Professional Documents
Culture Documents
The black line (darkest) represents the response of a well-tuned proportional control
system to an upset or change in the system, like a set point change for instance. Notice
how there are a few oscillations and then the system finds a new stable operating point at
a different valve position. The blue line is the response of this same system after the
throttling range has been decreased to the point where it starts to become unstable. The
red and the gray lines are the results of further decreases in the throttling range. Note
the valve was in a position where the flow temperature change at its input. If the
through it exactly matched the low load controller had a 5°F throttling range, a
condition. When there was no load, the 1°F temperature change at its input
system would have to allow the discharge would result in a 2.4 psi change in output.
temperature to drift down to 51.5°F to With a 2°F throttling range, the output
completely close the valve. Similarly, at would change 6 psi for every 1°F
high load conditions, the system would temperature change at the input sensor.
have to allow the discharge temperature So, as the throttling range of the
to drift up to 61.5°F before the control controller is decreased, the controller
valve would be driven fully open and becomes much more sensitive to changes
allow the design chilled water flow to at its input and responds much more
enter the cooling coil. strongly to them. This is good as long as
the system that the controller is
It is quite likely that the controller in the controlling can keep up with the
case study at the beginning of the paper controller’s changes. But, as you narrow
was a proportional controller with a 10°F the throttling range, most real systems
throttling range working as designed over will reach a point where time lags,
a range of load conditions. Early in the thermal inertia, physical inertia and other
day, when the load was low, a stable factors result in the over-all system
operating point was achieved with a response is slower than the controller’s
discharge temperature 4°F below the response to a change at its input. As a
controller’s set point. As the load result the system becomes unstable and
increased, the chilled water flow at this starts to hunt. (In human terms, the
valve position was inadequate and the system can’t keep up with the demands of
discharge temperature started to rise. the controller and goes crazy.) Figure 1
The rising discharge temperature illustrates what happens as the throttling
changed the offset in the system and thus range is reduced on a system.
output of the controller. This caused the
valve to open until a new point of stability
was achieved.
Figure 6 – The Effect of Proportional, Integral, and DerivativeControl Response on System Performance
In most cases, the biggest benefit associated with PID control for an HVAC application is the elimination of the
proportional offset via the integral function. Derivative action can minimize the process swing associated with a
system upset, but the benefits associated with this are often quite modest or insignificant in an HVAC application.
The open loop tuning technique will yield a response curve similar to the following. The parameters derived from the
response curve are used to determine the initial controller settings. They can also provide some insight into the
characteristics of the system. Notice how the apparent dead time is made up of a flat segment of pure dead time; i.e.
nothing literally happened, and a transition to a stable rate of change. In open loop tuning, you are most interested in
the apparent dead time and the slope of the linewhere the process is undergoing a stable rate of change.
systems using higher end controllers cycling. This is typically called the
often provided this sort of monitoring in natural period of the system and is a
the form of a circular chart, and some parameter that provides the basis for
stand-alone microprocessor based the subsequent steps and can also
controllers mimic this feature provide quite a bit of insight into the
electronically via a liquid crystal or CRT system’s characteristics.
display. If neither of theses options are 4. Set the controller settings to the
available to you, then you may want to following values.
bring a portable data-logger along that
has display capabilities to use while you Proportional gain = ¼ to ½ of the
are testing. At a minimum, it would need ultimate gain.
to have an input that could monitor the Integral time in minutes per repeat =
process variable for the control loop you 1.2 times the natural period.
are trying to tune. A second input that Derivative time in minutes = 1/8 of the
can document the signal to the final natural period.
control element is very handy, but not
necessary. 5. Monitor the performance of the loop
and make minor adjustments as
The general steps for the closed loop required to optimize the performance
method are as follows: and tailor it to the needs of the
system.
1. Turn off the integral and derivative 6. Subject the loop to upsets by making
gain. acceptable set point changes and/or
2. Increase the proportional gain of the shutting down and restarting the
loop in small increments until the loop system to be sure that stable
operation is achieved in a reasonable 9. Subject the loop to upsets by making
time and with-out excessive and/or large set point changes and/or
dangerous deviations in the process shutting down and restarting the
variable. system to be sure that stable
operation is achieved in a reasonable
The open loop technique involves placing time and with-out excessive and/or
the system in manual, and when the dangerous deviations in the process
process has stabilized, introducing a step variable.
