A FRAMEWORK FOR SELECTING AED DRONE LAUNCH SITE LOCATIONS

(1)

A FRAMEWORK FOR SELECTING

AED DRONE LAUNCH SITE LOCATIONS

Master’s Thesis by C.A.A. Hoormann 2311372 c.a.a.hoormann@student.rug.nl University of Groningen Faculty of Economics and Business MSc. Technology & Operations Management

Supervisor university: dr. ir. D.J. van der Zee Co-assessor university:

dr. N.B. Szirbik Supervisor field of study:

ir. J. Hatenboer UMCG Ambulancezorg

(2)

(3)

3

Abstract

Purpose – Mortality rates increases rapidly for out-of-hospital cardiac arrest (OHCA) patients if they are not treated within minutes. Early treatment with an automated external defibrillator (AED) for patients in rural areas, however, is stymied by its non-timely availability. AED delivery by drone shows high potential to overcome this challenge. Thus far, the choice of AED drone launch site locations in AED drone network design is hardly questioned. Launch site location selection is complex because the most optimal launch site location is determined by multiple cohesive decisions setting criteria, which are not covered in earlier literature. This study, therefore, aims at developing and evaluating a framework for selecting the location of AED drone network launch sites to facilitate an adequate AED response time in rural areas at affordable costs. The framework entails a stepwise approach to support quick, complete and effective decision-making on AED drone launch site locations, thereby contributing to AED drone network design.

Method – A design science approach is used to develop the framework. A literature review, interviews with domain experts and industrial practices of related industries provided input information for the framework elements, i.e., key decisions, relevant criteria, decision options and (model based) decision support. Finally, the use of the framework is evaluated by a case study. Results – The stepwise approach in this framework supports decision-making on AED drone launch site locations, thereby contributing to AED drone network design. Steps are linked to key decisions concerning search ring identification, identification and basic assessment of candidate sites, risk assessment, and testing and initial solution assessment. In a case study the framework has been evaluated and validated for serving as a decision support tool, giving guidance for quick, complete and effective decision-making for AED drone launch site locations.

Research implications – Although many challenges still have to be overcome, the development of the framework has laid a solid foundation for AED drone network design. In doing so, adding AED drones to the care network in order to facilitate adequate AED response times at affordable costs in rural areas became more attainable.

Keywords: AED drone network design; AED response time; framework; launch site location

(4)

4

Preface

I conducted this research in order to obtain my master’s degree in the field of Technology and Operations Management at the University of Groningen. It has been a pleasure to be given the opportunity to contribute to AED drone network design. The past months have been a challenging period where I had to face the complexity of AED drone network design.

In order to gather information and comprehend decision-making on location selection, I had to interview many domain experts of related fields of study. My thanks and appreciation to all those

kind people for being willing to offer some of their valuable time. My special thanks to R. Heidekamp from ANWB Medical Air Assistance and M. Boerman and F. van Selow from Allinq

Group for the many fruitful discussions. Without their help, it would not have been possible to obtain such a great result.

I gratefully acknowledge the invaluable feedback and suggestions from my supervisor dr. ir. D.J. van der Zee and my co-assessor dr. N.B. Szirbik, many thanks.

Furthermore, I would like to extend my sincere gratitude to ir. J. Hatenboer from UMCG Ambulancezorg for providing the chance to conduct my thesis on the subject of AED drones, the enthusiasm, and the valuable time invested in the many meetings we had.

Finally, I would like to thank my family, friends and fellow students for their support, guidance, and encouragement.

I hope reading my thesis will provide you with some valuable insight into AED drone launch site selection.

(5)

6.4 Framework output ... 37 7. Framework evaluation ... 38 7.1 Evaluation approach... 38 7.2 Input ... 38 7.3 Framework application ... 40 7.4 Output ... 42 7.5 Findings ... 42 8. Discussion ... 44 8.1 Research contributions ... 44 8.2 Limitations ... 45 8.3 Future research ... 46 9. Conclusion ... 47 List of references ... 48

Appendix A – Shockable heart rhythms and OHCA survival rates ... 52

Appendix B – Examples of Drone launch sites ... 53

Appendix C – Explanation (temporarily) no-fly zones ... 55

Appendix D – Determination procedure of robustness level ... 56

Appendix E – Assessment of Intrinsic Ground Risk Class ... 57

Appendix F – Mitigations of Ground Risk Classes ... 58

Appendix G – Tactical mitigations performance requirement levels... 59

Appendix H – Specific Assurance and Integrity Level (SAIL) ... 61

Appendix I – Illustration framework step 2 ... 63

(7)

7

List with abbreviations

AEC Airspace Encounter Categories AED Automatic External defibrillator ANSP Air Navigation Service Provider ARC Air Risk Class

BVLOS Beyond Visual Line-of-Sight CPR Cardiopulmonary Resuscitation

CTR Control Region

EASA European Aviation Safety Agency EMS Emergency Medical Service

JARUS Joint Authorities for Rulemaking on Unmanned Systems NOTAM Notice to Airman

OHCA Out-of-Hospital Cardiac Arrest OSO Operational Safety Objectives

RIVM National Institute for Public Health and the Environment SAIL Specific Assurance and Integrity Level

SORA Specific Operational Risk Assessment UAV Unmanned Aerial Vehicle

(8)

(9)

9

1. Introduction

In developed countries Out-of-Hospital Cardiac Arrest (OHCA) is the leading cause of death (Atwood, Eisenberg, Herlitz, & Rea, 2005) affecting 300.000 of Europeans yearly (Cleasson et al., 2016). The overall survival rate of OHCA is below 10% (Harnod, Ma, Chang, Chang, & Chang, 2013). To survive an OHCA, time to treatment is crucial since leaving OHCA untreated will lead to death within minutes (Larsen, Eisenberg, Cummins, & Hallstrom, 1993). Improving OHCA treatment outcomes requires efforts to shorten the time intervals between the following consecutive critical pre-hospital interventions: cardiopulmonary resuscitation (CPR), defibrillation, and advanced cardiac care by Emergency Medical Service (EMS) provider (Larsen et al., 1993).

Unfortunately, early defibrillation is stymied by the non-timely availability of the automatic external defibrillator (AED). The AED is required to administer a shock to re-establish the normal heart rhythm (Cummins, 1993). The non-timely availability of the AED is caused by multiple factors. Firstly, most of the OHCAs (80%) occur in private setting (Cooper, Swor, Jackson, & Chu, 1998), whereas most AEDs are located at public venues (Swor et al., 2003; Boutilier et al., 2017). The lack of near public venues in rural areas causes large distances to the nearest AED. In addition, AEDs within public buildings are only accessible during opening hours of the buildings. Furthermore, EMS providers are not able to improve their response time at reasonable costs to a level adequate for OHCA patients in rural areas (Van de Voorde et al., 2017). Consequently, the absence of a 24/7 accessible near public AED in combination with the long arrival times of the EMS provider is troublesome for OHCA patients in rural areas.

