Construction Delays Due to Weather in Cold Regions: A Two-Stage Structural Equation Modeling and Artificial Neural Network Approach

Abstract

Significant weather-induced delays often plague construction projects in India’s extremely cold regions, yet comprehensive studies addressing this issue remain scarce. This study aims to fill this gap by identifying key delay factors and proposing mitigation strategies for the construction industry. Through an extensive literature review, 42 delay factors were identified and categorized into four groups. A survey of 83 experts from cold regions was conducted to evaluate these factors’ significance to contractors and subcontractors. Employing exploratory factor analysis (EFA), structural equation modeling (SEM), and artificial neural networks (ANN), the study analyzed the relationships between these factors and ranked their impact. The findings reveal that snowfall, rainfall, and low temperatures are the most significant contributors to delays, with snowfall being the most influential (significance: 1.000), followed by rainfall (0.890) and low temperatures (0.790). This research establishes a risk hierarchy and develops a predictive model to facilitate the proactive scheduling of challenging tasks during favorable seasons. This study advances the understanding of weather-induced delays in India’s cold regions and offers valuable insights for project management in such climates. However, it underscores the importance of clearly articulating its novel contributions to differentiate it within the existing literature on weather-related construction delays.

1. Introduction

Construction projects are critical in national development, providing significant platforms for realizing essential dreams and goals [1,2]. Construction in India has grown dramatically in recent decades, becoming one of the country’s most vital economic engines [3]. Building construction is crucial in providing essential housing, water, and electricity services. As the population increases and the demand for these services grows, it becomes necessary to regularly upgrade the infrastructure that facilitates their delivery [4]. However, this vital industry is often plagued with delays; the expenses incurred by the construction industry due to schedule and cost overruns are estimated to exceed USD 80 billion [5]. Several sectors, including oil extraction and power generation, have experienced a 50% increase in schedule and cost overruns; on average, building projects have levels of 20% to 25% [6].

It is rare for a construction project to be finished within the initially determined time frame, even with innovative and experienced management, particularly in the construction sector globally. The major causal factors responsible for construction project delays (CPDs) include ineffective planning, financial problems, weather conditions, poor communication, material shortages, late payment, poor site management, construction labor shortages, equipment shortages, and the use of inexperienced workers [7,8,9]. It is crucial to note that these factors represent a subset of cases reported in the literature, and the contextualization of these challenges within specific project environments is essential for a comprehensive understanding. Though most causes of time and cost overruns have been investigated by researchers over the years and solutions have been proffered accordingly for their mitigation, the causes of overruns associated with weather or natural disasters remain underexplored.

Indeed, contractors and other project teams cannot control weather-related CPDs, negatively impacting project management and delivery [6]. Challenges from inclement and severe weather can be more difficult for some industries or countries to overcome than others, according to the corresponding climatic conditions. For instance, it is important to consider the diverse weather conditions that companies in various regions encounter. One such example can be found in India, where companies often encounter extreme weather conditions. During the winter season, certain regions of India, particularly in the northern parts, may experience snowfall and freezing temperatures, posing challenges for construction activities.

On the other hand, during the summer, these companies may face issues related to high temperatures, including the risk of fires and heat-related problems. By incorporating diverse illustrations of weather-induced predicaments faced by construction companies in various countries, the research problem can be more effectively underscored. This approach would enhance understanding of the broader impact and relevance of weather-related challenges in the construction industry [4].

The threats posed to the construction industry by weather are particularly related to workers’ health, well-being, and productivity at the project worksite [10]. Furthermore, the building sectors and infrastructure in various regions are underdeveloped, adversely affecting project cost, task productivity, and workers’ well-being, among other issues [11]. Within the realm of construction, cold regions present a unique set of challenges, including uncommonly poor weather conditions for most of the year, characterized by rainfall, snowfall, ground frost, and low-temperature differences [4,12,13]. This study focuses specifically on cold regions in India, encompassing parts of Jammu and Kashmir, Himachal Pradesh, Uttarakhand, and Sikkim. These regions provide distinct insights into the characteristics of cold environments and the challenges associated with delivering construction projects within the Indian context.

Conducting the study in India’s cold regions facilitates a localized and context-specific analysis of factors influencing construction projects in cold environments. In this context, “cold regions” refer to areas experiencing cold weather conditions, typically in high-altitude areas, northern latitudes, or locations with significant snowfall and freezing temperatures. Given the impact of cold regions weather-related factors on construction project delays, a comprehensive assessment of the influence of cold temperatures, snowfall, ice formation, and related elements on timely project delivery is crucial. Therefore, the following research questions (RQ) were addressed in this study:

• RQ1: What factors contribute to CPDs owing to weather conditions in cold regions?
• RQ2: What degree of association exists between these factors?

To address the research questions, this study establishes two primary objectives:

• RO1: To identify causal factors contributing to weather-related construction project delays in cold regions.
• RO2: To determine the most influential environmental factor affecting construction project delays in these regions.

This study contributes to the existing literature on construction project delays by offering a novel perspective on the challenges faced in cold-region construction, specifically within the Indian context. While previous studies have explored weather-related delays, limited research has focused on the distinct impacts of extreme cold conditions, such as snowfall, ground frost, and temperature fluctuations, on project timelines and workforce productivity. By integrating real-time data collection, predictive analytics, and adaptive mitigation strategies, this research provides a comprehensive framework for managing weather-induced delays. This predictive approach provides valuable insights for proactive scheduling during favorable seasons, aiding contractors and subcontractors in mitigating weather-related risks. Furthermore, it expands the global discourse on construction resilience by linking localized insights from India’s cold regions to broader industry challenges, offering valuable implications for policymakers, construction professionals, and researchers seeking to enhance project efficiency in extreme climates.

The remainder of this study is organized as follows. Section 2 provides contextual background by presenting the literature review and hypotheses underpinning this study. The methodology followed in this study is described in Section 3. The results are then discussed in Section 4, and their implications are presented in Section 5. Finally, the conclusions, limitations, and scope for future studies are provided in Section 6.

