%0 Journal Article
%K Windows
%K Daylighting
%K Switchable windows
%K Building energy-efficiency
%K Model predictive controls
%K Dynamic facades
%K Controller in the loop
%A Christoph Gehbauer
%A David Blum
%A Taoning Wang
%A Eleanor S Lee
%B Energy and Buildings
%D 2020
%G eng
%P 109762
%R 10.1016/j.enbuild.2020.109762
%T An assessment of the load modifying potential of model predictive controlled dynamic facades within the California context
%U https://linkinghub.elsevier.com/retrieve/pii/S0378778819317098
%V 210
%8 01/2020
%! Energy and Buildings
%X <p>California is making major strides towards meeting its greenhouse gas emission reduction goals with the transformation of its electrical grid to accommodate renewable generation, aggressive promotion of building energy efficiency, and increased emphasis on moving toward electrification of end uses (e.g., residential heating, etc.).&nbsp;&nbsp; As a result of this activity, the State is faced with significant challenges of system wide resource adequacy, power quality and grid reliability that could be addressed in part with demand responsive (DR) load modifying strategies using controllable building technologies.&nbsp; Dynamic facades have the ability to potentially shift and shed loads at critical times of the day in combination with daylighting and HVAC controls.&nbsp; This study explores the technical potential of dynamic facades to support net load shape objectives.&nbsp; A model predictive controller (MPC) was designed based on reduced order thermal (Modelica) and window (Radiance) models. Using an automated workflow (involving JModelica.org and MPCPy), these models were converted and differentiated to formulate a non-linear optimization problem. A gradient-based, non-linear programming problem solver (IPOPT) was used to derive an optimal control strategy, then a post-optimization step was used to convert the solution to a discrete state for facade actuation.&nbsp; Continuous state modulation of the façade was also modeled.&nbsp; The performance of the MPC controller with and without activation of thermal mass was evaluated in a south-facing perimeter office zone with a three-zone electrochromic window for a clear sunny week during summer and winter periods in Oakland and Burbank, California.&nbsp; MPC strategies reduced total energy cost by 9-28% and critical coincident peak demand was reduced by up to 0.58 W/ft<sup>2</sup>-floor or 19-43% in the 4.6 m (15 ft) deep south zone on sunny summer days in Oakland compared to state-of-the-art heuristic control.&nbsp; Similar savings were achieved for the hotter, Burbank climate in Southern California.&nbsp; This outcome supports the argument that MPC control of dynamic facades can provide significant electricity cost reductions and net load management capabilities of benefit to both the building owner and evolving electrical grid.</p>