Вы находитесь на странице: 1из 18

VOLUME 13, NUMBER 4 HVAC&R RESEARCH JULY 2007

581
A General Control Algorithm for
Cooling Towers in Cooling Plants with
Electric and/or Gas-Driven Chillers
James E. Braun, PhD
Fellow ASHRAE
Received September 13, 2006; accepted February 23, 2007
A hybrid chiller plant employs a combination of chillers that are powered by electricity and
natural gas. This paper presents an algorithm for determining cooling tower fan settings in
hybrid plants in response to loadings on individual chillers. Parameters of the algorithm are
evaluated using design information for the chillers and cooling tower fans. In addition to reduc-
ing operating costs, use of the open-loop control strategy simplifies the control and improves the
stability of the tower control compared with the use of a constant condenser water supply or
approach to wet-bulb. Simulated plant cooling costs associated with the algorithm were com-
pared with costs for optimized settings and were within 1% of the minimum costs. The control
method is general, in the sense that it also applies to cooling plants that have all electric or all
natural gas chillers.
INTRODUCTION
A chiller plant often consists of multiple chillers, multiple condenser water pumps, and multi-
ple cooling towers as depicted in Figure 1. In some situations, it is economical to employ a mix
of chillers that are powered by electricity and natural gas (e.g., absorption or engine-driven).
This is termed a hybrid chiller plant. The major advantage of using natural gas chillers in hybrid
central chiller plants is a reduction in peak electrical demand and on-peak energy usage, which
can reduce overall operating costs. The electric demand cost can often account for about half of
the total air-conditioning bill.
Proper control of cooling tower fans in cooling plants can have a significant impact on operat-
ing costs. For all electric plants, optimal control of cooling fans has been shown to result in sig-
nificant (e.g, 5%15%) savings in plant energy costs as compared with typical strategies that are
employed (see Sud [1984], Lau et al. [1985], Hackner et al. [1985], Klein et al. [1988], Braun
[1988], Braun et al. [1989a, 1989b], ASHRAE [2003]). The savings for plants employing
absorption chillers can be even larger because of higher heat rejection requirements. For
instance, Koeppel et al. (1995) simulated optimal control of tower fans and condenser pumps for
a cooling plant having a double-effect absorption chiller and determined a 20% reduction in
costs compared to using fixed speeds with a tower bypass control to maintain a constant cooling
tower water supply.
Although there is a large body of literature related to supervisory control for all-electric plants,
there is very little literature on supervisory control of absorption, engine-driven, and hybrid
chiller plants. Braun and Diderrich (1990) developed an algorithm for cooling tower fan control
for all-electric plants that is included in the 2003 ASHRAE HandbookHVAC Applications
James E. Braun is a professor of mechanical engineering, Ray W. Herrick Laboratories, Purdue University, West
Lafayette, IN.
2007, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in HVAC&R Research, Vol. 13,
No. 4, July 2007. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without
ASHRAEs prior written permission.
582 HVAC&R RESEARCH
(ASHRAE 2003). However, this strategy is not appropriate for hybrid plants because it is based
on minimizing input energy usage rather than cost.
Koeppel et al. (1995) developed a simplified strategy for cooling tower fan control for absorp-
tion cooling plants that involves the determination of a linear relationship between a setpoint for
cooling tower supply water temperature and ambient wet-bulb temperature. Optimization results
were used to determine a simple linear model for a case study involving a single double-effect
absorption chiller. The application of the simple strategy resulted in savings that were nearly
identical to the optimization results. However, its not obvious how linear control relationships
would be determined in practice and how to apply this method to hybrid plants.
This paper develops a general algorithm for control of cooling tower fans for cooling plants
that have any combination of electric and natural-gas chillers. The development follows an
approach that is similar to the development of Braun and Diderrich (1990). The control method
is evaluated through comparisons with optimal control for a range of different cooling plants
and operating conditions using a simulation tool. Optimal control means that the control is based
on minimization of an operating cost function that incorporates perfect information about the
plant. A nonlinear optimization was performed to determine the performance for the benchmark
optimal control using the simulation tool. The development of the near-optimal control algo-
rithm used several simplifying assumptions and heuristics in order to determine analytical
expressions for cooling tower control. In this context, near-optimal control implies that the strat-
egy has performance that is close to that associated with optimal control.
DEVELOPMENT OF A NEAR-OPTIMAL
COOLING TOWER CONTROL ALGORITHM
Chiller plants, such as that depicted in Figure 1, typically have multiple cooling towers with
fans that have multiple speeds of operation. In general, optimal control of cooling tower fans
Figure 1. Schematic of a chiller plant.
VOLUME 13, NUMBER 4, JULY 2007 583
results from a trade-off in the cost of operating the chillers and cooling tower fans. The energy
consumption of a chiller is sensitive to the condenser water temperature, which is affected by the
cooling tower control. Increasing the tower airflow reduces the chiller energy requirement but at
the expense of an increase in fan power consumption. For a given set of conditions, an optimal
tower control exists that minimizes the sum of the chiller and cooling tower fan power.
Braun and Diderrich (1990) described how the determination of optimal tower fan control can
be separated into two parts: tower sequencing and optimal airflow. For a given total tower air-
flow, optimal tower sequencing specifies the number of operating cells and the fan speeds that
give the minimum fan power consumption. Once the tower sequencing is specified, then the
optimal airflow can be determined by analyzing the trade-offs between the costs of operating the
chiller and the fan.
This section presents the development of an algorithm for near-optimal control of cooling
towers that is based upon a combination of heuristic rules for tower sequencing and an
open-loop control equation derived from a detailed analysis.
Optimal Fan Sequencing
Simple relationships exist for the best sequencing of cooling tower fans for towers having
multiple cells as capacity is added or removed. When additional tower capacity is required,
Braun et al. (1989a, 1989b) have shown that in almost all practical cases, the speed of the tower
fan operating at the lowest speed (including fans that are off) should be increased first. Simi-
larly, for removing tower capacity, the highest fan speeds are the first to be reduced. This leads
to the following general rules for sequencing of tower fans:
1. All Variable-Speed Fans: Operate all cells with fans at equal speeds.
2. Multi-Speed Fans: Increment lowest-speed fans first when adding tower capacity. Reverse
for removing capacity.
3. Variable/Multi-Speed Fans: Operate all cells with variable-speed fans at equal speeds.
Increment lowest-speed fans first when adding tower capacity with multi-speed fans. Add
multi-speed fan capacity when variable-speed fan speeds match the fan speed associated with
the next multi-speed fan increment to be added.
Criteria for Optimal Tower Airflow
Most cooling towers utilize single- or two-speed fans, such that the optimization problem is
discrete rather than continuous. However, for the purpose of estimating the control parameters,
it is sufficient to consider the flow as being continuously adjustable. Consider the problem of
determining the optimal tower control for continuously adjustable tower airflow. The minimum
combined chiller and tower fan cost occurs at a point where the rate of change of cost with
respect to changes in tower airflow is equal to zero, or
(1)
where C
twr
and C
ch
are instantaneous energy costs for the cooling tower fans and chillers and

