Attachment J.3.3.1 Longitudinal Tire Study-Research Design Plan.docx

Longitudinal Study of Commercial Motor Vehicles (CMV) Intelligent Tires

Attachment J.3.3.1 Research Design Plan

TABLE OF CONTENTS

LIST OF Figures (and Formulas)

List of Tables

v Research Design Plan This research plan is intended to support and guide Phase 2 of the FMCSA project entitled “Longitudinal Study of Commercial Motor Vehicle (CMV) Intelligent Tires.” This comprehensive research design plan aligns with the project’s goals and research questions.

This research plan continues to be an iterative process as we learn more information through various tasks conducted in Phase 1. The updated plan will incorporate insights into intelligent tire capabilities, limitations or requirements from fleets during fleet recruitment, capabilities of nationwide support from an install and maintenance team, as well as the expected outputs from tire analysis vendors. This approach ensures all information is captured before decision making occurs during the second phase of the research project.

This research design plan encompasses the goals of this project, as framed by the following research questions (RQs) presented in Table 1.

Table 1. Research questions considered in development of the research design plan.

a) Are some tires exceeding weight ratings under certain dynamic conditions?

i) Does this damage the tire?

b) Can the intelligent/smart tire sense an imbalanced load?

c) Can the intelligent/smart tire sense vehicle stability (e.g., equal weight distribution across all tires, imbalanced loads)?

d) When a truck sits during periods of time (8 hours+) in extreme temperatures (hot or cold), what is the pattern of normalization of tire temperature and pressure when the vehicle is put back in use?

e) How is management of tire pressure different in extreme environments?

f) How often does a tire run under low pressure?

g) How often does a tire run under excessive weight?

h) The drive axels of a power unit?

i) Overweight loads

ii) Improper inflation

i) The trailer?

i) Overweight loads

ii) Improper inflation

Building on the original research questions, further key sub-questions were operationalized to ensure they are clearly defined and actionable within the context of the study collection. This process involves translating the research questions into specific, measurable components.

Sub-questions (see Table 2) exclude questions relating to tire failure and tire damage (e.g., “does shifting weight increase wear on one side?”), certain dynamic conditions (e.g., “do road surface conditions impact tire damage”) and operational behaviors (e.g., “does frequent braking impact tire life?”). These sub-questions are meant for further exploration during analyses.

Table 2. Further defined research questions.

· Can we measure weight with more granularity?

· Weight on tire (single tire)

· Weight on axle shaft (two tires)

· Weight on axle (four tires)

· Weight on bi-axle (eight tires)

· Weight on trailer (trailer axle and kingpin)

· Can we measure when overall vehicle weight shifts?

· Trip measurement/calculation method

· Frequency with which sensors report weights

· Shift front/back or back/front during a trip

· Shift side to side during a trip

· Shift diagonally during a trip

· What do intelligent tires report about weight?

· Load distribution flags

· Imbalanced load flags

· Weight shifting flags

· How is temperature around a tire/engine differ from ambient air temperature?

· How frequently does tire pressure report out?

· How frequently does tire temperature report out?

· How is a trip measured or calculated?

· Can this method distinguish off from idle/nonmoving?

· Can we detect or determine when tires are filled (In comparison to warming up)?

· During idling (vehicle on)

· Between trips (vehicle off)

· How are operational behaviors defined (see operational behaviors table)?

· What are feasible ways to collect exposure?

· Miles driven on tire

· Total hours driven

· Total hours installed

· How does overall vehicle weight get reported?

· How often does tire pressure get reported/sampled?

· What constitutes a tire failure

· How will tire failure be collected?

· How can tire life be tracked?

· When is the life of a tire defined?

· Mid-study

· Study completion

· How does overall vehicle weight get reported?

· How often does tire pressure get reported/sampled?

· Can we detect the initial instance for which a tire becomes over/under inflated?

· Can we longitudinally track tire data across dual tire pairs?

· What types of effects are expected on dual tires?

· Tire damage and wear

· Over/under inflation

· Tire temperatures

· Can we detect the initial instance for which a tire becomes over/under inflated?

· Can we longitudinally track tire data across tire axles?

· What types of effects are expected on dual tires?

· Tire damage and wear

· Over/under inflation

· Tire temperatures

· Is a certain tire type retired from vehicles earlier than the other type?