Recent research in Stockholm and Toronto revealed that time to defibrillation can be drastically shortened by implementing the use of unmanned aerial vehicles (UAV) to deliver an AED, the so-called AED drone (Cleasson et al., 2017; Boutilier et al., 2017). Research by Cleasson et al. (2017) showed that in rural areas in Sweden the average time to defibrillation can be shortened with up to 19 minutes. The main benefits of a drone network are the increase of AED availability to 24/7, a decrease of travelling time since there is a straight-line travel and traffic avoidance, and relative high cost-effectiveness (Boutilier et al., 2017). Clearly, the choice of location is a main determinant of gains reported.

(10)

10

al., 2017), but does not explain why nor considers other options. Therefore, this research is aimed at:

Developing and evaluating a framework for selecting the location of AED drone network launch sites to facilitate an adequate AED response time in rural areas at affordable costs.

The framework boils down to a stepwise approach to support decision-making on AED drone launch site locations, thereby contributing to AED drone network design. Key steps are: search ring identification, identification and basic assessment of candidate sites, risk assessment, and testing and initial solution assessment. The framework will serve as a decision support tool, giving guidance for quick, complete and effective decision-making for AED drone launch site locations. To develop this framework a design science study will be conducted, according to the phases of the engineering cycle of Wieringa (2014). This concerns the phases: problem investigation, system characterization & solution exploration, framework design, and implementation of the artifact & implementation evaluation.

(11)

11

2. Theoretical background

This chapter reviews related literature concerning the development of a framework for launch site location selection. To foster reader understanding, OHCAs and its care path are considered. In addition, the concept of AED drone networks and its added value to implementing it into the chain

of survival are discussed. Furthermore, existing literature regarding AED drone launch site location

selection is reviewed and shortcomings are mentioned. We conclude by stating our research contributions.

2.1 Out of Hospital Cardiac Arrest – care needs and pathway

When seeking possibilities to shorten time to treatment, a more in-depth understanding of the pre-hospital care process, also known as the chain of survival, is required. Therefore, the next section will discuss the chain of survival. In addition to this, survival curves will be discussed to underpin the relevance of decreasing time to treatment.

2.1.1 Out of Hospital Cardiac Arrest

An OHCA is the “cessation of cardiac mechanical activity that is confirmed by the absence of signs

of circulation” (Roger et al., 2011, e37) in an out-of-hospital setting. To define whether a cardiac

arrest is treatable by means of an AED, it is important to identify the category of OHCA. Cardiac arrests can have a cardiac (80%) or non-cardiac origin (20%) (Kitamura et al., 2014). The two main categories of OHCA patients with cardiac causes that can be distinguished are those with ventricular fibrillation (70%) and pulseless ventricular tachycardia (30%), initial shockable and initial non-shockable heart rhythm respectively (Thomas, Newgard, Fu, Zive, & Daya, 2013). Only patients within the category of cardiac arrest with cardiac causes and initial shockable heart rhythm can be treated by means of AED.

2.1.2 Chain of survival

(12)

12

care is started as soon as the EMS provider arrives. The EMS provider starts with intubation, intravenous medication, and transport to the hospital.

Links (Process)

Initiation (Early Access)

EMS programs components (Structure)

- Bystander CPR - Volonteer network for CPR - Quick access to AED - Bystander or volonteer to perform Automated defifibrillation - Advanced defibrillation - Intubation - Intravenous access - Intravenous medication - Initiation of transport to emergency department - Early bystander recognition of the emergency - EMS activation by calling emergency number - Enhanced dispatch of ambulance Initiation

(Early Access) Early CPR

Early Defibrillation Early Advanced cardiac care

Figure 2.1: “The chain of survival. Separate EMS program components (structure) are required to produce strong links (process) in the chain.” Adapted from: Cummins (1993)

2.1.3 Survival curve

Time to pre-hospital interventions is one of the main factors determining the strength of the chain

of survival, and thus the survival rates. Larsen et al. (1993) developed a model that reflects the

relationship between the time proceeded from the moment of collapse of OHCA (with cardiac cause and shockable heart rhythm (Appendix A)) to the three main pre-hospital interventions and the average survival rate. Survival rates represent the percentage of patients discharged alive from the hospital after suffering an OHCA (Larsen et al., 1993). The survival rate decreases with the time to a certain pre-hospital intervention (ICPR, IDEF, IAC) multiplied by its relative strength of the effect.

This leads to the following model, predicting the average survival rate:

Survival rate = 67% - 2.3% ICPR -1.1% IDEF -2.1% IAC

ICPR = Time interval collapse to CPR, in minutes

IDEF = Time interval collapse to defibrillation, in minutes

IAC = Time interval collapse to acute care, in minutes

(13)

13

Figure 2.2: Survival curves OHCA. Adapted from: Larsen et al. (1993)

2.2 AED drone network

Treatment of OHCA in rural areas faces unacceptable long times to defibrillation due to non-timely availability of AEDs, caused by a lack of 24/7 accessible nearby public access AEDs (Cooper et al., 1998; Swor et al., 2003) and/or volunteers acting as carriers (Zijlstra et al., 2014). Momont (2014) introduced a possible solution to this problem by introducing UAVs, also known as drones, as carriers of AEDs to an emergency scene. Drones are aircrafts that do not have a human pilot on board (Clarke, 2014). Recent studies (Boutilier et al., 2017; Cleasson et al., 2017) show that the use of drones can shorten time to defibrillation drastically. Momont (2014) states that AED drones should not replace existing services, but that they should complement them. Hence, AED drones should be located in areas where an adequate time to defibrillation is not met by other healthcare services.

2.2.1 Laws and regulations for drone use

Current laws and regulations in the Netherlands do not permit the use of drones. Currently, UAV use is bound to the law of remote-controlled aircrafts. The main focus of this law is risk avoidance. This makes it unfeasible to integrate UAVs within regular aviation, by only accepting flights within visual line-of-sight (VLOS) with a maximal distance of 500 meters to the pilot (Art 13-17 Regeling op afstand bestuurde luchtvaartuigen 2017). This is not in line with beyond visual line-of-sight (BVLOS) requirement of the operational process of an AED drone (Ruijfrok, 2018). In order to adapt the laws and regulations of the European Union to current and future technological developments in the field of aviation, the European Aviation Safety Agency (EASA) is writing new European laws and regulations. These new laws allow all air traffic, under the condition that risks are mitigated to an acceptable level (JARUS, 2018). It enables the EMS to introduce AED drone networks as a new possibility to shorten time to defibrillation.

