How Do We Learn about Drivers for Industrial Energy Efficiency—Current State of Knowledge

Abstract

Drivers for industrial energy efficiency are factors that promote the sustained adoption of energy-efficient measures and practices. Leveraging drivers to overcome barriers and encourage action which improves industrial energy efficiency can contribute to closing the energy efficiency gap. In fossil-fuel-based systems, this will also contribute to greenhouse gas abatement. A systematic literature review was used to investigate how knowledge about drivers is generated and whether prevalent drivers can be mapped to existing taxonomies. The systematic literature review confirmed that surveys and/or interviews with managers from countries who are members of the Organisation for Economic Cooperation and Development (OECD) are the most common way to gather data on drivers for industrial energy efficiency. This means the extant knowledge on drivers may be incomplete because contributions from some stakeholders, industry types and company sizes may be missing. The review also found economic drivers are the most prevalent and that not all the drivers identified during the study can be mapped to contemporary driver taxonomies. Having an agreed-upon comprehensive taxonomy facilitates empirical research and comparison of studies. Further research into the views of frontline workers and the creation of a comprehensive driver taxonomy is recommended.

1. Introduction

Implementing energy efficiency measures is a cost-effective way for industries to save money and reduce carbon emissions [1] in systems reliant on fossil fuels. These same energy efficiency measures can also improve energy security [2] and provide a competitive advantage to the industry [3]. Additionally, energy efficiency is central to the International Energy Agency’s World Energy Outlook scenarios for limiting greenhouse gas emissions [2]. However, the industry fails to implement more than half of the cost-effective energy efficiency measures available [4]. Government-run energy efficiency programs in Australia [5] and Sweden [6] also found that approximately 50% of the projects identified were not implemented. The difference between the actual implemented measures and the identified cost-effective measures is commonly called the energy efficiency gap [7]. Barriers are commonly cited as the cause of the energy efficiency gap [8]. The installation of new technology and government policy supporting new technology are the most common methods adopted to overcome barriers and close the gap [9]. Involvement in energy networks [10] and utilising energy management systems [11] are also gaining prominence as methods to minimise barriers. Leveraging drivers is another way to overcome barriers and promote energy management to help close the energy efficiency gap [12].

Drivers, which are also described in the literature as motivators [13] or promoting factors [14], can be defined as “factors promoted by one or more stakeholders, stimulating the sustainable adoption of energy-efficient technologies, practices and services, influencing a portion of the organization and a part of the decision-making process in order to tackle existing barriers” [15] (p. 204). However other definitions, which are not dependent on barriers have been postulated [16]. For example, Ren [17] adopt a broad view, defining drivers as “factors that positively affect a firm’s intentions for innovation and therefore assist innovation activities” (p. 291). Cagno, Worrell [18] propose “factors facilitating the adoption of both energy-efficient technologies and practices” (p. 277) as a potential definition. Interest in drivers for industrial energy efficiency is relatively recent [16], so definitions may continue to evolve. The shared tenet, however, appears to be that drivers are a way to promote action. Accordingly, understanding drivers helps support the creation of good energy policy [19], makes energy management programs more successful [20] and assists with the creation of good energy culture [21].

Drivers can originate from sources internal or external to the company [22]. For example, “Efficiency due to legal restrictions” constitutes an external driver while “Cost reduction from lower energy use” is an internal driver [15] (p. 208). Cost-saving is a prominent driver in empirical studies for companies of differing sizes and industry types [15]. However, long payback periods and the need for high capital investment diminish the overall impact of cost savings as a driver [23]. The Solnordal and Foss [16] literature review investigating drivers for energy efficiency found management (which included awareness, long-term energy strategy and top management support) was the most common driver, with reducing operating costs and competence the second and third most prevalent drivers, respectively. Lawrence, Nehler [24] investigated drivers specifically for energy management in the Swedish pulp and paper industry. They found cost reduction to be the most important driver but access to internal competence and voluntary agreements were regarded “important” according to the Likert scale used and make up the top three. Drivers specific to adopting energy management programs at energy-intensive sites in Sweden have also been investigated. The study found awareness and alignment with core business were the primary drivers [25]. The varied nature of research conducted and the frameworks used suggest this field is still evaluating the effectiveness of the different approaches.

Contemporary work on drivers is derived from unrelated empirical studies without a unifying theoretical basis [15]. Drivers found during these studies are often organised into taxonomies, but there is currently no agreed-upon “best” taxonomy. Additionally, driver categories vary between taxonomies. For example, Solnordal and Foss [16] determined the main categories should be economic, organisational, market forces and policy instruments. Trianni, Cagno [15] propose economics, informative, regulatory and vocational training as overarching driver categories. For a taxonomy to support policymakers and industry, it needs to be able to accurately represent results from empirical research [18]. Buettner, Bottner [26] also found when investigating barriers, that using a single taxonomy facilitates comparisons between studies and empirical research. It follows, therefore, that an agreed-upon taxonomy for drivers would be advantageous.

Research on drivers uses surveys and/or interviews with industry participants as the major data source [15,27]. This means the drivers identified are subject to participant perception. Thus, to gain a complete understanding of the drivers, research needs to include all stakeholders. Trianni, Cagno [15] identified external stakeholders as governmental bodies, technology suppliers, technology manufacturers, installers, energy suppliers, energy service companies (ESCOs), financial institutions, industrial associations and groupings, clients, competitors and partners. Internal stakeholders for drivers are not well-defined in the literature. Yet, for successful energy management, input from multiple levels of an organisation is required [28]. This means, analysing data on who participates in the surveys and interviews may result in a more complete understanding of the drivers identified.

