Perceptual evaluation of tall building forms and facades in urban environments

ABSTRACT

The study investigates the effect of the environment on the perception of tall buildings with the fractal analysis method. According to the hypothesis, tall buildings are appreciated to the extent that their fractal dimension value (fdv) matches the fdv of the urban environment. If the fdv of the buildings is higher than the environment, complexity and impression increase, and liking decreases. Suppose the fdv of the buildings are lower than the environment; complexity, impression, and liking decrease. In the first stage, 24 tall building models with the same height (200 meters) were designed in the Sketchup program according to the variables of forms (8) and facades (3). These models were added in two different environments. Both environments are located on Atatürk Boulevard, the main artery of Ankara city, Turkey. The fractal similarity ratios of the models were measured with ImageJ and its plug-in Fraclac. In the second stage, the reliability of the data was questioned with a questionnaire. The survey was conducted with senior students of Gazi University, Department of Architecture. The students were shown 48 images of the models and asked to evaluate these images according to liking, impression, and complexity variables. The study found that images with a high fractal similarity ratio were the most liked. When fdv was higher than the environment, complexity and impression increased, and liking decreased. On the contrary, complexity, impression, and liking decreased. As a result, it was found that there was a significant relationship between fdv and the perception and appreciation of tall buildings.

KEYWORDS:

• Tall buildings
• perception
• aesthetics
• urban environment
• fractal analysis
• urban model

Introduction

The demand for tall buildings is increasing day by day. Population growth leads to urbanization and demand for more housing and workspace. The limited urban space also increases the demand for tall buildings. At the same time, the fact that tall buildings are prestige and status indicators for cities and reflect the modern, economic, and technological development of cities keeps the interest in tall buildings alive. However, this increasing demand for tall buildings brings many vital issues, such as urban planning, infrastructure, environmental impacts, and social factors [1]. Tall buildings have many impacts on the urban environment. This study focuses on the effects of tall buildings on historical city centers. Since the 21st century, the demand for urban centers has started to increase. Since urban centers are mainly composed of historical buildings, tall buildings’ interaction with them has increased more than in the past [2,3]. This situation poses a risk for urban centers. When tall buildings are not developed in harmony with the urban environment, they threaten cities. Based on this problem, this study investigates how a tall building is compatible with the urban environment.

A limited number of studies in the literature evaluate the relationship between tall buildings and the urban environment. In Bovill’s survey in 1996, it was found that fractals were influential in the perception of the urban environment [4]. While people like environments with textural continuity, on the contrary, people do not like environments without textural continuity. In another study by Bovill in 1996, the relationship between the architectural texture of a traditional city and the natural environment was examined. Results were obtained about the connection and impact of local ecology on conventional architecture. According to this study, it was concluded that geology, topography, and local environmental character affect the design of buildings. In Heath, Smith, and Lim’s 2000 study, the perceptual status of a tall building cluster was analyzed according to silhouette and facade complexity variables [5]. The study concluded that the more complex the skyline, the more aesthetically pleasing it would be. As a result, evaluating buildings as a whole with their surroundings has led to significant differences in perceptual appreciation. In a study conducted by Stamps in 2002, the relationship between buildings and their surroundings was questioned through the silhouette line [6]. It was assumed that the buildings would be liked to the extent that the silhouette lines of the buildings and their surroundings overlap. In his study, Stamps showed nine computer-generated images to a total of 63 participants and questioned the aesthetic preferences of the participants. As a result of the study, it was determined that the participants liked the images in which the fractal dimension of the built environment and the elements in the background matched more. The study of Hagerhall et al.. (2004) found that individuals’ level of appreciation increased in built environments with dense landscape elements [7]. In other words, the components of the physical environment were influential on the level of appreciation. Similar results were obtained in Cooper and Oskrochi’s (2013) study [8]. It has been stated that there is a positive relationship between the streets’ landscape density and the users’ perception. In Lionar and Ediz’s study in 2020, the relationship between a building and its historical context was mathematically examined by the fractal analysis method [9]. In the study, it has been proved that the design fiction, which is discursively expressed to be related to each other, is also mathematically related. From these studies in the literature, it is generally understood that the perception and appreciation of a building should be evaluated as a whole together with its environment. The components of the built environment affect perception and appreciation positively or negatively. In the literature, only Heath et al.. (2000) examined the relationship between tall buildings and their surroundings [5].