0 10 20 30 40 50 60 70 80 90 100 0 2 4 6 8 10 12 14 16 18 20 % Su rv iv al ra te Time (min)

Defibrillation after 15 minutes

No Treatment CPR Initiated Defibrillation Given Advanced care given

0 10 20 30 40 50 60 70 80 90 100 0 2 4 6 8 10 12 14 16 18 20 % Su rv iv al ra te Time (min)

Defibrillation after 6 minutes

(14)

14

This research only considers the new laws envisioned for the European Union. Other countries may set other laws that are not studied here. Therefore, the regionality of the framework should be considered before applying it. These new laws and regulations are considered as a critical

design factor. Before generalizing this framework to other countries outside the European Union,

relevant laws and regulations have to be accessed.

2.2.2 Location Selection AED drone launch sites

So far, only few articles have been published on AED drone network design (Pulver, Wei, & Mann, 2016; Cleasson et al., 2017; Boutilier et al., 2017). Those articles available tend to start from an a-priori choice of launch site locations, thereby providing little motivation for their choice. For example, existing literature assumes that the launch sites should be located at existing public buildings (Cleasson et al., 2017; Boutilier et al., 2017), but does not consider that rural areas may often lack public buildings. Alternatively, Pulver et al. (2016) identified and evaluated three strategies for locating AED drone launch sites:

 Using existing EMS stations as potential drone launch sites  Using new locations as potential drone launch sites

 Using new locations and EMS stations as potential drone launch sites

The scenario of only existing EMS stations was chosen since this is the most convenient way of implementing a drone network. But this contradicts the statement of Momont (2014) that drones should be used where EMS cannot provide adequate time to defibrillation. Pulver et al. (2016) also state that due to the fact that drones are small and do not need a lot of infrastructure, it is reasonable that drone stations can be placed almost anywhere. However, this does not seem reasonable considering among others no-fly zones and nature areas making possible location non-legit. Finally, it is also assumed that the AED drones can always fly an Euclidean route but recognizes that there may be obstacles such as buildings, no-fly zones, and other aircrafts that could hinder the drone (Pulver et al., 2016).

Kim, Lim, Cho & Côté (2017) developed an integer linear programming model to support strategic planning of launch site locations for medical delivery by drone. However, in order to avoid infeasible locations like the centre of a lake, they restricted the possibilities to locations of medical institutes and vacant lots adjacent to roads. It is not clarified sufficiently what the requirements are for the vacant lots making it a legit possibility, also no shared landsides are taken into consideration.

(15)

15

2.3 Summary of findings

In this chapter the relevance of a short time to defibrillation is stressed. The current system relying on the availability of public access AEDs is often not capable to provide short AED delivery times in rural areas. Due to technological and legal developments, drones are becoming an option to deliver AEDs. Therefore, AED drones could be a useful extension of the current network of public access AEDs and ambulances.

(16)

16

3. Research design

This chapter elaborates on the design of the research by explaining the motivation of research and choice of research method, providing a conceptual model that shows the role of the designed framework in the AED facilitation process, and highlighting the research outline.

3.1 Motivation of research

This thesis provides a design science research, motivated by a question put forward by the UMCG Ambulancezorg. The main goal of the UMCG Ambulancezorg is to provide emergency medical service (EMS) for the province of Drenthe and parts of Friesland. The Dutch EMS sector is experiencing problems in meeting the adequate time to defibrillation of six minutes (Van Drenth, 2016) for patients with an OHCA in rural areas. Recent technological developments in the field of drones, provide the opportunity to achieve this by delivering AEDs to the scene of the OHCA. The question of UMCG Ambulancezorg on “how to select the locations of AED drone launch sites?” initiated this research.

3.2 Research objective

The design problem requires the design of an artifact (framework) that contributes to the achievement of the goal (achieving adequate AED response times in rural areas):

Developing and evaluating a framework for selecting the location of AED drone network launch sites to facilitate an adequate AED response time in rural areas at affordable costs.

The performance of the AED drone network is determined by the performance indicators AED

response time and costs. The AED response time is defined as the length of the time interval

between the moment the chain of survival is initiated (call for help) and the arrival of the AED on the scene (Cummins, 1993). The costs are defined as total costs involved in establishing, operating and maintaining an AED drone network within the assigned budgets (domain expert EMS).

3.3 Conceptual model – Selection of AED drone launch sites

(17)

17

- Operational processes - Controls

- Network resources

- AED drone

- Control and dispatch centre - Staff

- Launch site - Network resource dimensions

- Launch site dimension

- Amount of sites - Location of sites

- Need for defibrillation - Patient characteristics

- Satisfied need for defibrillation - Patient characteristics OHCA demand Geography Stakeholders Regulations

AED response time Costs

Figure 3.1: AED facilitation process, adapted from Ruijfrok (2018)

Ideally, the public access AED network, EMS network and AED drone network are fully integrated and offer a sufficient regional coverage at affordable costs. To develop a successful AED drone network, decisions are required on operational processes, controls, network resources, and network resource dimensions. AED drone launch sites are one of the four main network resources of an AED drone network. Main decisions regarding this resource are the number of sites and the location of those sites. Choices to be made are influenced by a great many determinants; i.e. region-specific characteristics. To facilitate quick, complete and efficient decision-making on these dimensions a framework for selecting the location of AED drone network launch sites is needed.

3.3 Research outline

The design science method can be justified by the fact that knowledge about a non-existing system (AED drone network) needs to be developed by designing an artifact. The set-up of the research is based on the engineering cycle of Wieringa (2014). In line with this cycle four phases will be distinguished in the research set-up (Table 3.1).

Engineering Cycle (Research phase) Thesis Chapter

1. Problem investigation Chapter 2 Theoretical background

2. System characterization & solution exploration

Chapter 4 Characterization of launch sites

Chapter 5 Exploration of frameworks for location selection

3. Design of the artifact Chapter 6 Framework design

4. Implementation of the artifact and implementation evaluation

Chapter 7 Framework evaluation Chapter 8 Discussion

Chapter 9 Conclusion

(18)

18

Phase I – Problem investigation

First, a literature study is performed to review related literature regarding AED drone launch site selection. In addition to this, unstructured interviews with domain experts in the field of aviation, EMS and telecom are performed in order to get insights and understand the problem of launch site location selection.

