The concept of the built environment was first introduced by social scientists (The mutual interaction of people and their built environment: A cross-cultural perspective, 1976). Broadly speaking, it includes man-made buildings and infrastructure that constitute the physical, natural, economic, social, and cultural capital (Hassler & Kohler, 2014). The built environment forms the foundation of the urban texture, a complex socio-technical system comprising interrelated elements. These elements include building footprints, plots, blocks, streets, and building fabric, representing the dynamic three-dimensional form of the urban landscape (Conzen, 1960). Additionally, the built environment refers to specific elements related to buildings, such as doors and walls, or the spatial subdivisions of buildings involving their functions and arrangements (Lawrence & Low, 1990).

In the context of cities, buildings are the fundamental units that sit on plots of land. These plots, often with street access, play a vital role in shaping the urban texture (Garcia, 2013). The surrounding land formed by multiple plots constitutes blocks, and the size of these blocks is determined by the width of the streets and the distance between them. Several blocks can be combined into larger units, like communities. Each community exhibits its own unique history, distinct street patterns, and functional characteristics, and a city contains a series of communities (Garcia & Vale, 2017).

In-depth analysis of changes in the built environment is crucial for urban planning and design because it helps us understand changes that have already happened or are currently taking place in cities, thereby providing valuable insights for decision-makers and urban planners. Hence, in the context of rapid urban transformation, there is a need to explore methods and strategies that can guide cities towards more sustainable development.

The concept of resilience has its roots in engineering and is described as the ability of systems to absorb changes and persist (Walker et al., 2004). It has been developed since the 1970s in various fields, such as psychology and ecology. Three features related to resilience explain the functioning and development of systems. Firstly, changes are cyclical because the system is always in the process of exploitation, conservation, release, and reorganization (Holling & Gunderson, 2002). The adaptive cycle can be used to describe these cyclical changes. Secondly, such cyclical changes usually occur simultaneously on different scales and speeds (Garcia, 2020). Thus, a nested set of adaptive cycles forms a Panarchy, describing the dynamic hierarchy that involves adaptation across multiple scales (Garcia, 2013). Thirdly, changes across multiple scales result in a landscape that is both discontinuous and heterogeneous. They are critical to understanding the complexity and resilience of systems as heterogeneity and discontinuity contribute to the diversity of the system (Holling & Gunderson, 2002).

At the start of the 21st century, ecological resilience received attention in the field of urban studies because the city itself was in the process of constant change, and the built environment can be regarded as a record of long-term adaptation and overcoming the catastrophes of history (Garcia & Vale, 2017). Therefore, resilience theory can be used to explain the cyclical changes, renewal, restructuring, and development of cities. For example, the cyclical change in the urban landscape is called the burgage cycle, as it experiences phases of continuous production, growth, waste, and shrinkage (Feliciotti et al., 2018). Panarchy can also explain changes in elements of the built environment, such as streets, as these elements undergo discrete changes at different speeds and quantities (Garcia & Vale, 2017).

Scholars, like (Garcia, 2013), find that creating timelines and identifying key variables at different temporal and spatial scales is a reasonable way to assess the complexity of urban systems. This is because timelines provide important insights into the events and essential breakpoints that have shocked the system and the possible causes that triggered the changes. This process enables a deeper understanding of the heterogeneity and diversity of the built environment and the dynamics of change that occur at different scales (Dhar & Khirfan, 2017). Most existing research linked resilience in the built environment to alleviating shocks caused by earthquakes, floods, and tsunamis. However, few studies focus on analyzing and developing methods to improve the resilience of the built environment to the problems that arise due to its complexity. This study uses resilience theory to describe the changes in the built environments of the workers' residential area of ICs. Because resilience helps to understand the history and development of ICs, revealing the complexity, organization, and relationships of the built environment.

