Artificial intelligence and skyscraper design

ARTIFICIAL INTELLIGENCE & SKYSCRAPER DESIGN ABSTRACT Skyscrapers and tall structures are known as complex structures that are difficult to construct in terms of both construction technologies and aerodynamic solution mechanisms. The main reason for the difficulty in skyscraper construction systems is the wind engineering concept first put forward by Alan Davenport and the destructive power of the wind on the structure. Due to its speed increasing as the wind rises in skyscraper systems, it creates problems for both the façade construction system and the main structural system of the building to be kept under control. The wind, known as "along wind force" and "across wind force", remains under the effect of a balance moment between the two effective forces that the building creates in the same direction with the wind and in the direction of 90 degrees angle with the wind, and in order not to fall against the destructive effect of the wind force, it maintains its own structural system and uses aerodynamic solution mechanisms. The wind load, called along wind force, which is in the same direction as the wind, is the wind dragging force. Due to the natural friction force created by the wind between the building to the building, this force creates a natural resonance on the structure and creates a "sway" oscillation effect on the upper elevations of the building. It is essential that the sway oscillation effect remains within acceptable and controllable limits. The Sway resonance effect also has a significant impact on the comfort needs of people living in the building. In regions where there is a strong effective wind load such as Colorado, the active sway resonance effect creates a damaging effect on the building material when it is effective on the structure for a long time. The long-term exposure of the material to the resonance effect causes intermediate capillary cracks in the inner tissue and affects the strength of the building material. Although skyscraper systems were built with steel tubular systems in the 1950s, active composite systems in which reinforced concrete and steel are used together are preferred in skyscraper structural solutions today. In this sense, the deformation caused by the sway resonance effect in the structure also depends on the structural parameter changes, such as whether the structure was built with a steel system or a composite system. The section built as reinforced concrete in the composite system will be more affected by the damage caused by the sway resonance effect. On the other hand, if primary deformation occurs less in the sections built as steel in the structure, the oscillation and displacement factor of the steel under the displacement force will be higher. Along wind force that creates the sway resonance effect, the secondary force that creates a resultant force with a 90 degree angle is the lateral wind load across wind force. Across wind force works to create a direct destructive effect on the structure. This system, which works in the direction of the lift lifting force, cannot create a harmonious sway effect like along wind force, which works in favor of the friction force, on the contrary, it causes the structural die of the structure to be damaged. The system, which tries to stabilize with the foundations of the building under these two effective forces, and which is based on both lateral and main wind loads, is exposed to the destructive force of the wind under an axial rotation effect called torsional moment effect. For this purpose, the structure needs to protect itself with the help of aerodynamically compatible form and high structures are basically shaped by 6 main aerodynamically effective methods. The first of these is the "taper" crystal sharpening technique, which allows the structure to become sharper as it rises. Taper technique is the most effective aerodynamic resonance technique known, which has also been used in "The Shard London" structure. In the simplest sense, it can be thought of as the process of evolving from a rectangular prism to a crystal pyramid. The second aerodynamic technique is damper systems, which are among the most preferred aerodynamic techniques. Damper systems are systems that work with the + moment – moment principle and work by creating an equal but opposite effect (power force) against the wind load created by the wind. Based on techniques such as testing dampers with different masses and sizes by differentiating metals with different density values such as hydro-elastic damper systems to water-based damper systems or tungsten and iron, damper systems have been used effectively in many buildings today, including the 432 park avenue structure. The setback aerodynamic technique involves the structure having units that are gradually reduced in blocks as they rise. Usually, symmetrical decrements in regular rhythmic units are preferred. In the Openings system, it is aimed to leave different openings on the structure that will create a wind passage corridor. The wind finds its way through these openings and loses its strength before it can create a tipping effect on the structure, and becomes weak and controllable. With Openings, the concept of leaving a gap is either seen as a single and large gap, usually close to the upper peak of the tower, as in the Shanghai Fiannce Center, or a few gaps are left in the form of a rhythmic sequential series of small gaps, creating a porous structure that is presented to the wind. "Twist, rotate" axial rotation systems is another known aerodynamic resonance technique that we are used to seeing in industrial chimney systems. It is formed by rotating the basic base geometry system of the structure on its own coordinate plane with certain angles axially in the z axis direction, with the z coordinate system