2. Contextual Background

CPDs represent a pervasive challenge in construction management, warranting a comprehensive exploration of their conceptual underpinnings [14]. Understanding the intricacies of project delays is fundamental to navigating the complexities of the construction industry. The concept of CPDs encompasses disruptions or extensions in the scheduled completion time of a construction project beyond the initially agreed-upon timeframe [15]. Such delays can result from many factors, spanning issues related to planning, execution, external influences, and unforeseen circumstances. To embark on a nuanced exploration of CPDs, it is imperative to elucidate the definitions, classifications, and overarching implications associated with construction project delays [16]. CPDs are a global concern, impacting around 70% of construction projects worldwide [17,18,19]. In the case of India, statistical evidence indicates that a substantial percentage of construction projects experience delays, approximately 65–75%. These delays not only lead to cost overruns but also contribute to operational inefficiencies, posing significant challenges to timely project completion. Understanding and quantifying these statistical values are crucial for implementing targeted strategies, improving project management practices, and enhancing the resilience of the construction industry, both globally and within the Indian context [20].

The literature extensively documents various causes of CPDs. Commonly recognized factors include weather and climate conditions, weak correspondence, a lack of cooperation among stakeholders, inadequate preparation, material shortcomings, funding issues, installment delays, equipment/plant deficits, a lack of participation among partner organizations, and inappropriate site management [17,18,19,21]. Various challenges, including corruption, economic constraints, payment delays, inaccurate project assessments, land acquisition issues, and government procedural delays, are prevalent in construction projects worldwide. However, in cold regions, these challenges can be further intensified by extreme weather conditions, which uniquely impact project timelines, labor productivity, and resource availability. Addressing these factors in the context of cold regions is crucial for developing effective mitigation strategies to enhance project delivery efficiency [22].

The identification of the top ten causes contributing to CPDs encompasses factors such as weather and climate conditions, weak correspondence, stakeholder conflicts, inadequate preparation, material shortcomings, funding issues, installment delays, equipment/plant deficits, lack of participation among partner organizations, and inappropriate site management [21,23]. However, this study’s focus on weather/climatic conditions prompts a critical examination. The rationale for selecting weather-related factors over other contributors remains inadequately justified in the problem development and literature review. The study will benefit from a more explicit articulation of why weather conditions were prioritized among the myriad factors influencing CPDs in the existing body of knowledge. Therefore, weather/climatic conditions constitute one of the most disruptive factors causing CPDs [21,23,24]. Indeed, cold climatic conditions have been identified as exerting tremendous influence on the advancement and nature of development projects [14]. Doloi et al. [18] observed that unfavorable weather defers 45% of development projects worldwide, incurring significant costs for project owners and workers-for-hire in terms of additional expenses and loss of income. Researchers have also recognized that precipitation, excessively low temperatures, and strong winds are the most significant climatic obstacles hindering the completion of construction projects. The adverse consequences of CPDs include accusations between property owners/clients and project staff over misrepresentations of costs, value, or earnings, which can culminate in agreement termination [25].

After identifying difficulties associated with managing CPDs resulting from weather conditions in cold regions, Sharma and Thakur [26] developed potential game plans for developing small hydropower projects in such areas using Jammu and Kashmir as an example. In addition to the challenges posed by the steep terrain in Kashmir, the climatic conditions are especially poor for most of the year as precipitation increases and snow covers the entire land, including the roads. Furthermore, Braimah and Ndekugri [27] sought to understand the impact of various climatic conditions on the efficiency of work assignments, determining that the most significant condition was precipitation, followed by strong winds and low temperatures.

Previous studies examining the impact of construction project delay factors have primarily employed traditional statistical techniques or individual modeling approaches, such as regression or correlation analysis [7,28]. However, few studies have utilized SEM specifically for this purpose. SEM offers several advantages, including simultaneously modeling complex relationships among latent constructs and observed variables. However, its application in the construction industry, particularly regarding the analysis of the factors influencing project delays, remains limited. Similarly, while some studies have developed models using ANN to evaluate construction project performance and reduce rework incidents, their application in analyzing construction project delay factors has been scarce. ANN is a machine-learning technique capable of capturing non-linear relationships and patterns in the data [29]. By training the network on historical data, ANN can effectively predict the likelihood and extent of delays based on given input variables.

In this study, we propose a novel SEM–ANN two-stage approach to address this gap in the literature. By combining SEM with ANN, we aim to leverage both techniques’ strengths in articulating a profound understanding of the severity of these weather-related causal factors of CPD in cold regions. Furthermore, the SEM–ANN approach allows for a more comprehensive analysis of the factors affecting construction project delays by incorporating the complex interplay between variables, capturing non-linear relationships, and making predictions based on patterns in the data. By adopting this innovative methodology, our study offers a novel and rigorous approach to examining the impact of construction project delay factors, contributing to the advancement of research in the field.

Hypotheses and Model Construction

In the research examined and outlined in Table 1, approximately 30% of the projects analyzed experienced delays, with an average time overrun of 23.2%. This percentage was derived from studies that specifically reported project delays in cold regions, where the delay durations were calculated as a percentage of the total project timeline. It is important to note that the 30% delay rate is specific to projects affected by extreme cold-weather conditions, whereas the 65–75% delay rate for Indian construction projects and the 70% global delay rate include a broader range of delay factors, such as financial constraints, regulatory hurdles, and workforce challenges. The classification of cold-weather factors—temperature (T), rainfall (R), and snowfall (S)—was derived from primary meteorological variables significantly impacting cold-weather construction performance. These factors were strategically chosen to encompass the significant climatic elements affecting construction projects in cold regions. An elaboration of the rationale for this categorization is detailed below:

• Temperature (T): Recognized as a critical factor, temperature directly influences the freezing and thawing of construction materials, equipment operation, and laborer working conditions [30,31,32]. Cold temperatures can result in reduced productivity, prolonged concrete curing times, challenges in excavating frozen ground, and difficulties in maintaining optimal temperatures for specific construction processes [33]. By isolating temperature as a distinct factor, we aim to address its unique impact on cold-weather construction.
• Rainfall (R): Another crucial factor affecting construction activities in cold regions is that rainfall can lead to wet and slippery conditions, hindering excavation earthmoving operations and causing erosion or slope instability [34]. When combined with low temperatures, rainfall can contribute to ice formation, intensifying construction challenges [12]. By considering rainfall as a separate factor, we emphasize its specific implications for cold-weather construction performance.
• Snowfall (S): A distinct characteristic of cold regions, snowfall significantly impacts construction projects by obstructing transportation routes, covering sites, impeding access to work areas, and complicating snow removal operations [4]. The presence of snow can also affect construction schedules and elevate the risk of accidents [8]. Categorizing snowfall as a separate factor enables us to address the unique challenges associated with snow in cold-weather construction.