This study uses the Shanxi Knitting Factory (SKF) as a representative example of an IC. Founded in 1950 and situated in the south of Taiyuan, the SKF covers approximately 2.3 km2 and is home to about 6,800 residents. Currently, the factory of the SKF is gradually being replaced by new high-rise commercial complexes after its bankruptcy in 2007. Nevertheless, the workers' residential area remains the home of the former factory employees and their families. (Fig. 1)

As characterized by (Bonino & Pieri, 2015), such ICs operated much like encompassing “all the functions of a city”. In the case of SKF, both its factory and the workers’ residential area were constructed on the same site. This design facilitated easy and efficient commuting for workers and contributed to the compact landscape within the SKF. In the factory, giant office buildings and squares typically occupied central positions, symbolizing the supreme power of the Communist Party in socialist China. Workshops, warehouses, and equipment rooms surrounded these central structures, demonstrating that all production activities should support industrial solidarity and serve the supreme power. In the workers’ residential areas, a variety of facilities, including cafeterias, grocery stores, and schools, were available to enhance the living conditions and welfare of the residents and their families. (Fig. 2)

Scholars such as(Bjorklund, 1986) and (Bray, 2005) have observed that enclosed spaces, often surrounded by walls, were prevalent in traditional China because they efficiently organized various social activities. This insight sheds light on the fact that the construction of walls was usually the first step in establishing ICs. Beyond the conventional walls, building facades also served as boundary walls. In the case of the SKF, some of these buildings, positioned alongside city streets or internal roads, were specifically designed to integrate their facades with conventional walls. This design highlighted the most prominent feature of the enclosure for the SKF as an IC.

Gates, apart from walls, were another element contributing to the enclosure of the SKF. While walls symbolized strong state control and spatial division, gates managed the interactions between the enclosed SKF and the external city. Specifically, they reinforced the isolation of the SKF yet provided controlled access points, enabling necessary communication with the surrounding urban landscape. Thus, gates functioned as both barriers ensuring isolation and connectors facilitating interaction within the SKF. (Fig. 3) Moreover, the road network not only shaped the spatial organization of the SKF but also played a pivotal role in shaping, defining, and segregating various spatial functionalities. It contributed to the SKF as a multi-functional compound, including diverse land use for production, office, residential, and social activities (Zhang & Chai, 2014).

Since the economic transformation in the 1980s, the development of the SKF has been significantly influenced by the implementation of a series of reform policies related to enterprise restructuring, housing, and land. However, after the enterprise restructuring, the SKF factory faced challenges in adapting to the new economic environment, resulting in decreased production efficiency and massive worker layoffs, particularly after the 1990s. These difficulties eventually led to the factory's bankruptcy in 2007. Subsequently, the local government sold the brownfield to property developers to create new communities. Despite the demolition of the factory, the remaining workers' residential area persisted, continuing to accommodate former employees and their families.

In recent years, many buildings located close to main roads within the residential area and adjacent to the city streets have been demolished as part of efforts to reorganize the workers' residential area and widen the external city roads. Additionally, there has been a noticeable reduction of enclosure in the workers' residential areas, with increased gates and decreased walls, especially in response to the social appeal to "open up the enclosed community" after 2000. From this perspective, the decreased enclosure may allow the city to be more accessible to the workers' residential area of the SKF, allowing its boundaries to support a wider variety of functions and activities. It might also increase the complexity of the road network, given the rise in connections between various functional land uses and service facilities within the area.

The SKF was chosen as the case study because its factory underwent rapid development, followed by a decline and eventual bankruptcy. Considering its role as an industrial auxiliary facility, the workers' residential area likely experienced significant changes. Using resilience theory, this study analyses the changes in the built environment of the SKF workers' residential area impacted by the factory bankruptcy. The aim is to grasp the adaptive capacities of the workers' residential area when facing disturbances. This, in turn, provides insights into the complexities and sustainability challenges cities face during China's economic transformation.

