Engineering judgement in undergraduate structural design education: enhancing learning with failure case studies

ABSTRACT
 Universities face the challenge of developing undergraduate structural engineering students' design judgement. This study evaluates whether introducing ‘learning from failure', centralised around ‘real-world' case studies, serves to facilitate the development of engineering judgement in structural design. The study identifies the use of three characteristics of engineering judgement: diagnostic, inductive, and interpretive in the work of the first-year undergraduate structural design students. Thematic analysis, combined with a constant comparison method and the rigour of inter-researcher reliability, was used to develop coding and mapping to evaluate students' work. The majority of students correctly applied diagnostic engineering judgement to the definition of a problem for a failure case study; and displayed the inductive aspect of judgement. Students' interpretive understanding embraced multi-faceted considerations, with engineering practice, complexity in causality, and learning from history being dominant. Introducing case studies deepened students’ enquiry, stimulating the development of a more nuanced understanding of structural engineering judgement.


Introduction
postulates that the nature of structural design arises from a marriage of structural theory and practice. The theory embraces technical engineering knowledge, which must be harnessed by judgement, through appropriate practice to achieve satisfactory design outcomes. For some years, universities have sought to simulate this process intrinsic to professional practice in the structural design education delivered on undergraduate civil engineering programmes to shape young engineers for anticipated industry challenges. Zhou (2012) argues that 'engineering in practice' is difficult to define; it certainly embraces complex systems of uncertain heterogeneous interacting components, including 'science, technology, economic, human and sociology'. To deal with such complex systems the ability to make creative judgements, despite uncertainty, is accepted more or less as an axiom in the engineering profession, as an attribute of the practising engineer (Gunaratne 1995;Jonassen, Strobel, and Lee 2006;Kazerounain and Foley 2007;Cowan 2010). The tenants of uncertainty, complexity and constraint are never truer than in the discipline of structural engineering, where the design often has a bias towards bespoke solutions.
Within civil engineering undergraduate programmes, approaches to simulate design practice have generally focused on problem-solving (Donald 2011;Murray et al. 2019;Sheppard et al. 2009) and encouraging creativity (Lui and Schöwetter 2004;Daly, Mosyjowski, and Seifert 2014). McNeill et al. (2016),researched students' beliefs around problem-solving in the classroom. This study found the expectation that known concepts relating to the curriculum of a module would be tested, combined with the presentation of problems in a written form, led to classroom-based problems viewed as simple and less constrained than workplace problems from the perspectives of the nine students interviewed. Arguably the framing of classroom design problems can lead to the narrowing of opportunities for students to make judgements autonomously. Daly, Mosyjowski, and Seifert (2014), studying seven engineering courses developed to foster cognitive creative skills to solve design problems, found a lack of evidence of the development of skills revealing relationships or processes of elaboration and metaphorical thinking. Daly, Mosyjowski, and Seifert (2014) results suggest that those encouraging creative thinking in undergraduate civil engineering programmes must find new approaches to prompt deeper student enquiry, synergise ideas and prompt innovative judgement.
Recently, professional papers prepared by authors stemming from the industry have advocated that the education of civil engineering students should embrace learning from failure, centralised around 'real-world' case studies (Mottram 2013;Love, Lopez, and Edward 2013;Lewis 2012). Petroski (2006) argues that successful design and failure are inherently intertwined. The former springing from the necessity to improve on the limitations/failures of former design attempts. Within professional engineering practice designs are refined; a process that occurs progressively through a process of 'conceptual growth' (Bornasal et al. 2018) as a design develops through key stages from concept to detailed design, fabrication, construction and finally operation. At each stage, practising structural engineers will encounter design failures and setbacks that must be overcome.
A key requirement of engineering practice is, therefore, the acquisition of the cognitive ability to make appropriate and effective judgements in this environment. This research aims to explore whether judgement skills, here specifically in structural design, can be facilitated by the introduction of 'learning from failure' case studies into the structural design modules of the first year of an undergraduate civil engineering degree programme, through the thematic analysis of student coursework submissions.

