DESIGNING EVACUATION ROUTES AND ASSEMBLY POINT IN SCHOOLS FOR STUDENTS WITH DISABILITIES: A CASE STUDY OF YPAB SCHOOL IN SURABAYA, INDONESIA

Assembly points and evacuations can be a tremendous support for students with disabilities (SWDs) during disaster events. This paper presents a method for designing assembly points and evacuation routes in areas prone to earthquakes. As a case study, it uses a school for students with disabilities in Surabaya. The study employs a range of participatory methods, i.e., in-depth interviews, Focus Group Discussions, and field observations. The paper uses five key variables to determine the location of suitable assembly points: minimum area, safety from falling trees, safety from other hazards, access to the assembly point, and surface conditions. The study found two suitable assembly points that must be modified to ensure a minimum area of 21 m 2 to accommodate the school population. Evacuation routes to the two assembly points are defined using the nearest distance variable from 22 locations of origin. The routes vary in length from 4 to 58 meters. Lastly, the paper presents some strategies to increase the suitability and accessibility of evacuation routes and assembly points for SWDs. Four strategies are to familiarize SWDs with the assembly point locations, playing unique sounds at the assembly points, placing guiding blocks along the routes, and using a variety of bright colors to mark the assembly points, evacuation routes, and other relevant facilities.


Introduction
Surabaya is prone to various disaster such as urban fires, flash floods, strong winds, and earthquakes. Since 2017, The National Earthquake Center has warned that Surabaya faces the threat of an earthquake originating from the Surabaya and Waru Faults (PuSGeN, 2017). Consequently, Surabaya must prepare for regular small disasters (e.g., flash floods, urban fires, and strong winds) and sudden largescale earthquakes. However, a recent study found that Surabaya has low institutional resilience (Pamungkas et al., 2019), increasing disaster impacts.
The Indonesian Education Ministry's Roadmap on Safety Schools (2015) mentioned that 78,83% of schools for students with disabilities (SWDs) are situated in areas with high earthquake risks. East Java Province has around 30,522 children with disabilities, or about 3% of the Indonesian total (Riskesdas, 2018). Surabaya, the largest city of East Java Province, has 58 schools (Department of Education Surabaya, 2020) for around 1,089 children with disabilities (Mutia, 2015). Facilities to support the evacuation processes during disasters are compulsory in all of these schools. Evacuation systems in the schools significantly contribute to increasing urban resilience to disasters, especially for marginalized communities such as students with disabilities (Johnston et al., 2011).
Disaster risk reduction (DRR) initiatives and decision-making often neglects students with disabilities. These students regularly face difficulties in accessing resources and information on appropriate disaster responses (IFC, 2010;Ronoh, Gaillard, & Marlowe, 2015;Gates et al., 2006). Yet, this information is crucial for their safety as is evident from Indonesia's Metrological, Climatological, and Geophysical Agency (BMKG, 2018). Notably, BMKG suggested that self-assistance could have increased the number of survivors of the Kobe Earthquake in 1995 by about 34,9%. Jiang et al. (2012) stated that the longer the evacuation process takes for people with disabilities, the higher their vulnerability. Thus, promoting disaster risk reduction will enhance SWDs resilience during disaster events. Prerequisites to enhance SWDs resilience include the design of safe and accessible school buildings and facilities; children's involvement in DRR initiatives and decision-making processes; and collaboration with other stakeholders (Ronoh, 2017). The promotion of such principles is mandatory as stipulated in National Agency for Disaster Management (BNPB) Regulation No. 4 of 2012 concerning safe schools (BNPB, 2012).
This paper aims to determine assembly points and evacuation routes for schools for students with disabilities as part of an earthquake evacuation system. These assembly points and evacuation routes are expected to increase the SWDs' and teachers' self-assistance during earthquakes. The paper takes Yayasan Pendidikan Anak Buta (YPAB)-a school for students with visual impairments in Surabaya-as a case study (Figure 1). The school has four teachers with disabilities and eight non-disabled teachers to educate 45 students with full and partial visual impairments. Furthermore, the school complex consists of 14 buildings covering an area of 16.500 m 2 .

Data Collection
This research employs a combination of qualitative and quantitative approaches. The data are collected through in-depth interviews, Focus Group Discussions (FGDs), and field observations (FAO, 2006). Interviews were conducted with 43 students and 12 teachers. For this study, both teachers with and without disabilities were interviewed. These interviews aimed to explore the current availability and potential improvements of evacuation routes and assembly points. The study considered stakeholders' preferences for the routes and assembly points. Moreover, field observations were also carried out to understand the schools' layout and potential construction interventions. Lastly, the distances between points of interest (i.e., teachers' rooms and classrooms) were measured.
The paper considers six key variables in designing evacuation routes and assembly points based on relevant literature (Table 1). Five variables are related to the assembly points: minimum areas, safety from falling trees, safety from other hazards, access to assembly points, and surface conditions (Safrina, Hermansyah, & Aulia, 2015). Furthermore, route length is the main variable used for evaluating evacuation routes (Safrina et al., 2015). The variables are operationalized through a set of criteria.
The paper considers six key variables in designing evacuation routes and assembly points based on relevant literature (Table 1). Five variables are related to the assembly points: minimum areas, safety from falling trees, safety from other hazards, access to assembly points, and surface conditions (Safrina, Hermansyah, & Aulia, 2015). Furthermore, route length is the main variable used for evaluating evacuation routes (Safrina et al., 2015). The variables are operationalized through a set of criteria.

