Best Practices for Handling Data and Material Sharing in Engineering Research Papers

Engineering research depends on reproducible results and transparent methods. When researchers share their data and materials openly, they create a foundation that others can verify, extend, and apply to new problems. This practice has become a defining characteristic of rigorous engineering science, where the ability to replicate experiments and validate models directly affects the credibility of published findings.

The engineering disciplines present unique challenges for data and material sharing. Unlike some fields where data often takes the form of small tables or simple measurements, engineering research can involve massive datasets from sensor networks, complex simulation outputs, computer-aided design (CAD) files, source code for custom analysis tools, and physical prototypes. Sharing these resources effectively requires deliberate planning and a solid understanding of best practices.

Beyond the immediate benefits to individual researchers, widespread adoption of data and material sharing accelerates the pace of innovation across the entire engineering community. When one research group makes its experimental data available, other groups can test alternative hypotheses, apply machine learning methods, or combine datasets from multiple sources to draw broader conclusions. This collaborative dynamic reduces duplicated effort and allows the field to advance more quickly than it could through isolated work.

Best Practices for Handling Research Data

Data handling in engineering research involves the entire lifecycle of data: collection, processing, analysis, storage, sharing, and preservation. Each stage offers opportunities to improve quality, usability, and long-term value. The following practices address the most critical aspects of data management for engineering researchers.

Organize Data with Standardized Formats and Rich Metadata

Standardized data formats reduce friction when others attempt to use shared data. For numerical data, consider formats such as CSV, HDF5, or NetCDF rather than proprietary spreadsheet formats that may require specific software. For geospatial data, GeoTIFF or Shapefile formats are widely supported. The key is choosing formats that balance broad accessibility with the ability to represent the full complexity of the data.

Metadata transforms raw data into usable information. A CSV file of voltage readings means little without context. Metadata should describe what the data represents, how it was collected, what units are used, the date and location of collection, instrument specifications, and any processing steps applied. Standards such as the Dublin Core Metadata Initiative or domain-specific schemas like those from the NASA Earth Observing System Data and Information System provide structured frameworks for metadata creation.

Ensure Data Quality Through Validation and Cleaning

Sharing data without validation risks propagating errors that can undermine subsequent research. Engineers should establish quality control procedures at the point of data collection. For sensor data, this might involve calibrating instruments regularly and checking for drift or outliers. For simulation data, validation against analytical benchmarks or experimental results provides a check on the accuracy of the model.

Data cleaning should be documented clearly. If outliers are removed, the criteria for removal should be stated. If missing values are interpolated, the interpolation method should be specified. Ideally, both the raw data and the cleaned version are shared, allowing other researchers to evaluate the impact of cleaning decisions on the final results.

Select Reputable Repositories with Persistent Identifiers

Choosing where to deposit data is an important decision. Reputable repositories offer several advantages over personal or institutional websites. They provide persistent identifiers such as DOIs (Digital Object Identifiers), which ensure that data remains findable even if the hosting institution changes its web infrastructure. They also enforce metadata standards, manage access controls, and guarantee long-term preservation.

For engineering research, repositories such as Zenodo, Figshare, and domain-specific options like the Nature Scientific Data recommended repositories offer reliable platforms. When selecting a repository, consider whether it provides versioning capabilities, handles the file sizes common in your field, and allows for embargo periods if needed for patent or publication reasons.

Provide Comprehensive Documentation

Documentation should enable another researcher to understand and reuse the data without needing to contact the original authors. A data dictionary or codebook that defines each variable, its possible values, and its meaning is essential. For complex datasets, a README file in plain text format can describe the overall structure, relationships between files, and any assumptions made during data collection or processing.

For engineering research involving experimental data, documentation should include the experimental setup, equipment specifications, calibration records, environmental conditions, and any deviations from the planned protocol. The goal is to provide enough context that a competent researcher in the same field could reproduce the experiment or apply the data to a new problem with confidence.

Address Privacy, Confidentiality, and Security

While engineering data often does not involve human subjects, there are situations where privacy and confidentiality arise. Research involving human participants, such as user studies in human-computer interaction or biomechanics, may include personally identifiable information. In these cases, data must be anonymized or de-identified before sharing, and researchers must comply with institutional review board requirements and relevant regulations such as GDPR or HIPAA.

For proprietary or export-controlled data, researchers need to navigate restrictions carefully. In some cases, sharing derived data or aggregated results that do not reveal sensitive details can provide value while protecting intellectual property. Consulting with institutional technology transfer offices or legal counsel early in the research process helps avoid complications when it comes time to share.

Materials in engineering research extend beyond data files to include design files, software code, hardware specifications, experimental protocols, and physical samples. Sharing these materials fully enables other researchers to reproduce results and build on the work more effectively than data alone would allow.

