EP4302224A1

EP4302224A1 - Computer-implemented method and system for minimizing structural cost, maximizing free space and minimizing environmental impact in conceptual design of buildings

Info

Publication number: EP4302224A1
Application number: EP22709806.8A
Authority: EP
Inventors: Alper KANYILMAZ; Daniele Loiacono; Patricia Raquel NAVARRO TICHELL
Original assignee: Politecnico di Milano
Current assignee: Politecnico di Milano
Priority date: 2021-03-04
Filing date: 2022-02-25
Publication date: 2024-01-10
Also published as: IT202100005120A1; WO2022185170A1

Abstract

A computer-implemented method (1) for use in conceptual design of buildings comprises the steps of: generating (4) a random parent solutions population of size N, wherein each parent solution corresponds to a possible conceptual design of a building; calculating (6) dependent variables of each one of the parent solutions from the input variables and from the design variables; evaluation (7) of fitness functions; executing (8) genetic operators of a genetic algorithm configured for creating a child population solutions of size N starting from the parent population solutions; calculating (10) dependent variables of each one of the child solutions; evaluation (11) of fitness functions of the child solutions; combining (12) and ranking the parent and child solutions; selecting (13) the first N solutions of the ranking for the next generation; checking (14) a termination criterion, wherein if the termination criterion is met, the current combined solutions represent the population of solutions and, if the termination criterion is not met, the current population is considered the parents population and the process is iterated from the step (8) of executing genetic operator; wherein the fitness functions comprise the following steps: a cost evaluation step (71), a span evaluation step (72), an environmental evaluation step (73); and a solution sorting step (74).

Description

COMPUTER-IMPLEMENTED METHOD AND SYSTEM FOR

MINIMIZING STRUCTURAL COST, MAXIMIZING FREE SPACE AND MINIMIZING ENVIRONMENTAL IMPACT IN CONCEPTUAL

DESIGN OF BUILDINGS

Technical Field

The present invention relates to a computer-implemented method and system for minimizing structural cost, maximizing free space, and minimizing environmental impact in conceptual design of buildings.

Background Art

As known, the construction process in the building industry can be divided into three main stages: conceptual design, detailed design and construction stage.

The decisions taken at the first stage (conceptual design) drastically affect the rest, setting the direction of the project from the very beginning. Consequently, decisions in the early stages affect up to 80% of the resources involved in the project. Despite the importance of conceptual design, which has a major impact on the final cost and performance of the building, few resources and time are dedicated to it.

When starting a project, many design solutions exist for the given design constraints. Due to the impossibility of deeply evaluating them all, conceptual design is normally assigned to senior engineers who have the necessary expertise to select the best preliminary designs out of the existing alternatives. However, senior engineers are usually specialized in one single topic (e.g., steel design experts, concrete design expert, seismic experts, etc.) and they tend to choose the best solution inside their area of expertise. This specialization added to the fragmentation of the Architecture, Engineering and Construction (AEC) sector makes difficult to obtain coordinated and integrated solutions among all the involved parts.

With the introduction of Al in the engineering field some design assistants for conceptual design were developed. Most of them were created in the middle 80’ s applying Knowledge-Based Expert Systems (KBES). However, since the early 90’ s, researchers started to use Evolutionary Algorithms (EAs) to solve a large spectrum of engineering design problems such as pipe design or topological and shape optimization in structural design. EAs have been positioned since then as the best tool for this aim due to their exploration and optimization capabilities. They are based on the concept of ‘natural selection’ which states that the fittest individuals in a population survive.

Other Al tools such as Artificial Neural Networks (ANN) had also been applied to assist conceptual design. These methods try to find a relationship between the input parameters (design constraints) and the output (optimal design solution) learning through examples. Nevertheless, in order to achieve accurate solutions, a representative database of the design space is needed to successfully train ANNs, and this is not always the case during the conceptual design of buildings. Description of the Invention

The main aim of the present invention is to provide a method and a system for conceptual design of buildings that allows, at the same time, to minimize structural cost, maximize free space and minimize environmental impact in conceptual design of buildings with a reduced computational time.

The above-mentioned object is achieved by the present computer-implemented method for conceptual design of buildings according to the features of claim 1. Furthermore, the above-mentioned object is achieved by the present computer- implemented system for conceptual design of buildings according to the features of claim 13.

