Digitalization of the Workflow for Drone-Assisted Inspection and Automated Assessment of Industrial Buildings for Effective Maintenance Management

Abstract

Industrial buildings are a key element in the industrial fabric, and their maintenance is essential to ensure their proper functioning and avoid disruptions and costly economic losses. Continuous maintenance based on an accurate diagnosis makes it possible to meet the challenges of aging infrastructures, which demands a reliable data-based assessment for maintenance management implementing corrective and preventive actions, according to the damage criticality. This paper researches an innovative digitalized process for the inspection and diagnosis of industrial buildings, which leads to categorizing and prioritizing maintenance actions in an objective and cost-effective way from the inspection data. The process integrates some technical developments carried out in this work, aimed to automate the workflow: the drone-based inspection, the building condition assessment from the definition of a standardized construction pathology library, and a visual analysis of pathology evolution based on photogrammetry. The use of drones for digitalized inspection involves some challenges related to the positioning of the drone for damage localization, which has been herein overcome by developing a geo-annotation system for image acquisition. This system has also enabled the capture of geo-located images intended to generate 3D photogrammetric models for quantifying the pathological process evolution. Moreover, the assessment procedure outlined through multi-criteria decision-making methodology MIVES establishes a single criterion to automatically weight the relative importance of the damage defined in the library. As a result, this procedure yields the so-called Intervention Urgency Index (IUI), which allows prioritizing the maintenance actions associated with the damage while also considering economic criteria. In such a way, the overall process aims to increase reliability and consistency in the results of inspection and diagnosis needed for the effective maintenance management of industrial buildings.

1. Introduction

Industrial sites are highly complex built environments and fundamental elements of the industrial fabric. Their capacity to provide this service under safe conditions is key for the development of the industry, representing a significant source of investment to ensure their proper functioning. The complexity and diversity of these infrastructures, as well as the wide range of interconnected components, make surveying and maintenance activities specially challenging. Additionally, the aged deterioration of the constructive elements of the industrial buildings—such as roofs, trusses, or columns, among others—threatens the service life of the infrastructure and, accordingly, the industrial activity. Therefore, it is critical to perform and manage preventive maintenance on the industrial building elements, rather than the reactive and corrective approach commonly addressed so far. This will enable management of the risk status of these infrastructures and avoid situations that could result in negative economic impacts or personal injuries for workers and users of these infrastructures.

In this regard, it is worth noting that the Long-Term Strategy for Energy Rehabilitation in the Building Sector in Spain (ERESEE) [1] 2020 identifies a total of 1,715,782 industrial buildings, of which 775,900 are over 50 years old. Additionally, when considering the full life cycle of a building, it is estimated that 5% of its cost corresponds to the design phase, 20% to construction, 65–80% to maintenance and operating expenses, and the remaining 10% to rehabilitation or demolition costs [2]. Therefore, addressing this issue from an economic perspective is also essential to effectively manage maintenance and optimize the costs associated with the process. Worldwide, the market for infrastructure inspection services has experienced solid growth in recent years, exceeding $2.05 billion in 2023. It is forecast to reach $2.93 billion in 2028 at a compound annual growth rate (CAGR) of 7.5% [3].

Current research emphasizes the importance of efficiently interconnecting innovative methodologies in the inspection and maintenance of industrial buildings to address the inherent limitations of traditional practices. For instance, studies such as [4,5] highlight how preventive maintenance programs can significantly reduce operational costs and improve safety. Similarly, research by Srivastava et al. [6] demonstrates the potential of leveraging digital technologies, including drones and photogrammetry, for optimizing the inspection process, while Ly et al. [7] explore the use of algorithms for pathology assessment, underscoring their capacity to enhance objectivity and efficiency. Despite these advances, a notable gap remains in the combination of these technologies into a cohesive, automated workflow specifically tailored for industrial infrastructures. This gap limits the full realization of their potential, particularly in terms of cost optimization and lifecycle management of damage.

Nevertheless, the Inspection and Maintenance (I&M) sector demands innovative solutions for improving the efficiency of the process, and the gain in productivity is linked to a higher level of digitalization and automation of the process. In this respect, it is noticeable that the construction sector is currently undergoing a period of transformation and change. Approximately 98% of construction and engineering companies believe digital transformation of their services is necessary, but very few are implementing it, with only 28% of them claiming their company has a strategy towards digitalizing their services [8]. Furthermore, inspection and maintenance strategy rarely involves a methodology for the digitalization and automation of the process, as inspection and maintenance are often carried out manually and subjectively. The lack of a methodological framework for the digitalization of the process affects the quality, effectiveness, and efficiency of inspections, resulting in delays and, accordingly, additional maintenance costs. Moreover, there is a large component of subjectivity that primarily depends on expertise and experience, which undermines the objectivity of the results and complicates rational decision-making. Effectively managing this information, keeping accurate records, and ensuring traceability of actions are challenging for objective, informed, and cost-effective decision-making.

Whereas some stages of the process, such as assessment through multicriteria decision-making methodology, are automated in this work, others are being looked at for potentially being automated, creating the potential to leverage the use of innovative technologies. This is the case with the inspection phase, currently based on operator-driven survey, which is not fully automated at this stage but that may take advantage of technologies such as photogrammetry or machine learning to reach its full potential in the future, as presented in the research.

The need to maintain industrial buildings in proper working condition is based not only on safety and operational efficiency but also on maximizing the service life of assets. Various studies agree and have demonstrated that effective preventive maintenance can significantly reduce operational costs and the risk of failures in critical infrastructure [4,5]. Furthermore, implementing well-structured preventive maintenance programs can extend the life span of industrial assets, optimizing long-term investment, ensuring uninterrupted production without costly disruptions, and contributing to reducing the environmental impact of the construction sector by minimizing demolition waste generation.

Finally, it must be highlighted than the inspection, control, and maintenance of the structural elements of today’s industrial buildings require a labour-intensive process, which entails significant operational challenges. The need to access hard-to-reach or confined areas, the inaccessibility of certain parts of the structure, or its proximity to hostile or hazardous environments (corrosive, explosive…) all have an impact on the need to use specialized auxiliary equipment, large-scale safety measures, or even production stoppages, which increase the cost of the inspection processes. This situation not only compromises the safety of operators, but also implies errors and inefficient and unsustainable processes, a consequence of the low level of digitalization and automation of these tasks. As a result of this issue, and envisioning the potential and versatility that drones could provide to address this task, a remarkable increase in their application in this field has been noticed in recent years. In fact, the specific market for drones in energy infrastructures is foreseen to continue being the main market for drone technology worldwide, just beyond military applications. In fact, the drone inspection and monitoring market size was valued at USD 10.2 billion in 2022 and is expected to grow at a CAGR of 14.5% from 2022 to 2032 [9].

When discussing the preventive maintenance of industrial buildings [3,10], the process and workflow referred to should cover several stages: (1) data capture; (2) objective evaluation of damage; (3) definition of corrective and preventive actions for maintenance; and (4) short/medium-term planning of maintenance costs, including the quantitative management of the damage life cycle. The automation of the process and the interoperability of these phases will leverage the optimization of the maintenance costs, thereby ensuring the life span of industrial buildings.

