Note : Les descriptions sont présentées dans la langue officielle dans laquelle elles ont été soumises.
~36~
D17-4461-1
This invention relates to improvements in
air-conditioning system controls.
In particular, this invention relates to improvements
in the control of the output oE a variable inlet vane fan.
The present invention also relates to the positioning
oE the sensing probe in an air-conditioning system.
~ n addition, the present inven-tion provides
improvements in the stability of a control system of an
air-conditioning system.
PRIOR ART
For many years, variable inlet vanes have been a
commonly used device for controlling the output of a fan such as
the fans commonly used in air-conditioning systems. The use of
variable inlet vane fans in air-conditioning systems has
increased substantially since the popuiarity of variable air
volume air-conditioning systems has increased.
In controlling the movement of the variable inlet vanes
in order to adjust the fan output capacity, it has been found
that while the air flow is a function of the position of the
variable inlet vane, the actual air flow varies depending upon
whether the vanes are opening or closing. This difference in air
flow is as a result of hysteresis losses which are particularly
due to the linkage mechanisms required to control the movemen-t o~
the inlet vanes. The hysteresis is more pronounced on double
inlet fans than on single inlet fans mainly due to the more
' -'''F
D17-4461-1
extensive linkages required for the former. This hysteresis
would normally de-stabilize a conventional inlet vane con-trol
system. A conventional control system normally cycles about a
set point and the hysteresis which is inherent in the operation
of variable inlet vanes of a fan results in an unstable control
1 oop .
To prevent the system from cycling about it's control
or set point, the conventional practice has been to introduce
restrictors into the sensing loop of a pneumatic system or
damping circuits in an electrical or electronic control system.
The purpose of these control system modifications are to provide
system damping. These systems do not address the problem
correctly and therefore do not provide the required result. The
only efEective remedy presently in use is to decrease the control
system sensitivity, i.e. loop gain, to such an extent that the
control system cannot differentiate between the two extreme
sensed conditions. The disadvantage of this solution is that the
sensitivity is so low that the control system cannot follow
changes in fan system conditions and therefore the control system
is slow and sluggish and in some cases may be inoperative.
I find that I can improve that stability of the air
conditioning system of the present invention in circumstances
where the sensed variable which is usually the static pressure,
is not steady enough for conventional control. If the sensed
variable is not steady enough for control, I provide a filter
D17-4461-1
with a long-time constant. rhis filter serves -to ~ilter out
transients and to allow actual load changes, heavily filtered, to
pass from the signal transmitter to the controller. The time
constant should be adjustable and should normally be in the range
of 0 to 200 seconds. The time constant may then be adjus-ted on
site to be a good compromise between the desired system response
and the control system stability.
Alternatively, I may substitute an integrator for the
filter described in the preceding paragraph. In control systesm
requiring fast response times, cycling occurs at a specific
frequency. This frequency varies with the setting of -the
sensitivity or gain ad~ustments. If the control system is ~et
for the desired sensitivity and impermissable and excessive
cycling occurs, substituting an integ~ator for a filter would be
helpful. The integrator is inserted into the circuit between the
transmi-tter and the con-troller. As the transmitter senses the
system cycling, the integrator will act on the output. As long
as the cycling is of a set frequency, the integrator will
integrate the fluxuations out. The integral of a symmetrical
periodic function is 0 and consequently, the integrator output
will consist of the steady value only as sensed by the
transmitter. It is important that the frequency of the cycling
be measured on the job si~e and adjusted into the integrator.
