2018, Volume 1, Issue 2
Department of Civil, Environmental and Architectural Engineering, University of Padova, Italy
In the last decades, the themes related to comfort simulation have gained a central role in the building process, requiring increasingly thorough analyses. In this field, multidisciplinary simulation-
Keywords: Building Information Modeling, Visual Programming Language, Performance Based Building Design, Optimisation Algorithms, Multi-
The achievement of indoor comfort conditions should be one of the main goals in a user-
In this paper, the definition of a workflow able to optimize jointly thermo-
2. Indoor environmental quality
There is a great potential for energy savings on air heating and conditioning when using the ideal Window Area (WWR) in offices, depending on orientation and glazing types. Energy efficient window design should limit both cooling and heating demands. The brief addressed by this study was to evaluate internal conditions in an office space that have large glazed area in Constantine city, and to define an efficient window in terms of heating and cooling. This work intends to provide guidance to building designers with regard to the thermal performance of office buildings.
The deep comprehension of the relation between the users and the indoor environment in a living space is a fundamental aspect in the building process. The internal comfort, in fact, influences the life style of people that use a building, affecting their attitude and mood.
The Italian technical standard UNI 8289 on internal comfort defines five classes of requirement to ensure wellness: thermo-
In order to simulate the dynamics of thermo-
The PMV index is a mathematical function that express the average vote given by a group of people in a scale from -
For the aims of this study, in order to perform optimization based on simulation and according to ASHRAE 55, a space is considered comfortable when:
2.2 Visual comfort
Visual or visual comfort is a subjective impression related to quantity, distribution and color of light and is reached when all the objects in a room can be seen clearly and the activities to be carried out in a space can be pursued without any visual effort. Internal lighting can be natural or artificial and usually a combination of the two is used to allow the exploitation of the buildings for the whole day. In a sustainable approach to design, however, it is highly recommended to reduce to the minimum the needs for artificial lighting to ensure both better feelings and energy saving.
This study will use as the reference the daylight factor (DF), defined as the ratio between the illuminance measured in a point lying on a horizontal plane inside a building E_i and the illuminance measured in the same moment in a point outside the structure under an overcast sky E_0, both measured in lux.
Since the value of DF changes in each point of the room, the average daily factor in taken in account.
In this case, according to the indication of the CIBSE Lighting Guide 10 (LG10-
However, without taking in account the problem of glare, a higher value is preferred.
3. State of the art
While the purpose of this study is to connect BIM, simulation and optimization processes in an integrated workflow, in this section the state of the art on the topics and the already explored relations between them is presented.
3.1 Optimisation based on simulation
In last decades the use of computer-
The use of optimization methods in the studies involving the field of constructions have shown a great increase only in the last decades (Nguyen, Reiter, & Rigo, 2014). In fact, thanks to a rapid technological evolution, they have become very popular in the academic world and have been applied to solve a wide range of problems such as the shape of the building, the design of the building envelope, the management of the HVAC systems and the generation of renewable.
3.1.1 Optimisation based on simulation
Optimization algorithms are usually classified on the basis of six pairs of different non-
Looking at optimization processes related to the AEC industry in literature, about 60% of the cases use the mono objective method (Evins, 2013). However, in the real case of profession, the designers are requested to optimize simultaneously the performances referred to different aspects very far between them and sometime even odds. Most of the difficulties come from the fact that often the functions at stake cannot be minimized (or maximized) simultaneously. Generally speaking, in fact, a solution that determines the optimum of one objective doesn't do the same for the others. For this reason, multi objective optimization (MOO) appears to be more appropriate, at the expense of computational lightness.
There are different ways to solve a multi-
In another approach, which is the one that will be developed here, the concept of Pareto efficiency is used. Here a series of optimal solutions is examined to identify later the best one. This method leads to identify for every problem the set of all efficient allocations, which define the Pareto frontier (Fig. 1).
In this curve all the optimum points can be found, reflecting the solutions for which it is impossible to improve further all the objectives, with no mention to the weight of criteria. Once the Pareto frontier has been defined, the best solution should be chosen by the designer in the light of different aspects with a decision making multi-
This class of algorithms is widely used in the AEC industry, and chosen in this research, for several reasons:
They allow to solve multi-
They are an effective method to manage discontinuities and highly constrained problems;
They show high success rate while combined with simulation.
Evolutionary algorithms (EA), are based on the theory of evolution, referring directly to the one published by Charles Darwin on his "The origin of species" (1859). They are a stochastic method of optimisation for resolving complex problems and are part of the wider category of model based on natural metaphors.
