Comfort Prediction Model for Sports Car Seats

Abstract

Automotive seat comfort is a critical factor in enhancing driver satisfaction, especially in sports cars, where design must balance comfort features and performance-oriented features like lateral containment and anti-submarining. This study adopts an empirical-analytical approach for assessing and modelling perceived comfort in sports car seats using both objective and subjective data. A total of 64 participants (50 males, 14 females) evaluated two types of sports car seats—a road model (SEAT A) and a racing model (SEAT B)—during 15-min driving simulations using a dynamic simulator equipped with a full-body pressure mat (XSENSOR X3 PRO). Comfort was assessed through a postural comfort questionnaire using 10-point Likert scales. Statistical analysis revealed significant correlations between anthropometry, pressure distribution, and perceived comfort. In light of the correlation analysis, regression models were developed for four anthropometric percentile clusters (0–25th, 25–50th, 50–75th, 75–100th). Models were validated (accuracy > 75%) and one of them (named Model III) achieved accuracies of 95%, 96%, 90%, and 97% for its percentile clusters. The proposed models offer actionable insights for tailoring sports car seats to different user percentiles, enabling more personalized and effective seat designs that enhance both performance and comfort.

1. Introduction

There were about 249 million cars on the EU roads in 2023; on average, a person spends four years and one month in a vehicle during his/her lifetime [1]. This extended time spent in cars is typically characterized by static posture and isometric muscle contractions, resulting from the continuous need to reach commands, steering wheel, and pedals, while forwards to retrieve visual information from the environment and monitoring the conditions of the traffic and the road [2]. In this context, the car seat plays a pivotal role, not only as a physical support but also as an active interface between the occupant and the vehicle.

The challenge of creating comfortable seats has become increasingly complex due to the advancement of the automotive industry and the many factors involved in the human–seat car interaction [3,4,5,6,7,8,9,10].

Ergonomic research consistently highlights the importance of designing car seats based on the anthropometric variability of the user population to ensure optimal ergonomics, safety, and comfort [11]. A well-established practice in ergonomics is the use of stature- and body-dimension-based percentiles, commonly the 5th, 50th, and 95th percentiles [12]. These percentiles help accommodate a wide range of users, from the smallest to the largest, by guiding the design of key seat dimensions such as seat height, depth, backrest height, and seat width.

While sitting, the human body is in direct contact with the seat. Many researchers assume that the pressure distribution is the best objective measure for comfort. Less discomfort and higher comfort are related to a lower average pressure, a lower peak pressure, and a larger contact area(s) of the pressure map [13,14,15,16]. Moreover, posture and seat design have been shown to significantly influence interface pressure patterns and perceived comfort [17]. About the pressure map acquisition, the pressure mats are amply used because they are able to capture the interface pressure profile between the body and the seat [18]. Akgunduz, Rakheja, and Tarczay [19] also found correlations between perceived comfort and the peak and average pressures on the seat cushion.

Several scientific studies have emphasized the importance of dividing the human–seat interface into subareas to gain a more detailed understanding of pressure distribution and its impact on seating comfort. Mergl et al. [20] developed a widely adopted method that segments the contact surface into 17 subareas—11 on the seat cushion and 6 on the backrest—allowing localized analysis of pressure peaks and their correlation with discomfort. Similarly, Laintoine et al. [14] evaluated three different seats on driver’s sitting behavior and perceived discomfort by using two pressure mats and dividing the human-seat contact area into 10 regions. Zenk et al. [21] and Fiorillo et al. [22] also utilized subdivided contact maps to evaluate comfort across distinct anatomical zones, enabling more precise ergonomic interventions. These approaches consistently demonstrate that regional pressure metrics provide valuable insights for improving automotive seat design and optimizing comfort for a diverse user population [23,24,25].

Generally, two main categories of automotive seats are recognized: racing seats and utility (or commercial) seats. While both types must meet fundamental safety and comfort standards, some differences are notable. Racing seats are engineered for high-performance driving, offering enhanced lateral support, anti-submarining geometry, and rigid structural materials to ensure occupant stability and safety under extreme accelerations and cornering loads [2,26]. In contrast, utility seats—commonly found in commercial or passenger vehicles—emphasize comfort, adjustability, and ergonomic support. They are typically designed using softer foam densities, wider seating surfaces, and multi-directional adjustments to accommodate a broader range of body types and minimize driver fatigue during prolonged use [27,28,29]. Additionally, seat cover materials and surface properties can significantly affect comfort perception, particularly in long-term use and luxury applications [30].