change and observing the results. Figure
7 illustrates a typical response curve from There are some situations where these
an open loop tuning process. The general tuning process break down and won’t
steps are as follows: work, but they are rare. For instance,
some loops will exhibit a slope that
1. Place the controller in manual and continues to increase rather than
stabilize the process. Its important stabilizing in an open loop test. Other
that any changes introduced into the loops may exhibit an reversed response to
process during the test be changes an upset. These characteristics are
that you made. unlikely in an HVAC system, but are
2. Introduce a step change. discussed in Controller Tuning and
3. When you see the results of the Control Loop Performance, Second
change, make a step change in the Edition if you are interested in pursing
opposite direction with a magnitude of additional information in this regard.
twice that used for the original step
change. When tuning loops, the following general
4. Return the output to its starting value. concepts should be kept in mind.
Basically, you are trying to get the
result you want and put the process The natural period and the apparent
back in a safe and stable operating dead time of the control loop are very
mode with out exceeding critical important parameters. As a general
system parameters or tripping safety rule, the natural period will be about
equipment. 4 times the apparent dead time. The
5. Measure and document the apparent apparent dead time is the result of the
dead time. various lags in the system due to
6. Measure and document the slope of transportation times, slope in
the line in terms of rate of change per mechanism, thermal characteristics,
minute expressed as a percentage of etc. Generally, anything that you can
transmitter span divided by the step do to minimize these lags will improve
change you introduced expressed as a the performance of the control loop.
percentage of controller output span. If the control loop is subject to a
7. Set the controller settings to the periodic disturbance then the
following values. frequency of the disturbance might
have a very significant impact on the
Proportional gain = 1/slope to 1/(2 loops ability to control. The control
times the slope) loop will be helpless in dealing with
Integral time in minutes per repeat = disturbances that are short relative to
5 times the apparent dead time its natural frequency because they are
Derivative time in minutes = ½ the too fast for it to deal with. If the
apparent dead time. disturbance is at nor near the natural
frequency of the loop, then there is
8. Return the loop to automatic and strong likelihood that the control loop
monitor the performance of the loop will make things worse rather than
and make minor adjustments as better.
required to optimize the performance At first you may think that repetitive
and tailor it to the needs of the cyclic upsets are not a likely situation
system. in the HVAC industry. But, given the
configuration of the systems, there Given the variable performance
usually are many control loops that criteria seen by HVAC systems, you
interact through the dynamics of the may want to be a little conservative in
system. If one of the loops starts to your tuning parameter settings,
hunt for some reason, the result of it’s especially during the first year of
hunting will become a periodic operation. Otherwise, you may be
disturbance to other loops. Consider a faced with some major headaches
VAV fan system as an example. If the when an overnight weather change
system controlling the mixed air takes a system that was tuned to the
dampers starts to hunt, the dampers edge of stability over the edge.
will start moving around, This will Before tuning, you should have an
vary the static pressure requirements idea of what you expect to happen and
for the fan system, (especially if the what your desired outcome is as well
dampers have not been well sized) and as what the given system can be
thereby introduce a periodic expected to do. If you have reason to
disturbance into the static pressure believe that the valves and dampers in
control loop. If that disturbance the system have not been sized
happens to fall near the natural properly, then don’t expect high
frequency of the static pressure precision results. You should also try
control loop, it could cause that loop to get a feel for how fast the process
to suddenly begin to hunt. This would might respond to an upset.
lead to flow variations in the system You, and everyone else involved with
that could impact the performance of the tuning process should know and
the terminal unit flow controllers, the agree about how far you will let things
building static pressure control system go before you abort the test and shut
and the control loop controlling the down or otherwise take over control of
return fan. the system. This can be critical in
In general, for optimum performance, some cases if you want to avoid
you want to be just on the stable side damage to the system and equipment
of the ultimate gain point. or avoid unnecessary safety system
The ultimate gain of the system will trips. You will need to know how to
change as the characteristics of the respond quickly and with confidence
system change. In HVAC applications, to a situation that is going down hill
this can happen for a variety of rapidly. In situations where the risks
reasons including wear in the are high, it would be good for the
machinery and equipment and team to rehearse their actions in the
seasonal variations in the loads served event of a run-away process to be sure
and the heat transfer characteristics everyone knows what they are
of the equipment used to serve the supposed to do and can move quickly
load under various conditions. Loops and with confidence.
that are tuned in the winter may not You should have tested all safety
be stable in the summer or during the systems and interlocks and they
swing seasons. Variable volume should be set at the appropriate levels
system loops that were tuned with the required to protect systems, people
system operating at part load early in and equipment.
the day may not be stable at full load You should schedule your testing at a
later in the day. You should anticipate time when the system and the loads it
this and expect to retune frequently serves can tolerate some disturbance
during the course of the first year as and a shutdown down of the system
the systems go through all of their would not create a crisis. Tuning the
operating modes in real time for the discharge temperature control loop
first time. In addition, you should on a surgery air handling system
anticipate the need to retune during an open heart case is not a
occasionally as equipment ages and/or good idea.
systems are modified.
