Virtual Reality and Digital Twins for Mechanical Engineering Lab Education: Applications in Composite Manufacturing

Abstract

This study investigates the effectiveness of a virtual reality (VR) simulation for teaching the hand lay-up process in composite manufacturing within mechanical engineering education. A within-subjects experiment involving 17 undergraduate mechanical engineering students compared the VR-based training with conventional physical laboratory instruction. Task performance, cognitive load, and learner perceptions were measured using procedural accuracy scores, completion times, NASA-TLX workload ratings, and post-task interviews. Results indicated that while participants required more time to complete the task in VR, procedural accuracy was comparable between VR and physical labs. VR significantly reduced mental, physical, and effort-related demands but elicited higher frustration levels, primarily due to navigation challenges and motion discomfort. Qualitative feedback showed strong learner preference for VR, citing its hazard-free environment, repeatability, and step-by-step guidance. These findings suggest that VR offers a viable and pedagogically effective alternative or complement to traditional composite-manufacturing training, particularly in contexts where access to physical facilities is limited. Future work should examine long-term skill retention, incorporate haptic feedback for tactile realism, and explore hybrid models combining VR and physical practice to optimise learning outcomes.

1. Introduction

The rapid development of digital technologies has created new possibilities for enhancing engineering education, particularly in disciplines that require hands-on technical skills. Among these technologies, virtual reality (VR) has emerged as a powerful tool for simulating practical laboratory environments and complex manufacturing processes (Darejeh et al., 2024a). By immersing learners in an interactive three-dimensional space, VR enables them to explore procedures, manipulate virtual tools, and perform tasks that closely replicate real-world operations—without the constraints of physical resources or the risks associated with hazardous materials (Kapilan et al., 2021; Ma et al., 2019).

In mechanical engineering education, practical training is critical for developing both procedural knowledge and problem-solving skills. However, delivering effective laboratory-based learning often presents significant challenges. Physical labs for composite manufacturing, for example, require specialized equipment, costly consumable materials such as fiberglass and resin, and adherence to strict safety protocols. These constraints limit opportunities for repeated practice and can restrict access to students outside scheduled lab sessions. Moreover, traditional labs demand considerable institutional investment in space, equipment maintenance, and instructor supervision (Pfotenhauer & Gagnon, 2021; Srinivasa et al., 2020).

VR-based training offers a promising solution to these limitations by replicating immersive simulated environments that can be adapted for various engineering processes, including composite manufacturing, in a fully interactive format. In such settings, learners can engage in tasks such as material cutting, resin application, and vacuum setup step-by-step, with real-time guidance and feedback (Falcão & Soares, 2013). These immersive laboratory environments have been shown to reduce extraneous cognitive load, enhance learner engagement, and provide opportunities for safe, repeatable practice without consuming physical resources (Georgiou et al., 2007; Halabi, 2020).

The focus of this paper is on the application of VR to teach the hand lay-up process for composite manufacturing within a mechanical engineering context. This process involves layering fibers and resin in a mold to create strong, lightweight components—an essential manufacturing technique in industries such as aerospace, automotive, and renewable energy. A VR simulation was developed to mirror the physical composite lab, enabling students to practice the procedure in a safe, repeatable, and resource-efficient environment.

The study evaluates the effectiveness of this VR-based training compared to a traditional physical lab. Specifically, it examines cognitive load, task performance, and learner perceptions to determine whether VR can deliver comparable or improved educational outcomes. By focusing on a single, high-value manufacturing process, this research contributes to the growing body of literature on VR in engineering education and offers practical insights for integrating immersive technologies into mechanical engineering curricula.

The following research questions inform this investigation:

RQ1. 

How does VR-based training compare with traditional physical laboratory training in terms of procedural accuracy and task performance?

RQ2. 

How does VR-based training influence different dimensions of cognitive load—mental demand, physical demand, effort, and frustration—relative to a physical laboratory setting?

RQ3. 

How do learners perceive the usability, realism, and pedagogical value of the VR environment compared with the physical laboratory?

Composite manufacturing presents distinct pedagogical and cognitive challenges compared to other mechanical engineering domains, such as machining or welding. Its multi-stage nature, requiring precise fibre alignment, resin mixing, and vacuum curing, demands high procedural accuracy and sustained attention under safety constraints. These characteristics make it an ideal context for exploring how VR can reduce extraneous cognitive load while enabling safe, repeatable practice. These considerations motivate a closer examination of prior research on VR in engineering education, which is reviewed in the next section.

2. Literature Review

2.1. Background and Evolution of VR in Engineering Education

VR has evolved from a technical visualization tool into a powerful medium for experiential learning. Early research by Ellis (1991), Fisher (1999), and Seth et al. (2001) introduced VR as a means to represent complex spatial data and allow users to interact directly with simulated environments. These affordances laid the groundwork for its educational applications, particularly in engineering and technical domains that rely on spatial reasoning and process visualization.