Brief Description of the Drawings

Other characteristics and advantages of the present invention will become better evident from the description of a preferred, but not exclusive embodiments of a computer-implemented method and system for minimizing structural cost, maximizing free space and minimizing environmental impact in conceptual design of buildings, illustrated by way of an indicative but non-limiting example in the accompanying Figures, in which:

Figure 1 shows the overall structure of the computer-implemented method according to the invention;

Figures 2, 3, 4 and 5 shows different panels of the output user interface of the computer-implemented method according to the invention.

Embodiments the Invention

With particular reference to Figure 1, the computer-implemented method according to the invention is globally indicated with reference 1.

According to a preferred embodiment, the computer-implemented method 1 is composed of two main modules:

- a Genetic Algorithm module 2 (GA module): includes the material selection, grid and building dimensions and the floor system decisions;

- a Lateral Load System module 3: determines the best lateral system to use, sizes it and locates it in the building floor plan.

The GA module 2 has the function of exploring the solution space to find the optimal building configurations in order to optimize different objective functions. To ensure that the algorithm works properly the design problem has to be adequately modelled. The problem modelling involves the definition of:

- The design variables and how to encode them into a chromosome structure;

- The dependent variables and how are they calculated;

- The objective functions and how to evaluate each individual.

A target problem when designing a building is difficult to establish since the initial specifications can vary from one project to another.

For this reason, according to a preferred embodiment the computer- implemented method 1 has three slightly different working modes with different design variables.

I. Fixed floor plan dimensions: The building dimensions in the two orthogonal directions and the number of storeys are fixed. n. Fixed floor plan area: The number of storeys and the needed area per floor are fixed. in. Fixed total area: Only the total area of the building is fixed.

The overall structure of the computer-implemented method is shown in Figure 1.

Initially, the method comprises a step 4 of generating a random parent solutions population of size N, wherein each parent solution corresponds to a possible conceptual design of a building.

Each created parent solution is defined by a specific combination of design variables.

Particularly, the design variables considered for the genetic algorithm are the ones needed to encode the conceptual design decisions and comprise: material selection, grids and structural layouts, and floor system.

According to a preferred embodiment, the material selection can vary between steel or reinforced concrete. However, the selection of different type of materials is not excluded (for example, timber).

According to the preferred embodiment, the material selection has been encoded as a binary variable where the value “0” represents steel and the value “1” represents reinforced concrete.

Regarding the grid and structural layout, two variables are necessary: The distance between columns (Span distance) in the two orthogonal directions of the floor plan (x and y). These variables are continuous since they can take any value, but they have been discretized for simplicity purposes to multiples of 0.5m. Minimum and maximum values for these variables have been fixed to 3m and 18m according to the feasible and practical range of span lengths in real building projects.

Finally, the floor system is another categorical variable that can take multiple different values to account for the different systems included in the project scope. According to a preferred embodiment, the floor system variable has been represented by an integer value between 0 and 5 corresponding to: composite beams and composite slabs with steel decking; cellular composite beams with composite slabs and steel decking; precast concrete units; one-way slab; two- way slab; flat slab.

In addition to these decisions, the dimensions of the building that are not fixed by initial specifications also need to be determined. These variables vary from one working mode to the other and in consequence, are different for each one of them. Particularly, in case of fixed floor plan dimensions, no additional variables are needed since the dimensions of the building are completely fixed.

In case of fixed floor plan area and number of storeys, the dimensions of the building in the X and Y directions are not defined. Nevertheless, the area per floor is fixed, and consequently, the two dimensions are closely related. Considering them as two independent variables can lead to solutions that do not accomplish the area specifications. Therefore, just one additional design variable is needed, and the X dimension has been chosen. The other dimension (Y) can be directly assigned using the area per floor.

According to a preferred embodiment, to ensure that the X dimension is a multiple of the span distance it has been represented by the number of bays in this direction. The variable is discrete and can take values from 1 to a maximum that depends on the total area.

In case of a fixed total area, the total number of storeys is not specified. The building configuration can result in both a low-rise building with a large area per floor and in a high-rise building with smaller area per floor. The number of storeys has been considered as an additional design variable. This variable is related to the area per floor through the total area. Once the area per floor is determined the problem is the same as in the second mode.

According to a preferred embodiment, the number of storeys is a discrete variable and its maximum and minimum values have been fixed from 1 to 20.

Furthermore, each created parent solution is defined by a specific combination of input variables are those specified by the user.

