kato_multiojective_ga

8/8/2019 KATO_multiojective_GA

http://slidepdf.com/reader/full/katomultiojectivega 1/16

Optimal Design Method ofOptimal Design Method ofPassive and Active ControllingPassive and Active Controlling

System for Indoor Climate DesignSystem for Indoor Climate Designwith Fluctuating Outdoorwith Fluctuating Outdoor

ConditionsConditionsKATO, Shinsuke

IIS, Univ. of Tokyo

JAPAN, JN3



ObjectiveObjective• Developing the optimal design method of passive and

active controlling systems of indoor climate using GA(MOGAs) and CFD

• Optimization will be done on the basis of multi-objectives

• In the study, the rational restrict conditions for searching

optimal solutions set (Paleto set) are examined• Hybrid Air-conditioning (Combination of Wind induced

ventilation and Air-Conditioning) will be dealt withconsidering windows, room shape, AC conditions,

variations of (random) outdoor climate conditions,variations of random indoor conditions



MultiMulti--Objective OptimizationObjective Optimization

There are no singular solutions for

the multi-objective optimization

Energy use

Productivity

Comfort ability

Daylight

Solar heat

OutdoorTemperature

Solution set

Singular Solution

Energy use

c o m f o r t a b i l i t y

c o m f o r t a b i l i t y

energy use

Inferior solution

Paleto solutions set

Selective solution

Objectives for window design

Object is evaluated quantitatively

Elements for window design

Element is changed through design process



Approach of MultiApproach of Multi--ObjectiveObjective

OptimizationOptimization To get the Paleto solution set with the multi-objective genetic algorism

Once the Paleto set is obtained, then the cluster analysis is done foradvising possible selective solution to designer

Solution set

Inferior solutions

Paleto solution set Cluster Analysis

Understanding Characteristics of Paleto set

Understanding the relationship with designvariables



PaletoPaleto Solutions of Daylight, Ventilation andSolutions of Daylight, Ventilation and

Thermal EnvironmentThermal Environment

CLUSTER1

CLUSTER2

CLUSTER3

CLUSTER4

CLUSTER5

CLUSTER6

CLUSTER7

CFD for Thermal Environment

Daylight simulation



MethodologyMethodology• GA (Genetic Algorithm) and CFD

(Computational Fluid Dynamics) are used

• Searching the optimal design of the hybrid

system which uses both passive andactive methods for controlling indoorclimate strongly affected by fluctuating

outdoor conditions and other parameters



Research DescriptionResearch Description• Developing the CFD with active controlling of indoor

climate for fluctuating outdoor conditions and others

• The active system adjusts its output to keep the indoorcondition at the targeted state (feedback system)

• We develop the simulation system of indoor climate

with the active control for fluctuating outdoor conditions• The evaluation of the optima should be done from the

viewpoint of energy saving, cost, human comfort,uniformity of daylight and so on (multi-objectives)

• Applying the methods for hybrid ventilation whichutilizes both wind induced cross ventilation and air-conditioning with fluctuating outdoor



Two Step Optimization ProcedureTwo Step Optimization Procedure



Example of Objective Function• The amount of energy-saving sensible

heat removed by natural ventilation• E(kW) = Cp×ρ×ΔT ×Q



Examples of Restricting ConditionsExamples of Restricting Conditions

• The average temperature ranges from 23

ºC to 27 ºC in the task region• The average air velocity is below 0.5 m/s

in the task region• The vertical difference in temperature is

below 3.5ºC in the task region

• ...........



Hybrid AirHybrid Air--Conditioning ModelConditioning Model



Fluctuating Outdoor ConditionsFluctuating Outdoor Conditions

26.24 (M+1.5σ)0.136σ ― 2σ

24.47 (M+0.5σ)0.341M ― σ

22.70 (M−0.5σ)0.341-σ ― M

20.93 (M−1.5σ)0.136-2σ ― -σ

Random variable(ºC)

ProbabilitySampling interval



GA inquiryGA inquiry



Cases Selected in the First StepCases Selected in the First Step

Case A Case B Case C

GA and CFD with coarse grid systems

Higher evaluations for the objectives and passing the restrictions



Case B is selected in the second stepCase B is selected in the second stepCFD with fine grid systems and highestevaluation for the objectives

Provability

34.1%

Provability34.1%

kato_multiojective_ga

Documents