The conditions under which air flow takes place in an
air conditioning duct is subject to the following equation:
D17-4461-1
dP = Clq ~ C2q ~ C3q EQrJATIO~ 1
Where C1, C2, C3 are constants
q = air flow in suitable uni-ts
dp = pressure drop oE duct system in suitable
units
EQUATION 1 breaks up the pressure drop into components
due to totally turbulent flow depicted by the q2 term; due to
laminar ~low depicted ql term; and a component independent of
flow depicted by the q term. (Note q = 1). Experience
indicates that in the range of operation of air conditioning
systems, the laminar flow component is negligible. If we assume
this term to be zero, then Equation l-simplifies to the
following:
dP = Clq -~ C3q EQUATION ~
The characteristic curves of a typical fan are well
known and are provided by fan manufacturers. If a fan is
a-ttached to an air conditioning system consisting of filters,
coils, ducts, terminal boxes, dampers, etc., air flow will take
place at the point on the characteristic curve where the fan and
the system curves intersect. If the system is a constant volume
system, there would be one fan curve depicting the head-flow
condition at a particular constant speed. However, there are an
infinite number of system curves between a minimum and maximum
Elow. Each time a damper changes position, resulting in
different pressure in the duct, flow would be affected. Since
~ ~3ti7~3
D17-4461-1
air flow can only take place at the point of intersection of the
fan curve and the system curve, if the duct pressure changes,
this would have the effect oE moving away Erom -the point of
intersection. Since it is known that this cannot happen, the
condition must intersect. The only way this can happen is with a
new system curve intersecting the fan curve at the new
flow-pressure conditions. Hence, there can be an infinite number
of system curves.
If the fan is to vary capacity, there would likewise be
an infinite number oE fan characteristic curves between minimum
and maximum capacity; one for each speed. The fan system
operation is somewhere in the region defined by the minimum
capacity-flow and the maximum capacity-flow which is determined
by the system condition. That is to say the minimum capacity
flow is that which is achieved when the system is clean and the
maximum capacity-flow is that which is achieved when the system
is dirty. Dirty, in this context, refers to the clogging of
coils, filters, or other system components which result in a
higher pressure drop across them on the air side.
I have found that in order to keep the fan-system
operation steady, stable and accurate, it is important to locate
the tap of the static pressure sensing probe as short a distance
downstream of the fan as is practical and to locate the pressure
sensing device as close as possible to the tap. Ideally, the tap
would be placed at the fan discharge flange, however, this is
~9~JJ~
D17-4461-1
impractical since turbulence is too great at this point leading
to erroneous readings. Effectively, the same result may be
obtained if the tap of the sensing device is located at any point
in the duct system upstream of the first branch line.
PreEerably, the tap oE the sensing device is located within the
fan room and preferably less than 20 duct dia~neters downstream of
the discharge fan Elange. PreEerably, the sensing device is
located adjacent to the tap to eliminate control system time lags
and phase shifts associated with long sensing lines.
Traditional control -theory states that a control system
should react fast and it is generally taken that the faster the
reaction, the better. While this leads to good accurate control,
it is detrimental in air conditioning fan systems since there are
other control loops downstream of the fan, i.e. the terminal
boxes which discharge air into the space which is under control
of a room thermostat. The whole air conditioning system can
become unstable if the duct air pressure is caused to vary
considerably. This instability can cause undesirable temperature
changes in the space and provide a generally poorly operating air
conditioning system. Considerable difficulty has, however, been
experienced in attempting to maintain the duct pressure stable
and constant.
Most large air conditioning systems have a supply fan
and a return air fan. The supply fan is sized to deliver the
required air quantity at a pressure estimated to overcome losses
r~
D17-4461-1
of the filters, dampers, coils, ducts and other restrictors.
This means that the design fan pressure selected is normally
high, being about 4 to 6 inches WG (1 - 1.4 kPa) at the design
flow for medium pressure systems. The only purpose of the return
Ean is to overcome the duct losses of large duct systems.
Consequently, re-turn fans are selected to operate at a pressure
typically ranging from 0.5 to 2O0 inches WG (0.125 - 0.5 kPa).
Traditionally, the capacity control of the return fan has been a
problem. Early systems slaved the return air fan off the supply
air fan static pressure controller so that both fans tract. The
problem with this approach is that due to the vastly different
pressure selections of the fans, their capacity changes versus
linear control signal inputs are vastly different, i.e., their
curves have different slopes and sometimes are not linear.
Therefore, this matching of capacities only occured at the point
of calibration. This problem manifested itself in pressure
changes in the space as well as poor temperature regulati~n.