a. Assembly points
This paper uses the overlay function in ArcGIS with a binary value (0: inappropriate locations or 1: appropriate locations) for each variable to determine the assembly points. The potential assembly points were then evaluated using buffer analysis in GIS. Buffers of trees, poles, inundation areas, vehicle routes, and surface conditions were superimposed. The YPAB school property is scattered with various tall trees and electrical poles. During an earthquake, these trees and poles are likely to fall, thereby potentially injuring the students. Also, some areas experience inundation during heavy rainfall. These inundated areas inhibit evacuation processes. Moreover, a mix of traffic routes of vehicles and people may hamper evacuation processes in the case of an emergency. Such mixed traffic can also make evacuation shelters inaccessible. Safrina et al. (2015) suggested that shelters should be easily accessed in both normal and emergency conditions. The evacuation shelter must also be on a flat and dry surface because uneven and slippery surfaces can be a significant obstacle for people with low vision.

b. Evacuation routes
This paper maps all possible routes and calculates distances among origins and destinations. It uses the length of the route as the primary variable for selecting evacuation routes. The shortest route is selected as the proposed evacuation route for the school. Safrina et al. (2015) suggested that the shortest and fastest route is the key indicator of an effective evacuation system. Dito (2015) also emphasized that the duration of the route may also influence the evacuation system's optimization. In this school's context, the school's internal street network has no potential delays because there is no heavy traffic or other unfavorable conditions. Therefore, the duration of the evacuation depends primarily on the distance of the evaluation route. Thus, distance is the main key variable in defining the routes. The distance of the evacuation routes is calculated by adding all alternatives routes to the selected assembly points. The measurement starts from the locations of students and teachers (e.g., musical room) and ends at the nearest assembly point. Figure 3 illustrates the steps in determining the assembly points and evacuation routes.

Determination of Assembly Points
The paper evaluates three potential assembly points based on five variables ( Figure  4). The students and teachers proposed these three assembly points during the interviews. The primary consideration for these proposed locations is the familiarity of the students with disabilities (SWDs) with the sites. Students frequently use the areas to perform outdoor activities, or the recommended points are near commonly used rooms. The first point (AP1) is in front of the school hall (U) and the second point (AP2) is near the musical room (K). Lastly, the third potential location (AP3) is near the multipurpose building (P). The proposed sites are frequently used as outdoor spaces for gatherings. The nearby rooms (U, K, P) are used intensively, including as gathering places for students. The rooms have an ample empty space to accommodate evacuated students and teachers.
An overlay analysis was performed with possible constraining factors, namely trees and poles, inundation areas, vehicle routes, and surface conditions. Figure 5a illustrates the potential areas (in green) that are safe from falling trees and poles. This study deems areas that are regularly inundated unsuitable as assembly points (see Figure 5b). Moreover, figure 5c indicates the unsuitable areas for shelter due to vehicle routes. Furthermore, figure 5d shows the paved roads that are the most suitable locations for evacuation shelters. Most of the school's roads are paved (flat surface) and, thus, are ideal for evacuation routes.
The disadvantage of the first assembly point (AP1) is its small area of approximately 4.18 m 2 , whereas the second assembly point (AP2) is prone to collapsing trees and poles. Therefore, some modifications are needed. The third point (AP3) is prone to inundation, and students and teachers need to cross a small bridge to reach the spot. Consequently, the third point is unsuitable for SWDs during an evacuation. Crossing the bridge and the uneven and slippery surface conditions put the SWDs at harm during the evacuation. Moreover, in the event of an earthquake, the bridge may collapse, causing AP3 to be inaccessible. In conclusion, the modification of AP1 and AP2 is the most appropriate option in determining assembly points for this school.  After determining the two most suitable assembly points, some inhibiting conditions of these locations must be addressed. National Fire Protection Association (NFPA) Regulation 101 of 2000 states that a person requires 0.3 m 2 with a minimum height of 200 cm in an assembly point (Pratama, 2016). Since the school has approximately 70 students and teachers, the minimum area for each assembly point is 21 m 2 . Currently, the total area for the first point is 4.18 m 2 whereas the second point is 0 m 2 because of various constraints. As such, the total assembly point area is far below the minimum standard. Consequently, this paper proposes some adjustments to the current conditions. For AP1, the main constraints are trees, a school sign pole, and a flag pole. Relocating the trees, school sign, and flagpole may increase the area of this assembly point up to 27.87m 2 . For the second assembly point, the main inhibiting factors are trees and a school sign pole. Fortunately, an area west of the second point has good potential as an assembly point. Moving the second point westward may increase the capacity of AP2 up to 21 m 2 . These modifications will create two assembly points that are adequate to evacuate all the students and teachers of the school (Figure 7). The yellow dots in figure 7 indicate the new location of the school's signs poles to create two adequate assembly points for the school.   Figure 8 illustrates the length of the evacuation routes. Assuming that potential movement delay is insignificant, the route distance reflects the duration of the movement. Figure 8 shows 22 starting points, 57 possible routes to both assembly points, and 20 selected and shortest routes to the assembly points. The shortest routes of only four meters are from the year 12 classroom, year 11 classroom, and religion room. These three starting points direct to AP2. Furthermore, the longest route is 58 meters from the prayer room to AP2. It is important to note that all the rooms of this school are in separate buildings. Therefore, the evacuation route is outside rather than through a school building that can collapse. This study assumes that the longest-distance evacuation route is still feasible for SWDs. Table 1 gives an overview of the assessment of all possible routes to the two selected assembly points.    5,4,6,8,10,11,13,14,19,20,AP1 81 5,4 7 5,4 7 AP2 5,4,6,8,10,33,32,31,28,42,26,24,23,33,32,31,28,42,2 6,24,23,AP1 68 37,33,10,8,6,4 35 37,33,10,8,6,4 35 AP2 37,33,10,11,13,14 Figure 9 depicts the proposed evacuation routes. AP1 accommodates rooms A, B, V, U, R, S, and T. Among those building, buildings U (school hall) and T (Gamelan room) are used by all students simultaneously during gatherings. AP1 can accommodate all students and teachers with its area of about 27 m 2 . With an area of 21m 2 , AP2 can accommodate most teachers and students. Both assembly points have the advantage of access to the school's main vehicle route. This access supports the evacuation process from the school to safer locations in the case of earthquakes because vehicles can transport students and teachers from the assembly points. The selected points are for temporary assembly, as reported by teachers.