· Does a certain tire type accumulate damage faster than the other type?

· Does wear accumulate differently in new versus retread tires?

· Can we longitudinally track tire wear across individual tires?

· Does exceeding temperature ratings cause more rapid failure in retreaded tires compared to new tires?

· Are retreaded tires more prone to failure when air pressure is outside the recommended range?

· Does damage or wear accumulate differently in new versus retread tires?

· Does exposure to extreme conditions shorten the lifespan of retreaded tires more than new tires?

· How does the performance degradation of retreaded tires under extreme conditions compare to new tires under similar circumstances?

· Are retreaded tires more susceptible to delaminating under specific dynamic conditions or across operational behaviors?

· Can the data repository be used?

· Can additional experts be consulted about specific variables for collection?

Identifying Data Needs This section outlines the preliminary data needs as shaped by the research questions, establishing the minimum expected variables for addressing the study's objectives. A core set of minimum variables are identified as being required from data collection efforts to effectively address the research questions. These serve as critical variables necessary to answer the study's primary objectives. These minimum variables derived from the research questions are listed in Table 3.

Table 3. Minimum variable set and application towards research questions.

A secondary set of supplementary variables derived from the commercial motor vehicle (CMV) network (SAE J1939) could support data collection efforts, adding new parameters gathered to provide additional information for specific research objectives. These parameters (Table 4) further define a variable set necessary for answering certain research questions.

Table 4. Secondary variable set and application towards research questions.

Throughout the research questions, several iterations of dynamic conditions and other considerations for operational behaviors are mentioned. The conditions and behaviors listed are relevant to various research questions and are derived from questions and sub-questions. The variables described here are also restricted based on what is 1) potentially measurable during a study collection and 2) relevant to tire damage and tread wear. This restriction allows for a more refined set of analyses that can be conducted without losing too much power.

The research questions include iterations of dynamic conditions and operational behaviors, which are relevant across multiple inquiries and sub-questions. The variables tied to these considerations are constrained by two factors: 1) their measurability during the study collection process and 2) their relevance to tire damage and tread wear. These restrictions ensure a refined set of analyses can be conducted while maintaining statistical power. These variables are outlined below in Table 5, along with information regarding the collection source and sampling rates.

Table 5. Dynamic conditions and operational behavior examples and collection sources.

An important aspect of evaluating research questions is defining key concepts such as tire life, damage, failure, and wear. While this framework generally follows guidelines like the Radial Tire Conditions Analysis Guide[footnoteRef:2], it is essential to note that the approach can vary based on the specific vendor supplying tire damage reports. Variations in vendor methodologies may impact the definitions and the metrics used, which must be carefully considered when interpreting data for research purposes. Table 6 provides an overview of tire damage and wear types. [2: ATA Business Solutions. (2020). TMC Radial Tire Conditions Analysis Guide (5th ed.). Retrieved from https://www.atabusinesssolutions.com.]

Table 6. Types of tire damage and tire wear.

Determining Required Data Collection Tools and Data Sources This section outlines the methodologies and sources for data collection as well as where data are expected to come from. Throughout the completion of Phase 1, tables relating to vendor capabilities will be revised within the research design plan to provide updated recommendations for study collection. This will involve comparing the capabilities of intelligent tire sensor vendors and tire analysis vendors against the list of key variables needed to determine what supplemental data sources are required to answer the research questions.

This section outlines the methodologies and sources for data collection and the expected origins of critical data. During the completion of Tasks 4 and 7, tables detailing vendor capabilities will be refined within the research design plan to offer updated recommendations for study collection. This refinement involves evaluating the capabilities of intelligent tire and tire analysis vendors against key variables and identifying supplemental data needs to address research questions comprehensively.

Table 7 has been developed to solicit comprehensive vendor information, encompassing sensor capabilities, compatibility, cost breakdowns, installation requirements, operational durability, calibration needs, and other factors. An example of a completed entry from a fictitious vendor is included in the table for demonstration. In this example, the vendor reports that their system provides real-time PSI and temperature readings but lacks the ability to assess weight shifts or other dynamic conditions.

Table 7. Intelligent tire framework for market scan efforts.