The pressure changes were sometimes so bad that in
large buildings lobby doors could not be opened or kept closed.
In an attempt to overcome this problem, some engineers
have controlled the capacity of the re-turn air fan from a space
static pressure controller. This offered only limited
improvement and actually introduced problems of its own. First
of all, the space pressure controller had to control a very low
value, usually around 0.25 inches WG (62 kPa~. It is difficult
D17-4461-1
to find a controller that can control space static pressure
reliably. Usually, the opening of the door admitting a blast of
air or the wind genera-ted by a person running the sensor would
cause the controller to take corrective action even though none
was required. The second problem was the location oE the sensor.
If the sensor is located in the front lobby of a large building,
the pressurization problem would still exist in the back lobby
and vice ~Jersa. If two sensors were used, the questions arises
which should be used or should the signals be combined to provide
an average. These proposals have not provided a satisfactory
solution to this problem.
I have found that these difficulties can be overcome by
providing a method of capacity control of the return air fan that
matches the capacity of -the supply fan.
A further difficulty which is experienced in attempting
to air condition large buildings is that known as the "stack
effect". The stack effect is caused by temperature differences
between the constant temperature space and the changing outside
air temperature. As the outside temperature drops, the density
of the cold air increases, causing it to fall to the ground
causing an increase in this air pressure. Since the air
temperature in the building is kept constant, its pressure may be
considered constant. This higher density air then forces its way
inside the building, entering through cracks around the doors anc.
windows, which are recognized by the inhabitants of the building
D17-4461-1
as drafts. Sometimes the pressure difference is so great that
the building doors cannot be opened except with great difficulty.
The exac-t opposite occurs in summer. In this case, the
outside air temperature is higher than inside tempera-ture, so
that the air has a tendency to flow to the outside through the
building cracks. This problem is not noticeable since i-t does
not create uncomEortable "drafts" to the inhabitants. One
noticeable effect, however, is that the building doors may stay
open due to the higher internal pressure and the presence of a
whistling noise around tall shaft access doors like stairwells or
elevator shafts. The escape of conditioned air does, however,
represent a loss in efficiency.
I have found that these pressurization problems can be
solved by providing a stack effect compensator. A manual stack
effect compensator may be in the form of a potentiometer wired to
the return fan offset. In effect, operating the potentiometer
would offset the tracking of the return fan by an amount equal to
half of the rotation of the potentiometer. At 50%/rotation, the
AFC would track the return fan as if the SEC was not in the
circuit. Rotating i-t to one extreme would decrease the return
fan tracking ran~e and rotating it to the other extreme would
increase the return Ean tracking range. This could result in the
return of less air than would be delivered by the supply fan
resultiny in more air being available to pressurize the building,
as may be required in winter, to compensate for the higher
D17-4461-1
outside air pressure. This would eliminate the door operating
problem and whistling around the building shafts. Rotating it in
the other extreme would increase the amount of air being
returned~ This would reduce the amount of air exfiltrating from
the building and remove the door operating problems and the
whistling around the building shafts.
As previously, indicated that stack effect is caused by
a temperature differential between the constant building
temperature and the variable outside air temperature, I further
propose that an automatic stack effect compensator be provided.
The automatic stack effect compensator consists of an outside air
tempera-ture sensor supp~ying an adjustable range offset to the
AFC in the sa~e manner as the manual potentiometer described
above. This stack effect compensator has the ability to
automatically adjust the tracking range of the re-turn fan to
minimize space pressure fluctuations.
Difficulty is frequently experienced in attempting to
provide adequate air conditioning in buildings which have
supplementary exhaust systems such as laboratories which have
fume hood exhaust fans. These exhaust fans are normally under
the control of the occupant and may be started and stopped in any
order and they may be of different capacities. The total air
exhausted by these exhaust fans may represent a substantial, if
not total, percentage of the air supplied by the air supply fan.
