Avery 112

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

ASHRAE Journal

Chillers

Improving the Efficiency


Of Chilled Water Plants
and return water through the bypass
starts to mix with the supply water. This
raises the supply water temperature and
the air-handling unit control valves start
to open, thus increasing both the secondary and bypass flow. Temperature
control in the conditioned spaces is lost,
even though only about half of the central plant capacity is used. When chillers
are staged from flow and are controlled at
the design leaving chilled water temperature, the approximate maximum system
capacity is:

By Gil Avery, P.E.


Fellow ASHRAE

or the past 30 or more years most central chilled water plants have
used primary-secondary pumping arrangements as shown in
Figure 1. The primary system includes all equipment to the plant side
of Points A and B and the secondary system includes all equipment to the
load side of these points. The majority of these systems have the chillers
staged from flow. As the flow in the secondary starts to exceed the flow in
the primary, an additional primary pump and chiller is staged on. As the flow
(Operating T ) (Design Capacity )
in the secondary decreases, so that one less chiller can handle the reduced
Design T
flow, a pump and chiller is staged off. This method of control assumes that
Very often facilities with operating conthe load is proportional to flow and that the T between the chilled water ditions such as these will add an additional
chiller and pump even though the existing
supply and return is constant and at the design condition.

These assumptions are generally


wrong. Systems rarely operate at design
T because of coils that are dirty, inadequate coil control valves, thermostats
that are set wrong, excessive temperature
drops in the supply and return mains, etc.
Therefore, flow is not a reliable indication of load. As a result, the vast majority
of central chilled water plants operate inefficiently with lower than design Ts and
with more partially loaded chillers on-line
than would be necessary if the chillers
were properly staged.
The Way Its Been Done
The typical primary/secondary system
shown in Figure 1 is fully loaded at the
outdoor design conditions of 95F (35C)
dry bulb and 78F (25.5C) wet bulb. The
system is operating as designed with
6,000 gpm (379 L/s) of 45F (7C) supply
water and 57F (13.8C) return water (2
gpm/ton [0.03 mL/J]). Each of the three
1,000 ton (3517 kW) chillers and companion pumps are handling 2,000 gpm
(126 L/s). The fourth standby pump is off.
14

ASHRAE Journal

In Figure 2 the load is only 600 tons


(2110 kW). The T is 6F (3.3C) (4
gpm/ton [0.07 mL/J]), which is typical of
many systems designed to operate at a
higher T. Two pumps and chillers are
on-line, with each chiller inefficiently
loaded to 300 tons (1055 kW); 4,000 gpm
(252 L/s) is circulated in the primary, 1,600
gpm (101 L/s) through the bypass and
2,400 gpm (151 L/s) (600 tons 4 gpm/ton
[2110 kW 0.07 mL/J) in the secondary. If
the system were properly staged and
piped, one chiller could easily handle the
600 ton (2110 kW) load.
In Figure 3 the load has increased to
1,500 tons (5275 kW); 6,000 gpm (379 L/s)
is circulating in the primary and secondary systems with no flow in the bypass.
Three inefficiently loaded chillers and
three pumps are on-line. Two chillers
could handle the load.
To satisfy the 1,600 ton (5627 kW) load
in Figure 4, 6,400 gpm (404 L/s) (1,600 4
gpm/ton [5627 0.07 mL/J]) is required.
The flow in the secondary now exceeds
the primary flow of 6,000 gpm (379 L/s),

plant would have sufficient capacity if the


chillers were piped and staged properly.
Adding more equipment is not the solution. The secondary pump and piping is
not designed for the additional flow.
The capacity of a chiller, using hydrocarbon-based refrigerant, increases from
0.5% to 1.5% for every degree decrease in
entering condenser water temperature.
However, the increase in chiller capacity
at a lower kW/ton cannot be used when
chillers are staged from flow. Whenever
the flow exceeds the design flow of the
chiller in the typical primary/secondary
system, the next machine is staged on even
though the operating machine(s) are not
fully loaded. The maximum capacity of any
machine never goes above the 100% capacity abscissa line in Figure 5. Most chillers are rated at 85F (29C) entering condenser water temperature, a condition that
may exist less than 2% of the year. At other
times whenever the outdoor wet bulb and
About the Author
Gil Avery, P.E., is a principal in The
Kele Companies in Memphis, Tenn.

w w w. a s h r a e j o u r n a l . o r g

May 2001

Chillers

Figure 1: A typical primary/secondary chiller system fully


loaded at the outdoor design conditions of 95F (35C) dry
bulb and 78F (25.5C) wet bulb.

