Intelligent Automation in Knitting Manufacturing: Advanced Software Integration and Structural Optimisation for Complex Textile Design

Abstract

Automation in textile manufacturing plays a pivotal role in enhancing production efficiency, precision, and innovation. This study investigates the integration of intelligent technologies in the knitting sector, focusing on industrial flat knitting machines from a leading manufacturer and the use of the advanced software platform M1plus V7.5. The software’s capabilities for the digital design and simulation of complex patterned and structural knits are explored through the development and production of five experimental knitted designs. Each sample is evaluated in terms of its structural characteristics and dimensional behaviour after washing. The results highlight the potential of software-driven optimisation to improve product accuracy, reduce shrinkage variability, and support smart manufacturing practices in the textile industry.

1. Introduction

Industrial automation plays a key role in the development of the textile industry, aligning with the broader concept of intelligent manufacturing. It significantly enhances productivity, quality, and efficiency in manufacturing processes [1]. The implementation of modern automated systems and digital tools enables greater precision in textile production, reduces manufacturing defects, and ensures an optimal use of raw materials [2]. Thus, it contributes to more sustainable and optimised production cycles.

One of the most important benefits of automation is accelerated production, which enables rapid adaptation to market demands and supports product customization in line with modern industry trends [3]. Software-controlled knitting machines such as Stoll by Karl Mayer, Shima Seiki, and Steiger integrate seamlessly into intelligent manufacturing systems, providing high design flexibility and enabling the creation of complex textile structures with minimal human intervention [4]. These machines exemplify the shift towards cyber–physical systems and smart production environments in the textile sector.

Automation also reduces labour costs and the physical strain on operators [5], which is especially important in the context of sustainable and human-centred manufacturing. Automated machines are equipped with intuitive control interfaces and allow for quick, flexible adjustments in the production process [1,4], supporting adaptability and continuous improvement. Additionally, intelligent parameter control systems enhance quality assurance and ensure consistency in textile characteristics, thus contributing to smart quality control strategies [6].

Modern computerised knitting machines are equipped with innovative programming and control systems that ensure the precise reproduction of complex knitwear designs with minimal manufacturing errors. These systems form a core part of digital transformation in textile production. Software solutions such as M1plus V7.5 (Stoll by Karl Mayer) and Apex3 (Shima Seiki) offer integrated design environments with simulation and visualisation tools, enabling virtual prototyping, efficient design sampling, and substantial reductions in development time and resource costs [7,8]. Such tools play a vital role in the implementation of digital twins and cyber–physical systems within textile manufacturing.

The automation of knitting processes enhances yarn management efficiency and reduces material waste, contributing to more sustainable manufacturing practices while ensuring high repeatability of quality [9]. The ability to quickly adjust machine parameters through intelligent interfaces and software control facilitates the production of customised products. This positions modern knitting machines as key drivers of innovation and flexibility within the textile sector [4], in alignment with current advancements in smart manufacturing and adaptive production systems.

The textile industry exerts a significant impact on the environment, manifested through an excessive consumption of water and energy [10], the release of harmful chemicals into nature, and emissions of greenhouse gases [11]. Dyeing and other wet processing operations, as well as finishing treatments, contribute to the pollution of water resources [12]. The production of synthetic fibres is directly linked to the extraction and processing of petroleum. Textile waste, especially as a result of fast fashion, generates vast amounts of non-biodegradable materials, placing a heavy burden on landfills and ecosystems [13].

In response to these challenges, textile and apparel manufacturers are adopting various sustainable practices. Among the most common measures are the implementation of closed-loop water systems [14], the shift towards organic or recycled fibres [15], the optimisation of energy resources through automation [16], and the use of eco-friendly dyes and enzymatic treatments [17]. There is also a growing interest in developing products with extended life cycles [18], as well as in business models that promote reuse, recycling, and sustainable consumption [19].