Since a variable air volume system supplies a variable amount of
i7~
Dl 7 - d~ 4 6 1--1
air depending on space demand, and the fume hoods are made to be
started and stopped and random, the space pressure control
becomes a formidable problem. I have developed a simple and
effective solution to this problem. The basic problem is to keep
the total air exhausted and the return air about the same as -the
air supplied by the supply fan. There may be a slight offset
either negative or positive depending on whether it is desirable
to keep the space negative or positive with respect to the
outside. I provide a potentiometer for each exhaust fan. If all
of these potentiometers are wired in series and fed from a
constant current source, and the potentiometer value is adjusted
to be analogous to the air exhausted by the fan, then if that fan
were to be started by the occupant, an auxiliary contact would
short out that resistance, a voltage change proportional to the
air exhausted by the fan is obtained. The voltage can be used to
provide an offset to the return fan tracking circuit in the same
manner that the stack effect compensator does. The net effect is
that the return fan speed can be decreased or increase by an
amount corresponding to the air exhausted by the exhaust fans.
The present invention seeks to overcome the
disadvantages of the prior art described above and provides a
control signal which takes account of the inlet vane hysteresis.
According to one aspect of the present invention, there
is provided a method of controlling the air flow output of a
selected variable inlet vane fan comprising the steps of:
11
D17-~461-1
a) generating a flow control signal in a controller
which is programmed to selectively provide a first signal which
is proportional to the increasing response plot of air Elow with
respect to vane position when -the vanes of the selected fan are
opening or a second signal which is proportional to the
decreasing plot of the air flow with respect to vane position
when the inlet vanes of the selected fan are closing,
b) detecting the direction of response required to
meet the demand variation as between a requirement for increased
air flow and a requirement for decreased air flow,
c) activating the con-troller to provide said flow
control signal in the form of said first signal when a
requirement for increased flow is detected and to provide said
flow control signal in the form of said second signal when a
requirement for decreased flow is detected,
d) transmitting said full control signal to said fan
to control the position of the inlet vanes of the fan as required
in use.
According -to one aspect of the present invention, there
is provided in a variable air volume air conditioning system
having a variable air volume fan located to discharge air into a
duct leading to a plurality of branch lines, the improvement of
pressure sensing means communicating with said duct upstream of
said branch line.
12
;7~
D17-4461-1
According to a further ~spect of the p.resent
invention, there is provided in a variable air volume
air-conditioning system in which a controller, detecting the
direction of response required to meet the demand variation as
between a requiremen-t Eor increased air flow and a requirement
for decreased air flow,
Figure 1 is a diagramatic illustration of a building
having an air conditioning system of the type of the present
invention.
Figure 2 is a block diagram illustrating a control system
suitable for use in association with the fans illustrated in F.igure
1.
Figure 3 is a diagram illustrating a variable inlet vane
controller for controlling the supply fan of the air conditioning
system of Figure 2.
Figure 4 is a diagram illustrating a variable inlet van
controller for controlling the return air fan of Figure 2.
Figure 5 is a graphic illustration of the variable inlet
vane response curves of a typical variable inlet vane fan.
Figure 6 is a diagram illustrating a response curve of a
typical control system.
Figure 7 is a diagram of a system Eor generating an offset
signal to the controller when air exhaust systems are provided in
the space to be conditioned.
13
~9~
D17-4461-1
With reference to Figure 1 of the drawings, -the reference
numeral 10 refers generally to a building in the form of a
multi-storey structure which has a fan room 12 from which
conditioned air is discharged through a duct 14 from which a
plurality of branch ducts 16 extend at various levels through the
building.
As shown in Figure 2 of the drawings, a variable inlet
vane supply fan 18 is connected to the duct 14. A constant speed
motor 20 is connected to the fan 18 and is powered from a suitable
power source. An inlet vane operator 22 is connected to the vanes
24 and operates to move the inlet vanes to and fro between an open
and a closed position. In addition, a return air fan 19 is
connected to the return air duct lS. A constant spaced motor 21 is
connected to the fan 19 and is powered from a suitable power source.
An inle-t vane operator 23 is connected to the vanes 25 and operated
to move the inlet vanes to and fro between an open and a closed
position.
A sensing tap 26 is located in the supply duct 14.
