSA
will be ‘detectable’ by this technology for low winds speeds
≤
5 km/hr and leak rates
≥
0.5 lpm (4.5 bpd). Also, lower leak
rates are ‘detectable’ if wind speeds are higher.
Inspection frequency determination
Modern pipeline risk analysis recognises two forms of
assumed risk from leaks: the probability of a leak occurring
and the impact that the leak will have once it has occurred.
Whereas pipeline integrity management procedures and
other leak prevention measures have an impact on reducing
the pipeline failure risk, it can never reduce the probability of
a leak to zero. Given that there is always a likelihood of a leak
occurring, a leak detection system is the first line of defense in
reducing its impact, mostly by limiting the size of the eventual
total spill. API Standard 1160
2
covers this in Section 10,
Mitigation Options: 10.3 Detecting and minimising unintended
pipeline releases. It states that, in the event of an unintended
Figure 3.
Field setup of Bubbler System.
Figure 4.
Example of imaged/detected plume.
Figure 5.
Flowmeter and scale readout.
28
World Pipelines
/
JANUARY 2015