In this context, this research presents a methodology for streamlining the surveying and maintenance process of industrial buildings through the use of advanced technologies, such as drones, photogrammetry, and damage assessment algorithms, which are integrated into a comprehensive workflow to automate and optimize maintenance management for smart decision-making.

For that purpose, the methodology combines several developments into a common workflow aimed at streamlining the process: (1) a geo-annotation system for drones, that links the inspected damage to the drone flight path; (2) a procedure of digital comparison of pathologies through photogrammetry for analysing the evolution of damage over time; and (3) an algorithm for pathology assessment based on the MIVES methodology (Integrated Value Model for Sustainable Evaluation) from a construction pathology library specifically designed for industrial buildings.

In particular, the research outlines an innovative process for the digitalization of the inspection and diagnosis of industrial buildings, with the novelty lying in the following aspects:

• Systematization of the inspection through the use of drones, that can be even further automated.
• Overcoming key challenges such as the accurate localization of damage through a geo-annotation system for the acquired images.
• Facilitating the generation of 3D photogrammetric models, which improves the ability to assess damage evolution more accurately.
• Assessment of the damage and prioritization of the maintenance actions according to damage severity and economic criteria by using a multi-criteria decision-making methodology (MIVES) that can optimize maintenance resources in a cost-effective manner.

The innovative technologies above-mentioned, which can be used separately and standalone, reach their full potential for an innovative and more efficient way to improve maintenance once interconnected under an overall framework. Therefore, the innovation of this approach lies not only in the technologies used for covering the different phases of the surveying and maintenance (use of drones, photogrammetry, algorithms of damage assessment), but also in the combination of the phases towards the automation of the process itself, oriented to provide an agile, reliable, and effective system aimed at optimizing the management of preventive maintenance for industrial assets.

The rest of this article is organized as follows: Section 2 presents the methodology, including the development of an image geo-annotation system for drone inspection, the design of a standardized construction pathology library, a pathology assessment system, and the analysis of damage evolution through photogrammetry. Section 3 provides the results, detailing the procedure for drone-based inspection, the construction pathology library, and the system for damage assessment using the Intervention Urgency Index (IUI). Section 4 summarizes the conclusions, while Section 5 outlines future lines of research.

2. Methodology

The goal of this research is setting the framework for the digitalization and automation of the inspection surveying and diagnosis from the data capturing with drones and a data-based assessment of industrial buildings. For that purpose, the methodology of the research comprises some developments oriented to digitalize the key phases, which once integrated, will lead to the automation of the whole process.

The research provides a geo-annotation system for location of the images of damage taken by drone, as well as a system for analysing the evolution of the pathologies based on photogrammetry. Additionally, it offers a standardization of a construction pathology library, including all the elements of the assets, together with their potential damage as well as the development of an objective damage evaluation methodology that facilitates decision-making, while also considering cost criteria.

The methodology has been designed to address three key needs in the field of damage inspection and assessment, which have been identified by the industry sector: (1) improvement of the inspection and surveying; (2) automation of building assessment; and (3) easing of a decision-making process for maintenance. For that purpose, the methodology addresses the following challenges:

• Digitalization of the inspection, even in hard-to-reach zones supported by inspection drones, generating a standardized database of pathologies common to the industrial buildings, and systematically registering the detected damage according to the structure of the database. The library contains an inventory of potential pathologies that may occur in an infrastructure under maintenance.
• Automation of damage evaluation and prioritization of maintenance actions from the construction pathology library, through an evaluation system that allows for setting priorities for decision-making based on indicators, following the multi-criteria decision-making methodology MIVES [11,12]. This will allow the quantification of damage severity and the prioritization of pathologies through the so-called IUI (Intervention Urgency Index). Based on the database of pathologies, using algorithms, and weighting various characteristics and/or indicators, and following quantitative and objective criteria, a new urgency index evaluation system is generated. This system facilitates decision-making and enables the design of maintenance plans tailored to the inspection results. A system based on image analysis through photogrammetry for measuring damage evolution has also been developed in the current research, which can be helpful for providing qualitative information for the assessment.
• Facilitating an objective, data-based, and user-centred decision-making process, which provides the end-user with the ability to manage preventive maintenance in an informed, integrated, and automated manner.

The scheme below (Figure 1) depicts the methodology for automating the inspection, diagnosis, and assessment of industrial buildings presented in this research, as well as the technologies utilized for that goal, which will be further described.

Furthermore, for the inspection and recording of asset assessment information, the use of innovative technologies is applied by developing a geo-annotation system for the use of drones and a process to verify the evolution of the damage using 3D photogrammetric models.

2.1. Outlining an Image Geo-Annotation System for Drone Inspection

The inspection of industrial environments often requires access to hardly reachable, confined, or hazardous areas in order to survey certain parts of the structure, which occasionally involves an inevitable production shutdown, leading to significant operational costs. This need raises the possibility of using unmanned aerial vehicles (UAV), or drones, which enable the inspection of hard-to-reach areas, while allowing the inspesction to be automated. The drones’ capacity to embark image acquisition systems (RGB, thermal…) and sensors, as well as their flight capability, makes them useful to replace or complement traditional inspection methods. This fact makes drones especially suitable for inspections of inaccessible building elements.

The use of drones for capturing images of pathologies entails geo-annotation of the accurate position of the pathology when analysing a component of the building inspected. During the inspection work, it is necessary for each pathology to be associated with one or more geo-annotations, which describes the point in space where the pathology is located. The geo-annotation stores the image of the pathology obtained using the drone, which also serves as a reference for generating the photogrammetric model.

For this phase, a solution has been developed to control the drone in order to obtain as precise an annotation as possible, collecting a series of data from the drone’s sensors such as GPS coordinates, the frontal distance from the drone to the structure being observed, drone altitude, camera angles, etc. [13]. These data are associated with the images of the detected damage as metadata, which serve as localization information of the pathology. This set of data forms the geo-annotation.

To develop the geo-annotation solution, the DJI Mobile SDK framework [14] has been used, which allows the development of apps that control nearly all components of DJI drones, as well as creating a user-friendly interface for drone control.

The development of an Android-based application has been designed with an interface that includes a sidebar with camera and video controls. This allows, once the drone is positioned, the camera to be optimally configured before performing the geo-annotation.

The geo-annotation of the observed damage image is registered when the operator takes a photo. At this moment, the photo is recorded, and a file is created with the following data:

• Latitude.
• Longitude.
• Altitude.
• Orientation.
• Front distance to the element in meters.
• Camera pitch.
• Camera roll.
• Camera yaw.
• Drone pitch.
• Drone roll.
• Drone yaw.
• DATE.
• TIME.

In this way, the state of the drone as the photo is taken is recorded, with the GPS coordinates, altitude, and distance to the element providing the position, and the other parameters helping to determine the orientation of the device and the camera. This allows the precise position and orientation of the drone to be known, as well as the exact direction in which it is focused, to later pinpoint the pathology location, making it easier to find during subsequent inspections.