Genetic algorithms have been developed from evolutionary algorithms since 1975 by John Holland (Michigan University) and are their simpler subcategory. The functioning of a genetic algorithm can be subdivided into three parts (Tettamanzi, 2005). In the first phase, based on a random selection, an initial population of "n" individuals is chosen from the domain of the function. This set of elements constitutes the first possible solution to the problem, codified as a binary string and called chromosome. When the evolutionary cycle begins, firstly the operator of selection is applied, it simulates the Darwin's law of the survival of the fittest by applying a proportional selection based on the fitness value of each solution. While "n" parents have been chosen, the individuals of the next generation are generated by the application of recombination. In genetic algorithms two operators of reproduction, crossover and mutation, are used in order to change the genes of the solutions and explore new possibilities. Finally, the new generation of solutions replaces the previous one. The process is repeated x times until an acceptable approximation of the optimal solution or the maximum number of iterations is reached.
The use of evolutionary algorithms as a method to solve complex problems has both strengths and weaknesses and shows two principal vulnerabilities. Firstly, the convergence to the result is slower than with other optimisation techniques, at the point where the computation for the solution of some problems could even take days. Furthermore, due to their stochastic nature, this kind of algorithms do not guarantee the exact identification of the optimal solution, but most of the times they detect a good approximation, suitable to solve the problem. For these reasons the application of evolutionary algorithms is not suitable to every kind of problem, but they are particularly useful when the objective function is too complex to be rapidly maximized with non-
3.2 BIM tools for performance simulation
The achievement of the objectives related to internal comfort, in particular with a view to integrated performance-
A BIM is a building model based on data other than geometry that contains multidisciplinary information and includes all the links and the hierarchical relations between the elements. It is a shared digital representation of a construction with its physical and functional characteristics, based on open standard for interoperability. In this sense an informative model can be used as a central database able to communicate with external codes in order proceed performance analyses. While an informative model is intrinsically multidisciplinary, the real challenge is to link it to the processes of simulation and optimization, making the set of information stored on it available for a series of external platforms and readable by the algorithms involved in the calculations. A virtuous building process, in fact, should be characterized by an effective flow of information through different software tools to ensure the functionality of a cycle made of design, analysis and validation of each choice. In the field of energy simulation there are still lots of technical barriers that prevent the effective exchange of information between software and, thus, the achievement of an integrated multidisciplinary process between modelling and simulation (Zanchetta, Paparella, Borin, Cecchini, & Volpin, 2014). In the usual workflow the incompatibility between the tools leads to the need to define several times the same information in different platforms, making the design process more burdensome and highly error-
The application of algorithms in the field of energy building performance simulation began in the '70s, but only in more recent times the number of studies on the subject has really increased as demonstrated in "Improving the energy performance of residential buildings: A literature review". However, only few of them focus on the use of BIM to support the optimization process. For the purpose of this paper it is important to cite:
All those studies introduce a methodological framework useful both to mitigate the problems derived from the lack of interoperability between software and to link optimization processes to Building Information Modeling.
4. Methodology and tools
The set of tools implemented in the presented workflow are in a close relation that identifies them from the general to the specific. Starting from the authoring software, used in order to achieve the informative model of the building, a process based on Visual Programming Language (VPL) is activated and finally, within it, a series of plug-
4.1 Authoring software
One of the aims of this study is to connect BIM with algorithmic modelling in order to achieve a process of energy optimization while disposing of an informative model. This can be possible only by linking together a BIM software with a visual programming engine equipped with both environmental and optimization plug-
Actually, excluding the methods that involve IFC (Industry Foundation Classes), there exist three options to connect Archicad to Rhinoceros and Grasshopper:
Among the three, Grasshopper-
4.2 VPL environment
Visual programming language (VPL) is a simplified coding approach that helps user to design algorithms by manipulating graphic elements rather than writing text strings. Recently some CAD and BIM software provide internal Visual Programming interfaces, helping the professionals to define advanced design processes without the need to use scripting. Grasshopper (GH) is the VPL editor developed in 2007 by David Rutten and Robert McNeel & Associates and integrated in the NURBS modelling software Rhinoceros 3D. GH is open source, has a spontaneous attitude to interact with several external simulation tools and a wide number of add-
From a literature review, three studies very close to our purpose which link Rhinoceros with simulation tools thanks to Grasshoper, have been selected:
Despite their significance, it is observed that even though all these examples define a parametric workflow that involves performance simulation, they do not really exploit the access to all the information stored into an informative model, starting from a geometric more than an informative database.