Plan on being readily available for Non linearity – Non linear
some time after you make your characteristics abound in the process
adjustments in case problems show and equipment that is associated with
up. Making significant changes to a HVAC systems. Most heat transfer
control system late Friday afternoon devices have a non-linear relationship
on your way out of town on a two between heat transfer rates, flow rates
week vacation will not make you very and temperature differences. Many of
popular upon your return if the the sensing systems that provide
changes result in operational inputs to the controllers we use are
problems and you aren’t there to help non linear. Examples include the
correct them or return things to the output from the thermistors
way they were. commonly used to measure space
Document everything. This includes temperature and the output from
the settings that were in placed prior differential pressure based flow
to starting the process, the settings measuring elements. Velocity limiting
you left in place when you finished, is another non-linear system response
and any other pertinent observations that can cause havoc with HVAC.
that you might have made during the Actuators and final control elements
process. This information should be generally have some finite speed at
communicated to the folks who will be which they can move through their full
running the systems right after you stroke. If the process goes through a
are done tuning. If there are latent large change, and the rate of change
problems related to the final tuning is faster than the rate at which the
parameters you left in the system, final control element can move to
they will often (but not always) show make a correction, then the process
up in the first hours or days of becomes velocity limited. If the
operation subsequent to your change in the process is small, the
adjustments. problem usually doesn’t show up. As a
Proceed slowly. The best thing that result, the loop may be stable for
you can do after you make an small upsets but unstable for large
adjustment to a tuning parameter is ones.
wait and watch. Often, it is tempting Loop interactions – Sometimes,
to make a second change fairly quickly designs include interactive loops
if the initial change you made doesn’t where in the normal response of one
produce the anticipated result. This loop will upset the second loop, which
may be a satisfactory approach if you will upset the first loop, etc. A pipe
are very familiar with the system and line with a pressure control loop that
the process is not particularly critical. modulates a valve to maintain the
However, bear in mind that the lack of pressure ahead of a second valve that
anticipated response could be due to is controlled by a flow control loop is a
system lags or other phenomenon that good example of this. Many times,
you failed to consider and you may one of the control functions can be
suddenly find yourself suddenly eliminated to solve this type of
dealing with a run-away process when problem. Or, one of the loops can be
the accumulated impacts of your tuned to be very “loose” while the
changes finally take effect. more critical loop is tuned for “tight”
control.
RELATED ISSUES Auto tuning – Auto tuning is one of
those things that sounds like a really
There are several control loop issues that nice feature, but in fact can be of little
are not directly related to the actual use or benefit in some situations. It is
tuning process but which will certainly probably unwise to rely on it as a total
impact it. These include: loop tuning solution, despite what
sales literature and salesmen may lead
you to believe. The exact auto tuning
algorithm employed will vary from changes that minimize the hysteresis
manufacturer to manufacturer. There rather than additional controller
have been instances where an auto tuning.
tune controller from manufacturer A Matching final control element spans
would not be able to tune itself in a to output spans – If the output span
certain process, but when a controller from the control system is not
from manufacturer B was substituted, matched to the actual span of the final
the same process was able to stabilize. control element, a problem similar to
In a different process, the controller wind-up will occur because the control
from manufacturer A was able to self loop spends a portion of its time trying
tune, but the controller form to actuate something that can not be
manufacturer B only achieved further actuated. Then, when things
marginal results. It is also important change, the controller spends time
to understand exactly how the auto backing down from its limit to a point
tune algorithm works. Some where the final control element begins
algorithms make gradual adjustments to actuate again.
based on ongoing observation of the
process. Others will actually do a
Filters – Because current technology
controllers use digital technology and
variation of the open loop method and
operate at very high speeds, they are
slam the final control elements up
capable of detecting and responding
against their limits. This can be an
to very small and inconsequential
undesirable approach on some
processes.
Hysteresis – Packing friction and play The Impact of Sampling Time On Observed Data vs. What is Really Going On
in a linkage system can cause control
The rate at which data is sampled for analysis can result in an inaccurate picture of what is going on. Sampling
problems when small errors exist, rates that are an multiple of the frequence of the disturbance can mask instability. Other rates can distort the
One very practical way to check for 58.00 A sampling rate of 1/second reflects the real time data fairly accurately.