A common shortcoming in published research is incomplete material sharing. A paper might describe a novel sensor design but only include a photograph of the assembled device, leaving other researchers to guess at dimensions, material choices, and assembly procedures. Sharing the complete set of materials means providing CAD files, bill of materials, assembly instructions, and source code for any firmware or control software.

For software, sharing the complete codebase used in the research, including build scripts, test suites, and documentation, allows others to run the same analyses. Even better, sharing a containerized environment such as a Docker image ensures that the software runs identically on different systems. This level of completeness eliminates the "it works on my machine" problem that plagues computational research.

Choose Open and Non-Proprietary Formats

Open formats maximize the accessibility and longevity of shared materials. For CAD files, STEP or IGES formats are widely supported across different software packages, unlike proprietary formats that may require expensive licenses to open. For 3D printing files, STL or OBJ formats are standard. For images, PNG or TIFF formats are preferred over proprietary formats.

When proprietary formats are unavoidable, researchers should also provide an open-format conversion or a detailed specification that allows others to recreate the necessary tools. The principle is to minimize barriers to access. A researcher at a small institution or in a developing country should be able to work with shared materials without needing expensive proprietary software.

Establish Clear Licensing Terms

Without a clear license, shared materials exist in a legal gray area. Others may be uncertain whether they can use, modify, or redistribute the materials. Choosing an appropriate open license resolves this ambiguity and encourages reuse. For data, the Creative Commons CC0 or CC-BY licenses are popular choices. For software, open-source licenses such as MIT, BSD, or GPL provide well-understood terms.

The choice of license has practical implications. A permissive license like MIT allows commercial use and does not require derivative works to be open source. A copyleft license like GPL requires that derivative works also be distributed under the same terms. Researchers should understand these differences and select licenses that align with their goals for the work. Institutional legal offices can provide guidance when needed.

Document Procedures for Reproducibility

Experimental procedures in engineering research can be complex and subtle. A published paper may describe the general approach in a few paragraphs, but the actual protocol may involve specific steps, timing, environmental conditions, and troubleshooting procedures that are not captured in the main text. Sharing detailed protocols as supplementary materials or through platforms like protocols.io provides the level of detail needed for exact reproduction.

Video documentation is increasingly valuable for physical experiments and manufacturing processes. A short video showing how to align components, apply adhesives, or operate equipment can convey information that text and static images cannot. Researchers should consider creating video supplements that walk through the key steps of their procedures.

Ensure Long-Term Accessibility and Version Control

Shared materials need to remain accessible over time. Personal websites and institutional pages are prone to disappearing when researchers move institutions or when systems are upgraded. Using established repositories with preservation commitments protects against link rot. Materials should be assigned persistent identifiers, and the repository should have a clear policy for maintaining access over decades.

Version control is especially important for software and continuously updated resources. Git repositories hosted on platforms like GitHub or GitLab provide a complete history of changes, allowing researchers to reference the exact version used in a particular study. For large datasets that cannot be stored in Git, repositories that support versioning, such as Zenodo, allow researchers to update datasets while maintaining access to previous versions.

Despite broad agreement on the importance of sharing, researchers face practical obstacles. Addressing these challenges head-on makes it easier to follow best practices consistently.

Intellectual Property and Commercialization Concerns

Engineering research often has commercial potential, and researchers or their institutions may wish to file patents before sharing details publicly. Sharing does not need to conflict with commercialization when managed properly. Embargo periods on repositories allow researchers to delay public release until patent applications are filed. Sharing non-essential aspects of the work or providing reduced-resolution data during the embargo period can still advance science while protecting intellectual property.

Some funding agencies and journals now require data sharing within a specific timeframe after publication. Planning for these requirements from the start of a project helps researchers avoid last-minute conflicts. Including data management and sharing plans in grant proposals ensures that the necessary resources and timelines are established early.

Managing Large and Complex Datasets

Engineering datasets can be extremely large. High-resolution sensor networks, long-duration simulations, and video recordings of experiments can produce terabytes of data. Transferring, storing, and sharing data at this scale presents technical challenges. Researchers should consider whether the full dataset needs to be shared or whether derived or summarized data would serve the needs of reproducibility.

For very large datasets, repositories that specialize in big data, such as the DesignSafe Cyberinfrastructure for natural hazards engineering or the NOAA National Centers for Environmental Information, provide the necessary infrastructure for storage and access. Cloud-based sharing platforms can also handle large transfers efficiently. Including a plan for data size and transfer in the data management plan helps ensure that the chosen repository can accommodate the data.

Handling Proprietary and Third-Party Materials

Engineering research frequently uses commercial software, proprietary hardware, or third-party data that cannot be redistributed freely. When proprietary tools are essential to the research, sharing becomes complicated. Researchers should document the exact versions and configurations of proprietary tools used, so that others with access to the same tools can reproduce the work. Providing alternative implementations using open-source tools, where possible, increases the accessibility of the research.