Input variables comprises the following groups of variables:

- Geometric dimensions;

- Material;

- Loads;

- Occupancy;

- Ground conditions.

Concerning the geometric dimensions, depending on the working mode chosen, different inputs have to be specified. For the working mode I, the inputs are the X and Y building dimensions and the number storeys. For the working mode II, the area per floor and the number of storeys. Finally, the input for the working mode III is the total area.

Apart from the building dimensions, a tolerance value is required. The tolerance represents the admissible error between the dimensions specified and the dimensions of the solutions proposed by the program.

The last of the geometric inputs is the floor-to-ceiling height that should be initially defined by the user depending on the building final usage.

Concerning the materials, the user has to specify the materials to be considered between steel and reinforced concrete (or other materials). The user also has to indicate the availability of lightweight concrete for in-situ concrete slabs.

Concerning the load, the imposed load (IL) should be determined and introduced by the user according to the intended use of the building and following the requirements of the EN 1991. Besides, a superimposed dead load (SDL) has to be introduced to take into account the permanent load of the of cladding, ceiling and systems.

Concerning the occupancy, the intended use of the building is needed as an input to determine the best lateral load system strategy. According to a preferred embodiment, two options are allowed: Residential or office. The residential option sets the program preference to cellular spaces whereas the office one sets the preference to an open plan.

Concerning the ground conditions, the user can introduce the bearing capacity of the soil, if this is a known value. If not, the bearing stratum type shall be indicated between cohesive, granular or rock. Then a submenu allows the user to indicate the stratum characteristics. The program uses a predefined bearing capacity depending on the specifications chosen. If deep foundations (piles) are required for any external conditions the user can also specify it, so the program will just consider this type of foundations.

Furthermore, the method comprises a step 5 of discarding and replacing by means of a repair module the parent solutions with geometric incompatibilities.

Particularly, in the genetic operators that involve randomization of the variables (Initialization and mutation), the repair module checks the geometric incompatibilities, and if variables are incompatible between them, then the randomization process is repeated.

In the case of crossover operator, if the combination of two genomes leads into an incompatibility, the incompatible values are modified in order to obtain a similar solution with compatible values.

This step avoids a big number of unfeasible solutions to be evaluated and thus, reduces the computation time.

Then, the computer-implemented method 1 comprises a step 6 of calculating dependent variables of each one of the parent solutions from the input variables of the user and from the design variables.

In this step the lateral system module 3 is configured for selecting the dependent variables related with the stability system.

Particularly, the calculated dependent variables comprise the following groups of dependent variables:

- Plan dimensions;

- Floor system and secondary beams;

- Beams and columns sizing;

- Vertical dimensions;

- Material quantities;

- Foundations’ type and sizing;

- Lateral load system.

Concerning the plan dimensions variables, the dimensions and geometric parameters that are directly dependent from others are calculated at this point to facilitate the fitness evaluation. Depending on the working mode these parameters could be already determined as design variables.

According to the solving mode I, the number of bays in each direction is calculated as the rounded division between the total dimension (X and Y) and the length of the bays (L_X and L_Y).

To be able to compute this error between the target and the real dimensions, the real dimensions are calculated and stored.

In the modes n and in, the dimensions cannot be computed in the same way as before since XNBAY is a design variable and not a dependent one. In addition, X and Y dimensions are not fixed as input variables. The total X dimension is calculated in the first place using the number of bays and the length of these ones. Then, the Y dimension is computed using the area per floor.

Once arrived at this point the number of bays in Y and the real dimensions lacking are computed as done for mode I.

Concerning the floor system variables, the floor system direction (d) is determined according to the type of floor.

The floor system direction can take three different values depending on if the slab load is transferred through the minimum grid size (min(L_X, L_Y)), through the maximum grid size (max(L_X, L_Y)) or both.

In the case of composite beams, the secondary beams are placed to span the longest direction. This way the computer-implemented method allows to obtain primary and secondary beams of similar size. In these cases, the slab transfers the load in the shortest grid dimension (d=0). In contrast, precast composite slabs are normally located to span the largest direction and thus, the load is transferred in the long dimension (d=1).

For concrete one-way slabs beams are placed in the largest dimension and the slab spans the shortest (d=0).

Finally, flat slabs and two-way RC slabs are bidirectional and transfer the loads in both directions (d=2). The values of the variable d depending on the floor type are specified in the following equation:

This variable is used by the computer-implemented method to determine the direction of the loads and to know which dimensions to use when computing the contributing area.

