PLM IoT platforms: complexity and data scale

PLM IoT platforms: complexity and data scale

cloud-platform-scale

Manufacturing companies, analysts and software vendors are sharing excitement about huge potential of IoT and connected products. This is a very good news – I love the idea of things getting connected and optimized. I wish manufacturing industry will operate as smooth as my Waze navigation system by checking road condition, traffic jams and informing me in a following way – changing route, new ETA is 19:40p, you saved 5 min.

waze-routing-update

So, I’m dreaming about about product lifecycle management system that can tell me – “there is a better component selection for chosen BoM configuration, you saved 12’300$ in the next production batch“. This is probably still a dream in 2015.

However, dreaming during this Friday morning, made me think about complexity and scale of data problem can be discovered in order to make manufacturing work similar to way Waze navigation system works.

Yesterday, I shared some of my thoughts about how future of IoT – digital twin can crash PLM platforms with the scale of data. I want to continue this discussion with two examples I captured earlier this week during PI Congress event in Boston.

The first example came from Airbus presentation made by Tristan Gegaden, Head of Operation of PLM Harmonization Center. He spoke about leveraging of data in a modern digital environment and scale of data. Below you can see few data points about A350 digital model – 3 million part instances in 30’000 configurable items. This is just a single aircraft model.

airbus-complexity-data-scale

airbus-data-scale

The second example came from the Ford presentation made by Gahl Berkooz, Head of Data and Governance, Ford Motor Company. In his presentation he spoke about big data driven PLM systems and Hadoop based data technologies Ford is planning to use for data analytics. The following pictures shows how much data Ford vehicles are generating and comparing it to Google data scale. Actually, Google looks a smaller case of Ford data complexity.

ford-data-cloud

ford-data-scale-iot

What is my conclusion? I think, there is misalignment between “grand strategies” of future IoT driven PLM environments, digital twins, etc. on one side and data platforms and technologies PLM systems are using today in production. The future data scale of IoT enabled manufacturing products (starting from very complex avionic systems and ending up with connected toothbrushes collecting information about every single person combining it with your health history and adapting your dental insurance rate) can be overwhelming. A note before weekend for PLM IoT architects and technologists. Just my thoughts…

Best, Oleg

Share

Share This Post

  • Robert Brincheck

    You are mixing two different concepts that really don’t overlap. Streaming IoT data needs to be acquired and managed by IoT platforms that are built to deal with the challenges associated with acquiring and managing data on this scale. The platform also needs to be capable of managing this process so that only meaningful data is transmitted. Typically there is minimal value in a sensor continuously transmitting that it is ‘ok’ or operating in a normal range. Data that is in spec would be sent at less frequent intervals with out of spec of data being sent at high frequency to record the deviation. This data is acted upon if required and then gathered and stored in a data lake for later analysis.Big data analytics is used to understand and document actual product failure modes and their root causes. This processed data is used by the PLM system to help define proper part selection and to refine the overall process for developing new parts. There is no need to run tests that don’t simulate field failure conditions, for example.

    Digital twin and augmented reality are utilize subsets of this IoT data stream to perform explicit tasks. If I’m using augmented reality to gain insight on a failure mode on an aircraft before servicing it, I don’t need to load the entire aircraft into session and I don’t need to stream all the data from the aircraft. If the problem is in the engine, I only need to look at the engine. Same is true for digital twin. As these IoT enabled PLM applications mature, they will quickly adapt to enable this type of sophisticated data mgt. just like DMU applications did when they were developed. Despite having over 3 million part instances Airbus was apparently able to load the entire plane into DMU and section it right down the middle.

  • beyondplm

    Robert, thanks for your comment and clarification. Does it mean IoT data (and platform) has nothing to do with As-maintained BoM? In that case, two things will live in 2 separate silos – PLM for design data and IoT for the data about products in a real life. Then, what will be the process and technology to connect them together? Thanks, Oleg

  • Ganesh Salunke

    IoT will surely make significant impact on PLM in near
    future. I believe industries from various domains has to join hands with IoT to
    be a leader in market.

  • beyondplm

    Ganesh. I think you are right! IoT will change a lot.

  • Robert Brincheck

    Oleg, sorry for the delay in responding. No it does not mean that IoT data will not be associated with the As-maintained BOM, just that the IoT data, or at least the ‘raw’ IoT data will live in the IoT system and referenced in the PLM system. This is no different than an integration between PLM and ERP for As-Built BOM or the product cost for purchased parts. The one difference is that the PLM system could contain the relevant, or key field data, say the max temperature a part was exposed to, while the IoT system could contain all of temperature data for the product. If I’m servicing the product using the As-Maintained BOM I would only need the key information to decide if service was needed. If I’m an engineer trying to understand the environmental exposure of my part I could run a report or analysis on the IoT data.

  • beyondplm

    Robert, thanks for clarification. It makes sense. What you explained used many years by analytic software that created separate model of information that can used to explore data. While in theory it is okay- I’d like to see examples of “data reference” betweens systems. PLM-ERP integration by itself is a big problem in most of manufacturing companies. These solutions are replicating lots of data between ERP and PLM systems to stay in sync.

  • Robert Brincheck

    ERP-PLM integrations are often a problem because companies don’t take the time to develop an overall architecture plan for their business processes and define which system owns what data and if/when a change of ownership is required. There is little reason to replicate data and process it on both ends which is where most of the issues arise. With a proper process architecture and ownership clearly defined data can be referenced or at worst published from the master system into the viewing system. There are also issues with multiple ERP instances and even multiple ERP systems.

    PTC’s Blue Pump demo is a good example of ‘data reference’. It pulls data from CAD/PLM, SLM, ERP, CRM and IoT systems based on the use case. http://www.ptc.com/video?id=4539913896001&account=Demos

  • beyondplm

    Robert, very cool demo- thanks for sharing! Speaking about this pump- what system will store information about ALL pumps on all tractors with each serial number and related information presented in the video? Is it Windchill role?

  • Robert Brincheck

    Oleg, as the demo showed, each system would manage data about the pump that is appropriate for that system to own. As-built information would come from the ERP system with service information and as-maintained information coming from and SLM system like Windchill Service Information Manager. Since this is a service scenario, SLM system would manage the as-maintained BOM and the unique information for each specific instance. However, it references common information from the PLM and ERP systems. For example, the visualization data for the pump would come from the PLM system which owns the CAD and viewable info. The service procedure and video would live in the SLM system. The common reference is the part number and an as-maintained BOM that is linked to the e-BOM and updated as the eBOM changes. Windchill can be the PLM and SLM systems or the SLM system with a different PLM systems. There are customers using both configurations.

    Windchill Service Information Manager

  • beyondplm

    Robert, thanks for clarification! I see how information can be located in different systems. But since as-maintained information is in Windchill, does it mean it will have an instance for each physical pump in the field? How does it connect to ERP?

  • projet43

    I have doubts about the PLM to IoT integration. In between, there is typically an ERP and a maintenance system in between. Besides, many industries already have embedded software providers which answer to their specific needs.

    Also, I completly agree that current PLM systems are order of magnitude away from managing big data.

  • beyondplm

    Thanks for the comment and sharing insight. I think, IoT is a big opportunity, but PLM companies might have an over-promise. Therefore, it is very interesting to see PTC’s moves in this space as well as Autodesk jumping into IoT space with the acquisition of SeeControl.