The steel industry plays a pivotal role in the economic development of emerging nations, serving as a key driver of infrastructure, manufacturing, and employment. In developing countries, steel production has seen rapid expansion, with China, India, and Brazil accounting for over 60% of global output in recent years [1]. This growth is fueled by urbanization, population increases, and industrial policies aimed at self-sufficiency [13]. However, the sector's reliance on energy-intensive processes poses sustainability challenges, particularly in regions with limited access to advanced technologies and renewable energy sources [2]. Recent studies show that while developed economies have transitioned to more efficient EAF routes, developing countries predominantly use BOF methods, which are less efficient but better suited to available raw materials such as iron ore [14]. The OECD Steel Outlook 2025 emphasizes the need for policy interventions to balance economic growth with environmental sustainability in these contexts [6]. Overall, the literature underscores the dual opportunity and challenge: leveraging steel for development while mitigating its environmental footprint [13].

The intricate patterns of energy consumption in steel production are deeply rooted in the fundamental processes that define the industry, particularly ironmaking, steelmaking, and rolling. Ironmaking, often conducted in blast furnaces, relies heavily on chemical reactions to reduce iron ore into molten iron. This step demands intense thermal energy to sustain high temperatures and facilitate the necessary metallurgical transformations. This is compounded in steelmaking, where processes like the BOF or EAF convert raw materials into liquid steel through oxidation or melting, further amplifying energy needs due to the inherent inefficiencies in heat transfer and material handling. Rolling, the final shaping phase, adds another layer by requiring mechanical energy to deform hot steel into usable forms, often involving repeated heating to maintain malleability. At the core of these patterns is a profound dependence on fossil fuels such as coal for both energy and as a reducing agent, alongside electricity for powering furnaces and machinery, which creates a systemic vulnerability to resource availability and environmental impacts [15], [16].

In developing countries, these consumption patterns are exacerbated by structural factors that perpetuate inefficiency. Unlike in advanced economies, where modern technologies enable optimized energy flows, many facilities in emerging markets operate with legacy equipment that lacks advanced controls, resulting in excessive heat loss, suboptimal process integration, and prolonged production cycles. This is particularly evident in the dominance of primary production routes, which start from raw ores and consume vast amounts of energy in extraction and initial processing, compared with secondary routes that recycle scrap and inherently require less input. The reliance on outdated infrastructure not only inflates operational costs but also hinders the adoption of cleaner alternatives, trapping industries in a cycle of high energy intensity driven by limited access to capital for upgrades and a workforce that may lack specialized training in efficient practices [17].

The challenges embedded in these patterns extend beyond technical limitations to encompass economic and regulatory dimensions. Fluctuating energy prices, influenced by global market volatilities and geopolitical tensions, pose a constant threat to profitability, forcing producers to navigate unpredictable costs that can disrupt long-term planning. Supply chain disruptions, whether from raw material shortages or logistical bottlenecks, further complicate energy management, as inconsistent inputs lead to irregular operations and wasted resources. Increasing regulatory pressures for decarbonization add another layer of complexity, as governments in developing regions grapple with balancing industrial growth against international commitments to reduce emissions, often imposing standards that require transformative shifts without adequate support mechanisms. This interplay creates a paradox: while energy demand growth in emerging economies has shown signs of moderation in recent years due to efficiency gains and economic shifts, the absence of proactive interventions risks amplifying future pressures as population-driven demand escalates [18–20].

Moreover, the linkage between energy consumption and CO₂ emissions underscores a broader environmental challenge: heavy reliance on fossil fuels in developing contexts not only contributes to global climate change but also perpetuates local pollution and resource depletion. This dependency fosters a vicious cycle, as efforts to scale production for economic development inadvertently heighten vulnerability to climate-related disruptions. Addressing these entrenched patterns demands targeted innovations that rethink the entire value chain —from integrating renewable energy sources to fostering circular economies that prioritize recycling and waste minimization —while ensuring solutions are adaptable to the socio-economic realities of developing nations [21], [22].