Table 1. Weather-related factors that delay construction projects in cold regions.

Delay Effects	Code	Observed Factors	References
Low Temperature	T1	Choice of site location	[35,36,37]
	T2	Identification of project needs
	T3	Delays in transportation
	T4	Problems with equipment/machinery
	T5	Delays in the overall project *
	T6	Price escalation
	T7	Unsafe working conditions
	T8	Effects on concrete operations
	T9	Hiring and recruiting
	T10	Selection of construction materials
	T11	Structural damage
	T12	Risks to planning and designing methods
Rainfall	R1	Excavation and earthwork	[17,35,38,39,40]
	R2	High volumes of insurance
	R3	Delays in the overall project
	R4	Effects on the use of cranes/towers
	R5	Selection of construction materials
	R6	Account differences
	R7	Effects on elevation and topography
	R8	Structural damage
	R9	Delay in transportation
	R10	Storage of construction materials
	R11	Unsafe working conditio
	R12	Delays in handing over the project to the client
	R13	Including/excluding project groups
Snowfall	S1	Disputes among various groups	[8,18,41,42,43]
	S2	Storage of construction materials/equipment
	S3	Effects on elevation and topography
	S4	Delays in handling
	S5	Structural damage
	S6	Unsafe working conditions *
	S7	Effect on the choice of site location
	S8	Selection of construction materials
	S9	Additional maintenance work *
	S10	Excavation/earthwork difficulties
	S11	Complete suspension of work
	S12	Disasters
	S13	Selecting team members
Delay effects	D1	Time overruns	[19,44,45]
	D2	Cost overruns
	D3	Worker well-being
	D4	Task productivity

Note: * denotes that the respective factors are incorporated as per the experts’ opinions.

Table 1. Weather-related factors that delay construction projects in cold regions.

Delay Effects	Code	Observed Factors	References
Low Temperature	T1	Choice of site location	[35,36,37]
	T2	Identification of project needs
	T3	Delays in transportation
	T4	Problems with equipment/machinery
	T5	Delays in the overall project *
	T6	Price escalation
	T7	Unsafe working conditions
	T8	Effects on concrete operations
	T9	Hiring and recruiting
	T10	Selection of construction materials
	T11	Structural damage
	T12	Risks to planning and designing methods
Rainfall	R1	Excavation and earthwork	[17,35,38,39,40]
	R2	High volumes of insurance
	R3	Delays in the overall project
	R4	Effects on the use of cranes/towers
	R5	Selection of construction materials
	R6	Account differences
	R7	Effects on elevation and topography
	R8	Structural damage
	R9	Delay in transportation
	R10	Storage of construction materials
	R11	Unsafe working conditio
	R12	Delays in handing over the project to the client
	R13	Including/excluding project groups
Snowfall	S1	Disputes among various groups	[8,18,41,42,43]
	S2	Storage of construction materials/equipment
	S3	Effects on elevation and topography
	S4	Delays in handling
	S5	Structural damage
	S6	Unsafe working conditions *
	S7	Effect on the choice of site location
	S8	Selection of construction materials
	S9	Additional maintenance work *
	S10	Excavation/earthwork difficulties
	S11	Complete suspension of work
	S12	Disasters
	S13	Selecting team members
Delay effects	D1	Time overruns	[19,44,45]
	D2	Cost overruns
	D3	Worker well-being
	D4	Task productivity

Note: * denotes that the respective factors are incorporated as per the experts’ opinions.

The classification of observed factors under specific latent factors was based on their dominant environmental influence rather than overlapping multiple factors. For example, snowfall was prioritized over rainfall for excavation difficulties because frozen ground and snow accumulation present greater obstacles to excavation work than rainfall alone. Conversely, rainfall was linked to crane operations, since wet and slippery conditions significantly impact crane stability and safety, whereas snowfall’s impact on cranes is less prominent in the existing literature and expert evaluations. The various cold-weather factors contributing to cold-region C.P.D.s are classified into three types—(1) temperature (T), (2) rainfall (R), and (3) snowfall (S)—to introduce hypotheses and construct an examination model. Direct relationships among snowfall, rainfall, temperature, and CPDs are evident in the first-hand data, as depicted in Figure 1, illustrating potential connections between latent and observed factors. Table 1 clarifies how observed factors are systematically classified under latent factors. This clarification aims to enhance understanding and address potential ambiguities in the relationship between observed and latent factors. The hypotheses proposed to evaluate these connections are as follows:

H1. 

Snowfall and CPDs are positively correlated.

H2. 

Rainfall and CPDs are positively correlated.

H3. 

Temperature and CPDs are positively correlated.

Figure 1. Factors and hypotheses examined in this study.

Figure 1. Factors and hypotheses examined in this study.

Moreover, in cold regions, these latent factors and their critical related effects are hypothesized to impact the ability to achieve significant levels of construction work without obstacles. The examination model (Figure 2) considers both latent and observed factors from Table 1, indicating the directions in which observed factors affect latent factors. This construction aims to provide a comprehensive framework for understanding the interplay between observed and latent factors in the context of cold-weather construction delays.

3. Research Methodology

This study was divided into three stages, as shown in Figure 3: (1) the identification of factors contributing to construction project delays due to weather conditions (CPDDWCs), (2) data collection, and (3) data analysis. The philosophical stance of this study is rooted in pragmatism, emphasizing the practical combination of qualitative and quantitative methods. The chosen research approach is a mixed-methods design, employing a concurrent triangulation design to comprehensively explore weather-induced construction delays. The research strategy, a sequential explanatory strategy, involves the sequential collection and analysis of quantitative data through surveys and SEM, followed by qualitative data collection through interviews and analysis using ANN. This strategy contributes to a more nuanced and comprehensive understanding of the research questions. The research questions align with the pragmatic philosophy, mixed-methods approach, and sequential explanatory strategy, ensuring a robust investigation into weather-induced construction delays.

