I was extremely impressed by Karenann Terrell, CIO of Walmart, talking about Walmart’s partnership with SAP on stage at SAPPHIRE NOW.
She stressed that digitization is not a buzzword. “It is incredible real.” She sees Walmart transform into a tech company within a retailer – focused to lower the overall cost of goods, and to serve customers even better. Resonates with me – as a consumer, and from a mill perspective.
From my point of view, the Walmart partnership, and the Siemens announcement to use the HANA Cloud Platform to build their cloud for industry, mark the industrial internet of things, industry 4.0 and the digital transformation as very, very real.
Technology is now able to provide capabilities that can fundamentally transform businesses and create new levels of efficiency.
But how will digitization affect the mill products industry?
A recent study by Roland Berger Strategy Counsultants highlights that different industry verticals adopt digitization at different velocities, and are impacted differently. The mill products industries, like steel, paper or textiles, are at the intersection of process and discrete industries..They are already highly automated, but still they often struggle with fluctuations and quality issues in the manufacturing process.
Roland Berger highlighted 4 levers for transformation: digital data, automation, network and digital access. Let’s take a look at each from a mill products perspective.
Call it big data, or digital data: sensors provide new sources of data – in industrial machinery, like a paper machine or a fork lift, but also in logistics, and close to the customer.
- Logistics service providers use real time routing based on real time demand information and live traffic data..Especially in urban areas and with “perishable” goods like ready-mix concrete this makes a lot of sense as well.
- Better sensors in the paper machine may give you the granularity of information to optimize production in ways never possible before.
Equally important is the concept of the “digital twin” – to have a digital representation of the factory (and beyond). Every machine, fork lift, paper roll, pallet, truck and railcar has its digital twin – including precise location, “health” status, and business status (for which customer is this paper roll). In mining, but also in some warehouse scenarios even workers are monitored with their location, health status, and further environmental parameters.
- Such digital representation may visualize an entire steel plant including rail and logistics, creating a unified control room like the Usiminas video wall.
- Another European integrated steel plant strives to optimize their internal cross departmental supply chain integrating plant logistics with production
- Hamburg port authority integrates supply chain operations into real time hub operations – including external LSPs and real time traffic
Much of this has already been possible before – what is new are the adaptive real-time element like traffic, and the visualization through mobile devices and augmented reality, 3D visualization, GPS and in-door navigation.
Digital data close to the customer I have not seen yet. In the more end-consumer oriented building products industry,yes, but not in steel or paper. But I am curious to learn new examples.
Like the Chemicals industry and other process industries, the automation level is already quite high in steel or paper production. It is common to set processing parameters automatically. Many sensors are already in place measuring pressures, surface quality and much more. But typically this information stays close to the machine. It is visualized for the operator in the local operations context. Driver less fork lifts carry jumbo rolls through the mill, driver less trucks or trains and automated warehouse are becoming more and more common.
Bu there is still a lot of potential for more efficiency – through digital data enabling smarter manufacturing, smarter operators and more stable, more agile, more precise, more cost & efficiency aware production.
In paper manufacturing the paper strength or the brightness are key quality parameters.Today such quality parameters on a specific roll are fluctuating (hopefully within the allowed ranges), and therefore manufacturers usually run their machine in a “safe mode” where the fluctations do not touch the low quality threshold. As a result – a roll is in average typically much “better” than what the customer has requested.
Paper manufacturers have significant potential to save cost e.g. on new fibre or chemistry by operating more stable, with less fluctuations. This is possible through better sensors, and advanced analytical methods like data mining. Rather than having allowed ranges per each individual process parameter, more advanced models consider combinations of parameters, and thus allow much more stable processes – basically running the machine much closer to the lower limit of allowed quality – thus saving cost. A paper machine manufacturer called this effect the “Angstkosten” – the cost of playing too safe..
More context, please
Another interesting vision discussed at the German pulp and paper manufacturer association was called “situational production“. “Situational” means here to consider the context – overall utilization of the machine, health and wear of assets, maintenance plan, customer for which the current order is produced (and quality complaints of this customer in the past). In this broader context, it may be the case that it is smarter to operate the machine at lower speeds, and adjust other process parameters accordingly. To a certain degree this is common knowledge of the operator of a rolling mill. Certain grade – thickness combinations are only possible on a “fresh” rolling mill. When the rolls are already close to the next maintenance interval, it may no longer be possible to deliver highest surface qualities. Or it may be smarter to run the mill at a lower speed to postpone the next maintenance interval.because an important order needs to be completed before maintenance.
Speed matters, and better sensors
Raw materials like recycled paper or scrap metal differ a lot in quality and contamination. As a result, downstream operations are affected, quality deviates, and recipes need to be adapted. In some cases even within the pulper, the quality is not homogeneous.
Adaptive recipes with parameters that consider multiple sensor readings (e.g. combine human assessment, processing of digital images, and immediate down stream paper quality information), and a fast & robust predictive model allow to adjust process parameters much faster – thus driving further cost savings.
This applies similarly also to grade change transitions in paper-making. Advanced predictive models, and better and more sensors in the papermachine, will drive further reduce of paper loss and minimize the raw material usage.
Much of this is in the hands and the brain of an experienced operator. There is a lot of potential in a more data driven, fact based approach – that can be automated towards more stable, more reliable, and more cost-effective operations – and better guidance & recommendations for a “smarter workforce”.
Will this make the operator redundant? No, it will require a “new type of operator”.