Preferably, the sensing tap 26 is located at a distance L from the
discharge end of the supply fan 18 which is less than 20 D wherein D
is the largest duct diameter downstream of the fan.
The location of the sensing tap 26 is of considerable
importance. In order to keep the fan system opera-tion steady,
stable and accurate, it is more important to keep the fan from
operating in its unstable range than it is to satisfy the duct
1~
D17-4461-1
system requirements for good overall control. I have been able to
achieve this objecti~e by locating the static pressure sensing tap
26 as short a distance downstream from the Ean as is prac-tical.
Ideally, the static pressure sensing probe 26 would be placed at the
supply fan discharge flange, however, this is impractical since
turbulence is too great at this point leading to eroneous readings.
Effectively, the same result may be obtained by locating the static
pressure sensing probe 26 within about 10 to 20 duct diameters
downstream from the fan discharge. It will be noted, however, that
the static pressure sensing probe may be located atany position of
convenience along the discharge duct upstream of the first branch
line 16. The location of the static pressure sensing probe in this
manner is contrary to the conventional practice in the
air-conditioning industry. Conventional thinking is that it is the
space which is to be conditioned that must be satisfied and
therefore the sensor should be located about 2/3rd's of the way
along the longest duct run. While it may be difficult to argue
against this reasoning from an intuitive point oE view, in practice
it is virtually impossible to determine where such a location may be
found. The duct systems usually run from the fan discharge down one
or more shafts in a building with multiple take-oEfs on each floor
with the result that it is difficult, if not impossible, to
determine where the optimum location for the pressure sensing probe
is within the building. The main reason for selecting the 2/3
location, however, relates to the control system stability and is a
v
D17-4461-1
compromise between satisfying the longest duct run and the
practicality of identifying such a location. It has been generally
believed that iE the sensing point or tap is sufficiently far away
from the fan, the problems associated with the usual inlet vane
controls do not show up. The main problem, however, is the
hysteresis losses in the operation of the vanes of the fan with -the
result that locating the sensor in the location previously
considered to be the optimum location, does not solve the principal
problem~
HYSTERE~;IS LOSSES
In Figure 5 of the drawings, the curve 32 serves to
illustrate the airflo~ output of a typical variable inlet vane
fan as the position of the inlet vanes move from a closed
position to an open position and a curve 34 serves to illustrate
the air flow of the same fan as the inlet vanes move from a fully
open position to a fully closed position.
From Figure 5 of the ~rawings, it will be seen that for
a fan inlet vane position Y, air flow in the amount Ql or Q2 may
be obtained depending upon whether the inlet vanes are opening or
closing. The difference between these two values results from
the hysteresis of the control mechanism required to adjust the
position of the vanes. This hysteresis is more pronounced on
double inlet fans than on single inlet fans, mainly due to more
extensive linkages required for the former. This hysteresis
serves to de-stabilize any existing control system used for
16
D17-4461-1
controlling the position of the inlet vanes of a variable inlet
vane fan. If, for example~ it is desired to con-trol the air flow
at the value Qc illustrated in Figure 5 of the drawings, it
will be apparent that this air flow may be obtained with the
inlet vane position YCl if the vanes are opening or in the
position YCh if the vanes are closing. In view of the fact
that a conventional controller system normally cycles about a set
point, the addition of hysteresis results in an unstable control
loop.
VARIABLE INLET VANE CONTROLLER
The variable inlet vane controller 30 (Fig.2) receives
a signal from the sensor 26 through a line 28 which may be in the
form of a plastic tube or the like which conveys a pneumatic
signal to the controller 30. The controller 30 generates an
output signal 32 which is directed to the vane mo-tor operator 22
which is operable to open and close the vanes 24 as required in
use.
The variable inlet vane controller 30 will now be
described with reference to Figure 3 of the drawings.