3.1. Stage 1: Identification of Factors Leading to CPDDWCs

Based on the literature review, 35 weather-related factors that cause CPDDWCs in cold climates were identified. These observed factors and their sources are listed in Table 1. To confirm their relevance to Indian construction scenarios, the list was discussed with ten experts with more than 25 years of experience in the construction industry. Three additional factors—delays in the overall project, unsafe working conditions, and other maintenance works, indicated by asterisks in Table 1—were additionally incorporated per the experts’ opinions, resulting in 38 factors.

The study focuses on cold regions of India, such as Ladakh, Himachal Pradesh, and Uttarakhand, which experience extreme winter conditions. According to meteorological data, average winter temperatures in these regions range from −20 °C to −5 °C, with some areas experiencing even lower temperatures, particularly in high-altitude zones. Snowfall can exceed 10 feet in peak winter months, leading to disruptions in transportation and construction activities. Additionally, precipitation levels vary, with regions such as Uttarakhand receiving 1500–2000 mm of annual rainfall, whereas Ladakh remains relatively arid with less than 100 mm of annual precipitation, yet it faces challenges due to freezing conditions. These climatic conditions significantly impact construction activities, causing delays, material degradation, and safety risks. Including these meteorological data allows for a more comprehensive comparison with other regions worldwide that experience similar cold-climate-related challenges, ensuring broader applicability of the study findings.

Furthermore, comparative data on project delays in various regions of India highlight the disproportionate impact of cold-weather conditions. In prominent cold regions such as Himachal Pradesh and Ladakh, delays due to extreme temperatures, snowfall, and logistical challenges account for approximately 30–40% of the total project duration. By contrast, in relatively warm regions such as Maharashtra or Tamil Nadu, project delays due to weather factors are significantly lower, typically ranging from 10–15% of the total project timeline, which is primarily attributed to monsoons rather than prolonged freezing conditions. These data underscore the additional effect of snow, rain, and low temperatures, reinforcing the necessity of specialized strategies in mitigating cold-weather-induced delays in construction projects.

Additionally, the study examines the allocation of responsibility for weather-induced delays. In construction contracts, force majeure clauses generally exempt all parties from liability for extreme and unforeseeable weather conditions. However, if delays could have been mitigated through proactive scheduling, contingency planning, or resource allocation, partial responsibility may fall on project managers, contractors, or government agencies. Clearly defining accountability for such delays is essential for improving risk management practices and ensuring realistic project timelines in cold-climate construction.

3.2. Stage 2: Data Collection

An extensive review of the existing literature was conducted to identify potential causes of CPDDWCs. However, definitive conclusions could not be easily drawn from the few studies considering climatic conditions as a cause of CPDs. This difficulty culminated in using interviews to elicit expert opinions concerning the importance of these factors within the study environment, as expounded in [46,47]. Indeed, expert opinions are often sought in construction management, and questionnaires are typically used to gather random opinion data systematically [48]. For this study, the selection criteria for experts were based on multiple demographic and professional factors. Experts were chosen to ensure diversity in gender (male and female), sector (public and private), and professional experience (<5 years, 5–10 years, 11–15 years, and >15 years). Additionally, their designation within the industry was considered, including roles such as project managers, engineers, contractors, and other relevant positions.

A structured questionnaire was developed for data collection, consisting of two sections. The first section gathered demographic details of the experts, including gender (male/female), sector (public/private), years of experience (<5 years, 5–10 years, 11–15 years, >15 years), and designation (project manager, engineer, contractor, or other). The second section focused on assessing the effects of various delay factors related to construction projects in cold regions. Experts were asked to rate the influence of these factors, identified in Table 1, using a seven-point Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). The demographic details of the experts, as shown in

Table 2. Summary of respondent profiles.

Category	Subcategory	Total	Respondents %
Gender	Male	69	83.13
	Female	14	16.86
Sector	Public	35	39.75
	Private	48	55.42
Experience (in years)	<5	9	10.84
	5–10	15	18.07
	11–15	45	54.21
	>15	14	16.86
Designation	Project Manager	36	43.37
	Engineer	22	26.50
	Contractor	18	21.68
	Other	7	8.43

, indicate that the majority of respondents were male (83.13%) and from the private sector (55.42%). Most experts had significant experience, with 71% having more than 11 years in the industry, and 70% holding managerial positions, including project managers (43.37%), engineers (26.50%), and contractors (21.68%).

In the initial phase, a pilot study engaged twelve construction project experts and professionals, each possessing over 15 years of experience. Subsequently, potential respondents for the full survey were identified from construction-related databases, specifically focusing on organizations involved in Indian infrastructure projects within cold regions. Invitations for participation were extended to more than 150 eligible experts via email and phone. The respondents’ familiarity with infrastructure projects in the cold areas was pivotal for the quality and validity of their responses.

To ensure that variations in microclimatic conditions within the cold regions were adequately captured, the study incorporated responses from professionals working across multiple locations with different topographical and climatic conditions. These regions exhibit distinct variations in temperature fluctuations, localized snowfall intensities, and wind conditions. This approach allowed the model to integrate diverse climatic influences affecting CPDs. Additionally, meteorological data from local weather stations were referenced to cross-validate expert responses and enhance the robustness of the analysis.

Despite the general recommendation of a sample size of 50 or more for structural equation modeling (SEM) by [5,49,50], acknowledgment of the nuanced nature of sample size determination in SEM analyses is essential. The appropriateness of the chosen sample size is contingent on various factors, including the complexity of the model, the number of latent variables, and the expected effect sizes. Recognizing the variability in these factors, we aimed to capture a representative sample of expertise within the given parameters.

Out of the 150 responses received, 83 surveys were completed, resulting in a response rate of 64.6%, considered acceptable within research standards. Simultaneously, 35% of responses were deemed invalid or insufficient in some manner. The reliability of the survey findings is underpinned by the extensive work experience of respondents, with 71% possessing more than 11 years of experience and 70% occupying high managerial positions in their organizations, as detailed in Table 2. It is worth noting that prior studies, such as those conducted by [27,51,52,53,54], have successfully employed varying sample sizes to analyze quantitative data using SEM. The absence of prescriptive requirements for sample size relative to observed parameters in SEM analyses is acknowledged [55].

