Unintended consequences: institutional artefacts, closure mechanisms and the performance gap

ABSTRACT Renewable technologies often feature in policies to improve the energy efficiency of buildings. Designers introduce predicted energy values for specific technologies, but are surprised when the technologies fail to perform as expected. Three building projects are used to explore the effect of construction processes on the energy performance of building-integrated photovoltaic (BIPV) technology. In two cases BIPV failed to deliver expected energy generation, while in the third, dramatic changes in project processes and technical specifications were needed to achieve the specified output. A social construction of technology (SCOT) analysis documents how the energy generation of BIPV disappeared from view at certain points as actors focused on building features. A contribution is made to the theoretical development of SCOT by responding to two issues: privileging of cognitive closure mechanisms and the neglect of institutional analysis. The concept of inflection mechanisms is introduced as a second type of closure mechanism. More specifically, the role of institutional artefacts (e.g. planning requirements and schedules) in the construction process is found to contribute to the performance gap. To reduce the ‘performance gap’, practitioners need to focus on the distribution of design responsibility, sequencing of work and the location of expertise.


Introduction
This paper is about how plans for the energy generation of renewable technologies often fail to deliver due to a myriad of seemingly unconnected decisions and a succession of unintended consequences. The construction sector is consistently identified as critical for sustainable development in general and for energy savings in particular (IPCC, 2007). While a wide range of technical solutions have been proposed (including better fabric design and renewable technologies), policy-makers and sustainably minded professionals are increasingly concerned by the failure of many of these formulae to deliver on their promise (Palmer, Armit, & Terry, 2016;Zgajewski, 2015). The term 'performance gap' captures this concern. While it is generally defined as the gap between the energy performance of a building as designed and as built, the term has also come to signal a general frustration with the underperformance of supposedly green buildings. Within that conversation, renewable technologies occupy pride of place for both their promise and the disappointment over their performance in use.
Most discussions of the performance gap focus on either energy modelling or building occupants and their engagement with supposedly green buildings. More recently, construction professionals have begun to reflect on their own contribution to this phenomenon. A report by the Zero Carbon Hub (ZCH) focused on 'how and where the Performance Gap occurs within the current housebuilding process' (ZCH, 2014, p. 2). The report identified 15 issues for priority action, 17 issues as a priority for research and 23 issues to watch, each corresponding to different stages in the building process. Stages included: concept design and planning, detailed design, procurement, construction and commissioning, and verification and testing. This paper contributes to that work by analyzing the effects of project and construction processes on energy performance of BIPV systems at handover. Whereas the ZCH report sought to develop a comprehensive list of discrete factors, mapped onto a prespecified set of stages, this paper adopts a more holistic approach. As such, the current paper is both narrower and broader than the ZCH report. It is narrower as its focus is on a single technology (BIPV) and three building projects. It is broader as it explores the dynamic interaction between seemingly discrete issues and considerations, project stages and the resulting performance.
The choice of BIPV for this study lies in its integrated character, such that construction professionals and building design considerations are necessarily involved in the optimization of the technology. Far from a unique characteristic, a number of renewable technologies, including ground-source heat pumps and thermal mass-storage systems, share this feature. To signal the physical integration of the technology into the building, the paper takes as its technical object the BIPV system and its interfaces with the building (referred to as the BIPV/building).
The paper begins with a simple question: how does BIPV come to deliver less energy than initially expected (or than it potentially could) on three building projects? In two of the three cases, the energy generation of the BIPV/building was negligible, whereas in the third it was significant, but involved significant changes to the 'business-as-usual' project processes. The comparison across three cases serves to identify a number of construction-related considerations that affect the performance gap for BIPV in particular and building integrated renewable technologies in general.
A social construction of technology (SCOT) approach is adopted to explore the energy performance of BIPV. SCOT is one of a number of micro-level network theories used to explore the social construction of technology. An initial pilot study was used to develop the basic approach (Boyd, Larsen, & Schweber, 2015). It focused on the multiplicity of technological frames informing the ongoing development of BIPV/buildings. While it introduced the idea that institutional artefacts such as project schedules affect the configuration of BIPV/building, the absence of holistic case studies precluded an exploration of this suggestion. The current paper draws on the findings from a much more rigorous and extensive SCOT analysis of three building projects (Boyd, 2016). Theoretically, its contribution lies in the identification of a set of inflection mechanisms that capture the way in which institutional artefacts enter into ongoing negotiations over the BIPV/building and ultimately affect BIPV performance.
The paper is structured as follows. First, a brief overview is presented of the literature on green building and the challenges that construction professionals face. This literature underlines the importance of extra-technical considerations in the incorporation of renewable technologies into buildings. This is followed by a brief discussion of the literature on the performance gap and, more specifically, those studies highlighting the contribution of the design and construction process. The literature review concludes with a discussion of SCOT and its use in this paper. Key features include: a focus on a succession of problem/solution chains, the documentation of unintended consequences; attention to closure mechanisms; and a bounded concept of a network (which renders visible the effect of professional conventions and external requirements). For a more in-depth discussion of the difference between SCOT and other socio-technical network approaches including actor-network theory (Latour, 2005) and large systems technical analysis (Hughes, 1983), see Appendix A in the supplemental data online.