2.2. Design and Standardization of a Construction Pathology Library

A structured and systematized construction pathology library has been defined, aimed to respond to the damage assessment system. It includes the elements susceptible to damage, the common pathologies in those, and the appropriate interventions with cost estimation associated with each type of damage. The development of the library is the first step for the evaluation system which facilitates the decision-making for the maintenance and is defined according to the following approach:

• Definition of the data structure in the construction pathology library.
• Definition of the components set in the construction pathology library, including the following:
• Identification and categorization of constructive elements which compose the industrial buildings.
• Identification and classification of the pathologies affecting each construction element, for typology, material of elements, and origin of the damage.
• Definition of repair patterns and economic valuation associated with the pathological processes identified and collected in the library.

2.2.1. Data Structure of the Construction Pathology Library

First, in order to facilitate the generation of the data library, and to ensure that its result meets the objectives of the research, the necessary information specifications and their structuring have been analysed.

To ensure that the construction pathology library encompasses all possible pathologies and pathological processes that may occur in the different elements that make up a building, a pyramidal classification has been generated as shown in Figure 2. This classification covers the full range of construction elements, along with the various pathological processes associated with them, defining accordingly the optimal structure which will feed the digitized evaluation system.

In the first level, more general aspects have been considered, such as the category of the construction element, creating five main groups (external/interior enclosures, structural elements, carpentry, etc.).

In the second level, more specific definitions are provided within each construction category, such as the typology of the construction element, creating different subcategories for each of the higher-level categories, such as wall, column, pavement, roof.

In the third and fourth levels, the degradation processes associated with each construction element have been defined, including their cause/origin and the corresponding pathology.

Finally, a fifth level has been created for each identified pathology in the previous level, which includes information on the repair procedures and work for each pathology, along with an economic estimation of the repair costs. In this way, decisions can also be made considering economic criteria.

2.2.2. Definition of the Elements of the Construction Pathology Library

Once the information structure required was defined, and taking into account both the typology of the construction element and its material, a tree structure was generated including all the possible damage that may occur in the different construction systems composing a building.

Level 1. Construction Category: In this first level, five major groups of construction elements have been defined:

• External enclosures.
• Internal enclosures.
• Carpentry.
• Structural elements.
• Roofs.

The only elements excluded from this classification are the foundations of the building, as these elements cannot be inspected during the building’s life cycle, neither by visual inspection nor with the help of a drone.

Level 2. Construction Element: Based on the first level, the following subcategories of construction elements have been defined, considering systems belonging to each element of the higher categories, according to the typologies and materials typically used for construction elements, as shown below:

• External enclosures:

  ○

  Load-bearing stone walls.

  ○

  Load-bearing masonry walls.

  ○

  Concrete bearing walls.

  ○

  …

• Internal enclosures:

  ○

  Vertical partitions of blocks/ceramics.

  ○

  Vertical partitions of plasterboard.

  ○

  Vertical metal partitions.

  ○

  …

• Carpentry:

  ○

  Wooden windows.

  ○

  PVC windows.

  ○

  Metal windows.

  ○

  …

• Structural elements:

  ○

  Concrete columns.

  ○

  Concrete beams.

  ○

  Metal columns.

  ○

  …

• Roofs:

  ○

  Sloping roofs with ceramic finish.

  ○

  Sloping roofs with metal finish.

  ○

  Flat roofs with asphalt finish.

  ○

  …

Level 3. Cause/Origin: Based on the classification defined for Levels 1 and 2, Level 3 has been completed, identifying all causes (physical, mechanical, manufacturing and/or execution failures, etc.) that may lead to the pathological processes present in the elements defined in Level 2 of the library.

Level 4. Pathology: Based on the sets of causes and origins defined in Level 3, all pathological processes and damage that may occur and manifest in the different construction elements have been identified.

To complement the library, an exhaustive analysis of the state of the art in research and publications related to pathologies in construction has been carried out. This analysis has allowed the identification and referencing of previous studies that enrich the database, ensuring a broad and detailed coverage of potential pathological processes and their causes [15,16,17].

With the information from the 4 levels, the library has been completed with information on the different degradation processes associated with each of the defined construction elements, including both the cause/origin of these and the pathology produced in them, following the structure shown in

Table 1. Extract from the construction pathology library.

LEVEL 1	LEVEL 2	LEVEL 3	LEVEL 4
CONSTRUCTION CATEGORY	CONSTRUCTION ELEMENT	CAUSE/ORIGIN	PATHOLOGY
EXTERNAL ENCLOSURES	LOAD-BEARING MASONRY WALLS	MECHANICAL DAMAGE	CRACKS
			FRACTURES
			BLOWUPS/ DEFORMATIONS
		MOISTURE	CAPILLARITY
			FILTRATION
			CONDENSATION
		DETACHMENTS	CONTINUOUS FINISHING
			ELEMENTS
		DIRT	EFFLORESCENCE
			DEPOSITS
			DIFFERENTIAL WASHING
	CONCRETE BEARING WALLS	MECHANICAL DAMAGE	CRACKS
			FRACTURES
			BLOWUPS/ DEFORMATIONS
		MOISTURE	CAPILLARITY
			FILTRATION
			CONDENSATION
		MANUFACTURING AND/OR EXECUTION FAILURES	CONCRETE HOLLOWS
			LACK OF COVERAGE
			CONCRETE DEAGGREGATION
		DETACHMENTS	CONTINUOUS FINISHING
		DIRT	EFFLORESCENCE
			DEPOSITS
			DIFFERENTIAL WASHING

.

Level 5. Pathology Repair: Once the digital construction pathology library has been defined and refined, Level 5 of the identified pathological processes has been completed. This level includes a description of the repair work required to address the damage caused by the pathological process, along with an economic estimation of the cost of these repair works. Additionally, for each identified pathology, a sample image representing the produced pathological process has been included.

By directly linking the damage detected in an industrial building during maintenance work with the necessary repair tasks and their costs, it is possible to automatically generate the economic valuation of the intervention. In such a way, an estimated budget for the required repair work can be obtained based on the prioritized pathologies, and the economic factor can also be considered in the decision-making process for maintenance.

2.3. Pathology Assessment System

Once the construction pathology library has been systematically defined to automate damage evaluation, a damage evaluation system can be developed. This system allows the determination of the intervention urgency index (IUI) for the identified pathologies in various construction elements, thereby establishing intervention priorities.

The approach of the methodology for the assessment lies in a decision-making formulation based on the definition of KPIs (Key Performance Indicators) and subsequent determination of weights and alternatives. The weighting of indicators and alternatives has been set with the valuable help of an expert panel, composed of 20 specialists in the construction domain, particularly in the field of pathology analysis, which has been successively enquired until achieving fine-tuned results. Despite its limitations, this approach provides insightful outcomes when complex scenarios are addressed, ensuring consistency and transparency in guiding end-users through the decision-making process.

The following sequence represents the steps carried out to build the damage evaluation system. This structure is also the one used for describing the next subsections for Section 2.3:

• Definition of evaluation indicators and determination of individual evaluation criteria.
• Weighting of the indicators.
• Definition of indicator alternatives.
• Weighting of indicator alternatives.
• Calculation and classification of IUI—Intervention Urgency Index.