Alongside these efforts, recent years have seen a growing implementation of intelligent technological solutions associated with Industry 4.0 [20]. Examples include the use of sensor systems for monitoring production parameters, the application of artificial intelligence for process optimisation, and the introduction of digital twins for simulating and predicting the behaviour of textile structures [21]. Some studies have demonstrated the successful integration of automated design platforms such as Shima Seiki Apex3 and Stoll M1plus; however, comparative analyses involving different yarn types and knitted structures remain limited [4,22].

Recent developments in artificial intelligence (AI) are further transforming textile manufacturing, with applications ranging from the predictive maintenance of knitting machines [23] to automated defect detection [24] and digital twin integration [25]. In particular, AI-driven design tools are increasingly being used to optimise knit structures, analyse fabric behaviour under different conditions, and generate pattern variations based on performance requirements [26,27]. Studies published since 2023 highlight the integration of deep learning models into simulation platforms, enabling a more accurate prediction of structural outcomes and personalised product development [28,29,30,31,32].

The automated production of complex knitted structures offers significant advantages but also presents several challenges. One of the main issues is the precision of yarn handling—when working with intricate knits involving different types of yarns, accurate tension control, feeding coordination, and machine synchronisation are essential. These parameters must be finely tuned to prevent defects in the fabric, such as dropped stitches, distortions, and other structural irregularities [33,34], which can compromise both quality and functionality.

Another challenge is the optimisation of software programming. Although modern CAD/CAM systems provide powerful simulation and visualisation tools, the process of translating a complex design into an efficient and error-free knitting machine programme requires an in-depth knowledge of knitting structures, software functionalities, and textile material behaviour [35]. The choice of yarns and their interaction with machine parameters during the knitting process can significantly influence the final outcome [36]. Different fibres and blends exhibit varying levels of shrinkage, elasticity, and response to mechanical stress, making it difficult to predict the exact appearance and performance of the final product. This complexity necessitates specialised expertise from the designer and highlights the need for enhanced integration between software intelligence and material science within textile automation 37].

Finally, quality control remains a challenge, as even minor imperfections can lead to deviations in the appearance and functionality of the final knitted products [37,38]. Addressing these issues requires a combination of automated technologies and expert human oversight.

This study aims to explore the processes of automated design and production of complex knitted structures, with a focus on the application of specialised software tools and the influence of different design types on the final textile product. The research outlines a practical methodology for working with M1plus V7.5 software in an intelligent manufacturing context. Five complex knit designs have been developed and produced using a computerised flat knitting machine. Their structural characteristics have been analysed, and the shrinkage of the knitted samples after washing has been evaluated, with particular attention to the influence of the pattern on fabric deformation.

Although current platforms such as M1plus offer limited predictive capabilities—particularly for patterns involving loop transfers or compound textures—the results contribute a practice-based methodological framework for optimising design parameters in intelligent textile manufacturing. Based on the results, recommendations have been formulated to improve accuracy, repeatability, and efficiency within the automated knitting process.

In contrast to studies focused on predictive modelling, the present research adopts a practical and exploratory evaluation approach. It emphasises the application of CAD/CAM software within real-world manufacturing workflows and provides an experimental analysis of post-production structural behaviours in complex knitted textiles.

2. Materials and Methods

2.1. Design Methodology Using M1plus V7.5

M1plus V7.5 is a specialised CAD/CAM software developed by Stoll, designed for the creation, programming, and simulation of knitted structures [7]. As part of the digital workflow, the software enables detailed design of complex patterned and structural knits, providing precise control over stitch types, yarn interactions, and needle movements. It serves as a key tool in the development of knitting programmes for Karl Mayer Stoll CMS 503 ki industrial flat knitting machines, supporting automation and customisation in production.

The main stages of working with M1plus V7.5 are illustrated in Figure 1, representing a structured process within a smart digital environment.