Advances in head-mounted displays, motion tracking, and 3D rendering have since enabled VR’s integration into mainstream education (Hernández-Chávez et al., 2021). In engineering contexts, VR has been used to simulate laboratory experiments and industrial environments that are hazardous, costly, or logistically constrained (Halabi, 2020; Kapilan et al., 2021). Ma et al. (2019) demonstrated that VR laboratories can emulate manufacturing-system design processes with high ecological validity, while Srinivasa et al. (2020) found that students using VR for materials testing established stronger links between theoretical principles and physical phenomena.

The COVID-19 pandemic accelerated the adoption of immersive laboratories. Pfotenhauer and Gagnon (2021) highlighted how VR maintained continuity of laboratory learning when physical access was restricted, echoing institutional implementations described by Darejeh and Mashayekh (2025), who successfully used VR environments to support skill-based learning during remote instruction.

Recent applications extend VR’s reach to domains such as additive manufacturing, robotics, and composite-material fabrication (Akpan & Offodile, 2024; Nasri et al., 2024; Saravanan et al., 2025). These environments bridge theoretical understanding with practical skill acquisition by allowing repeated, risk-free practice. Beyond safety and accessibility, Bano et al. (2024) observed that VR-based training can also promote awareness of sustainability, encouraging students to visualize material waste and environmental impact. Similarly, Wang et al. (2023) emphasised that advances in nanomaterial-based sensing and adaptive feedback can enhance realism and precision in VR educational systems, enabling more accurate simulation of manufacturing processes.

While the practical advantages of VR are increasingly evident, its pedagogical mechanisms remain less clearly defined. Understanding why and under what conditions VR enhances learning requires a theoretical foundation that integrates cognitive, affective, and motivational perspectives. The following section explores these frameworks.

2.2. Theoretical Frameworks for VR-Based Learning

2.2.1. Constructivism and Situated Learning

Constructivist theory (Piaget, 1972; Vygotsky, 1978) asserts that learners actively construct knowledge through meaningful interaction and social mediation. VR environments naturally align with this perspective by enabling learners to manipulate virtual artefacts, test hypotheses, and receive immediate feedback (W. Li et al., 2023). By situating tasks in realistic contexts, VR promotes situated learning, where understanding emerges from experience rather than rote memorization. Collaborative VR environments also support Vygotsky’s (1978) principle of social constructivism, fostering peer dialogue and shared problem-solving that strengthen conceptual understanding (Kapilan et al., 2021).

2.2.2. Experiential Learning Theory

Kolb’s (1984) Experiential Learning Theory conceptualises learning as a cyclical process involving concrete experience, reflective observation, abstract conceptualisation, and active experimentation. VR offers an ideal medium for enacting this cycle by allowing students to engage in authentic tasks, observe outcomes, reflect on performance, and iteratively refine their approach (Halabi, 2020; Kapilan et al., 2021; Di Natale et al., 2020). In laboratory education, this cyclical engagement encourages deep learning and procedural mastery while reducing the logistical barriers of physical experimentation.

2.2.3. Technology Acceptance Model and Motivation

The Technology Acceptance Model (TAM) (Davis et al., 1989) explains user adoption based on perceived usefulness and ease of use. Recent extensions of TAM introduce additional predictors such as immersion, presence, flow, and enjoyment, which are highly relevant in immersive contexts (Fussell & Truong, 2022; Radianti et al., 2020; Petersen et al., 2022; Kaplan-Rakowski et al., 2022). Learners who experience presence and enjoyment are more likely to engage voluntarily, sustaining motivation through intrinsic rather than extrinsic means (Makransky & Petersen, 2021). These motivational factors bridge technology acceptance with self-determination theory, suggesting that VR enhances autonomy and competence.

In the present study, TAM was used as a conceptual lens to interpret learners’ motivational and affective responses to the VR system. Rather than serving as the basis for a quantitative questionnaire, TAM guided the development of interview questions exploring perceived usefulness, perceived ease of use, and enjoyment of the VR environment.

2.2.4. Cognitive Load Theory

Cognitive Load Theory (CLT) (Leppink et al., 2013; Sweller, 1988) provides a cognitive framework for understanding how immersive environments influence mental effort. CLT distinguishes intrinsic, extraneous, and germane loads. Well-designed VR can reduce extraneous load by embedding guidance directly within the environment, but excessive interactivity or sensory input may overwhelm working memory (Darejeh et al., 2024a, 2024c). Makransky et al. (2019) found that immersive simulations, while increasing presence, sometimes reduced learning due to higher cognitive load. Parong and Mayer (2021) similarly observed that learners expend more effort on navigation than comprehension when instructional scaffolds are weak. Makransky and Mayer (2022) refined this argument in the “immersion principle,” emphasising that sensory fidelity improves learning only when aligned with pedagogical relevance.

Integrating constructivism, ELT, TAM, and CLT produces a comprehensive model for analysing VR-based learning. Constructivism and ELT explain how knowledge develops through active participation; TAM elucidates why learners engage; CLT clarifies how much cognitive effort the interface demands. These frameworks collectively inform the design and evaluation of the present study, which examines cognitive load, procedural accuracy, and learner perception in a composite-manufacturing context.