Figure 2: A typical primary/secondary chiller system with


a load of only 600 tons (2110 kW).

Figure 3: A typical primary/secondary chiller system with


the load having increased to 1,500 tons (5275 kW).

Figure 4: A typical primary/secondary chiller system with


a 1,600 ton (5627 kW) load.

entering condenser water temperature is less than design, the


chiller capacity will increase and the kW/ton will decrease. This
higher efficiency and lower kW/ton operating range is in the
shaded area above the 100% capacity abscissa line in Figure 5.
A high-efficiency chiller may operate at 0.55 kW/ton with 85F
(29C) condenser water and at 0.34 kW/ton with 55F (12.7C)
condenser water. Most 1,000 ton (3517 kW) chillers may therefore produce more than 1,100 tons (3869 kW) when the entering
condenser water temperature is 65F70F (18C21C).
The majority of primary/secondary chilled water plants operate similar to those depicted in Figures 2 and 3, with too many
chillers on-line for efficient operation. Minimizing the number
of chillers on-line reduces the plant kW/ton since the unneeded
pumps and auxiliary devices are also off-line.

Improving The Efficiency


Although the remedy is relatively simple and inexpensive,
operators are hesitant to change, thinking, That was the way
the system was designed to operate. Initially everything possible must be done to correct the issues contributing to the low
T. Then change the way the chillers are staged on and off and
install a check valve in the bypass line at Point C as shown in
Figure 4 to prevent return water from mixing with supply water.
The check valve and bypass line should be sized to handle the
minimum flow of the largest chiller. A non-slamming check valve
with a manual opening device is generally used. A temperature
sensor in the secondary supply water, set low enough to satisfy the load requiring the coldest water, will stage the chillers
on. They are staged off by another sensor in the secondary

May 2001

ASHRAE Journal

15

ASHRAE Journal

Figure 5: Percent of rated capacity vs. outdoor wetbulb temperature.

Figure 6: A single pump primarysecondary chilled water system.

Figure 7: Multi-plexed sources and loads.

return water and by measuring the load (flow T). When the
load decreases below the capacity of an on-line chiller, that
chiller and pump is staged off. The sensors in the supply and
return lines must be matched sensors with an accuracy of
0.10F (0.5C). These changes generally have been shown to
reduce the kW/ton by 10 to 20%.
Most chillers will operate satisfactorily with evaporator tube
water velocities between 3 and 12 fps (0.9 and 3.6 m/s). Under
certain conditions (e.g., startup), the flow in the secondary will
exceed the primary design flow and the primary and secondary
pumps will then be in series. There will be a slight increase in
the flow through the chillers under these conditions but since
the flow varies as the square of the head, the actual increase
will rarely exceed 10% above design. An increase in flow even
this great would require an increase in pump head of over 20%
a condition unlikely to occur with the usual centrifugal pumps
that are chosen with flat pump curves.
Single Pump Primary/Secondary System
For new construction, the central plant can be designed us16

ASHRAE Journal

ing a single pump, as shown in Figure 6. Conventional primary


pumps are usually large, low-speed, low-head pumps that require an excessive amount of floor space. Their elimination can
materially reduce the initial cost of the system.
The differential pressure sensor DP-1 controls bypass
valve V-4 to maintain a minimum differential (minimum flow)
across the chiller evaporators. When a chiller is staged on and
the differential pressure decreases, the processor will modulate
valve V-4 open. When the coil valves or bypass valve V-4
opens, the differential pressure at sensor DP-2 will decrease
and the processor will ramp up the speed of pump P-1 through
the drive VFD to maintain a constant differential across the
ends of the supply and return mains. Flow at sensor FS-1 and
temperature at sensors T-1 and T-2 are used by the processor to stage the chillers on and off using the sequence described previously. Valves V-1, V-2, and V-3 are closed
when their respective chiller is off. Similar controls have been
applied for many years to constant flow systems that were
converted to variable flow and only one pump was available.
Although flow sensor FS-1 is shown, the differential pres-

w w w. a s h r a e j o u r n a l . o r g

May 2001

Figure 8: Tertiary pumping system.

