Impact de l'estimation solaire sur la performance de la régulation par modèle prévisionnel dans des bâtiments multi-zones.

Impact of solar estimation on MPC performance of multizone buildings.

Numéro : pap. 3635

Auteurs : KIM D., WITMER L., BROWNSON J. R. S., et al.


Local solar irradiation data has been used for Model Predictive Control (MPC) in supervisory level building control systems, with significant influence demonstrated on perimeter building zones with a high fraction of window area. This work investigates the impact of uncertain solar estimation on MPC performance with respect to energy consumption and comfort by comparing MPC responses for a multi-zone building based on two solar data sources: measured local Plane of Array (POA) solar data and satellite Global Horizontal Irradiance (GHI) based modeled POA solar data. Hourly solar irradiation on tilted surfaces can be calculated from local or offsite GHI measurements by a common algorithm that includes an empirical correlation to decompose the beam and diffuse components, followed by the application of an anisotropic sky model. Real-time control decisions based on the modeled solar data have been found to violate comfort conditions within zones, particularly if off-site or satellite derived GHI data is employed. The presented case study results inform the value of information, revealing that MPC decisions based on satellite derived solar data cause comfort violations that depend strongly on the building system characteristics, such as HVAC and internal gains.

Documents disponibles

Format PDF

Pages : 10 p.


  • Prix public

    20 €

  • Prix membre*

    15 €

* meilleur tarif applicable selon le type d'adhésion (voir le détail des avantages des adhésions individuelles et collectives)


  • Titre original : Impact of solar estimation on MPC performance of multizone buildings.
  • Identifiant de la fiche : 30013640
  • Langues : Anglais
  • Sujet : Environnement
  • Source : 2014 Purdue Conferences. 3rd International High Performance Buildings Conference at Purdue.
  • Date d'édition : 14/07/2014


Voir d'autres communications du même compte rendu (66)
Voir le compte rendu de la conférence