Building-integrated photovoltaic (BIPV) technology: background and context
Photovoltaic (PV) technology uses a suite of technologies to generate electricity from solar radiation. PV systems consist of: the PV cells, which convert solar radiation into electricity; the matrix, usually glass, in which they are embedded (often referred to as 'solar panels'); cables, which carry the direct current (DC) power from the panels; inverters, which convert the DC electricity to alternating current (AC); and cabling from the inverters to the standard supply metering system. To optimize electricity generation, each part of the system must be matched. The way in which strings of cells are wired together, the sizing of the inverters and the overall length of wiring runs have considerable impact on the overall generation potential of the BIPV system. Electricity from PV systems can be used to power the building where it is installed or can be exported to the grid. PV systems are installed in two main ways: building-applied photovoltaics (BAPV) and BIPV. The technologies remain similar, but the challenges of their installation differ greatly (Holden & Abhilash, 2014).
BAPVs are usually situated on roofs. The PV cells are mounted on top of the roof membrane and are not part of the structural element of the building. As the panels sit on the roof, often on a framework, there is little impact on the building structure and so BAPV are often installed as a retrofit technology to existing buildings. The major challenge for this technology is to maximize generation from the number and position of the panels and to minimize efficiency losses from cable runs by siting the inverter as close to the panels as possible. Contracts for BAPV installation in the UK are often turnkey and generally regarded as an add-on to the main design and construction of the building (Holden & Abhilash, 2014).
BIPV is very similar to BAPV in terms of its components. However, the key distinction for BIPV is that the PV panels are integrated into the fabric of the building rather than being placed on top of the structure. This integration into the building structure can include using BIPV in the roofs, windows, facades, louvres, brise-soleil and rain screens. With BIPV, the panels replace conventional building materials in part of the building, e.g. roof tiles, facade panels or window glass. The function of the panels is a combination of electricity generation, architectural aesthetic appeal and building function (in terms of water tightness, strength, durability etc.).
Unlike BAPV, BIPV has many interfaces with the rest of the building structure, which makes it both expensive (by virtue of its bespoke nature) and complicated (because of the number and type of interfaces with the building) (Henemann, 2008). In the UK, BIPV is generally restricted to commercial building projects, where each building is uniquely designed and where the adaptability of BIPV installations allows the technology to fit and contribute to the building architecture. The bespoke nature of the technology and the knock-on effects of its incorporation into a building project pose major challenges for construction professionals.

Literature review
The need for a socio-technical approach to the study of green building and the performance gap has been widely acknowledged. Within this literature, empirical case study research has proved invaluable in exploring the practical challenges that renewable technologies pose for construction professionals. The discussion that follows focuses primarily on this work, although an overview of the literature has also been introduced to identify the broader conversations to which these case studies contribute.