Sensor Capabilities and Specifications

Data Access and Support

Minimum Variables

Information gathered from vendors will be integrated into this table, identifying where it aligns with data requirements and pinpointing any gaps in minimum variable collection. This comprehensive approach ensures that all vendor strengths, weaknesses, and supplemental capabilities are documented, allowing for a robust comparison and precise recommendations for data integration.

As part of the data collection effort, the integration of intelligent tire sensor data with other available sources will be critical. These sources will be used for collecting additional variables (refer to Table 4 and Table 5), answering various aspects of the research questions, and further supplementing the intelligent tire sensors. Table 8 provides an overview of additional data sources.

Table 8. Supplemental sources of data.

While this list of data sources can capture most of the supplemental information required for answering research questions, it is not comprehensive and may be further iterated on. In addition, storage of these data elements may require a secondary data acquisition system to capture and record continuous data (e.g., J1939 does not innately record data).

Data Collection Tasks This section details an overview of the tasks required for study completion. Figure 1 displays a flow chart of expected high level tasks. The collection methodology is designed to address as many of the research questions as possible while striking a balance with collection feasibility and intelligent tire sensor capabilities.

Although all sources of data are unlikely to be attained through study collection, a prioritization of research questions and sub-questions may allow for an optimized path for designing the study collection methodology. Several considerations exist across data collection tasks regarding the types and amounts of information able to be captured throughout the study. Several research questions may not be able to be appropriately addressed.

Figure 1. Data Collection Task Flowchart Note that all data collection tasks listed are planned to be executed during Phase II of the “Longitudinal Study of Commercial Motor Vehicles (CMV) Intelligent Tires” project.

Intelligent Tire Selection and Single Unit Acquisition Based on the results from lab testing and on-road sensor trials, the research team will formalize a recommendation of intelligent tire sensors with FMCSA and integrate the recommendation into the research design plan. The recommendation is likely to prioritize sensor capabilities, maintenance, and costs. Once the intelligent tire sensor is chosen, the team will purchase that sensor for immediate system testing and characterization.

Intelligent Tire Sensor Characterization After acquiring an intelligent tire sensor, the research team will perform an evaluation to verify installation procedures provided by the vendor as well as record installation time and other information that would be helpful to disseminate to the installation and maintenance contractor. Following installation, the research team will perform lab and public road evaluations to ensure data variables from the sensor are being appropriately recorded as well as to establish or verify recording procedures. Additional tests will be performed to assess the accuracy of collected data compared to ground truth measurements, when possible. It will not be possible to test certain elements during this period, such as an evaluation of the battery life of the sensor. The specific tests will be fully designed during Phase 2 after reviewing the complete capabilities of the intelligent tire sensor.

Intelligent Tires Full Purchase and Acquisition Upon verification of sensor capabilities, the research team will purchase the full set of tires and tire sensors for the study collection. The shipping and purchasing logistics will be determined during Phase 2 and in tandem with the fleet partner(s) and installing contractors.

Fleet Selection and Vehicle Assessment Based on the recommendations provided during the Fleet Recruitment task, fleet partner(s) will be selected by the research team and FMCSA based on eligibility. Several considerations for fleet participation exist. Those considerations, in the general order of importance, include:

· geographic proximity to eligible fleet service company;

· applicability of stratification criteria (e.g., fleet location, size, and operation type);

· traceability of trailers (e.g., matched tractor trailers);

· ownership of vehicle technology;

· willingness to provide additional data (e.g., telematics, bypass systems);

· expected miles traveled during the course of the study; and

· diversity of operations.

During Phase I, the research team will develop a standard operating procedure (SOP) outline for participating fleets, which should be utilized to write a custom SOP for each selected fleet. The SOP should consider fleet demographics (number of power units, number of drivers, management of vehicle maintenance, fleet location, etc.) and include details on but not limited to how to communicate with the research team on tire maintenance, tire failures, technological issues, and sharing any fleet-managed sources of data.

Selection of Fleet Service Company During Phase 1, a fleet service company will be selected by the research team and FMCSA. Fleet service companies will primarily perform these main tasks:

1. install intelligent tires at the beginning of the study;

2. repair or replace damaged tires and retrieve tires debris and carcasses throughout the duration of data collection and return the carcasses to the tire analysis vendor;

3. remove the remaining intelligent tires at the end of the study and replace them with fleet-provided tires; and

4. perform maintenance or replace intelligent tire sensors depending on the intelligent tire sensor capabilities.

