TR2023013537A2 - PRE-APPLICATION SUSTAINABLE BUILDING ESTIMATION SYSTEM - Google Patents

Abstract

In the construction of sustainable buildings, the aim is to reduce the energy consumed by keeping the comfort of the users at the highest level in the interior and to use natural materials instead of materials that emit harmful emissions. All sustainable buildings are measured after completion and inspected to determine whether they provide energy consumption, carbon emissions or user comfort. In this case, when a building is planned, it cannot be checked whether the desired result will be obtained from the building before the interior space is used. Our invention is related to a prediction system that allows planning what a sustainable structure should be like before the intended structure is designed and deciding on the method that will consume the least amount of energy while the building is at the architectural design stage.

Description

DESCRIPTION PRE-IMPLEMENTATION SUSTAINABLE BUILDING ESTIMATION SYSTEM TECHNICAL FIELD In the construction of sustainable buildings, the goal is to minimize energy consumption by maximizing user comfort in the interior and to replace materials that emit harmful emissions with natural materials. All sustainable buildings are inspected after construction is completed, with measurements taken to assess their energy consumption, carbon emissions, and user comfort. Therefore, when a building is planned, it is impossible to verify whether the desired results will be achieved before the interior space begins to be used. Our invention relates to a estimation system that enables planning the sustainable building's intended form before the intended building is designed and enabling the determination of the method that will minimize energy consumption during the architectural design phase. PREVIOUS TECHNOLOGY The Passive House concept is a preferred method within sustainable construction, not because of its specific molds but because of its overall focus on energy efficiency. It reduces energy consumption without requiring additional mechanical requirements. In regions with hot climates, this significantly reduces electricity costs, particularly due to the need for air conditioning. In Türkiye and around the world, in regions with hot climates, the use of insulation materials is not considered necessary due to their high cost and the low heating load in buildings. In countries with hot climates, the high cooling load of hot climates results in increased use of mechanical systems such as air conditioners, refrigerators, and other systems, thus increasing the energy required. Therefore, the literature questions whether the energy-saving construction concept for cold climates is also applicable to hot climates. In this case, calculations and findings are presented for reducing energy needs and operating costs by applying the Passive House concept in hot climates. Our invention aims to guide the designer in architectural design within the framework of targeted parameters by applying the neuro-fuzzy logic method and achieve optimal results. This allows for the incorporation of energy-saving construction concepts into the planning process during the design phase, which is crucial for pre-determining and evaluating the factors that will affect the building's lifespan. The efficiency results of our invention demonstrate both the feasibility of the Passive House concept in hot climates and the potential use of artificial intelligence as a method for achieving accurate results in multi-criteria selection in architectural design. This concept, which will benefit both urban planning and individual building users, has been found to achieve efficiency levels of 50% in winter and 13.5% in summer for large structures like indoor pools. At smaller scales, this results in higher energy efficiency. Therefore, this AI-enabled system allows profits to be made even with low summer efficiency, especially for large-scale structures like indoor pools, which consume high amounts of electricity. This efficiency can also manifest itself in the form of natural gas or electricity consumption, depending on energy needs. Therefore, the Passive House concept, which reduces the consumption of natural resources, is easy to implement and adapts to nature. In today's buildings, the six parameters mentioned above are considered together, while building design is often based on consideration of landscape factors and snowfall in hot climates, while the profits of the user and the construction company are often overlooked. In this sense, achieving preliminary predictions of the most appropriate building design, aided by artificial intelligence, saves time and money both before and after construction. It has been concluded that reducing building operating costs, such as electricity, natural gas, and other expenses related to heating and cooling loads in summer and winter, can be achieved through the implementation of the Passive House concept. PURPOSE OF THE INVENTION Sustainable buildings include the construction concept known as the "Passive House." This construction concept utilizes natural methods and materials to maximize the building's harmony with nature. Passive Houses, which incorporate parameters such as natural ventilation, building orientation for natural heating, window/wall ratios for the most efficient use of lighting, the use of insulation materials according to climate conditions, and the use of shading elements to naturally reduce excessive light, were first implemented in Germany, a country with a cold climate. Over time, they have been implemented in other countries with cold climates and, due to their energy savings, are now being implemented in buildings of all functions. In the last 10 years, they have been implemented in buildings with warm climates, but measurements of the implemented buildings have revealed that they are not as efficient as structures in cold climates. Our invention aims to help determine parameters in warm climate regions by