best-practices guideline on advanced occupant modeling · advanced occupant modeling what are...

Best-practices Guideline onAdvanced Occupant Modeling

Sara GilaniPost-doctoral Fellow

Liam O’BrienAssociate Professor

Content

• Part 1: Basic theories

• Part 2: Implementation of advanced occupant models in EnergyPlus

• Part 3: Sensitivity analysis

2

3

Part 1:

Basic theories

4

Why do occupants matter in buildings?

BuildingOperating conditions

Occupant(s)

Performance

• Occupants are not passive recipients of buildings.

• They actively and continually attempt to improve comfort.

Why do occupants matter in buildings?5

(Saldanha, Beausoleil-Morrison, 2012)

factor of 4

6

Lights 10 W/m2

Blinds 10% transmittanceSouth-facing office in Ottawa

(Burak Gunay)


7

Energy Use

Pro

bab

ility Renewable energy generation

capacity for 50% chance of achieving net-zero energy

90% chance of achieving net-zero energy

Uncertainty from occupant behaviour

Occupant effects: 10 to 1000%


BehaviorsTriggers

88

Comfort:

• Thermal

• Visual

• Acoustic

• IAQ

Adaptive:

• Turning on lights

• Opening/closing windows/blinds

• Adjusting thermostats

• Changing clothing level

Non-adaptive:

• Occupancy (i.e. presence of occupants)

• Using plug-in equipment

• Turning off lights

Contextual factors:

• Control systems

• Cost of energy

• Social constraints

What are occupant behaviors?

What are occupant behaviors?9

Indirectly affects

heating/cooling

load

Directly affects

energy and affects

heating/cooling load

Affects heating/

cooling load

Occupancy

(presence)

Window shade/

blind use

Operable window

use

Office plug-in

equipment use

Thermostat use

Personal fan use

Personal heaters

use

Clothing level

Metabolic rate

Lighting use

Heating

Cooling

Lighting

Office equipment

Legend

Direct physical causal effect

Indirect physical causal effect

Comfort related causal effect

(O’Brien et al., 2018)

Key occupant behaviors and how they may impact building energy performance:

10

Two main occupant modeling approaches:• Standard-based models

What are occupant modeling approaches?

Lig

ht sc

hedul

e

Time of day (hours)

Static deterministic model

Uncertainty bounds for static stochastic model

Pro

bab

ility

of lig

ht

bein

g turn

ed o

n

Illuminance

Dynamic deterministic model

Dynamic stochastic model

0 24

1

0

Lig

ht schedule

Time of day (hours)

Static deterministic model

Uncertainty bounds for static stochastic model

Pro

babili

ty o

f lig

ht

bein

g turn

ed o

n

Illuminance

Dynamic deterministic model

Dynamic stochastic model

0 24

1

0

• Advanced occupant models


11

What are occupant modeling approaches?

• Standard-based models are the current practices in modeling occupant in buildings.

• Limitations:• Inaccurate prediction of building performance• Poor design decision-making based on the inaccurate

simulation results


12

(O’Brien, Gunay, 2015)

BuildingOperating conditions

Occupant(s)

Performance

Comfort conditions

Adaptive actions

Advanced occupant modeling

What are advanced models?

1. Dynamic

2. Stochastic

3. Agent-based

4. Data-driven

Pro

ba

bili

ty

Building performance(O’Brien et al., 2018)

What are advanced model forms?13

1. Markov chain:This model form predicts whether an occupanttakes an action in the next timestep or next event:1. Discrete-time2. Discrete-event

2. Bernoulli:This model from predicts the state of a buildingsystem or component.

3. SurvivalThis model form predicts the duration of a stateright before an event happens.

Which occupant models to use?

Occupant modeling strategies vary for different applications:

• What is our aim from simulating a building or a room-level model?

• How big is the building we simulate?

• What type of building we simulate?

14

What is our aim from simulating a building or a room-level model?

• Code compliance

• HVAC design

• Net-zero energy design

• Comfort assessment

• Façade design

15

How big is the building we simulate?

• The larger the building size, the lower the uncertainty in the predicted whole-building energy use

16

(Gilani et al., 2018)

What type of building we simulate?• Energy uses of various building types have different

sensitivities to occupant behaviors.

