1/8

Optimal Chilled Water Delta-T

Technical Assistance to Flow Control Industries (FCI)
Peter Armstrong and Dave Winiarski
30 September 2006
PNNL-16161
PNNL OSBP Technical Assistance Program
FCI markets a pressure-independent flow control valve for application at chilled water coils in
air handlers served by a chilled water loop. Proper control of chilled water to the air handlers
affects the staging of upstream pumps and chillers. A common symptom of pump control
sequence error is low chilled water temperature difference (Delta-T) at part load. In this
scenario, the flow rate and head remain high even when the cooling load is a fraction of design
load. Because flow rate and delta-T are inversely related, we can use the concepts of pump
sequencing, multi-speed control, and variable-speed control more or less interchangeably.
The objective of this Technical Assistance project is a set of analysis tools to help design and
operating engineers determine the best control sequence for pumps and fans and to estimate the
operating cost penalty for improper chilled water pump sequencing manifest as low Delta-T.
Delta-T is a shorthand term used in the HVAC industry for the chilled water temperature drop
across the evaporator, a function of cooling load and chilled-water flow rate.
A Chiller Plant Control Tool is needed to generate the control sequences and setpoints or reset
schedules that prevent excessive chiller compressor power on the one hand, and excessive fan
and pump power on the other. Under a given set of conditions (outdoor temperature and return
air or return chilled water temperature) a given cooling load will be met with minimum total
power when the relations between chiller, pump and fan power are properly balanced.
In addition to determining the best control strategy for efficient operation of a HVAC plant, the
designer should be able to estimate annual energy use and savings with respect to a faulty control
sequence. An Annual-Cost Estimating Tool is needed to estimate building loads and chiller plant
performance for a given control strategy.
It is desirable to structure these tools so that actual characteristics of existing equipment and
observed building load information can be plugged in to ensure that the resulting analysis applies
to the plant in questioni.e. has specific relevance to the building owner or bill-paying tenants,
and is not just a hypothetical paper study.
In this report, preliminary designs of the chiller plant control tool and annual operating cost tool
are presented. These preliminary designs illustrate the functions as well as the required input
data. However, the component models used at this stage are simplified and the results presented
for a simple test case are based on hypothetical plant parameters. Refinement of the models and
realistic test cases will be incorporated in the next phase of technical assistance.
Chiller Plant Control Tool
A prototype tool that may be used to find optimal speeds of the compressor, chilled water pump
and condenser fan has been developed. Although the use of a single compressor, single pump
and single fan is the simplest of possible topologies, it does illustrate the sensitivity of plant
performance to delta-T as pump speed is varied. This tool can be readily extended to model a
wide variety of plants with multiple chillers and multiple fans and pumps on the condenser and
evaporator sides of the compressors.

2/8
The plant parameters for an air-cooled chiller must be specified as follows.
Chiller performance maps:
Qcmpr (kBtu/h) = fc(s, Ps, Pd, Sc) and
Ecmpr (kW) = fp(s, Ps, Pd, Sc) where
s = suction density,
Ps, Pd = suction and discharge pressures, and
Sc = compressor shaft speed.
Evaporator pump and heat exchanger parameters:
Ceo (kBtu/h/F) = evaporator chilled water design thermal capacitance rate,
Eeo (kW) = pump power at design thermal capacitance rate, and
UAeo (kBtu/h/F) = evaporator heat exchanger conductance.
Condenser fan and heat exchanger parameters:
Cco (kBtu/h/F) = condenser fan design thermal capacitance rate,
Eco (kW) = fan power at design thermal capacitance rate, and
UAco (kBtu/h/F) = condenser heat exchanger conductance.
In addition, the conditions and cooling load are specified:
TODB = outdoor temperature
Tr = chilled water return temperature
Q = total chilled water cooling load.
The combination of pump, fan and compressor speed that satisfies load with the least total plant
power are solved for using the relationships developed in Appendix A. A typical solution is
shown in Table 1.
Table 1. Motor speed solution for one set of boundary conditions (BC).
Solution: Te,Tc
F
BC:Tr,TODB
F
BC: cooling load, Q
kBtu/h
Pump/fan parameters: Eeo, Eco
kW
HX Parameters: Ceo, Cco
kBtu/h/F
HX Parameters: UAe, UAc
kBtu/h/F
Q/dt (input to HX solver)
kBtu/h/F
HX solutions Ce, Cc
kBtu/h/F
3
Suction density,
Lbm/ft
Suction Pe, Discharge Pc
psia
Refrigt vapor enthalpies he,hc
Btu/lbm
Compressor speed Sc/So
Compressor capacity (Btuh), power (kW)
Pump, fan power Ee,Ec
kW
J = Ee + Ec + Ecmpr
kW
Delta-T
F
EER
Btu/Wh