2.3. Empirical Findings on VR Learning in Engineering and Manufacturing

Empirical studies across STEM disciplines consistently show that VR can improve engagement, motivation, and spatial understanding (Radianti et al., 2020; Makransky & Mayer, 2022). However, outcomes related to knowledge retention, skill transfer, and cognitive load vary substantially. Zhao et al. (2019) reported that immersive VR enhanced conceptual understanding but increased mental demand relative to desktop simulations. Halabi (2020) and Kapilan et al. (2021) found that VR training reduced anxiety and improved procedural accuracy in safety-focused tasks. These mixed results indicate that VR’s effectiveness depends on the alignment between task design, interactivity, and cognitive complexity.

Affective and motivational factors also play a key role. Studies grounded in TAM and self-determination theory suggest that presence and enjoyment strongly influence satisfaction and willingness to re-engage (Kaplan-Rakowski et al., 2022; Petersen et al., 2022). Conversely, frustration, motion sickness, and interface discomfort can undermine learning despite high engagement (Makransky & Petersen, 2021). Thus, the challenge lies in balancing cognitive load with emotional involvement to sustain attention without overloading learners.

In applied engineering education, VR is particularly effective for procedural and safety-critical learning. Abulrub et al. (2011) demonstrated improved hazard recognition and coordination in virtual welding environments. Ma et al. (2019) reported gains in design accuracy and system comprehension among students using VR-based manufacturing modules. Hernández-Chávez et al. (2021) observed enhanced spatial–visualisation skills in automotive-design laboratories, while Pfotenhauer and Gagnon (2021) confirmed equivalent performance between virtual and physical thermodynamics labs (V. Li et al., 2025). Darejeh (2023) found that immersive laboratory simulations enhanced both procedural confidence and engagement, supporting VR’s complementary role alongside traditional methods.

Recent studies have further explored machining and composite-material training. Chuang et al. (2023) observed improvements in task efficiency and error reduction in VR machining environments. Nasri et al. (2024) and Saravanan et al. (2025) identified positive effects of VR on collaborative problem-solving and error tolerance. Nonetheless, Makransky et al. (2017) and Parong and Mayer (2021) caution that excessive immersion may impair factual retention without adequate scaffolding.

Consistent with these observations, several studies have reported neutral or even adverse outcomes, underscoring the complexity of immersive learning. Makransky and Petersen (2021) demonstrated that high-immersion environments can generate greater extraneous load than desktop-based equivalents, reducing transfer performance despite increased enjoyment. Parong and Mayer (2021) found that VR learners often expend more cognitive effort on navigation and interpretation than on core learning tasks, leading to lower retention. Together, these findings suggest that VR’s educational value is contingent on effective cognitive-load management and usability design. For VR-based composite-manufacturing training, this means prioritising task-relevant cues, step-by-step guidance, and calibrated environmental complexity to optimise learning efficiency and retention.

In addition to individual outcomes, multi-user VR environments have demonstrated strong benefits for teamwork and communication. Kapilan et al. (2021) reported that learners interacted more equitably in collaborative VR, while Saravanan et al. (2025) found that shared virtual environments supported inclusive, low-anxiety communication. Such results highlight that VR’s value extends beyond cognition to social and emotional engagement, preparing students for collaborative industry contexts.

Despite these advances, critical gaps remain. Comparative within-subject studies that directly evaluate cognitive load and procedural accuracy between immersive and physical laboratories are scarce, particularly in composite manufacturing—an area involving multi-stage, safety-intensive operations requiring precise sequencing and hazard awareness.

2.4. Research Gaps and Contribution of the Present Study

Although extensive research has demonstrated VR’s effectiveness for enhancing engagement, motivation, and conceptual understanding in engineering education, several gaps remain in both theoretical grounding and application scope. Most existing studies focus on linear or repetitive processes such as machining, welding, thermodynamics, or material testing (Kapilan et al., 2021; Pfotenhauer & Gagnon, 2021; Srinivasa et al., 2020). These tasks involve relatively predictable sequences and stable environments, making them less cognitively demanding than procedures that require continuous adaptation. In contrast, composite manufacturing involves multi-stage, material-sensitive processes, cutting fibres, applying resin, layering, and vacuum curing, where each stage depends on the precision of the previous one. Errors in resin ratios or curing times can compromise the integrity of the final product, introducing unique cognitive and safety challenges (Halabi, 2020; Ma et al., 2019). Furthermore, composite laboratories are costly and subject to strict safety protocols, limiting opportunities for repeated hands-on practice. Despite the potential of VR to mitigate these constraints by providing a safe, resource-efficient, and repeatable learning environment (Falcão & Soares, 2013; Georgiou et al., 2007), empirical studies in this specific context remain scarce.