Contemporary research is also not clear on how often surveys and/or interviews are used to generate data on drivers, who participates in the studies and what other data sources are used. Knowledge of the data source used provides insights into the nature and applicability of the drivers. Knowing how the data were generated is also advantageous when comparing studies [26]. Drivers may be different for each organisational level because top management may be unfamiliar with day-to-day decisions made about energy efficiency [29]. As a result, information on the survey and/or interview participants provides insight into the completeness of the drivers found. For example, drivers may be different at each organisational level. Further investigation of the data sources used and the people involved may result in a more complete understanding of the drivers present in an industry and how they can be leveraged to close the energy efficiency gap.

Using a systematic literature review, this paper aims to answer two questions, namely “How do we learn about drivers to industrial energy efficiency?” and “Can contemporary drivers be authentically represented using existing taxonomies?”. A number of other systematic literature reviews investigate drivers. For example, Lawrence, Karlsson [30] review energy management in the pulp and paper industry. They identify barriers, drivers, research methods, publication years and success factors. The Solnordal and Foss [16] systematic review organises drivers found in literature into a taxonomy as well as commenting on research methods and publication timing. Likewise, May, Stahl [31] use their systematic review to organise the barriers and drivers into a framework, noting the publication years and prevalent journals. Trianni, Cagno [15] also conducted a significant literature review which resulted in a new definition and taxonomy for drivers. However, none of these reviews investigate in detail how knowledge of drivers is obtained nor map the drivers identified in the literature to existing taxonomies. Thus, the main sources of novelty in this work stem from generating quantitative data on how knowledge of drivers is generated. These include data on what information sources are used and who is involved. Mapping the drivers found in the literature to contemporary driver taxonomies is another source of novelty. A similar review on barriers completed by Smith, Wilson [32] investigated how new knowledge is created on barriers and mapped the barriers to a single taxonomy.

2. Method

2.1. Literature Selection

A systematic literature review was employed to answer the research questions. Systematic literature review is a transparent way to find, analyse and create meaning from all information pertaining to a particular research question [33]. The use of a documented process to source and choose literature for the review helps reduce researcher bias [34]. This review was guided by the PRISMA framework [35]. The PRISMA framework describes how to conduct and report literature reviews in a systematic manner [35]. The quality of the papers was maintained by searching academic databases for conference papers and academic journal articles.

Articles published between 2013 and 2019 were gathered in January 2021 and those published in 2020 were gathered in March 2021. Literature searches were conducted using Scopus, Web of Science and ABI Inform databases. Three different databases were searched to elicit information on the technical and economic dimensions of energy efficiency. Broad search terms were used to reduce the possibility of papers being excluded as a result of the many ways in which drivers, industrial energy management and industry are described. The base search string for finding relevant literature was (“energy efficiency” or “energy management”) and (indust* OR utilit* OR manfactur*) and (driv* OR motivate*). In order to focus the search, papers containing the words household, storage, transport*, shipping, grid, storage, solar, wind, geothermal, bio*, nuclear, certificate, distribution and/or construction were excluded at the search stage. Full search strings are contained in Appendix A (

Table A1. Full search strings for each database. The asterisk (*) represents a wilde card in the search string.

Database	Search Type	Full Search String
ABI Inform	Full text	((("energy efficiency" OR "energy management") AND (driv* OR motivat*) AND (indust* OR utilit* OR manufactur*) NOT (household OR renewable* OR transport OR shipping OR grid OR distribut* OR solar OR bio* OR nuclear OR wind OR geothermal OR certificate OR construction)) AND peer(yes)) AND yr(2013-2019) and stype.exact("Conference Papers & Proceedings" OR "Reports" OR "Scholarly Journals" OR "Dissertations & Theses")
Scopus	Title, Key Word, Abstract	(TITLE-ABS-KEY ("energy efficiency") OR TITLE-ABS-KEY ("energy management") AND TITLE-ABS-KEY (indust* OR utilit* OR manufactur*) AND TITLE-ABS-KEY (driv* OR motivat*) AND NOT TITLE-ABS-KEY (household) AND NOT TITLE-ABS-KEY (storage) AND NOT TITLE-ABS-KEY (transport*) AND NOT TITLE-ABS-KEY (shipping) AND NOT TITLE-ABS-KEY (grid) AND NOT TITLE-ABS-KEY (renewable) AND NOT TITLE-ABS-KEY (solar) AND NOT TITLE-ABS-KEY (wind) AND NOT TITLE-ABS-KEY (geothermal) AND NOT TITLE-ABS-KEY (bio*) AND NOT TITLE-ABS-KEY (nuclear) AND NOT TITLE-ABS-KEY (certificate) AND NOT TITLE-ABS-KEY (distribution) AND NOT TITLE-ABS-KEY (construction)) AND PUBYEAR > 2006 AND (EXCLUDE (SUBJAREA, "COMP") OR EXCLUDE (SUBJAREA, "MATE") OR EXCLUDE (SUBJAREA, "PHYS") OR EXCLUDE (SUBJAREA, "CHEM") OR EXCLUDE (SUBJAREA, "MATH") OR EXCLUDE (SUBJAREA, "EART") OR EXCLUDE (SUBJAREA, "AGRI") OR EXCLUDE (SUBJAREA, "BIOC") OR EXCLUDE (SUBJAREA, "MEDI") OR EXCLUDE (SUBJAREA, "NURS") OR EXCLUDE (SUBJAREA, "IMMU") OR EXCLUDE (SUBJAREA, "VETE") OR EXCLUDE (SUBJAREA, "NEUR") OR EXCLUDE (SUBJAREA, "DENT")) AND (EXCLUDE (DOCTYPE, "ch") OR EXCLUDE (DOCTYPE, "bk") OR EXCLUDE (DOCTYPE, "sh") OR EXCLUDE (DOCTYPE, "no") OR EXCLUDE (DOCTYPE, "ab") OR EXCLUDE (DOCTYPE, "bz") OR EXCLUDE (DOCTYPE, "ed") OR EXCLUDE (DOCTYPE, "er") OR EXCLUDE (DOCTYPE, "le")) AND (LIMIT-TO (PUBYEAR, 2020))
Web of Science	Topic	TS =("energy efficiency" OR "energy manage*") AND TS=(indust* OR utilit* OR manufactur*) AND TS=(driv* OR motivat*) NOT TS=(household) NOT TS=(storage) NOT TS=(tranport*) NOT TS=(shipping) NOT TS=(grid) NOT TS=(renewable) NOT TS=(solar) NOT TS=(wind) NOT TS=(geothermal) NOT TS=(bio*) NOT TS=(nuclear) NOT TS=(certificate) NOT TS=(distribut*) NOT TS=(construction) Indexes=SCI-EXPANDED, SSCI, A&HCI, CPCI-S, CPCI-SSH, BKCI-S, BKCI-SSH, ESCI, CCR-EXPANDED, IC Timespan=2020