The difference between this study and the studies in the literature is that similarity measurements were made over digital models. This is a method that has not been tried in previous studies. In the literature, measurements are usually made either through photographs [7,10–15], and [16] or architectural drawings [4,9,17]. Photographs have many physical components (landscape elements, urban features, traffic lights, vehicles, people, etc.) that affect the fractal dimension value. Therefore, urban photographs are not suitable for a reliable fractal measurement. Some studies have used architectural drawings to measure only buildings. However, this method is also inadequate due to the difficulty of accessing architectural drawings in urban scale studies. In Stamps’ study in 2002, where digital models were used in fractal measurements, topography and structures were modeled, but the evaluations were limited to silhouette lines [6]. In this study, a new dimension was brought to the fractal analysis method, and digital models were used for both the urban environment that forms the background of the buildings and the tall buildings. Thus, the effect of physical features on the fractal dimension value can be controlled.

The fractal analysis method is a method that examines the mathematical and geometric properties of fractal structures. Fractals are complex structures with unique scaling behavior. That is, the image of a fractal at a particular scale is similar to its image at smaller or larger scales. Therefore, fractal patterns are infinitely repeatable and have similarities at every level of detail. Due to these features, the fractal analysis method is applied in many fields. The fractal analysis method is used in architecture to understand the complexity of designs and structures, the variety of details, repeatable patterns, and scalability [7,18]. At the same time, fractals impact people’s perception and aesthetic appreciation. By nature, individuals perceive stimuli with fractal geometry more easily and quickly, and encoding the perceived information in memory is faster and easier. An individual’s sensory system can perceive fractal structures more effectively and comprehend them more efficiently [19]. Thus, the ability to easily perceive and quickly distinguish fractal structures positively affects the individual’s mental capacity and ability to be transferred to long-term memory [18]. Systemic hierarchy and complexity are also important parameters for environmental quality. In urban environments with high fractal dimensions, the visual quality of the space increases. Thus, more perceptible and livable urban environments are created [20]. Based on the effects of fractals on human perception and appreciation, this study focuses on tall buildings and their surroundings. The content of the study consists of the impact of the facades and forms of tall buildings on urban environments. The fractal analysis method is used to determine the similarity ratios of the buildings with their surroundings, and a survey study supports the consistency of the results. In the study where statistical analysis methods were used, correlation and regression analyzes analyzed the relationship between the values, and Anova analyzed the semantic differences between the values. Fractal measurements were calculated with the ImageJ program and its plug-in, Fraclac.

Material and method

Hypothesis 1:

The building is appreciated to the extent that tall buildings’ fractal dimension value matches the surrounding buildings’ fractal dimension.

Hypothesis 2:

When the fractal dimension value of tall buildings is higher than their surroundings, complexity, impression increase, and liking decreases. On the contrary, when the fractal dimensions of the tall buildings are lower than their surroundings, complexity, impression, and liking decrease.

To determine the validity of these hypotheses, tall building models and the regions where these models will be located were designed. First, a form category was created to design tall building models using the form classifications in the literature [21].

Vollers (2008) provides a comprehensive database for classifying tall building forms in the literature. Vollers’ classification is inspired by parametric modeling tools. In solid modeling software, volumes are shaped by commands like Shear, Twist, Scale, Unite, and Merge [22]. In Taghizadeh and Seyedinnoor (2013), the form classification of tall buildings is shaped according to sustainability and energy efficiency principles. Another factor that is effective in shaping tall buildings is aerodynamic modifications [23]. Studies in the literature have found that interventions in building form reduce the effect of wind [24-27]. In Tanaka et al.‘s study (2013), aerodynamic forms were developed and classified into seven categories: basic models, models with corner modification, curvilinear models, tapering models, spiral models, models consisting of openings, and composite models that combine several features [28]. The study by Ilgın and Günel in 2021 is a summary of the studies in the literature, and tall buildings are classified into six categories: prismatic form, curvilinear form, tapering form, layered form, helix form, and free forms [29]. This study identified eight form categories: simple form (covers square and rectangular buildings), circular form, hyperbolic form, tapering form, layered form, opening form, conoid form, and twisted form. After determining the form category for tall buildings, building models were designed. Each building was created in the Sketchup Program with the same floor area (30 meters * 30 meters) and height (200 meters). After the forms were designed, each building was reproduced according to the facade variables (simple-medium-complex). Thus, 24 tall building models (8 forms * 3 facades) were obtained (Figure 1).