Green building
The challenge of green building for construction professionals has been periodically noted. Rohracher (2001) published a general statement underlining the multiple product and process challenges associated with green building and calling for socio-technical approach. Häkinnen and Belloni's (2011) study of barriers and drivers for sustainable buildings took up a similar call, as did Schweber and Haroglu's (2014) research into variations in the 'fit' between the Building Research Establishment Environmental Assessment Method (BREEAM) and the building process. A number of studies focus on the role of one or more key actors in achieving this aim. Gluch (2009) explored the role of environmental managers, while Parag and Janda (2014) highlighted the role of middle managers. This work has similarly been matched by professional bodies interested in promoting their members' specific contribution. The Specialist Engineering Alliance (Macmillan, 2009) studied the complexity of the sustainable building supply chain, whilst more technical guides identify the complexity of interrelated components (e.g. BRE, EA Technology, and Sundog Energy, 2002).
In most of this work, 'green buildings' are defined as those that limit their negative environmental impact, generally with reference to either waste, energy and/or water; in almost all these studies, 'integration' is seen to be the primary condition for success. A key difference between these studies lies in the kind of 'integration' problem that they identify as critical and in the associated solution which they propose. Rohracher (2001) identifies the need to integrate different stakeholders within the design process, whilst Häkinnen and Belloni (2011) espouse the importance of integrating construction processes. The SEA used a banner of 'Sustainable Buildings need integrated teams' (Macmillan, 2009, p. 1) to highlight the need for integrated delivery teams, whilst the BRE (2002) guide signals a need for the integrated design of system components. This paper introduces another type of challenge by exploring the incorporation of integrated renewable technologies. While it starts with a concern for the physical integration of the BIPV technical system into the building, the approach and findings underline the interdependence of the different integration issues.
As these examples all illustrate, case study research has proved critical in developing a sector-specific understanding of the challenges of green building. Contributions can be divided into a managerialist literature focused on developing frameworks, decision-making tools and evaluation methods, and a more exploratory literature concerned with identifying the barriers to and opportunities for 'green building'.
In terms of research approach, managerialist studies tend to involve some type of experimental research design, be it a modelling or a simulation exercise. In these studies, empirical case studies provide an opportunity to develop and test management and assessment methods. For example Von Malmborg and Forsberg (2003) use life cycle analysis (LCA) to evaluate different heat and electricity mixes in three commercial buildings. Hassan (2006) builds on earlier attempts to integrate existing management tools, including total quality management (TQM), LCA and value for money (VFM), amongst others, to develop a managerial framework aimed at supporting green building. Other studies seek to develop multi-criteria decision-making tools (Langston, 2013;Matar, Georgy, & Ibrahim, 2008;Shen & Walker, 2001). In these studies, the technology is treated as a fixed component that, once selected, plays no further part in the development of the building. This is clearly evidenced in the neglect of challenges concerning the introduction or installation of renewable technologies. This omission can partly be explained by a radical distinction between technical and social dimensions and a privileging of the choice of either technologies or social factors, such as communication and skills.
In contrast, empirical case studies tend to analyse 'real-life' projects. A review of the literature revealed a surprisingly small number of this type of paper. Notable exceptions included Fedoruk, Cole, Robinson, and Cayuela (2015), Brown and Vergragt (2008), and Albino and Berardi (2012). Each of these papers documents the complexity of both the technology and the project environment. They also draw attention to the ongoing need for fine-tuning and to the obstacles that conventional construction management processes pose. This understanding, that both social and technical issues are at play, reinforces the need for further exploration. In particular, this creates a space for an approach that links the development of technology, ongoing building design, the network of actors involved and the various decisions that shape a building project.
The 'performance gap' The performance-gap literature differs from the work on green building in its exclusive focus on energy and in its framing of the challenge as one of 'sticking to the plan'. The concept of the performance gap refers to the gap between the intended energy performance of a building (as designed) and the energy performance in use. Research into the performance gap can be divided into three categories: work on the modelling of energy performance (De Wilde, 2014;Menezes, Cripps, Bouchlaghem, & Buswell, 2012), work on building occupants and the effect of their behaviour on energy performance in use (Ornetzeder & Rohracher, 2006;Sunikka-Blank & Galvin, 2012), and a third small but growing literature on the role of building delivery (Dainty, Thomson, & Fernie, 2013;Gorse et al., 2012).
Viewed from the perspective of the construction industry, the energy performance of buildings is often disappointing (Zgajewski, 2015). The concept of the 'performance gap' rests on a particularly rigid, stylized understanding of the construction process. It assumes that building designs are fixed early on in the process and treats subsequent changes (and in particular those that affect the energy performance of the building) as a problem. This image contrasts sharply with the experience of construction professionals whereby design decisions continue to be made throughout the delivery process, often for very good reasons, ranging from changes in client funding and goals to unanticipated problems with the overlay of systems or procurement issues (Hanna, Camlic, Peterson, & Nordheim, 2002). In this sense, the performance gap is better understood as a gap between energy performance as modelled (at a relatively early point in the design process) and energy performance in use. Moreover, performance-gap studies tend to assume that the energy performance of a building is a clearly understood parameter that is at the centre of professional attention from initial concept through to the commissioning of a building, whereas in fact the target is not always clear, measureable, visible or consistent.
While the concept of a performance gap may not be straightforward, empirical research into the problem has enriched understanding of the implementation of renewable technologies. Empirical case studies show that adoption is not a simple, one-way process and this points to the need to take into account standard building practices and performance measurement (Fedoruk et al., 2015).

