Skip to Content

SAPMMCONF_PIC.PNG

With the International SAP Conference for Mining and Metals 2016 still fresh on my mind, I traveled to the SAP Industry Analyst Base Camp to meet with Dan Miklovic from LNS Research. The conference was a real eye-opener for me. I was looking forward to an intense exchange on industry transformation with Dan to get an outside view:

  • Where is digital transformation happening in the metals industry?
  • Is it happening at all?
  • Where are the focus areas?

 

Customer centricity and E-Commerce

Overwhelmingly across regions and equally for special steel, to mass producers – customer centricity has been among the top 3 strategies in customer keynotes at the event. There are many trends & drivers that could explain this – from market overcapacity to customers expecting a state of the art, B2C like, seamless experience. Some companies wanted to attack new markets, expand their existing rather EDI-based and high-touch sales model to a more differentiated approach.

 

This still includes EDI, but does not stop there. Most complete visions spanned from a company web page, to an online webshop, self-service and value-add services for order tracking, certificates, fast re-order, and much more. Even the approaches to a webshop are rather sophisticated. Context-aware marketing and personalization of the shop experience, even dynamic pricing strategies, and predictive analytics have been mentioned.

 

Multi-brand management across shops and markets are another interesting aspect. How do you sell into a B2C-markets and the DIY-space, maybe even competing with existing channel partners in some countries, without hurting the main “upmarket” brand?

 

We also had interesting conversations about integration to B2B and B2C-marketplaces like Amazon, Alibaba or the Ariba network.

 

Which channel to use depends on one hand on the customer group, and on the other hand on the type of product: from pure make-to-stock standard products, to engineer-to.order special grades with their own specific chemistry – and everything between. One service center offers the upload of 2D drawings in the webshop for cut-to-size free-shape laser cutting.

 

My point of view: customer engagement and commerce is a very hot topic in steel. Most companies I talked to have either projects running, or seriously look at this.

 

I had a number of good conversations about “differentiating your strategy” between pure make-to-stock standard products (with typically lower profitability), make-to-order/finish-to-order (where it is possible to automatically determine “feasibility”e.g. checking allowed width and thickness combinations), and real engineer-to-order for new grades or applications. Pure MTS may be best sold “touchless” and highly scalable via a marketplace. MTO-type materials could be sold through own eCommerce-channels. ETO will require your sales experts and engineers to advice on options and discuss requirements.

 

The cornerstone for the large majority of projects presented at the conference is SAP Hybris Commerce.

 

 

The internet of things, digitalization and big data in the plant

No surprises here – IoT is the other big topic for many steel manufacturers. Most often predictive maintenance and quality have been identified as focus areas.
Some companies look at IoT much broader, from industrial drones to IoT-enabled in-plant logistics or remote operations and robotics as e.g. in mining.

We have seen some good projects.

 

Across several keynotes it was stressed to fast innovation, quick proof of concepts, early failures & quick learning are essential to identify feasible & valuable opportunities. Early proof of concepts and pilots have proven significant value potential.

 

Rather typical for the mill industry, and proven in multiple cases, is the notion of predictive quality. Actually, this term includes a  number of related, but slightly different scenarios:

  • root-cause analysis of quality deviations across a heterogeneous data set (including process & LIMS data, maintenance information, location & weather information, ERP and CRM aspects like complaints, cost impact, and defect & item traceability across production including cutting, slitting, rolling)
  • real-time prediction of potential quality issues based on machine learning & real-time process data. Mostly this is done to identify problems as early as possible to save cost. This could happen through early re-work, scrapping or down-grading. One manufacturer indicated that they want to utilize quality predictions to optimize & focus their quality checks on the likely outliers.

 

SAP’s IoT HANA platform has been used as the foundation for a number of proof of concepts and pilots. It is offered both on premise, as well as in the cloud.

SAP also offers SAP Predictive Maintenance and Service applications in both deployment models.

 

In my point of view, the traceability of defects across multiple production steps, is best addressed with a combination of strong platform capabilities, data science, and on-site work with the engineers. I do not believe in pure stand-alone data science & math. But then I am an engineer myself.

(You may be also interested in the analyst view on the Fedem acquisition.)

 

The topic is of course not entirely new – but the capabilities to combine data in very different formats and across individual domais, and analyse with advanced data science methodology, in a cost-effective manner – this is new, and has not been possible before. I had some good conversations on this topic with the colleagues from Osisoft. Their PI system integrates nicely with HANA through a new connector – again a more standardized approach to combine existing data for smarter analysis. I had a good and long conversation with Dan on exactly this topic. Thank you, Dan, for the valuable feedback and open discussion.

 

More predictable business, better real-time risk management

With the current level of commodity prices, and the overall difficult state of the mining industry, this has been a focus topic in several mining keynotes.

In my point of view, this is equally relevant in metals.

 

This included real-time risk reporting – intra-day – on commodity risk, having a central risk management platform, or “just” the possibility to do a soft-close any time during the quarter. In short – more real-time visibility & predictability.

Some of this was achieved through custom HANA-based enhancements, the “real-time” financials is based on S/4 HANA Finance.

 

More frequent planning cycles, simulation of alternative scenarios and possible responses have been mentioned several times as well. Joint planning across the demand and supply network, smarter demand forecasting, better inventory visibility were further capabilities highlighted.
A cornerstone to this more flexible planning approach is through SAP’s integrated business planning for supply chain in the cloud.

 

What are your top 3 initiatives?

What were your observations from the conference? What surpised you most? What did you put on your short term action plan?
Do you need a co-innovation partner along this journey – talk to us!

 

For more on how Digital Transformation helps transform the metals industry, check out this TED Style video and take this short survey to benchmark your company against your peers.

To report this post you need to login first.

4 Comments

You must be Logged on to comment or reply to a post.

  1. Chuck Pharris

    Great article Stefan.  To further reduce my ignorance of steel business, the issue of predictive quality caught my attention.

    a) What portion of the quality deviations typically is a function of the manufacturing process vs the transportation/logistics/storage degradation?

    b) What role is modern machine learning playing over traditional regression techniques in the real-time quality process prediction?  Is it incremental improvement “we can do it better now” or “we really couldn’t do it before and now we can”.

    Thanks for the great industry insight into the transformation of the industry and especially highlighting the role IoT is playing.

    Regards,

    Chuck

    (0) 
    1. Stefan Weisenberger Post author

      Hej Chuck,

      manufacturing steel is unfortunately not as predictable as discrete manufacturing. Liquid metal, made from ore and scrap of variying quality, plus heavy wear on manufacturing assets, results in a significant number of deviations from quality.

      Surface defects, like scratches on a coil or metals sheet, can both result from the rolling process, or the coating process, and later handling and warehousing errors – like a dent from a fork lift, either while liftiing the coil, or while not properly “navigating” in the warehouse.

      Modern machine learning indeed takes a very important role e.g. when analyzing surface scans, thermal imaging, and the likes. Traditional regression would not be able to do this.
      Other use cases like analyzing machine sensor and quality data could have been done before, especially when the datasets are closely related.

      It is getting more into “we really could not do it before” is when you explore broader sources of data – across multiple machine lines, multiple type of data sets, and data from the business layer like customer complaints. Add text, like in maintenance notifications, shift notes, or customer messages – not possible before.

      Also the data science plus engineering knowhow to sync & link the time series of sensors over a longer line and tracing across lengthening material, cut into strips etc. Tricky with the old means in my opnion.

      Thanks and regards,
      Stefan

      (0) 

Leave a Reply