2.3.1. Definition and Weighting of Evaluation Indicators

Firstly, it was necessary to define the indicators through which the damage would be evaluated, to determine the urgency of intervention. It is essential that the indicators selected for the evaluation system, which are used to obtain the IUIs associated with the pathologies present in an infrastructure, are quantifiable, objective, and understandable to anyone using the methodology. These indicators should be based on statistical and observable facts, not subjective opinions.

The importance of indicators in the context of pathology evaluation in construction lies in their ability to provide an objective and quantifiable basis for measuring and analysing the state of construction elements. Indicators serve as essential tools for identifying, monitoring, and predicting possible failures and degradations, facilitating informed decision-making and timely corrective actions. According to Nardo et al. [18], indicators allow complex information to be synthesized into a more manageable and comprehensible form, which is essential for evaluating various factors in an integrated manner in the context of construction and maintenance. Parmenter [19] emphasizes that the proper development and implementation of KPIs not only optimizes the visibility of the outcome and, as in our case, shows the potential risk of a particular type of damage, but also enables the coordination of maintenance strategies aligned with the objectives of industrial building owners, optimizing the management of construction pathologies and ensuring an adequate and timely response to repair and prevention needs.

To ensure the reliability of the pathology evaluation system presented, the indicators must be precise, relevant, and easily interpretable. Additionally, they must be specific to the preventive maintenance context, sensitive to changes in the evaluated conditions, and consistent in their measurement over time. The validity of the indicators should be strengthened through their alignment with the objectives of the developed evaluation system and their ability to accurately reflect the operational realities of the built environment.

It is also necessary that any person using the system clearly understands the objective or result sought by the indicator, as well as all the alternatives that compose it. Therefore, the selected indicators must be intuitive and easy to interpret. Simplicity and clarity in defining the indicators are essential to ensure their accessibility [19,20,21]. This implies using clear and precise language, as well as effective visual representations, such as graphs and tables, that facilitate the user’s understanding. By being comprehensible to all users of the system, from operators to managers, effective and consistent use of the indicators is ensured, which in turn contributes to a more reliable and accurate evaluation of construction pathologies.

Considering everything described in the previous paragraphs, it has been determined that the following four indicators are used to obtain the IUI classification that allows evaluating the damage in the building: Severity, Evolution, Impact on Others and Extension.

• Severity: Evaluates the decrease in the functional capacity of the construction element as a result of the damage.
• Evolution: Evaluates the likelihood of the damage progressing more or less rapidly if not intervened upon.
• Impact on Others: Evaluates the repercussions of the existing pathology in one element on other elements, the environment, third parties, critical machinery, and the functioning of the installation, etc.
• Extension: Evaluates the surface area affected by the pathology in relation to the total surface area of the analysed element.

Secondly, after defining the indicators, the relative weight of each indicator has been determined and assigned to enable the evaluation of pathologies. These weights are assigned using pre-established formulas defined by expert criteria. Linstone et al. [22] describe the use of expert judgment as a method that allows integrating multiple perspectives and specialized knowledge to reach informed and robust consensus. According to Murray et al. [23], the experience and judgment of the expert panel are crucial for conducting an accurate, rigorous evaluation that aligns with the indicators, as they provide a deep and nuanced understanding of the factors at play. This approach ensures that the resulting weights reflect not only empirical data but also the tacit knowledge and accumulated experience of experts in a particular field.

The weighting of the indicators has been obtained using the expert panel method, which draws on information from a group of people with expertise in a specific area. In this case, the collaboration of more than 20 experts with extensive experience in inspection, pathology, rehabilitation, maintenance, and repair of industrial buildings has been involved, from both the construction sector and academia.

To assign the weights of different indicators, and to ensure they align with their real importance and increase the transparency of the indicator determination process, the AHP (Analytic Hierarchy Process) system has been used. AHP is based on a group of decision-makers determining the relative importance of each indicator compared to others, using a scale of priorities. According to Sipahi et al. [24], AHP facilitates decision-making by breaking down a complex problem into more manageable hierarchies, allowing pairwise comparisons to quantify the relative importance of each element. The AHP decision method allows quantifying the relative priority of each alternative according to the scale, emphasizing the importance of the decision-maker’s intuitive criteria and the consistency of the comparisons between alternatives based on his or her judgment. This methodology also systematically organizes tangible and intangible factors, providing a structured and simple solution to the decision-maker’s problems. Omkarprasad et al. [25] emphasize that using a priority scale in AHP helps convert qualitative comparisons into quantitative values, thus improving the precision and coherence in assigning weights. Furthermore, Mendoza et al. [26] highlight that AHP not only systematizes the decision-making process but also increases transparency and understanding of the process, ensuring that the assigned weights accurately reflect the priorities and perspectives of the decision-making group.

In this way, weights are obtained from the subjective importance of each element in relation to others, through pairwise comparisons using a decision matrix. According to Saaty [27], the Analytic Hierarchy Process (AHP) is based on pairwise comparison of elements to construct a decision matrix, where each element is evaluated in relation to another, assigning numerical values that reflect their relative importance. Ishizaka et al. [28] stress that the pairwise comparison matrix is crucial in AHP, as it allows for a structured and systematic evaluation, facilitating the assignment of weights more accurately and reliably.

The pairwise comparison process, according to the proposal of T. Saaty [27], relies on a numerical scale from 1 to 9, where each value is associated with a verbal descriptor to help decision-makers express their judgments more intuitively. This verbal scale is not meant to imply mathematical or proportional relationships between the numerical values but rather serves as a qualitative guide to support the evaluation process. For example, ’Extremely more important’ corresponds to a numerical value of 9, while ’Slightly more important’ corresponds to 3, as shown in

Table 2. Classification for indicator weighting (As defined by Saaty [27]).

Relative Importance (x Compared to y)
Verbal Scale	Numerical Scale
Extremely less important	1/9
Much less important	1/7
Less important	1/5
Slightly less important	1/3
Equal importance	1
Slightly more important	3
More important	5
Much more important	7
Extremely more important	9

. This methodology not only enhances consistency in decision-making but also provides a transparent and understandable way to incorporate subjective judgments into the weighting process. (See Table 2 and

Table 3. Example pairwise comparison matrix.

Matrix-Verbal Scale
Researcher 1	SEVERITY	EVOLUTION	IMPACT ON OTHERS	EXTENSION
SEVERITY	Equal importance	More important	Slightly more important	Much more important
EVOLUTION		Equal importance	Less important	Slightly more important
IMPACT ON OTHERS			Equal importance	Much more important
EXTENSION				Equal importance

).

When conducting interviews with the various experts to complete the pairwise comparison matrix, the consistency and coherence of the results obtained have been considered, with the aim of evaluating the consistency of the values established by the experts in the decision matrix and avoiding inconsistent evaluations. Consistency is associated with two distinct characteristics: transitivity and proportionality [27,29,30,31].

Transitivity implies that the ordered relations between the compared elements are respected. For example, if it is considered that the importance of A is greater than that of B (IA > IB) and that the importance of B is greater than that of C (IB > IC), then it should hold that IA > IC. On the other hand, proportionality ensures that the proportions between the magnitude orders of the responses are maintained. For example, if IA is 3 times greater than IB and IB is 2 times greater than IC, then it should hold that IA is 6 times greater than IC.