In developing countries, the adoption of energy efficiency technologies in the steel industry is significantly hindered by a complex array of barriers. Economic constraints are among the most pressing challenges, with the high upfront costs of energy-efficient technologies and limited access to financing being key obstacles. This is especially problematic for small and medium-sized enterprises (SMEs), which often lack the financial resources to invest in advanced energy-saving solutions. Additionally, technological barriers further exacerbate the situation. The absence of a sufficiently skilled workforce, combined with inadequate infrastructure for integrating modern technologies, makes the transition to energy efficiency more difficult. The incompatibility of new technologies with existing equipment and systems also poses a significant challenge for many steel producers in developing nations [23], [24].

Regulatory and policy-related issues add another layer of complexity. In many cases, there is a lack of consistent, robust policy frameworks to support green investments, leaving industries without clear incentives to adopt energy-efficient practices. Furthermore, socio-cultural factors contribute to the slow adoption of these technologies. Resistance to change within companies, coupled with low awareness of the long-term economic and environmental benefits of energy efficiency, often delays decision-making. These challenges are compounded by broader geopolitical and supply chain vulnerabilities, especially pronounced in regions such as Sub-Saharan Africa and Southeast Asia, where external factors, such as political instability or fluctuating market conditions, further impede progress toward energy efficiency goals [25].

While these barriers result in significant energy inefficiencies, they also present opportunities for tailored solutions. Overcoming these challenges requires a comprehensive approach that addresses both the technical and non-technical aspects of energy efficiency adoption. Collaboration between governments, international organizations, and the private sector is crucial, as is the need for capacity-building programs to foster the development of local expertise and infrastructure. Only through coordinated efforts can developing countries overcome the barriers to energy efficiency in the steel industry and make meaningful progress toward more sustainable, cost-effective practices.

The literature reveals a broad spectrum of innovative technologies designed to reduce energy consumption in the steel industry, with a particular focus on the unique challenges and opportunities within developing countries. These technologies encompass a variety of approaches, including process optimization, material efficiency, and the integration of alternative energy sources. A central theme in this body of research is the potential for these technologies to not only reduce energy use but also enhance overall sustainability in resource-constrained environments. Among the most promising solutions are waste heat recovery systems, which capture and reuse thermal energy from exhaust gases, thereby significantly improving energy efficiency. Additionally, integrating hydrogen injection into blast furnaces as a substitute for carbon-based reductants offers a pathway to decarbonize steel production processes, aligning with global sustainability goals [26]. Further innovations include the application of advanced process controls powered by artificial intelligence (AI) and the Internet of Things (IoT), enabling real-time monitoring and optimization of production parameters and enhancing energy performance. In parallel, improvements in scrap recycling have been shown to increase the efficiency of EAFs, a vital process in steel production. Renewable energy integration, such as biomass co-firing, provides a means to reduce reliance on fossil fuels and further mitigate the environmental impact of steel production. Collectively, these technologies form comprehensive portfolios tailored to the needs of developing countries, emphasizing their feasibility and adaptability in environments where resources and infrastructure may be limited. The implementation of such technologies, however, requires careful consideration of local conditions and the unique challenges faced by steel producers in these regions [27]. To synthesize this, Table 1 presents a compilation of prominent energy-reduction technologies, including descriptions, potential savings, and supporting references.

Capture and reutilize heat from flue gases and Waste heat recovery cooling processes using organic Rankine cycles 10-30% [9], [28–30] systems or preheaters.

Inject hydrogen into blast furnaces as a Hydrogen injection reductant to minimize coke usage and CO₂ 20-40% [30–32] emissions.

Continuous casting Enhance casting efficiency through near-net-15-25% [33–35] improvements shape casting, reducing the need for reheating.

Produce thinner slabs directly from liquid steel, Thin-slab casting 20-35% [36], [37] minimizing energy in rolling mills.

Co-inject biomass with coal in furnaces to Biomass Co-firing 10-20% [38–40] partially replace fossil fuels.

Advanced process Implement AI-driven controls for optimizing 5-15% [41], [42] controls furnace operations and energy flows.

Scrap recycling Improve scrap preheating and sorting to 15-30% [43], [44] enhancements increase EAF efficiency.

This table draws from recent reviews and serves as the foundation for the screening and ranking in subsequent sections.