twr
is the relative tower airflow defined as ratio of the airflow to the maximum possible airflow
with all cells operating at maximum speed. In order to solve for the optimal tower control, it is
necessary to develop a functional relationship for the sensitivities of chiller and cooling tower
costs to tower airflow.
dC
twr
d
twr
--------------
dC
ch
d
twr
------------- , =
584 HVAC&R RESEARCH
Chiller Cost Sensitivity to Tower Airflow
The rate of change of chiller cost with respect to tower airflow may be expressed as
(2)
where T
cwr
is the condenser water return temperature. The rate of change of total chiller cost
with respect to condenser water return temperature is
(3)
where E
ch,i
is the rate of energy input and ER
i
is cost per unit energy input (heat or electricity)
for the ith chiller. The rate of change of energy input with respect to condenser water tempera-
ture can be written as
(4)
where COP is the chiller coefficient of performance (ratio of cooling to energy input), COP
rated
is the chiller COP at the chiller design conditions, PLR
ch
is the chiller part-load ratio (load rela-
tive to rated capacity), and S
ch
is the sensitivity of the chiller input energy to changes in con-
denser water temperature given as
(5)
For the purpose of determining tower airflow, it is reasonable to assume that the factor
S
ch
(COP
rated
/COP) is constant for a given chiller over different operating conditions. With this
assumption, Equation 4 can be simplified to
(6)
where S
ch,rated,i
is evaluated at the chiller design conditions.
The next step is to develop a relationship for the effect of the tower control on the condenser
water return temperature. From an effectiveness model for the thermal performance of a cooling
tower (Braun et al. 1989c), the condenser water return temperature may be expressed as
(7)
where T
wb
is ambient wet-bulb temperature, Q
twr
is the cooling tower heat rejection rate, m
a,twr
is
the tower air mass flow rate,
a,twr
is the an air-side effectiveness for heat and mass transfer within
the tower, and c
s
is a property of the air-water vapor mixture termed the saturation specific heat.
Equation 7 is simplified by relating the performance to design conditions, assuming that the
tower effectiveness and saturation specific heat are constants and utilizing the definitions for rel-
ative tower airflow, and tower approach and range.
dC
ch
d
twr
-------------
dC
ch
dT
cwr
--------------
dT
cwr
d
twr
-------------- , =
dC
ch
dT
cwr
-------------- ER
i
dE
ch,i
dT
cwr
--------------- ,
i 1 =
N
ch

=
dE
ch, i
dT
cwr
--------------- S
ch, i
COP
rated, i
COP
i
-------------------------- PLR
ch, i
E
ch, rated, i
, =
S
ch, i
1
E
ch, i
-----------
dE
ch, i
dT
cwr
--------------- . =
dE
ch, i
dT
cwr
--------------- S
ch, rated, i
PLR
ch, i
E
ch,rated, i
, =
T
cwr
T
wb
Q
twr

a, twr
m
a, twr
c
s
----------------------------------- , +
VOLUME 13, NUMBER 4, JULY 2007 585
(8)
where a
twr,rated
and r
twr,rated
are the tower approach and range evaluated at the rating condition
with the tower operating with maximum tower airflow at the chiller design load. The approach is
the difference between the condenser water supply and ambient wet-bulb temperatures
(T
cws
T
wb
), whereas the range is the difference between the condenser water return and supply
(T
cwr
T
cws
).
The ratio of the tower heat rejection to the design heat rejection can be approximated as
(9)
where
(10)
and where Q
cw,rated,i
is the heat rejection from ith chiller to the condenser water at the rating
condition.
Equations 8 and 9 lead to
(11)
Combining Equations 2, 3, 6, and 11 results in
(12)
Cooling Tower Fan Cost Sensitivity to Tower Airflow
The sensitivity of the tower fan costs to changes in control depends upon the type of fans and
motors employed. For variable-speed drives and with each tower cell operating at equal flows,
the fan power varies approximately with the cube of the airflow. For multi-speed fans and with
an optimal sequencing strategy, the fan power varies as a piecewise linear function that
approaches the variable-speed relationship as the number of speed settings increases. For sin-
gle-speed fans, the fan power increases as a single linear function of the total tower airflow.
For a system with discrete tower fan settings, it is adequate to assume a continuous variation
in order to perform the optimization and determine the discrete control that is closest to the
T
cwr
T
wb
Q
twr