As shown in Figure 3 of the drawings, the line 28
communicates with a transducer 40 which serves to convert the
pneumatic signal to an electrical signal. The electrical signal
is fed to a signal conditioner 42 which provides a signal having
an output of the order oE O to 5 volts or O -to 10 volts. This
signal is fed to an adjustable filter ~4. The Eilter 44 is
17
~ ~967~
D17-4461-1
adapted to overdampen the signal which it receives from the
conditioner 42 so that its output signal will change
progressively to a steady state over an extended time period to
ensure that a substantially stable air pressure is maintained in
the air duct system 14 in use. This overdamping feature will be
described hereinaEter with reference to Figure 6 of the drawings
in more detail. The outpu-t signal from the filter 44 is directed
to a controller ~6 which communicates through line 48 with a
driver 50 and through line 52 with a response de-tector 54. If the
response detector 54 detects an increase signal, the feedback
loop provides a feedback signal to the controller through line 56
and if the response detector 54 detects a decrease signal, this
is communicated -to the con-troller through the feedback line 58.
The controller 46 is programmed to generate a first signal output
which is proportional to the increasing response plot of air flow
with respect to vane position when the inlet vanes of the
selected fan are opening or to generate a second signal output
which is proportional to the decreasing response plot of air flow
with respect to vane position when the inlet vanes of the
selected fan are closing. The detector signals provided through
the feedback lines 56 or 58 serve to activate the controller 4~
to generate the first or second control signal. The value of the
Eirst or second control signal is determined by the controller
which compares the signal output from the filter 44 with the
18
Dl7-4461-l
curves 32 and 34 (Fig.3) and a set point signal which is provided
by the set point control 60.
As shown in Figure 3, a line 202 is taken from the line
28 after tile filter 44. As shown in Figure 2, this line leads to
a second controller 200 which serve to control -this operation of
the inlet vanes 25 of the return air inle-t vane operator 23
through line 201.
METHOD OF OPERATION OF VARIABLE INLET VANE CONTROLLE~
In use a drop in pressure in the air duct 14 will
signal a requirement for increased air supply and this signal
will be transmitted from the detector 26 throu~h the line 2~ to
the controller 30. The controller 30 is operable as previously
described to determine whether or not the signal is indicative of
a demand for increased air flow or decreased air flow. If as
previously indicated, a requirement for increased air flow is
detected, the controller determines the vane position requ.ired to
obtain the required air flow with reference to the curv 32 and
gener~tes a first signal output to the driver 50. If, on the
other hand, -the response detector 54 had detected a requirement
for decreased air flow, the controller 46 would cletermine the
required vane position with reEerence to the curve 34. The
output of the driver 50 is conveyed to the vane control motor 22
which moves the vanes 24 to the position required in order to
provide the desired air flow.
19
D17-4461-1
Various modifications of the controller of Figure 4
will be apparent to those skilled in the art. For example, an
integrator 62 ma~ be used in place of the filter 44 or an
integrator 64 may be used in addition to the filter 44. IE the
integra-tor 6~ is used in addition to the filter 44, it is
arranged in series with the filter 44 and may be located before
or after the Eilter 44. The integrator 62 or 64 may serve the
same function as the filter for periodic disturbances, that is to
say the integrator 62 and 64 may serve to integrate out
~luctuations in the signal prior to transmission of the signal to
the controller.
RETURN AIR FAN CONTROL
As previously indicated, the input signal to the return
air fan controller 200 is the signal convened by the line 202
from the controller 30. It will be understood that the return
air fan inlet vane characteristics are considerably different to
those of the supply fan with the result that before passing the
signal in the line 202 to the contxoller 246, it is necessary to
transpose or condition this signal be means oE a conditioner 20
so -that it varies in accordance with the output curve of the
return air fan rather than the output curve of the supply air
fan. This signal is transmitted through the line 228 to provide
the set point signal of the controller 246. The controller 246
driver 250 and response detector 254 operate in the same manner
as the corresponding cornponents of controller 30 and serve to
D17-4461-1
provide a feedback signal to the controller 246. Thus, i-t will
be seen that the output signal of the controller 200 which is
directed to the inlet vane operator 23 through line 201 serves to
control the operation of the return air fan in a manner which
will serve to insure that the operation of the return air fan
closely matches the operation o~ the supply air fan.