Social construction of technology (SCOT)
SCOT is one of a number of socio-technical network approaches that were introduced in the 1980s (see Appendix A in the supplemental data online). While the approach has developed considerably, this paper builds on the early version. At its most general, SCOT depicts the development of a technology as a contest between different actors with different visions for its form and use. Technological development is marked by negotiations over a succession of problems and associated solutions (Bijker, 2010).
Within SCOT, acknowledgement of both the physical aspect of technologies and their socio-technical nature can be found in the analytical distinction between technical artefact (the early focus of SCOT research) and technological system, both of which figure as possible units of analysis (Bijker, 2010). Whereas 'technical artefact' sets the shifting configuration of a set of interlocking physical parts as a research object, 'technological system' takes the heterogeneous network of artefacts, meaning and people as its object.
For the purposes of this paper, this distinction allows for an analysis of the changing network around BIPV/ building conceptualized as both a technical artefact and technological system. Analyzed as a technical artefact, BIPV appears as a collection of discrete component parts, the relationship between which changes as the BIPV/building develops. Components include: panels; inverters; wiring and control systems; as well as the parts of the building that are directly affected by BIPV, such as the building facade or electrical system. It is the panels in particular that are used to estimate the energygeneration potential of the technology. This paper explores the gap between initial expectations for the technology and its generation potential at handover. Analysed as a technological system, a BIPV/building is characterized by a heterogeneous network of human actors and physical and textual artefacts, which are constituted around and constitute specific project/solution chains. This model is useful as it allows for an identification of the succession of negotiations shaping the development of a BIPV/building and of the knock-on effects of one problem/solution chain on subsequent ones.
SCOT begins, like other network theories, with an assumption of 'interpretative flexibility' (Bijker, 2010). The concept refers to the multiplicity of different interpretations which are ascribed to a technical artefact. This means that for any given technological system, different actors will define the technical artefact (around which the technological system is elaborated), the problem under consideration and the range of possible solutions differently. Viewed from this perspective, a SCOT analysis focuses on how particular actors manage to impose their interests and associated problem definitions and solutions on the developing technological system.
As indicated above, theoretical generalization in SCOT tends to be around mechanisms of closure (Misa, 1992). The term points to the gradual movement from negotiations and even competition for control over the development of a technical artefact to (temporary) closure. The point is not that the development is fixed forever, but rather that at some point in time a particular version comes to be taken for granted, such that subsequent changes are defined relative to that version of the technical artefact.
Initially, SCOT scholars privileged the role of understandings and meaning in the fixing of a particular technological system. One of the key criticisms of SCOT concerns its neglect of structure. In a widely cited paper, Klein and Kleinman (2002) point to the way in which social structures 'explain' why some actors and technological frames 'win out' over others. A key point for this paper concerns the attention they draw to structural factors affecting closure. These include power and dependency relations between actors and institutional rules governing decision-making (Klein & Kleinman, 2002, p. 39). SCOT scholars responded to this and related criticisms by exploring the role of power (Bijker, 2010;Bijker, Hughes, & Pinch, 1987;Klein & Kleinman, 2002). This paper, in contrast, picks up on Klein and Kleinman's (2002) second point, regarding the role of externally established (institutional) rules. More specifically, it examines the way in which those rules enter into technological systems through the medium of textual artefacts. The term 'institutional artefact' is used to signal the grounding of artefacts such as contracts in broader institutional arrangements. A central argument in the paper concerns the way in which these artefacts inflect the ongoing definition of both problems and solution sets. This effect is referred to as inflection mechanisms to distinguish it from the more cognitively driven closure mechanisms that most SCOT theorists address.
In contrast to the more familiar concepts of 'intermediary' (Latour, 2005) and 'boundary object' (Star & Griesemer, 1989), which link actors or networks together without introducing new content or weighting outcomes, 'institutional artefacts' do both. They introduce rules that have been set outside of the technological system with the explicit intent of directing ongoing negotiations. While those rules can be modified, it is only with great effort and often involves an appeal to the relevant institutional body. A key contribution of this paper and of empirical case studies more generally is to draw attention to the numerous unintended consequences that such rules and associated artefacts produce.
In sum, this paper contributes to the development of SCOT by taking on two longstanding criticisms, namely its privileging of cognitive closure mechanisms and neglect of institutional analysis. It focuses on the effect of textual artefacts that figure in the course of negotiations around specific problems and solutions. While the value or content of the artefacts are produced by and through the network in which they figure, the type of artefact and taken-for-granted assumptions of what general form they should take are external to the network. An analysis of these effects introduces a number of often overlooked aspects of the performance gap; it also contributes to theory development by adding a second type of closure mechanism, namely inflection mechanisms, to the SCOT toolbox. For further discussion of the way in which SCOT informed the research and the difference between SCOT and other networked theories, such as ANT, see Appendix A in the supplemental data online.

Methods
As indicated above, this paper uses data from a much larger SCOT analysis. Whereas that broader study explored the co-development three BIPV systems and the buildings in which they were incorporated (Boyd, 2016), the focus of this paper is on the effect of that process on the energygeneration potential of BIPV. To select the cases, the first author contacted a manufacturer of BIPV laminate panels who provided contacts for five new-build commercial projects, three of whom agreed to participate in the study. As indicated in Table 1, the building projects shared certain features and differed in others. All three were commercial buildings, all used the same laminate supplier and all used design-and-build contracts (a procurement method that supports early contractor involvement). The projects differed in the function of the buildings, the physical component of the building into which BIPV was incorporated and the drivers for the specification of the BIPV system (Table 1).
Data for the study included 28 interviews and two extended e-mail correspondences, conducted between February 2013 and June 2015. For each project, a loose type of snowball sampling was adopted (Bryman & Bell, 2003); interviewees were asked for names of other professionals involved in the ongoing development of the particular BIPV/building. Sampling was considered complete when no new names were suggested (Table 2).
Semi-structured interviews were used to collect data on the co-development of each BIPV/building. The structured but flexible nature of the method allowed the interviewer to both explore the interviewees' experience and query developments identified in previous interviews (Bryman & Bell, 2003). Interviews lasted between one and two hours and were recorded, transcribed and anonymized.
Analysis initially focused on the development of detailed SCOT diagrams detailing the succession of problem/solution chains contributing to the ongoing design of three BIPV buildings. This use of SCOT diagrams is novel; it was initially used as a pilot study (Boyd et al., 2015) and mobilized in the broader research project (for further explanation, see Appendix B in the supplemental data online). For the purposes of this paper, the authors focused on those problem/solution chains that directly impacted on the energy-generation potential of the BIPV/building. This produced a set of four to five problem/solution chains for each case. Each chain was then analysed for its effect on the energy performance of the BIPV system. Findings were captured in a detailed table, which is reported in Appendix C in the supplemental data online.
This analysis led to a focus on closure mechanisms in general and the concept of institutional artefact in particular, and the table was revised to include these issues. As the discussion that follows suggests, the same set of

BIPV/building projects
This section presents each project in terms of the problem/solution chains that affected energy performance of the BIPV system. The findings document how institutional artefacts inflected the definition of the problem and the range of conceivable solutions, and the affect of these (re-)definitions on the potential energy performance of the BIPV/building system at handover.