If these two characteristics are fulfilled for all the elements in the decision matrices, the consistency would be 100%.

2.3.2. Definition and Weighting of Indicator Alternatives

Once the indicators that the system uses for the evaluation of the Urgency of Intervention Indexes (IUI) are defined and weighted, it is necessary to specify the alternatives available for each indicator.

To ensure a uniform criterion and to facilitate the technician’s work during the inspection, the number of available alternatives for each indicator has been standardized. Also, a total of four predefined alternatives for each indicator have been decided. This decision’s main objective is to simplify the inspection and evaluation process by providing a clear and consistent structure that the technician must follow. Limiting the number of alternatives to four minimizes complexity and facilitates the comparison between different options. This standardization allows the technician to focus the analysis more efficiently, avoiding ambiguities and reducing the risk of errors. Furthermore, by using the same number of alternatives for all indicators, it ensures that the evaluation is equitable and that all aspects are considered under the same reference framework. This way, greater consistency in applying the evaluation criteria is promoted, contributing to the reliability and validity of the results obtained. The alternatives available for each defined indicator are shown in

Table 4. List of indicators and alternatives.

SEVERITY	Evaluates the Decrease in the Functional Capacity of the Construction Element due to Damage
Null	Defects that do not affect the functional capacity of the construction element
Low	Defects that indicate the beginning of a pathological evolution that decreases the element’s function
Medium	Defects that affect the functional capacity of the construction element
High	Defects that lead to the element’s functional capacity approaching its limit state
EVOLUTION	Evaluates the Probability of the Damage Advancing More or Less Rapidly if No Intervention Is Made
Null	No damage evolution
Low	The damage evolves slowly
Medium	The damage evolves at a medium pace
Fast	The damage evolves rapidly
IMPACT ON OTHERS	Evaluates the Impact of the Existing Pathology on Other Elements, the Environment, Third Parties, Critical Machinery, and the Installation’s Operation, etc
Null	No impact on other construction elements, third parties, or installation functions
Low	The damage may cause minor impacts on other construction elements, third parties, or installation functions
Medium	The damage may significantly affect other construction elements, third parties, or installation functions
High	The damage may cause significant impacts on other construction elements, third parties, or installation functions
EXTENSION	Evaluates the area Affected by the Pathology, Relative to the Total Area of the Analyzed Element
<5%	The damage affects less than 5% of the surface of the construction element
5–25%	The damage affects between 5% and 25% of the surface of the construction element
25–50%	The damage affects between 25% and 50% of the surface of the construction element
>50%	The damage affects more than 50% of the surface of the construction element

:

After defining the alternatives for each indicator, it is necessary to weight each one of them, and thus obtain the value index for each of the alternatives proposed. This weighting has been done by defining the value function for each of the indicators according to expert criteria.

The main objective of the value function is to compare the assessments of indicators with alternatives of different measurements. The value function allows transforming a quantification of a variable or attribute into a dimensionless variable between 0 and 1. This way, a weighted sum of the different values of each indicator can be performed.

To define the different value functions for each indicator, the MIVES methodology [32,33] (Integrated Value Model for Sustainability Assessments) has been used. MIVES provides a model to support comparison and decision-making, allowing a single value index for the evaluated process. The MIVES methodology facilitates the conversion of various characteristics of the objects to evaluate into a series of homogeneous and quantifiable parameters, simplifying the objectification of the choice.

One of the main features of the MIVES methodology is that the valuation model is established before the creation of alternatives. This means that the criteria and weightings used to evaluate the alternatives are defined in advance, ensuring that decision-making is done impartially and without the influence of the specific evaluations of the available alternatives. This structured and transparent approach avoids any form of subjectivity and guarantees that decisions are based on a rigorous and consistent analysis of the previously established value indicators.

The MIVES methodology has been successfully applied in various contexts, demonstrating its capacity to integrate multiple criteria and to generate a value index that facilitates comparison and decision-making. By providing a quantitative and objective framework, MIVES not only improves decision quality but also enhances transparency and the justification of decisions to stakeholders.

Given the optimal results achieved with this methodology and the lack of others with a similar level of application, it has been chosen for this project. The value functions are defined using the expression shown in Formula (1):

$$V_{ind}=B·\left[1-e^{-K_i·{\left({\displaystyle \frac{\left|X-X_{mín}\right|}{C_i}}\right)}^{P_i}}\right]$$

(1)

Formula (1) used by MIVES to define the value functions [34,35].

• Where:

X _min is the abscissa value where the function value is zero (in the case of increasing value functions);

X is the abscissa of the evaluated indicator (variable for each alternative);

P_i is a shape factor that defines whether the curve is concave, convex, linear, or S-shaped. Concave curves are obtained for values of Pi < 1, convex or S-shaped curves for Pi > 1, and the curve tends to be linear for Pi = 1. It also approximately determines the slope of the curve at the inflection point with coordinates (Ci, Ki);

C_i approximates the abscissa of the inflection point;

K_i approximates the ordinate of the inflection point;

B is the factor that ensures the function stays within the value range of 0 to 1. This factor is defined by Formula (2).

$$B={\left[1-{e^{-K_i·\left({\displaystyle \frac{\left|X-X_{mín}\right|}{C_i}}\right)}}^{P_i}\right]}^{\left\{-1\right\}}$$

(2)

Formula (2) to obtain the factor B, which allows the value function to remain in the range [0.00–1.00].

Value functions can take different forms depending on the indicator being assessed, allowing for various shapes: linear, concave, convex, or S-shaped (Figure 3).

The weighting of the alternatives for the different indicators in the presented evaluation system has also been performed based on expert criteria, ensuring that the evaluations accurately reflect the experience and knowledge of specialists in the maintenance of industrial infrastructures. This process involved gathering and analysing responses from a panel of experts, who evaluated each of the alternatives defined for each indicator based on their relevance and contribution to the overall project objective. These value functions, like the one shown in Figure 3, allow the conversion of input data into a unified value index. This index facilitates the direct comparison of different alternatives and supports an informed and well-founded decision-making process.

The combination of expert knowledge with the MIVES methodology for weighting and evaluating the different alternatives for each indicator enables a rigorous and transparent evaluation process. This not only improves the quality and reliability of the decisions made but also increases transparency and justification to stakeholders, thus optimizing the management and maintenance of industrial infrastructures evaluated under the developed system.

2.3.3. IUI Classification–Intervention Urgency Index

Following the weighting of the indicators, the definition of the value function assigned to each one of them, and obtaining a value for each alternative, calculations can be made afterwards to evaluate the potential that pathologies (included in the library) could have in an industrial building, and thus to obtain the IUI to assist in decision-making.

In this phase, the results provided by the tool are obtained. The inspector, knowing the defined alternatives and based on the numerical value of each, will make a decision and assign an alternative to each of the types of damage recorded in the library.

For the creation of the IUI classification used in the developed evaluation system, it was necessary to assign an option from the defined indicator alternatives to every pathology included in the library.

The prioritization of actions is defined by the IUIs of the pathologies, and the IUI of each pathology identified during the inspection is defined by the 4 indicators described in the previous sections and the assignment of relative weights determined for each and their respective alternatives.

