Energy

DTU Compute
Richard Petersens Plads DTU - Building 324 DK-2800 Kgs. Lyngby
DTU

Dynamic Calculation Methods For Building Energy Assessment - Programme

The summer school is from 9AM to 5PM every day.

A more detailed programme will follow later.

  1. Preparation exercise for participants.

    The lecturers would like to get insight into the competence of the students at the start of the Summer School week.

    Homework has been prepared for the participants in order to a get a minimum homogeneous starting level with the objective of optimising the usefulness of lectures.

    An introduction to R and simple linear regression exercise is given as obligatory preparation. Participants will be asked to solve as homework a proposed common wall-exercise and report step by step, the analysis and validation carried out as clearly as possible. These reports must be submitted to the organisers before the start of the Summer School. 

  2. Building Physics and mathematical models.

    Lectures will provide the necessary background information on building physics to support the development of mathematical models for energy performance assessment. This includes basic knowledge of thermodynamic processes, in particular heat transfer and the impact of solar radiation. Topics like thermal conduction, convection and radiation will be presented as well as thermal mass. Using data-series for analysis the students will be introduced to the complexity of the physical process and how to translate the available information in mathematical models, e.g. the importance of model simplification of building physics represented by measured signals.

    Also the issue of standardisation will be presented, e.g. laboratory testing of building products and in-situ measurements for building element and whole building energy performance assessment.

  3. Models and model building

    Linear and non-linear state-space models. Topics such as identification, formulation, estimation, validation and Kalman Filter techniques are presented. In addition, lumped parameter models, RC-models, models, and combining information from data with prior information from physics are presented. Participants will learn how to use these techniques to estimate detailed physical quantities like the heat capacitance, window areas, solar aperture, effect of wind speed, nonlinear heat transfer, and non-stationary heat transfer.

    Simulation, Prediction and Control. A short introduction will be given on the representation of physical models within simulation programs and the use of these for prediction and control.. This will include consideration of uncertainty in inputs and algorithms.
    Linear transfer function models. Topics such as identification, formulation, estimation, and validation are presented. Furthermore, impulse response models, transfer function models, ARX, ARMAX and Box-Jenkins models and how to use these techniques to estimate values like the UA-value, gA-value and time constants of a building or a component will be covered.

  4. How to obtain results using different models and methods.
    The presented analysis and validation approaches will be step by step illustrated using a very simple and well documented case study.
    The U value of a well-known quite simple opaque wall will be estimated using different analysis approaches. The simplicity of this component allows the application of a wide variety of analysis approaches with different degree of complexity, capabilities and accuracy. This simplicity also allows isolate and highlight the effects of the different analysed aspects, from the effects of other assumptions about the test component and its boundaries that could be necessary if more complex case study were considered.
    The different approaches will be presented "bottom to top", starting from the simples and, gradually increasing complexity highlighting and discussing the main features added by each level by the corresponding modelling approach. The following approaches will be considered: average and pseudo-dynamic methods, transfer function models (using the statistical software R) and continuous-time state space models (using CTSM-R).
  5. Software
    Note that the software used during the Summer School is R. It is a free software environment for statistical computing and graphics. Find on the internet more information on R http://www.r-project.org/ and RStudio http://www.rstudio.com / (open source software).
    It would be wise to study the basics of R when you want to get the best out of the Summer School week.
    Examples will be presented and discussed for R and CTSM-R.
    CTSM-R is an R package in which an updated version of the Continuous Time Stochastic Modelling (CTSM) software is implemented. It can be used for grey-box modelling with stochastic differential equations and provides a very powerful framework for modelling dynamical systems.

For PhD students
In order to receive 2.5 ECTS credits at PhD level through DTU or KU Leuven, the students must:

  1. Hand in the solution to a preparing exercise before the Summer School starts. This exercise will be theoretical and will use simulated data
  2. Follow the Summer School all 5 days, 1st to 5th September 2014 in Leuven..
  3. Hand in reports on the exercises during the Summer School (appropriate time with assistance is reserved for carrying out the exercises).

Additional 2.5 ECTS points can be achieved by carrying out and reporting a proposed exercise or an individual project (related to the students own PhD project) after attending the summer school.

Note that you have to bring your own computer.