Vogue Terrace
Vogue Terrace is a commercial office building in Central London. It was part of a three-phase refurbishment project in which three adjacent blocks were reduced to a skeleton and then reconstructed. Although not exactly a new-build, the refurbishment was so extensive that it fulfilled the criteria for project selection. BIPV technology was incorporated in the brise-soleil louvres on the south elevation of the building. The development started in the mid-2000s, with Vogue Terrace being the last of the three buildings to be constructed. Initial planning permission for Vogue Terrace was granted in 2007; work on site began in August 2014 with work on the BIPV installation commencing in February 2015. Local planning requirements in 2010 did not establish particular generation targets for renewable energy; instead, they called for a sustainability review that included consideration of renewable technologies. The decision to incorporate BIPV into the brise-soleil louvres on the south elevation of Vogue Terrace was presented as both satisfying these planning requirements and providing the building with an up-to-date look. At this point, the energy-generation potential of the building was framed in terms of the number of louvres on the elevation of the building, rather than a specific generation figure.
During the initial design phase, the main contractor carried out a cost analysis and identified the BIPV brise-soleil louvres as a major source of capital expenditure. The client insisted that the BIPV system be retained, so the brise-soleil louvres were redesigned to increase the number of PV cells in each louvre but reduce the number of BIPV louvres. The intent was to maintain the initial design output at a reduced cost.
A key moment in the story of energy generation on the project came when the client insisted that the same contractors be used on Vogue Terrace as had been used in the previous phases of the project (which had not included BIPV). This led to a chain of decisions about who was responsible for what, which effectively masked the interdependence between the BIPV brisesoleil louvres and the BIPV wiring system. This decision to use the same contractor locked in the facade contractor who had no previous experience of using BIPV. Recognizing their own lack of experience, the facade supplier refused to include BIPV in their tender response. To accommodate the facade supplier, the main contractor redistributed the BIPV system across other work packages. This involved a further division of the BIPV contract into visible (panels and bracketry) and invisible (wiring, inverters and cabling) sections. As part of this decoupling of the BIPV system, and in an effort to maintain profit margins for the contractor, the PV panels were free-issued to the facade supplier and the electrical portion of the system was put together as a separate wiring package which was to be included in the main electrical contract for the project. As the project moved forward, project management conventions led the main contractor to issue the electrical contract work package as part of the main building electrical work package, which was after the design of the brisesoleil bracketry and frames had commenced. The subsequent refusal of the main building electrical contractor to take on the BIPV system design further blocked any possibility of an integrated design for frames, bracketry and wiring as the responsibility for the BIPV wiring was further subcontracted.
The result was that the BIPV was treated as a bolt-on installation, which lacked integrated design. The BIPV louvres were bolted onto the glazing units and the wiring was run vertically and externally up the building to the roof-mounted inverters, impacting the aesthetics of the building and increasing the length of wiring runs and reducing the efficiency of the BIPV system. The lack of an interface between the electrical and BIPV contracts meant that the electricity generated by the BIPV system was not part of the building's energy-management system and there were no integrated commissioning plans.
As this brief account indicates, in the case of Vogue Terrace there were no planning requirements for energy generation; the output of the BIPV system was only roughly estimated and the actual output was never measured. The second building project, Future Green, offers a different path to the final BIPV performance gap.