The slab depth (S_D) is essential to compute the slab load (S_L) and obtain the efforts over the beams.

Concerning beam and column sections variables, the computer-implemented method comprises a preliminary members’ sizing in order to ensure the structural feasibility of the building system. This step is also crucial in order to estimate later the material quantities.

Particularly, the method comprises calculating the actions over the structure using the following equation: where G stands for permanent actions, Q for variable actions and γG and γQ are the partial safety factors for load combinations.

For preliminary sizing only vertical loads have been included. The loads are supposed uniformly distributed over the beams. Vertical loads include slab self- weight (S_L), imposed loads (IL) and superimposed dead loads (SDL). The slab self-weight and the superimposed dead load are permanent actions over the structure as they represent the weight of permanent elements.

Permanent and variable loads are calculated as follows:

Once the combination of actions is computed, the specific loads over a single beam and column are calculated. The contributing area is determined according to the floor system direction.

In a conservative approach, all beams and columns have been designed as internal. Another consideration that has been taken for simplification purposes is that the section selected for the beams and columns would be the same for each of the storeys of the building.

The normal load over the columns is thus calculated as follows:

For beams, a uniform distributed load UDL is supposed. In the case of steel structures, the following formula is used to compute the UDL over secondary beams:

The load over primary beams is computed in the same way but adding the weight of the secondary beams according to the following formula:

Regarding the material, the partial safety factors indicated in the Eurocodes have been assumed.

Concerning the beam sizing variables, in the case of steel solutions, once the loads are estimated, restrictions over the strength in ULS conditions and displacement in SLS conditions are applied. These restrictions lead to a minimum moment and a minimum inertia needed. The beams are supposed simply supported. For ULS conditions the minimum required moment that the profile has to bear:

For SLS conditions the minimum inertia requested for the profile is:

Where δ_max represents the maximum displacement in beams for serviceability limit state and has been set to δ_max=L/250.

Concerning column sizing variables, for column sizing of steel elements, only ULS conditions have been checked since displacements are not significant. Stability phenomena have been considered using a safety factor.

For concrete elements, simple rules of thumb have been used.

Concerning vertical dimensions variables, the floor-to-floor height (h) of the building depends on the floor system used and the possible integration of the systems.

Concerning the material quantities variables, the weights or quantities of each construction material are important to calculate the overall cost of the project and its environmental impact.

The total steel weight is calculated taking into account the length of the beams and columns and the number of elements.

Particularly, the total steel weight is calculated by means of the following formula: where W_bs and W_bp are the weight per meter of the beams in each direction, w_c the weight per meter of the columns, L_s and L_p the length of the spans in each direction, SNBAY, PNBAY, XNBAY and YNBAY the number of spans in each direction, NS is the number of storeys and h the floor-to-floor height in meters. In case that braced frames are chosen as lateral load system, its weight is also computed. According to preferred embodiment, the bracing weight is calculated by means of the following formula: where X_NLLS and Y_NLLS are the number of reinforced bays in each direction.

The total steel weight is computed adding the previous weights by the following formula:

In the case of concrete, the unitary price is given per volume and thus, the total material quantity is calculated in cubic meters.

Volumes are calculated by means of the following formulas:

Where B_x and B_y are the height of the beams in each direction and C the width of the columns. If shear walls or concrete cores are chosen as lateral stability system, its weight is also computed.

The total concrete weight is computed adding the previous weights.

Finally, in the case of reinforced concrete, the quantity of reinforcement steel is also calculated using the reinforcement percentages of beams and columns specified in point D (Sections of beams and columns).

Concerning the type of foundation, four types have been considered: pad shallow foundations, strip shallow foundations, raft shallow foundations and deep foundations. According to the bearing capacity of the soil, specified by the user, and the building load over the foundations the area of foundation per column needed is computed. Geometric compatibilities are then checked, if the required foundation area occupies more than 2/3 of the distance between adjacent columns, then the foundation is chosen continuous in this direction. Following this criterion, a type of shallow foundation is chosen (Pad, strip in x direction, strip in y direction or raft). Finally, if the area needed for the foundation is bigger than the footprint of the building, deep foundations are chosen as the best option.

At this point, the computer implemented method 1 comprises a step of calculating the lateral load variables by the LLS module 3.

