POWER February 2011 - 52

INSTRUMENTATION & CONTROL
Because testing would be extremely
difficult to perform in the steam line
of interest, the tests took place in a
lab setting using a molten salt bath
at 350F.
bore. Wika also provided two sets of TCs-
one set of ¼-inch diameter and one set of
3/8-inch diameter. Each set contained an
exposed, grounded, and ungrounded TC.
Because testing would be extremely difficult
to perform in the steam line of interest,
the tests took place in a lab setting using
a molten salt bath at 350F. A room temperature
thermowell was submerged in the
molten salt. An array of five grounded 1/8inch
TCs (chosen for their short response
time) was submerged simultaneously. The
responses of all TCs (including the one in
the thermowell) were recorded. This test
was performed for all thermowell and TC
combinations-a total of nine test runs.
The results of these tests were compared
with the results of a separate set
of FEA models. These FEA models were
created specifically to model the lab tests
with initial conditions, bulk temperatures,
and convection values adjusted to the conditions
of each individual test. If necessary,
the FEA models were to be adjusted
until the FEA results and the lab results
correlated closely.
Possible areas of refinement for the FEA
models included the addition of contact resistance,
the adjustment of convection values,
and the addition of heat transfer to the
base of the thermowell (or heat " leaving "
the thermowell). It was found that none
of these refinements was necessary. However,
due to salt temporarily solidifying on
the thermowell, it was necessary to model
a dynamic load up to the point of when the
salt melted off of the thermowell. Beyond
this point, the FEA models were nearly
identical to the lab results, and it was determined
that the method of modeling and
calculating a convection value was valid.
Because the thermowells have internal
conditions that are nearly identical at low
and high temperatures, these validated
models were used to simulate the response
at high-temperature steam conditions
(with adjustments made for convection
and other fluid conditions).
5. On the fast track. The comparison of control performance with fast and slow inner loop
temperature measurement is shown. Courtesy: Southern Company Generation
Fast thermocouple
2.5
2.0
1.5
1.0
0.5
Slow thermocouple
Impact on Control System
Performance
It is known that adding a time lag in the
feedback of a control loop will degrade the
performance of the loop. In this study, it
appears that the time lag of the desuperheater
outlet temperature is actually about
100 seconds, when it was expected to be
about 30 seconds. This measurement is
used as the feedback signal for the inner
loop of a typical cascaded steam temperature
control strategy. To evaluate the
impact that the additional time lag has on
the overall loop performance, a simple dynamic
model of the control system and the
process was developed in the MATLAB/
Simulink environment.
The control system was equivalently
tuned for two different desuperheater outlet
temperature measurement response
times, 25 seconds and 100 seconds. The
final superheater temperature measurement
response model was the same for
both cases. Transients were performed on
both the fast and slow models to quantify
the impact of the change in the inner loop
process response on the control performance.
The transients consisted of four
step changes in setpoints and disturbances.
Throughout the transient analysis, the
control performance of the faster-responding
loop is better than that of the slowerresponding
loop (Figure 5), as expected.
Many factors influence the overall response
time of temperature measurement
in boiler steam lines. In this project, the
response time of one common boiler temperature
measurement, taken at the desuperheater
outlet, found that a thick-walled
thermowell was the primary culprit in the
longer-than-expected response time.
In fact, the thick-walled design resulted
in almost double the time constant of a
more typical tapered tip design. Laboratory
tests to verify the FEA models were
performed and confirmed these results.
Further tuning to allow for the thermodynamic
lag caused by the thermowell
improved the responsiveness of the plant
controls.
This article is based on a paper presented
by the authors at the 15th Annual
POWID/EPRI Controls and Instrumentation
Symposium. ■
-Cyrus Taft (cwtaft@taftengineering
500
1,000
52
1,500
2,000
Time (seconds)
www.powermag.com
POWER | February 2011
2,500
3,000
3,500 4,000
.com) works for Taft Engineering Inc. in
Harriman, Tenn. John Sorge (jnsorge@
southernco.com) works for Southern
Company Generation in Birmingham,
Ala. Jackson Willis (jackson@
roycmartin.com) is a sales engineer for
Roy C. Martin & Co.
Final superheat temperature (F)
http://www.powermag.com
POWER February 2011

Table of Contents for the Digital Edition of POWER February 2011
