Understanding Fire and Rescue Service Practices Through Problems and Problem-Solving Networks: An Analysis of a Critical Incident

The primary data consisted of semi-structured interviews with nine interview participants, see Table 1. The focus of the study was on the FRS and how they collaborate in a wider PSN to solve problems, i.e., the sample was chosen to get the FRS perspective. The single non-FRS participant was from the insurance industry, but this participant had extensive experience of the FRS having worked for over 20 years in the FRS.

The sampling procedure started with identifying persons in the FRS who were involved in the incident in a variety of ways based on a combination of the following:

The secondary data used in this study included the incident assessment report created on behalf of the Swedish Civil Contingencies Agency (MSB) [22], the local FRS assessment report [50], an educational video developed by MSB [51], and various media articles. See Table 1 for a summary of primary and secondary data used to develop the empirical data for analysis.

An interview guide was created for the semi-structured interviews, see supplementary material file 1. All individuals identified for interviews were contacted, received information about the study and, after informed consent, were given the opportunity to participate in an interview. Two researchers participated in each interview; the first author was present in all interviews while one of the other authors was an observer. The observer was specifically given the opportunity to ask follow-up questions that the primary interviewer may have missed at the end of each interview. All interviews were conducted in a hybrid setting where the person leading the interview was in-person and the observer was virtual. The interviews were all conducted over approximately 1 h. All interviews were transcribed. All interviews were confidential, and handling of personal details was in accordance with Lund University and RISE Research Institutes of Sweden policy for personal data handling.

The primary and secondary data collected as part of this study were analysed using the perspective of the problem space, including identification of the main problem and sub-problems, and application of the complexity framework described in Sect. 2. A PSN was defined as a network that arose in direct response to identified problems [20]. Breaking the main problem into sub-problems was viewed as a first step in dealing with this complex system of problems. Therefore, when analysing the rationale behind PSN using the complexity framework, we defined the scope in terms of what sub-problems were to be solved during the response operation (main problem). Thereby, the analytic boundary was drawn around what components and relationships were involved in solving those sub-problems (defined as dimension in the complexity framework), and the corresponding level of analysis (defined as resolution in the complexity framework). Consequently, the data were entered into NVivo 12 and coded according to the sub-problems (scope) together with relationships (dimensions) and a chosen level of detail (resolution).

During this coding process, two of the authors coded a small number of interviews independently and then compared their coding to ensure that these were similar. After this comparison the first author conducted the remaining coding independently while the analysis involved all authors. To gain a deeper understanding of the selected case study, secondary data was identified and included in NVivo for coding in the same way as the interview material. The results of the coding were discussed between the authors until agreement concerning the interpretation of the data was reached. Illuminating quotes are presented to clarify how the empirical data supports the results. Note that all quotes have been translated from Swedish to English by the authors.

Upon arrival, the FRS identified the current state of the main problem as a burning building with residents in danger, and the goal state as having extinguished the fire and ensured the safety of the residents [22].

Beyond the main problem, the data revealed a complex system of problems that could be further defined through different sub-problems. Respondents described the situation as complex due to the building construction and location in the central city. They further described the situation as chaotic and dramatic due to the large number of residents in need of evacuation and expressed that this incident was different from other incidents that they had encountered due to its chaotic nature and complexity. Due to the extensive resource demand, the respondents described problems of handling the personnel management both on site and in terms of being prepared for other incidents. The chaotic situation also created problems in understanding the situation, both on site and in the command centre. Handling documentation was also described as a problem, along with concerns regarding building stability, informing other actors and taking care of residents after the evacuation. These aspects were described by several respondents as well as in incident reports [22, 50]. The main problem and examples of sub-problems from the FRS perspective is illustrated in Fig. 5 using the concept of problem space previously presented. Potential dependencies between the difference sub-problems have not been investigated in this paper.

Despite the complexity of the situation, data indicated that the focus of the FRS was initially on the evacuation of residents and locating the seat of the fire. Therefore, these two sub-problems were chosen for further analysis in terms of which PSN were formed to solve these problems. Connected to the complexity framework, these sub-problems make up the scope of the two analytic interpretations. These scopes, together with relevant dimension and resolution (see Sect. 5.2 and 5.3), are summarised in Table 2.

The FRS prioritised the initial evacuation of approximately 60 apartments across three stairwells during the first two hours of the incident. All injured residents were taken from the scene by ambulance or bus, while uninjured residents were directed to a nearby church where their needs were addressed [22]. The PSN which formed to deal with the initial evacuation is illustrated in Fig. 6. Applying the complexity framework, this PSN consists of humans connected by relationships of coordination and different kinds of support (dimension). The resolution is on group level to illustrate that actions from the FRS, police, ambulance, and residents were executed in teams.