Intelligent Tire Sensor Installation Installations will be performed by a contractor with a regional or national presence. Installers will follow the procedures provided by the intelligent tire vendor, while adhering to other recommendations as captured by the intelligent tire sensor characterization study undertaken by the research team. The installation process should be streamlined for efficiency, with a focus on quick vehicle turnaround to minimize fleet or driver interaction. Each installation event must be documented with timestamps, technician details, and a checklist of completed tasks to ensure traceability and quality assurance.

Selection of Tire Analysis Vendor During Phase 1, the research team will identify a tire analysis vendor to evaluate all failed tires during the data collection phase and all remaining tires at the collection period’s conclusion. Failed tires include any carcasses retrieved from a tire blowout, as well as tires that reached end of life (e.g., irreparable tire damage or does not exceed minimum required tread depth). The selected vendor will generally be responsible for assessing variables listed in Table 6, and will be responsible for:

1. analyzing tread wear patterns and providing forensic results on causes of failure, and

2. assessing tire life and damage (e.g., sidewall or tread issues).

Collection and Analysis of Tires To obtain information related to tire damage and failure, all tires that reach their end of life and are replaced, or at the end of the study period, will be retrieved by the fleet service company. Tires may be held onsite at each fleet location, or delivered to the tire analysis vendor to assess for wear patterns and damage. All tires handled in this way should be carefully marked and recorded following proper data handling procedures.

For instances in which a tire blowout occurs, and the carcass is able to be recovered, the tire should be labeled and stored for the chosen tire analysis vendor. The research team will ensure appropriate traceability with the data logged regarding the tire’s operational history, such as mileage, load conditions, and environmental exposure. The tire analysis vendor will provide the research team with a report detailing tire carcass findings.

Data Collection, Storage, and Management To handle the data collected throughout the study, the research team will utilize secure servers to store collected data. Infrastructure will not only organize data systematically but also enforce strict access controls to protect sensitive information. Additionally, provisions for scalability and redundancy will be identified to ensure data integrity and resilience against potential system failures.

Sampled Fleet Data Integration The research team will request data from the fleet and from any external vendors they utilize and provide permissions to access. The collected data will then be aligned to ensure all information can be linked and synchronized across each of the different sources. The data will then be evaluated to determine if the information gathered is sufficiently robust for the planned analyses. For example, if the fleet is providing access to telematics data, the research team will need to identify how often that data gets updated and how to download and synchronize that data.

Data Processing The study will collect data from multiple sources and this data will need to be processed and linked using a systematic approach to maintain data integrity and expedite data processing and analysis. Participating carriers will be assigned non-identifying study carrier IDs. Each truck included in the study will also be assigned a non-identifying study truck ID. Data from different sources (such as data from individual tires or from the data acquisition system) will include the shared identifier for the truck and carrier, allowing the data to be linked prior to analysis. Date and time data will be listed in a common form (UTC or local time zone) across datasets from different sources. This approach will be useful in data augmentation (i.e., linking multiple datasets by time), to create an analysis dataset featuring a timeline per truck with information on dynamic driving conditions, tire pressure, tire temperature, etc.

During data collection and processing, new data must be routinely reviewed. The study team will systematically summarize distributions of key variables, using summary statistics (e.g., average, minimum, and maximum values) along with tabular and graphical representations of the data. This output needs to be examined to identify any irregularities in the data. Any issues in the data (e.g., missing data or unusual values) identified during review must be quickly addressed by the research team for correction at the data source or during data processing as soon as possible. The assessment of variable distributions will also be used to inform a final analysis plan, as certain analysis methods may be best suited for specific data distributions. Further, the study team should also systematically calculate tire reliability metrics, such as time between failures, failure rates, changes in treadwear over time, and others, to determine if there are irregularities with tire performance that should be addressed through single tire replacement. Identifying irregularities and assessing the nature of the irregularity should be supported through an analysis-based approach and investigation of the failed tire. The assessment of variable distributions should also be used to inform a final analysis plan as certain analysis methods may be best suited for specific data distributions.

Data Augmentation Effective augmentation of data consists of taking data from raw sensors (i.e., GPS, Intelligent Tires), and combining or linking that data with the supplemental datasets or additional information. During this task, tire sensor data will be merged with supplementary sources, such as GPS, weather data, and telematics. These integrations will align datasets collected at varying frequencies, such as maintenance logs and real-time tire performance metrics, ensuring a seamless representation of operational and environmental factors.

