global optimisation of chiller - airah · plant simulation 𝑄Ǘ...

Post on 17-Mar-2020

3 Views

Category:

Documents

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

Global Optimisation of Chiller Sequencing and Load Balancing Using Shuffled Complex Evolution

IAIN STEWARTA (MENG)LU AYEA (FAIRAH, FAIE, PHD)

TIM PETERSONB (PHD)

A Renewable Energy and Energy Efficiency Group, Department of Infrastructure Engineering,The University of Melbourne, Vic 3010, AustraliaB Environmental Hydrology and Water Resources Group, Department of Infrastructure Engineering, The University of Melbourne, Vic 3010, Australia

Evolutionary Chiller Optimisation (ECO)

• The why?

• A brief history of building controls

• Chillers and VSDs

• The how? Improving system efficiency

• Aims- Optimising equipment loading

• Model- Structure, inputs & outputs

• Results- Modelled savings

Why?

• Energy costs

• Carbon emissions

• Peak demand reduction

40%

35%

15%

20%

Typical Commercial Building End Use Energy

HVAC

Lighting

Equipment

Other

92PJ

22MtCO2

A Brief History of Building Controls

Pneumatic controls….

Direct Digital Control (DDC)

A Brief History of Building Controls

Building Management System (BMS)

A Brief History of Building Controls

Reciprocating vs Centrifugal Chillers

Chiller COP vs Loading

0

2

4

6

8

10

12

0% 25% 50% 75% 100%

Ch

iller

CO

P

Chiller Load %

New centrifugal variable speed chiller

Old reciprocating chiller

100%300%

Variable Speed Drives (VSDs)

Variable Speed Drives (VSDs)

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Ener

gy C

on

sum

pti

on

Pump Speed

Pump Speed vs Energy Consumption

<50%

10% 80%

Technological improvements

• Capabilities of Building Management Systems

• Improved chiller part load efficiency

• Affordability of Variable Speed Drives

Current Control Technique

• Return water temperature staging

Switch-on thresholds Switch-off thresholds

From 1 chiller to 2 chillers: Trtn > 11.94 °C From 2 chillers to 1 chiller: Trtn < 9.28 °C

From 2 chillers to 3 chillers: Trtn > 11.85 °C From 3 chillers to 2 chillers: Trtn < 10.10 °C

From 3 chillers to 4 chillers: Trtn > 11.81 °C From 4 chillers to 3 chillers: Trtn < 10.51 °C

Trtn = Temperature of return chilled water from the building

The Opportunity

• Maximize cooling system efficiency [COP]

– Non linear efficiency curves

– Multiple pieces of equipment

– Inherently difficult for humans

Aims

• Accurately simulate waterside plant equipment

• Optimise equipment loading for cooling loads

– Maximise plant COP for all cooling loads

”All models are wrong but some are useful” – George Box

ECO

1. Simulate waterside plant– Energy inputs

• Flow rates• Mechanical work• Efficiencies• Physical limits

– Refrigeration delivered2. Optimise equipment staging and loading using genetic algorithm for a range of cooling loads

Modelling assumptions

• Chiller turn down ratio = 10:1

• Condenser and chilled water pumps operate at minimum flow rate when chillers are at minimum output (eg. 50% pump speed when chiller is at 10% capacity)

• Pumps speed increases linearly with chiller energy

• Maximum temperature difference across chillers = 7°C

• Fans excluded from simulation

Plant layout

Plant Simulation

Inputs Simulation Output

Simulation of chilled water plant (Chillers &

Pumps)*Fans excluded

Cooling Load [kWr]

Low load chiller energy [kW]

High load chiller 1 energy [kW]

High load chiller 2 energy [kW]

Input energy [kWe]

System COP

Plant simulation

Ǘ𝑄𝑟 = Ǘ𝑚cw𝑐𝑤(𝑇𝑖 − 𝑇𝑜)

• where, Ǘ𝑚cw is the mass flow rate of the chilled water [kg s-1]

• 𝑐𝑤is the specific heat of chilled water [kJ kg-1K-1]

• 𝑇𝑖 and 𝑇𝑜 are the inlet and outlet chilled water temperatures [°C] across the chiller.