Key Clues for Evacuation Routes and Assembly Points
Based on FGDs with teachers and SWDS, several recommendations are proposed to improve the evacuation process. These include incorporating unique sounds, constructing guiding blocks, using distinct paint color, and familiarity with the designed evacuation route and assembly points.

a. Familiarity
Teachers mentioned that the SWDs must be familiarized with the selected assembly points and evacuation routes through regular emergency exercises. This finding is supported by Johnston et al. (2011). They mentioned that frequent, well-learned emergency practices are likely to increase the probability that, staff and pupils will respond in an informed and predictable manner during a real emergency at a school, and engage in behaviors that are recognized as best practices.

c. Guiding blocks at assembly points and evacuation routes
Guiding blocks improve people's accessibility with visual impairments in their daily activities (Ikaputra & Sholihah, 2001). Teachers and SWDs mentioned that guiding blocks will be useful during emergency evacuations.
"... whatever the disaster, students must follow the guiding blocks; they provide the directions. The students do not need to use sticks. They can walk together and follow the guiding blocks to the assembly point ... " (Teacher 1, 2019) Some locations in the school have been equipped with simple guiding blocks as markers (e.g., restrooms). The teachers confirm that the SWDs are aware of the function of the guiding blocks as walking routes. Thus, guiding blocks are likely to be useful in marking evacuation routes and assembly points.

d. Distinct bright colors
Since some students have low vision, teachers mentioned that colors play a role in assisting SWDs in finding rooms and locations. SWDs have different responses to various colors (Gaines & Curry, 2011). The use of particular colors is likely to aid SWDs in locating the assembly points. Teachers mentioned that dark-light differences could help students with low vision in distinguishing objects. The field observations revealed that the distinction between light and dark has been implemented by the school, which has successfully supported SWDs in locating and differentiating school buildings ( Figure 11). Color differences can also mark evacuation routes and assembly points. Based on the FGD, it is recommended to apply bright colors such as yellow.

Conclusions
This paper designed assembly points and evacuation routes for earthquakes in a school for students with disabilities (SWD) in Surabaya. The design of evacuation facilities for SWDs has some similarities to evacuation procedures for non-disabled students. This paper considers five variables for determining the assembly points: minimum areas, safety from falling trees, safety from hazards, access to the assembly point, and surface conditions. One variable (distance) was applied to determine the fastest evacuation routes.
Key considerations in designing assembly points and evacuation routes for SWDs are familiarity with the potential locations, unique sounds, guiding blocks, and distinct bright colors. Familiarity with the assembly points may increase SWDs' understanding of these destinations. Unique sounds at the assembly points will help guide SWDs to these points. Guiding blocks will assist the SWDs in reaching the assembly points. These guiding blocks help SWDs identify directions and speed up their travel time from their current locations to the assembly point. Lastly, distinct bright colors may also help SWDs to recognize their surroundings. Bright colors should be applied for the routes and assembly points and in other relevant locations (i.e., poles, buildings) to help SWDs to evacuate during emergencies. This article recommends that further studies assess the effectiveness of assembly points and evacuation routes as part of an earthquake evacuation system in schools for SWDs.