2.1.1. Creating a New Project

When launching the software, the user creates a new project file and defines the key parameters of the knitted macrostructure. This step initiates the digital design workflow and includes the following:

• Knit pattern—selection of single jersey, double jersey, jacquard, textured, and other knitting structures, depending on the design requirements.
• Machine gauge—setting the number of needles according to the specifications of the selected industrial knitting machine.
• Knitting density—a crucial parameter that determines the degree of yarn interlocking, directly affecting the structure and dimensional behaviour of the final product.
• Yarn types—selection of the yarn(s) to be used; the software supports a wide range of options including cotton, wool, synthetic fibres, and various blends of natural and synthetic materials.

This stage marks the foundation of an automated and adaptable design process within smart textile production.

2.1.2. Design Construction

The creation of the knitted macrostructure is the key stage in the design process. It begins with the selection of a base knit structure, which may be one of the following:

• Plain stitch—a uniform macrostructure with evenly distributed loops;
• Jacquard stitch—combining multiple colours and yarns to form intricate pattern motifs;
• Textured structures—involving loop transfers and needle skips to generate three-dimensional surfaces.

Once the structure is selected, the graphical design of the knit pattern is created. In M1plus V7.5, this can be performed manually, allowing the user to add or modify individual loops and needle actions. Alternatively, the software offers automatic pattern generation tools that enhance design efficiency. It also supports layer-based editing, enabling the separation and management of different fabric sections (e.g., background and motif layers). A built-in colour palette and symbolic system facilitate the clear differentiation of stitch types and yarns, contributing to a structured and intelligent design environment.

2.1.3. Machine Programming

After the design is completed, the graphical model is translated into machine code, enabling its execution by the knitting equipment. A key aspect of this stage is the definition of needle movements, including the sequencing for raising, lowering, and transferring loops. These instructions govern the interaction between the front and rear needle beds, which is especially critical for double-faced and structurally complex knitted fabrics.

M1plus V7.5 includes automated error detection algorithms that support quality assurance during programming. The software can identify inconsistencies such as improperly defined transfers, skipped or duplicated operations, and suboptimal row transitions. This built-in validation minimises the risk of production errors and ensures reliability throughout the automation process.

2.1.4. Design Simulation and Analysis

Before sending the programme file to the knitting machine, M1plus V7.5 provides a simulation of the final result through a built-in 3D visualisation module. This feature enables the designer to preview the knitted product’s structure, surface texture, and colour distribution in a realistic format.

The software also allows for structural analysis, including the behaviour of threads under different tensions and loading conditions. In addition, it supports predictive shrinkage analysis, helping estimate the post-wash dimensions of the fabric. This simulation stage contributes to more accurate forecasting, reduces the need for physical prototyping, and supports decision-making in the context of digital twins and smart textile development.

2.1.5. Export and Transfer to the Knitting Machine

After the final verification, the software generates a machine code, which is exported and transferred to the Stoll CMS 503 ki flat knitting machine. The code includes detailed instructions for needle movements, operating speed, and yarn carrierconfiguration, forming the basis of automated production.

Before starting the full knitting cycle, a test sample is typically produced to verify that all parameters function as intended. This test helps confirm that the digital simulation corresponds with the physical outcome.

The workflow with M1plus V7.5 ensures flexible and precise control over the design-to-production process. It allows for parameter optimisation of complex knitted structures and supports accurate simulation for better prediction of real-world results. Importantly, the methodology reduces the likelihood of production errors through integrated validation tools, thus contributing to a more reliable and intelligent manufacturing process.

2.2. Materials

Five knitted designs have been developed using complex crosswise knitting techniques. The production workflow follows the methodology outlined in Figure 1, with pattern digitisation and programming performed using the M1plus V7.5 software by Stoll [7].

Table 1. Characteristics of the threads used for the knit samples.