To streamline the field inspection process and ensure the objectivity of the result, some of these values are pre-set, and others will be evaluated by the inspector on-site. The evaluation of the indicators of IMPACT ON OTHERS and EXTENSION will be defined by the inspecting technician during the inspection, as their outcome cannot be assumed or anticipated and will depend on what the technician observes during the inspection. However, the values for the indicators of SEVERITY and EVOLUTION are defined in the tool, and their evaluation is based on expert knowledge. This way, the inspecting technician will only need to evaluate two of the defined indicators, as the tool will automatically present the values for the other two.

To prioritize the necessary actions after identifying the different pathologies that the inspected building may present and assigning the IUIs for each of them, the values defined for the different alternatives and their corresponding weights for each indicator, ranging from 0 to 1, must be translated to a scale that can define and prioritize the required repair activities. For this purpose, the formula shown in Formula (3) has been used.

$$V_{damage}=\sum _{i=1}^n{V}_{alternative}\times {Weight}_{indicator}=0-1$$

(3)

Formula (3) to transform the information of the indicators to a measurable scale between 0–1.

2.4. Analysis of Damage Evolution Through Photogrammetry

Besides the method used for a systematic assessment of pathologies, the visual examination of the images taken by drones also allows the monitoring of the pathologies’ evolution in time by comparing photogrammetric models generated from a vast range of images. Therefore, the generation of a 3D geometric model has also been tackled in this research in order to enable an objective and quantifiable evaluation of the pathology, through the application of photogrammetry. The analysis of a sequence of images taken periodically over time by photogrammetry could be further automated, but in the framework of this research, this activity contributes to qualitatively assessing the evolution of the pathology.

Photogrammetry allows for the creation of a 3D model of an object from multiple photos of that object taken from different angles [36]. An algorithm processes the images, detects overlaps between the photos (Figure 4), and can extract a three-dimensional model of the object with a high level of detail, depending on the number of photos used to extract and compare the point cloud generated.

In this sense, the developed digital evaluation system will include functionalities for uploading the photos obtained from a pathology for which a 3D model is to be created. Free software that supports the functionality for generating the 3D model of the pathology and comparing different models taken at different times will be used [37].

Thus, with these 3D models generated from the pathologies, it is possible to measure and compare the evolution of an instance of damage from one inspection to the next. The resulting 3D model is identical, without considering slight differences between the photos taken at different times.

The use of photogrammetry, as opposed to simple photography, allows for a more precise comparison of progress, even enabling the detection of damage development in depth, not just in the plane of the photograph.

3. Results

Based on the methodology described above, the main result of the research is the procedure for the objective diagnosis for industrial building, based on the IUI indicators, which establishes a unique criterion for identifying and weighting the results of inspections.

In addition to this main result, the research has produced intermediate results such as the construction pathology library, the geo-annotation solution for inspections carried out by drones, and procedures for the generation and comparison of photogrammetric 3D models.

The results of the research can be used standalone for inspection and assessment processes, mainly focused on industrial buildings, contributing to the state of the art. Moreover, they can be integrated in a seamless workflow for inspection, surveying, and diagnosis, leading to the automation of the processes intended to improve their efficiency and optimize the preventive maintenance of industrial buildings.

3.1. Procedure for Inspection of Industrial Building by Drones

The application of drones to inspect industrial buildings has proven to be a significant advance in terms of accuracy, efficiency, and safety compared to traditional inspection methods. This methodology takes advantage of the ability of drones to capture detailed visual information, which can then be processed into 3D models using photogrammetry. In addition, these models provide a better understanding of the actual condition of building elements, facilitating preventive maintenance management and informed decision-making.

The developed methodology emphasizes critical aspects to be considered when inspecting and capturing images by drone to create 3D photogrammetric models, such as the number and angles of photographs, technical parameters of camera configuration, and environmental factors such as lighting conditions. The results of this work also contribute to the refinement of imaging processes under various conditions and provide guidelines for optimizing the quality and usefulness of the generated models.

The quality of the 3D models generated by photogrammetry depends directly on the images captured during the inspection with drones. Therefore, and as part of the result of the methodology for the inspection by drones set forth in this work, good practices for the image acquisitio, have been defined in order to optimize this process:

• Extensive capture: Taking a large number of photographs is suitable, from different angles, heights, and positions, ensuring at least 65% overlap between successive images. It is preferable to move the drone rather than rotate it, to maintain consistency in image orientation.
• Quantity and detail: Although a greater number of images generates more detailed models, it is important to strike a balance between quality and quantity of images so as not to generate excessively large files.
• Camera settings: Optimizing camera settings is crucial:

  ○

  ISO: Keep values low (100–500) to reduce noise.

  ○

  Aperture: f/7 or higher, to focus on the whole element of interest.

  ○

  Speed: 1/200 or faster, to avoid blurred images.

  ○

  Format: RAW is recommended, although .jpg format is acceptable.

• Strategies according to the type of object: For surroundable objects (chimneys, wind turbines): Take shots from at least three heights and angles, avoiding turns greater than 15° between successive photos. For large, flat structures: Shoot horizontally and vertically, from different distances and positions.
• Lighting conditions: Minimize shadows and glare; outdoors. Best results are obtained on cloudy days or with the sun in a favorable position. Avoid the use of flash.

By addressing these technical considerations, the methodology ensures the generation of 3D models with sufficient detail for subsequent damage evaluation and comparison.

Based on the development of the geo-annotation system for images captured by the drone and the use of photogrammetry, an innovative solution for comparing damages using photogrammetric models has been developed. The implementation of the photogrammetry-based solution allows the automated generation of two main files on the drone during the image capture process: an image file and an associated metadata file. The image file, in standard format (.JPG), contains the visual capture of the inspected area and the observed damage, while the metadata file, in JSON format, stores additional relevant information for further analysis, such as geographic coordinates, capture angle, and camera parameters.

Once the drone completes the image capture, the files are automatically downloaded to the mobile device running the control application. This process ensures that each downloaded image is accompanied by a corresponding JSON file with the same name, making it easier to correlate images with their respective data. For example, an image captured with the name “DJI_0043.JPG” generates a metadata file called “DJI_0043.json” with the following data structure:

{

“datetime”: “25/03/2020 9:39”;

“gpslevel”: “1”;

“LAT”: 43.297144;

“LON”: −2.870884;

“ALT”: 0.6;

“ObstacleDistanceInMeters”: 0.5;

“WarningLevel”: 1;

“IsSensorBeingUsed”: true;

“DronePitch”: 0.5;

“DroneRoll”: 0.6;

“DroneYaw”: 0.2;

“GimballPitch”: 0.3;

“GimballRoll”: 0.1;

“GimballYaw”: 0.0

}

This process contributes to efficient data organization and facilitates the integration of results into an automated analysis system for damage evaluation. Subsequently, when the operator wants to register the inspection, the two files above-mentioned are attached to the observed pathology. The digitalized evaluation system stores the JSON information in the database, associating it with the respective pathology, and the image is stored on the server, keeping the reference in the database. The operator can then initiate the process to generate the photogrammetry. This process will perform all necessary calculations to generate the 3D model, and the result will be added as another data point in the repository. The input for this process will be the directory on the server where the images are stored. The photogrammetry will be generated and saved in the same directory, with this information being automatically added to the data model.