Among factors that drive the car manufacturers in seat design, comfort is the one that highly influences drivers’ and passengers’ experience, and the main goal for designers is to find effective methods to assess the subjective experience [31]. In the field of Ergonomics and Human Factors, questionnaires are a crucial tool for acquiring subjective opinions. These are used in different ways, such as to analyze the level of user satisfaction, to test prototypes, to compare products, or to obtain information from a large sample of the population. To evaluate the comfort of the seats, Helander and Zhang [32] used subjective evaluation tools such as the Chair Evaluation Checklist, the Body Part Discomfort, and the General Comfort Rating [33].

In the design process both these aspects have to be considered: the objective measurements, such as pressure distribution, have to be linked to subjective perceptions of comfort. The dual requirement of performance and comfort creates a particularly complex design challenge, especially for sports car seats, where aesthetic and lateral containment needs must be balanced with perceived driving comfort [34,35].

All the mentioned factors are widely discussed in the literature, but to date, no criterion reports all of them in a comfort assessment model. The purpose of this paper is the development of predictive comfort-assessment criteria for sporty cars, having as input the seat interface pressure measurements and the occupant anthropometry.

The main purposes of the study are:

• Define a methodology to acquire empirical data (objective and subjective) related to the comfort of sports car seats.
• Develop a mathematical model capable of predicting car seat comfort levels by correlating and estimating the influence of human–seat interaction factors (such as seat interface pressure measurements, occupant anthropometry, and subjective evaluations of postural comfort) in the context of sports cars.

In this framework, the study adapts and refines existing comfort assessment models, originally developed for conventional automotive seats, to the specific characteristics of sports car seats. The proposed approach aims to establish a stronger link between subjective comfort evaluations and objective pressure distribution data, providing a methodological contribution tailored to the analysis of comfort in sports car seats. This adaptation seeks to improve the understanding of how physical measurements relate to perceived comfort in a context where empirical evidence is still limited.

2. Materials and Methods

2.1. Setup

Sports car seats differ from traditional car seats mainly due to their performance-oriented design. While traditional seats are made for comfort and practicality during everyday driving, sports seats are designed to provide greater lateral support and stability, minimizing body movement during tight turns and intense acceleration. Sports seats typically feature reduced padding and a more rigid structure, unlike traditional seats that focus on comfort with softer materials. Additionally, the driving position in sports seats is more inclined to enhance stability during high-speed maneuvers, while traditional seats are designed for a more relaxed and comfortable posture. In short, sports seats sacrifice some comfort to ensure superior performance and control, whereas traditional seats prioritize comfort for daily driving.

The SDS (Static Driving Simulator) inside the “Human Centred Design and Vehicle Design by Simulation” laboratory at the Industrial Engineering Department of the University of Salerno, was used for the tests. The driving simulator was composed of a seating buck, a driving simulation software, and a pressure mat.

The seating buck was designed as a modular and adjustable structure by using Bosch^® aluminum profiles, connections, and fasteners. This structure lodged a steering wheel with sequential gear shift paddles and pedals, allowing users to interact with the driving controls as in a real car. Even fully adjustable, during tests, people were allowed only to change the horizontal position of the seat and the angle of the backrest (in a limited way).

For this study, two types of sports car seats were selected: a road-oriented model (SEAT A) and a racing-oriented model (SEAT B). The choice of these two models reflects the dual nature of sports car usage—everyday driving versus high-performance applications.

SEAT A: Road-Oriented Sport Seat. This seat represents a typical configuration found in production sports cars designed for both performance and daily usability. It features soft foam padding, wide lateral supports, and a pronounced lumbar bulge. Such seats aim to strike a balance between ergonomic comfort, long-distance support, and moderate lateral containment.

SEAT B: Racing-Oriented Sport Seat. This seat typifies the design philosophy of racing and track-focused vehicles, where the priority shifts toward maximizing body restraint, minimizing lateral movement, and enhancing driver-vehicle feedback. It is characterized by a rigid, fiberglass shell structure, minimal padding, more pronounced lateral supports, and a firm backrest with an upright angle, built to ensure maximum containment and reduced body movement during intense maneuvers. Both seats are covered in similar synthetic fabric for consistency.

To simulate the driving experience, “Assetto Corsa ^®, v1.16” software (a commercial software developed by Kunos Simulazioni, ACI Vallelunga Circuit, Rome, Italy), a racing simulator connected to a monitor to visualize the route, was used. The selected simulator was chosen for its ability to accurately reproduce driving dynamics in real-world scenarios, including aspects such as vehicle behavior, response to controls, and interaction with the surrounding environment. The simulator is equipped with advanced realism features, including high first-person visual and sound quality, which allow it to faithfully recreate everyday driving conditions.