The evacuation activities were mainly performed by four fire teams, as shown in the middle of Fig. 6. The relationship between these fire teams concerns coordination of activities. In terms of relationships of support, the Incident Command relieved the fire teams by undertaking tasks of a more comprehensive nature, such as overall organisation and information to other actors [22]. At the same time, however, one respondent from a fire team mentioned that “[The Incident Command] became very absent in relation to what we are used to [in this type of incident]. Therefore, there was some form of self-organisation at the scene.” Thus, in a sense, the four fire teams acted as one unit, which is why they are depicted with a dashed circle surrounding them.

The relationship of support between the fire teams and the other groups is demonstrated by the ambulance teams taking care of first aid and transportation of injured residents to local hospitals; police teams registering and redirecting residents after evacuation, provided they did not need medical assistance. These types of support allowed the fire teams to focus on searching the apartments rather than taking care of residents after evacuation. This means that the skill sets of the various organisations could be applied to solving problems for their particular skill sets. The fire teams were further supported by support personnel from FRS responsible for logistical matters such as equipment and water [22]. In addition to professional first responders, especially in the start-up phase, two residents offered support by providing their knowledge of the building, including familiarity with the building layout and keys to unlock gates blocking the passage to the building entrance. One of these residents (an off-duty firefighter) also spoke to residents on balconies to prevent them from jumping [53].

The FRS and the other first responders (police and ambulance) encountered a, for them, clear and known problem (apartment fires are common incidents). In some respect, the FRS fell into habitual patterns [22] as exemplified by one respondent: “… what experience do we have? So, references around past events are a big part of my thinking”. The fact that the residents were standing on their balconies surrounded by smoke made it clear to the FRS that the most urgent problem was evacuation, and it was also clear which resources were needed to solve this problem. As one respondent stated “It was quite obvious. It's so clear that lifesaving is our priority”. Similarly, one respondent identified the most important supporting partners by stating “I want the police's support here with parts of evacuation and registration of it. And ambulance is very important here…[taking care of injured residents]”. Previous experiences between the FRS, police and ambulance teams enabled the team members to comprehend each other's roles and responsibilities, facilitating the formation of the PSN. This is also described by one respondent as a capability of the PSN members to identify the right context for each actor in a particular moment.

On the other hand, while evacuation is something the FRS train for and are experienced with, the situation was nevertheless perceived as different from what they typically deal with, as exemplified by one of the respondents “In this case, the conditions for self-evacuation were eliminated and the situation was overwhelming for both the residents and for us at the same time”. Therefore, the outcome was a shift towards a PSN with self-organisation among the fire teams, as opposed to the traditional hierarchical command structure with the Incident Command at the top. One team member experienced that they had to establish a more comprehensive level of communication between themselves compared to in other incidents. In other words, they needed to work as a team of teams rather than as individual teams. This resulted in fire team members experiencing a PSN that was larger than usual and required more collaboration on the part of team leaders. The respondent described that the team leaders needed to take control of on-site activities without waiting for instructions from the Incident Commander. It was also mentioned that this self-organisation between the fire teams was a result of previous training with the police. One firefighter involved in the incident, who was not interviewed in this study, expressed this in an educational video that was developed after the fire [51]: “… education has given us the ability to work independently, you understand that you need to make many decisions yourself [and not wait for instructions from a higher Incident Commander]”.

Despite the unique scale of the incident, as described both in primary and secondary data, it could be noted that the Incident Command had confidence in the fire teams’ abilities to autonomously manage the on-site operations and that detailed management was neither necessary nor desirable. This assessment is reinforced by another respondent who noted that "In this context, if there had been detailed management, top management, central management, then it would have been very unfortunate. Here, it is important to have control over the right things further up”. The MSB incident report [22] established that there was an insufficient number of firefighters involved in the initial evacuation. Despite the Fire Chief's communication to the on-site Incident Commander that additional resources were available to assist the teams, it was not perceived that additional resources were needed, leading to a shared perception among the fire teams on-site in the courtyard that no further assistance could be expected. Therefore, the PSN was not developed further even if there was a need for it.

Locating the seat of the fire is a crucial prerequisite for effective suppression by the FRS. In this case, the fire was caused by an explosion in the basement and during the whole incident the fire was mainly contained within the basement and an adjacent store. However, the process of locating the fire was problematic due to two reasons, leading to a delay of almost four hours before the FRS could understand where the fire was and how it was developing. Firstly, FRS needed to prioritise evacuation of residents in the initial stages of the response and could not focus on locating the fire. Secondly, the explosion that caused the fire breached multiple fire cell boundaries which resulted in the dispersion of smoke throughout almost the entire building. This ultimately lead to the perception of multiple fires and, therefore, the FRS searched for several fires in different locations in the building [22]. Figure 7 present a visual representation of the proposed PSN that emerged to solve the problem of locating the fire. This PSN consists of both artefacts, i.e., floorplans and smoke emerging from the building, and humans, i.e., fire teams, residents, representatives from businesses and the building owner as components, connected by relationships of information flow. The resolution is both at single and group level as the actions needed to solve the problem involve both single components and groups of components.