). The search was constrained to articles published between 2013 and 2020 because only a small amount of literature was published on drivers prior to 2013 [16].

To progress to the full-text review stage, papers were assessed against some general and some content-related criteria. To meet the general criteria for inclusion, papers needed to have been published between 2013 and 2020 and be written in English. No restrictions were placed on the research methodology. Titles and abstracts of papers that met the general criteria were reviewed against content criteria. To meet the content criteria, and progress to the full-text review stage, papers needed to be focused on energy use at industrial or manufacturing sites and provide new insights on drivers. Papers were not progressed to the full-text review if they were based on:

• A single audit or calculation without providing new insights on drivers,
• Shipping or scheduling,
• Renewable energy sources,
• Electric vehicles.

Papers that were selected for the full-text review were assessed again using the content criteria. These criteria are the same as those used by Smith, Wilson [32] in a systematic literature review on barriers. Articles that only summarised existing literature were excluded during the full-text review stage.

The steps in the article selection process can be seen in Figure 1.

2.2. Data Classification

Data extraction for the classification of papers was conducted during the full-text review stage by the lead author. Excel and Nvivo were used for data collection and analysis. For each study, information was collected on the:

• Data source,
• Role description of people contributing the raw data,
• Study scope,
• Company size,
• Energy intensity,
• Country.

Information on data source, role description and study scope was selected as Smith, Wilson [32] found this information relevant in their study exploring how knowledge on barriers to energy efficiency is created. Company size, energy intensity and country were collected as these have been found to influence barriers and to be instructive in other systematic literature reviews (See [16,32]).

Data source refers to the origin of the information used for the research. The designation “survey and/or interview” was used when data were obtained using a survey, an interview or a combination of survey and interview. These collection methods were grouped together as the information produced is based on the individuals’ perception of drivers or motivating factors. “Case study for intervention” was used when researchers form partnerships with individual sites with the aim of implementing a particular process or technology to improve energy performance at the site. The process or technology is selected by the researchers before the site assessment commences. While this type of partnership may include data collection using interviews, plant data and documentation is available to the researcher(s) and may inform the findings. Additionally, having preselected an intervention may mean not all possible drivers were explored. “Case study for analysis” also involves an industry partner but the researcher(s) enters the partnership without a proposed intervention. In some cases, work described as a case study by the author was classified as “survey and/or interview” because interview responses were the only data collected. “Literature review” was used when the article used existing literature to generate new findings on drivers. “Other” was used for data sources that did not fit into the categories described. Examples of data sources in this class include the world bank database, an energy efficiency awards scheme and national databases of energy data.

The role or job description was noted for papers that used surveys and/or interviews, case studies for analysis and case studies for intervention as their data source. After collection, the role descriptions provided were coded as “manager”, “owner”, “not stated”, “mixed”, “person responsible for energy issues”, “engineers” or “frontline worker”. “Manager” was used when the role description contained the word management, whether it was top level or low level, production or finance. “Owner” was used where the article described participants as company owners and “not stated” means the authors did not describe participants’ roles. “Mixed” was used when the descriptions implied data were obtained from participants from more than one level in the organisation. “Person responsible for energy issues” was used in cases where it is a direct quote from the article or the description that implies some responsibility without using the term manager. Participants were designated as “engineers” if authors used the word engineer when describing them in the paper. “Frontline workers” are the maintainers, operators and other decision-makers who control when and how a plant is operated in real-time. This description was used if the authors referenced the shop floor, maintenance, production or operations personnel who were not described as managers. A full list of the descriptions supplied in the articles reviewed and the coding is contained in Appendix B (

Table A2. Systematic literature review data and coding for papers not using survey and/or interview as a data sources.