a, twr
m
a, twr
c
s
-----------------------------------

a, twr, rated
m
a, twr, rated
c
s,rated
T
cwr, rated
T
wb,rated
( )
Q
twr, rated
------------------------------------------------------------------------------------------------------------------------------------ + =
T
wb
Q
twr
m
a, twr
m
a,twr, rated
---------------------------- Q
twr,rated
--------------------------------------------------------

a, twr,rated
c
s, rated

a, twr
c
s
--------------------------------------------- a
twr,rated
r
twr,rated
+ ( ) + =
T
wb
1

twr
---------
Q
twr
Q
twr,rated
------------------------ a
twr,rated
r
twr, rated
+ ( ) +
Q
twr
Q
twr,rated
------------------------ PLR
ch, i
i 1 =
N
ch

f
twr, rated,i
, =
f
twr, rated,i
Q
cw, rated, i
Q
twr, rated
-------------------------- =
dT
cwr
d
twr
--------------
1

twr
2
--------- a
twr, rated
r
twr, rated
+ ( ) PLR
ch,i
i 1 =
N
ch

f
twr, rated, i
. =
dC
ch
d
twr
-------------
1

twr
2
--------- a
twr, rated
r
twr, rated
+ ( ) PLR
ch, i
f
twr, rated, i
( )
i 1 =
N
ch

PLR
ch, i
S
ch, rated i ,
ER
i
E
ch,rated,i
( ) .
i 1 =
N
ch

=
586 HVAC&R RESEARCH
estimated control. For single-speed fans, the power is assumed to be a continuous linear func-
tion. For two-speed fans, performance data suggest that a squared relationship for power con-
sumption as a function of airflow is adequate. For three-speed (or more) fans, a cubic variation
in power consumption with airflow is sufficient.
Thus, for variable-speed or three-speed fans, the sensitivity of the fan power consumption to
changes in tower control is assumed to be
(13)
while for two-speed fans,
(14)
and for single-speed fans,
(15)
where ER
e
is cost per unit electricity input and P
twr,rated
is the rated fan power with the fans run-
ning aT capacity.
Optimal Tower Airflow
Substituting Equations 12 and 13 into Equation 1 and solving for
twr
gives the following
relationship for near-optimal control of cooling towers with three-speed or variable-speed fans.
(16)
For two-speed fans, Equation 14 is used instead of Equation 13 and the near-optimal control
equation becomes
(17)
while for single-speed fans,
(18)
dC
twr
d
twr
-------------- 3
twr
2
ER
e
P
twr, rated
, =
dC
twr
d
twr
-------------- 2
twr
ER
e
P
twr, rated
, =
dC
twr
d
twr
-------------- ER
e
P
twr, rated
, =

twr
1
3
--- a
twr,rated
r
twr,rated
+ ( ) PLR
ch,i
f
twr,rated, i
( )
i 1 =
N
ch

PLR
ch,i
S
ch, rated, i

ER
i
E
ch, rated, i
ER
e
P
twr, rated
----------------------------------
\ `
| |
i 1 =
N
ch

)

`


1 4
=

twr
1
2
--- a
twr, rated
r
twr,rated
+ ( ) PLR
ch, i
f
twr,rated,i
( )
i 1 =
N
ch

PLR
ch,i
S
ch, rated, i

ER
i
E
ch, rated, i
ER
e
P
twr, rated
----------------------------------
\ `
| |
i 1 =
N
ch

)

`


1 3
, =

twr
a
twr, rated
r
twr, rated
+ ( ) PLR
ch, i
f
twr, rated,i
( )
i 1 =
N
ch

PLR
ch, i
S
ch,rated,i

ER
i
E
ch, rated,i
ER
e
P
twr,rated
----------------------------------
\ `
| |
i 1 =
N
ch

)