1. Defining judgement
Structural design practice, was defined as early as 1956 by Nervi (1956, 24), who stated that the mastering of structural knowledge is not synonymous with a knowledge of that mathematical development, which today constitutes the so-called theory of structures. It is a result of a physical understanding of the complex behaviour of a building, coupled with an intuitive interpretation of theoretical calculations.
Theory and analysis as taught in universities become 'first principles' within the workplace, executed by engineering judgement to produce successful designs. Structural engineers employ these scientific approaches to make judgements about the future performance and behaviour of a structure: will the structure maintain its stability in the temporary or permanent case? Neither structural theory or analysis, or the results derived from their application, can be taken to be uniquely or objectively true. Interpretive judgement is required to make sense of the outcomes of both; one method of analysis might yield an acceptable factor of safety, but not the location of the 'critical members', which have fundamental importance in maintaining stability. Therefore, choices must be made concerning what type of accuracy is important and which analytical methods should be applied.
Judgement has been defined as having 'three fundamental attributesit has a diagnostic character in problem definition, and inductive character in combination of evidence, and interpretative character in providing meaning and context to predictive conclusion' (Vicks 2002, 100). Examining this definition, it is furthermore apparent that judgement is a process that occurs at key stages of structural engineers' design practice. Judgement serves in the diagnostic forming of a hypothesis of how a structure will behave. Inductive reasoning gathers data and selects theories and analytical methods that are applicable to the problem. These could be characteristics of the form of the structure, the loading conditions, the applicability of elastic or plastic analysis, or the properties of the ground. Finally, interpretive judgement contextualises the results with wider understandings and the intuition of experience.
2. 2. The case for case-histories Khun (2012, 43) sheds some light on how judgement is developed, stating that a conceptual framework or paradigm develops within a professional community centred on the problem-solving achievement of its theories. 'Close historic inspection of a given speciality at a given time discloses a set of recurrent and quasi-standard illustrations of various theories in their conceptual, observational and instrumental applications. These are the community's paradigms … ' The community's paradigm adds value to engineering practice, such as reducing technical uncertainties and leading to efficiencies in using human effort, materials energy and environmental disturbance (Trevelyan and Williams 2019). Structural engineering practice has its own literature, chiefly case history publications encapsulated in accident reviews, consulting or contracting organisation 'watch-it' notes, and articles appearing in the magazines of professional bodies. These documents are subject to the peer-review scrutiny of the professional community; they serve to inform judgement in practice.
The case for considering failure case history in engineering education has been made by others (Love, Lopez, and Edward 2013;Petroski, 1991;Alexander 1964). Petroski (1991, 83) argues that the interpretive analysis of a design structure that has failed by collapse is accessible to all. Whether this be a full collapse of a building, a partial collapse of an element of a building, for example, an atrium, with accessibility is arguably made more synonymous and poignant where such collapse has led to human injury or loss of life. The judgement of structural error involves an intuitive and simple recognition of the 'misfit' between structural context and form, and as such offers a potential 'stepping stone' to understanding the fitness of a design to meet stated requirements: 'Engineering advances by proactive and reactive failure analysis, and thus at the heart of engineering method, is an understanding of failure in all its real and imagined manifestations'. While Alexander (1964, 102) proposes that 'we are never capable of stating a design problem except in terms of the errors we have observed'.
The concept of case-based learning is also not itself new, and it has been explored as an alternative to informative didactic delivery, with results confirming the relevance of the approach as a mechanism for increasing student engagement, attendance and fostering the relevance of realworld problems (Yadav, Shaver, and Meckle 2010;Scherer and Landel 1995;Fuchs 1970;Hoag, Lillie, and Hoppe 2005;Vesper and Adams 1969;Raju and Sankar, 1999;Thurston 1994). Casebased instruction has also been recognised to support the development of problem-solving skills (Chinowsky and Robinson 1997;Henderson et al, 1983). However, further work is still required to better understand the effectiveness of approaches that specifically mobilise case studies of failure and the outcomes of their adoption on the manifestation of engineering judgement within educational contexts.

