Strengthening Academic Makerspace Safety –
Creating a System Model for Hazard Analysis

Lawrence M. Wong¹, Edward A. Lamere², Tolga Durak², Mitchell S. Galanek²,
Nancy G. Leveson³

¹Dept. of Aeronautics and Astronautics, MIT; email: [email protected]

²Environment, Health & Safety Office, MIT

³Dept. of Aeronautics and Astronautics, MIT

Abstract:

Safety management in academic makerspaces poses particular challenges. Unconventional organizational structures and diverse user populations, among other factors, challenge traditional approaches. A systems theoretic hazard analysis with derived safety controls, however, can help to bolster the safety of such facilities. To enable this analysis, a multi-disciplinary team built a system model using a collaborative approach. The model consists of 10 system elements, 32 control actions, and 20 feedback. Preliminary safety implications were identified around coordination and missing feedback. The model elucidates the complexity of academic makerspaces and will allow further hazard analysis to identify potentially unsafe interactions in detail.

Introduction

Makerspaces, which are not only the facilities equipped with shared rapid prototyping equipment but also the community that embraces them, have grown in prevalence in the past decade [1]. This rapid expansion can be explained by looking at the origin of makerspaces through the lens of a larger, inherent making culture [2], modernized to our current digital age. As interest in makerspaces has grown, so has the variety of makerspaces, which now encompasses spaces ranging from textile labs1 to fully equipped prototyping design facilities at 65,000 square feet2. 

Creativity, community, and sharing remain central tenets to the “maker” culture [1], [3]. These ideals are evidenced by the response of makers across the globe to the recent COVID-19 pandemic. Makerspaces have played a major role in rapid development of scalable personal protective equipment to address severe shortages [4], as well as the partnership with industry to design ventilators that could be quickly produced to meet an anticipated shortfall for critically ill patients [5].  

Beyond their societal contribution in times of crisis, makerspaces can further benefit science, technology, engineering, and mathematics (STEM) students as dynamic and active learning environments [2], [6], [7], [8], [9], [10]. Increasingly, academic makerspaces have been developed at universities around the world to augment traditional engineering programs with opportunities to develop creativity and teamwork, skills highly valued by employers.

To fulfill their mission, most academic makerspaces are furnished with the types of advanced equipment often associated with machine shops [2]. While it offers versatile functionality, such equipment can cause severe injury or death. In 2018, the Bureau of Labor Statistics reported nine fatalities from accidents occurring in machine shops [11]. Outside of traditional workplace settings, this equipment has still proved dangerous. The US Consumer Product Safety Commission records that roughly 250 people per year are treated in emergency departments for injuries involving lathes alone [12], a type of machining equipment common to many makerspaces. Therefore, makerspace safety should be of utmost concern.

Moreover, managing the safety of makerspaces carries greater challenges than machine shop safety management. Since makerspaces strive to be open and inclusive, personnel of differing experience levels are often working side by side. Such diverse experience levels pose unique risks. Additionally, a fast-paced culture, coupled with the desire to pioneer novel and/or individualized design, hinders the feasibility of implementing conventional safety formalisms of design review, approval, etc. 

Further complicating the situation, a variety of makerspace organizational structures exist. Typically, the direct management of academic makerspaces involves three personnel categories: faculty, student, and specific support staff [2], [13]. A management structure that prioritizes incorporating students is significantly distinct from traditional machine shop management, but it has its merits. A student-embedded management structure allows greater creativity, customization by users, and reduced perceived barriers [14]-[16]. In addition to contributing to the success of the makerspace, student leaders can develop invaluable leadership and management skills [17]. On the other hand, trade-offs can include increased reticence from university powers (requiring increased burden of proof), inability to use traditional safety structures, or the need to enforce safety rules with social pressures/conventions [14]. 