Particularly, the LLS module 3 is configured in order to execute three main steps:

- determining which system is the most appropriate;

- determining how many bays need to be reinforced;

- determining where should these bays be located.

Regarding the step of determining which system is the most appropriate, three main types of systems have been considered: Moment-resisting frames, braced frames and shear walls. Shear walls have been also considered in its two main

3D configurations: Shear walls and concrete cores.

Several factors affect the lateral stability system decision. The most relevant ones which have been included in the assistant are: Material, number of floors, building occupancy and beams size. These factors have been integrated into a decision tree that has been introduced in the program for the stability system selection.

Regarding the step of determining how many bays need to be reinforced, the number of reinforced bays is estimated using height-to-width ratios. The number of reinforced bays can be different in each direction since the bay lengths are different. Regarding the step of determining where should these bays be located, uniformity over height is supposed. In direction with a lower number of bays, all the exterior walls are reinforced. In the orthogonal direction, all the bays of the perimeter walls are reinforced except for the ones concurrent with the comer columns. For the rest of the lateral load systems a procedure to locate the reinforced frames has been developed according to the following criteria:

- Symmetry to avoid torsional effects.

- Bays as far away as possible to minimize the eccentricity of the load.

The procedure to locate the braced frames and shear wall systems is done independently for each one of the dimensions of the building.

Once all the dependent variables are computed, the computer-implemented method 1 comprises a step 7 of evaluation of fitness functions.

A multicriteria optimization approach has been chosen and it has been implemented using the NSGA-II algorithm.

Particularly, the step 7 of evaluation of fitness functions comprises at least the following steps:

- a cost evaluation step 71 ;

- a span evaluation step 72;

- an environmental evaluation step 73.

The minimization of the structural cost is an essential objective since it increases the benefit of the project. Moreover, cost optimization normally implies an indirect structural optimization by reducing the material quantities needed. Another important factor, especially for architects, is the maximization of free space. To obtain larger spaces, span distances have to be maximized. Setting this as an objective function is the way of introducing the preference of larger spaces to the assistant. Finally, the increasing importance of environmental assessment has been already highlighted. To enhance this aspect of the design, the last optimizing variable included is the minimization of the environmental impact.

Particularly, the cost evaluation step 71 comprises at least the following steps:

- calculating the cost of land; - calculating the cost of the structure;

- calculating the cost of joints;

- calculating the cost of floor;

- calculating the cost of foundations. According to a preferred embodiment, the cost of land is calculated by the following formula:

Where FP is the footprint dimension in m2, uc_land the unitary cost of the terrain in €/m₂ and C_land the total land cost in €.

According to a preferred embodiment, the cost of the structure is calculated as follows: where umc_material is the unitary material cost and ulc_materia t_lhe unitary labor cost. Concerning the step of calculating the cost of joints, although the connections between beams and columns have been considered pinned (Simply supported beams), if moment frames are established as the optimal lateral load system, the impact in the cost will be significant. In this case, the percentage of the joints that are fixed is estimated as follows:

The total connection cost is calculated as follows: where the unitary percentual cost of fixed joints is upc_fixed=0.6 and the unitary percentage cost of pinned joints upc_fixed=0.2. In the case of concrete, the cost of joints has not been considered since no additional material or workforce is needed.

According to a preferred embodiment, the cost of the floor system is calculated with the floor surface and the unitary floor cost per quadratic meter, by the following formula: where umc_floor is the unitary material cost of the floor system chosen in €/m² and ulc_floor the unitary labour cost.

For concrete slabs the material cost per quadratic meter is highly related to the depth of the slab and the reinforcement quantity.

According to a preferred embodiment, the cost of the foundation system is calculated with the foundation volume and the unitary foundation cost per cubic meter in the case of shallow foundations, by the following formula:

For deep foundations the cost is computed considering the number of piles needed and the transportation and installation cost, using the following formula: where uc_piles is the unitary cost of prefabricated piles, n_piles the number of piles needed and utCpiies the cost of transport and installation.

Finally, the total cost C can be calculated as the aggregation of the different split costs: where C_{lateral system} is the lateral system cost. According to a preferred embodiment, the span evaluation step 72 comprises calculating the span diagonal according to the following formula:

The environmental evaluation is performed using the emissions data from the construction materials.

Particularly, the step of environmental evaluation comprises calculating the emissions caused by the structure.