This study focuses on the following research questions: ( 1) to what extent can resilience theory describe changes happening in the SKF? (2) How can resilience theory be applied to assess changes in the built environment of workers' residential areas of the SKF before and after the factory bankruptcy? ( 3) In what ways can resilience theory be used to evaluate the influence of changes in the level of enclosure on the built environment of the workers' residential area after the factory bankruptcy?

For the purpose of drawing comparisons across a time domain, this study started by creating a timeline based on key events associated with the development of the SKF. The bankruptcy of the SKF factory stands out as the most significant breakpoint, thus positioning 2007 as the midpoint of the timeline. Given these events, this study designates 1978 and 2022 as the starting and ending breakpoints of the timeline, respectively. The rationale for selecting 1978 as the starting breakpoint stems from the shift of the SKF factory from state financial support to independent operation after the economic transformation. This meant that the economic efficiency of the factory could directly influence the development pace of its surrounding residential area. Consequently, the timeline was divided into two phases: "before factory bankruptcy" and "after factory bankruptcy." Further essential breakpoints were incorporated into the timeline, including 1978, 2000, 2003, and 2007 (before the factory bankruptcy), as well as 2014, 2016, 2018, and 2022 (after the factory bankruptcy).

To address the research question, a quantitative study on the resilience of the built environment in the workers’ residential area of the SKF was conducted. Raw data were gathered using various methods, including library resources, land use maps, the Open Street Map API, and field research. Microsoft Office software was employed for statistical analysis. Further, the analysis of changes in the built environment was carried out by using ArcGIS and Adobe Illustrator software. Different data collection methods were applied for different development periods of the SKF. (Table. 1 ) Specifically, data before the factory bankruptcy were obtained from library resources and land use maps. On the other hand, data after the factory bankruptcy were obtained from library resources, land use maps, the Open Street Map API, and field research.

This study analyses ten elements of the built environment in the workers’ residential area of the SKF. Elements related to enclosure include GATE, boundary walls (BW), and buildings as boundary walls (BBW). Other elements influenced by enclosure comprise the total road length (TRL), building density (BD), green space (GS), commercial land (CL), residential land (RL), public service facility land (PSFL), and mixed-use land (ML). (Tab. 2)

To answer the first proposed research question, firstly, we aim to demonstrate that changes in the built environment are cyclical. To this end, we assess the land changes for residential purposes within the workers' residential area of the SKF to determine if these changes can be associated with phases of the adaptive cycle in resilience. Secondly, it is essential to create a diagram for the quantitative observation of relationships among elements of the built environment in the workers' residential area. If these elements reflect the independence of their respective development, it suggests a hierarchy in the built environment that can be compared to Panarchy in resilience. Thirdly, in Panarchy, small-scale changes occur more frequently than large-scale changes. Consequently, we need to track the quantity and frequency of change in these built environment elements of different scales to determine if these changes can be linked to resilience. Finally, if evidence from the three steps supports the idea that resilience theory can be applied to analyze changes in the built environment of the workers' residential area of the SKF, it would answer the first research question.

The differential analysis will be employed to address the second research question of this study. Specifically, this method facilitates a comparison of the built environment changes in the workers' residential area of the SKF before and after the factory bankruptcy. Our null hypothesis (H0) posits that there is no association between the factory bankruptcy and changes in the built environment of the workers' residential area. In contrast, the alternative hypothesis (H1) suggests that significant differences exist between the two periods. If the result shows that differences are significant, we would then accept the alternative hypothesis and reject the null hypothesis.

To address the third research question, we turn to multiple linear regression analysis. Here, the null hypothesis (H0) is that changes in the level of enclosure have no impact on the built environment of the workers' residential area after the factory bankruptcy. The alternative hypothesis (H1), on the other hand, contends that changes in the level of enclosure indeed affect the built environment of the workers' residential area after the factory bankruptcy. If subsequent findings demonstrate that the level of enclosure influences other built environment elements and significant correlations exist among them, it would validate the alternative hypothesis (H1) and reject the null hypothesis (H0).