4.3 Simulation plug-
To face the need of parametric design tools integrated with energy and comfort simulation engines, recently several studies applied to energy modelling have been developed, and most of them deals with Grasshopper (Jakubiec & Reinhart, 2011) (Lagios, Niemasz, & Reinhart, 2010) (Roudsari, Pak, & Smith, 2013). The common objective of these researches is to link a series of instruments to access performance based design processes in the early stages of the process, allowing to explore different design alternatives and giving in advance the results of performance analyses.
After a comparative analysis carried among several environmental plug-
4.4 Optimisation plug-
Galapagos and Octopus are two genetic solver that work with Grasshopper. Their application takes place especially in the first stages of the design process with the aim to define the parameters and the constrains of the project, and dispose of a preliminary representation of the related problems, in order to direct the designer toward the development of design alternatives that can be more suitable for the project requirements.
Galapagos, created by David Rutten, is an evolutionary solver used to develop processes of mono-
Octopus, as opposed to Galapagos, is a plug-
5. Case Study
The case study has been developed with an Italian engineering company, specialised in the field of engineering and project management for the building sector with special focus on sustainability and energy issues.
The object of analysis is a building designed to host offices and exhibition spaces, built on a single level with a rectangular footprint and characterized by glazed elevations.
Thanks to the combined use of the tools described in the previous paragraphs, an integrated framework for multi-
With reference to the case study the workflow has been divided into seven stages and, as it can be noticed, it starts and finishes inside Archicad in order to ensure the integration of the simulation and optimisation phases within a BIM-
Modelling of the building with the BIM authoring software Archicad
The first phase of the workflow consists in the informative modelling of the building. The BIM is an informative database, able to include all the multidisciplinary information useful for the design and the management of the construction, and it is the ideal starting point to activate different kinds of analyses on the building.
Definition of a simplified model suitable for the performance analyses
While the informative model of a building contains a plurality of multidisciplinary information, the one requested to implement energy and comfort analyses is a simplified version of it. The energy model should include geometrical and physical data related to building element, materials and spaces (Zanchetta, Paparella, Borin, Cecchini, & Volpin, 2014). In order to obtain it, a Model View Definition (MVD) is applied to the central model. This consist in a filter able to select the information that are relevant for a specific scope. For the purpose of this study a model able to satisfy simultaneously the requirement of a thermo-
Information exchange from Archicad to Rhinoceros
The flow of information is achieved using Rhinoceros Export Add-
Information exchange from Rhinoceros to Grasshopper and parametrization
The passage from the modelling software to the VPL environment is automated and don't cause the loss of any information.
Dynamic simulation and mono-
To better understand the results in this experimental phase of work it has been decided to carry out both the two processes of mono-
The dynamic analysis to determine the variation on thermo-
With the aim to activate a process of optimization linked to the model, some quantities have to be become parametric so that the calculation will be able to modify them and register the level of performance related to their state. In the case of this study the rate of glazing surface in relation to the dimension of the external walls is parametrized by the definition of a distinct variable for each orientation of the building. In this way, the result of the optimization process will show the ideal percentage of transparent area separately for the north, east, south and west elevation.
In order to make the model ready for the simulations all the properties of materials, building element and energy zones have to be set again because the ones coming from the informative model have been lost during the translation from Archicad to Rhinoceros. The integration of these data is implemented directly inside the VPL environment, thanks to the environmental plug-
Such as in the previous case, before the implementation of the lighting analysis some information have to be reintegrated with the method already described. In order to evaluate the internal visual comfort, the average Daylight Factor (DFM) is calculated. With this aim, inside Galapagos the objective function is represented by the DFM the objective function indicates that there are advantages only in the expansion of the glazed area. However, the optimization process, with its heuristic approach, will identify the minimal percentage of windows able to maximize the value of the average daylight factor. This means that values under the theoretical limit of 100% indicates that the maximization of DFM has already been reached and further enlargement of the openings will not improve the visual comfort.
Dynamic simulation and multi-
Starting from the same model that has been prepared for the two mono-
Information exchange from Grasshopper to Archicad
After the optimization stages have been processed and the ideal configuration has been chosen by the designer, the corresponding informative model have to be restored inside the authoring software. In this stage, by using the Live Connection tool, all the information related to the geometry are directly transferred into Archicad. With regard to the physical properties of material, however, a little work around has provided to be necessary, but in conclusion an effective BIM has been returned to the authoring environment.
In order to achieve thermo-
The results of the two processes of mono-
As it can be seen, in accordance with the hypothesis, the two sets of results identify a pair of very different optimal configurations due to the impossibility of maximize the two objective functions together.