`


1 2
. =
VOLUME 13, NUMBER 4, JULY 2007 587
The relative tower airflow is the ratio of airflow to the rated airflow with all fans operating at
maximum speed. The factors that affect the optimal relative tower airflow are (1) the part-load
ratio for each chiller, PLR
ch,i
; (2) the ratio of the operating costs per unit time for each chiller at
its rating conditions to the rated cooling tower operating costs per unit time,
(ER
i
E
ch,rated,i
)/(ER
e
P
twr,rated
); (3) the sensitivity of individual chiller input energy requirement
to changes in condenser water temperature at the chiller rating conditions, S
ch,rated,i
; (4) the ratio
of the ith chiller heat rejection rate at its rating condition to the rated cooling tower heat rejection
rate, f
twr,rated,i
; and (5) the sum of the tower approach and range at the tower rating condition
(a
twr,rated
+ r
twr,rated
). All of the factors in these expressions, except chiller part-load ratio and
utility rates, are based on design information and do not require measurements.
The design approach to wet-bulb, a
twr,rated
, is the temperature difference between the tower
sump water and the ambient wet-bulb for the tower operating at its air and water flow capacity at
the tower rating conditions. The rated range, r
twr,rated
, is the water temperature difference across
the tower at these same conditions. The sum of a
twr,rated
and r
twr,rated
is the temperature differ-
ence between the tower inlet and the wet-bulb and represents a measure of the towers capability
to reject heat to ambient relative to the system requirements. A small temperature difference
results from a high tower heat transfer effectiveness or high water flow rate and yields lower
condenser water temperatures with lower chiller energy consumption, resulting in a lower opti-
mal tower airflow. Typical values for the design approach and range are 7F and 10F.
Chiller part-load ratio measures the load for each chiller and influences the optimal tower air-
flow in two ways. First, chiller loading influences the total heat rejection requirements of the
cooling tower, which affects the optimal airflow. The optimal tower airflow increases with heat
rejection requirement. The first summation on the right-hand side of Equations 1618 character-
izes this effect. The factor f
twr,rated,i
weights the individual chiller loadings according to their
effect on the total heat rejection and is larger for chillers having greater design cooling capacity
or lower design COP (e.g., absorption).
The chiller loading also influences the chiller to cooling tower cost ratio along with the cost
ratio at the rating conditions. As the ratio of chiller to tower cost increases, it becomes more ben-
eficial to operate the tower at higher airflows. If the tower airflow were free, then the best strat-
egy would be to operate the towers at full capacity independent of the load. The cost ratio is
typically higher for hybrid cooling plants than for all-electric plants because of lower chiller
COPs and higher heat rejection requirements for gas-driven chillers.
The chiller sensitivity factor, S
ch
, is the incremental increase in chiller operating cost for each
degree increase in condenser water temperature as a fraction of the power, or
(19)
If the chiller input energy rate increases by 2% for a 1 degree increase in condenser water tem-
perature, then S is equal to 0.02. A large sensitivity factor means that the chiller input energy
rate is very sensitive to the cooling tower control, favoring operation at higher airflows. The sen-
sitivity factor should be evaluated at design conditions using chiller performance data. Typi-
cally, the sensitivity factor is between 0.01 and 0.03F
1
.
PERFORMANCE EVALUATION
The performance of the near-optimal control algorithm was evaluated through comparisons
with optimization results for typical hybrid cooling plants that were simulated using a tool
described by Braun (2006). A brief review of the modeling approach and plant characteristics is
provided in this section along with the performance evaluation.
S
ch
change in chiller input energy rate ( )
change in condenser water temperature ( ) chiller input energy rate ( )
--------------------------------------------------------------------------------------------------------------------------------------------------------------------- . =
588 HVAC&R RESEARCH
Chiller Plant Modeling
The cooling plant simulation tool neglects energy storage effects and determines hourly costs
of providing cooling for a specified combination of chillers, condenser water pumps, and cool-
ing towers. For any given hour, the cost of energy associated with operating the chiller plant is
(20)
where P
twr
is cooling tower fan power, P
cwp
is condenser water pump power, E
ch,i
is the rate of
energy usage for the ith chiller (gas or electric), N
ch
is the total number of chillers (gas and elec-
tric), ER
e
is the cost per unit of electrical energy, and ER
i
is the cost per unit energy input (heat
or electricity) for the ith chiller (gas or electric).
The rate of energy consumption for the ith chiller is determined as
(21)
where
ch,i
is a control variable (0 or 1) that indicates whether the chiller is operating or not,
Q
ch,rated,i
is the rated chiller capacity, COP
rated,i
is the rated chiller COP, and PLF
i
is the
part-load factor, defined as the ratio of energy usage to the value at the rating condition.
The chilled-water supply temperature and condenser water flow rate are assumed to be con-
stant, and the chiller part-load factor is correlated in terms of the chiller load and entering con-
denser water supply temperature. Figures 25 show part-load factors (PLF) as a function of
chiller part-load ratio (PLR) and entering condenser water temperature (T
cws
) for the four differ-
ent chillers considered within the simulation tool: (1) electric centrifugal chiller with vari-
able-speed motor, (2) electric centrifugal chiller with fixed-speed motor and inlet guide vane
capacity control, (3) single-effect absorption chiller, and (4) engine-driven centrifugal chiller.
These figures were generated by using correlations to manufacturers data. The PLR is the ratio
of chiller load to a rated capacity. Not surprisingly, the variable-speed electric chiller has the
best part-load performance, whereas the fixed-speed electric has the worst.
In order to simulate a cooling plant, rated cooling capacities (Q
ch,rated
) and COPs (COP
rated
)
are specified for each chiller. The model assumes that all of the energy input to the chillers is
rejected to the condenser water loop. The temperature of water leaving the condenser is deter-
mined from an energy balance on the condenser.
Individual condenser pumps are assumed to be dedicated to individual chillers and to provide
rated flow rates. The total condenser flow rate and pump power are simply the sums of the rated
values for the chillers that are operating. Water flows and power are scaled with chiller cooling
capacity according to user-specified rating values for volumetric flow rate per unit rated cooling
capacity (gpm/ton or L/skW) and pump power per unit volumetric flow rate (W/gpm or Ws/L).
An effectiveness model presented by Braun et al. (1989c) is used to represent the performance
of cooling tower cells. The model considers the impact of air and water flow rates on heat and
mass transfer rates using characteristics that are representative of commercial cooling towers.
For simplicity in developing and evaluating control strategy heuristics, the cooling tower is
modeled as a single cell having continuously variable airflow rate. The performance of a cooling
tower that has multiple cells that are operating identically (e.g., same flows) is equivalent to the
performance of a single larger cooling tower having the same total water and airflow rates.
There are two limiting cases that are considered for estimating the cooling tower fan power:
(1) variable-speed fans and (2) single-speed staged fans. These two cases represent the upper
C
pl
P
twr
P
cwp
+ ( )ER
e
E
ch, i
ER
i
,
i 1 =
N
ch

