POWER February 2011 - 49

INSTRUMENTATION & CONTROL
er, it passes through the final superheater
before leaving the boiler. The temperature
of the steam leaving the boiler is also
measured and referred to as the final superheater
outlet temperature. The flow of
water into the desuperheater is regulated
by a modulating control valve, which is
supplied with feedwater from the boiler
feedpump discharge or from the economizer
inlet.
Time Response Characteristics
To begin, assume that the initiating event
for the control system is a step change in
the position demand to the spray valve.
The valve position response for a pneumatic
valve can be approximated by 1 to
2 seconds of dead time followed by a 2to
5-second time constant (Figure 1). The
water flow into the desuperheater responds
almost instantly to changes in the spray
valve position. The desuperheating process
is a very quick process, so the actual
steam temperature leaving the desuperheater
changes almost as fast as the water
flow entering the desuperheater. The time
constant of the water flashing into steam
probably is less than 1 second.
The measured temperature of the desuperheater
outlet responds considerably
slower than the actual temperature. The
response characteristic for the measured
temperature should be approximately 2
to 3 seconds of dead time followed by a
30-second time constant based on tests
conducted on several nearly new units in
the 1980s. The 30-second time constant is
based on the combined effects of the thermowell
and the temperature sensor, with
the thermowell being the dominant effect.
Also based on past unit testing, the response
time of the final outlet temperature
at the boiler is typically about 60 seconds
of dead time followed by a 100- to
200-second time constant due primarily to
the large metal mass of the superheater.
Plant Test Results
Several open-loop step response tests
were conducted on the superheater spray
valves on the unit under study, a 270-MW
coal-fired drum unit. While the unit was
at steady-state conditions, the spray valve
demand on one side was stepped up by the
operator. The unit was given approximately
20 minutes to come to a new steady-state
condition. It was clear from the data that
there were disturbances present and that
these disturbances made it more difficult
to measure the actual response times. The
desuperheater outlet temperature response
time was approximately 90 to 100 seconds
2. Disturbing results. Spray valve open-loop step response test results are shown. The
desuperheater outlet temperature response time was approximately 90 to 100 seconds, which
was considerably slower than the expected response time of 30 seconds. This discrepancy
between the expected and actual response time initiated an investigation into the cause. Courtesy:
Southern Company Generation
65
60
55
50
45
40
780
770
760
750
1,010
1,000
990
980
970
(Figure 2), which was considerably slower
than the expected response time of 30 seconds.
This discrepancy between the expected
and actual response time initiated
an investigation into the cause.
Plant Findings on Thermocouple
Installation
As a result of the slow time response observed
on the desuperheater outlet thermocouple
(TC), plant staff began checking
the installation in an attempt to identify
the cause. Several problems were found,
including these:
■ The TC spring-loading mechanism appeared
to be insufficient, which may have
allowed the TC tip to not be in direct contact
with the bottom of the thermowell.
■ The inside diameter of the thermowell
was 0.5 inch and the outside diameter
of the TC was 0.25 inch, so there was a
considerable air gap around the outside
of the TC.
■ The TC was ungrounded, meaning that the
TC junction was not directly touching the
tip of the TC sheath. The thermal insulation
within the sheath slows down the response
of the TC.
■ The thermowell design was very heavy
duty and featured a straight shank rather
than a tapered shank. The wall thickness
of the straight shank was 0.437
inch throughout its length. The thickness
of the bottom of the thermowell
was not known but was thought to be
approximately the same as the wall
thickness.
■ The TC tip design itself seemed to be not
very conducive to quick response time,
and the condition of the lead wire insulation
inside the sheath was questionable
and had some indications of electrical
shorts between leads in the sheath.
500
1,000
1,500
2,000
2,500
To address these problems, plant staff
500
1,000
1,500
2,000
2,500
500
February 2011 | POWER
1,000
1,500
Time (seconds)
www.powermag.com
49
2,000
2,500
installed new TCs in the left hand and
right hand desuperheater outlet locations.
The thermowell was not changed. The new
TCs were grounded (Type K), 0.5 inch in
diameter, and were installed with spring
loading in the head to ensure that the TC
remained in direct contact with the bottom
of the well at all times.
After the new TCs were installed, the
open-loop step response tests were repeated
and, maybe surprisingly, the time constant
of the temperature measurement did not noticeably
change. The conclusion drawn from
this discovery was that the main culprit in
the slow response time was the heavy-duty,
straight-shank thermowell.
Final superheater
temp (F)
Desuperheater temp (F)
Valve demand (%)

http://www.powermag.com

POWER February 2011

Table of Contents for the Digital Edition of POWER February 2011