The pressure mat “XSENSOR X3 PRO^®” (a commercial instrument developed by XSENSOR Technology Corporation, Calgary, AB, Canada) was used to obtain the pressure distribution of the participants during the experiments [36]. The array, manufactured with Foresite SS technology, consists of 64 × 160 PX100 pressure sensors, each less than 1 mm thick. The pressure for each sensor was reported in N/cm². The proprietary acquisition system and software captured pressure maps at up to a sampling rate of 27 Hz, pressure range 0.07–2.7 N/cm², and measurement threshold: 0.7 kPa. The mat was placed on the driving seat, covering the whole area from the head to the lower thighs. Due to its flexibility and minimal thickness (0.8 mm), it was quite unperceivable by the subjects. Even the external texture and the friction are very similar to the seat covering one. To ensure stability, the sensor matrix was securely attached to the seat, preventing any relative movement between the seat and the mat (Figure 1).

2.2. Participants

A total of 64 participants (50 males and 14 females) were recruited to perform the tests. Since the scientific literature has demonstrated that the simultaneous use of percentile-based anthropometric data and pressure distribution measurements is essential for ergonomically and comfortably seat design [4], percentiles were calculated for each subject for use in subsequent analyses. The percentile calculation based on subjects’ stature was computed using the DELMIA^® (a commercial software developed by Dassault Systèmes, version V5-6R2017, Vélizy-Villacoublay, France) mannequin [12], considering a European population. Participants’ demographic characteristics are reported in

Table 1. Demographic data of the participants at the experiment.

	Mean	Std. Deviation	Minimum	Maximum
Age (years)	24	2.246	19	30
Mass (kg)	72	12.419	46	107
Stature (mm)	1730	84.410	1500	1950
Stature percentile	50	28.557	2.100	99.990

. The inclusion criteria were as follows: (1) Participants were required to have a valid driver’s license and to have engaged in regular driving activity within the last year; (2) participants were required to have completed one pilot test on the same simulator; (3) people have been included only if they were not affected by musculoskeletal disorders; (4) people suffering from sickness due to the simulator have not participated in the test (participants could stop the test at any time).

2.3. Experimental Protocol

Before starting the test, participants were informed about the test procedure but not about the exact aim of the study, to mitigate the potential influence of psychological biasing effects [37]. Their written consent, in accordance with ethical standards of the University of Salerno [38], was obtained.

Subsequently, each participant experienced each seat (SEAT A and SEAT B) through the Latin square method [39]. Specifically, a 2 × 2 Latin square design (

Table 2. 2 × 2 Latin square design for assigning the order of tests in the experiment.

	Test 1	Test 2
Even number tester	SEAT A	SEAT B
Odd number tester	SEAT B	SEAT A

) was employed. The Latin square method ensured that the order did not systematically influence perceived comfort and helped control for interindividual differences in the experimental sequence. To further minimize order and fatigue effects, a rest period of approximately 10 min was provided between the two simulations, allowing participants to recover before proceeding to the next test [40].

Simulation A (associated with SEAT A): Participants drove along a road route with a vast road and few wide curves. The simulation induced a relaxed driving style and therefore infrequent and large movements while driving. This test represents well a sports car on an urban route.

Simulation B (associated with SEAT B): Participants drove along a GTE (Gran Turismo Endurance) track, on a narrow road and numerous tight curves. The simulation induced a sporty driving style and therefore limited and frequent movements while driving. This test was used to test the variability of driving style.

Before starting the simulations, each participant adjusted their seating position, moving the seat toward or away from the steering wheel and changing the incline of the backrest. This was necessary because, given the different statures, the same configuration would not allow everyone to drive correctly and comfortably. The duration of the simulations was exactly 15 min each (the software forces the finish at a given time) to simulate a short trip. Fifteen minutes is about the average daily urban trip travel time in Italy, considering an average car speed of 48 km/h and 11.4 km as the average distance per person per day [41]. Furthermore, for sports cars, some manufacturers have informed us about the average usage of this type of car. Excluding people that use this kind of car for professional use (championship or professional riding) and for the so-called “cavalcades”, the average use is very short in time; according to insurance data aggregated by Hagerty Media [42], Ferraris’ average mileage is about 1814 miles (≈2920 km) per year for newer models, while according to “Motor and wheels” [43], the Ferraris are often used as “special occasion” vehicles, with typical annual usage of ~2000–10,000 miles (~3200–16,000 km) depending on the owner. Doing an average among sources and considering 100 hours/year, Ferraris are driven for about 16 min/day. Thus, 15 min is very well representative of a sports car. Tests of longer duration, which were useful for analysis of fatigue-related discomfort, were excluded to maintain uniformity among participants by preventing cumulative fatigue from affecting perceptions in the subsequent simulation.

2.4. Questionnaire

All participants, at the end of each simulation, filled out a questionnaire to record their subjective perceptions. The questionnaire, prepared using the Google Forms platform, was divided into two sections.