The new operator
A key success criteria for digitalization are the workers. This may sound counter-intuitive at first glance. Sure, their will be new skill profiles needed like data scientists in a centre of expertise in headquarters. But that kind of service could even come from external parties, service providers or the manufacturer of the machine.
A future operator will need to understand the upstream and downstream process, and the impact of his decisions. Decisions will become more complex, if you think of the entire value chain, but also decision support will become better.
Instead of visualizing current energy consumption, or an alert, digitalization will convert this information into a real-time recommendation to act. Knowing energy consumption is high, does not save energy cost by itself. The operator needs to get alerted, receive one or more proposed mitigation proposals – within the time frame where he still can re-adjust the parameters – in real-time.
The new type of operator will be more interdisciplinary, with more decision power, and better understanding of processes and value generated.
Networks and cyber-physical systems
New sensors are able to connect not only to fieldbus and SCADA, but also the cloud, the business layer, and to other machines – thus creating cyber-physical systems, basically a network of real, physical objects like machines, and their digital twins.
At the Hannover fair, I have seen self-organizing manufacturing “cells” where multiple robots talk to each other and the product they create. “Lot size one” is a big topic, enabled through machines that can change tooling extremely fast. In my opinion this is rather a discrete manufacturing scenario, especially in car and machinery manufacturing – not that relevant in mill products.
Set-up times in classical mill manufacturing are still rather long. There are examples like automatic adjustment of winder knifes in seconds. But whenever color is involved there is still the sequence from lighter to darker tones – and a lot of cleaning before you switch to lighter colors again.
Rolling mills already today operate in lot size one mode, one could argue, as each coil has different characteristics and its own future path through the manufacturing process.
RAMI 4.0 – the reference architecture for industry 4.0, or IIOT – extends to the business network
The ZVEI, German Electrical and Electronic Manufacturers’ Association, has created a reference architecture for Industry 4.0. Interestingly, they have extended the IEC 62264 / IEC 61512 hierarchy levels in both directions. Below the control and field device is the “Product”. And above the enterprise level is the “Connected World”.
The network – beyond mill borders – can coordinate demand across company (and mill) borders.It can automate and simplify transportation planning and execution between manufacturers and 3PLs. It can automate and streamline procurement of parts, services and raw materials through Ariba.
And with some mill having 60% of their workforce being external we should think outside industry 4.0, and take digitalization also into this domain – managing you “total workforce” from finding talent, providing payroll functionality, and ensuring adequate training & qualification for the job- through the network.
In the mining example for predictive maintenance demoed at SAPPHIRE NOW, we see another example well applicable also to mill. Remote sensing of truck (or any other asset) “health”, prediction of a breakdown in 5 days time, and counter-measure including maintenance activities – closing the loop into spare pars procurement through the Ariba network.
Linking the e-commerce side with a procurement network (e.g. through the hybris-Ariba punch out scenario) enables very effective new marketplaces e.g. for steel. We have discussed this with several mill customers as an additional channel to reach new customers. Beyond order and delivery information, companies already today exchange quality certificates, in some cases also detailed information on quality defects on a paper roll. Rather than trimming (cutting out) defects, production downstream in the value chain can consider the defects more efficiently.
Digital access to factories, services to optimize quality or production yield, maintenance may create entirely new business models e.g. for service providers (not the original manufacturer) who analyse asset health, offer predictive scenarios. Local (within the plant) scenarios work well with SAP MII, event stream processing and a local HANA installation. Digital remote access is required for remote monitoring of assets, a central CoE analyzing root cause analysis of quality defects or searching for correlations and patterns to enable predictive maintenance.
The concept of an asset intelligence network, where operators and original machine manufacturers collaborate on asset master data, quality and asset health data, allows for new business models and services. Machine manufacturers provide asset information that asset operators can utilize to create “correct” asset master data in their plant maintenance application. Machine manufacturers start offering already today condition-based and predictive maintenance services to the asset operators. Through better insight in usage patterns across many installations, they can derive better predictive models – for quality, maintenance or energy optimization.
Third parties could provide similar specialized services through the network.
Digital transformation enablers
I have discussed the Industry 4.0 transformation journey with my various mill products customers, and partners – most recently last week at the German SAP usergroup meeting for Mill Products. We all agreed that this is rather a evolution, than a revolution. Many solutions & capabilities are already in place, and broadly used across the industry:
- SAP MII is the foundation for the Usiminas video wall, is used with Wienerberger and many other to drive better visibility in energy usage, plant logistics and item traceability.
- SAP MII is used with event stream processing to capture & analyze continuous data from the shopfloor.
- A number of SAP customers use SAP Predictive Analysis on SAP HANA for predictive maintenance and predictive quality pilots and proof of concepts
- Others use the SAP HANA Cloud Platform to access and analyze manufacturing & asset data in a remote plant
- SAP Transportation Management, and Ariba drive the digitalization of procurement and tendering processes across enterprises.
- Mill products customer use GPS enabled packaging, on board units on trucks & rail cars for real time visibility – inside and outside the plant.
In my point of view, there are large saving financial potentials and the global opportunity to use raw materials and energy much more effciently. As with all transformations, there is change impact to the workforce. Automation and remote access, and a more decentral operation model may actually create a better work-life balance for operators. In the mining industry, remote operations monitoring and automation are helpful – as it is increasingly difficult to find employees for remote operations in a rather unsexy industry. We might see similar effects in mill products industries.