17

Regular staff/

Employees

Building owner/

manager

Visitors/guests/customers/

patients/students

Motel/hotel

Nursing home/care

facility

Retail

School

Medical office

Food/beverage store

Hospital

Non-med. office

Warehouse


Use cases:

• Whole-building energy prediction

• Building system and plant equipment sizing

• Net-zero energy buildings

• Occupant comfort

• Façade design

18

Which occupant models to use?

Whole-building energy prediction

• Current method: Standard schedules

• Proposed method:Small buildings:

Agent-based models

Medium to large buildings:Static-deterministic models for non-adaptive behaviors and dynamic-deterministic models for adaptive behaviors

19


Building system and plant equipment sizing20


Lack of potential to evaluate the risk of downsized building system and plant equipment

• Proposed method:Zone-level:

Static-deterministic (i.e. standard schedules)

Building level:Static-stochastic models

heatingheating (std. sched.)

cooling

cooling (std. sched.)


Net-zero energy buildings21

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0.58 0.66 0.74 0.82 0.9 0.98 1.06 1.14 1.22

Cu

mu

lati

ve

pro

bab

ilit

y

Annual electricity use GWh

90

% N

ZE

95

% N

ZE

99

% N

ZE

99

.9%

NZ

E

Standard

schedules

ASHRAE

Std. 90.1

Stochastic schedules

99.9

% N

ZE

99%

NZ

E

95%

NZ

E90%

NZ

E

Sta

nd

ard

AS

HR

AE

sched

ule

s

(Abdelalim, O’Brien, 2018)


Lack of accurate predicted building energy use

• Proposed method:static-stochastic or dynamic-stochastic models

Provide an insight into the impact of occupant-related uncertainties on building energy use

Occupant comfort22


Lack of distribution of energy use and number of occupants’ interactions with buildings

• Proposed method:Dynamic-deterministic or dynamic-stochastic models

Potential for robust design by reducing occupants’ interactions with buildings

(O’Brien, Gunay, 2015)

23


Façade design


Lack of providing an insight into occupant-related uncertainty and occupants' interactions with facades' components

• Proposed method:Dynamic-deterministic or dynamic-stochastic models

Potential for optimal design

24

Part 2:

Implementation of advanced models in EnergyPlus

Discrete-time Markov chain models

𝑝 𝑠𝑤𝑖𝑡𝑐ℎ 𝑜𝑛 𝑜𝑓𝑓 =𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑡𝑖𝑚𝑒𝑠𝑡𝑒𝑝𝑠 𝑤𝑖𝑡ℎ 𝑎 𝑙𝑖𝑔ℎ𝑡 𝑠𝑤𝑖𝑡𝑐ℎ 𝑜𝑛

𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑜𝑐𝑐𝑢𝑝𝑖𝑒𝑑 𝑡𝑖𝑚𝑒𝑠𝑡𝑝𝑒 𝑤ℎ𝑒𝑛 𝑙𝑖𝑔ℎ𝑡𝑠 𝑤𝑒𝑟𝑒 𝑜𝑓𝑓

𝑝 =𝑒𝛽0

+σ𝑖=1𝑛 𝛽𝑖𝑥𝑖

1 + 𝑒𝛽0+σ𝑖=1

𝑛 𝛽𝑖𝑥𝑖

(Burak Gunay)

25

Advanced occupant models from literature26

EnergyPlus EMS Application

(Gunay et al. 2015)

27

EnergyPlus EMS ApplicationEMS variables • Sensors

• Actuators• Built-in variables• Global variable

28

Daylight sensor in the center and at the workplane height (0.8 m)

South-facing shoebox model29

EnergyPlus EMS Application

• We will implement advanced occupant models for:

1. Occupancy: Wang et al.’s (2005) model

2. Light: Reinhart’s (2004) model

30

0.0

0.2

0.4

0.6

0.8

1.0

0 0.2 0.4 0.6 0.8 1

Vac

ancy

du

rati

on

Random number

Wang et al.’s occupancy model

Current time Occupancy stateWang et al.’s model

EMS Sensor EMS Program EMS Actuator

31

Vacancy duration: Exponential distribution

Event time (arrival, departure, lunch, two breaks): Normal distribution

0.00

0.20

0.40

0.60

0.80

1.00

1.20

1.40

1.60

6 7 8 9 10

Pro

bab

ility

Event time

Event time Mean (hr) Std (hr)