Reference	Year	Coded Data Source	Company Size	Energy Intensity	Coded Position	Economic Dev.	Scope	Participant Description from Pper
[41]	2017	Case study for analysis	Large	EI	Mixed	OECD	Technology only	Stakeholder, staff at the companies, plant manufacturers and governmental intuitions
[42]	2016	Case study for analysis	SME	Non-EI	Mixed	OECD	Managerial only	Management and employees and production associates and technical staff
[55]	2017	Case study for analysis	Not stated	EI	Not conducted	OECD	Technology only	No interviewing because answer was obvious
[56]	2015	Case study for analysis	Not stated	Non-EI	Frontline workers	OECD	Technology only	Plant and mine operators
[57]	2016	Case study for analysis	SME	Non-EI	Not conducted	OECD	All EEM	Not stated (interviewing was done)
[58]	2020	Case study for analysis	Large	Non-EI	Frontline workers	Non-OECD	All EEM	Maintenance and operational personnel
[23]	2020	Case study for analysis	Large	EI	Managers	OECD	All EEM	Talks to managers but also maintenance manager and technical controller
[59]	2013	Case study for intervention	Large	Non-EI	Not conducted	OECD	All EEM	No interview component
[60]	2019	Case study for intervention	Not stated	EI	Engineers	OECD	Managerial only	People with responsibility for energy—process engineers energy coordinators and manager
[61]	2020	Case study for intervention	Large	EI	Engineers	OECD	All EEM	Technical staff with significant knowledge about the operational and technical aspects of the refinery—multiple sorts of engineers
[62]	2016	Case study for intervention	Large	Non-EI	Managers	OECD	All EEM	Energy managers as well as independent energy audit experts
[43]	2019	Case study for intervention	Large	Non-EI	Mixed	OECD	Managerial only	Mixed energy specials, global leaders and project team members from different departments at the plant.
[63]	2018	Case study for intervention	Large	EI	Not conducted	BRIC	All EEM	No interview component
[49]	2019	Literature review	Not stated	Not stated	N/A	Not stated	All EEM	N/A
[24]	2018	Literature review	Not stated	EI	N/A	OECD	All EEM	N/A
[31]	2016	Literature review	Not stated	Not stated	N/A	Not stated	All EEM	N/A
[64]	2018	Literature review	Not stated	Not stated	N/A	Not stated	Managerial only	N/A
[65]	2018	Literature review	Not stated	Not stated	N/A	Not stated	All EEM	N/A
[16]	2018	Literature review	Not stated	Not stated	N/A	Not stated	All EEM	N/A
[15]	2017	Literature review	Not stated	Not stated	N/A	Not stated	All EEM	N/A
[66]	2013	Other	Mixed	Mixed	N/A	OECD	All EEM	N/A
[67]	2017	Other	Not stated	Mixed	N/A	Non-OECD	Technology only	N/A
[68]	2018	Other	Not stated	Mixed	N/A	OECD and Non-OECD	All EEM	N/A
[69]	2018	Other	Mixed	Mixed	N/A	OECD	All EEM	N/A
[70]	2016	Other	Mixed	Mixed	Managers	OECD	Technology only	Plant managers, manufacturing managers general managers
[71]	2016	Other	Mixed	Mixed	N/A	OECD	All EEM	N/A
[72]	2018	Other	Not stated	Mixed	N/A	BRIC	Not stated	N/A
[73]	2014	Other	Not stated	Not stated	N/A	OECD and Non-OECD	investment only	N/A
[74]	2017	Other	Not stated	Not stated	N/A	OECD and Non-OECD	All EEM	N/A
[75]	2017	Other	Not stated	Mixed	N/A	OECD	All EEM	Not Applicable

).

The scope of the studies presented in the papers reviewed was categorised as “all energy efficiency measures” (All EEM), “investment focus only”, “technical focus only”, “managerial focus only” and “not stated”. The scope was classified as “all energy efficiency measures” if the paper specifically stated all efficiency energy measures, including technologies and practices [19] were investigated, or if it could be deduced from the article. “All energy efficiency measures” was also used when a paper investigated installing a specific technology but addressed the systems and procedures required to operate the technology as well as the drivers to install the new technology. “Technology only” and “investment only” were used when the article said it was focusing on drivers for technology uptake or investment practices, respectively. The scope was designated as “Managerial only” if the paper stated it focussed on organisational behaviour or if the paper stated it was exclusively about energy management or energy management practices. “Not stated” was only used when the scope was not stated and it could not be inferred from the paper.

Size was categorised as per the authors’ description or the number of employees. A site with 250 employees or less was designated as a “small to medium enterprise” and those with more than 250 were described as “large” enterprises in line with the European Commission guideline [36]. If employee numbers were not listed and no description of the industry size was given, company size was categorised as “not stated”. The term “multiple” was used if the results came from different-sized companies.

The categorisation of energy intensity was based on the International Energy Agency (IEA) definition and the industry description provided by the authors. The IEA defines the energy-intensive sub-sectors as “basic metals manufacturing, non-metallic minerals manufacturing, paper and printing and chemicals and chemical products manufacturing” ([37], p. 70). Energy costs, as a percentage of turnover or production costs, can be used to determine energy intensity but this method is subject to time and regional fluctuations in energy prices. Energy-intensive (EI) was used if the participant’s company was a member of one of the sub-sectors described by the IEA and Non Energy Intensive (Non-EI) if it was not. The study was categorised as “Energy-intensive and Non-energy-intensive” if the study combined the results from energy-intensive and non-energy-intensive industries. The term not stated was used if the energy intensity could not be determined.

Economic development level was assigned using the country provided by the authors and the detail provided in the World Energy Outlook 2020 [38]. A country was labelled as OECD (Organisation for Economic Cooperation and Development) if it was included in the list of OECD countries in World Energy Outlook 2020 and as Non-OECD if it was not. OECD and Non-OECD were used where data from both sorts of countries were used in the study and not stated was used if the authors did not specify a country under study. BRIC (Brazil, Russia, India, China) was used if the study data were from Brazil, Russia, India or China.

2.3. Driver Mapping

The three most prevalent or important drivers identified in papers that used surveys and/or interviews were recorded by the lead author. Articles using survey and/or interview were investigated further because survey and/or interview was the most common source of data. Additionally, comparing studies with different data collection methods may not produce valid comparisons. If less than three drivers were presented, then all the drivers were recorded. If the paper presented more than three unprioritised drivers then none of the drivers presented were included in the data mapping exercise to stop less important or prevalent drivers from being included in the analysis. The extracted drivers were then inspected for common keywords and/or phrases. Excel was used to record, map and calculate percentages during driver mapping.

After inspecting the extracted text on drivers for repeated words or phrases, they were mapped to the taxonomies created by Solnordal and Foss [16], Trianni, Cagno [15] and Lawrence, Nehler [24]. These taxonomies were the three most recent driver taxonomies found during the systematic literature review. Drivers were mapped to more than one taxonomy because while all were created with the intent of capturing drivers in a meaningful way there is currently no agreed-upon “best” taxonomy and they contain different driver categories. Drivers were usually mapped on a category level because in a similar study on barriers, by Smith, Wilson [32] found data needed to be aggregated to a category level to allow insights to be drawn. The mapping was guided by the operational statements or sub-categories contained in the taxonomies but a broad interpretation of the drivers was taken. For example, the driver “external pressures” [39] (p. 279) was not present in any of the taxonomies but is well-represented by the economic category in Lawrence, Nehler [24], which contains operational statements including “demand from customer”, “pressure from environmental organisations” and “international pressure” [24] (p. 72). Thus, it is a conceptual rather than semantic mapping. A “not mapped” option was allowed to record drivers which did not fit into a driver category.

