- ERT Test Results
The OIWM was tested against a range of parameters at ERT as shown in
the following test matrix:
|
Oil
concentration
(20,
100, 500, 1000, 2000 ppm)
|
Pressure
(5, 10, 20 bar)
|
Flow
rate (1.5, 2.5, 4, 5m/s)
|
|
Oil
type (Flotta, Foinaven)
|
Oil
droplet size (20, 50, 100 microns)
|
Solids
(25, 100, 300 ppm)
|
|
Temperature
(15, 30, 45, 90 °C)
|
Chemical
additions (corrosion inhibitor, demulsifier and scale inhibitor)
|
Fouling
(effects of three months build-up studied)
|
|
Salinity
(0, 33, 66, 99 g/l)
|
Gas
content (0.05, 0.1, 0.2% gas)
|
Noise
(gate valve noise studies)
|
|
Table
1: Test matrix used in ERT test programme
|
The following table provides a summary of the results from the testing:
|
|
12
reproducibility runs were made during the test period. The PA
signal increases in a linear response between 20 and 1000 ppm
and this is unaffected by other process parameters (see Figure
6). The 2000 ppm response fell below the straight line and is
due to agglomeration of droplets - in a field installation a shear
mixer would be mounted upstream of the OIWM.
|
|
|
As
the process temperature increased, the PA response also increased
at a given oil concentration. The curves closely matched those
achieved under static conditions in the laboratory. The temperature
sensor in the head will enable on-line correction of the measured
PA response.
|
|
|
The
PA response increased with salinity for a given oil concentration.
The salinity of the process water was related to the signal time
delay under dynamic and static conditions - this enables accurate
calibration of the OIWM to account for the effect of salinity
on the ppm measurement.
|
|
|
Very
small increases in PA response were registered in the 5-20 bar
range - initial static tests had indicated a 7% increase in response
per 100 bar. The pressure sensor on the head will enable on-line
correction of the PA response.
|
|
|
Fouling
was discovered to be the cause of shifting the linear response
of the OIWM for an given set of conditions. The presence of fouling
and its extent is revealed by a 'pre-peak' in the recorded response.
This pre-peak can be used to trigger a cleaning mechanism, the
implementation of which is a matter for this JIP. The fouling
at ERT was largely corrosion products from the upstream pipe,
rather than oil filming.
|
|
|
The
OIWM was in use for 7 months continuously at ERT. For 3 months
the monitor was regularly removed, cleaned and replaced. No failures
were recorded during the testing (beyond initial infant mortality
of a hard disk).
|
|
|
No
quantifiable response.
|
|
|
Apparent
effects on one oil type were due to fouling. No effect was seen
on OIWM performance with second oil type at up to 4m/s, although
further testing is required to confirm.
|
|
|
No
discernible effect at range tested (<100ppm). At 2000ppm, some
signal attenuation was caused by agglomeration of droplets.
|
|
|
Installation
noise caused by near-closed control valves can prevent correct
operation of the OIWM. Normal levels of installation noise have
no effect on OIWM operation (OIWM tested alongside Troll subsea
pumps with no effect).
|
|
|
Some
evidence of reduced response at high solids loading. Further testing
and analysis required.
|
|
|
Free
gas at concentrations used did not significantly affect the monitor.
|
|
Table
2: Summary of test results at ERT
|
In summary
therefore, excellent reliability, accuracy and reproducibility was demonstrated
by the monitor during 7 months of independent testing. The data acquired
was used to develop a calibration algorithm for the clean head data.
The calculated measurements at 1000ppm and below (see table 3) are well
within +/- 10% of the averaged measurements from the measurement of
samples taken at the flow loop outlet. The 2000ppm calculation was affected
by the agglomeration of droplets (see table 2) attenuating the PA signal.
|
|
|
|
|
|
20
|
22
|
22
|
-0.3
|
|
100
|
115
|
110
|
4.4
|
|
500
|
471
|
499
|
-5.5
|
|
1000
|
1015
|
1000
|
1.4
|
|
2000
|
1607
|
2068
|
-22.3
|
|
Table
3: Oil concentration prediction by OIWM
|
|