Knit Macrostructure	Sample 1	Sample 2	Sample 3	Sample 4	Sample 5
Threads	Cotton 100%, NE 60/4	Cotton 100%, NE 20/2	Cotton 100%, NE 20/2	Cotton 100%, NE 60/4	Cotton 100%, NE 20/2
				Cotton 100%, NE 20/2

presents the characteristics of the threads used in the creation of the knitted samples.

2.3. Methods

The type of knitting technology is selected according to the desired product format—knitted fabric, cut panel, or seamless garment. Based on this, the number of rows and active needles is defined for each design, as presented in

Table 2. Numbers of needles and rows for each of the designed knit macrostructures.

Knit Macrostructure	Sample 1	Sample 2	Sample 3	Sample 4	Sample 5
Number of needles	172	170	74	120	140
Number of rows	274	218	132	377	187

. The Karl Mayer Stoll CMS 503 ki flat knitting machine used in this study supports a maximum of 620 needles and enables high-precision control within automated production workflows.

A view of the machine setup is shown in Figure 2.

The knit setup parameters for the five macrostructures are summarised in

Table 3. Knit setup parameters for the five macrostructures.

Parameter	Setting
Rib binding setting	Stoll with protection thread
Needle binding type	Standard
Number of knitting systems for binding	1 system
Type of thread/stitch for binding	With elastic thread
Type of transition row	Transition DJ (for rib to double-faced knit)
Rib type	2 × 2

.

For each design variation, a single knitted sample was produced and analysed. All dimensional measurements and shrinkage evaluations were conducted on these individual specimens. This approach was chosen because this study focuses on structural comparison under uniform production and washing conditions, rather than statistical variation between multiple samples.

Due to the high elasticity of knitted fabrics, dimensional changes in shape and size typically occur during the first washing or wetting cycle [39]. In this study, such changes are expected as a result of the high hygroscopicity and swelling behaviour of the cotton yarns used [40]. These changes affect not only the yarn dimensions but also the interactions between individual elements of the knit structure and the contact points of the loops. In addition, the loop geometry is altered, primarily due to changes in the ratio between course height and wale density [41,42].

All knitted samples were washed together in a single cycle using a standard domestic washing machine at 40 °C with a liquid detergent. A standard wash, rinse, and spin programme was used, with a spin speed of 900 rpm. No fabric softeners or optical brighteners were applied. Performing a single wash ensured that all samples were subjected to identical conditions, thus supporting the repeatability and comparability of shrinkage measurements. After washing, the samples were left to dry under controlled atmospheric conditions: temperature 22 °C and humidity 60% [43].

To evaluate the dimensional changes, the length and width of the finished samples were measured before and after washing, and the shrinkage (Sh, %) was calculated using the following formula [44]:

$$Sh={\displaystyle \frac{X_{bw}-X_{aw}}{X_{bw}}}100, \%$$

(1)

where X_bw is the parameter (length or width) before washing, cm, and X_aw is the parameter (length or width) after washing, cm.

The shrinkage of each sample was also calculated in terms of area (Sh_area, %) using Equation (2):

$$S{h}_{area}=100\left(1-{\displaystyle \frac{L_{aw}·W_{aw}}{L_{bw}·L_{bw}}}\right), \%$$

(2)

where L_bw is the length before washing, cm; W_bw is the width before washing, cm; L_aw is the length after washing, cm; and W_aw is the width after washing, cm.

3. Results and Discussion

3.1. Design of the Knit Macrostructures

Sample 1 is developed by combining Aran and Cable modules to form a complex embossed structure. The design is divided into four logical sections, each applying specific module combinations to achieve distinct visual patterns.

The detailed breakdown of the design structure is presented in

Table 4. Sample 1: Design sections and applied modules.

