When I started this series in July, I expected to talk mostly about new ways of working, enabled through a data-driven and model-based approach. However, when analyzing what is needed for such a future (part 3), it became apparent that many of these new ways of working are dependent on technology.

From coordinated to connected sounds like a business change;

however, it all depends on technology. And here I have to thank Marc Halpern (Gartner’s Research VP, Engineering and Design Technologies)  again, who came with this brilliant scheme below:

So now it is time to address the last point from my starting post:

Configuration Management requires a new approach. The current methodology is very much based on hardware products with labor-intensive change management. However, the world of software products has different configuration management and change procedures. Therefore, we need to merge them into a single framework. Unfortunately, this cannot be the BOM framework due to the dynamics in software changes.

Configuration management at this moment

PLM and CM are often considered overlapping. My March 2019 post: PLM and Configuration Management – a happy marriage? shares some thoughts related to this point

Does having PLM or PDM installed mean you have implemented CM? There is this confusion because revision management is considered the same as configuration management. Read my March 2020 post: What the FFF is happening? Based on a vivid discussion launched by  Yoann Maingon, CEO and founder of Ganister, an example of a modern, graph database-based, flexible PLM solution.

To hear it from a CM-side,  I discussed it with Martijn Dullaart in my February 2021 post: PLM and Configuration Management. We also zoomed in on CM2 in this post as a methodology.

Martijn is the Lead Architect for Enterprise Configuration Management at ASML (Our Dutch national pride) and chairperson of the Industry 4.0 committee of the Integrated Process Excellence (IPX) Congress.

As mentioned before in a previous post (part 6), he will be speaking at the PLM Roadmap & PDT Fall conference starting this upcoming week.

In this post, I want to talk about the CM future. For understanding the current situation, you can find a broad explanation here on Wikipedia. Have a look at CM in the context of the product lifecycle, ensuring that the product As-Specified and As-Designed information matches the As-Built and As-Operated product information.

A mismatch or inconsistency between these artifacts can lead to costly errors, particularly in later lifecycle stages. CM originated from the Aerospace and Defense industry for that reason. However, companies in other industries might have implemented CM practices too. Either due to regulations or thanks to the understanding that configuration mistakes can cause significant damage to the company.

Historically configuration management addresses the needs of “slow-moving” products. For example, the design of an airplane could take years before manufacturing started. Tracking changes and ensuring consistency of all referenced datasets was often a manual process.

On purpose, I wrote “referenced datasets,” as the information was not connected in a single environment most of the time. The identifier of a dataset ( an item or a document) was the primary information carrier used for mentally connecting other artifacts to keep consistency.

The Institute of Process Excellence (IPX) has been one of the significant contributors to configuration management methodology. They have been providing (and still offer) CM2 training and certification.

As mentioned before, PLM vendors or implementers suggest that a PLM system could fully support Configuration Management. However, CM is more than change management, release management and revision management.

As the diagram from Martijn Dullaart shows, PLM is one facet of configuration management.

Of course, there are also (a few) separate CM tools focusing on the configuration management process. CMstat’s EPOCH CM tool is an example of such software. In addition, on their website, you can find excellent articles explaining the history and their future thoughts related to CM.

The future will undoubtedly be a connected, model-based, software-driven environment. Naturally, therefore, configuration management processes will have to change. (Impressive buzz word sentence, still I hope you get the message).

From coordinated to connected has a severe impact on CM. Let’s have a look at the issues.

Configuration Management – the future

The transition to a data-driven and model-based infrastructure has raised the following questions:

  • How to deal with the granularity of data – each dataset needs to be validated. For example, a document (a collection of datasets) needs to be validated in the document-based approach. How to do this efficiently?
  • The behavior of a product (or system) will more and more dependent on software. Product CM practices have been designed for the hardware domain; now, we need a mix of hardware and software CM practices.
  • Due to the increased complexity of products (or systems) and the rapid changes due to software versions, how do we guarantee the As-Operated product is still matching the As-Designed / As-Certified definitions.

I don’t have answers to these questions. I only share observations and trends I see in my actual world.

Granularity of data

The concept of datasets has been discussed in my post (part 6). Now it is about how to manage the right sets of connected data.

The image on the left, borrowed from Erik Herzog’s presentation at the PDM Roadmap & PDT Fall conference in 2020, is a good illustration of the challenge.

At that time, Erik suggested that OSLC could be the enabler of a digital CM backbone for an enterprise. Therefore, it was a pleasure to see Erik providing an update at the yearly OSLC Fest conference this week.

You can find the agenda and Erik’s presentation here on day 2.

OSLC as a framework seems to be a good candidate for supporting modern CM scenarios. It allows a company to build full traceability between all relevant artifacts (if digital available). I can see the beauty of the technical infrastructure.

Still, it is about people and processes first. Therefore, I am curious to learn from my readers who believe and experiment with such a federated infrastructure.

More software

Traditional working companies might believe that software should be treated as part of the Bill of Materials. In this theory, you treat software code as a part, with a part number and revision. In this way, you might believe configuration management practices do not have to change. However, there are some fundamental differences in why we should decouple hardware and software.

First, for the same hardware solution, there might be a whole collection of valid software codes. Just like your computer. How many valid software codes, even from the same application, can you run on this hardware? Managing a computer system and its software through a Bill of Materials is unimaginable.

A computer, of course, is designed for running all kinds of software versions. However, modern products in the field, like cars, machines, electrical devices, all will have a similar type of software-driven flexibility.

For that reason, I believe that companies that deliver software-driven products should design a mechanism to check if the combination of hardware and software is valid. For a computer system, a software mismatch might not be costly or painful; for an industrial system, it might be crucial to ensure invalid combinations can exist. Click on the image to learn more.

Solutions like Configit or pure::variants might lead to a solution. In Feb 2021, I discussed in PLM and Configuration Lifecycle Management with Henrik Hulgaard, the CTO from Configit, the unique features of their solution.

I hope to have a similar post shortly with Pure Systems to understand their added value to configuration management.

Software change management is entirely different from hardware change management. The challenge is to have two different change management approaches under one consistent umbrella without creating needless overhead.

Increased complexity – the digital twin?

With the increased complexity of products and many potential variants of a solution, how can you validate a configuration? Perhaps we should investigate the digital twin concept, with a twin for each instance we want to validate.

Having a complete virtual representation of a product, including the possibility to validate the software behavior on the virtual product, would allow you to run (automated) validation tests to certify and later understand a product in the field.

No need for inspection on-site or test and fix upgrades in the physical world. Needed for space systems for sure, but why not for every system in the long term. When we are able to define and maintain a virtual twin of our physical product (on-demand), we can validate.

I learned about this concept at the 2020 Digital Twin conference in the Netherlands. Bart Theelen from Canon Production Printing explained that they could feed their simulation models with actual customer data to simulate and analyze the physical situation. In some cases, it is even impossible to observe the physical behavior. By tuning the virtual environment, you might understand what happens in the physical world.

An eye-opener and an advocate for the model-based approach. Therefore, I am looking forward to the upcoming PLM Roadmap & PDT Fall conference. Hopefully, Martijn Dullaart will share his thoughts on combining CM and working in a model-based environment. See you there?

Conclusion

Finally, we have reached in this series the methodology part, particularly the one related to configuration management and traceability in a very granular, digital environment.  

After the PLM Roadmap & PDT fall conference, I plan to follow up with three thought leaders on this topic: Martijn Dullaart (ASML), Maxime Gravel (Moog) and Lisa Fenwick (CMstat).  What would you ask them?