A combination of quantitative and visual data software was used to map changes in the built environment of the workers’ residential area of the SKF. If a diverse trend of functional land use is observed as the enclosure declines, the association between heterogeneity and resilience can help explain the ongoing changes in the built environment. Finally, this study concludes by discussing whether the workers’ residential area of the SKF is gradually integrating into the surrounding urban neighborhoods.

(1) Analysing changes in the built environment of ICs

Fig. 4 depicts the land changes for residential purposes in the workers’ residential area of the SKF from 1978 to 2022. In 1978, the percentage stood at 33.14%. After that, it gradually increased to 46.74% in 2007 when the factory went bankrupt. Since then, the land allocated for residential purposes sharply declined to 34.65% in 2016, before rebounding to 38.63% in 2022. These fluctuations suggest a cyclical change comprising four phases: exploitation and conservation from 1978 to 2007, followed by release and reorganization from 2007 to 2022.

The changes in the built environment of the workers’ residential area of the SKF are depicted in Fig. 5, with each line representing a built environment element from 1978 to 2022. Despite belonging to the same landscape, these elements are independent and do not intersect. The observed changes offer evidence of the complexity of the built environment, which can be associated with the Panarchy theory.

Tab. 3 presents the percentage of changes in the built environment elements of the workers’ residential area. It is

observed that TRL (33.49%) and BW (-36.86%) act as slow variables, contributing to landscape stability, while GATE (125%), BBW (679.37%), and various functional lands such as ML (308.97%), PSFL (101.65%), CL (68.24%), and GS (47%) are fast variables. Additionally, RL (-16.65%) demonstrates a decreasing trend due to the local government’s planning after 2007, which aligns with the concept of Panarchy. Overall, the three steps confirm that resilience can be employed to analyze changes in the built environment of the workers’ residential area of the SKF, thus supporting the first hypothesis.

(2) Measuring the impact of the factory bankruptcy on the built environmen.

Tab. 4 provides a differential analysis comparing data from two distinct periods: before and after the factory bankruptcy. The results highlight a variation in the mean value (M) between these two periods. After the factory's bankruptcy, all elements increased, except for the decrease of BW and RL. Among all variables, BW and BBW showed the most pronounced changes (t>±5). Moreover, the P-values for all variables are consistently <0.05,

suggesting a significant association between factory bankruptcy and changes in the built environment. Consequently, we accept the alternative hypothesis (H1) and reject the null hypothesis (H0). The results of the differential analysis indicate that changes in the operating status of the factory, specifically the factory bankruptcy, have a significant impact on the changes in the built environment of the workers' residential area of the SKF.

(3)Assessing the impact of changes in the level of enclosure on the built environment

The multiple linear regression analysis reveals the relationship between the level of enclosure and the elements of the built environment in the workers’ residential area (Tab. 5). Initially, we focused on the influence of GATE (independent variable) on the other built environment elements (dependent variables). Except for RL (Beta=-0.52, p=0.183), all elements showed a positive association with GATE (Beta>0, p<0.05 or p<0.01). For example, 95% of

the changes in mixed-use land (ML) (R²=0.95) can be attributed to GATE. As for BW, while RL displayed a

significant positive relationship (Beta=0.90, p=0.002), most other built environment elements showed notable correlations with BW (Beta<0, p<0.05 or p<0.01). Regarding BBW, it consistently impacted all elements of the built environment. An exception to this trend was found in RL, which had a minor negative correlation (Beta<0, p=0.032). On the other hand, all other elements had significant positive correlations with BBW, indicating its consistent positive influence (Beta>0, p<0.05 or p<0.01). Therefore, based on the results from the multiple linear regression analysis, we accept the alternative hypothesis (H1) and reject the null hypothesis (H0).

Not applicable.

The authors declare that there is no competing interest.