Beyond this domain-specific limitation, broader pedagogical and methodological gaps also persist. Much of the existing VR research emphasises affective and motivational outcomes while neglecting procedural cognition and cognitive efficiency—how learners process, retain, and accurately execute complex procedures. Few investigations integrate complementary theoretical frameworks such as constructivism, experiential learning theory, the TAM, and CLT within a unified analytical lens. Consequently, while user satisfaction and motivation are frequently reported, these outcomes are rarely linked to measurable learning performance or workload metrics. This lack of theoretical integration constrains understanding of why and under what conditions immersive environments effectively support procedural learning.

The present study responds to these interrelated gaps by examining how immersive VR influences cognitive load, procedural accuracy, and learner perceptions during composite-manufacturing training. The developed simulation is grounded in constructivist and experiential learning principles, enabling learners to actively experiment, reflect, and iterate within a controlled environment. At the same time, the system embeds cognitive-load-informed scaffolding to balance realism with mental effort. Learners’ perceptions of usefulness and ease of use, derived from TAM, provide additional insight into motivation and acceptance, bridging cognitive and affective dimensions.

Building on earlier work by Darejeh and Mashayekh (2025), this research extends the evaluation of immersive environments to a safety-critical and resource-intensive educational domain that has received little empirical attention. By combining quantitative measures of workload and procedural accuracy with qualitative interview data, the study offers a holistic perspective on VR’s pedagogical value.

3. Methodology

3.1. Research Design

This study employed a within-subjects experimental design to compare the effectiveness of VR-based and physical laboratory training for teaching the hand lay-up process in composite manufacturing. The within-subjects design was chosen to control for individual differences in prior knowledge, technical ability, and learning style, as each participant experienced both training conditions. The study measured task performance, cognitive load, and learner perceptions across both environments.

3.2. Participants

Seventeen undergraduate mechanical engineering students from the University of New South Wales (UNSW) participated in the study. Participants ranged in age from 18 to 25 years (M = 21.2, SD = 1.9) and had no prior experience with composite-manufacturing processes. Recruitment occurred through course announcements and direct invitations to students enrolled in relevant engineering subjects. Participation was voluntary, and students received no academic credit for involvement.

All participants provided written informed consent before taking part in the study. The research protocol was approved by the UNSW Human Research Ethics Advisory Panel and adhered to institutional and national ethical guidelines for human research.

3.3. VR Application Development

A high-fidelity virtual-reality simulation of the composite-manufacturing hand lay-up process was developed using Unreal Engine 5 (UE5) as part of a broader metaverse platform initiative at UNSW. The application was deployed on HTC Vive Pro 2 headsets, each providing a native resolution of 2448 × 2448 pixels per eye and a refresh rate of 120 Hz to ensure visual clarity and minimise motion discomfort. The system ran on high-performance desktop computers equipped with an Intel i9 processor (4.9 GHz), 32 GB RAM, a 2 TB SSD, and an NVIDIA GeForce RTX 3070 graphics card with 8 GB of dedicated VRAM, enabling real-time rendering of complex physics simulations and maintaining consistent frame rates above 90 FPS. These specifications were selected to ensure low-latency performance and stable interaction fidelity during extended immersive sessions.

Unreal Engine 5 was chosen after preliminary prototyping in both UE5 and Unity 3D. While Unity offers advantages in rapid prototyping and mobile deployment, UE5 provides superior photorealistic rendering, advanced lighting (Lumen GI), and precise physics simulation, capabilities that are critical for accurately replicating resin flow, material layering, and fibre alignment during composite fabrication. Its Nanite geometry system ensured detailed object visualisation without compromising performance, producing the spatial realism necessary for learners to perceive fine-grained details of composite textures and tool manipulation (Venter & Ogterop, 2022).

The HTC Vive Pro 2 was selected following an evaluation of several commercially available head-mounted displays, including the Meta Quest 2 and 3. Although stand-alone devices such as the Quest 3 offer greater portability, they are limited by lower GPU capacity and reduced tracking precision, which can compromise interaction accuracy in high-fidelity engineering simulations. In contrast, the Vive Pro 2’s external lighthouse tracking and high refresh rate provided sub-millimetre positional accuracy and reduced motion latency, essential for maintaining the hand–eye coordination required in composite lay-up tasks (Le Chénéchal & Chatel-Goldman, 2018).

Furthermore, the chosen hardware and software combination allowed seamless integration with existing laboratory infrastructure and provided scalability for future collaborative modules. The PC-tethered configuration supports the computational demands of photorealistic rendering and real-time physics while enabling future interoperability with digital-twin data pipelines.

The virtual environment accurately replicated the physical composite-manufacturing laboratory (Figure 1 and Figure 2), including workstations, tools, consumable materials, and simulated equipment. It was intentionally designed to remain clean and free of distractions, keeping students focused on the task. Guided learning instructions and interactive features provided an authentic yet hazard-free simulation of real-world manufacturing processes.

Within the simulation, users employ VR controllers to navigate the laboratory environment and interact with virtual objects in a manner closely resembling real-world procedures. An overview of the virtual lab environment is presented in Figure 3.