The first section focused on gathering demographic information (nationality; age; gender; weight and stature, both measured), obtained through an open-ended question, and personal driving habits (how often, in a week, participants use the car and for how long and whether as drivers or passengers), collected through a multiple-choice question.

The second section focused on assessing postural comfort experienced during the simulations. Subjective comfort data were acquired using structured 10-point Likert scales, which ensured high sensitivity to perceived comfort (ranging from 1 = minimum comfort to 10 = maximum comfort) [44,45]. This methodology aligns with validated protocols from previous studies using body part-specific comfort assessments, ensuring both internal consistency and relevance for seat evaluation [46,47]. Specifically, the tester was asked to rate the following:

• Overall perceived comfort
• Comfort related to seat cushion and backrest (Figure 2a)
• Comfort of specific body parts (Figure 2b), based on the body region framework proposed by Mergl [23], which divides the seat–user interface into 17 areas covering buttocks, thighs, and the back.

2.5. Pressure Acquisition

Pressure maps have been acquired through the XSENSOR X3 PRO^® pressure mat that provides, as output *.CSV files.

Participants drove for 15 min during which 900 pressure maps (each map is obtained as the average of 27 acquired maps per second) were acquired (Figure 3a).

The pressure maps were subsequently processed as follows:

• The 900 frames were merged to obtain only one map, which is the averaged pressure map (Figure 3b).
• The average pressure map was divided into 17 areas, as the Mergl model suggests (Figure 4).
• For each area, the following parameters were collected and calculated: contact area, average pressure, peak pressure, and load percentage.
• The same parameters, calculated for each area, were calculated for the whole backrest, for the whole seat cushion, and the entire seat.

3. Data Analysis and Results

To ensure the reliability of the questionnaire, the Cronbach-α reliability coefficient has been calculated through IBM SPSS Statistics^® software, version 26 (a commercial software developed by IBM, Armonk, NY, USA), which produced a value of 0.781, indicating good internal consistency.

Data were collected for two series of tests regarding SEAT A and SEAT B.

Table 3. Statistical data analysis.

	Average Overall Comfort	Standard Deviation
Seat A	8.27	1.36
Seat B	8.25	1.48
Seat |A−B|	0.02	0.12

shows the mean and the variance of the comfort scores reported for the two test campaigns. The results confirmed the absence of significant differences between the two groups (the difference between the mean was 0.02 and between the variance was 0.12), so the data can be considered as a single dataset, thus doubling the sample size (64 participants × 2 = 128).

A preliminary ANOVA analysis was conducted to examine the effect of the independent variable “gender” on the dependent variable “comfort”, due to gender imbalance. The results showed a p-value of 0.164 for SEAT A and a p-value of 0.126 for SEAT B, so, although the sample was not gender-balanced, the absence of a significant difference suggests that, in this experiment, gender did not have a significant impact on perceived comfort.

3.1. Correlation Analysis

Subjective data (comfort perceived obtained through the questionnaires) and objective ones (contact area, peak and average pressure obtained through the pressure map) were processed on IBM SPSS Statistics^® software, version 26. Through this software, the correlation calculation was performed to check eventual relations among the data. As the data were not normally distributed, Spearman’s index was used. In

Table 4. Correlation data. Correlations are significant (two-tailed) at the 0.05 level (*) and at the 0.01 level (**).

Variable 1	Variable 2	Correlation Coefficient	Significance Level
Stature percentile	Overall perceived comfort	−0.256	**
Stature percentile	Comfort related to backrest	−0.252	**
Back length	Overall perceived comfort	−0.223	*
Back length	Comfort related to backrest	−0.183	*
Chest width	Comfort related to backrest	−0.177	*
Thigh length	Comfort related to seat cushion	−0.176	*
Comfort related to backrest	Overall perceived comfort	0.810	**
Comfort related to seat cushion	Overall perceived comfort	0.435	**
Seat contact area	Overall perceived comfort	0.215	*
Backrest contact area	Overall perceived comfort	0.210	*
Zone 6 contact area	Overall perceived comfort	0.197	*
Seat peak pressure	Comfort related to seat cushion	−0.239	**
Seat cushion peak pressure	Comfort related to seat cushion	−0.267	**
Zone 8 peak pressure	Comfort related to seat cushion	−0.178	*
Zone 9 peak pressure	Comfort related to seat cushion	−0.177	*
Zone 11 peak pressure	Comfort related to seat cushion	−0.178	*

are reported the most significant correlations resulting from the software data processing (between objective and subjective data).