Phase II – System characterization & Solution exploration

In the second phase, the elements and key decisions of AED drone networks and launch sites are identified and described. Unstructured interviews with domain experts (in the field of EMS, emergency helicopter operations, drone operations, and telecom) are performed in order to extract information and get insights in the required characteristics of a drone launch site location and its direct surroundings. Furthermore, procedures and criteria for location selection in related industries will be explored starting from literature, and relevant industrial practice. Interviews will be held with domain experts in the field facing similar types of decisions, i.e., ambulance care, aviation, and telecom. Finally, the laws and regulations for UAV are studied for their possible impact on drone launch site location selection. The gained knowledge and insight of this phase is the basis for developing the framework in the next phase.

Phase III – Design of the artifact

During this phase, the artifact is designed and validated. All the relevant identified procedures and elements of the previous phase are used for the framework design. The development is an iterative process by entailing a continuous feedback loop of domain experts. The framework is revised during unstructured interviews until the domain experts expect that the framework will satisfy all the requirements for location selection of AED drone launch sites.

Phase IV – Implementation of artifact and implementation evaluation

(19)

19

4. AED drone networks – choice of launch sites

To develop a framework that builds a stepwise approach deciding upon launch site location, knowledge about the AED drone launch sites and functioning of the site in the AED drone network is required. Therefore, this chapter will characterize the key elements of an AED drone network and clarify how decisions regarding those key elements influence the location of the launch site and the characteristics of the site itself.

4.1 AED drone network

Ruijfrok (2018) identified the main system elements of an AED drone network in terms of operational processes, network resources, and the control structures. Each of those elements will be described in the following section.

The main operational processes of an AED network identified by Ruijfrok (2018) will form the bedrock for the development of the framework (Figure 4.1). Underneath the eight phases are described succinctly.

1. Recognition: During the recognition phase, a bystander has to recognize the emergence of the incident and make a call for help. This call for help initiates the process.

2. Dispatching: In the dispatching phase, the triage determines whether an AED is required. The location of the patient, in combination with the factors influencing the possibility of flight and availability of an AED, determine whether the AED drone can and should be dispatched. 3. Preparation of the flight: This phase is initiated at the moment that the ground control centre

receives a dispatch request for the AED drones. During this phase, the possibility of a flight is checked, followed by the determination of the flight path and possible landing zone.

4. Activation of the AED drone: Subsequently, the drone is activated in order to allow the take-off.

5. Flight: The drone will fly at a very low level altitude. By doing so, other aircrafts flying at higher altitudes can be avoided and the flight time is minimized by avoiding unnecessary vertical movement. It is assumed that the drone will be flying beyond visual line-of-sight (BVLOS). This implies that neither the pilot nor a remote observer will have continuous visual sight on the drone. For flying BVLOS technological support is needed. In this process a pilot is required to monitor the flight and intervene when needed.

6. Delivery of AED: This process involves the delivery of the AED to the bystander. This can be performed by means of landing, line or parachute.

7. Return of drone and AED: When delivery is performed, the drone will either fly back itself or brought back by the EMS provider.

(20)

20

Figure 4.1: Overview of operational processes identified, adapted from Ruijfrok (2018).

Physical resources support the operational process and are required to establish successful AED drone networks. The five main resources are AED drones, AEDs, launch sites, ground control stations and staff (Ruijfrok, 2018). Their influence on launch site design and the direct surroundings will be discussed any further in Section 4.2.

The control structures are the procedures and criteria determining the organizational goals to be met. The five different main categories of control structures are: dispatch procedures, training, emergency procedures, maintenance and safety management (Ruijfrok, 2018). The control structures influencing AED drone launch sites are the dispatch controls (response time standards & coverage) and safety management controls (privacy protection & security of physical attributes). Those will be elaborated any further in the next section.

4.2 Influence of choice of system element on launch site selection

Each AED drone network element may be implemented in various ways. These decisions on the network elements will influence the design and requirements of the drone launch site and its location. Respective decisions are assumed to influence launch site location selection by setting demands to physical dimensions, direct surroundings and resources of the launch site. Table 4.1 shows how characteristics of the launch site (first column) are influenced by decisions on network elements (second and third column). Appendix B provides the reader with examples of existing drone launch sites.

Criteria set to Criteria Depends on decisions on …

Physical dimensions Size of property Drone (size)

Launch system type (size) Garaging facility (size)

Direct surroundings Nearby obstacles Wing type of the drone

(Drone flight path)

Data connection Required bandwidth and frequency

Weather monitoring New or existing weather stations

Degree of accessibility Horizontal accessibility

(21)

21

Coverage Acceptable level of coverage

Privacy Drone flight path

Resources required Energy source New or existing energy infrastructure

Safety of personnel Vertical accessibility

Weather monitoring New or existing weather stations

Security of physical attributes Level of protection from theft and vandalism Table 4.1: Overview of criteria set to launch site influenced by decisions on network elements

4.2.1 Operational processes leading to launch site requirements

During almost all phases of the operational process, a data connection to a triage nurse or pilot is required. A monitoring tool on the launch site has to provide the triage nurse with information about the availability of the drone and possibility of a flight. Further, the pilot in the control centre has to assess the flight possibility based on weather monitoring tools at or near the station (domain expert aviation). Moreover, the pilot also has to monitor and control the drone during the flight. For flying the drone a lot of data capacity is required (domain experts telecom). This implies that a high bandwidth and frequency is necessary everywhere the drone flies, including the launch site itself.

To prepare the drone for its flight, a mechanism has to place the drone on the take-off location and automatically activate the drone. This automated launch system is also required if the drone has to return automatically to the launch site (Ruijfrok, 2018). If the drone needs to be returned by the EMS providers, the launch site should be easy to access by car. Furthermore, during the return and reinstalling, the EMS and/or maintenance personnel should be able to reach the location easily, in order to stay safe. When the launch site is located at a high or difficult to reach place, the maintenance personnel should be able to manually fly the drone from its launch site to himself and then re-install the drone (domain experts aviation & telecom).

4.2.2 Network resources leading to launch site requirements

(22)

22

4.2.3 Control structures leading to launch site requirements

The response time standard and coverage is leading in finding the optimal location to cover the demand (domain experts EMS and aviation). According to Van Drenth (2016), a response time standard below 6 minutes is key in surviving OHCA. For coverage, a cost-benefit analysis should be used to determine whether the benefits of covering an area outweigh the costs of constructing, operating and maintaining an AED drone launch site.