Solnordal and Foss [16] and Trianni, Cagno [15] created taxonomies for energy efficiency based on reviews of literature using empirical studies. Lawrence, Nehler [24] take a slightly different approach presenting a taxonomy for energy management practices, based on the authors’ experience and previous research. Energy management practices are largely concerned with the processes, technical or managerial, used to support energy management [40].

Once the initial mapping exercise was complete, the prevalence of drivers for different company sizes, energy intensity and economic development were compared to investigate the impact of these contextual factors. The impact of contextual factors was considered to be different if the sub-category contained more than 10% for one classification and the larger percentage was more than double the smaller one.

3. Results

3.1. How Do We Learn about Drivers

Data were collected from nine conference papers and 52 journal articles. All conference papers were peer-reviewed. Results from the data collection can be seen in

Table 1. Summary of results from systematic literature review on how academia learns about drivers showing the number of papers for each classification.

Data Source (n = 61)		Role Description for Survey and Interview (n = 31)		Role Description for All Case Studies (n = 9)	Study Scope (n = 61)		Company Size (n = 61)		Energy-Intensity (n = 61)		Economic Dev. (n = 61).
Survey or/and Interview	31	Manager	13	2	All energy efficiency measures	45	Small to medium enterprise	10	Energy-intensive	18	OECD	36
Case study for intervention	6	Frontline workers	0	2	Managerial only	4	Large enterprise	15	Non energy-intensive	13	Non-OECD	11
Case study for analysis	7	Person responsible for energy issues	7	0	Technology only	6	Multiple	16	Energy-intensive and Non energy-intensive	18	OECD and non-OECD	3
Literature Review	7	Mixed	7	3	Investment only	4	Not stated	20	Not stated	12	BRIC	7
Other	10	Owner	1	0	Not stated	2					Not stated	11
		Engineers	0	2
		Not stated	3	0

. The results show that survey and/or interview (31 of 61) is the most common way to generate data on drivers for industrial energy efficiency. Three of the articles using survey and/or interview data were conference papers. The majority of studies investigate drivers for all energy efficiency measures (45 of 61) in energy-intensive companies (18 of 61) of unknown size (20 of 61) in OECD countries (36 of 61).

Additionally, more than 10% of studies did not include data on the size (20 of 61), energy intensity (12 of 61) or country (7 of 61) under study. Not including the size or energy intensity of the company means valuable context may be lost and makes the applicability of findings difficult to ascertain.

The bars in Figure 2 are the percent of studies that fall into a particular classification for the whole sample or the subset of studies that use survey and/or interview as a data source. Figure 2 shows that the subset of data for survey and/or interview gathers results from sites of different sizes (39%), which are energy-intensive (35%) and study all energy measures (84%) in OECD countries (61%).

Figure 3 shows that frontline workers are not explicitly represented (0) in the survey and/or interview data set and that managers are the most common group consulted (13 of 31 articles). Additionally, in seven of the papers, participants were a “person responsible for energy issues”, which suggests some degree of supervisory or managerial responsibility. The focus on managers may be justified if the studies were predominantly investigating only technology or investment. However, most studies investigate all energy efficiency measures (84%).

After noting the absence of frontline workers’ views in research based on survey and/or interview, an investigation into their inclusion in case study research and representation in the “mixed” classification was conducted. Nine of the 13 case studies included in this review have an interview or workshop component. Two of these eight studies list frontline workers as the participants in interviews or workshops. Additionally, case studies that include workshops or interviews appear to include participants from a greater number of organisational levels than are detectable in studies reliant on survey and/or interview. A summary of the roles of case study participants is shown in Table 1.

Table 2. Participant descriptions for studies where the role description was coded as mixed. Entries where frontline workers are included have been highlighted.

Author	Data Source	Role Description	Participant
[41]	Case study for analysis	Mixed	Stakeholders—for example, staff at the companies, plant manufacturers and governmental institutions
[42]	Case study for analysis	Mixed	Management and employees and production associates and technical staff
[43]	Case study for intervention	Mixed	Mixed energy specialists, global leaders and project team members from different departments at the plant
[44]	Survey and/or interview	Mixed	Staff from production and quality departments
[45]	Survey and/or interview	Mixed	Owner, energy management, accounting production workers, maintenance engineering, human resources, marketing, trainees
[46]	Survey and/or interview	Mixed	Experienced workers in the field from mining companies vendors engineering companies and academia (field refers to field of study)
[47]	Survey and/or interview	Mixed	Managers are the typical respondent (comment is made on shop floor perspective)
[14]	Survey and/or interview	Mixed	One from top management and one from the energy group
[25]	Survey and/or interview	Mixed	One from top management and one from the energy group
[48]	Survey and/or interview	Mixed	Energy management personnel

shows frontline or shop floor workers are apparent in three (highlighted) of the 10 studies where the role description was described as mixed. Some descriptions, such as “energy management personnel” or “energy group” lack sufficient detail to determine whether the groups contain frontline workers.

Some studies where frontline workers were included, did not highlight their views or particular motivations. Concepts particular to frontline workers can be found in Cosgrove, O’Neill [42] and Pusnik, Al-Mansour [47] but not in Konig, Lobbe [45]. Cosgrove, O’Neill [42] recognised frontline workers did not have the authority to enact a proposed energy-saving measure (turning off idle machines). Pusnik, Al-Mansour [47] concluded that “recognition of effort” is a driver for shop floor employees.