They begin by selecting and cutting fiberglass sheets to specified dimensions, ensuring precision in measurement and alignment. This step introduces learners to the importance of accuracy in preparing raw materials, while simultaneously familiarising them with VR-based hand–eye coordination. Once the fibers are prepared, participants proceed to apply resin evenly across the sheets, an operation that demands attention to coverage consistency to prevent defects in the final composite structure. The simulation provides visual cues, such as color changes and texture updates, to indicate proper resin saturation, thereby reinforcing correct procedural knowledge.

Following resin application, users layer fibers within a mold to replicate the stacking process fundamental to composite strength and durability. The system requires students to correctly orient the fiber layers, mirroring real-life challenges where orientation directly affects material properties. To complete the procedure, students assemble and connect a virtual vacuum pump system, which includes attaching hoses and sealing the mold. This stage exposes learners to the critical curing process, demonstrating how air extraction and pressure application consolidate the composite layers.

The workflow also incorporates logistical elements such as transporting prepared materials to the workstation and arranging them in the correct order, highlighting organizational and sequencing skills that are often overlooked in traditional teaching methods. The guided environment ensures that each action contributes to the overall understanding of the composite-manufacturing process while minimising the risks associated with handling sharp tools, volatile chemicals, and heavy equipment.

Figure 4, Figure 5 and Figure 6 illustrate the key stages and visual components of the VR environment. Figure 4 presents an overview of the virtual workspace, displaying the materials shelf containing labelled rolls of Breather, Pill Ply, Glass Fibre, and Mesh. These materials replicate those used in the physical lab, and participants select them from the shelf to complete each stage of the task.

Figure 5 shows the intermediate stage in which the participant, guided by on-screen instructions, arranges the composite layers (Glass Fibre, Pill Ply, Mesh, and Breather) in the specified order using the VR controllers to grasp and position each layer accurately.

Figure 6 depicts the final stage of the process, where the participant connects tubing from the resin bucket to the sealed composite lay-up on the table, completing the vacuum setup required for resin infusion and curing. Together, these images represent the entire procedural workflow implemented in the VR simulation and mirror the equivalent stages in the physical laboratory.

3.4. Experimental Procedure

The experiment was conducted in three structured phases designed to ensure methodological rigor and allow for both quantitative and qualitative comparisons between the VR and physical laboratory conditions.

(a) 

Pre-Experiment Phase

Participants first completed a short eligibility questionnaire to confirm the absence of prior experience in composite manufacturing. This step was essential to control for prior knowledge, ensuring that performance differences could be attributed to the training environment rather than pre-existing expertise. Basic demographic data (age, gender, year of study) were also collected to contextualise findings and allow for the identification of potential sample biases.

Following the questionnaire, a brief orientation session was conducted. Participants were introduced to the physical laboratory facilities and given a demonstration of the VR equipment, including headset usage and controller functions. This familiarisation was critical for reducing potential extraneous cognitive load during the training phase, as it ensured that difficulties with equipment handling would not confound performance outcomes. The orientation also aimed to minimise anxiety or discomfort associated with VR use, such as motion sickness, which has been identified as a barrier to learning in immersive environments.

(b) 

Training Phase

Participants were randomly assigned to one of two counterbalanced condition orders to control for order effects:

VR-first group: Completed the VR simulation followed by the physical lab.

Physical-first group: Completed the physical lab before the VR simulation. This stage was conducted in the Mechanical Engineering Laboratory within the School of Mechanical and Manufacturing Engineering at UNSW (see Figure 1 and Figure 2).

In both conditions, participants were required to perform the complete hand lay-up process, replicating the sequential steps of composite manufacturing. The VR environment included an embedded, on-screen, step-by-step instructional guide designed to scaffold learning and provide real-time guidance. By contrast, the physical lab condition relied on traditional paper-based instructions, reflecting common instructional practices in engineering education. This difference allowed the study to evaluate not only the environments themselves but also the role of instructional delivery methods in shaping learner outcomes.

The distinction in instructional modality was intentional: the VR condition integrated step-by-step, on-screen guidance designed to minimise split attention and support procedural sequencing, whereas the physical laboratory relied on printed instructions that students consulted between stages. This contrast allowed examination of whether embedded, real-time guidance in VR influences learning efficiency or cognitive load compared with traditional paper-based formats.

It is important to note that participants in both conditions had continuous access to task instructions throughout the activity. In the VR condition, guidance was presented as embedded, step-by-step visual prompts within the environment, whereas in the physical laboratory, the same content was available as printed sheets placed beside the workstation. Consequently, procedural accuracy in this study reflects participants’ task performance under two instructional modalities, embedded digital guidance versus paper-based reference, rather than their retention of procedural steps from memory. The comparison therefore highlights the effect of instructional format on task execution, not differences in memory recall.

(c) 

Post-Task Data Collection

Upon completion of each condition, participants immediately completed the NASA Task Load Index (NASA-TLX), which measures perceived cognitive workload across six dimensions: mental demand, physical demand, temporal demand, performance, effort, and frustration. Administering the instrument after each condition enabled direct within-subject comparisons and reduced recall bias.