+ =
E
ch,i

ch, i
PLF
i
Q
ch,rated, i
COP
rated, i
-------------------------- , =
VOLUME 13, NUMBER 4, JULY 2007 589

Figure 2. Part-load performance of variable-speed, electric-driven centrifugal chiller.
Figure 3. Part-load performance of fixed-speed, electric-driven centrifugal chiller.
590 HVAC&R RESEARCH
Figure 4. Part-load performance of engine-driven chiller.
Figure 5. Part-load performance of absorption chiller.
VOLUME 13, NUMBER 4, JULY 2007 591
and lower bounds for performance associated with a particular tower design. For variable-speed
fans, the fan laws are employed and the power varies with the cube of the airflow. This charac-
terizes the behavior of either a single large tower with a variable-speed fan or multiple smaller
tower cells with variable-speed fans all operating at the same speeds. For staged fans, it is
assumed that the fan power varies linearly with airflow. This characterizes the behavior of mul-
tiple tower cells having single-speed fans that are staged on and off.
The cooling tower size and design fan airflow rates and power consumption are scaled
according to the plant heat rejection requirements. Rated airflows and power are specified in
terms of volumetric airflow rate per unit condenser heat rejection rate at design (cfm/ton or
L/skW) and fan power per unit volumetric flow rate (W/cfm or Ws/L).
For given cooling load, ambient air conditions, and plant control variables, the cooling plant
model iteratively determines the condenser entering and leaving water temperatures that balance
condenser and tower heat rejection. The primary ambient condition that influences plant cooling
performance is the air wet-bulb temperature, whereas the dry-bulb temperature has a minor
effect. The control variables are the sequencing and relative loadings for individual chillers and
the relative tower fan speed or airflow.
Overall Plant Characteristics
Table 1 gives parameters used in system simulations to evaluate the performance of the
near-optimal tower control algorithm. The design ambient conditions, performance ratings, and
embedded component performance characteristics were used along with a specification of the
number, type, and cooling capacities of chillers to determine the cooling tower design airflow
and fan power requirements. For all cases considered in this study, two 500-ton chillers were
employed. All six possible combinations of two different chillers were chosen from the four
available types: (1) fixed- and variable-speed electric, (2) fixed-speed electric and absorption,
(3) fixed-speed electric and engine, (4) variable-speed electric and absorption, (5) vari-
able-speed electric and engine, and (6) absorption and engine. For each of these chiller combina-
tions, all other combinations of cooling tower fan types, utility rates, plant part-load ratios (ratio
of load to total chiller capacity), and ambient conditions given in Table 1 were considered, lead-
ing to 768 operating points (6 2 2 2 4 4) for evaluation of the control algorithm.
Figure 6 shows the effect of tower airflow and electric chiller part-load ratio on plant energy
costs per ton of cooling provided for a plant that utilizes equally sized electric (variable-speed)
and absorption chillers with an overall plant part-load ratio of 0.5 at the design ambient condi-
tions (T
wb
= 80F, T
db
= 95F). Lines of constant plant costs are shown over the range of about
0.0.085 to 0.125 $/ton for the entire range of tower airflow (20% to 100% of the design tower
airflow) and electric chiller part-load ratios (0.1 to 0.8).
Figure 6 indicates that the optimal tower airflow for this operating condition is about 50% of
the design airflow and relatively independent of the relative chiller loadings. Greater airflow
results in improved chiller performance but at the expense of increased fan power. Lower air-
flow results in lower fan power but with higher chiller energy input requirements. The optimum
results from these trade-offs, which depends on the total plant load. The penalty associated with
operating the cooling tower at the design airflow as compared with the optimal flow would be
about 15% for these operating conditions if the electric chiller is heavily loaded. Although the
penalty is much smaller if the absorption chiller loading is maximized, the optimal chiller policy
for this case is to maximize loading on the electric chiller.
Benchmark Comparisons
The performance of the cooling tower control algorithm was evaluated using the simulation
tool. Simulated plant costs for the simple control algorithm were compared with those for
592 HVAC&R RESEARCH
optimal tower airflow. The benchmark optimal results were determined by minimizing the
plant costs of Equation 20 with respect to tower airflow using a one-dimensional golden-sec-
tion search algorithm applied to the plant model. The comparisons were performed for the
systems and operating conditions described in the previous section (Table 1) and assuming
that the multiple chillers were loaded evenly. The comparisons were insensitive to whether
the chillers were loaded evenly or not.
Figure 7 shows comparisons between the cooling plant costs for optimal and near-optimal
control when applied to the hybrid, all-electric, and all-gas-driven chiller plants. Overall, the
