Automatic Optimisation of Window Shading to Achieve Net-Zero Energy

Figure 1: Parametric optimisation of the solar shading for the same building in different climates. Green is the shading, White is the building volume, blue are the windows.

The internet is full of fancy visuals of useless parametric studies designed to wow clients rather than to draw design insights. In this blog post we present a simple framework on how to avoid producing fluff and get useful results from your computational optimisation. The article is based on a preliminary shoebox study on solar energy optimisation for a residential building located in the mountains of Saudi Arabia where two important climatic zero-energy building design strategies are to maximise solar heat gain in the winter and effectively shade the sun in the summer. For this task we chose parametric optimisation as a design generator due to the complex solution space and architectural constraints.

A good optimisation study should include following steps:

1.       Specify design objective and motivation 

2.       Concept sketching and qualitative analysis

3.       Specify boundary conditions

4.       Specify design parameters

5.       Optimise

6.       Visualise, conclude and communicate

1 | Specify design objective and motivation

This case study focuses on a residential building in the mountains of Saudi Arabia. Cold nights and hot days give energy designers an unique opportunity to design for minimal cooling and heating use, something that is a luxury when we typically design in a hot and humid climate. The crux of the simulation is to balance the heat gains during winter and summer through carefully designed solar shading. However, how much and where to shade is non-trivial when building thermal mass is included as a parameter.

Objective definition: Design solar shading such that a minimum of annual hours of active heating and cooling is needed. Heating is needed when indoor operative temperature falls below 20°C Cooling is needed when indoor operative temperature rises above 27°C

2 | Concept sketching and qualitative analysis

To understand the problem and verify solutions, a useful approach is to sketch your idea of what the optimal solution might look like. This is rarely pretty (with my drawing skills), but an essential step. This helps understand design constraints, realistic proportions and give a benchmark solution for the optimisation to beat. Any optimisation process should be held against a realistic benchmark to prove any gains, not just a worst-case assumption.

Figure 2: Hand sketches are important for understanding before simulation and cannot be skipped – You might even be able to make it prettier than me. My approach is printing out the solar path and building shape and then sketching what I believe is the optimal shading with green for shade and blue for window.

3 | Specify boundary conditions

Boundary conditions should be as close to a design situation as possible and should represent the underlying conditions which govern the design. This is often a challenge for interconnected systems such as energy design, but much easier for daylighting optimization. Even a preliminary study should at least incorporate typical local design traditions. In many cases some preliminary analysis is required to get realistic boundary condition values. At this stage, internal loads, operational schedules, space volumes, opening sizes, building envelope constructions, thermal mass and thermal night flushing with outdoor air is specified.

This case study used a high thermal mass reinforced concrete building with insulation on walls, roof and below slab. The building uses fans to get 6 air changes per hour (ACH) as night flushing during summers months. Double-glazing with a 50% solar heat gain coefficient is chosen. The benchmark shading is a simple 4 meter wide overhang above all glazed walls. Architecturally defined wall-to-window ratio is defined to preserve certain views on site.

Figure 3: Base Case design of building before optimisation

4 | Specify design parameters

The parameters of the study should be constrained to realistic values. These parameters should become apparent from the initial sketches. In this case 18 degrees of freedom are given by allowing X,Y,Z translation of 6 nodes on the solar shading, with the other nodes being fixed at the roof.

Figure 4: Degrees of freedom for the optimisation of the window shading  

5 | Optimise

The geometry is optimised using Honeybee as energy simulation interface to EnergyPlus and evolutionary optimisation engine Galapagos within Grasshopper in Rhino3d. Each iteration simulates the annual building temperatures hour-by-hour given a solar shade setup. The optimisation criterion is to minimise the amount of annual hours outside the thermal comfort zone of 20 – 27°C. Convergence was found after roughly 2000 iterations for 2 hours of computation time (Request for the nerds: EnergyPlus within Honeybee runs on a single CPU core by default. It should be possible to allow parallel simulations through scripting, please let me know if you have made it work).

Video 1: It is helpful to capture the optimisation process. Small videos are a good tool to impress and communicate the process

Figure 5: Parametric modelling for annual building simulation and automatic optimisation through grasshopper environment.

6 | Visualise, conclude and communicate

Studies only exist once they leave the computer. If the architect does not understand the outcome, the work is wasted. The communication can come in the form of the literal shape or the underlying design logic, which can be extracted from the simulation results. While the optimal design can look overwhelming and foreign at face value, important design lessons can be extracted, which guide the design further in the project. It is an important task for the energy designer and architect to understand this logic together. Once you understand why a solution looks the way it does it can be translated into buildable components. West facing facade: Large horizontal overhang protect against the afternoon sun, which is the most critical for overheating. South-western corner is for angles that open to allow winter sun to enter while blocking out the summer sun. To translate this the angles for a shading overhang would have to be replicated. East facing facade: Long overhang which becomes slimmer towards the south east corner. The long overhang on the north east corner shows a high need for sloar shading. In practical terms this could be replaced with a tree, a neighbouring building or other ways of shading a facade. The slimming towards the south east corner should be interpreted as allowing low angle winter sun to heat up the building in the morning. The angles should be incorporated into the final design South facing facade: Closed towards east and open towards west. The optimisation provides the outline for an organic shape which could be translated into a curved textile canopy in a final design or planting trees where the solar shading is closed.

Video 2: Rotational video of the optimal design. Most notable design feature is the slope of the solar shading around the south-western corner allowing specifically only winter sun to enter.

Interestingly, chosing different climates also give different simulation outcomes and different lessons for the energy designer about a certain climate. Without changing the boundary conditions we ran the model for a hot and temperate climate (Kuala Lumpur and Copenhagen) resulting in an extremly closed and open designs. Note that the energy model is tuned for the Saudi Arabia Climate, which will bias the simulation. Good luck on your optimisation!