PLM Transformation Requires Data Transformation

March 22, 2019 Doug Halverson

By transforming unstructured data into usable information, manufacturers can reinvigorate their businesses for increased profitability and enhanced regulatory compliance.

Today’s complex manufacturing processes rely on increasingly integrated technologies that manage the entire product lifecycle. Product lifecycle management (PLM) is a practice that governs products from their inception, through design and manufacture, to their obsolescence and withdrawal from the market. PLM spans people, processes, technology, and data related to the product. This data forms the foundation of future research and development, supply chain optimization, customer relationship management, and regulatory compliance.

Often existing only in non-traditional or unstructured sources such as CAD drawings, PDFs, text documents or spreadsheets, PLM data consists of items such as specifications, packaging details, artwork, and labels. The vastness of systems spanning the product lifecycle is often incompatible with one another, leading to silos of this unstructured product data and the inability to merge, manage and derive meaningful analytics and insights from it.

As organizations turn to specialized software to unify the PLM process, there’s a growing recognition of the challenge and potential presented by unstructured product data. A successful PLM program must prepare unstructured data for migration to a central location. As part of a comprehensive PLM initiative, data cleansing, governance, and transformation are essential.

Going after unstructured data

The first step in any data cleansing and migration project is to identify the types and locations of existing data. Is it already digital, or do paper documents need to be scanned? Which departments or business units own the various silos? Long-established manufacturing companies may have product data on outdated storage devices or in legacy applications. Examples of unstructured data are almost endless and include documents like product specifications, technical specifications and artwork for packaging brand marks and product artwork, recipes, formulas and ingredient, and part lists, regulated labels, and engineering drawings.

Assessing the current state of the data is key to determining the overall scope of the project. Experts recommend using automation to identify the number and relevancy of objects, fields, and records that currently exist and whether they all need to migrate to a new PLM system. Performing this assessment also helps determine the completeness of the data set and how to best allocate resources to resolve discrepancies.

Since PLM requires the extraction of data from documents, images, and other non-traditional sources, there needs to be a metadata schema that describes the contents of these sources so they can be accessed and searched. There are any number of existing schemas for various industries and types of assets, such as video, images and geospatial, so it pays to research whether one already exists before creating one.

Powerful machine learning algorithms are what make the magic happen. Rather than having teams of “human middleware” analyzing documents or labels and rekeying the relevant data, artificial intelligence is now able to do the job. With natural language processing and pattern matching, AI programs analyze the structure of a document or image and effectively decompose it to extract data and arrange it into a structure that can be used to form linkages among materials.

Of course, artificial intelligence algorithms must be guided by the subject matter experts who understand the content contained in the data sources. A PLM program rarely succeeds without the involvement of the people who are in a position to validate the rules in order for the rules to evolve in real-time thanks to AI.

Today’s complex manufacturing processes rely on increasingly integrated technologies that manage the entire product lifecycle. Product lifecycle management (PLM) is a practice that governs products from their inception, through design and manufacture, to their obsolescence and withdrawal from the market. PLM spans people, processes, technology, and data related to the product. This data forms the foundation of future research and development, supply chain optimization, customer relationship management, and regulatory compliance. Often existing only in non-traditional or unstructured sources such as CAD drawings, PDFs, text documents or spreadsheets, PLM data consists of items such as specifications, packaging details, artwork, and labels. The vastness of systems spanning the product lifecycle is often incompatible with one another, leading to silos of this unstructured product data and the inability to merge, manage and derive meaningful analytics and insights from it. As organizations turn to specialized software to unify the PLM process, there’s a growing recognition of the challenge and potential presented by unstructured product data. A successful PLM program must prepare unstructured data for migration to a central location. As part of a comprehensive PLM initiative, data cleansing, governance, and transformation are essential. Going after unstructured data The first step in any data cleansing and migration project is to identify the types and locations of existing data. Is it already digital, or do paper documents need to be scanned? Which departments or business units own the various silos? Long-established manufacturing companies may have product data on outdated storage devices or in legacy applications. Examples of unstructured data are almost endless and include documents like product specifications, technical specifications and artwork for packaging brand marks and product artwork, recipes, formulas and ingredient, and part lists, regulated labels, and engineering drawings. Assessing the current state of the data is key to determining the overall scope of the project. Experts recommend using automation to identify the number and relevancy of objects, fields, and records that currently exist and whether they all need to migrate to a new PLM system. Performing this assessment also helps determine the completeness of the data set and how to best allocate resources to resolve discrepancies. Since PLM requires the extraction of data from documents, images, and other non-traditional sources, there needs to be a metadata schema that describes the contents of these sources so they can be accessed and searched. There are any number of existing schemas for various industries and types of assets, such as video, images and geospatial, so it pays to research whether one already exists before creating one. Powerful machine learning algorithms are what make the magic happen. Rather than having teams of “human middleware” analyzing documents or labels and rekeying the relevant data, artificial intelligence is now able to do the job. With natural language processing and pattern matching, AI programs analyze the structure of a document or image and effectively decompose it to extract data and arrange it into a structure that can be used to form linkages among materials. Of course, artificial intelligence algorithms must be guided by the subject matter experts who understand the content contained in the data sources. A PLM program rarely succeeds without the involvement of the people who are in a position to validate the rules in order for the rules to evolve in real-time thanks to AI.

What are the outcomes of a successful PLM transformation initiative?

Ultimately, a successful PLM transformation initiative is going to result in improved product profitability by optimizing the entire product lifecycle. This can take the form of:

  • Optimized component costs – by understanding everything that goes into a product, manufacturers can make informed changes that reduce overall cost.
  • Better change control – beyond cost reduction, it’s critical to know how an upstream component change will affect the overall performance of a product – particularly in a highly-regulated industry.
  • Waste reduction and sustainability – visibility into packaging design can drive standardization and promote reuse and recycling.
  • Reduced inventory – being able to determine whether parts from different suppliers are identical can reduce inventory and associated carrying costs. This is especially important when merging supply chains as companies get acquired.
  • Accelerated innovation – a complete product repository makes it possible to quickly identify and improve upon successful products while avoiding reinvention and engineering rework.

Product specifications, components, and ingredients are the starting point for the manufacturing process and are the drivers for every aspect of the ERP, SCM and CRM systems that run the business. If product data is inaccurate, redundant or incomplete there will surely be systemic problems that cause disruptions.

For those industries that are highly regulated—such as pharmaceuticals, life sciences, food and beverage, and aerospace—PLM is more than a competitive advantage, it’s a requirement. At any time, a manufacturer in a regulated industry must know the provenance and composition of every item on the bill of materials for every assembly and sub-assembly or every ingredient on a material safety data sheet.

By transforming unstructured data into usable information, manufacturers can reinvigorate their businesses for increased profitability and enhanced regulatory compliance.

> To view the original article on New Equipment Digest, click HERE.

About the Author

Doug Halverson

Vice President, BackOffice Associates

More Content by Doug Halverson
Previous Article
Yell.com
Yell.com

A BackOffice Associates story of Yell.com’s SAP S/4HANA Migration to the Microsoft Azure Cloud

Next Article
What Does It Mean To Be A Data-Driven Enterprise Today?
What Does It Mean To Be A Data-Driven Enterprise Today?

Digitalist Magazine spoke with our CEO, Kevin Campbell, about how organizations can map their digital trans...