Theoretical framework
Fundamentally, this work is grounded in a constructivist approach to learning, accepting that students construct new knowledge and understanding as they participate in lectures, seminars and assessments. More significant here is the positioning of engineering judgment as a critical aspect in the development of engineering design skills. It is suggested that developing students' understanding of engineering failure will, in turn, reinforce the foundations of their knowledge and the position from which they can consequently apply engineering judgement to future practice.
The aim of this study to understand whether introducing 'learning from failure', centralised around 'real-world' case studies, serves to facilitate the use of engineering judgement in structural design among undergraduate students. We, therefore, ask the research question 'does the introduction of learning from failure real life case studies facilitate students' use of engineering judgement?' To answer this question, we look to the empirical space of a specific module on a UK undergraduate degree course, and its assessed coursework for our data. Analysis of this data will enable the identification and exploration of demonstrable uses of engineering judgement, independently and in combination, as part of these coursework submissions. This enables us to answer our research question bluntly, in terms of yes or no and with nuance, as to how this use manifests within the data is explored.
Due to the exploratory nature and methodological underpinning of the study, no claim is made to generalisation, and we duly acknowledge the inherent limitations in our small sample size. However, we suggest there is novelty and contribution in the way in which we have explored the different types of engineering judgement within student practice and added a further facet to the conversation on-going around learning from failure and how it can best inform and shape engineering education going forward.

Empirical context
Here, case-based learning is examined as introduced to a first-year structural design module for civil engineering students, which ultimately formed part of their module assessment. The students were enrolled on an undergraduate bachelor's degree in civil engineering at a UK university. One of six first-year engineering modules, 'Introduction to Structural Design' was undertaken over a full academic year. Students had three hours of direct instructional contact a week with the same academic tutor who had gained structural engineering experience in professional practice. Sessions were delivered in the format of a two-hour lecture and one-hour seminar, allocating a further five hours and twenty minutes for self-study weekly. The learning aim of the module was to disseminate the foundational principles of structural design and the communication of design work. The cohort consisted primarily of direct school entrants aged eighteen, with a predominantly 90% to 10% male to female gender split. Within the first two weeks of their studies as an inaugural activity, students were asked to autonomously research case histories of previous engineering failures within a peer group setting. At this stage in the academic year, students had no experience of structural design. The cohort (n = 60) formed into groups (n = 15), and each group was provided with two case-study project titles and the dates of the failures that had occurred by the lecturer. One project title was selected from a recent failure occurred in the last twenty years and the other an older project. Students were then asked to define the principal structural form and behaviour of the structure involved in the failure, thus mirroring the diagnostic element of engineering judgement. Furthermore, they evaluated the project against various themes of failure, as given in Table  1 (a topic on which lectures were also provided), thereby requiring the students to use inductive reasoning, considered the second aspect of engineering judgement, to assess the engineering parameters of the case history project. Finally, the students were asked to interpret the outcomes of the evaluated themes and to determine the key lessons to be learned from the failure, and whether the project could be considered 'good design'.

Data collection
The data used for the study were the student's group assessment submissions (n = 15 groups) for the module. These were in the form of a poster presentation in which each group set out their analysis of two distinct failure case studies. This resulted in n = 28 case studies for analysis as one group combined their two case studies into one, and one group only provided one case study within their submission.

Analytical framework
Evaluation of the students' knowledge and understanding has been explored through the three aspects of engineering judgement found within their assessment submissions, revealed by the ways in which they presented their research of specific case-history failure. This approach is summarised in Table 2.
As noted in the final column of Table 2, this study adopts content and thematic analysis in its approach to the data, informed by a wider constructivist methodology. For the first aspect of engineering judgement, content analysis was applied to reveal how the students defined the problem itself (the diagnostic aspect). More nuanced thematic analysis was then used to reveal how they brought together different evidence in the form of the 'themes of failure' to support this positioning (the inductive aspect). Patterns were revealed by this analysis through an inclusive coding process, which was also able to reveal their relative prominence and relationships within the data as a whole. The final aspect of engineering judgement, the interpretive aspect, was explored through a thematic approach that looked to the data as a whole, able to illuminate and present the ways (as distinct from the content) in which the students positioned and understood failure within wider engineering contexts. In mobilising these different approaches to qualitative data analysis in these complementary ways, we can determine the use of engineering judgement in practice while also revealing how it has been used by the students.