3.3. Stage 3: Data Analysis

3.3.1. SEM

The SEM approach was chosen because it can determine the sequence of interconnections between latent or independent variables to address the objective of this study. The variables used to examine and visualize a structured model include delay factors and their relationships [56]. Indicators can subsequently be analyzed to determine their theoretical foundations and the degree to which they are correlated. As a result, SEM is increasingly used by marketing and management experts to examine the causal relationships among latent features [57]. As an extension of the standard regression model used for poorly measured independent variables, SEM may benefit engineers and management experts in the construction sector [54]. The combination of structural equation and measurement models produces a system in which the former considers all interactions between latent variables and their corresponding observed variables, while the latter only addresses the correlations among observed variables [58].

As suggested by [59], SEM was used in this study to evaluate and quantify the direct and relative impacts of latent variables (i.e., temperature, rainfall, and snowfall) on observable phenomena (i.e., CPDs). The essential weather conditions that drive CPDs in cold regions were subsequently clarified by observing the relationships between the three latent factors and their underlying characteristics (observed factors).

3.3.2. ANN

This study employs a multi-explanatory approach, combining PLS-based SEM with neural network analysis for robust results. Neural networks, a powerful tool for forecasting complex real processes and uncertain phenomena, stand out in artificial intelligence applications. They excel in identifying non-linear relationships, contrasting with simpler methods such as SEM or multiple regression analysis [29]. Modeled after the human brain, ANN handles complexity and non-linearity adeptly, making it ideal for predicting construction project delays. ANN excels in capturing intricate factor relationships through data training, offering valuable insights into the complex interplay of variables [60]. This involves the dataset’s network learning patterns and connections, contributing to a deeper understanding of the relationships among factors influencing construction project delays.

Artificial neural network (ANN) analysis enhances understanding, complementing PLS-SEM results and handling non-linear challenges [55]. Its accuracy surpasses SEM and regression methods. The synergy of PLS–SEM and ANN analyses strengthens findings, reflecting the brain’s intricate information processing [61]. ANN encompasses learning rules, transfer functions, and network architecture, including subcategories such as feed-forward multi-layer perceptron (MLP) [62,63]. Among these, the widely adopted feed-forward MLP, characterized by input, output, and hidden layers, prevails among researchers [64]. This fusion of technology and cognition yields novel insights, deepening comprehension of complex phenomena. After completing the PLS-based SEM analysis, the subsequent phase involved the development of an artificial neural network (ANN) model. The salient factors identified through the SEM analysis were incorporated as input neurons for the ANN, representing a purposeful integration of both methodologies [65]. While PLS-based SEM contributed valuable confirmatory insights, the ANN was crucial in addressing specific challenges, such as non-normal data distributions and intricate non-linear relationships between exogenous and endogenous variables. The concurrent application of PLS–SEM and ANN in this research was carefully orchestrated to harness the synergistic strengths of each method, thereby elevating the analytical robustness. This methodological fusion facilitated a comprehensive exploration of intricate relationships within the dataset, providing nuanced insights into the complex phenomena under investigation.

4. Results

4.1. Exploratory Factor Analysis

An exploratory factor analysis (EFA) was conducted to analyze the corresponding variables and reduce the quantity of data in the climatic CPD factor model. EFA is a statistical technique to uncover the underlying factor structure within a set of observed variables. In the context of this study, EFA can help identify and extract the latent factors or constructs that contribute to construction project delays. By examining the relationships between variables, EFA provides insights into the underlying dimensions or factors influencing the delays, thereby aiding in a more comprehensive understanding of the factors at play. Rotating the primary component matrix is a typical practice employed to facilitate the comprehension of components. The varimax rotation method provided in the SPSS software package version 22 was therefore used to rotate the solutions of the primary components. The variables (factors) were turned to extend the squared factor loadings and clarify their meanings. Weak indicators were excluded from these components owing to the small sample size; factors with loadings below 0.40 (the significance cut-off) were discarded [66]. Three problematic elements were deleted from the list based on their loadings: the risks to planning and designing methods (T12), selecting team members (S13), and including/excluding project groups (R13).

Table 3. Total variances of the measured factors.

Component	Initial Eigenvalues			Rotated Sums of Squared Loadings
	Total	Variance %	Cumulative %	Total	Variance %	Cumulative %
1	15.136	38.809	38.809	4.420	11.333	11.333
2	3.939	10.099	48.908	4.364	11.190	22.524
3	2.044	5.241	54.149	4.178	10.712	33.236
38	0.000	0.001	100	-	-	-

shows the eigenvalues for a selection of components. Figure 4 depicts the scatter plots for all components, representing the eigenvalue distribution of each. First, the dataset was divided into 38 components according to the relevant factors. The eigenvalues, which quantify the variation in each element, were used to express the variance of the delay. The results indicate that a substantial amount of variance could be described by just three components [67]. In summary, 35 observed factors (selected according to the hypothetical model) exhibited meaningful relationships with the three latent factors (temperature, rainfall, and snowfall), as shown in

Table 4. Factor loadings and KMO values extracted from the EFA.

Latent Factor	Observed Factor	Factor Loadings	Rank	KMO Value
Temperature	Choice of site location	0.59	25	0.783
	Identification of project needs	0.565	27
	Delays in transportation	0.714	41
	Problems with equipment/machinery	0.596	24
	Delays in the overall project	0.756	5
	Price escalation	0.637	20
	Unsafe working conditions	0.641	21
	Effects on concrete operations	0.769	2
	Hiring and recruiting	0.523	32
	Selection of construction materials	0.747	10
	Structural damage	0.692	19
Rainfall	Excavation and earthwork	0.791	8	0.736
	High volumes of insurance	0.578	26
	Delays in the overall project	0.763	3
	Effects on the use of cranes/towers	0.477	30
	Selection of construction materials	0.704	13
	Account differences	0.682	28
	Effects on elevation and topography	0.759	4
	Structural damage	0.733	7
	Delay in transportation	0.7	15
	Storage of construction materials	0.807	1
	Unsafe working conditions	0.738	6
	Delays in handing over the project to the client	0.669	17
Snowfall	Disputes among various groups	0.486	31	0.779
	Storage of construction materials/equipment	0.665	18
	Effects on elevation and topography	0.624	29
	Delays in handling	0.627	23
	Structural damage	0.646	19
	Unsafe working conditions	0.713	12
	Effects on the choice of site location	0.639	22
	Selection of construction materials	0.695	14
	Additional maintenance work	0.681	16
	Excavation/earthwork difficulties	0.728	9
	Complete suspension of work	0.72	11
	Disasters	0.767	11

