Pongsak Chaisuparasmikul, Raymond J Clark, Robert J Krawczyk

1
Solar Energy Incorporated Day-lighting Prediction
Model Using Hypothetical Module

2
Pongsak Chaisuparasmikul, Raymond J Clark, Robert
J Krawczyk College of ArchitectureIllinois
Institute of Technology
ISES 2003 Solar World CongressGvteborg, Sweden
June 14-19, 2003.
3
KEY ISSUES IN SOLAR ENERGY PREDICTION MODEL

• Building consumes 35 of total energy
  consumption.
• Need the simplified method and tool for solar
  heating, cooling, and
• daylighting during the schematic design
  process.
• Modelling becomes the major issue of providing
  knowledge based information for
• designing solar energy efficient buildings.
• Measurable digital studying of the approximate
  method for the solar energy.
• Ability to identify the potential and problems
  related the functions and parameters.
• Addressing the issues of development the model
  into the software.

4
ABOUT THIS RESEARCH

• Looking at the building solar cooling model
  (which is more important for the office
• buildings in the U.S.A.)
• Finding could be iterative and alternative
  solutions, during the conceptual design
• process.
• Dealt with the interactive process, many unknown
  variables, energy approximation
• and probability model.

5
OBJECTIVES

• Create the solar cooling prediction models or
  equations.
• Find the approximate methods for the measurement
  of the solar heating, cooling and daylighting.
• Study the solar energy model without having to
  use the complicated building energy design
  software.
• Providing a mean of simplifying the calculation
  of solar cooling.

6
METHODOLOGY

• Wrote the source code program to create the DOE-2
  input simulation from the nested loop to generate
  the meaningful data for the multiple parameters.
• Assessment the influencial data from each of the
  parameters resulted from the DOE-2 simulation.
• Sensitivity analysis and correlation methods were
  used to select the most significant design
  variables.
• Regression procedure was using SPSS statistical
  software tp conduct and identify the principal
  form of relationships.
• Data from these selected parameters were
  generated for developing the multiple non-linear
  regression models (least square models) that can
  fit into the linear regression models.
• Multiple linear regressions were performed to
  derive the prediction model to obtain the
  best-fit equations.

7
MODULE

• The module was developed as a simulation input
  model for the DOE-2 processor.
• Intelligent module worked as an environmental
  interaction building (space, size,
• envelope materials, temperature,
  condition), and can be programmed as any the
• building types).
• Module concept was derived from the finite
  element theory (Raymond J. Clark)
• Any building forms or geometries can be
  approximately studied if they were divided into
  the smallest workable parts.
• Work well in studying the relative probability of
  influential parameters that are affected from the
  solar energy.

8
MODULE

• Module size 15 ft x 15 ft (4.50 m x 4.50 m)
  14ft (4.20 m) floor to floor high,
• 10 ft (3.00 m) floor
  to ceiling high.
• Climate Location Chicago, Illinois, USA.
• Weather file TRY (Test Reference Year)
• Sample
• Module schedule used the typical office-building
  schedule.
• Module wall, floor, partition, ceiling, and roof
  used the typical office building size, material,
  transmission U-value, insulation.
• Control strategies temperature, light (1.2W/m²),
  daylight, ventilation rate, number of people and
  equipment.
• Day-lighting schedule
• From 8.00am-6pm each day into the middle of the
  window façade, 5 ft (1.52 m) deep to the inside,
  3 ft (0.91 m) above the floor.
• Infiltration rate 0.06 cfm per window perimeter.

9
MODULE
10
MODULE
Pongsak Chaisuparasmikul, Raymond J Clark, Robert
J Krawczyk College of ArchitectureIllinois
Institute of Technology
ISES 2003 Solar World CongressGvteborg, Sweden
June 14-19, 2003.
11
MODULE
Pongsak Chaisuparasmikul, Raymond J Clark, Robert
J Krawczyk College of ArchitectureIllinois
Institute of Technology
ISES 2003 Solar World CongressGvteborg, Sweden
June 14-19, 2003.
12
DESIGN PARAMETERS

• Deterministic
• Variability

13
DESIGN PARAMETERS

• Orientation 8 orientation
• north, north-east, east, south-east, south,
  south-west, west, north-west.
• Fenestration 4 window to wall ratio
• 40, 50, 60, 80.
• Glass types 8 glass types
• Monolithic clear tinted Insulated clear
  tinted Low-E clear, tinted, green, reflective
• Overhang shading 4 overhangs shading ratio
• none, 25, 50, 0.75
• Fin shading 4 fins shading ratio
• none, 25, 50, 75

14
(No Transcript)
15
SIMULATION MODEL

• Customized input model
• Interactive and interfacing model with program
  Front end software
• Library entry data
• Programming the model
• Parametric analysis was assessed to obtain the
  most influential of each design
• variables to the annual solar heating.