taking climate into account, predicting the appropriate structure before implementation, and facilitating decisions during the architectural design phase. The parameters that should be considered in buildings to be implemented in hot climates have been determined as follows: - Temperature difference between interior and exterior - Thermal conductivity of insulation (window frame, glass, wall, ceiling, floor) material - Presence/absence of natural ventilation - Presence/absence of shading elements - Building orientation (north, south, east, west) - Window-to-wall ratio. These parameters allow artificial intelligence to find the path that will minimize a building's annual energy needs. The output from these parameters is the desired and obtained data required for a Passive House. Obtaining this data is also referred to as training data for teaching artificial intelligence by collecting real data by comparing measured houses in the literature. To ensure accurate results were obtained from the learned data using artificial intelligence, a trial test was conducted and the results were compared. The data obtained for the trial test consisted of data not included in actual measurements, but whose results were known. This data, defined as test data, was used to test whether learning had occurred. First, the error value was measured, and the learned data was compared with the actual data. According to our invention, the training root mean square error (RMSE) value for the learned data was 0.17755. Because the error value was low, the learning of data sufficient to verify the result was completed. Thus, rules were created to reach the result within neuro-fuzzy logic, and through these rules, the optimal ratios were derived to meet the minimum energy requirement. Regression analysis was conducted to determine the accuracy of the results obtained, and information was obtained about the existence and strength of the relationship between the variables, and as a result, the R2 value was determined to be 0.9997 (99%). If we correctly apply the invention, considering the requirements of our own climate region in hot climates, it will minimize energy consumption and provide a change in the construction sector in terms of economics and unnecessary consumption of energy resources. LIST OF FIGURES Figure 1. ANFlS Error Graph Figure 2. Comparison of output obtained and estimated outputs in education DETAILED DESCRIPTION OF THE INVENTION Neural fuzzy logic is a method presented by synthesizing the learning ability of artificial neural networks with the decision-making ability of the fuzzy logic method. Fuzzy logic differs from other methods because there is no specific method for determining input parameters, such as by trial and error. Neurofuzzy logic, on the other hand, utilizes learning methods to determine parameters, adjust them, or define the structure. To illustrate this point, when setting the structure, we determine the number of variables to be calculated, the number of rules in the system, and the range of membership functions for the inputs. In the variable setting process, we determine the centers of membership functions, start and end points, and connection weights. Therefore, different learning types are applied at different layers within the neurofuzzy logic method. Learning methods in neurofuzzy logic systems, combined with fuzzy logic modeling principles, enable the development of a model that learns through training using a dataset. Combining fuzzy logic's robustness in working with linguistic variables and decision-making with the learning, flexibility, speed, and adaptability of artificial neural networks creates neural fuzzy logic. While fuzzy logic directly utilizes expert knowledge and experience, neural fuzzy logic, thanks to its learning capabilities, can classify and adapt to problems, and thanks to its decision-making capabilities, it can make decisions similar to humans. Fuzzy logic uses linguistic variables, while neural fuzzy logic uses linguistic and numerical variables. While rules are generated by machines based on data learning in neural fuzzy logic, rules in fuzzy logic are generated with the help of expert knowledge. Within the scope of our invention, we first investigate the factors that influence passive house requirements and energy efficiency, and then identify the factors that will affect outputs for their inputs. The aim of the neuro-fuzzy logic method is to find some inputs and outputs that will meet passive house requirements in buildings. In the first stage, the input for the outputs to be obtained with the neuro-fuzzy logic method was prepared. The neuro-fuzzy logic inputs used are as follows; - Outdoor temperature value - Total energy required in the building - Insulation U value - Shading element (present/absent) - Building orientation (South / North / East / West) - Window-to-wall ratio (PDO). The following outputs were obtained as a result of processing the data with the neuro-fuzzy logic method. There are many examples of window-to-wall ratio in the literature. Marino et al. (2017) conducted a study in Italy, demonstrating that the correct design of the building envelope, which should appropriately consider the climate of the building's location, is a crucial phase of the construction process. Xue et al. (2019) conducted a study in China on optimizing the window-to-wall ratio of sunshades to meet daylighting and energy-saving requirements. Considering hotel room views, they concluded that the window-to-wall ratio minimizes daylighting requirements while maximizing energy consumption. Studies have also investigated the effect of lighting on the window-to-wall ratio (Zolfaghari and Jones 2022) and the effect of the window-to-wall ratio on energy requirements for space heating and cooling, using applications in office buildings in Turin and Helsinki (Chiesa et al. 2019). The parameters used for these results are insulation level, building orientation, shading, and controlled natural ventilation. These real-world data obtained