In the absolute, the balance between safety risks and the multiple opportunities afforded by advanced equipment, creative freedom, unconventional organizational structure, etc. in academic makerspaces is not definitively known. However, safety risks can be mitigated with a safety-guided design [18], combined with the implementation of designed safety controls (e.g., accountability mechanisms, etc.), to build a robust safety ecosystem.

This safety-guided design is predicated on an institution’s having undertaking a hazard analysis that considers not only the technical aspect of makerspaces, which has been the focus of a few other safety assessments (e.g., [19], [20]), but also the human-organizational (social) aspect. This sociotechnical systems approach has overcome some the major limitations that characterize historical approaches [21] and is particularly fitting given the novelty and complexity of makerspaces.

Building a system model is a stepping-stone to the use of a sociotechnical systems approach. The Systems Engineering Book of Knowledge [22] provides a few definitions of “model.” The definition that most closely aligns with this work reads:

“An abstraction of a system, aimed at understanding, communicating, explaining, or designing aspects of interest of that system.”

As described, a model portrays a system at a particular level of detail appropriate for the usage. Instead of focusing on the physical attributes, as in dimension drawing or scaled models, a system model explicates the interactions between elements in the system. Specifically, it identifies the elements, as well as the behavior and information flow between these elements.

To our knowledge, a system model that facilitates systems theoretic hazard analyses of academic makerspaces has not been developed. Our objective is to develop one such model and to explore whether any high-level safety implications can be derived from its development.

Methodology

We purpose-built the system model as an integral part of a System-Theoretic Process Analysis (STPA). Beyond incorporating social and organizational factors, a particular aspect of STPA enables more comprehensive results to be generated. Specifically, flawed interactions are understood to be the mechanism leading to accidents. This approach surpasses the narrow focus on the frontline and/or failures associated with physical components of other techniques. STPAs have been successfully conducted in fields including aerospace, nuclear, maritime, health care engineering, and beyond. A comprehensive discussion is beyond the scope of this article. The interested reader is referred to the publications by Leveson [18], [23], [24].

The graphical formatting of the system model spawns from the theoretical underpinnings of STPA, in which the concept of control takes a foundational role [18]. Therefore, the system model, known as a safety control structure, incorporates control loops as a building block. A generic control loop (Fig. 1) has the following parts: a controller – human or artificial – who bears the responsibility of keeping a process within acceptable (safe) limits and is depicted by a rectangular box. The controller’s responsibility is fulfilled through control actions, depicted with a downward arrow. The decision for what control action to execute can be informed by the current state of the process through feedback, depicted with an upward arrow.

Hierarchy, or the arrangement of elements at various levels of detail, is another concept that helps organize the safety control structure. Many real-world systems are complex and comprise multiple controllers. What is controlled by a controller may itself be a set of controllers, and may control other elements lower still in the hierarchy. In other words, the rectangular boxes in the safety control structure are arranged vertically and can be expanded to show additional details. This top-down approach enables a cogent and manageable representation of the system.

The development of our safety control structure was undertaken by a multi-disciplinary team with a systems engineer experienced in conducting STPA, and industrial safety professionals responsible for managing makerspaces at the institutional level. First-hand knowledge of daily operations was gathered through site visits to the premises and participation of the makerspace membership process. Additional inputs were obtained from makerspace managers.

As with any modeling exercise, decisions about which details to capture versus which to omit is crucial in determining the accuracy of the subsequent analysis using the model. To this end, we followed the process suggested by Leveson [24]. The system-level losses and hazards were first defined. Correspondingly, system-level safety goals – safety constraints – were derived based on the hazards. The safety control structure can then depict the details insofar as to maintain the safety constraints. Initially, an abstract model with just a few general controllers, control actions, and feedback was built through adaptation of the general form of sociotechnical control published by Leveson [18]. The model was then iteratively refined to increase the number of controllers, control actions, and feedback at different hierarchical levels.

As mentioned, makerspaces adopt different organizational arrangements – some purely student-run, some purely managed by paid staff, and some a hybrid of the two. We chose to model the hybrid form of makerspaces, in order to capture its additional interactions and complexity.