Task performance was recorded in two forms: (1) completion time, measured in minutes and seconds to capture efficiency, and (2) procedural accuracy, defined as the number of correctly executed steps performed in the prescribed sequence. These metrics provided complementary insights—completion time reflecting efficiency, and procedural accuracy reflecting the quality of skill acquisition.

Finally, semi-structured interviews were conducted to gather richer insights into the learner experience. The purpose was to explore user perceptions of task difficulty, cognitive effort, usability, and educational value. While the quantitative component focused on workload and performance measures, the qualitative interviews were designed to capture learners’ acceptance of the VR technology in line with TAM constructs. Specifically, questions targeting perceived usefulness, ease of use, and enjoyment were included to reflect TAM’s motivational dimensions within the broader theoretical framework. The interview guide included the following questions:

• Did you find any of the activities especially difficult? Why?
• Did you feel overwhelmed at any point during the activities? When and why?
• Which parts of the system did you like the most? Why?
• Which parts of the system were the easiest to use? Why?
• Do you believe this system could replace in-person laboratories after being further refined? Why?
• Do you believe completing these tasks in person would have been easier or harder? Why?

Each interview lasted approximately 10 min and was conducted immediately after participants completed both training conditions. All interviews were conducted by the student researcher. Rather than audio recording, detailed notes were taken during each interview to capture participants’ responses as accurately as possible.

3.5. Data Analysis

Quantitative data were analysed using paired-samples t-tests, which enabled direct within-subject comparisons between the VR and physical lab conditions. This design controlled for inter-individual variability in baseline ability, allowing statistical significance to be attributed to the training condition rather than participant differences. Significance thresholds were set at p < 0.05, with effect sizes considered to contextualise the magnitude of observed differences.

All statistical analyses were conducted on the complete dataset, including both sequencing subgroups (VR-first and Lab-first). The design was within-subjects, meaning that each participant contributed paired data from both instructional conditions. Paired-sample t-tests were therefore performed across the entire participant pool rather than within isolated subgroups.

Qualitative data were analysed using an inductive thematic analysis approach (Braun & Clarke, 2006). All transcripts were imported into NVivo 12 for systematic data organisation, coding, and theme development. Researchers read and coded the transcripts to identify recurring ideas and patterns related to usability, workload, and perceived realism. Initial codes were derived directly from participants’ statements and subsequently grouped into broader conceptual categories. The resulting themes were refined and validated through iterative comparison with the raw data to ensure internal consistency and representativeness. Illustrative quotations from participants are presented in the Results section to exemplify each theme.

4. Results

4.1. Task Performance (Quantitative Results)

A paired-samples t-test revealed a significant difference in task completion times between the VR and physical laboratory conditions. Participants took longer to complete the task in VR (M = 12.46 min, SD = 2.31) than in the physical lab (M = 10.00 min, SD = 1.24), t(16) = 6.00, p < 0.001.

Procedural accuracy, measured as the number of correctly executed steps, was slightly higher in VR (M = 4.59, SD = 2.14) than in the physical lab (M = 3.94, SD = 0.97). However, this difference was not statistically significant, t(16) = 0.07, p = 0.945, indicating that both environments supported similar levels of procedural retention.

These results suggest that while participants worked more slowly in the VR condition—likely due to navigation and control familiarity—this did not negatively affect their ability to remember and correctly execute the manufacturing steps.

4.2. Cognitive Load

NASA-TLX scores indicated significant differences in perceived workload between the two conditions (

Table 1. Mean NASA-TLX Scores (15-point scale) by Condition.

Workload Dimension	VR (M ± SD)	Physical Lab (M ± SD)	t(16)	p
Mental Demand	11.23 ± 3.35	13.82 ± 1.67	−4.28	0.001
Physical Demand	6.06 ± 2.46	15.00 ± 1.44	−9.75	<0.001
Effort	3.41 ± 1.85	10.29 ± 2.31	−8.42	<0.001
Frustration	5.06 ± 2.25	1.71 ± 1.83	4.86	0.0002

).

• Mental Demand: Significantly lower in VR (M = 11.23, SD = 3.35) compared to the physical lab (M = 13.82, SD = 1.67), t(16) = −4.28, p = 0.001.
• Physical Demand: Markedly reduced in VR (M = 6.06, SD = 2.46) compared to the physical lab (M = 15.00, SD = 1.44), t(16) = −9.75, p < 0.001.
• Effort: Lower in VR (M = 3.41, SD = 1.85) than in the physical lab (M = 10.29, SD = 2.31), t(16) = −8.42, p < 0.001.
• Frustration: Higher in VR (M = 5.06, SD = 2.25) compared to the physical lab (M = 1.71, SD = 1.83), t(16) = 4.86, p = 0.0002.

No statistically significant differences were observed in temporal demand or perceived performance between the two environments.

These findings indicate that VR significantly reduced mental, physical, and effort-related workload compared to the physical lab but also increased frustration levels—likely linked to motion discomfort and navigation challenges. Table 1 presents a summary of the NASA-TLX scores.

