Model-Based Control of Flexible and Responsive Buildings by pptfiles

VIEWS: 0 PAGES: 28

									Model-Based Control of Flexible
     and Responsive Buildings

     With Application to Desiccant
        Dehumidification Systems

          Masters Thesis Proposal
                     Brian Coffey
Contents
 Introduction and Overview
 Precedents / Literature Review
 Detailed Problem Definition
 Methodology
     Simple Case Study for Concept
      Development
     Detailed Case Studies

   Summary
                                      Contents
 Flexible and Responsive
     Change the                        Outdoor
                                                  Building
                                         Enviro
                                                             Occupant’s
      building system                             System
                                                              Desired
      to suit the constantly                                   Enviro

      changing environment
      and user desires




http://kdg.mit.edu/Projects/p07.html   Introduction and Overview
Responding to What?
   Weather
   Signals from Occupants
   Electricity and Gas Prices
   Signals from an Electrical Grid Manager



Why?
   Maximize Comfort
   Minimize Energy Consumption or Energy Costs
   Minimize Cost or Dissatisfaction While Avoiding Blackouts


                              Introduction and Overview
         Outdoor             Building
          Enviro             System


    Input                                           Output to
                         Control Unit
   Signals                                        Control System


           Rule-Based Control Unit
…if outdoorTemp > 25C and timeOfDay > 11am then Close Blinds
 if priceElectricity < 5 c/kWh then Turn Off Microturbine …


                               vs

         Model-Based Control Unit
 conditions are {C}, possible building states are {S1, S2, S3 …}
 use model to test possible states under given conditions
 choose the building state that uses the least energy


                                Introduction and Overview
Examples of Model-Based
   Solar Shading
      • Technical University of Vienna
   HVAC Control
      • EFPL’s NEUROBAT
      • Michael Kummert’s work
   Integrated System Control
      • EFPL’s EDIFICIO



                    Precedents / Literature Review
Examples of Model-Based
              Solar Shading
                 (TU Vienna)
   EDIFICIO (EPFL)




                   Precedents / Literature Review
System Description
    Integrated      Output
   Control Unit     Signals



      Input
     Signals

                     Building
                  Control System

      External
                              Building
     Conditions


                   Detailed Problem Definition
System Description
   n elements, {x1,x2,x3, … xn}
   xi has mi possible states,
        {si1,si2,si3, … sim}
                                  n
                                 m
    possible system states M = i=1 i
   at the beginning of every time interval t,
    integrated control unit must determine
    the system state S for that time period,
        S = {s1?, s2?,s3?, … sn?}

                      Detailed Problem Definition
Optimization Problem
            At a particular time t
       (where t = integer multiple of t)

Given input signals (weather, energy costs, etc.)
              C = {c1, c2, c3, … cl}

                Determine S to
           minimize Z = f(C,S)

                      Detailed Problem Definition
Optimization Approach
   f(C,S) ~ approximated by simulation

   If (M*simulationTime < t), then
      all the possible states can be tested with a
       simulation, and the state S which results in
       the lowest Z is chosen
   Else
      Only some of the the states can be tested,
       so which ones to test?


                         Detailed Problem Definition
Key Considerations
   Running simulations for more than
    just current timestep
     Cost of configuration change
     Transient effects, energy storage




                     Detailed Problem Definition
Study Methodology
Develop a control function in Matlab to act
 as the integrated control unit
  Integrated           Output
 Control Unit          Signals


                Make it general.
    Input
   Signals      Allow it to collect inputs, interface
                with TRNSYS models, use
                appropriate optimization
                strategy, find S, send outputs

                                        Methodology
Case Studies
   Use a simple case study to develop and
    test the Matlab control module
       Simplified Building (exists on paper only)
   Then apply the method and module to
    more detailed case studies to further
    refine and test the approach
       PWGSC Building (extant), with added
        cogeneration-desiccant system (not extant)
       LEPTAB Building with solar-desiccant system
        in Chambery (extant)
       Other possibilities
         • building-in-a-box, Annex20    Methodology
Simplified Building




          Methodology: Simple Case Study
Simplified Building
   2 elements, {x1,x2}
   x1 has 3 possible states, x2 has 2
   possible system states M = 6
   t = 1 hr
   integrated control unit must determine
    the system state S for that time period,
        S = {s1?, s2?}


                 Methodology: Simple Case Study
Simplified Building
   Input signals C = {c1, c2, c3, … cl}
       c1 = Toutdoor
       c2 = Tindoor
       c3 = solar radiation on south wall
       c4 = electrical load
       c5 = number of occupants in space
       etc...
 Find S to minimize Z = f(C,S)
 Z = Total energy consumption
 f(C,S) calculated by simulation
                       Methodology: Simple Case Study
Simplified Building
   Text Model

Shading Input
Window Input

TRNSED Model

      Output

     Example
      Results



                Methodology: Simple Case Study
Simplified Building
   Integrated Control Unit (Matlab)
 Integrated              Output
Control Unit             Signals


          • Determines what configurations to test
 Input
Signals   • Runs simulation for each configuration
               • Modifies Input Files (txt files)
               • Calls TRNSYS
               • Reads Output Files (txt files)
          • Compares outputs, chooses best
                      Methodology: Simple Case Study
Detailed Case Studies
 PWGSC Building for
  Responsive Buildings Project
 LEPTAB Building with
  Solar-Desiccant System
 Other possibilities




          Methodology: Detailed Case Studies
PWGSC Building
with Cogeneration-Desiccant
                Responsive Buildings
                             Project




         Methodology: Detailed Case Studies
PWGSC Building
with Cogeneration-Desiccant
   More than 5 elements, {x1,x2,x3, ...}
       x1   =   generator dispatch level
       x2   =   desiccant dispatch level
       x3   =   room temperature setpoint
       x4   =   room humidity setpoint
       x5 = light dimming control
       etc...
   Each xi has numerous possible states
   t = 1 hr? 15 min? 5 min?
   Inputs C will depend on building system
   Minimize Z = Energy Cost ?
                         Methodology: Detailed Case Studies
LEPTAB Building with
Solar-Desiccant System




         Methodology: Detailed Case Studies
LEPTAB Building with
Solar-Desiccant System
                     4 operating modes (M=4)
                         Ventilation Mode (S1)
                         Direct Humidification (S2)
                         Indirect Humidif. (S3)
                         Desiccant Mode (S4)
                     Currently rule-based control
                      if (time>9AM) & (TF-TE>1C)
                          if (Troom>26C)
                            Desiccant Mode
                          else
                             Indirect Humid.
                      …

         Methodology: Detailed Case Studies
Other possible case studies
   ‘Building in a box’
       Controls researchers at PWGSC have a box that tests
        actual building control systems, but with the control
        signals going to and from computers instead of a
        building and occupants
   Other buildings with desiccant systems in
    Germany that have already been modeled
    extensively, and for which I can get lots
    of data (part of IEA Annex 20 –
    Sustainable Cooling with TES)


                   Methodology: Detailed Case Studies
Expected Contributions
 Codification of model-based
  approach
 Application to desiccant systems
 Consideration of future timesteps in
  determining state at current
  timestep


                               Summary
Expected Challenges
 Dealing effectively with future states
 Performance Mapping or ANN
  creation for buildings under
  consideration
 TRNSYS modeling of desiccant
  systems


                                Summary
   ??


Questions?
Comments?
Feedback?



             Thank You

								
To top