Arrival 8

0.25

First break 10Lunch 12

Second break 15Departure 18

Vacancy duration Mean (hr) Std (hr)

First break 0.25 0.1Lunch 1 0.25

Second break 0.25 0.1

0.0

0.2

0.4

0.6

0.8

1.0

0 200 400 600 800 1000

Pro

bab

ility

of

ligh

t sw

itch

-on

Workplane illuminance (lx)

Reinhart’s light switch model

•Current time•Occupancy•Daylight

Light stateReinhart’s model

EMS SensorEMS Program EMS Actuator

𝑝 =𝑒𝛽0

+𝛽1𝑥1

1 + 𝑒𝛽0+𝛽

1𝑥1

32

Light switch-on:

x = workplane illuminance

Light switch β0 (μ±σ) β1 (μ±σ)

Switch-onArrival 1.6 ± 0.3 −0.009 ± 0.002

Intermediate -3.9 ± 0.5 −0.002 ± 0.0005Switch-off at departure -1.3 ± 0.3 0.003 ± 0.001

33

Part 3:

Sensitivity analysis


Glazing systems design parameters

Type U-factor (kWh/m2) SHGC VT

1 1.82 0.36 0.642 1.42 0.48 0.69

34

• Two window shade control systems:

(1) blinds are closed all the time

(2) blinds are open all the time

35


36

(Abuimara et al., 2018)


• Impact of energy conservation measures (ECMs) on the predicted building energy use


• Various parameters:

o ECMs

o Climate zones

oBuilding types

oBuilding sizes

oBuilding users-related domains

37

ReferencesAbdelalim A, O’Brien W. 2018. An approach towards achieving net-zero energy buildings based on a stochastictenant model. In: eSim 2018. Montreal, Canada.

Abuimara T, O'Brien W, Gunay HB, Carrizo JS. 2018. Assessing the impact of changes in occupants on designdecision making. In: eSim 2018. Montreal, Canada.

Gilani S, O'Brien W, Gunay HB. 2018. Simulating occupants' impact on building energy performance at differentspatial scales. Building and Environment 132:327-337.

Gilani S, O’Brien W, Gunay HB, Carrizo JS. 2016. Use of dynamic occupant behavior models in the buildingdesign and code compliance processes. Energy and Buildings: Special Issue on Advances in BEM and Sim117:260-271.

Gunay HB, O'Brien W, Beausoleil-Morrison I. 2016. Implementation and comparison of existing occupantbehavior models in EnergyPlus" Journal of Building Performance Simulation.

O'Brien W, Abdelalim A, Abuimara T, Beausoleil-Morrison I, Carrizo JS, Danks R, Gilani S, Gunay HB, Kesik T, OufM. 2018. Roadmap for occupant modelling in building codes and standards. In: eSim 2018. Montreal, Canada.

O’Brien W, Abdelalim A, Gunay HB. 2018. Development of an office tenant electricity use model and itsapplication for right-sizing HVAC equipment. Journal of Building Performance Simulation.

O’Brien W, Gunay HB. 2015. Mitigating office performance uncertainty of occupant use of window blinds andlighting using robust design. Building Simulation:1-16.

Reinhart CF. 2004. Lightswitch-2002: a model for manual and automated control of electric lighting and blinds.Solar Energy 77:15-28.

Saldanha N, Beausoleil-Morrison I. 2012. Measured end-use electric load profiles for 12 Canadian houses athigh temporal resolution. Energy and Buildings 49:519-530.

Wang D, Federspiel CC, Rubinstein F. 2005. Modeling occupancy in single person offices. Energy and Buildings37:121-126.

38

Tareq AbuimaraPhD studentCivil Engineering

Aly Abdelalim, PhDPost-docEnvironmental Engineering

Sara Gilani, PhDPost-docCivil Engineering

Mohamed Ouf, PhDPost-docCivil Engineering

Prof. Liam O’Brien Prof. Burak Gunay Prof. Ian Beausoleil-Morrison

Carleton project team39

Partners and Sponsors40

Thank You

Sara GilaniEmail address: [email protected]

41

mailto:[email protected]

best-practices guideline on advanced occupant modeling · advanced occupant modeling what are...

Documents