When using third-party data subject to licensing restrictions, researchers should clearly state the source and terms of use. In some cases, the data owner may grant permission for limited redistribution for research purposes. Establishing these permissions early and documenting them in the shared materials avoids confusion and potential legal issues for downstream users.

Navigating Policies and Community Standards

Data and material sharing does not happen in a vacuum. Researchers must work within the frameworks established by journals, funding agencies, and their professional communities.

Journal Requirements

An increasing number of engineering journals require authors to include data availability statements and to deposit supporting data in a recognized repository. Some journals go further, requiring that reviewers have access to data and code during the review process to verify claims. Authors should check the specific requirements of their target journal early in the writing process and plan accordingly.

Journals in different engineering subfields have varying expectations. The Journal of Engineering Mechanics may have different requirements than IEEE Transactions on Signal Processing. Familiarity with the norms of the specific journal and community helps researchers meet expectations without last-minute scrambling.

Funding Agency Mandates

Major funding agencies, including the National Science Foundation (NSF), the National Institutes of Health (NIH), and the European Research Council (ERC), have data sharing policies that apply to funded research. The NSF requires that all proposals include a Data Management Plan that describes how data will be managed, shared, and preserved. These plans are evaluated as part of the proposal review process.

Compliance with funding agency mandates is not optional. Failure to comply can affect future funding eligibility. Researchers should treat data management planning as an integral part of project planning rather than an afterthought. Many institutions offer resources and templates to help researchers create compliant plans.

Discipline-Specific Norms and Standards

Different engineering disciplines have developed their own norms and standards for data sharing. In structural engineering, for example, the DesignSafe platform provides a community standard for sharing experimental and simulation data in natural hazards research. In the bioengineering community, standards like ISA-Tab provide structured formats for describing experimental metadata.

Participating in the standards development process within a researcher's community is a valuable way to contribute to the field while ensuring that the standards meet real needs. Even for researchers who do not participate in standards bodies, adopting existing community standards makes shared materials more useful and reduces the effort required to document data and protocols.

The Role of Repositories, Identifiers, and Metadata Standards

The infrastructure for data and material sharing has matured significantly in recent years. Understanding how to use this infrastructure effectively is a core skill for modern engineering researchers.

Persistent Identifiers and Their Value

A persistent identifier such as a DOI or Handle ensures that a dataset remains findable and citable even if its physical location changes. When a dataset has a DOI, researchers can reference it in their papers with confidence that the link will continue to work. Publishers and indexing services recognize DOIs, and they are increasingly required by journals for data citations.

Persistent identifiers also support proper attribution. When a dataset is cited with a DOI, usage can be tracked, providing metrics that demonstrate the impact of the data. This is particularly important for early-career researchers who need to show the broader influence of their work beyond traditional publications.

Metadata Standards for Engineering Research

Metadata standards provide common vocabularies and structures for describing data, making it easier to discover and integrate datasets from different sources. The FAIR principles (Findable, Accessible, Interoperable, Reusable) provide a high-level framework, and domain-specific standards implement these principles for particular types of data.

For engineering research, metadata standards vary by subfield. The NASA Earth Science Data and Information System (ESDIS) provides standards for geospatial and environmental engineering data. The International Committee on Materials and Mechanical Engineering has developed standards for materials data. Using the appropriate standard for the research domain maximizes the discoverability and usability of shared data.

The landscape of data and material sharing continues to evolve. New technologies and changing expectations are shaping the future of how engineering research is conducted and disseminated.

Automated workflows for data management are becoming more sophisticated. Tools that automatically generate metadata from instrument outputs, validate data against community standards, and deposit data in repositories are reducing the burden on researchers. Adopting these tools as they become available will make it easier to integrate data sharing into regular research workflows without adding significant overhead.

Machine learning and artificial intelligence are creating new demands for high-quality, well-documented datasets. Engineering researchers who share their data according to best practices are contributing to a growing pool of training data that can accelerate progress in areas such as predictive maintenance, materials discovery, and design optimization. The value of well-curated engineering datasets is only going to increase as these methods mature.

Open science practices, including data and material sharing, are becoming expected rather than exceptional. Early-career researchers who develop strong data management habits will be well-positioned to meet the standards of their fields and to take advantage of the opportunities that open sharing creates for collaboration, visibility, and impact.

Conclusion

Effective handling of data and materials in engineering research is a skill that rewards both the individual researcher and the broader scientific community. Following the best practices outlined here organizing data with standardized formats and rich metadata, ensuring data quality, selecting reputable repositories, providing comprehensive documentation, addressing privacy and security, sharing complete materials in open formats, establishing clear licenses, and documenting procedures in detail creates a foundation for reproducible, trustworthy research.

Researchers who invest in these practices find that the effort pays dividends. Their work reaches a wider audience, their findings are more readily verified and built upon, and their contributions to the field are more clearly recognized through citations and reuse of their data and materials. In the rapidly evolving landscape of engineering research, the ability to share effectively is not just a compliance requirement it is a competitive advantage.