According to a preferred embodiment, the emissions caused by the structure are calculated as follows: where e_s and e_c are the total embodied carbon factors for steel and RC.

Furthermore, the step 73 of environmental evaluation comprises calculating the emissions caused by the type of floor.

According to a preferred embodiment, the emissions caused by the type of floor are calculated as follows:

The total emissions are calculated as follows:

Furthermore, the step 7 of evaluation of the fitness functions comprises a step 74 of sorting of the solutions according to the non-domination.

Following the step 7 of evaluation of the fitness functions, the computer- implemented method 1 according to the invention comprises a step 8 of executing genetic operators of a genetic algorithm configured for creating a child population solutions of size N starting from the parent population solutions.

Particularly, the step 8 of executing genetic operator comprises the following steps:

- a step 81 of binary tournament selection;

- a step 82 of crossover;

- a step 83 of mutation.

The mutation operator is applied over the offspring population once the crossover process has finished. It consists in the alteration of the value of one or more genes of the offspring population. In order to decide if an individual is mutated or not a mutation probability is set. In this care the mutation probability has been set up to 0.2 after the parameter tunning.

According to a preferred embodiment, said genetic algorithm is the NSGA-II (Non-dominated sorting genetic algorithm II).

Furthermore, the computer-implemented method 1 comprises a step 9 of discarding and replacing by means of the repair module the child solutions with geometric incompatibilities.

The computer-implemented method 1 further comprises a step 10 of calculating dependent variables of each one of the child solutions.

Once all the dependent variables are computed, the computer-implemented method 1 comprises a step 11 of evaluation of fitness functions of the child solutions.

It is pointed out that the steps 9, 10, 11 of discarding and replacing, calculating dependent variables and evaluation of fitness functions of the child solutions are implemented as the steps 5, 6, 7 of discarding and replacing, calculating dependent variables and evaluation of fitness functions disclosed above for the parent solutions.

Then, the computer-implemented method 1 according to the invention comprises a step 12 of combining and ranking the parent and child solutions according to non-domination and according to a crowded distance operator.

Then the computer-implemented method 1 comprises a step 13 of selecting the first N solutions of the ranking for the next generation.

Elitism is present at this point since parent and child populations are compared together.

Furthermore, the computer-implemented method 1 comprises a step 14 of checking a termination criterion.

If the termination criterion is met, the current individuals represent the population of solutions.

If not, the current population is considered the parents population and the process is iterated from the step 8 of executing genetic operator.

According to preferred embodiments, the termination criteria comprise at least one of the following:

- total number of iterations;

- number of iterations without better solutions.

However, other criteria are possible, for example it is possible the definition of a goal in terms of performance and number of solutions found by the algorithm. Furthermore, the computer-implemented method according to the inventions comprises a graphic user interface (GUI) provided with four main elements:

- an inputs framework configured to introduce input variables into the algorithm;

- an outputs framework configured to show to the user the algorithm results;

- a RUN button configured for executing the main algorithm; after pressing it, all the variables specified in the inputs’ framework are passed to the algorithm and it is executed. Once the algorithm has finished the outputs framework is actualized with the result values;

- an EXIT button configured to stop the program execution and close the window.

According to a preferred embodiment the inputs’ framework comprises at least the following tabs: design parameters, cost parameters, GA configuration and environmental parameters.

In the design parameters tab, the user has to specify the inputs required by the GA module and the lateral system module.

The cost and environmental parameters tabs allow the user to modify the constants used for the objective functions’ evaluation.

Finally, The GA configuration tab enables to modify the algorithm parameters. However, these parameters have been set up to their optimal values.

According to a preferred embodiment, the outputs framework comprises at least the following tabs: pareto front, solution summary, 3D representation and solution plans.

The Pareto front tab comprises a Pareto graph of the solutions and a toolbox to explore the graph (Figure 2).

The Pareto graph comprises a representation of the solutions as points in a 2D graph where the cost of each solution is represented in the ordinate axis and the span measure (Span diagonal) in the abscissa axis.

The abscissa axis values have been represented as categorical (Short, medium, and long span) to avoid misinterpretation of the values.

The Pareto graph further comprises a representation of the environmental impact by means of the color and size of the points.

Also, the Pareto graph comprises a shape differentiation for steel and concrete solutions to enable an easier interpretation of the results. Only the solutions with the penalty function equal to zero have been represented (Feasible solutions).