The significant correlations between objective parameters (such as contact area and pressure) and perceived comfort and, notably, the correlation between stature percentile and comfort, justified the clustering of the sample into four stature percentile-based groups: 0–25th, 25–50th, 50–75th, and 75–100th. This approach allows for a more detailed investigation of how body size influences comfort perception and enables the development of more targeted comfort models. Similarly, variables related to pressure distribution, in particular contact area and peak pressure, were found to be influencers of comfort perception, reinforcing the idea that a well-distributed pressure profile contributes to a higher comfort feeling [8,27]. Furthermore, for the sports car seats, the backrest appears to have a greater impact than the seat cushion itself, highlighting that this area exerts a stronger influence on overall comfort perception. This is due to the role of the backrest in providing upper body support and maintaining posture during dynamic driving [28,29,31]. In comparison, standard car seats typically prioritize comfort through softer padding and more generalized support, without offering the same level of lateral containment or rigidity in the backrest. This makes them more suitable for moderated driving, while sports car seats are specifically designed to enhance stability and anti-submarining during high-performance driving, where precise body support is essential.

In

Table 5. Correlation data of the sample 0–25th stature percentiles. Correlations are significant (two-tailed) at the 0.05 level (*) and at the 0.01 level (**).

Variable 1	Variable 2	Correlation Coefficient	Significance Level
Seat contact area	Overall perceived comfort	0.401	*
Backrest contact area	Overall perceived comfort	0.540	**
Backrest contact area	Comfort related to backrest	0.576	**
Zone 3 contact area	Overall perceived comfort	0.353	*
Zone 4 contact area	Comfort related to backrest	0.436	*
Zone 5 contact area	Overall perceived comfort	0.363	*
Zone 6 contact area	Overall perceived comfort	0.508	**
Seat peak pressure	Overall perceived comfort	−0.384	*
Seat peak pressure	Comfort related to seat cushion	−0.722	**
Seat cushion peak pressure	Overall perceived comfort	−0.405	*
Seat cushion peak pressure	Comfort related to seat cushion	−0.549	**
Zone 8 peak pressure	Overall perceived comfort	−0.421	*
Zone 8 peak pressure	Comfort related to seat cushion	−0.562	**
Zone 9 peak pressure	Overall perceived comfort	−0.559	**
Zone 9 peak pressure	Comfort related to seat cushion	−0.471	**
Zone 5 average pressure	Overall perceived comfort	−0.383	*

, the most significant correlations for the 0–25th stature percentile sample are reported. Stature percentiles 0–25 correlation analysis shows results consistent with the general trends observed in the entire sample, confirming the robustness of the results. In addition, some areas of the seat have a significant impact on the comfort perception. In particular, regarding the contact areas, Zone 6 (the lateral support of the back) emerges as the most influential, presenting a notable correlation with comfort. On the other hand, for pressure, Zones 8 and 9 (respectively, the buttocks and upper thighs) have a stronger effect on comfort perception compared to other zones. These three zones should be given special attention in seat design to maximize comfort for individuals in this stature percentile range.

The analysis performed for the 25–50 stature percentile group, although showing fewer significant correlations compared to the other groups, aligns well with the general trends observed in the entire sample, confirming the consistency of the results (

Table 6. Correlation data of the sample 25–50th stature percentiles. Correlations are significant (two-tailed) at the 0.05 level (*) and at the 0.01 level (**).

Variable 1	Variable 2	Correlation Coefficient	Significance Level
Comfort related to backrest	Overall perceived comfort	0.907	**
Comfort related to seat cushion	Overall perceived comfort	0.507	**
Zone 11 contact area	Comfort related to seat cushion	0.326	*
Seat-cushion peak pressure	Comfort related to seat cushion	−0.340	*
Zone 11 average pressure	Overall perceived comfort	−0.430	**

). In this group, Zone 11, corresponding to the lateral thighs, emerges as the most significant area, with both contact area and average pressure showing notable correlations with comfort. Additionally, the strong correlation between backrest comfort related to backrest and overall perceived comfort reinforces the consistent importance of back support across stature percentiles, in line with the general trend.

For the 50–75th stature percentile group, the correlation analysis shows strong results that match with the general trends observed in the overall sample (

Table 7. Correlation data of the sample 50–75th stature percentiles. Correlations are significant (two-tailed) at the 0.05 level (*) and at the 0.01 level (**).