At the end of the calculation 15 solutions belonging to the Pareto's frontier have been identified.
Starting from these results, and on the basis of the relative weights of the two objective functions, the designer can choose which configuration to select. In particular, from Figure 9 three alternatives are highlighted. Alternative n. 6 provides the highest value of DFM, but shows a poor number of comfort hours according to Fanger's model, on the contrary alternative n. 14 corresponds to the best thermos-
Conclusions and future works
Performance optimisation based on simulation while integrated in a BIM process shows a great potential in the field of comfort-
However, an effective integrated workflow is prevented by a series of obstacles fundamentally linked to interoperability. A BIM model, in fact, could contain all the data needed to activate multidisciplinary analyses, but at the state of the art the opportunity of a perfect exchange of information is not effective. To build a functional workflow with present tools, the problem must be discretized, broken down and then reassembled. In a solid integrated framework, the flow of information should be automated without any need of intervention by the designer in order to both prevent the possibility of errors and make the process accessible to a larger pool of professionals. Nowadays, in fact, the difficulties of operation typical of multi-
With this idea, then, the workflow could be extended to support more than two objective functions because, as explained above, the achievement of internal comfort would request the consideration of a wider set of aspects. Indeed, even though with the present tools the multi-
Alanne, K., Salo, A., Saari, A., & Gustafsson, S. I. (2007). Multi-
An, J., & Mason, S. (2010). Integrating Advanced Daylight Analysis into Building Energy Analysis. IBPSA-
Asadi, E., Da Silva, M. G., Antunes, C. H., & Dias, L. (2012). Multi-
Asl, M. R., Zarrinmehr, S., Bergin, M., & Yan, W. (2015). BPOpt: A framework for BIM-
Bahara, Y. N., & Nicolle, C. (2017). Building Energy Optimization through Thermal Efficiency Determination using Digital Mock-
Carlucci, S., Cattarin, G., Causone, F., & Pagliano, L. (2015). Multi-
Deb, K. (1999). Multi-
Diakaki, C., Grigoroudis, E., Kabelis, N., Kolokotsa, D., Kalaitzakis, K., & Stavrakakis, G. (2010). A multi-
Evins, R. (2013). A review of computational optimisation methods applied to sustainable building design. Renewable and Sustainable Energy Reviews, 22, 230-
Flager, F., Basbagill, J., Lepech, M., & Fischer, M. (2012). Multi-
Garber, R. (2009). Optimisation stories: The impact of building information modeling on contemporary design practice. Architectural Design, 79(2), 6-
Griego, D., Krarti, M., & Hernández-
Harmathy, N., Magyar, Z., & Folic, R. (2016). Multi-
Haupt, R. L., & Haupt, S. E., 2004. Practical genetic algorithms. John Wiley & Sons.
Jakubiec, J. A., & Reinhart, C. F. (2011). DIVA 2.0: Integrating daylight and thermal simulations using Rhinoceros 3D, Daysim and EnergyPlus. Proceedings of building simulation, 20(11), 2202-
Konak, A., Coit, D. W., & Smith, A. E. (2006). Multi-
Lagios, K., Niemasz, J., & Reinhart, C. F. (2010). Animated building performance simulation (ABPS)-
Lin, S. H., & Gerber, D. J. (2014). Evolutionary energy performance feedback for design: Multidisciplinary design optimization and performance boundaries for design decision support. Energy and Buildings, 84, 426-
Machairas, V., Tsangrassoulis, A., & Axarli, K. (2014). Algorithms for optimisation of building design: A review. Renewable and Sustainable Energy Reviews, 31, 101-
Nguyen, A. T., Reiter, S., & Rigo, P. (2014). A review on simulation-
Roudsari, M. S., Pak, M., & Smith, A. (2013). Ladybug: a parametric environmental plugin for grasshopper to help designers create an environmentally-
Tettamanzi, A. G. (2005). Algoritmi evolutivi: concetti ed applicazioni. Mondo Digitale, 2005(1), 3-
Welle, B., Haymaker, J., & Rogers, Z. (2011). ThermalOpt: A methodology for automated BIM-
Wright, J. A., Loosemore, H. A., & Farmani, R. (2002). Optimisation of building thermal design and control by multi-
Zanchetta, C., Paparella, R., Borin, P., Cecchini, C., & Volpin, D. (2014). The role of building energy modeling to ensure building sustainability and quality in a whole system design process. Recent Advances in Urban Planning, Sustainable Development and Green Energy (USCUDAR '14). 87-
Journal of Buildings and Sustainability -
INSIGHTCORE ® -