Regarding the tracking of photogrammetry and its utility, it is agreed that photogrammetry from different inspections carried out over time can be compared to observe the evolution of the pathology. This comparison is considered superior to that obtained from a single photograph, which does not provide as much data due to the differences in viewpoints.

The following images (Figure 5) are obtained with the developed solution, showing the same pathology at different times, to analyse its evolution over time, and how, by comparing the photogrammetric models, the difference between both is observed and can be quantified.

By comparing the 3D photogrammetry models obtained from the images captured by the drone during the pathology registration over time, the evolution of the pathology can be checked, and the difference of the surfaces affected quantified by pixel analysis (Figure 6). This confirms that the solution will allow such analysis to be conducted in the future for the building’s inspection, tending to a smart, objective, and quantifiable assessment of the damage, using techniques such as photogrammetry for measuring the evolution.

3.2. Standardized Construction Pathology Library

From the categorization of information carried out, as a partial result of this research the construction pathology library for industrial buildings has been achieved. Completing the information of the four levels in which the library schema has been structured, the construction pathology library has been defined with information on the different degradation processes associated with each of the defined construction elements, covering both their cause/origin and the resulting damage.

Once all the types of damage and defects making up the pathology library have been defined, Level 5 has been completed. This describes the procedure and repair work to address the damage and includes example images of each type of damage.

The result of the research is the construction pathology library that includes up to 300 different pathological processes to which the elements of industrial buildings are susceptible, together with the maintenance action associated and cost estimation, which will support the decision-making for the maintenance.

An extract of the library with the five levels is depicted below for illustrative purposes (

Table 5. Extract of the five levels of information from the construction pathology library.

LEVEL 1	LEVEL 2	LEVEL 3	LEVEL 4	LEVEL 5
CONSTRUCTIVE CATEGORY	CONSTRUCTIVE ELEMENT	CAUSE/ ORIGIN	PATHOLOGY	COST	REPAIR DESCRIPTION	IMAGE
External enclosures	Load-bearing masonry walls	Mechanical Damage	Cracks	1000 €/m	Restoration of cracks in ceramic brickwork for cladding, diagnosed by an approximate opening of 1 cm and an apparent depth of 1 foot. This process involves chipping the edges of the crack to fully expose it, demolishing the bricks of the first interior and exterior leaves located on either side of the crack, injecting epoxy mortar for filling, removing pieces for re-plastering, and constructing new brick leaves with solid ceramic bricks of 25 × 12 × 5 cm, similar to the existing ones, following CTE DB SE-F, DB SE, DB SE-AE, and NTE-FFL guidelines, with original bond, set with lime mortar at a 1/3 ratio to create interlocking, and absorbing the width of the crack. The process also includes re-planning, leveling, and plumbness checks, proportional plastering, handling of waste and debris, lifting and unloading equipment, work platform, piece wetting, debris removal, and cleaning
External enclosures	Load-bearing masonry walls	Mechanical damage	Fractures	50 €/m	Sealing of generalized cracks and fissures in brickwork, using lime mortar with a 1/2 ratio of natural color, including samples of finish, color, and texture to be selected. The existing mortar residue will be removed using pressurized air, after which prepared mortar will be injected using a gun to fill up to flush level, removing excess mortar, and cleaning the surface as the sealing is carried out
External enclosures	Load-bearing masonry walls	Mechanical Damage s	Blowups/ Deformations	100 €/m2	Demolition and reconstruction of walls. Demolition of solid brick walls with a thickness of one and a half feet using a compressor, including cleaning and removal of debris at the loading point, without transporting it to the landfill and with auxiliary equipment, without collective protection measures. The subsequent reconstruction will involve solid bricks of 24 × 11.5 × 8 cm, 1/2 foot thickness, set with CEM II/B-P 32.5 N cement mortar and M-5 river sand, prepared centrally and supplied on-site, for cladding, re-planning, leveling, plumbness checks, jointing, cleaning, and auxiliary means.
External enclosures	Load-bearing masonry walls	Moisture	Filtration	45 €/m2	Plastic paint in a color selected by the Technical Director, with a matte-satin finish, suitable for exterior or interior use in damp areas, with fungicidal and antibacterial additives. No solvents, great coverage, non-splashing, and resistant to wet rubbing according to DIN 53778. Prevents mold growth. On very porous surfaces, a coat of transparent, non-film-forming water-based primer will be applied first

).

3.3. Damage Assessment System for Buildings Based on Intervention Urgency Index (IUI)

As the main result of the research, the specific procedure for damage evaluation has been developed, particularly for industrial buildings, culminating in the classification of damage severity using the Intervention Urgency Index, referred to as IUI.

Following the proposed methodology and based on the results of the analysis using expert criteria, the weighting of the defined indicators has been obtained. After verifying that the pairwise comparison matrices created by the expert panel present adequate consistency, and following their numerical transformation, the weighting of the defined indicators was obtained, with the weights for each being recorded in

Table 6. Indicator weighting.

	Weight
SEVERITY	0.56
EVOLUTION	0.09
IMPACT ON OTHERS	0.28
EXTENSION	0.06

:

Once the weighting for each indicator was obtained, the value curves for each indicator were derived. As indicated in the methodology section, this process was also carried out using expert judgment, and different alternatives for each indicator were weighted, resulting in a value for each defined alternative as shown in Figure 7.

By transforming the obtained value functions into numerical values to generate evaluation algorithms that facilitate decision-making, the corresponding results for the different alternatives were calculated. The results from this process are presented in

Table 7. Weighting of alternatives for each indicator.

SEVERITY		EVOLUTION		IMPACT ON OTHERS		EXTENSION
VALUES		VALUES		VALUES		VALUES
Null	0	Null	0	Null	0	<5%	0
Low	0.26	Low	0.22	Low	0.28	5–25%	0.38
Medium	0.83	Medium	0.60	Medium	0.80	25–50%	0.71
High	1	Fast	1	High	1	>50%	1

.

Based on this information, following the methodology described, and depending on the quantitative evaluation of the detected damage on a scale from 0 to 1, the Intervention Urgency Index (IUI) can be obtained. The IUI (

Table 8. IUI Index Rating.

IUI Index	Rating
1	0–0.2
2	0.21–0.4
3	0.41–0.6
4	0.61–0.8
5	0.81–1

) is then classified, according to the damage’s criticality and the urgency of intervention, into 5 categories.

As a consequence, the analysis leads to a damage classification based on the Intervention Urgency Index (IUI) as shown in Figure 8:

4. Conclusions

A seamless data-based workflow for surveying and assessment intended to facilitate effective maintenance is drafted in the current research. The workflow is upgraded, taking leverage from new developments, based on inspection technologies and data analysis techniques as presented in the article. The developments addressed together with the process itself are promising for enhancing safety, quality, and efficiency in inspections and for an objective decision-making process.

This research is particularly meaningful for industrial buildings, due to the high potential of replicability because of the large number of those buildings which require upgrading, involving huge maintenance costs, but also because of the impact that the failure in some of the elements might have on industrial activity.