The solution summary, the 3D representation and the solution plans tabs are empty in the first instance. When exploring the Pareto graph if a solution is selected (clicked) the three tabs are fulfilled with the values for this solution.

The solution summary tab shows the design variables values indicating the essential information to define the conceptual design solution (Figure 3). The 3D representation and solution plans tabs show to the user a graphic representation of the solution (Figures 4 and 5).

The computer-implemented system according to the inventions comprises at least an elaboration unit configured for executing the steps of the computer- implemented method 1 disclosed above.

In practice it has been found that the described invention achieves the intended aim and objects.

In particular the computer-implemented method and system according to the invention allow, at the same time, to minimize structural cost, maximize free space and minimize environmental impact in conceptual design of buildings with a reduced computational time (less than 2 min per run).

The reduced computational time is due to: The NSGAII capabilities to explore the design space in a reduced amount of computation time;

The simplified approach chosen to ensure structural constraints that does not involve performing numerical simulations with an external software;

The deterministic approach chosen for the LLS module that does not involve performing numerical simulations with an external software to compute lateral loads.

Claims

1) Computer-implemented method (1) for minimizing structural cost, maximizing free space and minimizing environmental impact in conceptual design of buildings, characterized in that it comprises at least the following steps:

- a step (4) of generating a random parent solutions population of size N, wherein each parent solution corresponds to a possible conceptual design of a building;

- a step (6) of calculating dependent variables of each one of said parent solutions from said input variables and from said design variables;

- once all the dependent variables are computed, a step (7) of evaluation of fitness functions;

- a step (8) of executing genetic operators of a genetic algorithm configured for creating a child population solutions of size N starting from the parent population solutions;

- a step (10) of calculating dependent variables of each one of the child solutions;

- once all the dependent variables are computed, a step (11) of evaluation of fitness functions of the child solutions;

- a step (12) of combining and ranking the parent and child solutions;

- a step (13) of selecting the first N solutions of the ranking for the next generation;

- a step (14) of checking a termination criterion, wherein if the termination criterion is met, the current combined solutions represent the population of solutions and, if the termination criterion is not met, the current population is considered the parents population and the process is iterated from said step (8) of executing genetic operator.

- wherein said fitness functions comprises at least the following steps for each of said generated parent solutions and for each of said created child solutions: a cost evaluation step (71), a span evaluation step (72), an environmental evaluation step (73); a solution sorting step (74). 2) Computer-implemented method (1) according to claim 1, characterized in that each of said created parent solution is defined by a specific combination of design variables and input variables specified by the user.

3) Computer-implemented method (1) according to claim 2, characterized in that said design variables comprise: material selection, grids and structural layouts, and floor system.

4) Computer-implemented method (1) according to one or more of the claims 2 and 3, characterized in that said input variables comprises the following groups of variables: geometric dimensions, material loads, occupancy.

5) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that it comprises a step (5) of discarding and replacing, by means of a repair module, the generated parent solutions with geometric incompatibilities.

6) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that said calculated dependent variables comprises the following groups of dependent variables: plan dimensions, floor system and secondary beams, beams and columns sizing, vertical dimensions, material quantities; lateral load system.

7) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that said step (8) of executing genetic operator comprises the following steps: a step (81) of binary tournament selection; a step (82) of crossover; a step (83) of mutation.

8) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that it comprises a step (9) of discarding and replacing by means of a repair module the child solutions with geometric incompatibilities.

9) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that said termination criteria comprises at least one of the following: total number of iterations; number of iterations without better solutions.

10) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that it comprises two main modules:

- a Genetic Algorithm module (2);

- a Lateral Load System module (3).

11) Computer-implemented method (1) according to claim 10, characterized in that, in the steps (6, 10) of calculating dependent variables, said lateral system module (3) is configured for calculating the lateral load variables related with the stability system.

12) Computer-implemented method (1) according to one or more of the preceding claims, characterized in that it comprises a graphic user interface provided with a Pareto graph of the solutions, wherein said Pareto graph comprises a representation of the solutions as points in a 2D graph where the cost of each solution is represented in the ordinate axis and the span measure (Span diagonal) in the abscissa axis, a representation of the environmental impact by means of the colour and size of the points.

13) Computer-implemented system for minimizing structural cost, maximizing free space and minimizing environmental impact in conceptual design of buildings, characterized in that it comprises at least an elaboration unit configured for executing the steps of the computer-implemented method (1) according to one or more of the preceding claims.