3.2. Can Contemporary Drivers Be Authentically Represented Using Existing Taxonomies?

To answer this question, drivers need to be identified and then mapped to driver taxonomies from the literature. A total of 88 drivers were identified in the 31 papers that used survey and/or interview as a data source. A full list of the drivers can be seen in Appendix C (

Table A3. Systematic literature review drivers for papers using survey and interview as a data source.

Ref.	Year	Coded Data Source	Size	Energy Intensity	Coded Position	Economic Dev.	Scope	Participant	3 Most Prominent or Important Drivers
[44]	2020	Survey and/or interview	SME	EI	Mixed	Non-OECD	All EEM	Staff	Cost reduction from lowered energy use	Improved working conditions	Threat of rising energy prices
[20]	2013	Survey and/or interview	Not stated	Mixed	Managers	Non-OECD	All EEM	Middle managers or technical directors and engineering staff who dealt directly with energy issues at the firm	Cost reduction lowered energy use	Increasing energy prices	Requirements by government
[76]	2014	Survey and/or interview	Large	EI	Managers	OECD	All EEM	Mixed managers	Commitment from top management/energy management	Cost reduction lowered energy use	Long-term energy strategy
[39]	2013	Survey and/or interview	SME	Mixed	Person responsible for energy issues	OECD	All EEM	Person in charge of energy issues	Allowances or public finances	External pressures	Long term benefits
[77]	2015	Survey and/or interview	SME	Mixed	Person responsible for energy issues	OECD	All EEM	People know legible and responsible for energy issues.	Cost reduction lowered energy use	Long-term energy strategy	Clarity of information
[12]	2017	Survey and/or interview	SME	Mixed	Person responsible for energy issues	OECD	All EEM	Person in charge of energy-efficient investment	Information about real costs	Clarity and trustworthiness of information	Public investment subsidies
[78]	2019	Survey and/or interview	Mixed	Mixed	Not stated	OECD	investment only	Not stated	Cost reductions resulting from lower energy use	Enhancing the positive image and reputation	Enhanced competitiveness
[79]	2019	Survey and/or interview	Mixed	EI	Managers	OECD	All EEM	Energy managers or CEO or production manager but they could pass the survey on	People with real ambition	Full support from top management	Knowledge about daily operations
[80]	2018	Survey and/or interview	Large	EI	Managers	Non-OECD	All EEM	Production manager or plant manager	Cost reductions from lowered energy use	Long term energy strategy	Threat of rising energy prices
[50]	2019	Survey and/or interview	SME	Non-EI	Person responsible for energy issues	Non-OECD	All EEM	Person mainly responsible for energy-related issues	Owners requirement	Expense minimisation due to lower energy use	Risk of high energy prices in the future
[81]	2019	Survey and/or interview	Large	Non-EI	Managers	Non-OECD	All EEM	Concerned plant manager, chief engineer	Energy management Scheme	Risk of higher energy prices in the future	Assistance from energy professionals
[82]	2020	Survey and/or interview	Large	EI	Managers	Non-OECD	All EEM	Plant managers, manufacturing managers general managers	Risk of high energy prices in the future	Highly motivated employee	High demand from consumer and NGO
[83]	2020	Survey and/or interview	Large	Not stated	Managers	Non-OECD	Not stated	Lower, middle and upper level managers	Environmental strategies	Entrepreneurial innovation	Entrepreneurial orientation
[45]	2020	Survey and/or interview	SME	Mixed	Mixed	OECD	All EEM	Owner, energy management, Accounting production workers, maintenance engineering, human resources, marketing, trainee	Embedding of energy efficiency in corporate strategy	The use of a broad spectrum of different practices	The empowerment and involvement of employees
[24]	2019	Survey and/or interview	Mixed	EI	Managers	OECD	All EEM	Energy manager or similar	Cost reduction from lower energy use	Access to internal competence with knowledge of the processes	Voluntary agreements with tax exemptions
[84]	2015	Survey and/or interview	Not stated	EI	Managers	Non-OECD	All EEM	Senior executives and managers	Cost savings from lowered energy use	Demand from owner	Energy tax
[85]	2017	Survey and/or interview	Mixed	Not stated	Managers	OECD	All EEM	Senior managers who were in charge of implementing ISO50001	Improve energy efficiency	Enhance employee awareness	The rise of energy prices/image improvement
[86]	2013	Survey and/or interview	Large	Non-EI	Not stated	OECD	All EEM	Not stated	Reductions of costs	Reduction of environmental impact	Image improvement due to enhanced reputation
[46]	2015	Survey and/or interview	Not stated	EI	Mixed	OECD	All EEM	Experienced workers in the field from mining companies vendors engineering companies and academia	Economic	Community expectation
[87]	2018	Survey and/or interview	Large	Non-EI	Managers	OECD	All EEM	Industrial energy managers and audit experts	Commitment from top management	People with real ambition	Cost reductions from lowered energy use
[88]	2014	Survey and/or interview	Mixed	Mixed	Not stated	OECD	All EEM	Not stated	Cost reductions resulting from lower energy use	Increase in energy prices	Fiscal arrangement
[47]	2016	Survey and/or interview	Mixed	Mixed	Mixed	OECD	All EEM	Managers are the typical respondent but some shop floor information is represented	Recognition for personal contributions	Motivation
[89]	2013	Survey and/or interview	Mixed	Not stated	Managers	Not stated	investment only	Managers (sustainability, energy, plant, presidents and vice president)	Competitive performance	Proactive risk containment	Top down direction
[14]	2018	Survey and/or interview	Mixed	EI	Mixed	OECD	Technology only	One from top management and one from the energy group	Cost reductions from lowered energy use	People with real ambition	Long term energy strategy
[25]	2017	Survey and/or interview	Mixed	Mixed	Mixed	OECD	All EEM	One from top management and one from the energy group	Drivers cannot be extracted.
[90]	2015	Survey and/or interview	Mixed	Non-EI	Person responsible for energy issues	Non-OECD	All EEM	Key person who is charge of energy efficiency of each company	Potential to reduce energy costs
[91]	2020	Survey and/or interview	Mixed	EI	Managers	BRIC	All EEM	Middle managers	Environment	Competitiveness	Economics
[19]	2013	Survey and/or interview	Mixed	EI	Person responsible for energy issues	OECD	All EEM	People responsible for energy at their site (interview)	Cost reductions resulting from lowered energy use	Threat of rising energy prices	Commitment from top management
[22]	2016	Survey and/or interview	SME	Mixed	Person responsible for energy issues	OECD	investment only	People responsible for energy efficiency investments	Economic external	Economic internal	Regulatory external
[92]	2014	Survey and/or interview	SME	Non-EI	Owner	OECD	All EEM	Owners/Managers	Economics internal	Economics external	Regulatory external
[48]	2018	Survey and/or interview	Not stated	Not stated	Mixed	BRIC	All EEM	Energy management personnel	Economic	Informative	Regulatory