Regression analysis

Ǘ𝑄𝑟 = aP + bP2

where, Ǘ𝑄𝑟 is the refriegeration energy [kWr], P is the chiller input power [kWe], and a & b are scalars to be

calculated in the regression analysis

y = -0.0185x2 + 8.8461xR² = 0.9224

0

200

400

600

800

1000

1200

1400

0 50 100 150 200

Co

olin

g ca

pac

ity

[kW

r]

Input Energy [kWe]

Regression of Chiller [kWr vs kWe]

Results of chiller regression analysis

Chiller #Input Power

[kWe]Nominal

Capacity [kWr] a b R2

1 59 289 8.7957 -0.0699 0.9236

2 178 988 8.8461 -0.0185 0.9224

Modelling Structure

InputSimulation & Optimisation

Outputs

Simulation of chilled water plant (Chillers &

Pumps)*Fans excluded

Cooling Load [kWr]

Low load chiller energy [kW]

High load chiller 1 energy [kW]

High load chiller 2 energy [kW]Shuffled complex

evolution algorithmObjective function:

∑ Plant equipment energy

Shuffled Complex Evolution

• Deterministic & probabilistic

• Competitive evolution

• Clustering

• Steepest decent

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

0 500 1000 1500 2000 2500 3000 3500

Ch

ille

r In

pu

t P

ow

er

[kW

r]

Cooling Load [kWr]

Chiller Input Power [%] vs Cooling Load [kWr]

Low LoadChiller

High LoadChiller 1

High LoadChiller 2

5.9

5.95

6

6.05

6.1

6.15

6.2

6.25

6.3

6.35

0 500 1000 1500 2000 2500 3000 3500

Syst

em

CO

P

Cooling Load [kWr]

System COP vs Cooling Load [kWr]

5.9

5.95

6

6.05

6.1

6.15

6.2

6.25

6.3

6.35

0 500 1000 1500 2000 2500 3000 3500

Syst

em

CO

P

Cooling Load [kWr]

System COP vs Cooling Load [kWr]

Response surface, COP vs Chiller Loading [%]

0

20

40

60

80

100

120

140

0 500 1000 1500 2000 2500 3000 3500

Ene

rgy

Savi

ngs

[kW

e]

Cooling Load [kWr]

Energy Savings [kWe] vs Cooling Load [kWr], aggregated into bins of 100kWr

0

20

40

60

80

100

120

140

0 500 1000 1500 2000 2500 3000 3500

Ene

rgy

savi

ngs

[kW

e]

Cooling Load [kWr]

Energy savings [kWe] vs Cooling load [kWr]

148,269

18,683

166,952

21,579

11,594

33,173

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

100%

Chillers Pumps Total

Modelled Energy Savings, 2016 [kWh]

Modelled energy consumption 2016 Energy savings from observed consumption

Results

Chiller 1 Chiller 2 Chiller 3 Pumps Total

2016 actual [kWh] 95,254 71,480 3,114 30,277 200,125

Modelled control strategy [kWh] 82,014 50,877 15,378 18,683 166,952

Energy savings [kWh] 13,240 20,603 (12,264) 11,594 33,173

% Savings 13.9% 28.8% -393.8% 38.3% 16.6%

40%

35%

15%

20%

Typical Commercial Building End Use Electricity

HVAC

Lighting

Equipment

Other

92PJ

22MtCO2

Savings

15PJ

4MtCO2

15-25% Peak

demand reduction

Next steps…

• Test ECO in real world

• Represent chillers and fans using neural networks to increase optimisation opportunities (adaptive control)

• Incorporate ambient conditions and fan/condenser pump speed modulation (dynamic control)

top related