Figure 7 provides a visual summary of these results. The bar chart displays mean scores for task performance and NASA-TLX workload dimensions in both conditions, with error bars representing standard deviations. Asterisks indicate variables where statistically significant differences were found between the VR and physical lab conditions (p < 0.05).

To examine potential order effects, performance and workload data were compared between participants who completed the VR condition first and those who began with the physical laboratory. Independent-samples t-tests revealed no statistically significant differences between order groups on any variable (all p > 0.10), indicating that the counterbalancing procedure successfully controlled for learning or fatigue effects.

4.3. Learner Perceptions (Qualitative Results)

The semi-structured interviews provided valuable insights into participants’ perceptions of the VR training compared with the physical laboratory. Four overarching themes emerged: (1) initial motor and coordination difficulty, (2) occasional cognitive overload when managing guidance and manipulation simultaneously, (3) high usability and engagement after familiarisation, and (4) recognition of VR’s pedagogical value as a safe and motivating learning environment.

Participants initially described some physical and coordination challenges when interacting with the virtual materials. One student noted, “The layering part was tricky at first because you have to be precise with the controller, it’s not like handling real fibre.” Others mentioned brief moments of cognitive strain while trying to follow on-screen instructions during task execution: “At the start I was trying to read the instructions and move things at the same time, so it was easy to lose focus for a moment.” Despite these minor difficulties, most participants reported that they quickly adapted to the controls, finding them increasingly intuitive and responsive over time. As one explained, “Once I understood the controls, everything felt very intuitive, placing and mixing materials was straightforward.”

Positive feedback centred on VR’s interactivity, safety, and integrated instructional guidance. Many participants highlighted that the on-screen, step-by-step prompts made the experience feel supported and engaging: “I liked that the system guided me step by step; it felt like having a tutor next to me.” Students also appreciated the ability to repeat tasks multiple times without material waste or safety risks, which encouraged deliberate practice and confidence building. One participant commented, “It’s easier in VR because you can try again if you make a mistake, and you don’t have to deal with the mess.” Another added, “This could easily be used for training before going into the real lab; it helps you memorise the steps.”

Participants consistently emphasised the pedagogical and practical advantages of VR. They valued the hazard-free setting that removed concerns about chemical exposure or equipment misuse, particularly for beginners who might feel anxious in a high-risk lab. The integrated, real-time instructions were also seen as reducing cognitive load by eliminating the need to alternate between reading paper-based instructions and performing the task: “Having the instructions inside the VR made it much easier to focus on what I was doing.”

Although the majority expressed a strong preference for using VR in future training, several participants noted that the system could not yet fully replace the physical lab. As one explained, “It’s great for practice, but you still need to feel the actual materials to really learn.” A few also reported usability and ergonomic limitations, including difficulty manipulating smaller objects and occasional motion discomfort. These issues sometimes interrupted concentration and contributed to higher frustration levels, consistent with the elevated NASA-TLX frustration scores recorded in the VR condition.

5. Discussion

This study examined the effectiveness of a VR-based training simulation for teaching the hand lay-up process in composite manufacturing, comparing it with traditional physical laboratory training. Although participants completed tasks more slowly in VR, they reported significantly lower mental, physical, and effort-related demands. Procedural accuracy was comparable between VR and physical lab conditions, and learner feedback suggested a strong preference for VR as a training modality, supporting earlier evidence that immersive environments can enhance motivation and learner satisfaction (Halabi, 2020; Kapilan et al., 2021).

The increased task completion time aligns with previous findings that learners in immersive VR often require additional time to navigate and interact due to unfamiliarity with the interface (Kamińska et al., 2021; Ma et al., 2019). Importantly, this slower pace did not negatively impact procedural retention, consistent with studies showing that VR can match or exceed traditional methods in supporting technical skill acquisition (Pfotenhauer & Gagnon, 2021; Srinivasa et al., 2020). The reduced cognitive and physical demand in VR reflects the predictions of Cognitive Load Theory (Sweller, 1988), which suggests that lowering extraneous load—such as physical fatigue, environmental hazards, and material handling complexities—frees cognitive resources for essential learning. The embedded on-screen, step-by-step guidance in the VR simulation likely contributed to this effect by reducing split attention and supporting germane cognitive load, enabling deeper processing and knowledge consolidation despite the high sensory demands (Darejeh et al., 2022; Georgiou et al., 2007).

These cognitive processing patterns are consistent with recent research showing that immersive VR can enhance motivation and engagement through heightened presence (Huang et al., 2020) but may also risk increasing extraneous load without adequate guidance (Makransky & Petersen, 2021; Parong & Mayer, 2021). In our study, careful instructional design appears to have balanced sensory immersion and environmental complexity, supporting learning without overloading working memory.