. The table lists all 35 factors along with their respective factor loadings (>0.4), rankings, and Kaiser–Meyer–Olkin (KMO) values. The KMO sample adequacy score was used to assess the suitability of factor analysis for this dataset. While KMO values for latent factors were explicitly calculated (temperature: 0.783, rainfall: 0.736, snowfall: 0.779), additional KMO values for individual observed factors were also included where available. Previous recommendations suggest that factors with a KMO value below 0.5 should be omitted from analysis [68]. The inclusion of these KMO values ensures transparency in the selection process and confirms that factor analysis is appropriate for identifying the key effects of cold-weather conditions on construction projects. Among these, the top five risk factors with the highest factor loadings were identified as the most critical in influencing construction project delays in cold regions. Storage of construction materials (0.807) emerged as the most significant factor, emphasizing the impact of logistical challenges in ensuring timely access to materials. This was followed by effects on concrete operations (0.769), highlighting the adverse influence of cold weather on concrete setting and overall structural integrity. Delays in the overall project (0.763) ranked third, reflecting the cumulative effect of multiple weather-related disruptions. Effects on elevation and topography (0.759) were also significant, as changes in terrain and site conditions posed substantial risks to project progress. Lastly, unsafe working conditions (0.738) were identified as a critical factor, reinforcing the role of extreme weather in increasing safety hazards and, consequently, causing project delays due to workforce disruptions.

4.2. Structural Equation Modeling

SEM was applied to fulfill the study’s second objective, examining hypothesized relationships among factors affecting construction project delays. By modeling these relationships, SEM assesses direct and indirect effects, enhancing our comprehension of the dynamics at play. Researchers adopted Variance-Based Structural Equational Modeling (VB-SEM), particularly PLS-SEM, to delve into the intricacies of their research model. PLS-SEM outshines Covariance-Based SEM (CB-SEM) with its ability to handle complex higher-order models, perfectly aligning with this study’s three higher-order variables. It is known for its precision in revealing the complete model, as confirmed by Jafar et al. [69].

The PLS analysis merged two models: one assessed through factor analysis and another via path analysis, an integration unique to SEM, surpassing multiple regression [70]. Additionally, Cronbach’s alpha measured internal consistency after SEM classification and labeling were conducted using SmartPLS 3 [71]. SEM’s analytical prowess proved indispensable in this transformative research journey, concisely and robustly facilitating a deeper understanding of the proposed model’s intricacies [69].

As the fitness indices for the initial baseline model failed to meet the standards (

Table 5. Model fitness indices.

Fitness Index	Recommended Level	Baseline Model	Final Model
Chi-squared	Less than 2	2.76	1.97
p value	Less than 0.05	0	0
RMSEA	Less than 0.08	0.134	0.072
CFI	Greater than 0.9	0.844	0.915
SRMR	Less than 0.08	0.105	0.078

), data with a lower R² were discarded. Indeed, Beatriz et al. [72] suggested that deleting links and factors with extremely low correlations to their latent component might refine the model. The baseline model was therefore examined using SmartPLS, as shown in Figure 5. As a result, the first three versions of the framework had S13, R13, and T12 removed. All five model fitness indices—the Chi-squared distribution, p-value, root mean square error of approximation (RMSEA), comparative fit index (CFI), and standardized root mean square residual (SRMR)—improved significantly after the removal of these factors, with the model fitness improving with each iteration. Indeed, the obtained index values indicated that the model exhibited significant statistical fitness with a 95% confidence level.

As the PLS-based SEM represents an effective tool for modeling adoption behavior [73] and can resolve issues associated with non-normality and small datasets, it was considered a good fit for this study. A Cronbach’s alpha coefficient of greater than 0.7 represents internal consistency when applied to a wide range of latent factors, as depicted in

Table 6. Latent factors with their validity and reliability statistics.

Latent Factor	Cronbach’s Alpha	CR	AVE
Temperature	0.864	0.889	0.425
Rainfall	0.906	0.920	0.496
Snowfall	0.880	0.901	0.438

[74]. The composite reliability (CR) was also used to evaluate internal consistency; the results presented in the table show that all values were larger than the 0.7 cut-off value [67]. Finally, all average variance extracted (AVE) values were greater than the cut-off of 0.04 [75], ensuring convergence validity. The SEM shown in Figure 5 represents the final result of all modifications.

4.3. Artificial Neural Network

Only significant factors (snowfall, rainfall, temperature) from SEM were inputs for the one-hidden-layer ANN architecture [64]. The sigmoid function was the activation function for both layers, with neuron values constrained within [0, 1] for optimization [76]. Multilayer perceptrons were chosen as the suitable neural system [77]. A two-stage approach was applied: SEM determined factor significance, guiding their importance assessment in predicting CPDs [78].

The final SEM results in Figure 6 and the path estimates in

Table 7. Hypothesis testing.

Path	β	Significance (p)	Result
Snowfall → CPDs	0.576	0.000	Supported
Rainfall → CPDs	0.216	0.000	Supported
Temperature → CPDs	0.096	0.004	Supported

show that the results obtained for the snowfall, rainfall, and low-temperature factors verified hypotheses H1, H2, and H3. Snowfall, with a β value of 0.576 (p = 0.000), was the primary contributor to CPDs in cold regions, followed by rainfall with a β value of 0.216 (p = 0.000). Although low temperature showed a weak correlation with CPDs (β = 0.096, p = 0.004), its statistical significance was evident in the model dimensions.

The final SEM standardized path coefficients, R² multiple correlation values, and indices of path significance (t-values) for weather-related CPDs in cold regions are listed in

Table 8. Standardized path coefficients, R² values, and t-values of the observed variables.