near-optimal control algorithm gives performance that is within 1% of the minimum power con-
sumption. The method worked extremely well in all cases considered.
CONTROL STRATEGY IMPLEMENTATION
The cooling tower control algorithm involves determination of a relative tower airflow using
Equations 16, 17, or 18. The different equations are for different fan types and give tower air-
flow relative to the maximum tower airflow if all tower fans were operating at the highest fan
speed. As a result, the maximum relative airflow must be constrained to 1. There are additional
constraints on the temperature of the supply water to the chiller condensers, which are necessary
Table 1. Parameters for Hybrid Cooling Plant Simulation Studies
Parameter Value
Design conditions
Wet-bulb temperature
Dry-bulb temperature
80F
95F
Electric chillers
Motor
Rated COP
Condenser-water flow rate
Condenser-water pump power
Fixed or variable-speed
6.0
3 gpm/ton
15 W/gpm
Absorption chiller
Rated COP
Condenser-water flow rate
Condenser-water pump power
1.0
4 gpm/ton
15 W/gpm
Engine-driven chiller
Rated COP
Condenser-water flow rate
Condenser-water pump power
1.5
3 gpm/ton
15 W/gpm
Cooling tower
Motor
Rated airflow
Rated fan power
Staged or variable-speed
200 cfm/ton
0.4 W/cfm
Utility costs
Electrical energy
Gas energy
0.05 or 0.15 $/kWh
0.40 or 1.20 $/therm
Plant operating conditions
Plant part-load ratio
Wet-bulb temperature
Dry-bulb/wet-bulb difference
0.25, 0.5, 0.75, 1.0
50F, 60F, 70F, 80F
15F
VOLUME 13, NUMBER 4, JULY 2007 593
to avoid potential chiller maintenance problems. Some chillers have a low limit on the con-
denser water supply temperature that is necessary to avoid lubrication migration from the com-
pressor. A high temperature limit is also necessary to avoid excessively high pressures within
the condenser, which can lead to compressor surge. If the condenser water temperature falls
below the low limit, then it is necessary to override the open-loop tower control and reduce the
tower airflow to go above this limit. Similarly, if the high limit is exceeded, then the tower air-
flow should be increased as required.
At each decision interval (e.g., five minutes), the following steps can be executed to deter-
mine the setpoint for the relative cooling tower airflow:
1. Evaluate the time-averaged values of the condenser water supply temperature and overall
plant cooling load over a fixed time interval (e.g., five minutes).
2. If the condenser water supply temperature is less than the low limit, then reduce the setpoint
for the relative tower airflow by a fixed increment and exit the algorithm. Otherwise go to
step 3. For fans with discrete settings, the setpoint increment should be chosen so that a single
fan changes speed by a single step.
3. If the condenser water supply temperature is greater than the high limit, then increase the set-
point for the relative tower airflow by a fixed increment and exit the algorithm. Otherwise go
to step 4.
4. If the chilled-water load has changed by a significant amount (e.g., 10%) since the last con-
trol change, then go to step 5. Otherwise exit the algorithm.
5. Use the current measured cooling load to estimate the PLR (load relative to rated capacity)
for each chiller.
Figure 6. Contours for plant energy costs per unit of cooling ($/ton-h) as a function of
the electric chiller part-load ratio and relative cooling tower airflow (equally sized vari-
able-speed electric and absorption chillers, plant part-load ratio of 0.5, T
wb
= 80F,
T
db
= 95F, variable-speed cooling tower fans, 0.10 $/kWh, 0.80 $/therm).
594 HVAC&R RESEARCH
6. Use individual chiller part-load ratios, performance information at rating conditions for the
chillers and cooling tower, and utility energy rates for the current rate period to determine a
relative cooling tower airflow setpoint using Equation 16, 17, or 18.
The relative tower airflow must be converted to a specific set of tower fan settings. The total
cooling tower airflow is approximately linear with individual tower cell fan speed and the num-
ber of fans operating, such that
(22)
where
twr,i
is the fan speed control function (0 to 1) for the ith cooling tower cell fan and N
twr
is
the number of cooling tower cells.
Equation 22 can be used to determine the required fan settings for a given total relative air-
flow. However, first it is necessary to have specific sequencing rules for the order in which fans
should be turned on and off. As previously discussed, the best fan settings for a given airflow
result from operating the maximum number of fans at the lowest possible speeds. In other
words, the fans operating at the lowest speeds should be incremented first when adding fan
capacity. For fans with discrete speed settings, this means that all tower fans should be operating
at low speed before any fan speeds are increased to the next speed increment. Fan capacity
should be removed in the reverse order in which it was added.
The process for converting from relative tower airflow setpoint to specific fan settings
depends on the type of fan as described below:
1. Fans with Discrete Speed Settings. For multiple cells having fans with discrete speed set-
tings (e.g., two-speed), Equation 22 and the sequencing rules can be used to construct a table
Figure 7. Comparison of near-optimal and optimal cooling tower costs.