Higher frustration scores in VR mirror earlier findings that motion discomfort, technical limitations, and less intuitive controls can hinder user experience (Ghazali et al., 2024; Kamińska et al., 2021; Makransky & Petersen, 2021). Addressing these usability challenges—through improved interface design, motion tracking, and adaptive interaction methods—will be important for broader adoption. Nonetheless, these challenges did not significantly undermine performance outcomes, suggesting that well-designed VR training can achieve procedural learning outcomes equivalent to physical labs while offering reduced cognitive and physical demands.

Recent bibliometric research highlights the rapid expansion of digital-twin applications in manufacturing and materials engineering (Vergara et al., 2025). While most of this work has focused on system-level and industrial implementations, the present study demonstrates how digital-twin principles can also be applied pedagogically through immersive VR. By replicating real-world composite-manufacturing processes and feedback loops, the developed simulation illustrates how digital-twin technologies can enhance learning, safety, and cognitive efficiency in engineering education.

The implications of these findings are relevant to both education and industry. In higher education, VR provides a scalable, hazard-free training solution that can supplement or partially replace physical lab sessions, particularly in institutions with limited access to specialised composite-manufacturing facilities (Kapilan et al., 2021; Srinivasa et al., 2020). The capacity of VR simulations to replicate the entire hand lay-up process enables consistent, standardised instruction across large cohorts, thereby reducing variability in learning outcomes (Ma et al., 2019). Introducing VR early in a course could help students develop procedural familiarity before entering the physical lab, reducing errors and increasing efficiency in hands-on sessions (Falcão & Soares, 2013).

From an industrial perspective, VR-based composite training could be integrated into professional development programs to upskill technicians without disrupting production schedules or incurring the cost of consumable materials (Alpala et al., 2022; Hosseinia et al., 2024). The hazard-free environment also facilitates experimentation with process variations, encouraging innovation and process optimisation in manufacturing (Halabi, 2020). Furthermore, VR enables consistent training delivery to geographically distributed teams, overcoming logistical and safety challenges associated with centralised training (Saravanan et al., 2025).

The observed differences in participants’ results likely stem in part from the differing instructional formats rather than from differences in skill retention. In the VR condition, participants referred to step-by-step visual prompts embedded within the environment, whereas in the physical lab, they accessed the same instructions on printed sheets placed beside the workstation. Consequently, procedural accuracy in this study reflects performance under two modes of guidance delivery rather than memory-based retention.

Despite the promising outcomes, this study has several limitations. The small sample size constrains the generalisability of results, and the participant pool consisted solely of undergraduate mechanical engineering students with no prior experience in composite manufacturing, meaning findings may differ for skilled practitioners (Srinivasa et al., 2020). The absence of haptic feedback is another limitation, as tactile sensation is critical for realistic simulation of resin application and fibre placement (Wang et al., 2023). Additionally, the higher frustration scores suggest that motion sickness and navigation difficulties remain barriers to adoption, echoing usability challenges documented in earlier VR studies (Ghazali et al., 2024; Kamińska et al., 2021).

While the results demonstrated notable short-term improvements in engagement and procedural accuracy, these outcomes capture only immediate effects following a single VR training session. The short duration and narrow scope of the intervention were intentionally designed to isolate initial usability and cognitive impacts rather than long-term learning trajectories. Consequently, this design limits conclusions regarding knowledge retention, continued engagement, and transfer of conceptual understanding to new contexts. As highlighted in prior research, sustained exposure and spaced practice are essential to determine whether early performance gains persist beyond the novelty phase (Makransky & Mayer, 2022; Parong & Mayer, 2021).

Future research should address these limitations by involving larger and more diverse participant groups, including both novice and experienced users, and by adopting longitudinal or multi-session designs with delayed post-tests to assess retention, transfer, and enduring motivation. The integration of advanced haptic devices should also be explored to enhance the realism of tactile interactions, and the application of VR to more complex composite-manufacturing methods beyond the hand lay-up process warrants investigation. Hybrid models that combine VR-based preparation with subsequent physical lab practice may further enhance efficiency, safety, and learning outcomes. In addition, incorporating AI-driven intelligent tutoring agents could provide adaptive, real-time feedback and personalised learning pathways in future studies (Darejeh et al., 2024b; Sargazi Moghadam et al., 2024).

6. Conclusions

This study compared a VR simulation of the hand lay-up process in composite manufacturing with traditional physical laboratory training. The results showed that VR achieved comparable procedural accuracy while reducing mental, physical, and effort-related workload. Although tasks took longer in VR, this likely reflected unfamiliarity with the interface rather than reduced learning efficiency.

The findings support the potential of VR as a scalable and hazard-free alternative or complement to physical lab sessions in both educational and industrial contexts. In universities, VR can help standardise training and extend access to complex manufacturing processes in resource-limited settings. In industry, it can enable safe, repeatable practice and remote training without production interruptions.

Future work should explore larger and more diverse participant groups, examine long-term skill retention, integrate haptic feedback for greater realism, and evaluate hybrid training models that combine VR with hands-on practice. As VR technology advances, its application in composite-manufacturing training has the potential to enhance both accessibility and skill acquisition.