Variable 1	Variable 2	Correlation Coefficient	Significance Level
Comfort related to backrest	Overall perceived comfort	0.755	**
Seat contact area	Overall perceived comfort	0.585	**
Backrest contact area	Overall perceived comfort	0.559	**
Backrest contact area	Comfort related to backrest	0.430	*
Zone 7 contact area	Overall perceived comfort	0.398	*
Zone 7 contact area	Comfort related to backrest	0.496	**
Seat-cushion peak pressure	Comfort related to seat cushion	−0.473	*
Zone 7 peak pressure	Comfort related to backrest	−0.472	*
Zone 11 peak pressure	Overall perceived comfort	−0.505	**
Zone 11 peak pressure	Comfort related to seat cushion	−0.430	*
Seat average pressure	Overall perceived comfort	−0.430	*
Seat average pressure	Comfort related to backrest	−0.454	*
Zone 7 average pressure	Comfort related to backrest	−0.467	*
Zone 8 average pressure	Comfort related to seat cushion	−0.488	*
Zone 9 average pressure	Comfort related to seat cushion	−0.610	**
Zone 11 average pressure	Comfort related to seat cushion	−0.426	*

). In this range, Zone 7, corresponding to the pelvic girdle, stands out as the most significant in terms of contact area, with a notable correlation with comfort. Regarding pressure, both Zone 9 (upper thighs) and Zone 11 (lateral thighs) emerge as the most impactful, with peak pressure and average pressure in these zones showing significant correlations with comfort.

In the case of the 75–100th stature percentile range, the correlation analysis confirms consistency with the general trends observed across the entire sample (

Table 8. Correlation data of the sample 75–100th stature percentiles. Correlations are significant (two-tailed) at the 0.05 level (*) and at the 0.01 level (**).

Variable 1	Variable 2	Correlation Coefficient	Significance Level
Comfort related to backrest	Overall perceived comfort	0.903	**
Zone 7 contact area	Overall perceived comfort	0.383	*
Seat cushion contact area	Comfort related to seat cushion	0.475	**
Zone 8 contact area	Comfort related to seat cushion	0.600	**
Zone 10 contact area	Comfort related to seat cushion	0.503	**
Seat peak pressure	Overall perceived comfort	−0.458	*
Seat peak pressure	Comfort related to backrest	−0.520	**
Backrest peak pressure	Comfort related to backrest	−0.462	*
Zone 5 peak pressure	Comfort related to backrest	−0.408	*
Zone 5 average pressure	Overall perceived comfort	−0.470	**
Zone 8 average pressure	Comfort related to backrest	−0.386	*

). Within this range, zones 8 and 10 stand out as the most significant contact areas, showing a particularly strong correlation with comfort. Regarding pressure, zone 5 should be closely examined, as both peak and average pressures in this zone have a notable impact on comfort perception.

3.2. Data Processing

The correlation analyses highlighted important relationships between the observed parameters (pressure, anthropometry, comfort), paving the way for the development of comfort evaluation models. These models were developed by considering, for the subjective aspects, the overall comfort perception offered by the two tested seats, and for the objective aspects, the contact area and pressures at the user–seat interface. The samples were grouped according to the significant correlation observed between stature percentiles and overall perceived comfort. This grouping was done to obtain more homogeneous data sets, in which the relationships between the variables are more stable and predictable.

The models were developed in MATLAB^®, version R2024a (a commercial software developed by MathWorks, Natick, MA, USA), using a multiple-stepwise nonlinear regression approach. The most influential variables, as identified through correlation analysis, were included in the regression process. For each variable, the best mathematical formulation was determined, considering linear and nonlinear terms (e.g., first-order, squared, or cubic powers, as well as interaction terms such as products with other variables or their squared values, etc.). The general structure of the regression models adopted in this work can be expressed as (1):

$$\mathrm{C}\mathrm{o}\mathrm{m}\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{t}=\mathrm{a}+{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{b}}_{\mathrm{i}}\cdot {\mathrm{f}}_{\mathrm{i}}\left({\mathrm{x}}_{\mathrm{i}}\right),$$

(1)

where ${\mathrm{x}}_{\mathrm{i}}$ represents the selected variables (e.g., average pressure, contact area, peak pressure in specific zones, or subject stature percentile), and ${\mathrm{f}}_{\mathrm{i}}\left({\mathrm{x}}_{\mathrm{i}}\right)$ corresponds to their best-performing mathematical transformation, including linear, squared, cubic, or interaction terms. Each model (I, II, III) differs in the specific variables and transformations employed, according to the subset of data and the relevant stature percentile group. To ensure the robustness and predictive ability of the model, the data were divided into two sets: 75% was used for model training, while the remaining 25% was reserved for validation. This procedure allowed the performance of the model to be evaluated on independent data, reducing the risk of overfitting. At each iteration, the accuracy error was computed to fine-tune the regression and optimize model performance.

Model I: Incorporates average backrest and seat-cushion pressure, backrest and seat-cushion peak pressure, and stature percentile.