Evap-side
35.6968741
70
2.69
4.32
12
2.33214898
2.3462468
77.095644
20.2008627
74790.7801
0.43094527
34.32710

Cond-side
115.919812
90
80
0.81
8.64
15
3.63017512
3.69383818
1.41203864
260.452879
44.3584305
1.070
4.12934608
0.06329616
4.62358752
22.57433
17.3025815

Solutions for a range of loads and conditions are presented in Figure 1. The response surfaces
represented in Figure 1 can be evaluated from tabular data (Appendix B) by linear interpolation.
The following trends are noted.
Efficiency. With optimal control of fan, pump and compressor, the overall plant efficiency,
EER=(cooling effect kBtuh)/(total kW), increases as load decreases. This is primarily the result
of reduced flow losses (fan and pump power decrease faster than cooling load) and closer
approach temperatures (compressor power decreases slightly faster than load).

3/8
40

35

30

25

20

15

EER (Btu/Wh)
delta-T
compressor power
total power (kW)

10

Input Power (kW)

EER (Btu/Wh), delta-T (F)

nf\smprj\FCI2\compr3.xls P|Q,T

0
0

10

20

30

40

50
Qload (kBtuh)

60

70

80

90

0
100

Figure 1. Solutions for range of cooling loads and TODB=60:10:100F. Within each family of
curves, power and delta-T increase while EER decreases with TODB.
Plant Electrical Load. Total plant power is almost linear with load because, under optimal
control, it is dominated by compressor power--even at the lightest of partial cooling loads.
Delta-T. Chilled water temperature delta-T is relatively constant under optimal control of fan,
pump and compressor. There is only a 17% reduction from design delta-T at the 50% part-load
mark.
Annual Cost Tool
A bin method used to estimate annual operating cost of package equipment (www.pnl.gov/uac)
has been adapted to estimate chiller plant operating costs.
To estimate seasonal performance we must specify, in addition to chiller plant performance, a
climate and a load. We must also specify energy prices.
A spreadsheet version of the UAC calculator is shown in Table 2. The load model assumes a
sensible cooling load directly proportional to outdoor temperature and a latent load directly
proportional to the product of outdoor humidity (mass ratio of water vapor to dry air) and outside
air flow rate. An ideal enthalpy control is assumed and a fixed minimum outside air flow rate
(10% in this analysis) is specified by the user. The peak load (for sizing purposes) is assumed to
occur at the ASHRAE 0.4% dry- and coincident wet-bulb temperatures with the minimum (40
scfm/Ton) 1 outside air-flow setting.
The cooling balance point must also be specified. Here balance point is defined as the thermostat setpoint minus the ratio of average solar and internal gains 2 to envelope UA. For example,
with a 75F setpoint, average gains of 150 kBtu/h and a UA of 60 kBtu/h/F, the balance point is
75 150/60 = 50F.
1
2

40 scfm per Ton of cooling capacity corresponds to about 10% outside air, a typical minimum outside air fraction
Average over occupied hours, including metabolic heat output of occupants.

4/8
Table 2. Chiller Plant Operation Cost Calculator.
CHILLER PLANT ANNUAL ENERGY, DEMAND, AND COST CALCULATION
Location:
Atlanta, GA

Balance Point

Economizer

Constant-speed fans & pumps

Demand Charges

Energy Charges

Plant kW, (f(I,A,B))

Ratchet ($/kW)

Remaining Load After

Economizer

Economizer Cooling
Capacity kBtu/h

Seasonal Cooling Hours,

TMY

0
0

Total Heat Gain, kBtu/h

(E+F)

Number of Months with

this Peak hour, TMY

C
Outdoor Temperature
Difference (ATbalance)

Outdoor Temperature 5F
Increments

Cooling Load

B
Coincident Wet Bulb
Temperature, F

Weather Conditions
A

52

Winter demand ($/kW)

Ventilation Heat Gain,

kBtu/h

Summer demand ($/kW)

Sensible Heat Gain,

kBtu/h

75

0.08

kBtuh
scfm

30

27.27549

-22

276

-46.5

35

31.64923

-17

424

-35.9

40

35.31892

-12

473

-25.4

45

40.87396

-7

629

-14.8

50

45.05447

-2

639

55

49.50995

748

60

54.32564

65

58.80224

Sequenced fans & pumps

O

Q
Energy Charges

Wet Bulb:

115

Demand Charges

Design Load
Ventilation CFM

Plant kW, (f(I,A,B))

Electric rate ($/kWh)

ASHRAE 0.4% Design Conditions
Dry Bulb:
93 F

-11.4

-26.2

-113.7

0.0

-4.2

-7.8

-12.1

-78.5

0.0

6.3

-3.7

2.6

-37.4

0.0

0.000

0.00

0.00

0.000

0.00

0.00

828

16.9

1.0

18.0

0.0

18.0

3.562

0.00

235.96

0.846

0.00

56.04

13

926

27.5

6.0

33.4

0.0

33.4

3.928

0.00

290.98

1.643

0.00

121.69

70

64.56173

18

1319

38.1

12.9

51.0

0.0

51.0

4.270

0.00

450.62

2.629

0.00

277.37

75

67.23051

23

983

48.6

16.5

65.2

0.0

65.2

4.522

0.00

355.63

3.512

0.00

276.21

80

69.00217

28

696

59.2

19.1

78.3

0.0

78.3

4.747

0.00

264.29

4.399

0.00

244.96

85

71.13729

33

449

69.8

22.2

91.9

0.0

91.9

5.407

0.00

194.22

5.407

0.00

194.22

90

73.31966

38

114

80.3

25.6

105.9

0.0

105.9

6.512

0.00

59.39

6.512

0.00

59.39

95

75.65207

43

16

90.9

29.3

120.2

0.0

120.2

7.671

0.00

9.82

7.671

0.00

9.82

100

$0.00

$1,624.95

$0.00

$1,183.66

105
110
115
Total Annual Costs
Outdoor temperature bins represent the median temperature (i.e. 60F is the range of temperatures between 57.5F through 62.5F)

5/8
The hourly coil load is evaluated by the load model for a given bin temperature The chiller plant
model is then applied to determine average kW for the bin and this is multiplied by the number of
annual bin hours to get bin kWh. The resulting kWh numbers are then summed over all bins to
get the annual operating energy. The bin energy charge is the product of bin kWh and price. For
time-of-use rates one must use a blended rate in which the summer on-peak rate is given extra
weight appropriate for the seasonal and time-of-day distribution of chiller plant electrical use.
The monthly peak demands can be estimated from the bins corresponding to the 12 monthly peak
cooling load conditions. The bin demand charge is the product of bin demand, its corresponding
number of months, and the demand charge. The numbers of months may be decomposed into
separate summer and winter columns to allow for different summer and winter demand rates.
Annual energy and demand charges are summed and appear at the foot of each column.
Conclusions and Recommendations
This report contains the preliminary draft versions of a chiller plant control sequence tool and an
annual operating cost tool. Following a review of this report by FCI, further refinement of the
component models used in the tools and more realistic test cases will be added in the next phase of
the technical assistance project.
Bibliography
Armstrong, PR, DW Winiarski, and S Somasundaram, 2004. Preliminary Analysis of PressureIndependent Flow Control for Cooling Coils, TGA PNNL-14417
Bahnfleth, William P. and Eric B. Peyer, 2004, Varying views on variable-primary flow: chilledwater systems, HPAC, March.
Bahnfleth, William P. and Eric B. Peyer, 2001, Comparative analysis of chilled-water-plant
performance, HPAC, April.
Bellenger, Lynn G., 2003, Revisiting chiller retrofits to replace constant volume pumps, HPAC,
September.
Braun J E., Klein S A., Beckman W A., Mitchell J W. 1989. Methodologies for optimal control of
chilled water systems without storage, ASHRAE Trans. 95(1): 652-662.
Braun J E., Klein S A., Mitchell J W., Beckman W A. 1989. Applications of optimal control to
chilled water systems without storage, ASHRAE Trans. 95(1): 663-675.
Braun, J, SA Klein and JW Mitchell. 1989. Effectiveness models for cooling towers and cooling
coils. ASHRAE Trans. 95(2): 164-174.
Braun J E., Diderrich G T. 1990. Near-optimal control of cooling towers for chilled water systems,
ASHRAE Trans. 96(2): 806-813.
Coad, WJ, 1985. Variable flow in hydronic systems for improved stability, simplicity and energy
economics, ASHRAE Transactions, 91(1) 224-237.
Ding, X, JP Eppe, J Lebrun, M Wasacz. 1990. Cooling coil model to be used in transient and/or
wet regimes: theoretical analysis and experimental validation. Proc. Third IBPSA Conf., Liege.
405-441.
Elmahdy, AH and GP Mitalas. 1977. A simple model for cooling and dehumidifying coils for use
in calculating energy requirements in buildings. ASHRAE Trans. 83(2): 103-117.