sure measured by DP-1 across the chiller evaporators can be


used to calculate the flow.
The accuracy of this method of calculating flow increases in
proportion to the number of chillers in the central plant. Each
evaporator circuit acts as an orifice. As long as the system is
closed to the atmosphere, there will not be a significant change
in pressure drop through the evaporator tubes as the system
ages. On plants with three or more chillers, the accuracy approaches that of most flow sensors.
The bypass valve V-4 and bypass line should be sized to
handle the minimum flow of the largest chiller. The drop across
the bypass and valve must be less than the minimum operating
differential (setting of DP-2). Valve V-4 must be coupled to
an actuator that can close off against the full head of the secondary system. Correctly sized industrial metal-seated butterfly or ball valves are a good choice if they are selected to withstand the system dynamic forces.

Advertisement in the print edition formerly in this space.

Decentralized Systems
Outlying buildings on campuses and institutional complexes
may have chilled water sources that can be connected to the
chilled water system through the supply and return manifold
as shown in Figure 7. Building A houses the chilled water or
ice storage facility. Building B has an existing chiller and load.
Buildings C and E have a load and source respectively. Building D houses the central chiller plant. All sources pump out of
the return manifold through the cooling source and into the
supply manifold. All loads are pumped out of the supply manifold through the load and into the return manifold. Two position valves V-1 are used to isolate inactive loads and sources.
Multiplexing all of the chilled water sources and loads in this
manner has many advantages. Any source or combination of
sources can serve any load or combination of loads any place
on the manifold. Loads and sources can be matched to provide
the most efficient operation. Manifold piping is smaller and
less expensive than conventional loop piping. Chiller redundancy is maximized. A detailed method of controlling loads and
sources is found in (Avery 1987).
Tertiary Pumping Systems
Tertiary loads on central systems are generally piped in the
customary fashion as shown in Figure 8. Temperature sensor
T-1 controls valve V-1 to maintain the design supply chilled
water temperature. Differential pressure sensor DP-1 conMay 2001

ASHRAE Journal

17

ASHRAE Journal
trols the speed of the building pump through the variable frequency drive VFD. This method of piping and control does
not take advantage of the differential pressure that is available
between the supply and return mains. The pressure is wasted
across valve V-1. A preferred method of piping the tertiary
system is shown in Figure 9. The differential pressure sensor
DP-1 controls valve V-1 in sequence with the speed of the
pump through the variable frequency drive VFD. As the load
in the building increases, valve V-1 modulates open. On a
continued drop in differential pressure, pump P-1 will start
and ramp up to the speed required to maintain the design differential pressure at DP-1. The cycle reverses on a rise in differential pressure. Using the pressure between the supply and
return mains reduces the building pumping energy.
Summary
The use of the typical primary/secondary paradigm shown in
Figure 1 may be costing building owners millions of dollars in
additional operating costs because of the inefficient nature of
these systems. There are better ways of designing chilled water
systems. Its time to get our heads out of the 1950s and start
using what weve learned since then.
Bibliography
Avery, G. 1987. Microprocessor control for large chilled water

Figure 9: Tertiary booster pumping system.


distribution systems. Heating/Piping/Air Conditioning Engineering.
Avery, G. 1993. Designing and commissioning variable flow hydronic systems. ASHRAE Journal 35(7):4549.
Avery. G. 1998. Controlling chillers in variable flow systems.
ASHRAE Journal 40(2):4245.
Avery, G. 2000. Selecting valves and piping coils. ASHRAE Journal 42(4):2327.

Advertisement in the print edition formerly in this space.

18

ASHRAE Journal

w w w. a s h r a e j o u r n a l . o r g

May 2001

You might also like