Future Green
Future Green is a commercial science hub building set on a 24-acre site in a large science park development in northern England. The BIPV system was incorporated into the windows of the south elevation of the building. Design for the project started in 2010; construction began in late 2013 and the installation of BIPV was completed by August 2014; the Future Green project was completed by November 2014. The project was a joint partnership between a university, the city council and several other partners.
Future Green used BIPV to win funding from the European Regional Development Fund (ERDF); in addition, BIPV was a sustainability statement and attracted tenants to the building. Minimum requirements for ERDF funding included the achievement of (at least) BREEAM Excellent and a rated energy performance certificate (EPC) of at least 'B'. The EPC criteria for the inclusion of BIPV is expressed in area (m 2 ) of solar panels, rather than specifying energy generation in kilowatt-hours (kWh). This shift in the measurement unit focused the project team on the physical attributes of the solar panels, rather than on the electrical output of the PV system as a whole. This shift in focus was important because it signalled the moment that energy generation was no longer a key factor for the project team.
When it came to procurement decisions, a reliance on pre-existing relationships masked the failure to design the BIPV system. Towards the start of the project, the main contractor apportioned work packages as though the BIPV system were 'just a set of windows'. The mechanical and electrical (M&E) engineer, with whom the main contractor had worked previously, was given the task of apportioning responsibility for the BIPV design, which he divided into two parts. Responsibility for design and procurement of the panels was assigned to the facade contract, whilst responsibility for the electrical aspect was included in the main project M&E contract for internal work. Like the electrical contract, the facade contract was awarded to a contractor with whom the main contractor had collaborated in the past. One result of this process was that the main contractor was not aware of the requirement for design portions in either of the subcontracts and missed the failure of both contractors to design his assigned portion of the BIPV system.
During the detailed design phase the architect (who was not novated to the main contractor and therefore acted independently), asked the facade supplier to reposition the BIPV cells within the glazed units so that the cell spacings were even and aesthetically pleasing. The facade supplier deferred to the architect in this decision even though as a result of the changes the output from the cells was reduced. The facade contractor subcontracted the glazing panels (including the BIPV panels), assuming that they would be fitted into the facade supplier's frames on site. This division of tasks and need to keep the project on schedule obscured the need for detailed design of the BIPV cell string configuration, which in turn led to a loss of generation potential as the string configuration was not optimized.
In order to reduce the effect of glare on the south and west elevations, and in keeping with current architectural practice, the architect had designed deep window reveals in the facade without considering the effect of these reveals on the BIPV generation. From the perspective of BIPV generation, this had serious consequences as it resulted in a total generation loss when the reveals cast a shadow over any of the PV cells in a string.
The overall result was that the BIPV windows made a strong, visible 'green' statement but their PV functionality was severely compromised.

Synergy Court
Synergy Court is an interdisciplinary biomedical research building in Central London that serves a medical research partnership between three national research organizations and three universities. The BIPV system was incorporated into roof fins on the building. Project planning began in 2001 and planning permission was granted in December 2010. Ground works began in April 2011 and BIPV installation began in 2014. The estimated completion date was early 2016.
Synergy Court was intended to be a flagship research centre. Local, negotiated planning requirements demanded that 1% of the electricity requirements of the building be generated from renewable technology onsite. BIPV was included in the building to meet these conditions and as part of the client's sustainability strategy. The planning requirement fixed the energy generation target to 120 MWh.
An initial problem arose with the insistence of the planning authority that the proposed shape of the building be modified. The redesigned building shape met the planning authority's requirements, but it also reduced the area available for PV generation, making it difficult to meet the generation targets. A fortunate by-product of this tension was that it forced the detailed design of the BIPV system on the agenda before the main contractor put the work packages out to tender. The work packages initially included BIPV fins within the roof louvre contract and the electrical work within the general electrical contract. Because of the particularly stringent generation target, all but one of the roof louvre suppliers contacted refused to quote on the package. This led the main contractor to issue a pre-contract design order (PCSA) to one of the roof louvre suppliers for a more detailed design of the BIPV system before the tender documents were finalized.
During the PCSA the roof louvre contractor suggested an innovative redesign using micro-inverters and subcollectors which would meet the BIPV-generation target. The contractor also insisted that the electrical work was included within the louvre work package as a turnkey contract. The main contractor agreed and as a result generation output and the BIPV system as a whole were taken into account in each subsequent design and installation decision.
The result of this procurement decision was that the BIPV system became an integral part of the building and a flagship technology within the flagship building. The generation potential of the BIPV system was designed to meet 1% of the building's energy needs and was connected to the building's energy management system, allowing web monitoring of the electricity generated.

Discussion
The comparison of three BIPV buildings presented above draws attention to the relevance of planning requirements, generation targets and the contractual distribution of responsibility for BIPV generation in particular and for the performance gap more generally. It also highlights the reluctance of many subcontractors to take responsibility for the design of this new technology and the distortions introduced by the successive passing on of responsibility. Every time contractual responsibility for the BIPV system design was either divided up or passed on, the risk of invisibility and, as a consequence the performance gap, increased. A summary of these three examples lays a basis for theorization and more practical recommendations.
In Future Green the BIPV system was translated into 'just a set of windows'. The result was that by the time someone thought to evaluate the energy-generation potential of the building, it was too late to intervene. In Vogue Terrace, planning requirements and the client's brief initially kept the use of renewable technologies on the agenda. The client's concern for a modern look led to the choice of BIPV brise-soleil louvres. However, at a certain point, the client's preference for a general contractor with whom he had already worked and a complicated set of contractual arrangements (motivated in part by the reluctance of any of the relevant actors to take responsibility for the BIPV design) effectively removed the design of the BIPV system and energy generation from the agenda. However, the Synergy Court building evidenced a different approach. The energy generation remained visible throughout this project. This effect can be attributed to the way in which externally imposed and challenging energy targets, together with early recognition of the challenge which this posed (in part because responsible contractors refused to take responsibility for something they could not deliver) pushed the project team to privilege BIPV system design over conventional ways of working and contending considerations.

