The science behind Metabuild

Our cloud platform was built around a new simulation-based optimization algorithm that was developed to meet the needs of the building industry.

Creating perfect buildings is about making sure that buildings remain fit for purpose for years to come, offer a high level of comfort, minimize environmental impact and keep operating cost low.

But considering the vast number of design options available today, there are billions of possible configurations for every building project. With today’s tools it impossible to test all these options and find the optimal configuration. That is why most experts rely on experience and gut-feeling – which results in sub-optimal building designs.

Metabuild’s algorithm was created to identify optimal building designs by testing thousands of design options for each project.

Truly holistic: All relevant building data is included in Metabuild

Metabuild connects artificial intelligence, BIM and building simulation

In artificial intelligence, an evolutionary algorithm is a subset of evolutionary computation. Evolutionary algorithms use mechanisms inspired by biological evolution, such as reproduction, mutation, recombination, and selection for complex optimization tasks.

Metabuild’s evolutionary algorithm was tailor-made to fit the needs of BIM and building simulation. Its unique features allowed us to create the first platform that connects evolutionary multi-objective optimization with the power of building simulation and cloud computing.

Following a hybrid optimization approach, Metabuild’s algorithm performs optimization in two phases: a deterministic and a stochastic one. The role of the first phase (the preparation phase) is to prepare a good collection of solutions and supply them to the second phase (the genetic algorithm phase) as an initial population rather than starting with a random sample. A good collection of solutions is prepared by using a single-objective sequential quadratic programming algorithm. This algorithm minimizes one objective function using the other one as a constraint.

Considering different constraint values, the minimization is repeated several times. According to diverse and non-domination concepts, the minimization results are sorted and a good collection of solutions is selected. A controlled elitist genetic algorithm is then used in the second phase. In order to avoid losing good solutions during the two-phase optimization, active archiving is used for storing the evaluated solutions. This two-stage approach is eliminating the negative characteristics of the random behavior of genetic algorithms and obtains optimal solutions with a significantly lower number of simulations.

Proven simulation engine

Metabuild uses EnergyPlus as a core simulation engine. EnergyPlus implements detailed building physics simulation algorithms for heat transfer, radiation, convection as well as air and moisture transfer, light distribution, and water flows. EnergyPlus has proven its effectiveness in thousands of projects and is widely accepted in the industry as being one of the most advanced building simulation engines available today.

Cutting-edge technology. Backed by science.

  • Our software was developed by leading scientists in the field of building simulation and optimization.
  • Studies on the impact of our algorithm have been published in the leading peer-reviewed journals.
  • Our partnerships with TU-Berlin and NTNU Trondheim enable us to constantly improve our software.
  • We’re actively collaborating with leading building industry experts through EU funded projects.