A decade ago BP Canada started installing HART-enabled
temperature and pressure transmitters in Ft. Saskatchewan.
Today a majority of the 600 transmitters in the plant
are HART-enabled. As part of the upgrade package, the
company also installed flow computers, up to four transmitters
per computer.
In the initial configuration, Boisvert says the 4-20
mA analog signal output from a transmitter was fed into
a flow computer. The digital signal superimposed atop
the analog signal as part of the HART standard functioned
as a second communication channel to and from the transmitter
but was not being used by the flow computer.
For
audit and other purposes, flow-computer readings were
and are compared to that of a standard, certified gauge.
One plant staff person handled transmitter calibration
while another did the meter proving. The accuracy in
the mass delivery demanded by contract was one or two
tenths of a percent, a figure that translated into a
need for highly accurate temperature and pressure readings.
For example, a temperature error of 0.25 °C results
in up to a 0.07% net flow error. Multiplied by the amount
of product Ft. Saskatchewan ships, that error meant
a potential loss of $250,000 a year. For NGL, the corresponding
figure was $350,000, creating strong incentives for
accurate measurements. In addition, those receiving
the product would make their own readings and take action
if a large enough difference was found. “There
will be a correction, if you have somebody at the other
end able to do the measurement as well as you do or
better,” says Boisvert in explaining what would
happen.
Speaking
one language
Problems, however, cropped up when Boisvert and coworkers
tried to reduce the measurement error. Despite diligent
transmitter calibration efforts that should have kept
the devices operating within a tenth degree, meter proving
continued to indicate an inaccuracy of up to half a
degree. The employee proving the meter would then adjust
the transmitters or ask for them to be calibrated. Calibration
showed that the transmitters were working well within
tolerances. The sequence would then repeat for each
employee's area until various team members witnessed
each he team finally found the problem to be within
the flow computer, specifically in the transformation
of the analog signal into its digital equivalent. “The
analog to digital conversion was not as accurate as
what we were looking for,” notes Boisvert.
The errors weren’t great, perhaps 0.2 to 0.5 °C
on a range of -18 to +65.5 °C. On the pressure side,
the discrepancy might be up to 50 kiloPascals on a range
of 10,000 kPa. Those small errors, though, were enough
to eat up a significant part of the overall allowable
error budget.
Faced with this problem, the team started looking for
a solution. A series of tests convinced them the readout
at the transmitter agreed with analog and digital signals.
That was when the group decided to sidestep conversion,
notes Boisvert. “If the flow computer speaks HART
and the transmitter speaks HART, let's read the digital
value rather than the analog value,” he says,
thinking through the process.
The simple-sounding idea required a bit of work. For
one thing, the flow computer's model didn’t natively
read or write HART commands. However, it was programmable
at a low level. The vendor provided what was essentially
a translator, allowing Ft. Saskatchewan’s team
to worry only about the higher level application software.
Part of the solution also involved upgrading the transmitters.
A typical specification for a transmitter of 0.5-degree
accuracy wouldn’t work. “We were looking
for that 0.2 degree accuracy,” says Boisvert.
Working with vendors, they found transmitters that would
hit the more stringent spec. Such a tightening only
made sense, though, because the conversion-induced error
had been eliminated.
Highly accurate results
With transmitters and flow computers talking without
an analog step in the middle and with more accurate
transmitters, the Ft. Saskatchewan plant achieved the
desired performance. It’s hard to pinpoint an
exact dollar figure saved due to the improvement because,
as Boisvert notes, the error could have been positive
or negative. The total amount of the possible adjustment,
however, runs in the hundreds of thousands of dollars
a year. Since the setup has been in place for a decade,
potential savings are in the millions of dollars.
Another benefit even harder to quantify has been removal
of internal inconsistencies. That has contributed to
creation of a well- deserved reputation that helps in
certifying measurements to external parties. “There's
never any issues about our accuracy,” says Boisvert.
There’s also been some savings of engineering
and technician time. To take one example, the old meter
proving required one employee to take a reading from
a certified device placed next to the transmitter. The
staffer would then have to go to where the flow computer
was located and see what it measured. Finally, the employee
would have to travel back to the certified instrument
and hope there hadn’t been a process change during
this back and forth. Now the task is much easier because
the transmitter's digital display shows what the flow
computer sees. “He can look at the certified device
sitting right next to the digital display on the transmitter
and compare his readings,” notes Boisvert.
There are also time savings because instruments aren’t
being tweaked, and other adjustments aren’t being
made only to be undone later. BP Canada Energy estimates
total savings of at least an hour a week of engineering
and technician time.
As for lessons learned by the whole process, Boisvert
cites one that could apply to a wide variety of situations.
“HART is not just a maintenance tool. It is a
process improvement tool as well,” he says.
SUBMIT
YOUR NOMINATION FOR PLANT OF THE YEAR AWARD |