Section	Applied Modules and Structure
1	Aran modules “Aran” type mesh, forming multiple diamonds that create “X” patterns arranged in three columns.
2	Cable modules “Cable 3 × 3<” and “Cable 3 × 3>”, forming embossed wave-like lines.
3	Aran modules “Aran 2 × 1 <” and “Aran 2 × 1 >”, resulting in embossed broken lines where the peaks touch, forming diamonds. The pattern continues with wave-like lines, using Cable modules “Cable 3 × 3<” and “Cable 3 × 3>”.
4	Cable modules “Cable 2 × 2<” and “Cable 2 × 2>” to complete the design.

.

A symbolic view of the layout is shown in Figure 3, while the loop simulation of the front-side configuration is illustrated in Figure 4, following finalisation in the Shape Editor module.

Sample 2 is constructed as a fabric composed of alternating modules selected from the Module Explorer of the CMS 503 ki software. The structure is divided into four distinct sections, each employing different types of Cable and Aran modules.

The modular breakdown is presented in

Table 5. Sample 2: Design sections and applied modules.

Section	Applied Modules and Structure
1	Cable modules “Cable 8 × 8<” and “Cable 8 × 8>”, forming a group of embossed wave-like lines.
2	Cable modules “Cable 1 × 1<” and “Cable 1 × 1>”, forming straight embossed lines.
3	Cable modules “Cable 6 × 1<” and “Cable 6 × 1>”.
4	Cable modules “Cable 1 × 1<” and “Cable 1 × 1>”, as well as Aran modules “Aran 2 × 1 <” and “Aran 2 × 1 >”, creating embossed broken lines with touching peaks forming diamonds. “Cable 1 × 1<” and “Cable 1 × 1>” were used to complete the design.

.

A visual representation of the design is shown in Figure 5 (Symbol View) and its corresponding loop structure simulation is presented in Figure 6 (Fabric View).

Sample 3 is constructed by alternating Pointelle modules from the Module Explorer of Database, specifically designed for programming half-loop transfers and creating openwork knit structures. The model consists of four main sections.

The structure of the design is described in

Table 6. Sample 3: Design sections and applied modules.

Section	Applied Modules and Structure
1	Pointelle modules “Pointelle_v_Repeat <=”
2	Pointelle modules “Pointelle_v_Repeat =>”
3	Pointelle modules “Pointelle_^_Repeat <=”
4	Pointelle modules “Pointelle_^_Repeat =>”

, while a symbolic view is presented in Figure 7, and the simulated macrostructure is shown in Figure 8.

Sample 4 shows a design of a scarf featuring the logo “Design and Technologies for Clothing and Textile.” The effect is achieved by alternating knit and purl stitches, forming the text element directly within the knitted surface.

The inscription is generated using the “Text as pattern element” tool in the CMS 503 ki software. From the “Needle Actions—Stitch Lengths” menu, specific stitch types are applied, including “Front Stitch with Transfer” and “Rear Stitch with Transfer”, allowing for accurate and readable lettering.

The simulated loop structure of the design is shown in Figure 9.

Sample 5 is developed as a patterned fabric, composed of alternating Pointelle modules from the Module Explorer of Database. Specifically, the modules “Pointelle_v_Repeat <=” and “Pointelle_v_Repeat =>” are used to generate a variety of geometric shapes, arranged in a specific and repeating sequence.

The symbolic view of the design is presented in Figure 10, while the loop structure simulation is shown in Figure 11, visualising the knitted macrostructure created by the Pointelle configuration.

3.2. Manufacturing of the Knit Macrostructures

Once the design programming is complete, the corresponding machine file is loaded into the CMS 503 ki flat knitting machine. The desired design is selected using the machine’s built-in controller (Figure 12), initiating the production process.

Before knitting begins, the machine performs a series of empty runs to calibrate key mechanisms, including the positioning of the yarn carriers, alignment of the needle beds, and system readiness. These preparatory steps ensure the accurate execution of the programmed pattern. The number of active needles and required rows, defined during the design stage, are automatically recognised during this phase.

If a yarn break occurs during knitting, the system halts automatically and displays a warning message on the controller. Once the process is resumed, the finished knitted pieces are extracted using the take-down system and transported to the collector for knitted items.