Figure 1. Tall building Models Designed According to Form and Facade Variables (drawn by authors).

After the tall building models were developed, the lands where these structures would be located were determined. Two contrasting regions suitable for fractal measurement were identified. To compare fractal dimension values, the environments of low-rise historical buildings and tall modern buildings constitute an excellent sample area. In this context, Atatürk Boulevard, which forms the backbone of Ankara, is a suitable sample area for the study as it hosts historical and modern high-rise buildings. The study focuses on the Ulus District and Kavaklıdere District, two important Atatürk Boulevard sub-districts. In these two contrasting districts, Ulus is characterized by many low-rise, elaborate, and ornamented historical buildings from the early Republican period. At the same time, Kavaklıdere is home to modern high-rise buildings in the International Style, which began to develop after the 1950s. After the regions were determined, the lands where tall building proposals would be placed were also identified. For the Ulus region, the land is located at the intersection of Atatürk Boulevard, Cumhuriyet Street, and Anafartalar Street, which is planned to be demolished and hosts the 100th Year Bazaar, was chosen (Figure 2).

Figure 2. Land in the Ulus Region and Its Surroundings (drawn by authors).

In the Kavaklıdere region, the vacant land between the Vakıflar Genel Müdürlüğü and Celal Bayar İş Merkezi on Atatürk Boulevard was chosen (Figure 3).

Figure 3. Land in the Kavaklıdere Region and Its Surroundings (drawn by authors).

In the first experimental part of the study, fractal measurements of the buildings that make up the urban environments (20 buildings in Ulus and 13 buildings in Kavaklıdere) and tall building models were carried out. The average fractal dimension value of both urban environments and the fractal dimension values of each tall building were calculated, and according to these values, the similarity ratios of tall buildings to urban environments were determined. Ostwald and Vaughan (2009) developed a scale for the relationship between the fractal value of the building and its surroundings [30]. It was suggested that the maximum difference between the fractal dimension values should be 4% for the building and its surroundings to be considered similar, and the maximum difference should be 1% for it to be considered very similar. In this study, similarity ratios were determined based on the scale developed by Ostwald and Vaughan. In addition, ideal fractal dimension values used in evaluating structures in the literature were determined. While structures with values of 1.1 and below are monotonous and boring, structures with values of 1.8 and above are very complex [17]. Values between 1.1 and 1.5 refer to structures with little detail diversity, while values between 1.6 and 1.9 refer to detail richness [31]. Values of 1.3 and above are preferred in some studies; ideal values are between 1.2–1.6 [32,33]. In light of the data in the literature, the ideal values in this study are considered as values between 1.3 and 1.7; structures below 1.3 are considered monotonous and simple; structures above 1.7 are considered very complex.

The validity of the data obtained in the first study is analyzed in an experimental study. A survey was conducted with the 4th year students of Gazi University, Department of Architecture. The images were shown to the students in the classroom, and they were asked to evaluate each image according to the dependent variables of liking, impression, and complexity. The results of the study are evaluated with statistical methods in the SPSS Program. The relationship between the independent variables (form, facade, and environment) and the dependent variables (liking, impression, and complexity) was evaluated. Correlation and regression analyzes were conducted between the variables. Anova Test was also performed for each variable containing semantic differences.

Results

Table 1 shows tall building models’ similarity ratios (SR) according to urban environments. In both Ulus and Kavaklıdere districts, medium façades are most similar to the urban environment. Among the medium complex façades, hyperbole form (1%), tapering form (1%), and layered form (2%) have the closest similarity with the Kavaklidere District, respectively. In Ulus District, circular form (2%), conoid form (2%), twisted form (3%), simple form (4%), and opening form (4%) have the closest similarity rates with the urban environment. Regarding ideal fractal dimension values, values between 1.3 and 1.7 are considered ideal. Medium facades have the most ideal values in the table when evaluated in terms of ideal fractal dimension values. Therefore, since it is foreseen that the data obtained will overlap with individuals’ perception and appreciation preferences, a survey was conducted in the second stage of the study.