Institutional artefacts and inflection mechanisms
From a theoretical perspective the concept of institutional artefacts captures these dynamics and their consequences for the performance gap. The term refers to (predominantly) textual objects that introduce rules or conventions into the development of the BIPV/building. Examples include planning requirements, client requirements, cost analysis, work packages and schedules. As these examples suggest, institutional artefacts are shaped by rules or conventions, which exist prior to and independent of any particular project. Relevant rules may be formal or informal; the important point is that they are socially recognized, such that deviation from them is an active choice that needs to be justified.
The term 'inflection mechanism' points to the way in which these institutional artefacts contribute to closure. Whereas most SCOT analyses of closure mechanisms focus on closure around a technology as a whole, the focus in this paper has been on the closure of specific problem/solution chains which contribute to that broader process. Also, where most SCOT closure mechanisms work through the achievement of consensus, inflection mechanisms affect technological development through their effect on taken-for-granted assumptions. More specifically, they affect decision-making by shifting attention from one definition of a problem to another, by drawing attention to certain issues and obscuring others, and by circumscribing the set of conceivable options. In contrast to cognitive closure mechanisms, their consequences are often unintended and unanticipated.

Three types of inflection mechanisms
A review of institutional artefacts at play in the three BIPV/building cases suggests three such mechanisms: the (re)-specification of the unit of analysis, the imposition of new parameters and recourse to convention. In the projects described above, these mechanisms worked to obscure and render visible both the BIPV system design and its energy generation potential.
The (re)-specification of the unit of analysis refers to the role of an institutional artefact in shifting the unit used in the evaluation of energy generation. For example, in the case of Future Green, the introduction of an EPC requirement of an 'A' or 'B' rating meant that area (m 2 ) was substituted for performance (KWh) in the specification of installed cells. This effectively redefined the problem and eliminated the energy target. Conversely, in Synergy Court the Merton Rule (a planning requirement) fixed a kWh target for PV, protecting the energy target from attempts to unseat it. Finally, in Vogue Terrace the unit of analysis shifted from energy generation to the number of brise-soleil louvres on the south elevation. A key consequence was to shift the set of conceivable solutions from BIPV system design to bracketry and framework issues, which in turn delayed consideration of BIPV wiring design until after louvre frames and bracket design had been fixed.
The second, related mechanism involves the imposition of new parameters. In this inflection mechanism, institutional artefacts introduce additional parameters into the problem/solution chain. This effect can be seen in the way in which a cost analysis introduced specific budgetary constraints into what had previously been a technical discussion over energy generation in both Future Green and Vogue Terrace. The result in Future Green was to shift the range of conceivable solutions to those that met the less expensive EPC 'B' rating. In Vogue Terrace a parallel exercise by the main contractor led to the free issue of BIPV louvres to the facade contractor, at the expense of an integrated technical design. Finally, in Vogue Terrace, the client's brief introduced aesthetic considerations that primed over economic ones. More specifically, the client's insistence on a 'modern'-looking building ensured the retention of BIPV brise-soleil in the louvres and kept energy generation on the agenda, at least for the short-term.
A third inflection mechanism involves the primacy of conventions, whether this is design conventions, project conventions or simply past practice. This inflection mechanism can be found in numerous moments in all three projects. In Future Green, scheduling conventions dictated a very short lead-in time for tendering. This in turn deprived the team of time for reflection needed to recognize and compensate for the way in which the work packages cut across the BIPV system. The effect was that both the electrical and the facade packages failed to take into account the BIPV/building design. Similarly, in the same project, conventional guidelines for how to cope with glare for east/west facades and shading for south-facing facades informed the set of conceivable solutions to the profile of the window reveals (obscuring the effect of shading on the energy generation potential of the BIPV system). Whereas in these examples professional conventions excluded energy generation from the ongoing definition of problems and set of conceivable solution, in Synergy Court planning requirements in the form of the Merton Rule kept them on the agenda.
One of the more striking indirect effects of this mechanism concerns the way in which professional conventions shape the types of expertise available at any given point in time and mask the absence of BIPV knowledge. In Vogue Terrace, the client relied on the well-tested method of hiring a general contractor with whom they had worked previously. While this may have reassured the client, it also created an expertise gap. In what seems from the outside like a jumbled succession of subcontracts and work packages, the design of the BIPV brise-soleil was passed like a hot potato from the architect to the main contractor to the facade supplier to the laminate supplier to the electrical contractor and ultimately to the BIPV wiring contractor. With each pass, contractual arrangements decoupled the system design, further diminishing the possibility that the experts, when they were finally brought onto the project, could salvage the energy-generation potential of the BIPV system.
Similar impacts of conventions on the presence or absence of technical expertise can be found in Future Green. In keeping with convention, the main contractor relied on the M&E design engineers to define and split up the work packages, and the M&E engineers relied on the M&E and facade subcontractors to each design parts of the BIPV system, although neither had had experience of BIPV systems. The scheduling conventions of design-and-build contracts (a procurement type in which the contractor is brought on relatively early in the process and represents the client) relied on fast turnaround of the tender process, which precluded detailed design of the BIPV elements and masked the effect of the deep window reveals on the energy-generation potential. Each actor in the chain was convinced that the non-existent BIPV expert was in charge of the system design and that all would be well. In both Vogue Terrace and Future Green, the way in which conventions shape the types of expertise available resulted in BIPV systems being installed without ever having been designed. For a complete table of these mechanisms and their effect on specific projects and problem/solution chains, see Appendix C in the supplemental data online.