6/8
Erpelding, Ben, 2006, Ultra-efficient all-variable-speed chilled-water plants, HPAC, March
Gordon, J.M. and K.C. Ng, 2000. Cool Thermodynamics, Cambridge Intl Science Publishing.
Hardaway L R. 1982. Effective chilled water coil control to reduce pump energy and flow
demand, ASHRAE Trans., 88(1): 331-341.
Haines, Roger W., 2002. HVAC Controls through the years, HPAC, March and April
Hartman, Thomas, 2003, Direct network connection of variable-speed drives, HPAC, March.
Hartman, Thomas, 2001, Ultra-efficient cooling with demand-based control, HPAC, December.
Hartman, Thomas, 1998, Packaging DDC Networks with Variable Speed Drives, HPAC, Nov.
Hartman, Thomas, undated. A Hartmann Loop Example, www.hartmanco.com/pdf/a36.pdf
Henze, G.P., R.H. Dodier, and M. Krarti, 1997. Development of a predictive optimal controller for
thermal energy storage systems. HVAC&R Research, 3(3) 233-264.
Hung, C.Y.S, H.N. Lam, A. Dunn, 1999. Dynamic performance of an electronic zone air
temperature control loop in a typical VAV air conditioning system, Intl J HVAC&R Research
5(4):317-337
Jawadi, Z. 1988. A simple transient heating coil model, Proc. Winter Annual Mtg., ASME, 63-69.
Kelly, David W. and Tumin Chan, 1999, Optimizing chilled water plants, HPAC, January
Lau A S., Beckman W A., Mitchell J W. 1985. Development of computerised control strategies
for a large chilled water plant, ASHRAE Trans., 91(1B): 766-780.
Link J, Pepler R. 1970. Associated fluctuations in daily temperature, productivity and
absenteeism. No 2167 RP-57, ASHRAE Trans. 1970, vol 76, Part II, , pp 326-337,
Maxwell G M., Shapiro H N., Westra D.G. 1989. Dynamics and control of a chilled water coil,
ASHRAE Trans. 95(1): 1243-1255.
Mitchell J. 1988. Research Note 63 - analysis of energy use and control characteristics of a large
variable speed drive chiller system, ASHRAE Journal, 30(1) 33-34.
Pearson, JT, RG Leonard and RD McCutchan. 1974. Gain and time constant for finned serpentine
crossflow heat exchangers. ASHRAE Trans. 80(2): 255-267
Rishel J B. 1991. Control of variable-speed pumps on hot- and chilled-water systems, ASHRAE
Trans. 97(1): 746-750.
Schwedler, Mick and Brenda Bradley, 2000, Variable-primary-flow systems, HPAC, April
Schwedler, Mick and Brenda Bradley, 2003, Variable primary flow in chilled-water systems,
HPAC, March
Tamm, H and GH Green. 1973. Experimental multi-row crossflow heat exchanger dynamics.
ASHRAE Trans. 79(2): 9-18
Thielman D E. 1983. Chiller optimisation by energy management control systems, ASHRAE
Journal, 25(11)L, 60-62.
Utesch A L. 1990. Direct digital control of a large centrifugal chiller, ASHRAE Trans., 96(2): 797799.
Utesch, A L., 1995, Chilled-water distribution re-examined, HPAC, April
Zhou, X and JE Braun. 2004. Transient modeling of chilled water cooling coils. Intl
Refrigeration and Air Conditioning Conf. at Purdue, July 12-15, 2004.

7/8
Appendix A. Chiller Plant Optimizer.
Boundary conditions are total cooling load and evaporator- and condenser-side temperatures. The
condenser-side temperature of interest is ambient dry bulb for air cooled chillers and package
equipment or wet bulb temperature for chillers with cooling towers.
The evaporator side temperature of interest depends on the system boundary selected for the
analysis. If latent load is satisfied the return-air temperature should be used and the only other
evaporator-side boundary condition to satisfy is total cooling load. To satisfy the latent load it
may be necessary to specify the chilled water supply temperature as a constraint.
To illustrate the chiller plant analysis we consider an air-cooled chiller with outdoor dry-bulb and
returning chilled water temperatures specified. The total cooling load is also specified. The
evaporator and condenser saturation temperatures are unknown but we can provide a feasible
initial guess because the condensing temperature must be higher than ambient and the evaporating
temperature must be below the chilled water return temperature. The boundary conditions and
unknowns are thus defined as:
Q = total cooling load,
Tr = chilled water return temperature,
TODB = ambient dry bulb temperature
Te = evaporator saturation temperature (initial guess of Te < Tr),
Tc = condenser saturation temperature (initial guess of Tc > TODB).
(Note that, except when a water-side economizer is modeled, the initial guesses must also satisfy
Te < Tc even if Tr > TODB and must lie within the region modeled by the compressor maps)
The chilled-water thermal capacitance rate, Ce, is determined by solving:
Q = e(Ce,UAe)UAe(Te-Tr),
and the chilled water pump power is given by:
X Ee