To operate a successful AED drone network, the network should be protected from improper use and harm to human population and the environment should be minimized. To accomplish this

security of physical attributes is required (Ruijfrok, 2018). By means of launch site design, this

(23)

23

5. Exploration of frameworks for location selection

This chapter aims to foster a solid basis for the framework design by discussing location selection approaches found in similar industries, starting from a literature review and by consulting domain experts. Moreover, regulations that may be applicable to AED drone network design are identified and analysed for influences on decision-making on AED drone launch site locations. When considering those AED drone regulations and decision approaches in other industries, attention is payed to information regarding decision approach (steps taken), decision criteria, decision options, and support of decision-making.

5.1 Procedures and decision criteria in related industries

In this section location selection procedures and decision criteria of related industries are reviewed in order to extract information for the designed framework. Those industries are chosen based on similarities to AED drone networks. AED drone networks are characterized by the presence of one or more launch sites, spread over the region such that they allow for a timely response by AED drones to a call for help. Ambulance and public access AED networks share these characteristics by selecting locations based on this same principle. Furthermore, a cell tower network consists of multiple sites spread over an area to achieve a required coverage, a so-called cellular network.

5.1.1 Ambulance station placement

When selecting locations for ambulance stations, the goal will be minimizing the total travelling time of all ambulances to all incidents, subject to several constraints. These constraints include among others, site accessibility, construction cost and operational costs (McAleer & Naqvi, 1994). In the Netherlands, the National Institute for Public Health and the Environment (RIVM) calculates whether an ambulance station should be located in a region to achieve travelling time standard of 15 minutes in 95% of the cases. The RIVM uses allocation models based on travelling times and demographic data to determine the regions where a station should be positioned (Kommer & Zwakhals, 2013). Based on the outcome of the allocation models, the budgets for ambulance service are assigned to the EMS providers. Next, the ambulance service providers have to analyse the region-specific factors to identify feasible locations for ambulance stations. For selecting the optimal ambulance station location within this predetermined area, underlying factors of fast travelling times of ambulances should be identified.

(24)

24

and the costs of the property must be taken into consideration. Moreover, ambulances operation needs quite some space for e.g. garaging facilities for ambulances, therefore, the location needs to be spacious. (McAleer & Naqvi, 1994). Apart from this, ambulance operation can be noisy, due to the fact that they have a roaring siren (Geersing, 2013). Combining all these factors with short travelling times makes that ambulance posts are often located on the edge of a village or central location (“De ambulance bij u in de buurt”, n.d.).

5.1.2 Public access AED placement

To strengthen the link early defibrillation, it is advised to increase the number of AEDs in public and potential AED operators (Gratton, Lindholm & Campbell, 1999; Winkel, 2010; Mao & Ong, 2016). These public access AEDs need to be placed strategically since the coverage range is limited. The range of an AED is about a one-way 90-second quick walk in all directions. One should consider delays like elevators and stairs (Winkle, 2010). According to the cost-effectiveness principle, potential locations are those with the greatest likelihood of benefit (Gratton, Lindholm & Campbell, 1999; Van de Voorde et al., 2017), by way of example: office buildings, train stations and concert arenas. However, these locations differ per community and therefore they should be identified through data analysis of population demographics (Gratton et al., 1999; Mao & Ong, 2016). Gratton et al. (1999) state that locations with more than one arrest per year are high-risk public locations, and thus high-yield locations for AED placements. However, research of Cooper et al. (1998) points out that only a limited number of locations can be identified as high-risk locations. The lack of those locations in combination with the limited covering range restrains cost-effectiveness in public access AED placement (Van de Voorde, 2017).

According to Mao & Ong (2016), besides effectiveness, social impact has to be considered in location selection. For example, the placement of a public AED in a school is not that cost-effective, because the likeliness of young people having an OHCA is low, however, the social impact of the unexpected death of a child can outweigh the financial considerations.

5.1.3 Cell tower site location selection

According to Sigh, Bhatt and Maheshwari (2015), the process of location selection starts by identifying an area that needs coverage, a so-called black spot. After that, a search ring is determined. This search ring is a circular area with a radius depending on its surroundings (e.g. geography, population density, federal/local restrictions), where a new cell tower site should be located to achieve coverage over the black spot (Inside Towers, n.d.).

(25)

25

performed on the characteristics of the available property to estimate the cost of construction. When analysing the geography, it is checked whether the property would be suitable for construction. The ground is checked: the soil structure needs to be strong enough to form the cell tower, as the properties topography: no nearby structures that hinder the signal. Furthermore, it is important to minimize resistance. Inhabitants can fear radiation of the cell towers (domain experts telecom). The last characteristic, access to the cell tower is required for when staff, mostly for maintenance, has to reach the property. As a result, a nearby road is preferred. Finally, the entire project from the initial idea to site usage should be executed within a reasonable amount of time, therefore the identification of possible delays in the project (e.g. required changes in zoning plan or construction of new sites/energy sources) is essential (domain experts telecom).

5.1.4 Summary of findings: decision approaches and decision criteria

In Table 5.1 an overview and comparison of the found decision approaches and decision criteria of choice of location for ambulance stations, public access AEDs and cell towers is shown. It is reasonable that these findings are to a certain extend also applicable to location selection of AED drone launch sites. Each of the findings in Table 5.1 are individually discussed in the next paragraph.

Table 5.1: Comparison decision approach and decision criteria of location choice for ambulance stations, public

(26)

26

1. Two-layer decision-making: Previous research optimizes the drone network by purely calculating the location where a drone needs to be placed without considering other criteria. It is reasonable to assume that searching for a suitable location within a predetermined area fulfilling all the criteria is a more desirable decision approach to achieve coverage of a region. 2. Costs: Since healthcare is a public service, presumably the government will fund AED drone

networks. Therefore, the cost of the network should be kept within the assigned budget. This stimulates to seek for the most cost-effective way of operating the AED drone network. 3. Travelling time: Time is one of the most important factors determining OHCA survival, and

must be used for AED drone launch site location selection.

4. Social impact: When implementing a new system, it is important that there is social acceptance towards the system. In the case of AED drones, this involves people living in the vicinity of the launch sites and the potential users of the AED drones embracing the system in order to achieve a smooth implementation process. This implies that, people acknowledge the need of the AED drone system and the nuisance of the system should minimal. Combining both, the resistance of people within the vicinity will be minimized.

5. Operational requirements: It is assumable that the property of the AED drone launch site is subjected to a zoning plan. Furthermore, the location’s direct surroundings has to meet requirements that allow the AED drone to operate properly.

6. Time to establish location: It can be assumed that the launch site should be established within a feasible amount of time (Pulver et al., 2016).