Development of an Analysis Approach The following section presents initial analysis recommendations for each research question. The proposed analysis approach is dynamic to allow for adjustments based on the variable characteristics (e.g., frequency at which data is collected, continuous or categorical data, etc.). During Phase II, the study team will need to consider the data collection tools, study sample, and features of the collected data to determine the most appropriate analysis method for each research question. Research team members with advanced knowledge of traditional statistical analysis techniques and machine learning modeling 1) reviewed published literature focused on tire performance and 2) discussed established and novel approaches to assessing each research question. Generally, the approach includes summary statistics, tabular and graphical presentations of data distributions, regression modeling, hypothesis testing, and advanced modeling techniques to assess the relationships between key variables relevant to each research question. The approach also incorporates data visualization techniques to illustrate relationships between variables. Finally, the analysis approach considers the stratification criteria and potential covariates.

Hypothetical Analysis Plan A hypothetical analysis plan is included below. The plan will continue to be refined during Phase 1, to include additional analysis methods or streamline current options based on information gathered on data collection sources and data collection tasks. Analysis methods to consider include the following.

Logistic regression models are used to predict the probability of a binary or multinomial outcome based on one or more explanatory variables. Individual models can be built for specific stratifications (e.g., tire size) and to include covariates of interest (e.g., season) and/or random-effects terms to control for correlated data (e.g., a term for the participant carrier) in a mixed-effects logistic regression model.

Generalized linear mixed models may be used to address research questions involving lifespan or time to failure (e.g., predicting the life of a tire based on one or more explanatory variables).

Survival analysis is another approach for assessing lifespans or time to failure. This approach allows the comparison of survival curves from multiple groups to assess differences in lifespan under multiple conditions.

Ordinal regression can be used to assess ordinal response variables (e.g., no damage, slight damage, severe damage) in relation to explanatory factors.

Advanced data visualization techniques can show how tires are affected by dynamic conditions and temperature.

Data simulations and advanced modeling approaches can be applied to large-scale multimodal data (e.g., to study factors influencing tire failure or normalization of tire temperature).

Frequency and rate calculations are used to describe data distributions with stratified calculations by tire size, carrier, and other conditions of interest.

Analysis options are presented per research question in Figure 2 below and described in more detail below.

Figure 2. Analysis options for Phase II research questions.

Research Question 1 Research Question 1 states:

How does the overall weight of the vehicle shift under driving conditions and various load distributions?

a) Are some tires exceeding weight ratings under certain dynamic conditions?

i) Does this damage the tire?

b) Can the intelligent/smart tire sense an imbalanced load?

c) Can the intelligent/smart tire sense vehicle stability (e.g., equal weight distribution across all tires, imbalanced loads)?

Tire pressure (1.a) will be addressed through data simulations and advanced modeling approaches that can model tire pressure by dynamic conditions over the course of a trip or data collection period. These models can summarize patterns of tire pressure changes by continuous and categorical dynamic condition variables. Further, analysis of incidents where tire pressure exceeds recommended ratings can be analyzed using generalized linear mixed models, which can be used to model frequency of events by independent variables, while controlling for correlated data from individual carriers and vehicles. Tire damage (1.a.i) under changing conditions and operating behaviors can be modeled using machine learning models.[footnoteRef:3] Distinct patterns of driving under dynamic conditions over the course of the study can be identified, and associated tire damage per pattern can be compared using these advanced modeling approaches. Differences in tire damage severity or frequency could also be modeled by patterns of driving under dynamic conditions using generalized linear mixed models, ordinal regression models, and logistic regression mixed models, depending on the tire damage measure. [3: Ye Liu, Haibo Chen, Sijin Wu, Jianbing Gao, Ying Li, Zihao An, Baohua Mao, Ran Tu, Tiezhu Li, Impact of vehicle type, tyre feature and driving behaviour on tyre wear under real-world driving conditions, Science of The Total Environment, Volume 842, 2022, 156950, ISSN 0048-9697, https://doi.org/10.1016/j.scitotenv.2022.156950.