The digitalization of the inspection process with drones, the automated damage evaluation system based on a standardized construction pathology library, and the photogrammetry analysis of the pathologic processes, all addressed in this work, pave the way for streamlining the inspection and diagnosis process. The integration of those developments into a workflow can offer significant advantages in the maintenance of industrial buildings, such as the following:

• Process optimization: The use of a database that includes all possible types of damage associated with the various building components in an industrial facility will allow for the simple and clear registration of detected damage.
• Reduction in training time for inspectors: The extensive damage library allows inspectors to clearly and easily distinguish pathologies.
• Objectivity and transparency of results: The algorithms generated through expert knowledge eliminate the subjectivity that inspections were previously subject to, where each inspector could assign different values to a pathology.
• Facilitation of decision-making: The automation of the damage assessment process together with the generation of different economic scenarios of the intervention, streamline the decision-making, boosting an efficient maintenance management, which incorporates the cost as a new criteria for decision-making.

Solving the current barriers to automating the maintenance and inspection process: Combining different technologies to automate the maintenance process, from data capture to objective decision-making, allows for more efficient, higher quality, and less user-dependent continuous process. It must be highlighted that the use of the standardized construction pathology library and the evaluation system developed in the research leads to a reliable assessment of building condition, not dependent on the user’s subjectivity. That is to say, the inspector’s work will be strictly focused on identifying the type of damage in the library, evaluating its potential impact on other elements or people, and determining its extent. Then, the system will automatically evaluate the damage through weighting of different pathologies and the value functions assigned to each alternative. Those inputs for the algorithm based on multi-criteria analysis, give as a result an Intervention Urgency Index (IUI) of each instance of damage, which will be extremely valuable for prioritizing maintenance interventions. The outcome is, thus, an objective and reliable assessment system for effective maintenance management.

In addition, the procedure for inspection using drones investigated in this research plays a fundamental role in achieving proper results of inspection in an effective way. By establishing good practices for image acquisition, such as capturing images with adequate overlap, optimizing camera settings, and strategies adapted to the different types of objects to be inspected, as well as setting a geo-annotation system for the drone positioning, the methodology ensures the generation of high-quality 3D photogrammetric models. Emphasis on capturing data under optimal lighting conditions and maintaining consistency in image orientation further enhances the reliability of the models. These methodological advances not only improve the accuracy of inspections, but also facilitate the subsequent steps of damage assessment and decision-making, contributing significantly to the optimization of the maintenance workflow.

Moreover, the use of photogrammetry for 3D models’ generation from the images taken by drones is particularly promising for examination of defects, opening new insights for the automated analysis. Precisely, the integration of 3D photogrammetric models tackled in the research allows a detailed analysis of damage evolution over time. By capturing and comparing high-resolution images, it will be possible to generate three-dimensional models of the inspected areas and precisely track the progress of the pathologies detected in successive inspections. This approach not only complements the damage evaluation system, but also provides a dynamic visual representation of the deterioration of building elements, enabling a proactive and preventive management of damage rather than a reactive response.

5. Future Lines of Research

When it comes to the inspection by drones, it must be pointed out that nowadays the sector presents some limitations, mainly related to legal barriers as well as technical ones. Therefore, the forthcoming research in this field should advance together with progress in regulation about the restricted areas for flying to make the developments applicable in an extensive way. Moreover, the future lines of investigation for drone-based inspection should be focused on a higher automation of the process leading to an automated recognition of damage using machine learning and its evolution by using image detection models with VI-SLAMS applying Deep Learning technologies. Studies such as Dais et al. [38] highlight the growing importance of Deep Learning techniques and automated classification systems for damage recognition, such as crack detection. For that purpose, the images taken by drone must be captured with the required accuracy and quality, involving deep research in flight control algorithms to ensure the drone’s stability and insulate the sensors from external disturbances, improving the quality of performance; operational and asset condition data would significantly improve operational investment strategies.

When a sufficient amount of data becomes available in the future, new lines of research can even be explored, including machine learning-driven predictive models for the vulnerability assessment, using promising binary classification models for reliable predictions, as pointed out by Aloisio et al. [39]. Moreover, the application of advanced data analytics techniques, such as predictive analytics, machine learning, and artificial intelligence, potentially could improve the accuracy and efficiency of damage assessment and decision-making in the maintenance of industrial buildings to predict damage evolution.

A further step of the research is explorating the integration of machine learning algorithms for automated damage recognition and the development of 3D models using advanced image detection techniques and deep learning. These advancements will progressively automate the process, enabling a more autonomous and reliable system in the future. We firmly believe these future lines of research will lead to significant automation in the inspection process, further enhancing the system’s capabilities.

The process automation also involves the automated generation of a digital 3D model, preferably using BIM methodology, employing techniques such as photogrammetry and, in further steps, 3D Gaussian Splatting. The goal is to achieve a single centralized model to be managed collaboratively, which includes the captured information from the survey, duly positioned in the digital model. The implementation of Augmented Reality (AR) or Mixed Reality (MR) techniques combined with Deep Learning techniques can also help to the generation of 3D model elements and damages in augmented images. Thus, new techniques for automated generation of the 3D model of the existing building with semantic information about the pathologies must be also explored in the future for the visualisation and interpretation of the information.

On the other hand, regarding the assessment, the reliability of the diagnosis and prognosis of the state of conservation of the infrastructure can be upgraded considering not only the multi-criteria analysis addressed in the paper, but also Machine Learning (ML) solutions on captured data, if the amount of reliable data is available. Highly reliable algorithm for objective diagnosis fed with heterogeneous datasets in format (image, measurements…) and frequency (continuous real-time sensors, sporadic captures by drone…) can also provide the prediction capacity of the algorithm, based on ML techniques, to predict the evolution of pathological processes. Additionally, the inclusion of a broad approach to consequences (cost/benefit, cost of non-action, etc.) in the algorithm for assessment and prioritising maintenance actions could be explored for increasing the decision-making value, integrating the users’ requirements.

A further line of research for optimizing maintenance from inspection and diagnosis tasks could involve the integration of maintenance tasks derived from the construction pathology library and the automated assessment into a CMMS (Computerized Maintenance Management System). This will allow work orders to be defined and scheduled according to the prioritization achieved and will even allow for the planning of drone missions and organising the fleet for activating interventions.

Finally, all the advancements outlined in this research, along with the above-mentioned ambitions, could converge into a comprehensive digital tool. Such a tool would ensure the continuity of the inspection and diagnosis workflow, enabling stakeholders to reliably assess infrastructure conditions based on drone-captured data and prioritize interventions. By providing a data-driven decision-making system, this tool would facilitate effective maintenance management. Moreover, by incorporating into the tool the ability to analyse the temporal evolution of instances of damage through 3D photogrammetric models, this system would provide a more comprehensive and accurate approach to managing the lifecycle of industrial buildings. Such integration would allow for a holistic view of the state of the infrastructure, facilitating long-term decision-making and enabling the implementation of proactive strategies to maintain building performance. The result would be a robust tool offering all necessary functionalities for preventive maintenance while centrally managing information about the asset, its instances of damage, and their evolution, ultimately ensuring sustainable and effective infrastructure management.