Table 1. Similarity ratios of tall buildings according to Ulus and Kavaklıdere Regions.

Forms		Simple Form	Circular Form	Hyperbol Form	Tapering Form	Layered Form	Opening Form	Conoid Form	Twisted Form
Simple Facade		D=1.07	D=1.07	D=1.08	D=1.08	D=1.1	D=1.08	D=1.17	D=1.17
Ulus	SR	%61	%61	%60	%60	%58	%60	%51	%51
Kavaklıdere	SR	%52	%52	%51	%51	%49	%51	%42	%42
Medium Facade		D=1.64	D=1.70	D=1.60	D=1.60	D=1.61	D=1.64	D=1.66	D=1.65
Ulus	SR	%4	%2	%8	%8	%7	%4	%2	%3
Kavaklıdere	SR	%5	%11	%1	%1	%2	%5	%7	%6
Complex Facade		D=1.84	D=1.82	D=1.78	D=1.78	D=1.79	D=1.84	D=1.85	D=1.84
Ulus	SR	%16	%14	%10	%10	%11	%16	%17	%16
Kavaklıdere	SR	%25	%23	%19	%19	%20	%25	%26	%25

The data obtained from the classroom interviews between 30 November 2022 and 15 December 2022 with 52 senior architect candidates, were transferred to an Excel table and then analyzed with statistical methods in the SPSS analysis program. Students were shown 48 images in which tall building models were added to urban areas and asked to evaluate these images in terms of complexity, impression, and liking.

Of the students who participated in the survey, 65% were female students, and 35% were male students. Women were less affected by images than men. The degree of complexity of images is at the same level for both women and men. Women’s liking of the images is lower than men’s. Therefore, according to the study data, men had a more positive approach to interpreting images than women. The degree of complexity of images is at the same level for both women and men. Women’s liking of the images was lower than that of men. Thus, men had a more positive approach to interpreting images than women.

The reliability analysis of the questionnaire study was made with the SPSS program. Statistically, this measure is called Cronbach’s Alpha coefficient. The alpha coefficient is between 0 and 1; the higher the value, the more reliable the scale is. Reliability analyzes for each dependent variable (impression, complexity, and liking) were analyzed in detail (Table 2).

Table 2. Reliability values of the variables.

Variables	Scale reliability
Impression	0.95
Complexity	0.89
Liking	0.90

The study analyzed the proportional relationship between the participants’ answers to the dependent variables of liking, complexity, and impression and the fractal dimension values of the models by correlation and regression methods (Table 3). In the correlation evaluation, the correlation coefficient is between 1 and −1. A value of 1 indicates a strong positive relationship, 0 indicates no relationship, and −1 means a strong negative relationship. Regression analysis tries to determine how the dependent variable depends on one or more independent variables and how strong this connection is. When the regression between the impressiveness variable and the independent variables is analyzed, it is seen that there is a strong relationship between complex facades, opening form, twisted form, and impression. When the regression between the complexity variable and the independent variables is analyzed, it is understood that there is a strong relationship between complex facades, layered form, opening form, conoid form, twisted form, and complexity. When the regression values of the liking variable are analyzed, it is seen that there is a strong relationship between medium facades and liking. When the correlation of the dependent variables with fractal value is analyzed, liking has the highest value at 0.45, followed by complexity at 0.28 and impressiveness at 0.12.

Table 3. Correlation and regression values of the variables.