twr
1
N
twr
-----------
twr,i
,
i 1 =
N
twr

=
VOLUME 13, NUMBER 4, JULY 2007 595
relating airflow to fan-speed settings for individual cooling tower cells. The minimum tower
airflow would be that associated with a single cell operating at its minimum fan setting (e.g.,
half-speed). The next increment of airflow would be associated with two cells operating at
their minimum speed and so on. The conversion process between a relative airflow setpoint
and the discrete fan control involves choosing the set of discrete fan settings from the table
that produces a tower airflow closest to the desired flow, assuming that airflow is propor-
tional to fan speed. In general, it is better to have greater rather than less than the optimal air-
flow. A good rule of thumb is to choose the set of discrete fan controls that results in a
relative airflow that is closest to, but not more than 10% less than, the target relative airflow.
2. Variable-Speed Fans. If the cooling tower cells have only variable-speed fans, then all oper-
ating fans should be set at identical speeds. If possible, all tower cells should be operated
unless the required fan speed would fall below a minimum allowable setting (e.g., 0.2).
According to Equation 22, the fan-speed setting if all cells are operating is equal to the rela-
tive airflow setpoint. If the calculated setting for this case is below the minimum allowable
setting, then the number of operating cells should be reduced by one and Equation 22 should
be used to determine the required setting. This process should be repeated as necessary until
the calculated fan setting is above the minimum allowable.
3. Discrete and Variable-Speed Fans. If the cooling tower cells have a mix of fans with vari-
able-speed and discrete settings, then the goal is to operate the fans as close to the same speed
as possible. For the discrete fans, Equation 22 and the sequencing rules can be used to con-
struct a table relating airflow to fan-speed settings for individual cooling tower cells (see
step 1). The first entry in the table should be for all discrete fans turned off. Then, for each
entry in the table, the required fan control settings for the variable-speed fans that give the
required relative airflow are determined using Equation 22, assuming that all the vari-
able-speed fans operate at the same speed (see step 2). Then, the fan settings are selected
from the table using the criteria that the variable-speed setting should be closest to, but less
than, the maximum discrete fan setting.
COOLING TOWER CONTROL EXAMPLE
Consider an example plant consisting of one variable-speed electric and one engine-driven
chiller, each having a rated cooling capacity of 500 tons. The rated COP for the electric chiller
is 6, resulting in a rated electrical power input requirement of
The rated COP for the engine-driven chiller is 1.5. Then, the rated requirement for energy
input to the gas chiller is
The cooling tower has two cells, each having two-speed fans. The tower design approach and
range from manufacturers data are 7F and 10F. The rated airflow and fan power are
200 cfm/ton and 0.4 W/cfm. Therefore, the fan power for the total rated plant cooling capacity is
E
ch, rated,E 1
Q
ch, rated,E 1
COP
rated, E 1
----------------------------------
500 tons
6
-------------------- 3.517
kW
ton
-------- 293.1 kW . = = =
E
ch, rated,G 1
Q
ch,rated, E 1
COP
rated, E 1
----------------------------------
500 tons
1.5
-------------------- 3.517
kW
ton
-------- 1172 kW . = = =
P
twr,rated
200
cfm
ton
--------- 0.0004
kW
cfm
--------- 1000 tons 80 kW . = =
596 HVAC&R RESEARCH
From energy balances on the chillers operating at rated capacities and COPs, the fraction of
the cooling tower heat rejection associated with each chiller is:
From the characteristics given in Figures 2 and 4, it is found that the rated sensitivity factor
for chiller input energy with respect to changes in condenser water temperature is about
0.015F
1
for both chiller types. This is a relatively typical value for chillers.
Consider a time during the on-peak period where the total plant load is 700 tons, electrical
energy costs are $0.1/kWh, and gas energy costs are $0.80/therm. If the chillers are equally
loaded, then both chillers have a part-load ratio of 0.7 under these conditions. For two-speed
fans, the relative tower airflow is then determined using Equation 17 so that
At this load, the cooling tower fans should operate at approximately 78% of the maximum
tower airflow. In order to convert
twr
into a specific tower control, it is necessary to define the
tower sequencing. Table 2 gives this information in a form that specifies the relationship
between
twr
and tower control for this example.
For a specific chilled-water load, the fan control should be the sequence of tower fan settings
from Table 2 that results in a value of
twr
that is closest to, but not more than, 10% less than the
value computed with Equation 17. For this example, sequence number 3 would represent the
best choice.
CONCLUSIONS
The algorithm developed in this paper is similar in nature to the approach developed by Braun
and Diderrich (1990) for all-electric plants that appears in the 2003 ASHRAE HandbookHVAC
Applications (ASHRAE 2003). However, the current algorithm is general for cooling plants that
f
twr,rated, E 1
Q
ch, rated,E 1
1
1
COP
rated, E 1
---------------------------------- +
\ `
| |
Q
ch,rated, E 1
1
1
COP
rated, E 1
---------------------------------- +
\ `
| |
Q
ch, rated,G 1
1
1
COP
rated, G 1
----------------------------------- +
\ `
| |
+
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- =
500 1
1
6
--- +
\ `
| |
500 1
1
6
--- +
\ `
| |
500 1
1
1.5
------- +
\ `
| |
+
----------------------------------------------------------------- 0.412 = =
f
twr, rated,G 1
1 f
twr, rated,E 1
0.588 = =

twr
1
2
--- a
twr,rated
r
twr, rated
+ ( ) PLR
ch, i
f
twr, rated, i
( )
i 1 =
N
ch

\ `
]
| |
PLR
ch,i
S
ch, rated, i

ER
i
E
ch,rated, i
ER
e
P
twr, rated
----------------------------------
\ `
| |
i 1 =
N
ch

)

`


1 3
=
1
2
--- 7F 10F + ( ) 0.7 0.411 0.7 + 0.588 ( )
)
`

1 3
=

0.7 0.015
0.1
$
kWh
----------- 80 kW
-------------------------------------------- 0.1
$
kWh
----------- 293.1 kW 0.80
$
therm
-------------- + 0.03412
therm
kWh
-------------- 1172 kW
\ `
| |
)