Data analysis
Content and thematic analysis was used to explore the text and images found within the students' group poster presentations. A data-driven coding process was used to ensure the approach was as inclusive as possible and allow major themes and interpretations associated with engineering failure to be identified.
To provide inter-rater reliability in this subjective process, two researchers coded the data independently, to explore the patterns of use with regard to the ways in which the students defined the Thematic analysis of the assessments as a whole problems, and drew on combinations of evidence for failure. The coding was initially framed through the aspects of engineering judgement, as in Table 2, with sub-codes generated by the data themselves, as specific to each case study. Once the coding process had been undertaken independently, the two coding matrices were shared, discussed and ultimately combined. For the diagnostic aspect of engineering judgement, content analysis was used to initially determine their delivery of the criteria as required by the marking scheme and the patterns found within the data as a whole. This relatively simple analysis was also retrospectively checked against the formal marking of the assessments (undertaken prior to this study) and was found to correlate precisely.
For the inductive aspect, more complex coding naturally emerged; and the subsequent discussion process between the researchers distilled the codes into a manageable number of inclusive categories, with relevance to engineering judgement and the wider context of construction, while retaining the finer-grained analysis beneath. This approach necessitated drawing on the experience and knowledge of the two researchers: one is a Chartered Structural Engineer and the other a Chartered Construction Manager, to enhance validity. The code of 'excessive working hours' was included within the category of 'management', as was the 'lack of a clear reporting structure on the site', while a 'lack of connection detailing' was included within the category 'design'. This approach reveals the dominance of particular patterns and associations linked to the diagnostic and inductive elements of engineering judgment. The appropriate allocation of sub-codes to the higher level categories was consequently able to provide a determination of 'quantification' within the data to enable evaluation of student judgement and an evaluation of its depth and complexity.
To reveal the interpretive aspect of engineering judgement through the context and influences on failure, a less structured approach was used. This involved the development of detailed notations on copies of the posters themselves, which over time supported the development of a 'mapping' of the dominant themes, again developed first independently and then collaboratively by the researchers. These themes, once crystalised within the mapping, were reconsidered within a holistic interrogation of the data, which focused on patterns of consistency and inconsistency, variations, emerging themes or representations, and patterns of nuance, contradiction and repetition (Wildemuth 2016). Again, a flexible approach allowed for the dominant themes labelled by the researchers to emerge or disappear as the process continued.
Throughout both analyses, a constant comparison method (Silverman 2001) was used, with evaluation within and between the data sources part of an on-going process which led to the development or collapse of the coding framework/mapping over time. Consequently, multiple and repeated passes were made of the data, developing a high level of individual researcher confidence in its processing and enhancing the validity and accuracy of the findings as they emerged, as supported by the use of inter-rater reliability. A systematic investigation is essential to ensure rigour within the analytical process (Taylor 2001), and although thematic analysis is an interpretive process, and the researchers' skill (and in this instance, their own professional knowledge and judgement) in the identification of patterns and variations is itself critical (Potter and Wetherell 1992).
Results are presented here in a narrative form although, where relevant, numerical values are also used to provide context for the thematic interpretation presented. Where examples are used, they are representational of the wider data as a whole. This approach has been taken in large part to reflect the exploratory nature of this work, and thus to tell the story of our findings, rather than present quantitative analyses, for example, that would lack validity given the small sample size.

The diagnostic element of engineering judgement
Analysis of the diagnostic element of engineering judgement revealed how students defined the case study failure event. Benchmarked against the assessment criteria 'to define the principal structural form and behaviour of the structure involved in the failure', findings showed that over half (n = 19) of the case studies submitted, the students correctly applied diagnostic engineering judgement in their definition of the problem. The group presenting the Charles De Gaulle Airport roof failure in 2004 noted that there were signs of deformation that continued until the collapse … a result of creep and shrinkage in the concrete' [and that] 'cracks had developed where the footbridges were fixed … cut into the concrete shell [which] severely weakened the structure.
identifying the behaviour of the structure that led to the collapse, alongside further consideration of the causes and consequences of that behaviour.
Where students did not achieve this, the analysis showed that in all cases, the students' explication of the case studies was descriptive not analytical, and they presented their research as the 'story' of the case study without further evaluation or a stated problem definition, despite the explicit inclusion of this element within the marking scheme for the assessment. In their consideration of the Can Tho bridge collapse, the group noted that … a section of the approach ramp collapsed from over 30 meters in the air over an island in the river … after 8 months the investigation revealed a number of factors that may have caused the disaster … With the noted factors then listed in the post text. However, this list was published following the formal investigation and lacked any additional analysis from engineering perspectives by the students, with reference to the principal structural form and its behaviour during the failure. It should be noted here that three groups failed to achieve the demonstration of diagnostic engineering judgement in both of their case studies, while the other three groups only lacked this element within one of their submitted case studies, meaning from a cohort perspective, the majority of the groups demonstrated this aspect of engineering judgement within at least one aspect of their submitted work.