16
Solar radiation variation of solar radiation
assessment is considered as a probability
Module with The design parameters
Module as a finite element black box
SIM File SUM File
Customized Input Model
DOE-2 Simulation 49,135 runs
Parametric analysis to obtain the most
influential variables
Regression procedure to identify the principal
form of relationships
Sensitivity analysis and correlation method to
select the most significant design variables.
Testing the model
Prediction Model
Multiple non-linear regression models fit on
linear regression model
Form of equations
17
Solar radiation variation of solar radiation
assessment is considered as a probability
Module with The design parameters
Module as a finite element black box
SIM File SUM File
Customized Input Model
DOE-2 Simulation 49,135 runs
Parametric analysis to obtain the most
influential variables
Regression procedure to identify the principal
form of relationships
Sensitivity analysis and correlation method to
select the most significant design variables.
Testing the model
Prediction Model
Multiple non-linear regression models fit on
linear regression model
Form of equations
18
Solar radiation variation of solar radiation
assessment is considered as a probability
Module with The design parameters
Module as a finite element black box
SIM File SUM File
Customized Input Model
DOE-2 Simulation 49,135 runs
Parametric analysis to obtain the most
influential variables
Regression procedure to identify the principal
form of relationships
Sensitivity analysis and correlation method to
select the most significant design variables.
Testing the model
Prediction Model
Multiple non-linear regression models fit on
linear regression model
Form of equations
19
Solar radiation variation of solar radiation
assessment is considered as a probability
Module with The design parameters
Module as a finite element black box
SIM File SUM File
Customized Input Model
DOE-2 Simulation 49,135 runs
Parametric analysis to obtain the most
influential variables
Regression procedure to identify the principal
form of relationships
Sensitivity analysis and correlation method to
select the most significant design variables.
Testing the model
Prediction Model
Multiple non-linear regression models fit on
linear regression model
Form of equations
20
Solar radiation variation of solar radiation
assessment is considered as a probability
Module with The design parameters
Module as a finite element black box
SIM File SUM File
Customized Input Model
DOE-2 Simulation 49,135 runs
Parametric analysis to obtain the most
influential variables
Regression procedure to identify the principal
form of relationships
Sensitivity analysis and correlation method to
select the most significant design variables.
Testing the model
Prediction Model
Multiple non-linear regression models fit on
linear regression model
Form of equations
21
Solar radiation variation of solar radiation
assessment is considered as a probability
Module with The design parameters
Module as a finite element black box
SIM File SUM File
Customized Input Model
DOE-2 Simulation 49,135 runs
Parametric analysis to obtain the most
influential variables
Regression procedure to identify the principal
form of relationships
Sensitivity analysis and correlation method to
select the most significant design variables.
Testing the model
Prediction Model
Multiple non-linear regression models fit on
linear regression model
Form of equations
22
PARAMETRIC ANALYSIS TREND OF DISTRIBUTION
Solar heating and cooling curve
23
PARAMETRIC ANALYSIS TREND OF DISTRIBUTION
Solar daylighting curve
24
PARAMETRIC ANALYSIS TREND OF DISTRIBUTION
Solar cooling peaks in June, July, or August
Solar heating peaks in December, or January
25
SENSITIVITY ANALYSIS AND CORRELATION SPREE PLOT
Correlation is significant at the 0.01 level
(2-tailed).
26
SENSITIVITY ANALYSIS AND CORRELATION SPREE PLOT
Significant 5 variables
27
RESULTS
Mean was closer than median
Histogram
Scattered outliers
Scattered Curve Fit
28
RESULTS
Data Residual Identification
Solar cooling
S distribution
Solar heating
S distribution with a significant on line
plotted for both of the prediction model
29
RESULTS
Regression Line
R² 0.21 R² 0.46 use Variable addition
multiplication R² 0.79 use Z transform Z
transform X - Mean
Estimator
of scattered data from regression line Std 1 ,
Mean 0
Comparison the predicted value and standard
residual with DOE-2 simulation
S distribution with a significant on line
plotted for both of the prediction model
30
RESULTS
Regression Model after Z transform
a Predictors (Constant), Standardized
Residual b Predictors (Constant), Standardized
Residual, Standardized Predicted Value c
Dependent Variable COOLING
31
FORM OF THE MODEL

• Form of equation was polynomial probability
  distribution.
• The variables were transformed (multiplication)
  one by one, and add new variables into the
• equations.
• Y(0.254OR0.065FN)-(0.025OR²0.004GL²0.008OH²)
  (0.044ORWR-0.005ORGL-
• 0.034OROH-0.010ORFN)-(0.002ORWRGL0.00
  3ORWROH0.003ORWRFN-
• 0.003ORGLOH-0.006OROHFN0.002GLOHFN)
  ____Least Square Line-1
•   Y(0.108WR-0.087GL-0.343OH)(0.023OHGL)(0.048O
  H²-0.003OH²WR-0.003OH²GL)
• 
  
  ____Least Square Line-2
• Y(0.216WR-0.147OH-0.125FN)-(0.141OHGL)
  ____Simplified Model

Where Y Maximum solar heat gain
OROrientation WRWindow to wall ratio
GLGlass types OH Overhang shading FN Fin
shading.
32
VALIDATION TESTING THE MODEL

• The valid testing of this model was conducted to
  see how this model was deviated from the
• standard errors and compared with DOE-2
  simulation results
• The test had S distribution with a significant
  line plotted
• Y(Model) Y(DOE-2) Model deviated
  from standard error
• N-2

33
CONCLUSIONS

• The results were to find the best-predicted
  value of Y (solar cooling energy) that
• respond well to the changing of the combination
  of design parameters X variables
• The model was in the form of quadratic and cubic
  equations
• Testing of this model was proposed to see how
  these models were deviated from
• the standard error and compared with DOE 2
  results.
• These methods were able to include all of the
  envelope design parameters.
• The equation models can be developed to provide
  the effective simplified tools for
• solar cooling.
• Future research include using these models in
  the measurement and verifiction of
• Midwest Chicago Green Technology Project.

34
THANK YOU FOR BEING HERE
FOR MORE DETAILED INFORMATION PLEASE SEE MY PAPER
NO. 06 22
OR YOU CAN CONTACT ME AT
EMAIL chaipon_at_iit.edu
WEBSITE www.iit.edu/chaipon/