C
where
E e = E eo e
C eo
Eeo, Ceo = pump power and thermal capacitance rate at reference conditions and
xEe 3 = the pump and chilled-water loop subsystems characteristic load-speed exponent.
The compressor capacity, power and mass flow rate at reference speed are:
Qo = fQ(, Pc/Pe) = chiller capacity at compressor reference speed,
mo = fm(, Pc/Pe) = refrigerant mass flow rate at reference speed, and
Eo = fE(, m, Pc/Pe) = compressor power input at reference speed
where
= suction vapor density,
Pc is the saturation pressure corresponding to Tc and
Pe is the saturation pressure corresponding to Te.
The compressor speed required to satisfy the load is:
Sc = SoQ/ Qo
where So is the reference speed. Compressor power is given by:
Ecmpr = Sc fE(, m, Pc/Pe) where m = (Sc/So)fm(, Pc/Pe)

8/8
The condenser air flow rate, Cc, is determined by solving:
Q + Ecmpr = c(Cc,UAc)UAc(TODB -Tc),
and the condenser fan power is given by:
X Ec

C
where
E c = E co c
C
co

Eco, Cco = condenser fan power and thermal capacitance rate at reference conditions and
xEc 3 = the fan and air-cooling loop subsystems characteristic load-speed exponent.
There are two unknowns, chilled water delta-T and condenser fan delta-T. When these are
specified all other intermediate variables (flow rates, and the speeds and electrical loads of fan,
pump and compressor) can be evaluated. The solver, given a feasible initial guess, performs a
search to find the values that minimize the total plant electrical load,
J = Ee + Ec + Ecmpr

Simplifying assumptions. To obtain a useful prototype of the plant control tool we have made
several simplifying assumptions. These simplifications can be addressed with more realistic,
complete, and rigorous modeling in the next generation tool:
-superheat is zero (entire cooling effect is delivered at the evaporator saturation temperature),
-the evaporator sensible heat ratio is 1,
-all heat rejection occurs at the condenser saturation temperature,
-no minimum flow requirements for chilled-water or condenser air,
-the chilled-water loop load curve (pressure drop vs. flow rate) is fixed,
-compressor flow and capacity both taken as proportional to shaft speed,
-chilled-water return (or return-air) temperature is fixed,
-plant consists of only one each: supply pump, compressor, and condenser fan.
Appendix B. Test Case Plant Performance Maps
The following plant performance maps were produced for a single load-side condition of Tr=70F.
Aggregate kW for plant comprised of VSD compressor with VSD pump & VSD condenser fan
Cooling Load (kBtu/h)
TODB

10

20

30

40

50

60

70

80

90

100

60
70
80
90
100

0.046
0.046
0.046
0.047
0.047

0.463
0.467
0.472
0.476
0.480

0.945
0.958
0.965
0.976
0.986

1.449
1.467
1.485
1.503
1.520

1.979
2.010
2.032
2.059
2.091

2.534
2.570
2.607
2.644
2.681

3.116
3.164
3.212
3.261
3.310

3.725
3.786
3.848
3.910
3.973

4.364
4.439
4.515
4.593
4.671

5.033
5.123
5.216
5.310
5.407

5.733
5.841
5.952
6.065
6.182

Aggregate kW for plant comprised of VSD compressor with fixed-speed pump & condenser fan
Cooling Load (kBtu/h)
TODB

10

20

30

40

50

60

70

80

90

100

60

2.864

3.330

3.622

3.854

4.057

4.242

4.415

4.580

4.739

5.033

5.733

70
80

2.864
2.865

3.334
3.337

3.629
3.633

3.861
3.869

4.068
4.075

4.253
4.265

4.429
4.442

4.596
4.612

4.757
4.775

5.123
5.216

5.841
5.952

90
100

2.866
2.867

3.340
3.343

3.638
3.643

3.876
3.883

4.085
4.096

4.276
4.287

4.456
4.469

4.627
4.643

4.793
4.811

5.310
5.407

6.065
6.182