The inductive element of engineering judgement
Analysis of the themes of failure codes and the patterns of their use in terms of complexity and combination enabled evaluation of the inductive element of engineering judgement, exploring how the students positioned the case study within the broader scope of engineering failure. The data ultimately generated five codes for the different themes of failure used within this diagnostic element: Design, Management, Loading, Materials and Profit. For coding to be assigned, the students were deemed to have clearly articulated this activity appropriately within their poster submission and positioned it as a causal factor for failure within the case study. The analysis showed that in all case studies the students had identified and explicated at least one appropriate theme of failure, as found in Table 3.
The students also demonstrated a multiplicity of themes within their analysis. In most of the case studies, n = 22 or 79%, the students had drawn on more than one theme of failure within their assessments, bringing them together in a synthesis of the evidence (Vicks 2002), thus demonstrating the inductive aspect of engineering judgement. Within the majority of the case studies where inductive engineering judgement was exercised, the students actually mobilised between 2 and 5 themes (n = 14 or 64%), and in over one-third, they drew on more than 5 (n = 8 or 36%).
The two most frequently mobilised themes were those of Management (51 instances) and Design (29 instances), arguably reflecting the realities of structural failure within civil engineering, which readily result from a combination of design and site management issues. However, this can also, in part, be explained by the finer-grade analysis and sub-codes that sit behind these themes of failure. This analysis was able to reveal how the students demonstrated their awareness and understanding of the different aspects of the practices of Management and Design, adding depth and detail to the inductive aspects of their engineering judgement. Within the Newport Docks case study, Management was underpinned by the three sub-codes below, with quotes from the text also included for illustration: i project blame culture: 'Workers were concerned about reporting possible dangers … as they could have lost their jobs' ii reporting issues: 'the lack of a defined route of action for the walking ganger to take upon receiving reports of movement … ' iii drive for profit: 'the wealth increase in import and trade suggested the people were more driven to expand the docks, arguably this lessened their attention to detail in design and implementation'.
While on the Tay Bridge case study, Design was underpinned by four sub-codes: (i) inaccuracy in loading calculations: 'the design was well known … but these structures didn't face the same forces that a railway bridge would … ' (ii) a lack of design experience and knowledge: 'Bouch had not assessed the new loads on the bridge so that the strength of the bridge could be adjusted subsequently'. (iii) inaccuracy during changes to the original design: 'reduced the number of piers making the spans of the superstructure girders much longer … without taking into account wind loadings' (iv) under specification of connection details: 'Bouch didn't specify well enough if he wanted the holes drilled or not, so the metalwork company used the cheaper option which subsequently had two-thirds of the strength calculated'.
It must be noted that the complexities of this judgement were naturally dependent on the bespoke nature of the case studies themselves, and thus not all case studies can generate equal levels of complexity behind their failures; hence no further analysis of the strengths or details of these relationships has been carried out, and we have refrained from more detailed quantitative analyses here. However, this does suggest that the use of failure case studies is, to some extent, effective in facilitating students to develop their diagnostic engineering judgement, considering, analysing and combining a number of factors and different sources of evidence from their case studies, and this is unpacked further in the following section.

The interpretive element of engineering judgement
By mobilising thematic analysis in a way that encompassed the data as a whole, taking a broader and more holistic point of departure for the coding process, the students' interpretive understandings of the wider contexts of failure were explored. This approach was able to provide broader insights from 'Wind speeds were calculated over an average of a one-minute period and the engineers did not consider that the structure may be vulnerable to strong short gusts. The safety factor had not covered the uncertainties in the variable action due to the wind'. Profit 'The store's owner, Joon replaced the original contractor, Woosung Construction, as their employees tried to inform Joon the modifications he wished to put in place would put far too much strain on the structure's supports. Joon then dismissed the company, replacing them with his own construction company (Sampoong Construction) to finish the construction of the building. The design modifications were illegal' Materials 'Concrete typically is considered to be at its full strength after 28 days however lift 28 had been in position for less than 24 hours before it was loaded. This may have resulted in the compressive strength being far too low and contributing to the collapse' the ways in which the students drew upon and positioned various influences, including those revealed through the previous analysis, within their case study assessments. Building upon the foundations of diagnostic and inductive judgement, this analysis sought to illuminate the contextualisation and 'meaning' the students then presented for case study projectsthe reasons why things went wrong. The dominant themes that emerged from this part of the data analysis were engineering practice; causality; and learning from history.