• 0–100th stature percentile: accuracy 22%
• 0–25th stature percentile: accuracy 99%
• 25–50th stature percentile: accuracy 98%
• 50–75th stature percentile: accuracy 72%
• 75–100th stature percentile: accuracy 83%

Model II: Utilizes backrest and seat-cushion contact areas, along with average backrest and seat-cushion pressure

• 0–100th stature percentile: accuracy 26%

Model III: Employs seat contact area, average backrest and seat-cushion pressures, and the peak pressure from the zones that were found to be most impactful during the correlation analysis, such as zone 8 in the 0–25 sample or zone 11 in the 25–50 sample.

• 0–100th stature percentile: accuracy 35%
• 0–25th stature percentile: accuracy 95%
• 25–50th stature percentile: accuracy 96%
• 50–75th stature percentile: accuracy 90%
• 75–100th stature percentile: accuracy 97%

An example ANOVA table for one of the models (Model I 75–100th stature percentile) is shown in

Table 9. ANOVA table of the Model I sample 75–100th stature percentiles.

Source	SS	df	MS	F	p-Value
Model	44.188	19	2.326	2.757	0.0442
Residual	9.279	11	0.844
Total	53.467	29

. Other models showed similar trends in terms of model significance and explained variance.

The ANOVA confirms the model’s statistical significance (p < 0.05), with a good balance between explained variance and model complexity.

The greater accuracy obtained in the clustered samples than in the whole sample can be interpreted as a direct consequence of the greater homogeneity within the groups. In fact, models applied to homogeneous samples are able to make more accurate predictions because the characteristics of the data are less variable.

Compared to Models I and II, Model III achieves a predictive accuracy of no less than 90% while maintaining a limited number of input variables, thus ensuring a good balance between model complexity and performance (Figure 5).

The general model can be expressed as (2):

$$\mathrm{C}\mathrm{o}\mathrm{m}\mathrm{f}\mathrm{o}\mathrm{r}\mathrm{t}={\mathrm{a}}_1+{\mathrm{b}}_1·{\mathrm{A}}_{\mathrm{s}\mathrm{e}\mathrm{a}\mathrm{t}}+{\mathrm{b}}_2·{{\mathrm{p}}_{\mathrm{a}\mathrm{v}\mathrm{g}}}_{\mathrm{b}\mathrm{a}\mathrm{c}\mathrm{k}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{t}}+{\mathrm{b}}_3·{{\mathrm{p}}_{\mathrm{a}\mathrm{v}\mathrm{g}}}_{\mathrm{s}\mathrm{e}\mathrm{a}\mathrm{t}-\mathrm{c}\mathrm{u}\mathrm{s}\mathrm{h}\mathrm{i}\mathrm{o}\mathrm{n}}+\sum _{\mathrm{i}=4}^{\mathrm{n}}{\mathrm{b}}_{\mathrm{i}}\cdot {{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_{\mathrm{i}},$$

(2)

For each percentile cluster, the model structure remains similar to the general one, but in addition to variations in coefficient values, the summation term includes the peak pressure of the body region most influential for each case, as identified through correlation analysis of comfort perception, as summarized in

Table 10. Peak pressure zones included in Model III for each stature percentile cluster.

Stature Percentile	${{\mathbf{p}}_{\mathbf{p}\mathbf{e}\mathbf{a}\mathbf{k}}}_{\mathbf{i}}$	Description
0–25th	${{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_8,{{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_9$	Buttocks and upper thighs, associated with localized load concentration in smaller users.
25–50th	${{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_{11}$	Lateral thighs, where average and peak pressures are strongly related to comfort.
50–75th	${{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_9,{{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_{11}$	Upper and lateral thighs, confirming the relevance of thigh support in mid-size users.
75–100th	${{\mathrm{p}}_{\mathrm{p}\mathrm{e}\mathrm{a}\mathrm{k}}}_5$	Lower backrest, indicating the importance of lumbar support and pressure balance for taller users.

.

Model III thus captures both the effects of distributed pressure (through average pressures and contact area) and localized discomfort effects (through peak pressures in the most relevant zones). Thanks to this integrated approach, Model III effectively represents the combined influence of global and local pressure features on perceived comfort. Therefore, it is considered the most suitable model for practical ergonomic applications, such as seat comfort evaluation and digital human simulation.

4. Discussion

The applied methodology revealed that comfort perception is significantly correlated with pressure distribution at the user–seat interface, with comfort being negatively correlated with pressure (mean and especially peak) and positively correlated with contact area. This indicates that greater contact area enhances comfort perception, while higher pressure tends to reduce comfort. These results are in line with standard-seat results reported in the literature [3,48,49].

Depending on the considered stature percentile group, different areas of the seat were identified as having a strong impact on the perception of comfort, emphasizing the importance of body size and shape in influencing comfort while using a sports car seat.