Variables	Impression		Complexity		Liking
	Estimate Value	Pr (>|t|)	Estimate Value	Pr (>|t|)	Estimate Value	Pr (>|t|)
FA	−0.03009	0.802403	0.431674	1.81e-06***	−0.068478	0.426664
Region 2	0.20112	9.92e-06***	−0.008013	0.81417	0.141827	1.37e-05***
Facade 2	0.31558	0.188219	−0.503314	0.00519**	1.709771	2e-16***
Facade 3	0.53130	0.002750**	1.145656	2e-16***	0.069427	0.584654
Form 2	0.19872	0.028797*	0.439103	1.42e-10***	−0.214744	0.000983***
Form 3	0.21755	0.017539*	0.309069	7.12e-06***	−0.237198	0.000304***
Form 4	0.34576	0.000162***	0.180864	0.00851**	0.157032	0.016717
Form 5	0.36218	8.10e-05***	0.697544	2e-16***	0.004099	0.950279
Form 6	0.66707	2.84e-13***	0.821167	2e-16***	0.109887	0.091521
Form 7	0.38823	2.52e-05***	0.589225	2e-16***	−0.065501	0.320361
Form 8	0.68670	1.34e-13***	1.131546	2e-16***	−0.192793	0.003565**
Correlation coefficient	0.1227677		0.2821997		0.4566783
P-value	7.569987e-10		6.429328e-47		7.478467e-129

In the study, ANOVA was applied to reveal the semantic difference of each variable. The mean values of the dependent variables (impression, complexity, liking) were correlated with the independent variables (fractal analysis, region, facade, form). These ANOVA analyzes are shown in Table 4. p values in the Pr(>F) column indicate whether the effect of the independent variables on the dependent variables is statistically significant. The stars () in the output indicate the significance of the p-values. For example, if ’**’ is represented by three stars, the p-value is less than 0.001, which is statistically significant. Accordingly, as seen in Table 4, the variables contain significant differences due to ANOVA analysis. The effect of the façade variable on the dependent variables is exceptionally high compared to the other variables.

Table 4. Anova values of the variables.

	Df	Sum Sq	Mean Sq	F value	Pr (>F)
Fractal Value	1	27.8	27.76	76.946	2e-16***
Region	1	8.5	8.54	23.670	1.22e-06***
Facade	2	336.3	168.15	466.043	2e-16***
Form	7	21.7	3.10	8.589	1.93e-10***

Accordingly, Figure 4 shows that the middle facades are the most liked, while the fractal similarity ratios have the highest values. Thus, the study’s first hypothesis, ‘tall buildings are liked to the extent that their fractal dimension value matches the fractal dimension of the surrounding buildings’, is confirmed. Medium facades have the highest similarity ratio and have the highest appreciation values. The appreciation values of simple and complex facades are close to each other and have low values. The study’s second hypothesis is ‘when the fractal dimension value of the tall buildings are higher than their surroundings, complexity, and impression increase and liking decreases. On the contrary, when the fractal value of the tall buildings is lower than their surroundings, complexity, impression, and liking decrease’ is also confirmed. The fact that the fractal dimension value of tall buildings is higher than their surroundings is possible due to the complexity of the building façade. The building facade becomes complicated; complexity and impression increase while liking decreases. The fact that the tall buildings’ fractal value is lower than the environment is possible with the simplification of the building facades. The façade becomes simple; complexity, impression, and liking decrease. As complexity increases, liking first increases reaches saturation at medium façades, and then decreases. An inverted ‘U’ relationship between liking and complexity confirms Berlyne’s theory (1974) [34]. According to his theorem, satisfaction increases to a certain level as complexity increases and then decreases. Accordingly, a medium level of complexity shows maximum satisfaction, while the least complex and the most complex situations show the least satisfaction. Impression increases linearly as complexity increases. However, the most affected images are usually not the most liked. Berlyne also stated that the aesthetic appearance of a texture will vary depending on the stimulus and non-stimulus states of the values. In parallel with this situation, an increase in stimulation or a decrease in overstimulation will bring happiness.

Figure 4. Relation of the facades with the dependent variables (drawn by authors).

Figure 5 shows the relationship between each form variable and the dependent variables. It is seen that the forms take close values in terms of fractal similarity ratios. The forms become complicated; the fractal similarity ratio partially increases. This result is related to the fact that the forms generally have low values (mean ∑D = 1.10, minimum ∑D = 1.07, maximum ∑D = 1.17). Regarding the impression variable, the forms become complicated as the impression increases. The least impressive form is the simple form. The most impressive forms are the opening form and the twisted form. There is an increase in the graph regarding the complexity variable, but it is not a regular increase. The simplest forms are the simple form and the tapering form. The most complex forms are the opening form and the twisting form. The results are pretty different when the graph regarding the liking variable is evaluated. It is seen that there are differences between impression and liking variables. According to the graph, the most admired form is the tapering form. The simple form and the opening form follow it. The least liked forms were hyperbole and twisted forms. When the graph is analyzed regionally (Figure 5), fractal values are almost identical for simple, opening, and twisted forms. The fractal similarity rate for circular, hyperbole, tapering, layered, and conoid forms differed in both regions.