OFFSET TRACKING OF THE RETURN AIR FAN
In order to overcome the difficulties associated with
the "stack effect" previously described, I provide a manual stack
effect compensator potentiometer 270. This signal is conditioned
by a signal conditioning or translating device 272 and is then
fed to the controller 246 through the line 274. The signal
generated by the stack effect compensator is received by the
controller 24~ and serves to offset the tracking of the return
fan by an amount equal to half the rotation of the potentiometer.
A 50~ rotation of the AFC would track the return fan as if the
special effect compensator was not in the circuit. Rotating it
to one extreme would decrease the return fan tracking range and
rotating it in the other extreme would increase the return fan
tracking range. This would result in the return oE less air than
would be delivered by the supply Ean resulting in more air being
available to pressurize the building, in winter, to compensate
Eor the higher outside air pressure. This would eliminate the
door operating problem and whistling around the building shafts.
Rotating it in the other extreme would increase the amount of air
21
Dl7-4461-l
being returned. This would reduce the problem oE air
exfiltrating from the building and would eliminate the door
operating and whistling problems.
The stack efEect problem is caused by a temperature
difference between the constant building temperature and the
variable outside air temperature. This problem is overcome by
providing an automatic stack efEect compensator. This device is
illustrated in Figure 4 oE the drawings and includes an outside
air tempera-ture sensor 280 which provides a signal which is
conditioned by a signal conditioner 282 which has a range
adjustment potentiometer 284 and a 0 set potentiometer 286. The
output signal from the signal conditioner 282 may be Eed to the
conditioner 272 through line 288 in the same manner as the signal
of the manual potentiometer. This signal has the ability to
automatically adjust the tracking range of the return fan to
minimize space pressure fluctuations.
A further signal may be used for the purpuses of
oEfsetting the controller 246. This signal is supplied through
line 290. This signal can be used to compensate for air which is
exhausted from the air space which is to be conditioned such as
by way of the fume hood exhaust Eans which are widely used in
laboratories. A typical Eume hood control system is illustrated
in Figure 7 oE the drawings wherein an auxiliary contact 301,
302, 303, 304, 305 and 306 is provided and a line which
communicates with the fume hood e~haust fan motor starter of each
22
D17-4461-1
of six different fume hoods. Potentiometers 310, 312, 313, 314,
315 and 316 are provided, one associated with each Eume hood
motor. The poten-tiometers 11~ to 116 are wired in series and fed
from a constant current source 320. The potentiometer values are
adjusted to be analogous to the air exhausted by the fan with the
result that when the fan is started by the occupant, the
auxiliary contacts associated with the fan short out that
resistance and a volt changed proportional to the air exhausted
by the fan is obtained. The signal translator 322 translates
this signal to a signal which can be ied through line 290 to the
signal conditioner 27~. It will ~ appararent that by this
system, it is possible to keep the total air exhausted and the
returned air about the same as the air supplied by the supply
fan. There may be a slight offset either negative or positive
depending on whether it is desirable to keep the space negative
or positive with respect to the outside. This system is
particularly well suited for use in controlling serious air
control problems which plague laboratory air supply systems. It
will be understood that this system is not restricted to fume
hood exhaust systems but is applicable to any system in which the
air is discharged from the space which is to be conditioned by
way of a separate exhaust system.
OVER-DAMPENING OF CONTROL SIGNAL
According to general control theory, the optimum
control system is one that is op-timally dampened, i.e. a system
23
D17-4461-1
wherein the output amplitude cycling fails within a decay
envelope within approximately 3 excursions. In this regard,
reference is made to Figure 5 of the drawings which illustrates
optimal damping of the amplitude of the conven-tional control
signal illustrated by the line 70. The signal 70 is normally
dampened within the decay envelope illustrated by the lines 72
and 74 within approximately 3 excursions.
In air-conditioning systems, any changes that take
place in the system loading are due to the sun shining in
windows, due to people moving into and out of the space to be
conditioned or due -to lights and heat producing equipment such as
computer terminals or the like being turned on or off. If, for
example, a computer is running and producing heat, when it is
turned off, it ~unctions as a heating element. The residual heat
reduces very slowly. In addition, there is the thermal mass of
the building structure that dampens all air-conditioning load
changes. By the time the terminal air conditioning boxes
collectively react to these load changes, and they should
collectively react slowly, sometimes minutes will pass.