being the height elevation value. For this purpose, Shanghai Tower is one of the best known twist techniques. The technique of creating corner modifications known as "Modify the edges" means creating a different set of angular changes at the corner points at each floor plan scale of the structure. In the logic of the Rotate technique, it is aimed to lose the wind power by changing the geometry of the wind on the basis of each floor, while the logic of the corner modification technique, which is called modify the edges, is to affect the friction force of the wind with the facade and ensure that it reaches controllable speed values. Prof. Alan Davenport, known as the wind wizard, explains the parameters that enable the wind to be actively effective on the structure, in the wind chain of Alan Davenport, which he designed and named after himself. In this ring, which consists of 5 parameters like the Olympic chain, the first link of the chain is defined as understanding what the character of the wind is in the region where the structure known as "local wind climate" is located. In the second link of the chain, there are criteria for what the building environment is known as "terrain surrounding". In this context, whether the building as a stand-alone structure is located in the middle of a large space or on the shore of the ocean, or whether it is located in the middle of a dense area within a metropolis, the structure will be driven by the effect of the surrounding buildings on the structure. What the floor heights are, changes the aerodynamics of the building by directly affecting the wind load on the building. In the third ring of the wind chain, there is the aerodynamic response quality, which is defined as the aerodynamic response. The aerodynamic response of the building to the wind is related to basic parameters such as what the building facade material is and what the building form is. What the building facade material is will affect the coefficient of friction, the building form and geometry are also the basis of aerodynamic optimization. Another link is the mechanical response quality as the 4th link of the wind chain. (mechanical response criteria) The mechanical response of the structure and its response to the wind are related to the structural solution system of the structure. When the reinforced concrete system is considered as a steel system or a composite system in which both are used, the response of the building to the wind varies according to the structural construction system. While steel construction systems show a more intense sway oscillation resonance effect against the wind, reinforced concrete construction systems create a more rigid form and create a damper-like effect. In the 5th ring of the wind chain, there are design criteria known as "design criteria". Here it is important to know what the main design parameters of the building are. For example, in the façade analysis of the building, deciding to what extent the minimal movement between the main frame skeleton system of the building and the façade mulion systems will be allowed, parameters such as flexibility coefficient and stiffness affect the resulting aerodynamic response quality of the building. In order to investigate the aerodynamic response quality as the 3rd basic parameter of the wind chain, the phenomenon known as the artificial intelligence LSTM technique – transfer learning – has been established. With this logic, in order to examine whether a solution algorithm that works in the solution of a known problem can also be useful in the solution of another defined problem, the operation logic of the aircraft wing tip optimization systems, which has been proven to be useful in aircraft wing optimization systems, is examined in the skyscraper building systems and skyscraper construction systems. A hypothesis has been put forward that it should be modified with the aim of creating an efficient aerodynamic form. With this support, it will be examined whether it is possible to reduce the need for damper systems, to reduce dampers, to minimize them, or to completely eliminate the use of dapmer as in The Shard London structure. In the thesis, which will consist of design and simulation stages, 3D visual designs of each skyscraper prototype whose design is considered, were first made in Rhino Grasshopper 3D environment, then mesh (mesh - grid mesh) was provided on the formed 3D visual prototypes and simulation from Matlab & Paraview / ANSYS from Paraview / ANSYS programs. It is aimed to make the decision of the most suitable simulation system according to the density of the point and the solution cell volume (grid mesh volume size) and to reach the conclusion through this appropriate program. In the examples examined in the literature research, the examples made by the Japanese Tanaka Tamura group were seen as the most closely related work with this thesis study, but the fact that the Japanese research group only included structures with symmetrical cross-sections in their prototypes and no studies were made on free plan diagram structures is a critical aspect. On the other hand, in studies on symmetrical plan diagram structures, simple plan diagram structures with full square or completely rectangular cross-sections have been found in the literature. In addition, in the analysis studies conducted in the literature, it has been observed that only the pressure changes on the basis of a single parameter are calculated over Bernoulli differential equations in CFD-based simulations. In the context of the thesis, the solution was provided on the k – omega turbulence model in the Navier Stokes equation, and it was concluded to which prototype the aerodynamically efficient model belonged, based on both pressure, velocity and vorticity calculations in 3 coordinate axes (x, y, z). F14 and F18 