Skip to content

2018 11 21 Task 3 Building modelling subgroup meeting

Jump to bottom Edit New page
InaDJ edited this page Nov 22, 2018 · 9 revisions

Task 3 Building modelling subgroup meeting

Date: November 21, 2018, 10am Brussels time (UTC+1)

Agenda

  • Single building of first dummy district was modelled in different libraries and its simulation results are compared > discuss probable causes of deviations
  • Discuss is how to coordinate our DESTEST building models with WP1 building "emulators" that are used in the BOPTEST

Meeting information

Minutes

Present

Present: Ina De Jaeger (notes), Alessandro Maccarini, Ilaria Vigna, Michael Mans, Igor Sartori, Haris Shamsi, Øystein Rønneseth, Rémi Patureau

Discussion of building comparison

  • Single building of first dummy district was modelled in 5 different libraries and its simulation results are compared. The plots of this comparison can be found here.
    • Modelica Aixlib (by Michael Mans),
    • Modelica IDEAS (by Ina De Jaeger),
    • Modelica Buildings (by Alessandro Maccarini),
    • IDA ICE (by Øystein Rønneseth),
    • DIMOSIM (by Enora Garreau)
  • Peak power: differences in calculation method
    • Aixlib: peak power is based on steady-state heat loss calculation, but reheating is not taken into account
    • IDEAS: peak power is based on steady-state heat loss calculation, and reheating is taken into account
    • Buildings: fixed value, equal to Ina's calculation
    • IDA ICE: fixed value, equal to 50 kW, so the building just takes what it needs
    • DIMOSIM: ?
  • Igor: for our case and our goals, the focus is on the peak power and the temporal behaviour (time profiles & load duration curves), not the yearly values of the energy demand for space heating
  • Heating power profiles:
    • IDA ICE: some negative numbers in the heating power (91 values), due to poor arithmetrics. There is only an ideal heating system modelled.
    • Do some models consider a heating period and a summer period, during which there is no heating? > find out
  • Temperature profiles:
    • IDEAS & Buildings: they seem to have really other dynamics, other time constants.
      • Ina thinks it is because of the ideal heating system that works on the operative temperature rather than the air temperature
      • Igor: are you sure that it is really and ideal heating system? The time constant and dynamics of the heating system
      • Igor: it is not really common to work on operative temperature. Your profiles look smoother and smarter, but it does not correspond to reality > proposal to work with air temperature
    • DIMOSIM: differences are really too big. What kind of model is it? > find out
      • Igor: Models for certification or steady-state models are incapable to represent the summer behaviour of buildings. That's the difference between high-order and low-order models (good vs. bad). Within the low-order models: physics-based models (calculate R & C) OR data-driven models (grey-box models, based on measured data).
  • Load duration curves:
    • Buildings & IDEAS: higher peaks and faster to zero > influence of operative temperature!
    • Aixlib: flatter peak > max heat power is too low (indoor comfort requirements will not be met at certain moments)
    • Michael: total duration is not that important. Maybe run a simulation for a certain heating period and make a load duration curve of that limited time frame.
    • Focus on peaks !
    • How to properly compare?
      • Igor: Throw away the first percentile of the load duration curve, so ignore the very first peak and then compare?
      • Michael: OR limit the maximal power and have a flat beginning of the load duration curve. This is more like in reality? Drawback: at certain moments, the indoor comfort requirements won't be met (Igor: real heating season is approx. 4000 hours, it is allowed for 50 hours that the set-points are not met by some standards, so there are guidelines for that)
      • Deciding for a rule ? First simulation: infinite power, to check the required power for the building. Then calculate the 99th percentile (= power at 87.6h), the 97.5 percentile (= power at 219h) and the 95 percentile (= power at 438h), also take power at 40 hours and power at 50 hours = 5 possibilities to use as a maximal power. Then run 5 simulations and send to Ina a CSV-file with air temperature of the day zone, air temperature of the night zone and heating power for each of these 5 simulations. Each CSV file has the following columns: Datetime_s; Tair_nightzone_K; Tair_dayzone_K; Qheating_building_W. Each CSV file has the name: 'simulation_power40hours.csv', 'simulation_power50hours.csv', 'simulation_power87hours.csv', 'simulation_power219hours.csv', 'simulation_power438hours.csv'.
  • General remarks:
    • Alessandro: Convective and radiative heat gains through heating system: proportion is defined in Ina's model. Set to same values for all models (70 % convective and 30 % radiative).

Actions

  • Make all models work based on air temperature and compare again > Ina & Alessandro re-run simulations + Ina includes results in comparison before Friday, December 7 (next Coordination meeting)
  • Restructure presentation of comparison results > ALL: please comment on the file that you find here and send comments to Ina by Monday, December 3 (presentation will be shown at Coordination meeting)
  • After coordination meeting, Ina, Alessandro, Enora, Michael and Øystein by January XX, 2019 (next Buildings meeting):
    • Convective and radiative heat gains through heating system: proportion is defined by Ina. Set to same values for all models (70 % convective and 30 % radiative)
    • Re-run with maximal powers of ideal heater (see above)
Add a custom footer

Wiki

Main project site

https://ibpsa.github.io/project1

Clone this wiki locally