7. Synergies: Pulver et al. (2016) mention the use of existing locations can reduce the time and costs involved with establishing and operating the network. Not only the construction expenses will be lower by sharing sites, but also operating costs can be shared. Furthermore, the use of synergies influences the required time to establish the network. Concluding, the option of a synergy is a relevant factor in AED drone launch site location selection.

5.2 Specific requirements aviation industry

Besides the important elements identified during the assessment of location selection procedures and criteria in other industries, the law is dictating also some. Accordingly, in this section the current and new laws with respect to air classifications, no-fly zones and drones will be discussed.

5.2.1 Air classification

(27)

27

to enter the CTR, pilots are allowed to fly within this area. All the other airspace from the ground until 1500 foot is classified as airspace class G. In this airspace class, air vehicles are only allowed to fly with a minimum height of 500 foot, with the exception of runways. Drones are allowed to fly at a maximum height of 400 foot, so a natural separation between the drone and other air traffic is created. Only helicopters of the medical air service, police, defence and coast guard are allowed to fly in this region. Moreover, air balloons are allowed to take-off through this area. No two-way communication system is required. Furthermore, some special areas (Appendix C) can be identified (Aeronautical Information Publication, n.d.). It is always or when activated by NOTAM (Notice to Airman) forbidden to fly within these areas.

Before selecting a launch site location, it is important to consider the air classification and other areas that might complicate the flight of an AED drone. Therefore, the chance an AED drone will have to pass through or the consequences of dodging those areas has to be considered when selecting the launch site (Table 5.2).

Decision criteria Dimensions

Launch site in/near air classification Air classification

1. Classification C, 2. Classification D, 3. Classification G

Launch site in/near area Area (Appendix C)

1. Prohibited area, 2. Restricted area, 3. Danger area, 4. Low flying route, 5. Heliport, 6. Field for areal sporting and recreational activities, 7. Parachute jumping area, 8. Temporarily Segregated Areas, 9. Low flying areas, 10. Nature area, 11. A combination, 12. None of the above

Table 5.2: Overview of the key decision criteria for location selection regarding air classification

5.2.2 Specific Operational Risk Assessment (SORA)

In order to allow drone flights the EASA has established the Joint Authorities for Rulemaking on Unmanned Systems (JARUS) to write new European regulations. The regulations allow drones to fly when the operations can be proven safe. Safety is defined as the state in which risk is acceptable (JARUS, 2018). From the air risk point of view, this would mean 10-7 _{as the number of} mid-air collisions per flight hour as the target level of safety. From the ground risk point of view, this would mean 10-6_{as numbers of fatal injuries on the ground per flight hour. Following the} guidelines of the Specific Operations Risk Assessment (SORA) one can access the risk.

Figure 5.1 shows the principle of the SORA. When the drone is operating as planned it will fly in

(28)

28

Figure 5.1: Graphical representation of SORA sematic model (JARUS, 2018)

For determining the safety of the (AED) drone operation, the SORA made a procedure to access the risk of the operation. At first, the initial risk is accessed by scaling the drone operation based on criteria. Then the mitigation of the risk is access by determining the level of robustness of that mitigation. The robustness is determined by the level of integrity (safety gain = how useful is the barrier to improve the safety of the operation) and the level of assurance (proof = is the claimed safety gain proven) of the mitigation (Appendix D).

The SORA approval will only be granted for the drone operation if the operator has provided all relating document regarding air risk and ground risk to the full satisfaction of the approving authority.

5.2.2.1 Ground Risks

First, the intrinsic ground risk class should be assessed, by combining the drone size and the operational scenario (Appendix E). According to the interviewed domain experts in the field of aviation and drone operations, only two drone sizes categories are applicable for AED drones; smaller than 1 meter and smaller than 3 meters. It was also stated that flying BVLOS is a must for AED drones. According to JARUS (2018), four operational scenarios determining the ground risks can occur. These directly involve location selection.

The intrinsic ground risk can be reduced by all kinds of interventions, the so-called mitigation of the risk (Appendix F). The final ground risk is determined by the initial ground risk in combination with the modifiers (mitigations with robustness level), and it may not be more than 7.

(29)

29

Initial ground risk class Span of the drone

1. Small <1 meter, 2. Medium 1-3 meter

Drone operational scenario

1. BVLOS over sparsely populated environment

2. BVLOS over controlled area, located inside a populated environment

3. BVLOS over populated environment 4. BVLOS over gathering of people

5. Combination of the above

Possibilities for ground risk mitigations Types (Appendix E)

1. Emergency response plan, 2. Reducing effects of ground impact, 3. Technical containment in place and effective

Level of robustness

1. Level of integrity 2. Level of assurance

Table 5.2: Overview of the key decision criteria for location selection regarding ground risk

5.2.2.2 Air Risk

The SORA (JARUS, 2018) defines different kind of locations where drones could possibly fly, Airspace Encounter Categories (AEC) (Figure 5.2), and translated them to Air Risk Classes (ARC). In practice, the AED drone will not fly above 500 foot. Therefore, the ARCs will vary between B, C & D.

Figure 5.2: Airspace Encounter Categories (JARUS, 2018)

(30)

30

flight rules (e.g. implicit coordination) and common airspace structure (e.g. common procedures). By applying strategic mitigations the initial ARC is reduced to the final ARC. Depending on the final ARC some tactical mitigations (Appendix G) are needed to cover the residual risk that is not covered by strategic mitigations. Table 5.2 provides an overview of the key decision criteria for AED drone location selection regarding air risk.

Initial air risk class Airspace Encounter Category

1. AEC 6a 2. AEC 6b 3. AEC 6c, 4. AEC 7, 5. AEC 8, 6. AEC 9, 7. AEC 10

Strategic mitigations by operational restrictions

By boundary

Volume of airspace

By chronology

Timeslots e.g. between 17:00 & 07:00

By behaviour

1. Announcing the presence of the drone, 2. Others

By time of exposure

In minutes

Strategic mitigations by structures and rules

Common flight rules

1. Flight rules, 2. Right way, 3. Implicit coordination, 4. Conspicuity rules, 5. Others, 6. Combination

Common airspace structure

1. Common airways, 2. Procedures, 3. Airflow management, 4. Others, 5. Combination

Tactical mitigation Tactical mitigation performance requirements

1. High, 2. Medium, 3. Low, 4. No requirements

Level of robustness

1. High, 2. Medium, 3. Low, 4. No requirements

Table 5.3: Overview of the key decision criteria for location selection regarding air risk

Combining both, the final ground risk class and air risk class, the specific assurance and integrity level can be determined (Appendix H). This is the level of confidence the operation will stay under control (green/yellow area in Figure 5.1).