`


1 3

0.784 . =
VOLUME 13, NUMBER 4, JULY 2007 597
incorporate any combination of electric and gas-driven chillers. The control algorithm deter-
mines cooling tower fan settings in response to loadings on individual chillers. Parameters of the
algorithm are evaluated using design information for the chillers and cooling tower fans. In
addition to reducing operating costs, use of the open-loop control strategy simplifies the control
and improves the stability of the tower control compared with the use of a constant condenser
water supply or approach to wet-bulb.
ACKNOWLEDGMENTS
The financial support of ASHRAE under RP-1200 and the technical support provided by the
Project Monitoring Subcommittee that included Jay Kohler, Paul Sarkisian, and Dharam Pun-
wani are greatly appreciated.
NOMENCLATURE
a
twr
= cooling tower approach to wet-bulb
temperature (T
cws
T
wb
)
C
ch
= instantaneous cost of operating
chillers
C
twr
= instantaneous cost of operating cool-
ing tower
COP
i
= coefficient of performance for the ith
chiller
E
ch,i
= rate of energy input for ith chiller (gas
or electric)
ER
e
= cost per unit of electrical energy
ER
i
= cost per unit energy input (heat or
electricity) for the ith chiller (gas or
electric)
f
twr,i
= fraction of cooling tower heat rejec-
tion associated with ith chiller
N
ch
= total number of chillers (gas and
electric)
N
twr
= number of cooling tower cells
P
cwp
= total condenser water pump power
P
twr
= total cooling tower fan power
PLF
i
= part-load factor for ith chiller defined
as the ratio of energy usage to the
value at the rating condition
PLR
ch,i
= part-load ratio for ith chiller (gas or
electric)load relative to rated chiller
capacity
PLR
plt
= part-load ratio, planttotal cooling
load relative to rated cooling capacity
for plant
Q
ch,i
= cooling load for ith chiller
Q
twr
= cooling tower heat rejection rate
r
twr
= cooling tower range (T
cwr
T
cws
)
S
ch, i
= sensitivity of ith chiller input energy
to changes in condenser water tem-
perature
T
cwr
= condenser water return temperature
T
cws
= condenser water supply temperature
T
db
= ambient dry-bulb temperature
T
wb
= ambient wet-bulb temperature

ch,i
= control function that determines
whether the ith chiller (gas or electric)
is on or off (0 or 1)

twr
= control function that specifies the rela-
tive airflow for the cooling tower

twr,i
= control function that specifies the rela-
tive fan speed or airflow for an indi-
vidual cooling tower cell
Additional Subscript
rated = evaluated at rating conditions
Table 2. Example Cooling Tower Fan Sequencing
Sequence Number
twr
Tower Fan Speeds
Cell #1 Cell #2
1 0.25 Low Off
2 0.50 Low Low
3 0.75 High Low
4 1.00 High High
598 HVAC&R RESEARCH
REFERENCES
ASHRAE. 2003. 2003 ASHRAE HandbookHVAC Applications, Chapter 41, Supervisory Control Strat-
egies and Optimization, pp. 41.141.39. Atlanta: American Society of Heating, Refrigerating and
Air-Conditioning Engineers, Inc.
Braun, J.E. 1988. Methodologies for the design and control of central cooling plants. PhD dissertation,
University of WisconsinMadison.
Braun, J.E., S.A. Klein, J.W. Mitchell, and W.A. Beckman. 1989a. Applications of optimal control to
chilled-water systems without storage. ASHRAE Transactions 95(1):66375.
Braun, J.E., S.A. Klein, J.W. Mitchell, and W.A. Beckman. 1989b. Methodologies for optimal control of
chilled-water systems without storage. ASHRAE Transactions 95(1):65262.
Braun, J.E., S.A. Klein, and J.W. Mitchell. 1989c. Effectiveness models for cooling towers and cooling
coils. ASHRAE Transactions 95(2):16474.
Braun, J.E., and G.T. Diderrich. 1990. Near-optimal control of cooling towers for chilled-water systems.
ASHRAE Transactions 96(2):80613.
Braun, J.E. 2006. Optimized operation of chiller equipment in hybrid machinery rooms and associated
operating and control strategies, ASHRAE RP-1200 final report. Atlanta: American Society of Heat-
ing, Refrigerating and Air-Conditioning Engineers, Inc.
Hackner, R.J., J.W. Mitchell, and W.A. Beckman. 1985. HVAC system dynamics and energy use in build-
ingsPart II (RP-321). ASHRAE Transactions 91(1B):78195.
Klein, S.A., D.R. Nugent, and W.A. Beckman. 1988. Investigation of control alternatives for a steam tur-
bine driven chiller. ASHRAE Transactions 94(1):62743.
Koeppel, E.A., J.W. Mitchell, S.A. Klein, and B.A. Flake. 1995. Optimal supervisory control of an absorp-
tion chiller system. HVAC&R Research 1(4):32542.
Lau, A.S., W.A. Beckman, and J.W. Mitchell. 1985. Development of computer controlRoutines for a
large chilled-water plant. ASHRAE Transactions 91(1B):76680.
Sud, I. 1984. Control strategies for minimum energy usage (RP-253). ASHRAE Transactions
90(2A):24777.

Вам также может понравиться