Figure 5. Relation of the forms with the dependent variables (drawn by authors).

Regarding fractal value, the difference between the form values in both regions does not give a significant result. When the graph is evaluated in terms of impressiveness, tall building proposals in Kavaklıdere are more impressive than those in Ulus. This situation is related to the fact that tall building construction overlaps more with the Kavaklıdere region, which consists of multi-story buildings. In terms of complexity, the values are close to each other, and the values are almost the same for the simple form, tapering form, layered form, and conoid form. The values differ in complexity in circular form, hyperbole form, opening form, and twisted form. When evaluating the graph regarding the liking variable, Ulus has lower values than the Kavaklıdere region. Tall building proposals in Kavaklıdere were more appreciated. This result supports the previous one and is related to the fact that tall building proposals are not adopted in Ulus, surrounded by low-rise historical buildings. Therefore, although the fractal similarity ratio influenced the results, different qualitative characteristics (height compatibility between buildings) also influenced perception.

Conclusion

The study hypothesizes that ‘the building will be appreciated to the extent that the fractal value of a tall building matches the fractal value of the surrounding buildings’. Studies in which the environment impacts the perception and appreciation of buildings have been examined in the literature. In addition, literature studies have found that fractal dimension value and liking are related. In this study, a result parallel to the studies in the literature was reached. Images in which the fractal dimension value of the buildings overlapped with the fractal dimension value of the environment were the most liked cases. Medium facades are the ones where the fractal dimension values have the highest similarity rate with the environment. These facades were the most appreciated by the participants. In the study, the scale developed by Oswald and Vaughan (2009) was taken as a reference in establishing the similarity relationship between the building and its environment. Accordingly, the maximum difference between the fractal dimension values should be 4% for a building and its surroundings to be considered similar, and the maximum difference should be 1% for it to be considered very similar. Based on these data, the similarity ratios of each building model according to urban regions were determined. As a result, the data obtained have results identical to those of the studies in the literature.

The sub-hypothesis of the study is “when the fractal dimension value of tall buildings are higher than their surroundings, complexity, and impression increase and liking decreases. On the contrary, when the fractal value of tall buildings is lower than their surroundings, complexity, impression, and liking decrease. According to Berlyne’s inverted U relationship, liking does not increase with complexity as in the linear relationship. As complexity increases, liking increases but decreases after a certain saturation level. Medium facades are the most appreciated, while simple and complex facades are less appreciated. Impression increases parallel with complexity, but the most admired images are not the most impressive. The data of this study confirms the inverted U relationship between complexity and liking. According to the data obtained, the hypothesis was supported. In cases where the fractal dimension value of tall buildings is higher than their surroundings, complexity and impression increased while liking decreased. In cases where the fractal dimension value of tall buildings is lower than their surroundings, complexity, impression, and liking decrease. Therefore, the results confirm Berlyne’s theorem.

The findings reveal whether there is a significant relationship between the value of the fractal dimension and the perception and appreciation of the buildings. In this way, the fractal similarities behind the designs are revealed in the visual relationship of the buildings with the environment rather than formal imitations and similarities. The study offers a new design methodology for the unique and characteristic buildings, especially in historical centers. The study brings a fresh perspective for designers and architects in constructing tall buildings harmoniously with the environment. When building in harmony with its surroundings, tall buildings are generally expected to be similar to the surrounding buildings in terms of criteria such as proportion, scale, height, use of materials, and form. However, a building contrary to the historic environment (i.e. not harmonizing in terms of any criteria) can also be appreciated. In the background of this kind of appreciation, the similarity of the fractal qualities of the buildings or their equivalent levels of complexity may be practical. Therefore, this study develops a new perspective for designers and architects regarding building-environment harmony beyond formal similarities.

Disclosure statement

No potential conflict of interest was reported by the author(s).