Frequently the load changes in one part of a system are cancelled
by load changes in another part of the system which are nearly
identical in magnitude but opposite in direction. For example,
the load resulting from the movement oE direct sunlight around a
building in the course of a day can result in changes in the load
24
D17-4461-1
applied to various parts of a building while the total load may
remain substantially the same over the full day.
Conventionally, control system response times are
measured in very small fractions of a second, frequently in milli
or micro seconds. ~owever, air-conditioning load changes take
place in much longer time periods as for e~ample, periods as long
as ten minutes or more. When these changes take place, the
control system must be sensitive enough to sense them and to
react accordingly. I have recognized these characteristics and
for this reason I provide a control system in which the control
signal is over-dampened, thereby to increase the controller
sensitivity. Over-dampening is achieved by the filter 4~ or
integraters 62 and 64 previously described so as to generate a
signal to the controller 46 which follows the curve 76
illustrated in Figure 5 oF the drawings.
~ he slow system response which I am able to generate is
important for other reasons. Terminal boxes are connected to the
fan duct which are usually under the control of a room
thermostat. Such a thermostat will allow a typical box to pass
just the correct amount oE air to the space to satisfy the load.
The correct amount oE air depends on the pressure upstream oE the
box remainincJ constant. If the fan capacity control system which
-tries to keep this pressure constant is "live" and allowed to
cycle, causing the duct pressure to vary, the box will allow more
or less air to pass to the space causing temperature
~ . J
D17-4461-1
fluctuations.These fluctuations would be sensed by the thermostat
which would try to correct them. The net result of this scenario
is that the two con-trol systems fight each other and the whole
air-conditioning system would be unstable. The over-dampening of
the fan capacity controls helps to prevent this from occurring
and greatly assists in providing stable controls.
As previously indicated in Figure 2 oE the drawings,
the sensed variable which is detected by the detector 26 is the
s-tatic pressure. Generally, this static pressure is no-t steady
enough for control purposes and the filter 44 which has a long
time constant serves to filter out transients and to allow only
actual load changes heavily filtered to pass to the controller
46. The time constant is adjustable with the adjustment
providing a range of 0 to 200 seconds. The time constant may
therefore be adjusted on-site to be a good compromise between the
desired system response time and control system stability.
As previously indicated, integrators o2 or 64 may be
substituted fGr or used in conjunction with the filter 44. In
control systems requiring fast response times, cycling occurs a-t
a specific frequency. This frequency varies with the damping
ra-tio of the sys-tem. If the control system is set for the
desired sensitivity and impermissable and excessive cycling
occurs, substituting an integrator for a filter may help to
overcome this difficulty. The integrator must be inserted into
the circuit as illustrated in Figure 3 of the drawings between
26
D17-44~1-1
the signal conditioner 42 and the controller 46. As the
transmitter senses the system cycling, the integrator will act on
the output. As long as the cycling is of a se-t frequency, the
integrator will integrate the fluctuations out. The integral of
a symetrical periodic Eunc-tion is 0. Therefore, -the integrator
output will consist of a steady value only as sensed by the
transmitter. It is important that the frequency of the cycling
be measured on the job site and adjusted into the integra-tor. An
alternate position of the integrator is in the Eeedback loop, as
shown by integrator 16 in Figure 3. This device may be used with
or without filter 44 and integrators 62 and 64.
Controller 30 may also be constructed with a
micro-processor replacing some or all signal processing,
conditioning, filtering, integrating, controlling, feedback, set
point and detecting functions by means of software algori-thms
contained in the micro-processor's memory.
Various modifications of the present invention will be
apparent to those skilled in the art. It will be apparent that
the air conditioning system oE the present invention may be used
to advantage in a single storey building in which case it serves
to overcome the difficulties experienced with lateral rather than
vertical air disturbances.
27