through a literature review were used as input. After preparing the data as input and digitizing the verbal data, rules must be specified for processing the input data, as shown in Table 1. The greater the number of rules, the more accurate and realistic the results. After defining the rules, the software hosted by the inventive system evaluates the inputs according to the rules and creates a model of the problem. These rules can be modified and monitored as needed, and new rules can be added while the system is operating. These additions and changes are made via an input device within the system. The device can be any device with a processor capable of running software, such as a computer, phone, or tablet. Input can be provided via any touch or mechanical interface device. No Rule 1 If (Temperature is low) and (Window U value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is south) and (Window to wall ratio is small), then (Annual Energy Requirement 2 If (Temperature is low) and (Window U value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is south) and (Window to wall ratio is medium), then (Annual Energy Requirement is medium) 3 If (Temperature is low) and (Window U value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is south) and (Window to wall ratio is large), then (Annual Energy Requirement 4 If (Temperature is low) and (Window U value is low) and (Ventilation is open) and (Shading is present) and (Building Orientation is north) and (Window to Wall Ratio is small) and (Annual Energy Requirement) In order to create the numerical equivalent of the data, the input data must first be normalized. The first step in preparing the collected data in two parts, training and testing, is to determine the numerical values of the inputs. The value ranges to be used for each input were normalized between 0 and 1 using a linear transformation through the software according to the formula 1 below. According to the formula applied for normalization, the results were given values between 0 and 1. XO - xml-n xOmax _ xmin Formula 1: A total of 200 normalized input data were prepared for the model, which was prepared with a total of six inputs and one output, 176 for training and 44 for testing. The next step for using the prepared inputs and numerical values is to determine the membership functions. The membership function is one of eight membership functions named triangular, trapezoidal, bell-shaped, Gaussian, Gaussian2, pi-shaped, dsigmodial, and psigmodial. There may be differences in the intended use of membership functions due to reasons such as whether the data on the line in the graph of these membership functions is an exact result or has a fuzzy value between 0 and 1 on the line. Membership function Formula Error value çg _ y trimf 0.0036261 Function Function www "M" `%1 Bell-shaped Membership gbellmf 0.0033403 Function WT" "M" `%1 Gaussian Membership Function (full gaussmf 0.0034226 symmetric) Function H y pimf 0.003715 Function Membership . n› _ dsigmf 0.0033692 o. Function (full symmetric) gg Membership psigmf 0.0033692 Table 2. Error values of membership functions In order to find the membership functions with the least error value and to proceed in that way, the training of the inputs to be used according to the function types and the comparison of the results are provided with the RMSE values given in Table 2. The performance evaluation criteria of the model to be created for the prepared inputs and their values are given in Formula 2 below. The root mean square error (RMSE) value shown is used; RMSE = Jüai - a1)2 + n - +(pn - (in)2 Formula 2. According to the comparison made, gauss2mf, which has the least error rate among them, was determined as the membership function to be used in the model creation process. According to Table 2, the RMSE value was determined as 0.0033174. Thus, in order to reach the outputs, the training process was continued by continuing with the Gauss2 membership function. As shown in the graph in Figure 1, the error magnitude decreases to 0.0033174 at the end of the 16th cycle using the Gauss2 membership function. To obtain results using the neural fuzzy logic method, an artificial intelligence method, the results were compared using two different data sets, separated into training and test sets, to determine whether the results were consistent during the machine learning process. Comparing the outputs of the training and test sets, the close relationship between the values and the actual values is shown by the overlapping circles and star shapes in Figure 2. In this case, it was determined that the correlation between the actual data and the outputs of the data predicted by the neural fuzzy network is strong. Test error Six inputs determined Training error value value Temperature / Insulation U value / Ventilation use - Wall ratio (PDO) Table 3. Input error value table The error value in the test data with the determination of the inputs is shown in Table 3 as 0.17755. After finding the error value of the trained data, some rules were generated. These rules are verbal expressions that show the effects of inputs on the output. The expressions mentioned are determined as "if" / "then". On a model with a total number of inputs (six inputs for our invention) and number of outputs (single output for our invention), different rules were added as many times as the number of intervals (there are three different intervals for our invention) for the determined membership functions, the number of intervals to the sixth power of three. In other words, during the invention's testing phase, a total of 36,729 rules were created in the model. This number of rules varies depending on the number of inputs, outputs, and intervals. For each output rule, the verbal values and expressions were determined differently. These values are labeled low-medium-high, low-medium-high, small-medium-large, and so on. When these verbal values and inputs are combined, the output rule is formed as low-medium-high. If the first of the rules is examined as an example, it considers the first two inputs: low outdoor temperature, the selection of insulation material to minimize heat transfer, and its evaluation. The rule that the annual energy requirement will be low is established by using open windows to allow natural ventilation in the space, using shading elements on the facade, orienting the building towards the south, and expressing the window to wall ratio as small. 