Engineering practice -'the design was wrong'
Perhaps unsurprisingly, prominent within the data surrounding the case-history failures was a theme of 'engineering practice', closely associated with the two themes of failure of Management and Design. This engineering practice theme drew on a professional context from structural engineering and construction management activities and mobilised them through practical and tangible examples as contributory 'factors' of failure. This was also often associated with blame, at times even directed at specific individuals such as the engineers who led the projects, and responsibility for the failures positioned as either a consequence of individual poor practice or in the mismanagement of subordinates. As highlighted above in the Tay Bridge case study, the students were quick to judge the lead engineer personally, stating that 'poor workmanship and management from Thomas Bouch had allowed corners to be cut … ' The need for experience, training and qualification to make engineering judgements was prominent within the data, again often closely associated with assigned ownership of the decisions that led to the failure. In the case of the Barton Bridge case study, the students highlighted that the 'tower designer was a 24 year old draughtsman at the scaffolding company with no formal qualifications', while in the case of the Quebec Bridge, the lead engineer delegated site management to another who 'was not up to the task of supervising the construction on site due to previously being a desk engineer' This multi-faceted theme drew on a language of calculations, stresses, loads, stability, geotechnical knowledge, restraints and detailing to develop highly technical design discussions, able to generate relatively simple cause and effect pathways to the failure itself, as one student group summarised quite simply: the design was wrong And by logical extension, so was the designer(s). A further consideration was the positioning of rigour in the processes of engineering as a contributory factor to support the technical aspects of practice. The need for checks, approvals and appraisals of design, both initially and after any project or design change, was frequently positioned as a potential point of failure by the students, adding further aspects of professionalism and 'good practice' to the wider theme of engineering practice. In the case of Barton Bridge, the students noted that " … such operations should not take place without the approval of a structural engineer".

Causality: 'no single fault'
Alongside the theme of engineering practice, other themes were identifiable that positioned this technical engineering discourse within a more practice-based, real-world context. Dominant here was the consideration that causality was and is itself complex, countering the relatively simplistic allocation of failure to design 'fault' and instead of developing more nuanced understandings of failure as a complex and multi-faceted 'thing', in and of itself. This theme was very much connected to the inductive aspect of engineering judgement, where students first realised the variation and nuance in the evidence, linked back further to the failure itself as diagnosed through the problem definition. The prominence of this theme within the data, therefore, suggests that interpretive judgement is itself facilitated and strengthened by the outcomes of the first diagnostic then inductive judgements.
Although students' ability to explore and unpack such complexities varied, this theme could be found in some form within all the case studies, drawing on 'a combination of errors' to illuminate the various 'factors' that created such complex causality in practice. The students examining the West Gate Bridge case study concluded that 'the failure of the bridge was down to uncertainty in extreme loading which led to serviceability and elastic instability failures. Those on the other hand were caused by the unusual method of construction … ' This theme of complexity in causality mobilised two key strands for its development: the consideration of aspects associated with construction management and aspects associated with construction practice, as identified within the wider themes of failure. As with the sub-themes found within the inductive theme of Management, here production pressures in the form of time and money as contributory factors to failure were identifiable sub-themes, as well as procurement routes, client decisions, project change, subcontracting, contractual arrangement and other more intangible aspects, such as project prestige. Several of these could be identified in the Tacoma Narrows Bridge case study, for which the students noted: 'social demands … economic need … cost savings' as drivers within the project, but also aspects of poor construction practice, notably 'the seals were damaged when the bridge was sand-blasted before being painted so the effectiveness of the hydraulic dampers was nullified'.
The influence of such aspects on engineering practice, including specifically that of engineering design in the form of 'value engineering', could be found throughout the data. Failure within this theme becomes a consequence of (often commercial) practice, as drawn on by the students in their understandings of failure within the wider construction industry context.
Construction practice more specifically focused on the site itself and was used by students to bring considerations of site conditions, workforce competency, material quality, and the influence of proceeding and subsequent trades to their understandings of failure in practice. In the Cần Thơ bridge case study, the students explicitly noted the ' … contractor had decided to cut costs by not using the correct number of supports per bridge section and removing them before the concrete had cured'. Within both sub-themes, the understanding of construction team coherence and good communication was also prominent, particularly in case histories where early warnings of failure were evidenced, yet unacted upon.
Interestingly, both of these themes were also frequently interwoven with notions of blame and responsibility for the failure. In some instances, blame was also allocated by students onto those outside of the engineering profession, such as it was associated with 'poor workmanship', 'poor leadership', or with other named parties in the project, such as subcontractors or fabricators and those responsible for the design.
However, this segregation of engineering and construction practice also at times led to the development of a schism between the themes, which instead of acknowledging the role of complexity within the causes of failure, sought to other blame away from engineering practice and design. In the case of the Almuñécar Falsework collapse, despite the students state that: 'load transfers that needed to occur during construction were difficult' they then only included discussion of construction management and construction practice failure, opting out of wider considerations of multiple causality and with no recourse to engineering practice. In the Almuñécar case, the students swiftly moved to 'failure of workmanship' and focused on the construction processes. Such disassociation was, however, actually very limited within the data, and only four instances of such positions were identified within the data as a whole.