The analysis plan for parts 1.b and 1.c will be updated depending on the available data for imbalanced loads. However, analysis options may include calculating the per trip rate of detected imbalanced or stable loads across sample stratification criteria, calculating the per trip rate of imbalanced or stable loads using data from weigh stations, and comparing these measures by modeling the likelihood of intelligent/smart tires to detect a true imbalanced load using a logistic regression model. Individual models can be built for specific stratifications (e.g., tire size) and to include covariates of interest (e.g., season) and/or random-effects terms to control for correlated data (e.g., a term for the participant carrier) in a mixed-effects logistic regression model.

Research Question 2 Research Question 2 states:

How does ambient temperature affect tires under very hot or very cold conditions?

a) When a truck sits during periods of time (8 hours+) in extreme temperatures (hot or cold), what is the pattern of normalization of tire temperature and pressure when the vehicle is put back in use?

b) How is management of tire pressure different in extreme environments?

RQ2 will use data simulations and advanced modeling approaches that can be applied to large-scale multimodal data and/or time series analysis.

For RQ 2.a, all trips following a period of non-driving for 8 or more hours will be identified. These trips will be binned by their average or highest/lowest temperature value during the non-driving period (e.g., -32 to -16 degrees Fahrenheit, -16 to 0 degrees, 0 to 16 degrees, 16 to 32 degrees, 32 to 48 degrees, etc.). The pattern of normalization of tire temperature will be plotted for each of the non-driving period temperature bins using all trip data associated with that temperature bin. The expected time to reach normalized tire temperature will be compared across the non-driving period temperature bins. The process will be repeated to assess patterns of normalizations of tire pressure by non-driving period temperature bins.

RQ 2.b will assess variations in tire pressure while driving in extreme environments. This analysis will use data on tire pressure and ambient temperature while driving. Ambient temperature conditions while driving will be binned as in the previous analysis. Tire pressure data under each ambient temperature bin will be plotted to identify different patterns of tire pressure management in different temperature conditions. Variation in tire pressure will be compared across the ambient temperature bins.

Research Question 3 Research Question 3 states:

What is the “life of a tire” under various operational behaviors?

a) How often does a tire run under low pressure?

b) How often does a tire run under excessive weight?

To address RQ3, operational behaviors will be summarized for each trip (e.g., number of stops, average speed, braking pattern). The proportion of time with low pressure or excessive weight will also be calculated per trip. The proportion of time with low pressure will be modeled by each operational behavior using a generalized linear mixed model. The model will control correlated tire data from each vehicle. The approach will be repeated to model proportion of time with excessive weight.

Generalized linear mixed models will also be used to measure the life of a tire, in terms of estimated life remaining at the end of data collection or by a measure of tire damage, by the total driving time or proportion of driving time the tire was under low pressure. A second model will quantify changes in the expected life of a tire by total driving time or proportion of driving time the tire was under excessive weight. Another option to continue investigating for applicability to this research question is survival analysis, which may be used to assess lifespans, time to failure, or time to near-failure (tire replacement needed for safe operation of the vehicle).

Research Question 4 Research Question 4 states:

What is the “life of a set of tires” on:

a) The drive axels of a power unit?

i) Overweight loads

ii) Improper inflation

b) The trailer?

i) Overweight loads

ii) Improper inflation

The generalized linear mixed models used in RQ3 will be expanded to include terms for tire location (drive axels and trailer) to address RQ4. In this analysis, unique models will be developed to assess tire lifespans for different tire locations. If survival analysis is a viable approach for the collected data, it should also be considered for RQ4.

Research Question 5 Research Question 5 states: What is the effect of an improperly inflated tire on the adjacent dual tire?

The analysis approach for RQ5 depends on how “effect” is defined and measured. For example, if measured effects include adjacent tire failure (yes/no) or adjacent tire damage (yes/no) or adjacent sensor damage (yes/no), a logistic regression model will be used to assess changes in the probability of observing these effects for the adjacent dual tire. If tire lifespan is a measured effect, generalized linear regression models can be used to quantify differences in adjacent tire lifespan by the presence of an improperly inflated tire. Further, survival analysis should also be considered, as should data simulations and advanced modeling approaches.

Research Question 6 Research Question 6 states: What is the effect of an improperly inflated tire on the seven adjacent tires on the same power unit or trailer?