After production, visual quality control is performed. The knitted structure is evaluated for conformity with the programmed shapes and dimensions. Any surface irregularities or technical defects are identified at this stage.

The total production time for each of the five samples is summarised in

Table 7. Production time.

Knit Macrostructure	Sample 1	Sample 2	Sample 3	Sample 4	Sample 5
Production time, min	3	9	3	10	5

, and the completed knitted samples are displayed in Figure 13.

3.3. Shrinkage Results and Analysis

The dimensional changes in the knitted samples were evaluated by measuring their length and width before and after the first washing. The process was identical for all samples: machine washing at 40 °C using liquid detergent and a standard wash–rinse–spin cycle, followed by air-drying under controlled ambient conditions (22 °C, 60% humidity).

The length measurements before and after washing are shown in Figure 14, and the width measurements are presented in Figure 15.

To support a more comprehensive interpretation of the results, a correlation analysis was performed between pre- and post-wash dimensions. The findings are illustrated in Figure 16, where blue dots represent length and red squares represent width. The black dashed line (y = x) serves as a reference indicating no dimensional change. Points below the line indicate shrinkage, while those above suggest expansion.

The actual shrinkage percentages for both length and width were calculated using Equation (1) and visualised in

Table 8. Shrinkage of the samples.

Knit Macrostructure	Sample 1	Sample 2	Sample 3	Sample 4	Sample 5
Shrinkage Length, %	26.170	4.76	−6.52	−7.29	11.11
Shrinkage Width, %	36.25	−6.28	−7.50	6.25	10.005
Area Shrinkage, %	30.78	−1.22	−14.51	−0.58	20.00

and Figure 17.

To better capture the combined dimensional changes in both directions, the shrinkage of each sample was also calculated in terms of area (Equation (2)). This approach offers a more comprehensive measure of overall deformation, especially in cases where shrinkage occurs in one direction and expansion in the other. The results, presented in Table 8, show the total percentage of area lost or gained after washing for each structure.

The results highlight significant differences in post-wash characteristics among the samples.

Sample 1 (Sh_av = 30.69%) exhibits the most severe reduction in total fabric area. The high-density Aran and Cable modules introduce substantial internal tension and compact loop structures, which contract heavily when exposed to washing. Such a dramatic loss of area suggests that the structure is not suited for applications where size consistency is critical unless pre-treatment or compensation is applied. It also reflects how dense, textured designs amplify the effects of yarn swelling and loop pullback.

Sample 2 is the most dimensionally stable sample in the group, with its area shrinkage of Sh_av = 30.69%. The near-zero change in area is remarkable given the relative complexity of the symmetrical Cable-Aran layout. It indicates a well-balanced stitch geometry, allowing for even distribution of tension and controlled deformation. Sample 2 could be considered a reference model for structure types that require size retention—e.g., shaped garments, panels, or knit-to-measure applications.

Sample 3 has a negative area shrinkage (Sh_av = −13.64%), which means area expansion after washing. This sample is the least stable. The negative shrinkage reflects a net gain in fabric area—a clear sign of relaxation and expansion. The openwork Pointelle design, while visually delicate and breathable, lacks the structural tension needed to resist dimensional spreading after washing. This behaviour compromises fit and consistency and may require stabilising strategies such as tighter gauge or additional reinforcing rows.

The area shrinkage of Sample 4 is Sh_av = 0.01%. Although the overall area remains virtually unchanged, this is a mathematically neutral result masking physical asymmetry. The sample expands in length and shrinks in width, which can distort the overall proportions of the design. This is particularly relevant for structured motifs such as text, where geometric fidelity is critical. In applications such as scarves, where visual and symmetrical form are important, this asymmetrical change may disrupt the proportions of the text motif, lead to distortion of the letters, and finally render the product visually unsatisfactory. Thus, area stability does not always equate to shape stability—a key distinction in textile design.