This study employs a mixed-methods approach to identify, screen, and rank energy-consumption reduction technologies in the steel industry, with a particular focus on developing countries, including Iran. In the qualitative phase, a Systematic Literature Review (SLR) is utilized to compile a comprehensive set of energy-saving technologies applicable to steel production in resource-constrained settings. This involves analyzing global and region-specific studies to ensure relevance to developing economies. In the quantitative phase, the FDM is applied, with an expert panel screening the identified technologies for suitability to the Iranian steel industry. During this process, some technologies may be excluded, while others may be added based on expert consensus and contextual factors. To rank the screened technologies under uncertainty, the Interval Type-2 Fuzzy Best-Worst Method (IT2-FBWM) is employed, providing a robust multi-criteria decision-making framework that effectively handles vagueness and imprecision in expert judgments. The detailed methodology is outlined below.

The Delphi method seeks to establish expert agreement on a particular subject by conducting several rounds of anonymous surveys, gathering insights from a chosen panel of professionals who possess expertise and experience in a specific domain. Nevertheless, the conventional Delphi approach faces challenges such as uncertainty and a time-intensive process [45].

To address these challenges, the FDM was created by combining fuzzy theory with the traditional Delphi approach. As a collaborative decision-making tool, the FDM emphasizes anonymity, consensus-building, controlled feedback, and the synthesis of participants’ responses using linguistic preferences, making it highly effective for predicting outcomes [46].

The process for applying the FDM involves the following steps [47]:

Step 1. Perform a literature review to pinpoint and gather pertinent criteria for the research issue.

Step 2. Experts assess the selected criteria through a fuzzy Delphi questionnaire to evaluate their significance. Linguistic terms (as presented in Table 2) are employed to articulate the importance of each criterion. Triangular fuzzy numbers (TFNs) are chosen for their simplicity and prevalent use in research. A triangular fuzzy number F=(l,m,u) comprises three components: the lower bound l, denoting the smallest possible value; the upper bound u, representing the largest possible value; and the most probable value m.

Step 3. The definitive set of criteria is established by comparing each criterion's weight against a predetermined threshold. The weight for each criterion is computed using the equations provided below:

a ̃ij = (lij, mij, uij) for i = 1 … n; j = 1 … m, (1)

τ ̃j = (lj, mj, uj) = (min{lij}, (∏ mij) , max{uij}). (2)

Where i represents the expert, n is the number of experts, j is the criterion, and m is the total number of criteria. ̃ represents the fuzzy value assigned to each criterion by an expert, and ̃ is the aggregated fuzzy value for each criterion.

l + m + u

crisp value = . (3)

If the defuzzified value surpasses the established threshold (e.g., 0.7), the criterion is retained; otherwise, it is excluded. The threshold value may be set using different approaches, depending on expert agreement and the study's requirements.

Interval type-2 fuzzy numbers build upon type-1 and type-2 fuzzy logic, offering an enhanced approach to modeling uncertainty in intricate decision-making scenarios. These fuzzy numbers are instrumental when the degree of membership is itself uncertain. Nonetheless, there is also a requirement to decrease computational complexity relative to complete type-2 fuzzy sets [48].

In type-1 fuzzy logic, every value is assigned a precise membership degree within the range [0,1]. In contrast, in type-2 fuzzy sets, the uncertainty in the membership function leads to each value having a secondary membership function. This characteristic enhances decision-making under uncertainty. However, the high computational complexity of full type-2 fuzzy sets restricts their broad adoption. To address this issue, interval type-2 fuzzy numbers were developed, where the secondary membership function is assigned a constant value, thereby defining a range of membership degree values for each element [49].

A key benefit of employing interval type-2 fuzzy numbers is their capacity to capture uncertainty in pairwise comparisons and qualitative evaluations more precisely. This characteristic is particularly valuable in multi-criteria decision-making approaches where linguistic data and expert insights are central. By adopting this method, human judgments can be represented with greater accuracy, leading to more dependable outcomes [50].

In the fuzzy best-worst method, linguistic scales are commonly employed to compare inherently uncertain criteria. Utilizing interval type-2 fuzzy sets within BWM facilitates a more precise representation of ambiguity in expert opinions, especially when there are notable differences in perspectives.

The following steps outline the structured process of IT2F-BWM [51]:

Step 1. Selecting the best and worst criteria Specialists determine the most significant (best) and least significant (worst) criteria from the chosen set.

Experts provide linguistic evaluations to compare the best criterion against all other criteria and all criteria against the worst criterion using interval type-2 fuzzy numbers.