5.4 Summary

(31)

31

6. Framework design

In this chapter, a framework for AED launch site location selection is established. This framework entails a stepwise approach and decision criteria helping to select launch site locations and develop an AED drone network. Firstly, the approach of the framework’s development is discussed, followed by a description of the framework input in Section 6.2. Furthermore, Section

6.3 presents and elaborates on the framework. Finally, the framework output is be addressed in Section 6.4.

6.1 Design approach

Basically, the framework is a decision tool, containing several steps and decision criteria that can be used to transform input into output (Figure 6.1). The primary goal of the framework is to enable quick, complete and effective decision-making in launch site location selection. The framework will be developed based on earlier extracted knowledge (Table 6.1). The development of the framework and its validation has been an iterative process, ending at the moment no further alterations are suggested. The validation of the framework is executed by means review and feedback sessions with domain experts.

Information on region-specific characteristics

Network of AED drone launch site locations Choice of launch site location

Broad location selection Specific location selection

Testing and evaluating

Figure 6.1: Systematic overview of the design approach

The input of the framework are region-specific characteristics (Section 6.2). Those region-specific characteristics will influence the chosen location of the AED drone launch site. The developed framework (Section 6.3) distinguishes among three phases in decision-making. The first phase addresses the overall region and identifies areas in which launch sites are to be located. This is followed by the second phase, the specific location selection, identifying and assessing all possible locations within selected areas, eventually leading to a launch site location. Finally, in the third phase, the proposed network of launch site location(s) is tested and evaluated. Ultimately, the framework will have as output the design of an AED drone network with respect to launch site locations (Section 6.4). The last phase of the framework provides whether the network launch site locations is indeed meeting the targets.

Framework part Knowledge extracted from

Input (6.2) Section 4.2, 5.1.4 & 5.2

Framework (6.3) Step 1 Section 5.4.1

Step 2 Section 4.2 & 5.4.1

Step 3 Section 5.2

(32)

32

6.2 Framework input

The input of the framework is translated into output by applying all the framework steps. As a result, the region-specific characteristics (input information) should be studied well. The main input categories are discussed in this section. Underneath, Table 6.2 provides an overview of those categories and their possible options.

Laws and Regulations – The most important part to consider are the laws and regulations. The

framework is based on the new regulations of the EASA (JARUS, 2018), and therefore if the region has different laws and regulations, those have to be studied thoroughly. Other laws and regulations could mean that the framework needs to be adapted or even no drone operation is possible.

Demographics – This category is the study of the characteristics of groups of people. When

analysing a region, it is important to look at the location and occurrence of OHCA so the AED launch site can be located strategically to locations with high OHCA frequency. This frequency depends on multiple characteristics of groups of people; age, welfare and/or lifestyle. Moreover, information about population density is required. This information is crucial for estimating the risk of the operation.

Air structure – For determining the risk of the operation in the air based on the chance of collision

it is essential to study the structure of the air and other air traffic. Ergo, the division of air classifications over the region serves as input for the model. Likewise, (temporarily restricted) no-fly zones should be known to determine the suitability of a location to become an AED drone launch site.

Stakeholders – Performing a stakeholder analysis before starting to develop an AED drone

network for a region is essential. The project manager must identify the stakeholders and understand the interests/stake of each of them. Only then the manager can strategize on how to address them and manage the stakeholder expectations. In Table 6.2 possible stakeholders have been identified.

Geography – Geography can be divided into four different categories; the landscape, structure of

the ground, available physical structures and characteristics of the property. The obstacles in the landscape mainly influence the route a drone can fly, the risk of collision and travelling time. The structure of the ground influences the degree of difficulty to build a launch site.

Climate – The climate of a region determines the to be expected weather conditions. Those

(33)

33

Earlier decisions towards AED drone network – In the thesis of Ruijfrok (2018) multiple decisions

towards AED drone networks are mentioned. Most those decisions are intended before launch site locations are selected. Some decisions can influence the location selection. Therefore, decisions with respect to drone type, coverage, average response time and operational scenario (flying BVLOS), are used as input information for location selection.

Table 6.2: Overview framework input categories and their corresponding types of information

6.3 Framework

The framework provides the user with a stepwise structure (decision approach) in order to determine drone launch site locations. Its aim is to facilitate quick, complete, and effective decision-making. The framework acts as a funnel decision-making model, narrowing down the options for possible launch site locations until only one location is left, the most well-found location.

(34)

34

Step 0: Identification of regions suitable for implementation of AED drone networks

Step 1: Identification of search ring Decision 0

Step 2: Identification and basic assessment of candidate launch site locations

Decision 1

Framework output

Decision 2

Step 3: Risk assessment of candidate launch site locations

Decision 3

Step 4: Testing and initial solution assessment Decision 4 Phase 1 Phase 2 Phase 3 Determinants step 1 Determinants step 2 Determinants step 3 Determinants step 4 Framework Input

(Regional characteristics) Framework

Design of AED drone network with respect to

locations Figure 6.2: Visual representation of the framework for AED drone launch site selection

In an earlier phase of decision-making (step 0) an input region is identified (Westendorp, 2018). In essence, the feasibility of implementing a drone network at the input region is tested on a basic level. As a result, this region is a legitimate case identified by a combination of the demographical information (regional level), current times to defibrillation (Public access AEDs & AED provided by EMS), and desired level of performance. Furthermore, prior using the framework it is checked whether the new laws of the EASA apply to this region. If this is not the case, the laws of that region have to be evaluated first and the framework has to be readjusted, if necessary.

The framework entails three phases:

Phase 1: Broad location selection (step 1) – The first step identifies a set of search rings covering a

region based on the performance indicators of the AED drone network, travelling speed of the drone and detailed demographic information (local level).

Phase 2: Specific location selection (step 2 & 3) – During the second phase the locations are

(35)

35

assessment. During step 2 an assessment will be performed based on travelling time, social impact, operational requirements, and the involved time and costs. Whereas step 3 involves an assessment based on ground risk & mitigations, air risk & mitigations, and its involved time and costs.

Phase 3: Solution testing and evaluation (step 4) – During the final phase the proposed location

should be tested and assessed by means of a simulation study and/or pilot. This can among others be done by means of performing test flights from the selected locations.

6.3.1 Detailed description of the phases of the framework

This section provides the reader with a detailed description of the four steps of the framework required to determine the best launch site location of an AED drone network. For each of the steps of the framework, the following questions are answered: What does the step entail, who is involved, where is the research executed, what are the criteria candidate locations are assessed on, what input information is required and, finally, what will be the decision outcome.