Sample 5 presents a significant (Sh_av = 20.0%) but symmetric shrinkage. The deformation is balanced between length and width (11.11% and 10.00%, respectively), which helps preserve the original shape, even as the overall size is reduced. This makes Sample 5 suitable for products where proportional scaling is acceptable (e.g., scarves, panels), but dimensional predictability is still required.

Area shrinkage is not just a summary metric—it provides meaningful insight into how evenly and predictably a structure responds to washing. Some samples may lose area but maintain shape, while others retain area but deform visibly. For intelligent textile design, both numerical shrinkage and deformation symmetry must be considered in the evaluation of post-production behaviour.

3.4. Discussion

This study follows a practical evaluation methodology rather than a predictive modelling approach. By physically producing and analysing five distinct knitted designs, the work explores the structural responses of complex knit macrostructures under standardised washing conditions, highlighting practical insights for intelligent textile manufacturing.

A comparative summary of the key characteristics of all five samples, including production parameters and shrinkage performance, is provided in

Table 9. Summary of structural and performance characteristics of the samples.

Sample	Main Modules Used	Visual Complexity	Production Time (min)	Shrinkage Length (%)	Shrinkage Width (%)	Area Shrinkage (%)	Dimensional Stability	Notes
1	Aran + Cable	High	3	26.17	6.25	30.69	Low	Dense structure, vertical shrinkage
2	Cable + Aran (symmetrical)	Moderate	9	4.76	−6.28	1.22	High	Most stable overall
3	Pointelle (openwork)	Moderate	3	−6.52	−7.50	−13.64	Low	Structure expands after wash
4	Text motif (knit/purl)	Moderate–High	10	−7.29	6.25	0.01	Low–Moderate	Unstable in length
5	Pointelle (geometric)	Moderate	5	11.11	10.00	19.99	Moderate	Balanced shrinkage, good control

.

The five knitted samples demonstrate how structural design, module selection, and stitch configuration directly affect production parameters and post-wash dimensions.

Sample 1 incorporates complex Aran and Cable modules, forming highly embossed patterns with dense loop formations. The sample exhibits the highest shrinkage in length (26.17%), which can be attributed to the structural tension resulting from the compact stitch configuration. The relatively short production time (3 min) reflects the linear construction of the design, despite its visual complexity, and is not associated with the observed shrinkage. The moderate shrinkage in width (6.25%) indicates that the deformation is predominantly vertical. These findings suggest that visually rich and dense macrostructures may require compensation in product sizing or the use of alternative yarns to achieve greater dimensional stability.

Sample 2, also based on Cable and Aran elements but arranged more symmetrically and with consistent density, shows minimal shrinkage in length (4.76%) and expansion in width (−6.28%). Although it has the longest production time (9 min) due to the repetition of symmetrical cable elements, its post-wash behaviour indicates high structural stability. This balance between complexity and consistency makes it a promising option for precise, size-dependent textile products.

Sample 3 employs Pointelle modules, generating a lightweight openwork structure through frequent half-loop transfers. The sample shows negative shrinkage (−6.52% in length, −7.50% in width), which in this context indicates dimensional expansion. This is probably due to internal relaxation of the structure during washing. With its low production time (3 min) but poor post-wash stability, such a structure may require reinforcement or pre-treatment when used in functional garments.

Sample 4 includes a textual motif within the scarf structure, created through alternating knit and purl stitches. The pattern results in length expansion (−7.29%) and moderate shrinkage in width (6.25%), revealing asymmetrical dimensional properties. The longest production time (10 min) is expected due to the added complexity of stitch sequencing and motif placement. This design approach is suitable for visual or branding elements but may need structural compensation or stabilisation to ensure size retention.