Path	Standardized Path Coefficient	R2	t-Value
D1 ← CPDs	0.77	0.60	11.231
D2 ← CPDs	0.73	0.54	9.819
D3 ← CPDs	0.67	0.45	6.605
D4 ← CPDs	0.79	0.63	12.716
R1 ← Rainfall	0.82	0.67	23.336
R10 ← Rainfall	0.84	0.70	24.573
R11 ← Rainfall	0.69	0.48	7.893
R12 ← Rainfall	0.69	0.48	5.376
R2 ← Rainfall	0.54	0.30	5.239
R3 ← Rainfall	0.78	0.61	13.364
R4 ← Rainfall	0.46	0.21	4.331
R5 ← Rainfall	0.65	0.43	6.603
R6 ← Rainfall	0.69	0.48	6.843
R7 ← Rainfall	0.78	0.61	19.318
R8 ← Rainfall	0.69	0.47	7.426
R9 ← Rainfall	0.72	0.52	5.749
S1 ← Snowfall	0.36	0.13	3.174
S10 ← Snowfall	0.72	0.52	7.619
S11 ← Snowfall	0.74	0.55	10.645
S12 ← Snowfall	0.78	0.60	10.029
S2 ← Snowfall	0.67	0.45	8.319
S3 ← Snowfall	0.61	0.37	5.172
S4 ← Snowfall	0.61	0.37	6.077
S5 ← Snowfall	0.64	0.40	4.856
S6 ← Snowfall	0.72	0.52	6.985
S7 ← Snowfall	0.65	0.42	5.692
S8 ← Snowfall	0.69	0.48	10.177
S9 ← Snowfall	0.67	0.45	8.074
T1 ← Temperature	0.53	0.28	4.474
T10 ← Temperature	0.75	0.57	11.616
T11 ← Temperature	0.72	0.52	15.466
T2 ← Temperature	0.51	0.26	6.007
T3 ← Temperature	0.69	0.47	8.742
T4 ← Temperature	0.63	0.40	7.899
T5 ← Temperature	0.75	0.56	12.501
T6 ← Temperature	0.59	0.34	5.373
T7 ← Temperature	0.65	0.42	8.003
T8 ← Temperature	0.75	0.56	11.457
T9 ← Temperature	0.55	0.31	5.682

Note: Significant factors at the p < 0.05 probability level.

, which reports positive and statistically significant outcomes (p < 0.05). To analyze the association between the observed and latent factors, the t-value was compared with the error level. All relationships were validated as their t-values were larger than the minimum value (1.64) recommended by Ma et al. [79].

The SEM and factor analysis results reveal that weather delays substantially impact time overruns, cost overruns, worker well-being, and task productivity. An ANN analysis conducted with SPSS 26 employed the three significant SEM factors (snowfall, rainfall, and temperature) as input layers and climatic delays as the output layer (see in Figure 7). To avoid overfitting, 90% of the dataset was used for training and 10% for testing, ensuring model accuracy, as measured by R² [3].

The root mean square errors (RMSEs) and mean square errors (MSEs) for both the training and testing data were computed for all ten neural networks, confirming an excellent model fit (average RMSE 1.715 for training and 1.597 for testing), as shown in

Table 9. RMSE and MSE of evaluated neural networks.

Neural Network	RMSE		MSE
	Training	Testing	Training	Testing
ANN 1	1.653	1.495	2.732	2.235
ANN 2	1.806	1.538	3.262	2.365
ANN 3	1.814	1.616	3.291	2.611
ANN 4	1.819	1.721	3.309	2.962
ANN 5	1.496	1.341	2.238	1.798
ANN 6	1.668	1.687	2.782	2.846
ANN 7	1.810	1.675	3.276	2.806
ANN 8	1.753	1.692	3.073	2.863
ANN 9	1.557	1.500	2.424	2.250
ANN 10	1.770	1.708	1.330	2.917
Mean	1.715	1.597	2.772	2.565
Standard deviation	0.116	0.124	0.635	0.387

. The R² value demonstrated the ANN’s accuracy in predicting CPDs, achieving a 72% accuracy rate [61]. Furthermore, the importance of each normalized variable was determined, indicating that snowfall, rainfall, and temperature are significant predictors with normalized importance values of 1.000, 0.890, and 0.790, respectively, underscoring their relevance in forecasting construction project delays (see

Table 10. Normalized variable importance.

Factors	Normalized Importance
Snowfall factor	1.000
Rainfall factor	0.890
Temperature factor	0.790

).

5. Discussion

Construction projects in cold regions face unique challenges due to adverse weather conditions, including snowfall, rainfall, and low temperatures. This study’s findings shed light on significant factors contributing to cold-region CPDs and propose effective mitigation strategies. Snowfall is a primary driver of delays, leading to work suspension and reduced productivity [8,12]. Proactive planning, incorporating weather forecasts, is essential to anticipating snowfall periods, strategically scheduling activities, and allocating resources, thereby minimizing potential delays [8].

Similarly, rainfall, the second most influential factor, impacts operations, site selection, and construction progress [4,80]. Considering rainfall patterns in project scheduling, strategic planning effectively aligns construction activities with anticipated weather conditions, reducing the likelihood of disruptions. Additionally, material procurement based on weather forecasts ensures timely availability, preventing delays caused by unexpected rainfall [8]. Training programs on weather conditions enhance personnel adaptability, fostering continued work in changing weather conditions [81].

Low temperatures, while weakly correlated with CPDs, consistently affect labor productivity, risks, structural integrity, and project costs [82]. Risk assessment, structural considerations, and personnel safety measures are crucial in mitigating low-temperature challenges [83]. Thorough risk assessments identify potential issues, allowing the development of proactive strategies. Designing structures with low temperatures in mind, including material choices and insulation, helps minimize the impact of cold weather. Implementing safety measures, such as appropriate clothing and training, ensures personnel safety in low-temperature conditions [84].

The study’s emphasis on proactive planning, construction techniques tailored to weather conditions, and strategic resource allocation provides actionable insights for overcoming delays caused by snowfall. Similarly, strategic planning, material procurement based on weather forecasts, and training programs effectively address challenges posed by rainfall [8]. The subtle yet consistent impact of low temperatures underscores the importance of risk assessment, structural considerations, and personnel safety measures in managing construction projects in cold regions [8,33,85].