Step 1: Identification of the search ring

During this step the search rings of drone launch sites locations are identified concerning the criteria and input determinants. Using mathematical programming, the best locations of the search rings can be placed in such a manner, that a network of a minimal amount of search rings will be able to have an appropriate AED drone response time and level of coverage of the region. This data analysis contains no information that has to be gathered through field research, and therefore, this research is categorized as desk research. Table 6.3 depicts the criteria that determine the location of the search rings and the information (input determinant) that is required for determining the criteria score. This step results in a set of search rings that cover a region.

Decision criteria Input determinants Input determinant details

Travelling time Demographics Distance of launch site to population

Earlier decisions Drone travelling speed Adequate coverage level Adequate response time

Table 6.3: Overview decision criteria and determinants of step 1

Step 2: Identification and basic assessment of candidate site locations

(36)

36

lead each search ring to a set of feasible potential launch site locations scoring high on the assessment criteria of step 2 (Table 6.4).

Table 6.4: Overview of decision criteria and determinants of step 2

Step 3: Risk assessment of candidate site locations

(37)

37

Table 6.5: Overview of decision criteria and determinants of step 3

Step 4: Testing and initial solution assessment

During this step the initial solution of a network of the selected AED drone launch sites is tested and the performance is analysed. This means that a pilot and/or simulation study has to be performed. Measurements should show whether the network of launch sites indeed performed as predicted. Table 6.6 shows some examples of performance indicators used to measure the performance of the AED drone network.

Table 6.6: Overview of performance indicators step 4

6.4 Framework output

(38)

38

7. Framework evaluation

In the following sections of this chapter, the designed framework will be applied to a pilot region. Firstly, the input region of the framework is discussed. Subsequently, the framework will be evaluated by a walk-through of all the individual steps. Finally, the framework output and the findings of the evaluation are presented.

7.1 Evaluation approach

This chapter will inspect the framework by illustrating its function on a test region. By doing so, the performance of the framework are assessed. Domain experts will evaluate based on this illustration whether the goal of quick, complete, and effective decision-making has been reached. Due to constraints in time, resources, and information, an in-depth evaluation of the framework is not feasible. The fourth step, testing, and initial solution assessment, cannot be executed at all since this would require simulation study, and therefore is skipped.

The pilot of the framework will be delimited to a region which suits the purpose of illustration, the island of Schiermonnikoog. Schiermonnikoog is a demarcated area, and therefore ideal for framework illustration. However, Schiermonnikoog will have a small scale AED drone network, and therefore may not be highly representative for bigger and more complex regions. On the other hand, using this region as a case provides way more insights due to its comprehensibility.

7.2 Input

In this paragraph, the region-specific characteristics will be described according to the seven main categories identified in Section 6.2.

Determining input per category

Laws and Regulations – It is important to keep in mind that under the current laws and regulations it would

not be feasible to implement an AED drone network. This case is studied under the assumption of the new laws and regulations of the EASA (JARUS, 2018) are implemented.

Demographics – Analysing the population characteristics, it is noticeable that 59.8% of all the inhabitants

older is than 45, of which 28.1% older is than 65-year (Figure 7.1). Those peop are more at risk for suffering OHCA (Zijlstra et al., 2016). In Figure 7.2, the population density of the island is depicted. About 89.7% of all the islanders live in the village Schiermonnikoog, the other 10.3% lives outspread over the remainder of the island (AlleCijfers, n.d.).

Figure 7.1: Age distribution population Schiermonnikoog (AlleCijfers, n.d.)

0 100 200 300 400 0 - 15 15 - 25 25 - 45 45 - 65 65+ Inha bi ta nt s Age

(39)

39

Figure 7.2: Population distribution Schiermonnikoog (AlleCijfers, n.d.)

Air structure – Above Schiermonnikoog all the airspace till 1500 meters is categorized as air classification

G. At the island of Schiermonnikoog there is one heliport located near the village Schiermonnikoog and the whole island is a Natura2000 zone (Kadaster, n.d.).

Stakeholders – Beneath an overview of some possible stakeholders and their stake in an AED drone

network at Schiermonnikoog is depicted (interviews with domain experts).

Geography – The island of Schiermonnikoog is characterized by dunes, the shallows, polder, “kwelders”

and woods. Furthermore, only a few high structures can be found on the island, such as the two lighthouses. The soil structure varies between sandy near the dunes and beach, polder and swampy at the “kwelders”.

Climate – The island of Schiermonnikoog has an oceanic climate, leading to cool summers and mild winters.

Furthermore, the amount of wind varies a lot between calm and very stormy. (Ruijfrok, 2018)

Earlier decisions towards AED drone network – During the interviews with the domain expert in EMS, it

came forward that a response time of six minutes for 97% of all cases is the most appropriate for an AED drone network. Furthermore, the expert in aviation also stated that the operational scenario of flying BVLOS is the only feasible option. Furthermore, according to Ruijfrok (2018), a rotary-wing with a top speed of 100 km/h would be most suitable to use at Schiermonnikoog in order to provide fast travel and save landing.

(40)

40

7.3 Framework application

In this section, the framework will be applied to develop an AED drone network on the island of Schiermonnikoog. During the first step, the search rings will be identified. In the second step zooms in at one of the search rings and assesses some possible locations at a basic level. In the final phase, the outcome of step two will be assessed more thoroughly based on the risks involved, and the feasibility of that location is determined.

Step 1: Identification of the search ring

The Ambulance station on the island of Schiermonnikoog is located on the edge of the village Schiermonnikoog. The mean response time of the ambulances is 5.5 minutes (domain expert EMS). The ambulance service of this region dispatches 80 times on a yearly basis. Most of the incidents are located at the village, for which they arrived far below the advised 6 minutes. However, deployment at the eastern part of the island and the beaches is way more difficult. At those areas an AED drone could be helpful.

Some of the 6 minutes response time, is required for earlier preparatory stages in the drone operations process than the flight and delivery. According to domain experts in the field of EMS and aviation, about 3.5 minutes will be left to actually transport the AED by drone. The average flight radius of an AED drone with a maximum speed of 100 km/h will be 5 kilometres. When applying the weight of the importance of the beaches in the western part of the island and the coverage in the eastern part (except for the “kwelders” at the top right) of the island two possible search rings can be identified (Figure 7.3).

Figure 7.3: Identification of search rings

Step 2: Identification and basic assessment of candidate site locations

A FRAMEWORK FOR SELECTING AED DRONE LAUNCH SITE LOCATIONS