Results

To focus the development of the safety control structure, the losses, hazards, and safety constraints were defined as in Table 1 to Table 3. The losses included physical harm to human and property, as well as the equivalent of loss of mission – the maker initiative being in jeopardy. Correspondingly, the system-level hazards included not only dangerous exposures, but also over-imposing or detrimental conditions threatening the viability of the initiative.

Table 1 Loss statements for the hazard analysis

Table 2 Hazard statements for the hazard analysis

Table 3 Safety constraints in response to hazards

The safety control structure was then modeled to capture the system elements responsible for maintaining the safety constraints. The high-level view of the safety control structure depicts the executive management of the makerspace initiative (Fig. 2). The key system elements in this view include the university administration who tops the campus governance structure, campus partners that shape makerspace safety (in this case the Department of Facilities), and the Environment, Health & Safety (EHS) Office. The inner workings of the makerspace initiative were treated as a black box thus hidden in this view, along with the details of external interactions.

The interactions highlighted in the high-level view heavily influence the establishment of the makerspace initiative and the subsidiary makerspaces. The administration sets the operational goals. Department of Facilities, and the EHS Office define requirements on space design, emission control, pollution prevention, etc. based on government regulations (and permitting requirements) and professional guidelines. While their work is most involved in the development phase, continued monitoring is provided on the status of the initiative and individual makerspaces over time for goal adjustments, regulatory compliance, permit maintenance, and change management.

Displaying the inner workings, the mid-level view sheds light on the intra-initiative management of a makerspace (Fig. 3). The system elements highlighted in this view include the management of the initiative, makerspace managers, and non-salaried senior mentors who are students or alumni with extended experience in mentorship. It is here that the hybrid organizational structure first becomes apparent.

The interactions captured in the mid-level view provide nearer-term management of a makerspace. Bearing in mind the strategic goals from executive management, the initiative management set operational constraints, appoint local management staff, and approve a facility for inauguration. Coordinating with senior mentors, a makerspace manager develops procedures and training, and monitors their continued effectiveness. Makerspace management personnel engage with the EHS Office to develop and enhance safety practices. In addition to receiving modification requests and incident reports in return, the EHS Office is informed by site inspections of facility and equipment. Crossing the organizational boundary, makerspace management interfaces with equipment manufacturers for procurement and maintenance information.

Filling in more specific details, the low-level view shows the day-to-day management and operations of makerspaces (Fig. 4). The system elements of interest include the physical space and equipment, space users, makerspace manager, and team of mentors.

There are diverse interactions at this level. The mentors and makerspace manager are responsible for ensuring safe makerspace operations. In brief, this responsibility translates into nurturing a pool of knowledgeable space users and providing real-time oversight. Beyond interacting with space users, the mentors and makerspace manager also perform maintenance and repair on the physical space and equipment. Space users constitute yet another controller vital to makerspace safety. Not only does safety hinge upon the obvious activity of equipment usage, it involves additional actions by space users. For instance, tagging out when a machine exhibits anomaly, and reporting the incident to makerspace management are just two critical interactions.

Discussion

In summary, we built a safety control structure with three views corresponding to the strategic, long-term management; intra-initiative, medium-term management; and intra-makerspace, day-to-day management and operations. With all the details expanded, there were 10 system elements, 32 control actions, and 20 feedback.

Preliminary safety implications

The process of detailing the safety control structure – defining explicitly the controllers, their control actions, and the feedback they receive – informs the analysis team the complexity of a given system. The multiple views also help focus attention while preserving the overall system interactions. This layout can bring to light previously overlooked safety issues. For example, potentially conflicting control actions to the same element from multiple controllers and missing feedback, which hampers safe decision making, can be more easily observed.