). The repeated words or phrases used in papers can be seen in

Table 3. Prevalence of keywords and/or phrases identified in drivers extracted from articles using survey and/or interview as a data source.

Repeated Words and Phrases	Number of Papers	% of Total Drivers
Cost-saving/cost reductions	14	16
Rising/higher energy prices	7	8
Long term energy strategy	5	6
Company image or reputation	4	5
Information	3	3
Environment	3	3
Management commitment	3	3

.

Results for mapping drivers to published taxonomies are shown in

Table 4. Mapping of drivers from papers using survey and/or interview to contemporary literature taxonomies.

Mapping to Lawrence, Nehler [24]		Mapping to Solnordal and Foss [16]		Mapping to Trianni, Cagno [15]
Driver Category	% of Drivers	Driver Category	% of Drivers	Driver Category	% of Drivers
Economic	60	Economic	41	Economic (internal and external)	32
Organisational	16	Organisational	25	Regulatory (internal)	17
		Policy Instruments	8	Regulatory (external)	19
Knowledge-based	14
		Market	18
				Vocational Training (internal and external)	0
				Informative (internal and external)	14
Not mapped	10 (6 papers)		8 (6 papers)		18 (13 papers)

. Driver categories that represent similar concepts are shown in the same row.

Mapping the results to the three different taxonomies also shows that economic drivers are the most prevalent. Discerning the next most common driver is difficult due to the different categories, but two of the three taxonomies show that organisation-related drivers as the next most prevalent. It can also be seen for two of the taxonomies that more than 10% of the drivers identified could not be mapped. Solnordal and Foss [16] had the lowest number of drivers which could not be mapped. Drivers were also mapped at a sub-category level to the Solnordal and Foss [16] taxonomy to investigate the impact of size, energy intensity and economic development. See

Table 5. Sub-category level mapping with results divided by size, energy intensity and economic development. Categories which show a significant difference have been highlighted. Driver categories and sub-categories are from [16].

		Size		Energy Intensity		Economic Development			Tot. l
Driver Category	Driver Sub-Category	Small to Medium Enterprises (n = 8)	Large (n = 7)	Energy- Intensive (n = 11)	Non Energy- Intensive (n = 6)	OECD (n = 19)	Non-OECD (n = 9)	BRIC (n = 2)
		%	%	%	%	%	%	%	%
Economic	Technology	0	0	0	0	0	0	0	0
	Operating Costs	33	33	41	39	33	44	33	35
	Finance	17	0	0	6	9	0	0	6
Organisational	Organisational Structure	0	0	0	0	0	0	0	0
	Management	13	29	31	6	17	19	0	17
	Competence	0	5	10	0	8	4	17	8
Market	Market Forces	4	19	6	22	13	7	0	10
	Ownership	4	0	3	6	0	7	0	2
	Network and Information	13	5	0	5	6	4	17	6
Policy Instruments	Policy and Regulation	8	0	6	11	6	11	17	8
Not mapped		8	9	3	5	8	4	16	8
	Total	100	100	100	100	100	100	100	100

.

Approximately half of the papers based on survey and/or interview can be categorised by size, energy intensity or economic development. Four sub-categories of drivers appear to be different (highlighted in grey) when the results are segregated by size. When segregating data by energy intensity, two driver sub-categories are different. One sub-category appears to be different when OECD and Non-OECD countries are compared. BRIC countries were not included in the comparisons because the sample contains two articles.

4. Discussion

4.1. Limitations

Limitations to the systematic literature review include search types, constraining the timeframe, reviewing only articles in English, not pursuing grey literature, sample size, having only one researcher map the drivers and researcher bias. While the search type may mean some relevant literature was missed the impact of this was minimised by using broad search terms and more than one database. Constraining the timeframe to 2013–2020 should limit the impact of drivers changing over time while not unduly screening out relevant contributions. Solnordal and Foss [16] determined there was limited literature on drivers prior to 2013. The review did not include contributions from grey literature. This means the contributions published by well-regarded institutions, such as the International Energy Agency and others, who publish outside the traditional publishing methods are not included. However, limiting the searches to academic publications is consistent with similar published reviews (e.g., [16,49]) and means the work is not produced by an institution with a defined remit. The final sample size is also consistent with reviews of this type (e.g., [16,49]). Researcher bias and the inclusion of articles that do not meet the criteria being included in the review were minimised by using a transparent process.