The obtained IT2F best-to-others (IT2FBO) and IT2F others-to-worst (IT2FOW) vectors are:

A ̃ = (A ̃ , A ̃ , … , A ̃ ), (4)

B B1 B2 Bn

A ̃ = (A ̃ , A ̃ , … , A ̃ ). (5)

W W1 W2 Wn

Clearly, A ̃ = A ̃ = [(1,1,1,1), (1,1,1)].

BB WW

The definition of a consistent IT2F preference (IT2FP) is as follows:

The IT2FP ̃ is consistent if

A ˜ × A ˜ = A ˜ , A˜ × A , Worst ˜ = A˜ , j, k ∈ N. (6)

Best ,j jk Best ,k lk k jW

Step 3. Constructing the IT2F-BWM model The obtained fuzzy comparisons are used to formulate a set of optimization equations to determine the optimal weight vector.

minmax{|w ˜ /w ˜ = A ‾ |, |w ‾ /w ‾ = A ‾ |},

B W Bj j W jw

s.t.

∑ C(w ˜ j) = 1

U L L U

wj1 ≤ wj1, wj3 ≤ wj4

L L L

wj1 ≤ wj1 ≤ wj3

U U U U

wj1 ≤ wj2 ≤ wj3 ≤ wj4 {wj1 ≥ 0, j = 1,2, … , n,

where

U U U U L L L

w ‾ B = [(w ‾ B1, w ‾ B2, w ‾ B3, w ‾ B4), (w ‾ B1, w ‾ B2, w ‾ B3)], w ‾ j =

U U U U L L L U U

[(w ‾ J1, w ‾ J2, w ‾ J3, w ‾ J4), (w ‾ J1, w ‾ J2, w ‾ J3), w ‾ w[(w ‾ W1, w ‾ W2, U U L L L ‾ U U U w ‾ W3, w ‾ W4), (w ‾ W1, w ‾ W2, w ‾ W3)], AB.J = [(w ‾ B.J1, w ‾ B.J2, w ‾ B.J3, U L L L ‾ U U U w ‾ B.J4), (w ‾ B.J1, w ‾ B.J2, w ‾ B.J3)], AJ.W[(w ‾ J.W1, w ‾ J.W2, w ‾ J.W3,

U L L L

w ‾ J.W4), (w ‾ J.W1, w ‾ J.W2, w ‾ J.W3)].

To avoid obtaining multiple optimal solutions from the model, one can minimize the maximum absolute gaps between | × ‾ | and | × ‾ |. To solve the model under the assumption that the maximum absolute

‾∗ ∗ ∗ ∗ ∗ ∗ ∗ ∗

gap is = [( , , , ), ( , , ), the model can be transformed through the following programming model:

s.t.

U U U U U U U U U

|wB1 − wJ1wBJ,1| ≤ δ, |wB2 − wJ2wBJ,2| ≤ δ, |wB3 − wJ3wBJ,3| ≤ δ,

U U U L L L L L L

|wB4 − wJ4wBJ,4| ≤ δ, |wB1 − wJ1wBJ,1| ≤ δ, |wB2 − wJ2wBJ,2| ≤ δ,

U U U U U U L L L

|wJ3 − wW3wJW,3| ≤ δ, |wJ4 − wW1wJW,4| ≤ δ, |wJ1 − wW1wJW,1| ≤ δ,

L L L L L L

|wJ2 − wW2wJW,2| ≤ δ, |wJ3 − wW3wJW,3| ≤ δ, ∑ C(w ̃j) = 1,

U L L U U U U U U

{wj1 ≤ wj1, wj3 ≤ wj4, wj1 ≤ wj2 ≤ wj3 ≤ wj4, wj1 ≥ 0, j = 1,2, … , n.

Step 4. Solving the optimization model The objective function minimizes the maximum deviation between the pairwise comparisons and the calculated weights, ensuring consistency in expert judgments.

Step 5. Defuzzification and normalization The final interval type-2 fuzzy weights are defuzzified and normalized to obtain crisp weights for decision-making.

Step 6. Consistency check The consistency of expert judgments is evaluated to ensure reliable weight calculations. The CR inspects the degree of consistency and the reliability of the obtained weights through:

CR = δ /CI. (7)

The value of CR ranges from 0 to 1, where a CR approaching 0 signifies higher consistency, while a CR closer to 1 indicates lower consistency. [51].