1. lf (Temperature is low) and (Insulation U-value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is south) and (Window to wall ratio is small) then (Annual energy need is low) (1) 2. lf (Temperature is low) and (Insulation U-value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is south) and (Window to wall ratio is medium) then (Annual energy need is medium) (1) 3. lf (Temperature is low) and (Insulation U-value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is south) and (Window to wall ratio is large) then (Annual Energy Requirement is high) (1) 4. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is present) and (Building Orientation is north) and (Window to Wall Ratio is small) then (Annual Energy Requirement is out1mf4) (1) . lf (Temperature is low) and (Insulation U value is low) and (Ventilation is on) ant (Shading is on) and (Building orientation is north) and (Window to wall ratio is medium) then (Annual Energy Requirement is out1mf5) (1) 6. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is north) and (Window to wall ratio is large) then (Annual Energy Requirement is out1mf6) (1) 7. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is on) and (Shading is on) and (Building orientation is east) and (Window to wall ratio is Ratio is small) then (Annual Energy Requirement is out1mt7) (1) 8. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is on) and (Shading is on) and (Building Direction is east) and (Window to Wall Ratio is medium) then (Annual Energy Requirement is out1mf8) (1) 9. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is on) and (Shading is on) and (Building Direction is east) and (Window to Wall Ratio is large) then (Annual Energy Requirement is out1mf9) (1) . lf(Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Building orientation is south) and (Window to wall ratio is small) then (Annual Energy Need is low0) (1) 11. lf(Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Building orientation is south) and (Window to wall ratio is medium) then (Annual Energy Need is low1) (1) 12. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Building orientation is south) and (Window to Wall Ratio is large) then (Annual Energy Requirement is low2) (1) 13. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Orientation is north) and (Window to Wall Ratio is small) then (Annual Energy Requirement is a23) (1) 14. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Orientation is north) and (Window to Wall Ratio is medium) then (Annual Energy Requirement is low4) (1) . lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Building orientation is north) and (Window to wall ratio is large) then (Annual Energy Requirement is a25) (1) 16. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Building orientation is east) and (Window to wall ratio is small) then (Annual Energy Requirement is low) (1) 17. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Building orientation is east) and (Window to Wall Ratio is medium) then (Annual Energy Requirement is low?) (1) 18. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (Shading is half open) and (Direction of the building is east) and (Window to Wall Ratio is large) then (Annual Energy Requirement is a28) (1) 19. lf (Temperature is low) and (Insulation U value is low) and (Ventilation is open) and (No Shading is) and (Direction of the building is south) and (Window to Wall Ratio is small) then (Annual Energy Requirement is a29) (1) . lf (Temperature is low) and (Insulation U-value is low) and (Ventilation is on) and (Shading is not available) and (Building orientation is south) and (Window to wall ratio is medium) then (Annual Energy Requirement is medium0) (1) 21. lf (Temperature is low) and (Insulation U-value is low) and (Ventilation is on) and (Shading is not available) and (Building orientation is south) and (Window to wall ratio is large) then (Annual Energy Requirement is medium1) (1) 22. lf (Temperature is low) and (Insulation U-value is low) and (Ventilation is on) and (Shading is not available) and (Building orientation is north) and (Window to wall ratio is small) then (Annual Energy Requirement is 0middle2) (1) 729. lf (Temperature is high) and (Insulation U value is high) and (Ventilation is closed) and (Shading is absent) and (Building Direction is east) and (Window to Wall Ratio is large) then (Annual Energy Requirement is high) (1) Table 4. Example of rules for 23 out of 729 created. With the creation of rules, the output value becomes accessible. These results are estimated results generated by artificial intelligence. For the six inputs and results specified in Table 3, the output data is normalized and converted to values between 0 and 1. In order to calculate the output of these numerical data with neuro fuzzy logic, all data were normalized according to the formula applied, and the results were given values between 0 and 1. For normalization, the highest and lowest values of the input must be known. Therefore, with the help of real data, all inputs and outputs are converted to numbers between 0 and 1. The formula used is as follows; XO - xml-n xOmax _ xmin An example of obtaining the result after normalization can be shown as follows; the first input is temperature 0.83 (hot), the second input is 0.982 (high insulation U value), the third input is 0 (ventilation is on), the fourth input is 0 (shading element is present), the fifth input is 0.1 (building orientation is south), and the sixth input is 0.5 (window-to-wall ratio is 50%). In this case, the output value (annual energy requirement) was determined to be 