Locating the fire was centred around the Incident Command which was responsible for the overall situation understanding and for coordinating the FRS overall response [22]. The figure indicates two important processes where the Incident Command needed to create a flow of information between different components (both artefacts and humans) in order to solve the problem of locating the fire. The two processes were to locate the fire by dispatching fire teams to report on potential fire locations and contacting different actors to gather information about the building [22].

Finding information about the building is one process which is shown in the upper part of Fig. 7. This included different components e.g., the FRS command centre, with help from the municipality, provided information about the building layout by finding technical drawings. The property owner, found through the insurance representative, was able to provide further insights regarding the building layout. Moreover, since the building had undergone renovation, it was imperative to contact the relevant parties involved in the renovation process for input on changes relative to the technical drawings [22].

The lower part in Fig. 7 shows the other process, where the information flow included evaluating each presumed location of the fire within the building. First, smoke was observed emanating from multiple apartments in the courtyard side of the building. Although the primary focus for the fire teams was on initial evacuation, they also attempted to locate the source of the fire in the apartments. The dashed line in Fig. 7 illustrates the lack of information between Incident Command on the front side and the fire teams in the courtyard as exemplified by one respondent “We had no idea there was a fire on the front side…We wondered why our extinction efforts didn’t have expected effect…It was the lack of this information [that it was a fire in the basement] that would have been of value to us.”

Early in the incident, a crackling sound was identified from the basement by a person walking by (named individual in the figure), and a fire team was sent to evaluate the sound. The fire team found a fire in the basement but was unable to extinguish it. Later, a person in the Incident Command observed smoke from a store located next to the basement. Subsequently, a fire team was dispatched to the location, whereupon they successfully located that the fire in the basement had spread to this location. After a couple of hours into the incident, the Incident Commander discovered smoke from the roof which prompted the FRS to prepare for a possible attic fire, but this risk was soon dismissed due to effective communication between the UAV (Unmanned Aerial Vehicle), fire teams and Incident Command [22].

The PSN developed to locate the assumed fires was formed because of the need to obtain a better understanding of the unusual smoke dispersion and the complicated building layout. Although the incident was atypical and did not conform to previously encountered scenarios [22], the Incident Command handled the incident as usual and predicted that where there is smoke, there should be a fire. Two respondents sum this up well: “it seemed that there were several apartment fires going on together with maybe some type of fire in the basement so that you are very focused on finding these apartment fires at the beginning then. But then after a while you realise that you are inside these apartments and there is no fire but just smoke from somewhere else…” and, “the whole house was leaking smoke in a way that it normally shouldn't”. The complicated building layout is exemplified by the following quote: “It's a tricky building… If it had been a normal Swedish apartment building with three staircases and four floors, that we could have run around with ease, then we would have understood the incident almost immediately”.

This PSN indicates that artifacts can be integral components and can have an impact on the problem-solving process. It seems especially important to obtain an understanding of the building through visual observations, which include on-site inspections and the examination of the building layout and UAV images. Furthermore, the visual dispersion of smoke played a critical role in shaping the network, as shown by the network's formation.

The purpose of a PSN is to solve a particular problem. In this paper, we have shown that the FRS develop PSN by breaking down complex problems into manageable sub-problems. In these sub-problems, components and relationships within the PSN can be found more easily. Dividing complex problems into sub-problems is a way to match the situation to previous experience and to more easily identify which actions to take. Actions in this context can be interpreted as the FRS searching for which resources (or components in the network) are needed to solve the problem. Our results indicate that the matching with previous experience [54] will affect the formation and development of PSN, e.g. in the context of locating the fire where the layout of the building and the dispersion of smoke presented a challenge. The Incident Command created a PSN by dividing the problem into several known sub-problem, finding resources that could assist in understanding the building layout and sending resources to investigate each location emitting smoke. This is consistent with previous studies where it is argued that it is possible to divide complex problems into more manageable sub-problems [6] and manage the sub-problems through sub-networks that operate within the larger response network [13, 16, 55]. Even though the initial interpretation of the fire location was incorrect, the use of sub-problems allowed the ultimate outcome of the fire location.