Effects of inflection mechanisms on the performance gap
The identification of three common inflection mechanisms helps to shift the analysis of construction process and the performance gap from a list of discrete issues to an analysis of processes and unintended consequences. As the above examples illustrate, a shift in the unit of analysis is often the result of a new policy document or externally set directive. The contribution of this paper is to draw attention to the way in which the choice of units in client briefs and planning requirements serve either to obscure or to keep energy generation on the agenda. In contrast, the imposition of new parameters is often more internally driven. In the three cases examined, it involved an appeal by one or more stakeholders to externally established rules and types of artefacts and was driven by particular interests. In SCOT terms, institutional artefacts were used to carve out a space for the imposition of one technological frame over another in negotiations around a particular problem/solution. While the introduction of financial or aesthetic considerations is generally explicit and even strategic, the knockon effects of these moves for the performance gap were unintended. Finally, the primacy of conventions highlights the pervasive role of 'business as usual' in the adoption of new technologies. Whilst the effect of taken-for-granted, dominant practice has begun to be remarked and theorized in the literature on renewable technology (Fedoruk et al., 2015;Lees & Sexton, 2013) and is at the centre of analyses of user behaviour (Gram-Hassen, 2010;Shove, Pantzar, & Watson, 2012), it is relatively neglected in the growing managerialist literature on the performance gap. A key contribution of this paper lies in the detailed documentation of the unintended consequences of schedules, work packages, cost analyses and even reliance on established relationships.

Conclusions
The exploration of the three cases focused on the gap between early expectations for BIPV energy generation and generation as designed at the point of handover. It investigated the particular (extended) moments in the production of the performance gap over which construction professionals have control (and for which they are responsible). Given the importance and promise of renewable technologies, these moments are important. Theoretically, the paper contributes to the development of SCOT by expanding the range of closure mechanisms identified from those that depend on negotiation and consensus to more indirect inflection mechanisms. These inflection mechanisms highlight the role of broader institutional arrangements on everyday decision-making and their consequences for the incorporation of new technologies whose systems cut across established conventions.
The report by the Zero Carbon Hub (2014) identified a number of discrete construction-related factors that contribute to the gap between building-as-designed and as-built. The present research builds on that systematic analysis by exploring how these different factors came together in three commercial projects. In doing so, it documents the consequences of a large number of seemingly small, independent (non-)decisions about things which ostensibly have nothing to do with energy generation on the performance of BIPV/buildings. More generally, it identifies some of the overlooked challenges involved in keeping the design of the BIPV system and energy generation on the agenda and the role of institutional artefacts such as work packages, schedules and client requirements in either obscuring or maintaining that visibility Theoretically, the use of SCOT, and more specifically an analysis of problem/solution chains and closure mechanisms at play, provides a basis on which to expand the types of closure mechanisms involved in the stabilization of new technologies and their associated networks. In addition to the well-studied cognitive closure mechanisms generally discussed in SCOT research, the paper introduces three types of inflection mechanisms. The analysis points to the way in which institutional artefacts shift the unit of analysis, introduce new parameters and introduce organizational conventions in ways which compromise the energy performance of the BIPV system as initially anticipated.
Conversations with colleagues and professionals suggest that the findings extend to the introduction of any new integrated technology. As scholars and policymakers are fond of saying, construction is a very complex, highly fragmented sector (Gann, 1996;Fernie, Green, & Weller, 2003;Reichstein, Salter, & Gann, 2005). The claim is generally followed by a list of problems linked to fragmentation and a call for integration.
The contribution of this paper is to explore in detail what that integration involves at the project level. Instead of looking for who can best play the essential integrator role, the present research asked what gets in the way of the best laid plans (for low-energy buildings). The main finding concerns the role of dominant ways of working and, more specifically, seemingly unrelated institutional artefacts, which privilege certain criteria over others, introduce units of analysis, and contribute to the location of expertise and the sequencing of decisions. These often have far reaching, but often unintended, consequences for the energy performance of renewable technologies and the building as a whole.
Practically, the detailed analysis of the ways in which these different considerations enter into the everyday work of developing a building suggests a list of issues that policy-makers, clients, construction professionals and promoters of BIPV, integrated technology and innovation will want to take into account. These include: a systematic reflection on the fit between the system requirements of the new element and conventional divisions of labourbe it work packages or schedules; explicit reflection on the fit and consequences of different metrics and parameters; and an awareness of the unintended consequences of contractual divisions of responsibility for the location of expertise. One of the main responses to the growing recognition of the role of construction professionals in the performance gap has been to call for someone, be it the project manager or an integrator, to take ownership of energy generation and keep it on the agenda. Without weighing in on whether this needs to be one person or a more distributed responsibility, this paper contributes to that argument by drawing attention to the myriad of often apparently disconnected micro-level processes and decisions that need to be taken into account to render that role effective.