Notably, Zone 6 (lateral back support), Zone 8 (buttocks), and Zone 11 (lateral thighs) were identified as critical areas, which were correlated with comfort across most stature percentiles. Larger contact areas in these zones were associated with greater comfort, while higher pressure was linked to discomfort.

Despite these variations, the backrest consistently emerged as a key factor influencing comfort across all stature percentiles. The backrest’s role in providing upper body support and maintaining posture during dynamic driving conditions proved to be more impactful than the seat cushion itself. This finding highlights the importance of lateral containment and rigidity in the backrest for sports car seat design, particularly when aiming to support drivers in high-performance scenarios, in contrast to the more generalized support typically offered by traditional car seats [50].

The statistical analysis, combined with the correlation findings, enabled the development of comfort evaluation models using regression. These models, based on pressure distribution, anthropometric data, and comfort perception, were found to have high accuracy and provide valuable insights for the design of sports car seats. The main finding is about the wide difference between the validity of the regression model evaluated for the entire sample and the clustered sample. Due to the highlighted relevance of anthropometric factors while using sports car seats, it has been demonstrated that the regression models can provide a good predictive model, for comfort perception, for each cluster but not for the entire sample. Thus, the results underline the importance of designing seats tailored to the specific needs of different stature percentile groups, as the comfort experience varies according to body size and shape. By considering these variations in pressure distribution and body support, it is possible to create sports car seats that maximize comfort for a wide range of users.

5. Conclusions

This study provides a comprehensive framework for evaluating sports car seat comfort, combining both objective and subjective data to better understand the factors influencing comfort perception in automotive seats. The methodology, which involved a driving simulator, pressure mats, and subjective comfort questionnaires, enabled the collection of detailed pressure distribution data and user comfort evaluations from 64 participants.

The results demonstrated strong correlations between the data, with significant findings related to the interaction between anthropometric characteristics, pressure distribution, and comfort perception. Notably, the study highlighted the critical role of the backrest in influencing overall comfort, with larger contact areas proving to be more beneficial. Conversely, for the seat cushion, a more uniform pressure distribution was identified as key to improving comfort.

The analysis also revealed the importance of considering user anthropometry in seat design. By dividing participants into stature percentile clusters based on their body dimensions, the study showed how comfort perception varies across different body types. This approach allows for the development of targeted comfort models tailored to specific user groups.

The regression models developed from the collected data offer a practical tool for manufacturers to predict and optimize comfort based on both objective measurements and subjective comfort evaluations. These models could assist in designing seats that better accommodate a diverse range of users, improving the overall driving experience.

6. Limitations

While the study provides valuable insights into seat comfort, certain limitations should be admitted.

Although the sample size was representative of a variety of anthropometric characteristics, it was limited to a narrow age range (20 to 30 years), which may restrict the generalizability of the findings. As different age groups may experience varying physical characteristics, musculoskeletal conditions, and comfort needs, further research is needed to explore how age influences comfort perception. Moreover, considering that the average age of sports car buyers is decreasing [51,52], it becomes even more important to investigate how different age groups affect comfort perception. Expanding the age range in future studies to include a broader spectrum of participants would help better understand the role of age-related factors in seat comfort and lead to more inclusive solutions for consumers of all ages.

A further limitation concerns the gender imbalance in the sample. Although the statistical analysis showed no significant differences between genders, the disparity in the number of participants may have reduced the statistical power of the analysis. Indeed, the literature suggests that gender may influence the perception of comfort [53], indicating that the imbalance in the sample may still have affected the results at the individual level. To improve the generalizability of the results, future studies should consider more gender-homogeneous samples to more accurately explore gender differences in relation to comfort and other relevant variables.

Additionally, the study was conducted in a controlled simulator environment, and future work could benefit from testing in real-world conditions to validate and refine the findings.

Moreover, even if the short duration of the driving simulations (15 min) corresponds to the average urban car trip time in Italy, it only allows the evaluation of the short-term comfort perception. Further studies involving prolonged driving sessions (e.g., 30–60 min) could better reflect long-distance driving conditions and allow deeper insight into cumulative discomfort or fatigue-related seat performance. Such studies could also contribute to refining the proposed model by considering long-term comfort factors such as fatigue accumulation and posture-related discomfort, thereby improving its applicability to extended driving scenarios.

The study focused primarily on significant correlations to identify the most relevant variables in order to develop a robust model. However, this approach left some nonsignificant correlations unexplored. Although these correlations did not reach statistical significance, they could still indicate underlying factors or dynamics that warrant further investigation.

Although a strong correlation was found between the comfort related to the backrest and overall perceived comfort, further studies are needed to fully understand how seat shape influences comfort. To evaluate this in more detail, different seat configurations should be tested, considering various ergonomic and biomechanical factors.