Sample 5 is developed using Pointelle modules in geometric arrangements. It displays balanced and moderate shrinkage: 11.11% in length and 10% in width, with a medium production time (5 min). The symmetry of the design and consistent loop repetition may contribute to its structural predictability. Among the five, Sample 5 offers a reasonable trade-off between design interest, production efficiency, and post-wash dimensional reliability.

In summary, the analysis reveals the following:

• Dense structures (Sample 1) are prone to vertical shrinkage;
• Balanced, repetitive structures (Sample 2) perform best in terms of dimensional stability;
• Openwork and decorative designs (Samples 3 and 4) may expand or distort due to loop relaxation;
• Geometric pointelle structures (Sample 5) can offer stable shrinkage ratios if well distributed.

These findings demonstrate the critical role of design logic and module type in smart and sustainable knitted textile development.

Future research may investigate the behaviour of complex knitted macrostructures under repeated washing cycles and mechanical stress using area shrinkage and geometric deformation as indicators of long-term dimensional stability. In addition, the incorporation of specific yarn types (e.g., elastane, high-twist, or recycled fibres) could be explored to optimise both dimensional resistance and functional performance.

Further studies could also focus on the role of artificial intelligence and machine learning in the automatic generation of sustainable design alternatives, structural optimisation, or prediction of production outcomes based on selected materials.

The application of digital twin technologies, supported by advanced simulation software, also presents a promising direction for quality control, predictive evaluation of product behaviour, and reduction in production waste.

It is important to acknowledge that while simulation tools such as M1plus V7.5 provide valuable predictive capabilities, they cannot fully replicate the complex interactions between yarn properties, stitch tension, and real-world finishing processes. A mismatch often arises between the simulated visual/textile structure and the final manufactured result, particularly in complex designs involving loop transfers or textured modules. This discrepancy can impact industrial adoption, as manufacturers must calibrate design parameters manually through sample testing. Further advancements in material behaviour modelling and digital twin technologies are required to bridge this gap and support fully predictive virtual development workflows.

4. Conclusions

This study examines the role of automation in the design and production of complex knitted structures, focusing on the use of the specialised M1plus V7.5 software and computerised flat knitting machines. Through the development and physical manufacturing of five experimental designs, the research demonstrates how digital tools support precise loop formation, efficient programming, and high repeatability in production with minimal human intervention.

While it is well known that macrostructure geometry affects post-wash deformation, this study offers a detailed comparison of five structurally diverse designs and demonstrates how area shrinkage analysis can reveal asymmetries not evident through linear measurements. Sample 2 showed the highest dimensional stability, suggesting potential for reducing material waste in knit-to-measure production where size accuracy is critical. In contrast, Samples 1 and 5 exhibited significant shrinkage, while Samples 3 and 4 experienced elongation in length. The deformation observed in Sample 4 indicates that additional design compensation may be necessary in applications where visual alignment is essential, such as branded knitwear incorporating text or logos. Area shrinkage proved to be a more comprehensive indicator of structural deformation, particularly in cases involving asymmetric dimensional change.

This study focuses on experimental evaluation rather than predictive simulation. Although M1plus offers certain simulation capabilities, its accuracy in forecasting shrinkage—particularly in complex structures—is limited. Nevertheless, the results contribute a practical methodological framework for optimising design parameters within the context of intelligent textile manufacturing. The integration of CAD/CAM software significantly reduces trial-and-error during development, shortens production cycles, and minimises material waste, in alignment with smart manufacturing objectives.

Future research may investigate the behaviour of such structures under repeated washing and mechanical stress, as dimensional changes may continue to evolve over multiple laundering cycles. Area shrinkage and structural distortion can serve as indicators of long-term durability and product performance. The inclusion of specific yarn types—such as elastane, high-twist, or recycled fibres—may support the development of more resilient and functional knits. Furthermore, the use of artificial intelligence and machine learning for structural optimisation, as well as the implementation of digital twin technologies for predictive modelling and quality control, represent promising directions for the advancement of intelligent textile manufacturing.