Examining our results, we found that the former category of potential safety issues is particularly relevant in makerspaces. The embrace of democracy and equality is fundamental to the maker movement and shapes a flat/horizontal organization. Many controllers interact on the same or similar levels in the safety control structure with mutual control actions. Specifically, “User Equipment Training” is a hallmark control action with the potential for flawed coordination. This control action can be provided by either the mentors or makerspace manager. Of course, providing clear and consistent equipment training is a safety critical action. Misalignment between controllers on the necessary content or the need for a given space user to be trained could lead to confusion or missing training in such a vibrant and dynamic makerspace user community.

Next, missing feedback is another category of potential safety issues that can be identified with the safety control structure. An example is provided by the control action “correct/recognize mentor practices,” which strives to build a safety-minded and proficient space user community. Because of the fluid and dynamic nature of the activities in a makerspace (e.g., absence of a structured curriculum), there is no guarantee of the ability of a makerspace manager to effectively interact with space users and observe each mentor. The lack of a proactive, systematic feedback mechanism handicaps the effectiveness of this control action.

Limitations

To model the system, we conducted site visits, interviews, and utilized institutional and professional knowledge (e.g., documentations). However, there may remain aspects of the makerspaces that were not captured. We strived for completeness by iteratively examining and refining the result, and by taking a collaborative approach.

Likewise, a question may be raised on how representative the model is of makerspaces in general. A partial answer can be given based on our modeling choices. As mentioned, the hybrid form of makerspaces was chosen to capture a superset of the interactions. Another partial answer comes from the merits of taking a systems approach to modeling. With this approach, the focus was on illustrating the interactions between system components instead of the sequential steps in a given process, which may be unique to a particular makerspace. Therefore, the abstraction as performed enhanced the generality and universal applicability of the model. One final remark must made on the temporal validity of the system model. Inevitably, systems change over time, and so should the system models. For this reason, it is crucial that the safety control structure is updated as part of the change management process. This not only enables the safety control structure to be current, but also assists in the safety evaluation of change. The interested reader is referred to the publications by Leveson, Thomas, and Castilho [18], [24], [25]. Our model is available <link> to enable this process.

Applications

Apart from making apparent the preliminary safety implications, the safety control structure supports the subsequent parts of an STPA analysis to detail how and why accidents may occur. In fact, the preliminary safety implications are to be further examined as well, e.g., by identifying the conditions that render coordination strategies infeasible. The results provide a pathway towards mitigating hazards while enabling the aforementioned merits of makerspaces to materialize.

In parallel, the safety control structure depicts graphically how different elements relate to one another for a diverse audience to build a common understanding. Notably, the model expands the perspective beyond frontline operations and unveils the involvement of the larger system. This perspective can serve as a basis for training and development, particularly for leaders at various parts of the system, and facilitate communication and coordination.

Conclusions and future work

We built a system model of academic makerspaces in the form of a safety control structure to facilitate a systems theoretic hazard analysis through STPA. The system is complex, as reflected in the safety control structure comprising diverse controllers both internal and external to the university, and various interactions.

Some preliminary safety implications were identified. First, as some control actions are common to multiple controllers, coordination must be made. Second, some feedback was identified to be missing, which hampers safe decision making. In other words, safety hinges upon both actions to shape the behavior of other controllers and processes and feedback such that informed decisions can be made.

Detailed scenarios of potential unsafe interactions in makerspaces are being identified as the wider STPA proceeds. These scenarios will provide insights to mitigate hazards not only on the frontline or the technical aspect of makerspaces, but also within the larger system incorporating the human-organizational aspect. We will report these findings in a subsequent publication.

Acknowledgement

We would like to thank Patrick White for his initial work on the safety control structure, as well as Seth Avecilla and Jonathan Hunt for their feedback and willingness to share their intimate knowledge of makerspaces. We are also grateful to Michael Labosky and Andrew Kalil for their insights into makerspace management and operations at MIT, along with their help assessing safety in a university setting. Finally, we would like to express our gratitude to MakerWorkshop for providing training and graciously welcoming the authors into their makerspace.