The results indicate that the rationale behind the PSN could not be found in the main problem itself, nor the sub-problems per se, but rather in the relationships which contribute to the process of using actions to reduce or eliminate the difference between the current state and goal state. For instance, in the context of the initial evacuation, a clear problem for those directly involved, and the corresponding PSN, was developed based on previous experience. The fire team knew which support they needed and the PSN was formed around the relationships associated with this support. Similarly, the need for coordinated actions between the fire team and the lack of support from the Incident Command shaped the PSN, with the relationship of coordination between the fire teams and surrounding support functions at the heart of the PSN formation. In the context of locating the fire, the incident command function needed different types of information to understand the fire behaviour within the building. This in turn led to the creation of the PSN for the purpose of gathering information, thereby establishing relationships of information flows between different components. An understanding of network relationships can be beneficial to FRS organisations, as it enables them to identify the key actors to approach for information, personnel support, and collaboration on tasks [15].

The main practical contribution of the present study is the application of the conceptual framework in the context of learning from incidents.

When it comes to learning from incidents, Frykmer [23] has noted that emergency evaluations often lead to conclusions that are too broad or general to be readily operationalised. Conversely, by applying the conceptual framework in this study and zooming in on different sub-problems in an incident, organisations can identify important components and types of relationships needed to solve the problems and use this as an input to their learning process. The learning outcome from this incident can be summarized in four main lessons learned:

By identifying relationships and components needed to solve problems, the FRS could plan and exercise for important components for problem-solving during emergencies [13]. Using the relevant sub-problems to determine related parameters based on the chosen scope, dimension and resolution [48] can improve the learning potential in an exercise [56]. This approach could also assist the FRS in managing relationships as vital resources, and better comprehending the efficacy of different relationships in addressing emergency challenges [15].

This study contains several theoretical contributions. First, we have addressed the call for more knowledge of what lies in the relationships within a network (as presented in Hu, Yeo [15]). We see the relationships within a PSN as tools for reducing or eliminating the difference between the current state and goal state [30, 44]. For example, this study has illustrated that the relationships can consist of different types of support, information flow and coordination, that are used to solve the specific problems.

Second, we have contributed to the literature on PSN, as exemplified with Milward and Provan [20]. We argue that it is not only inter-organisational relationships that could explain the rationale behind PSN (as in the current definition by Milward and Provan [20]) instead we should include all components and relationships that affect the process of finding a solution to the sub-problem. To obtain a better understanding of the purpose of a PSN in emergencies, we suggest that the definition of a PSN is expanded to include not only organisations and humans within organisations, but also different artifacts relevant for reducing the difference between the current state and the goal state.

Third, the study contributes to developing the complexity framework by Bergström et al. [48] We believe that the framework can be expanded to incorporate multiple levels of resolution within the same scope, such as components on single and group level. For example, the resolution needed for analysing the PSN in the case of locating the fire contained system components at both single and group level to ensure that interactions between the Incident Command and various components could be captured. The study has also introduced an additional view on scope that can be applied in the framework; that of problems and sub-problems.

The perception of a given situation or circumstance can vary widely between individuals due to their unique experiences, biases or role in a specific emergency situation. This implies that what one individual considers to be a problem, another may not [8, 9]. Therefore, the problems identified in this study would most likely have been impacted if more or other respondents had been interviewed.

Moreover, it should be noted that the PSN depicted in this study constitutes a condensed representation. The dynamic nature of a PSN implies that its configuration is in a constant state of change, and a particular representation of the PSN is only applicable within specific temporal and spatial limitations [48], which is a limitation in our study.

In our work, we used two different interpretations of sub-problems. We acknowledge that there will always be additional interpretations. We however maintain that these interpretations captured the most important sub-problems perceived by the respondents and were thus the main problems to cover.

Last, our findings are based on a single case. While the findings may not be directly generalisable, they have contributed valuable insight into how the conceptual framework can be used to gain a more comprehensive understanding of FRS practices through problems and problem-solving networks.

Collaboration in networks has been widely investigated by numerous researchers (e.g., [13‐18]. In this paper, we have taken a first step of going beyond collaboration relationships in networks. Our study can be seen as a first step towards developing a better understanding of what network relationships actually mean in problem solving networks. This is relevant for other kinds of network research as previously stated by Hu et al. [15]. Further research could focus on the deeper understanding of the constituents of the relationships, to gain a more profound understanding of the rationale behind the PSN.

In this paper, we have seen how PSN are formed by both formal and informal relationships between the components. Examples are formal relationships between the FRS and other first responders and informal relationships between the fire teams and the off-duty firefighter in the case of initial evacuation. Further research could be conducted to examine whether PSN exhibit formal or informal characteristics. Such research would serve to evaluate the significance of formal organisational structures or prior experience in shaping effective incident management practices.

Finally, investigating the dynamic nature of PSN can provide important insight into how these networks develop over time. Such insight can, for example, be useful for understanding which components and relationships might be valuable at what time during an emergency, to further improve the solving of complex problems in this context.