model prototypes gave the best matlab CFD simulation results in the trials carried out from F9 to F23 throughout the F series model prototypes. In the F14 model, the speed coefficient remained limited to 1.53, while the Cp pressure coefficient value gave the result of 1.64. In F18, the Cp pressure coefficient remained at 1.66 and the speed coefficient at 1.50. During the F series, the effects of angle and axial rotation were studied on the aircraft wing typology, which is similar to a right triangle modified at 4 different angles (between 50 degrees - 80 degrees), and in the next stage, on the double wing and single wing combination model prototypes in the D and E series. has been trained. While a speed coefficient of 1.70 was achieved in the E24 prototype, a pressure coefficient of 1.64 was achieved, the vortex intensity was obtained in this model (E24 in which single wing and axial rotation techniques were tested) and ¼ of the F18 model with double blade system and axial rotation. The vortex turbulence current intensity, which was measured at 48 units of density in F18, remained limited to 12 units of density in E24. When the X model prototype series was started, this time the coordinate orientation technique was activated in the Rhino 3D design environment and the starting point of the axial rotation coordinate was modified. In the coordinate orientations made from X1 to X7, the speed coefficient value reached 2.80 in the modified prototype design in the northeast direction in the X4 model, creating active turbulence, but the pressure coefficient value remained at 1.60. It is important that the vortex intensity remained at a density value of 10.92 units in the system formed under the K-omega turbulence model and vortex effect. Although the coordinate orientation technique caused an increase in the wind speed, it did not create a dangerous increase in the pressure value, but it provided results similar to the current turbulence intensity values in the D series and E series on the vortex current intensity. In order to improve the studies carried out by the University of Hertfordshire on aircraft wing systems, UH series geometries were obtained based on the skyscraper form of aircraft wing prototypes with 1-2, 1-3 and 1-5 layers parallel plate interface relationship. geometry systems were modeled in 3D and simulated in matlab CFD environment. Against the 1.78 speed coefficient obtained in the UH_1_3 prototype model series, 1.54 Cp pressure coefficient value and 9.22 (unit density) vortex current intensity value ensured that a result was ahead of all test tests performed before the UH series. Moving on to the M wing model series, the M basic, MT and MD model series were designed in 3D in the Rhino 3D environment. Among the prototypes, the details of which are clearly explained in the 4th chapter of the thesis, the M4 prototype remains at a speed coefficient of 1.31 against a pressure coefficient of 1.12 in the M basic model series. it is solid. The speed coefficient for the MT3 model was also obtained as 1.21. When switching to the MD model series, a speed coefficient of 2.11 was obtained against the pressure coefficient of 1.89 in the MD3 prototype. While a pressure coefficient of 1.73 and a speed coefficient of 1.78 were obtained in the simulation results of the GX1 prototype in the GX model series, which was created by modifying a symmetrical airfoil on the basis of pain and length, in the GY model series; A speed coefficient of 1.54 was obtained against a pressure coefficient of 1.65. In the final, two different series were designed using the concepts of openings and setback & taper, and it was aimed to compare the series based on the optimization of the aircraft wing with these two series. For these series, which are called Space and Willis series, 4 model prototypes from each group were produced and these models were solved using the k-omega turbulence model in the ANSYS CFD environment due to the density of the total mesh and number of nodes. As a result, the "Willis 3 – WL3" prototype on the wind tunnel model from which the surrounding buildings were extracted, was characterized as the most aerodynamic skyscraper form design among the 8 model prototypes in the group, with a pressure coefficient of 0.753 Cp. Among the SP1, SP2, SP3 and SP4 model prototypes produced as Space series, on the approach based on creating wind passage corridors by leaving regular rhythmic gaps in openings, the model that gave the most aerodynamically effective form design result was the SP1 model with a pressure coefficient of 0.95 Cp. As a result, the path created by following the steps of the data flow diagram, through the algorithm written in Chapter 3, and the use of the LSTM model for the purpose of transferring the aerodynamic efficient design principles from aerospace engineering to architectural design with a multidisciplinary perspective, and questioning its effectiveness in the Rhino 3D environment. Among all skyscraper prototypes F, D, E, X, UH, Mbasic, MT, MD, GX, GY, SP and WL series, the design that gave the best results was determined to be MT3 with a pressure coefficient of 0.7404 Cp. The design logic of the MT3 model is the form created by modifying the concepts of angle length – "camber angle – chord length" on NASA aircraft wing systems, which constitutes the main hypothesis of the thesis, and it questions the concepts of "wind corridor effect in the structure" and "rising tapering effect" as an operating mechanism, so important. The MT3 model created with the logic of aircraft blade optimization systems and the concept of creating a wind corridor has been proven in the thesis context as a new solution technique produced on artificial intelligence LSTM systems, going beyond the previously known and used openings – setback skyscraper aerodynamic modification techniques.