Learning from history?
An interesting shift in the students' analysis and demonstrations of interpretive judgement occurred depending on the age of the failure case history. For older projects (notably pre-2000) students mobilised a more 'dismissive' attitude to the failure, suggesting that such actions and consequences would simply not happen within contemporary construction operations. However, this was itself countered by an identifiable theme of 'surprise' when the failures were found in more recent case histories. As one student group noted: [it was] a surprise that the [company name] were responsible for such big failures. This is a welcome understanding to emerge from this analysis that the use of real-life case studies has created the space for students to learn that all engineering design, be it historic or contemporary, is vulnerable to failure in practice.

Discussion and implications for teaching practice
Overall, the findings from this study support suggestions made by others (e.g. Love, Lopez, and Edward 2013;Petroski 1991;Alexander 1964) that the use of failure case studies could be beneficial for engineering students, and here more specifically the manifestation of engineering judgement skills within undergraduate students. The diagnostic, inductive and interpretive elements of design judgement (Vicks 2002) were all identifiable within the analysis, further indicating that this approach would support Daly et al's (2014) call for new approaches to prompt deeper student enquiry to synergise ideas and prompt innovative judgment. Specifically, 69% of the posters demonstrated diagnostic judgement, 64% inductive judgement and the emergence of three dominant themes with further associated complexity and depth indicates the presence of interpretive judgement.
The diagnostic element of engineering judgement was explicitly requested in the assessment criteria for this submission, and as such should have been responded to directly by the students. This initial step in the analytical process is critical in supporting more sophisticated applications of engineering judgement, yet in some case students did not specifically focus on this element, describing rather than analysing the engineering failures in the cases. This could be a potential issue with the use of case studies, which always contain a descriptive 'story to tell', and indeed a detailed and chronological description of the case study (even if not presented explicitly in the submission) should be undertaken prior to the application of engineering judgement to ensure all facts and evidence associated can be collated and reviewed as part of this process. However, the fundamental difference between description and analysis can be a problematic distinction for the first-year students to make, and the relative ease with which the case study stories could be told could have blurred the line between description and analysis, as perceived by the students. It was this ultimate analytical step that was missing from some of the submitted work, and it is suggested that clearer direction within the brief to that end could resolve this issue. However, despite such concerns, the majority of the students undertook such analysis and so successfully demonstrated the diagnostic element of engineering judgement within their submitted case studies.
The use of case studies has been recommended as a valuable approach in terms of learning from failure (e.g. Mottram 2013;Love, Lopez, and Edward 2013;Lewis 2012).. This study further supports this notion, also finding them to be beneficial in enabling the students to identify and analyse the themes of failure found therein. More specifically, and perhaps more importantly, this study found that the use of case studies also provided the students with a rich context for them to unpack and consider, with regard to the potential combinations and interactions of evidence that contributed to such failure. While all students successfully applied this inductive element of engineering judgement in the analysis of at least one theme of failure within their case study, the majority brought together more than one theme in their analysis, suggesting that this use of case studies was able to support, if not encourage, enhanced evaluations of failure, drawing on the inductive element of engineering judgement to combine themes as appropriate. This has implications for the case studies used, and these should be selected to optimise the complexity that surrounded the failure, and thus provide students with the opportunity to enhance and optimise their learning through the assessment process. It also recommended that a combination of historical and contemporary case studies are used, to ensure students can appreciate the developments in the industry that have been undertaken for the better (such as those around workplace and labour conditions), and realise that such factors have long shadows and traditional ways of working (such as lowest cost tendering) do still hold influence in contemporary construction activities, avoiding complacency in that regard.