Prioritizing weather-related factors over financial constraints or material shortages in this analysis is justified by their direct and uncontrollable impact on construction timelines, safety, and structural integrity, particularly in cold-region projects [86]. Unlike financial constraints, which can be mitigated through phased budgeting or alternative funding sources [87], and material shortages, which can be managed through strategic procurement and inventory management [35], adverse weather conditions pose immediate and unavoidable risks that disrupt project schedules, reduce labor productivity, and escalate costs due to delays [18]. Furthermore, in remote or extreme climatic conditions, weather-related disruptions often exacerbate other challenges, such as transportation difficulties and supply chain breakdowns, amplifying their overall impact on project feasibility [41]. Thus, addressing these factors first ensures a more resilient and adaptive construction approach, minimizing cascading effects on other project constraints [44].

By integrating these mitigation strategies into construction project management practices, stakeholders enhance their ability to navigate the complexities of cold-region projects. Contractors, especially in remote areas with limited resources, can take specific steps to implement these strategies effectively. These include leveraging local materials and workforces to reduce dependency on external supply chains, adopting modular construction techniques to minimize on-site work, investing in weather-resilient technologies, and establishing contingency plans for logistical disruptions. Past studies have extensively explored the role of big data and UAV technologies in optimizing decision making and data collection in complex environments. For instance, research on big data analytics highlights its potential in predictive modeling and real-time decision support, which can enhance construction project management by mitigating uncertainties [88]. Similarly, UAV-enabled multi-target tracking and sensing frameworks have demonstrated their effectiveness in automated data collection, particularly in dynamic and resource-constrained environments [89]. Linking these insights to the present study, integrating such technologies can enhance the accuracy and efficiency of assessing construction project delays, especially in cold regions where real-time data and predictive analytics are crucial for proactive decision-making. This holistic approach, combining informed decision making and adaptive strategies, is essential for effectively managing and mitigating delays in such challenging environments.

5.1. Implications

5.1.1. Implications for Policy

The findings of this study have significant implications for government, policymakers, and industry stakeholders. Identifying snowfall, rainfall, and low temperature as influential weather-related factors affecting CPDs underscores the need for tailored strategies to weather-proof projects in cold regions. Policymakers should prioritize the development of specific institutional policies and technological support to assist construction workers in adapting to cold weather conditions. Concrete recommendations include establishing training programs, financial aid, and providing adaptable technologies. By addressing these specific needs, policymakers can effectively reduce economic losses and foster a resilient construction industry that is capable of adapting to changing weather patterns.

5.1.2. Implications for Practice

The practical implications for the construction industry, especially in cold regions, are noteworthy. Project managers and industry professionals need to recognize the impact of weather-related factors on project delays, specifically snowfall, rainfall, and low temperatures. This awareness allows stakeholders to allocate resources and plan construction activities effectively. By establishing a hierarchy of delay-related factors, this research provides valuable insights for project managers to effectively prioritize mitigation efforts. Identifying the most critical risk factors—such as the storage of construction materials, effects on concrete operations, overall project delays, elevation and topography challenges, and unsafe working conditions—enables construction professionals to allocate resources strategically and implement targeted interventions.

The prediction model developed in this study offers a proactive approach to risk management, empowering project managers to identify high-risk periods and implement measures to minimize delays. This involves strategic rescheduling, resource allocation, and alternative construction methods. By focusing on the highest-ranked risks, project managers can develop data-driven, adaptive strategies to mitigate weather-related delays and optimize construction efficiency in challenging environments. Overall, this study provides actionable guidance for enhancing project efficiency, optimizing resource allocation, and achieving anticipated benefits in construction projects subjected to weather-induced delays in extremely cold regions.

6. Conclusions

This study investigated the factors affecting CPDs due to weather in cold regions of India. To the best of the authors’ knowledge, no studies with this particular focus have been conducted. This gap in research was addressed using a novel two-phase hybrid SEM–ANN approach informed by the selection of 38 factors that influence CPDDWCs in cold regions based on a thorough review of related research and data collected from a survey of qualified experts with extensive experience in the field. The conclusions of this study are as follows:

(1)

This study’s analysis indicated that snowfall, rainfall, and low-temperature support hypotheses H1, H2, and H3, directly affecting project delays. Snowfall (β = 0.576, p = 0.000) had the largest impact on delays, followed by rainfall (β = 0.216, p = 0.000), and then low temperature (β = 0.096, p = 0.004).

(2)

The ANN analysis revealed that snowfall, rainfall, and low-temperature factors play significant roles as predictors of CPDs, with normalized importance values of 1.000, 0.890, and 0.790, respectively.

Construction project delivery and management are directly affected by CPDs. Contractors can minimize CPDs by planning and utilizing resources efficiently and effectively. This study used SEM to model the primary factors influencing CPDs to inform such planning efforts. Empirical data from a survey of 83 professionals in India’s construction industry were used to authenticate the model using the PLS-based SEM method.

This study has certain limitations that should be acknowledged. First, the sample size is relatively small, which may affect the generalizability of the findings. Expanding the sample to include a more diverse set of participants could provide deeper insights into the factors influencing construction project delays in cold regions. Second, the study focuses specifically on cold regions within India, which may limit the applicability of the results to other geographic locations with different climatic, economic, and regulatory conditions. The unique environmental and construction challenges in Indian cold regions may not fully align with those in other cold-climate areas, such as Alaska, Canada, or Greenland, where distinct construction practices and policies are in place. Additionally, while this research highlights critical delay factors, future studies could integrate comparative analyses across multiple cold regions globally to enhance the robustness of the findings. Despite these limitations, the study offers valuable insights that contribute to a better understanding of weather-induced delays in construction projects and lays the groundwork for future research aimed at improving project resilience in extreme climatic conditions.

Future research in this area should consider larger sample sizes to enhance the generalizability of findings on the impact of cold region weather factors on construction project delays. Additionally, investigating factors influencing project delays in cold regions outside India, particularly in regions closer to the North Pole or far south, would provide valuable comparative insights. Furthermore, exploring the effectiveness of specific mitigation strategies and technologies tailored for cold environments could improve project efficiency. Longitudinal studies tracking construction project delays over time would comprehensively understand trends and recurring challenges. By addressing these research gaps, future studies can contribute to a more robust understanding of construction project delays in cold regions and inform the development of effective mitigation strategies.