The analysis approach for RQ5 depends on how “effect” is defined and measured. For example, if measured effects include adjacent tire failure (yes/no) or adjacent tire damage (yes/no) or adjacent sensor damage (yes/no), a logistic regression model will be used to assess changes in the probability of observing these effects for the adjacent dual tire. If tire lifespan is a measured effect, generalized linear regression models can be used to quantify differences in adjacent tire lifespan by the presence of an improperly inflated tire. Further, survival analysis should also be considered, as should data simulations and advanced modeling approaches.

Research Question 7 Research Question 7 states: What is the likelihood of a tire failure on a new tire versus a retreaded tire?

To answer RQ7, tire failure for each tire (yes/no) will be noted by the end of data collection. The failure outcome data will be modeled by the tire type (new/retreaded), using a mixed-effects logistic regression model. Individual models can be built for specific stratifications and to include covariates of interest and random-effects terms to control for correlated data (e.g., a term for the participant carrier and vehicle). The model will be used to calculate odds ratios for tire failure by tire type.

If tire failure rates are low, measuring lifespan of a new tire versus a retreaded tire using the same approaches outlined in Research Questions 3, 4, 5, and 6 may provide important insight on any potential differences in new tires and retreaded tires.

Research Question 8 Research Question 8 states: Are retreaded tires more likely to fail under conditions that are outside of the ratings (load, temperature, air pressure) of a tire?

If RQ8 seeks to assess what conditions are more likely to lead to tire failure in retreaded tires, trips that ended with a retreaded tire failure will be compared to trips without a retreaded tire failure for differences in tire operating conditions. The data can be compared using mixed-effects logistic regression models to predict the likelihood of retreaded tire failure by operation outside of ratings (e.g., low pressure yes/no). Similarly to RQ7, if tire failure rates of retreaded tires are low, measuring lifespan of a retreaded tire using the same approaches outlined in Research Questions 3, 4, 5, and 6 may be an opportunity to understand conditions that may lead to failure. Patterns of operation outside of ratings over the course of data collection could be compared for retreaded tires that failed versus retreaded tires that did not fail, with analysis options including survival analysis and data simulations and advanced modeling approaches.

If RQ8 instead seeks to assess whether retreaded tires are more likely than new tires to fail under operating conditions outside of ratings, the models described above could be expanded to include data from new tires as well. Models would include a predictor variable for tire type (yes/no) and assess the significance of tire type in measuring tire failure or tire lifespan in various operating conditions.

Research Question 9 Research Question 9 states: Are retreaded tires more likely to delaminate the tread under similar conditions?

To answer RQ9, the delamination status of each tire will be recorded by the end of data collection, if not sooner, should delamination occur during the study. The likelihood of delamination of retreaded tires can be modeled using a mixed-effects logistic regression modeling approach using data on operation outside of ratings (e.g., low pressure yes/no) as independent predictor variables. Differences in time to delamination can be assessed using survival analysis models with data on operation outside of ratings as predictor variables. If rates of delamination are low, each tire can be assessed for risk of delamination based on tire condition at the end of data collection. This data can be analyzed using the same approaches outlined in Research Questions 3, 4, 5, 6, and 7. Other analysis options include data simulations and advanced modeling approaches.

Research Question 10 Research Question 10 states: How can this data be used by other researchers to understand tire life in wear models and tire research?

By incorporating continuously monitored variables like tire pressure, temperature, load distribution, and road conditions in real time, the dataset to be collected in Phase II will allow researchers to further analyze how tires wear under diverse operational and environmental circumstances. The data collected in Phase II will provide insights that enhance the precision of wear models to facilitate accurate predictions of tire durability and performance. Moreover, these data can be leveraged to develop sophisticated algorithms for predictive maintenance, enabling timely interventions that effectively extend tire lifespan. This comprehensive dataset will also support investigations into the effects of varying weather conditions, driving habits, and terrain on tire performance, thereby fostering the development of resilient and adaptive tire technologies. To address RQ10, the study output should include a spreadsheet documenting all data collection tools and collected variables, along with a data dictionary. In addition, deidentified analysis datasets should be prepared for the data repository. Further, the study report should include recommendations on how the data and analysis results can be used by other researchers to further tire research.

Development of a Sampling Plan Many of the RQs utilize data collected at high frequencies—on a trip level or even continuously. However, some analyses will look at patterns of use over time and the impact on tire lifespan and for these RQs, the sample size is dependent on the…