4.2. How Do We Learn about Drivers?

The majority of information on drivers for industrial energy efficiency is generated from energy-intensive sites of unknown size, in OECD countries, using surveys and/or interviews. The prevalence of energy-intensive sites is not unexpected as energy costs are usually a significant part of their operating costs. The lack of data on company size is notable. Additionally, the current focus on the potential savings available in small- to medium-sized enterprises [27] suggests that size is material to the results. As with barriers [32], investigations of drivers were largely conducted on OECD countries. There is, however, some evidence that drivers differ between OECD and non-OECD countries [44] and that non-OECD countries have the potential for improved energy management [50]. Additionally, there have been calls from other authors to do more research on non-OECD countries [16]. Thus, the current view of drivers may be incomplete due to a lack of information about the size of companies being studied and a lack of input from non-OECD countries. Limited inclusion of these groups may also constitute a missed opportunity to find new ways of closing the energy efficiency gap.

The dominance of surveys and/or interviews with managers as a way to generate new insights is consistent with barrier research [32]. The focus on managers means the frontline worker perspective may not be recognised in the literature and so the current view of drivers may be incomplete. Incomplete knowledge of drivers may contribute to the energy efficiency gap because companies are unaware of ways that could motivate frontline workers to work on improving energy performance. For example, Pusnik, Al-Mansour [47] identified recognition of effort as a driver for shop floor workers (frontline workers). Recognition of effort is not a commonly reported driver and could not be mapped to a driver category during the mapping exercise. Frontline workers may provide new or different insights on drivers because previous research has found that frontline operations personnel have been delegated much of the work of controlling energy consumption [51]. Yet, it is often low on their priority list [52] or viewed as part of management’s remit [53]. Whether focusing on the views of managers is deliberate or a function of the data collection method is unclear. Frontline worker participation, although still not common, appears to be more likely in case study research. This may be because researchers are more likely to spend time at the site. This suggests maybe barriers to gathering frontline workers’ views using surveys and/or interviews.

Reviews of empirical research are used to generate and conceptualise a new understanding of drivers including energy management frameworks (e.g., [54]) and taxonomies (e.g., [16]). Frameworks can be used to create new systems, benchmark existing systems or as the basis for further research. Similarly, taxonomies can be used for further empirical investigation and as the basis for policy decisions. Thus, if frontline workers are not included in the initial empirical studies, they may be excluded from further investigation and also from the mechanisms created to support energy management.

4.3. Can Contemporary Drivers Be Authentically Represented Using Existing Taxonomies?

Economics, including internal costs and rising prices, were the most prevalent drivers found in the literature. This is consistent with much published literature [15] but different from Solnordal and Foss [16] whose work found organisational drivers to be the most prevalent.

Economic drivers, possibly because of their prevalence, can be reliably mapped to each of the three taxonomies used. Drivers regarding concern for the environment, as distinct from being motivated by an environmental management system or company image, were the most difficult to map. Two of the three taxonomies did not provide a suitable category. Concerns for the environment were mapped to the organisational category, when using the Lawrence, Nehler [24] taxonomy. This was based on the organisational category containing the operational statement “company’s environmental profile” [24] (p. 72).

Overall, the Trianni, Cagno [15] mapping had the greatest number of unmapped drivers in terms of overall percent and the number of papers with unmapped drivers. The number of papers and overall percentage of drivers were examined when investigating taxonomy useability to limit the influence of individual papers on the apparent usability of the taxonomy.

It is difficult to draw strong conclusions about the impact of context because many studies either did not include contextual information or combined results for sites of different size, energy intensity and economic development. However, it would appear size has a larger impact on drivers than energy intensity or economic development. This is consistent with results from Smith, Wilson [32] on barriers.

5. Conclusions

The aim of this paper was to answer the two research questions, “How do we learn about drivers to industrial energy efficiency and “Can contemporary drivers be authentically represented using existing taxonomies?”. The first research question was addressed using a systematic literature review. The review found new information or learnings on drivers for industrial energy efficiency are predominantly created using surveys and/or interviews with managers. Using surveys and/or interviews as a data source finds perceived drivers, which are subject to the perception of the survey and interview participants. Driver research mainly investigates the perspective of managers at energy-intensive sites, of unknown size, in OECD countries. This means drivers relevant to practitioners other than managers, non-energy-intensive sites or non-OECD countries may be missing or under-represented in the current understanding of drivers. This suggests the current knowledge of drivers is incomplete. Drivers that are not recognised by the company cannot be leveraged by the company to improve energy performance and so may contribute to the energy efficiency gap. The work also shows a number of studies do not include enough detail for the company size, energy intensity, scope or program participants to be identified. This makes the applicability, validity and novelty of the work difficult to determine.

Frontline workers are more likely to be included when case studies are used to learn about drivers. Single or small number case studies may be a more appropriate method to learn about drivers as it enables access to difficult to reach populations who are integral to energy performance. Representing and investigating frontline workers’ views explicitly will allow a better understanding of whether this group is influenced by the same drivers as managers.

The second research question was addressed by mapping the drivers found during the systematic literature review to three contemporary driver taxonomies. The number of drivers that could not be mapped to existing taxonomies and the difficulty in using the taxonomy to identify prevalent drivers suggest that existing driver taxonomies are not always able to authentically represent contemporary drivers. The driver mapping exercise also showed that economic drivers were the most prevalent regardless of analysis method or contextual factors and suggested that company size has a bigger impact than energy intensity or economic development.

This study points to a number of areas for further research about how to address the lack of information about driver research. More research into the drivers in non-OECD countries is warranted as there is potential to improve their energy performance and there are a relatively small number of studies. It is also recommended that future studies include contextual information so more definitive conclusions can be drawn about the influence of context. A complete understanding of the effect of context could allow companies to easily identify likely energy efficiency drivers at their sites. Investigating the influence of time, as more studies become available, is also recommended. Additionally, more research on the drivers for frontline workers is warranted. This could initially be directed at seeing whether the drivers for frontline workers are different from those faced by managers. Finally, the mapping exercise shows more work is required to create a taxonomy that can authentically represent driver research. The new taxonomy should be able to represent frontline workers’ views, concerns about the environment and describe drivers in a way the intent is readily understood by industry and academia, now and into the future.