The prominence of the failure themes of 'Management' and 'Design' within the inductive judgement displayed by the students, in the single root cause analysis and within the more complex analyses of failure, is promising as it reflects the critical role played by engineers as managers and designers, and thus the acknowledgement of the consequences of a lack within for engineering failure. The use of various combinations of other themes and sub-themes alongside Management and Design within this aspect of engineering judgement was also promising, reflecting deeper and more nuanced analysis by the students as they unpacked the more complex case studies.
The appreciation and understanding of complexity carried through to the findings focused on the interpretive element of engineering judgement, as the dominant themes within the work also reflected the wider activities and factors that, through combinations of errors or failures, resulted in the case study failure itself. This aspect of engineering judgement is built upon the appropriate exercise of diagnostic and inductive judgements, which, in turn, combine and support the application of interpretive judgement specifically and more broadly.
The students' work showed understanding of the interrelationships between structural engineering practice and Management (both professional and site), thus situating their own role and associated responsibilities (Design) appropriately within wider industry practice and enhancing the associated learning. This is important for practice, given the need for designers to fully appreciate the close relationships between design and construction and use their structural design judgement appropriately, duly cognizant of the potential consequences for practice. The use of blame allocation for failure by the students was also interesting, demonstrating a reductionist analysis of the overall failure to those at fault, while also reinforcing the role of the professional engineer and their responsibilities in practice. Taken together, this suggests the use of case studies actually enhances the exercise of interpretive engineering judgement through this assessment, as students naturally developed a narrative able to support specific lessons learnt (Vicks 2002) from each failure case study.
Without case studies, it is arguable that the depth and nuance found here within the students' assessments (and thus learning through assessment) would be much harder to achieve. Classroom-based problems are often more simple than workplace problems (McNeill et al. 2016), inevitably limiting the learning potential therein. Findings show that in this study the use of specifically selected case studies enabled the students to better appreciate and analyse the impact of perceived 'non-engineering' considerations, such as site management or profit prioritisation, on engineering practice, and more importantly how they had impact. As the influence of such factors for failure can be significant, this is arguably vital learning essential in helping students develop a deeper structural engineering judgement, able to find resonance and relevance with real-life situations and the actual environment of engineering practice.

Conclusions
This research aimed to explore whether the development of judgement skills in structural design could be facilitated by the introduction of 'learning from failure' case studies into the structural design modules of the first year of an undergraduate civil engineering degree programme. Analysis of the students' group assessment submissions found that their use of case studies demonstrated all three aspects of engineering judgement: diagnostic, inductive and interpretive. In addition, how diagnostic and inductive judgement subsequently informs and supports the exercise of interpretive judgement within this context is also worthy of note.
Further work is required to refine this approach and to determine which types of a case study are able to maximise student familiarisation with, understanding of and ability to adopt the different elements of engineering judgement. The strengths and weaknesses of this approach, compared to other methods, should also be explored through comparative control group analyses to enable the development of an optimal approach to the teaching analytical skills to students and thus optimise the development of their structural design judgement throughout their courses as a whole. It is also accepted that engineering judgement is not constrained to structural design, and thus this work can inform other aspects of engineering education where judgement is also necessary and can be informed by learning from failure, in consideration of ethics within engineering.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Notes on contributors
Vikki Edmondson is a Senior Lecturer in Civil Engineering and teaches on BEng and MEng Civil Engineering degrees, with a focus on Highway Engineering, Structural Design and Transportation. She is a Chartered Engineer and Fellow of the Institute of Civil Engineers.
Fred Sherratt is a Professor of Construction and has taught on BEng degrees, with a focus on construction management and civil engineering technology. She is an Associate Member of the Institute of Civil Engineers.