0.508. Similarly, by holding some input variables constant and varying others, the resulting data will indicate the annual energy requirement. Regression analysis was conducted to compare the actual values provided during the training with the obtained results. Regression analysis is an analysis performed to provide information about the existence and strength of the relationship between variables. According to the obtained result, the R2 value was determined to be 0.9997. A high R2 value indicates good fit of the regression model. This indicates that the fit between the actual inputs and the inputs in the algorithm in question is very good. To obtain numerical results from verbal data, it is necessary to hold some values constant and examine the effects of variable inputs. Eight different comparison data are created and their differences are examined. Eight different methods for changes to be made to a building are listed below; . Measuring the effect of ventilation and shading on annual energy requirement in hot weather conditions. Measuring the effect of no ventilation and shading on annual energy requirement in hot weather conditions. Measuring the effect of ventilation and no shading on annual energy requirement in hot weather conditions. Measuring the effect of ventilation and no shading on annual energy requirement in hot weather conditions. Measuring the effect of ventilation and no shading on annual energy requirement in hot weather conditions. Measuring the effect of ventilation and shading on annual energy requirement in cold weather conditions. Measuring the effect of no ventilation and use of shading on annual energy need in cold weather conditions. Measuring the effect of presence of ventilation and no shading on annual energy need in cold weather conditions. Measuring the effect of no ventilation and no shading on annual energy need in hot weather conditions In this way, the differences between them and the changes that affect energy consumption can be measured and the correct building design can be achieved. If there are mandatory situations, preliminary information about what the result may be can be obtained using this method. Detailed sample results are given in Tables 5-12. The U values of insulation materials were matched with real materials and the results were evaluated. The verbal results obtained from the numerical data are as follows; The effect on output is significant when the temperature is very low or very high, but the effect decreases when the temperature is moderate. Poor or no insulation, depending on the thermal conductivity of the insulation material, has a significant effect on output, but good insulation halves the effect. There isn't much difference between a moderate and good insulation U-value. While there isn't much difference between open and closed natural ventilation, open ventilation requires more energy than closed ventilation. - There's little difference in the energy consumption between the presence and absence of a façade shading element, but the presence of shading increases energy consumption more than no shading. - There is no significant difference in energy consumption between building orientations, such as south, north, east, and west. However, numerically, it has been determined that buildings oriented north require more energy, while buildings oriented south require less. - PDO, the ratio of the window area to the wall area on a building's facade, results in a higher energy consumption. Findings from the graph indicate that a 10% window ratio is the lowest energy requirement, followed by a 90% window ratio. - Other inputs, along with temperature, reduce annual energy consumption. Temperature input is a significant input affecting the amount of energy used. - The second most important input is insulation. It has a significant impact on output, particularly when combined with shading and ventilation inputs. - Ventilation and shading inputs do not significantly impact output. It provides some change, but not as effective as insulation and temperature inputs. Findings indicate that energy needs can vary between buildings in cold weather conditions, depending on heating and cooling loads. Furthermore, the lack of ventilation reduces annual energy needs in both hot and cold weather conditions. Parameters for the training data and output data process are determined through a literature review. Assessments are being made using Passive House criteria to reduce energy consumption and support natural conditions for the efficient use of existing buildings. Data is collected to determine the inputs and determine the desired outcome. After determining the obtained parameters, the process begins by collecting data to determine the results by performing measurements in buildings with hot and cold climates. The step-by-step process is as follows: Collecting the measured values in real buildings. Separating some of the real data into training and some into testing to test whether the machine has successfully learned in neuro-fuzzy logic. If the trial fails during this phase, the machine continues learning. In the invention, based on the RMSE value obtained from the prepared real and test data, the machine has completed learning on the 16th trial and is ready to find the result. If the accuracy margin is high for the next trial, regression analysis is performed to confirm the accuracy margin (99% for the invention). After completing the machine learning process, neurofuzzy logic generated 729 rules. This aims to create verbal expressions that specify the conditions and obstacles associated with the desired outcome. This also reveals the parameters that affect the outcome. The data intended to be obtained with the result was generated by holding certain parameters constant. This way, the output is shown for each fixed value. By comparing the normalized data in the table with each other, it is determined which situations will have the greatest impact on energy consumption and which will have the least impact.

Claims (1)

1.