Realising the Potential of Smart Manufacturing

Sai Prakash R. Iyer

October 21, 2019

What are we missing in our current approaches to smart manufacturing and all its fabulous benefits

Digital technology has been in the shop floor and the continuous process manufacturing plant for many decades now. In its early avatars, digital technology entered manufacturing in the form of Programmable Logic Controllers (PLCs) that replaced analog logical control circuits. Soon, continuous process manufacturing industries were leveraging digital to put the supervision and direction of a production plant in a control room, where process managers could monitor the production real-time through digital systems that sat on top of an array of PLCs as well as analog controllers hooked up to assets in the field that spewed mostly analog data. Computer aided manufacturing, robotics, machine tool automation etc., have been some of the other avatars of digital in manufacturing. While these initiatives were discrete applications focused on assets in the field (Levels 0 to 2 in Purdue Reference Enterprise Architecture or PERA), the 1980s and 1990s saw digital initiatives to integrate manufacturing into rest of the enterprise, in the form of MRP/MRP2 (Materials Requirement Planning/Manufacturing Resources Planning), ERP (Enterprise Resources Planning) and MES/MOMS(Manufacturing Execution Systems/ Manufacturing Operations Management Systems). MES/MOMS were particularly focused on real-time data capture from assets in the field, providing the production- and process managers near-real-time analysis and insights. These sought to link field level digitalization (PLCs, SCADA, DCS) with enterprise-level deployments (Levels 3 and above in PERA), facilitated by data repositories called Historians that maintained structured time series data on the entire production flow that can be accessed by a range of systems that managed operations, logistics and supply chain, production, customers, etc. All these deployments of digital technology have invariably focused on either improved efficiency or higher effectiveness (yield, quality and so on) or both which resulted in a lower cost or better quality (or both) of the output from the production process. Digital has seen sophisticated use cases in manufacturing for over five decades now.

In recent years, ever cheaper computing and storage, abundant bandwidth for data exchange and high speed, low latency connectivity and mobility has created new opportunities to endow more smartness to the shop floor. The advent of cloud computing now turns the business case for investments in digital from CAPEX-heavy to near-CAPEX-free. This idea – digital transformation of manufacturing – has been proposed under different names such as smart manufacturing, smart factories, Industry 4.0 etc. The baseline benefit seems to be to endow more smartness to the shop floor via the ability to derive near-real-time insights on production that can further enhance the level of automation of live production decisions resulting in significantly improved cost-quality proposition for the producer. Evangelists envision smart factories, where not only are the shop floors smart, but these are also connected real-time to other smart shop floors, smart warehouses of vendors/ customers and the smart logistics networks that manage the complex flow of inputs and outputs, creating an ecosystem of players involved in value addition that spans organisational-, industry- and geographic boundaries. In short, a radically new way of producing and delivering value.

While this vision is awe-inspiring, most of the benefits from digital transformation of manufacturing remain a promise. For sure, there are many manufacturing firms that are talking about investing in smart manufacturing. Some of them are making modest investments in experiments and pilots. Most of them are attending conferences and participating in panel discussions. Conferences and panel discussions are great ways to explore, debate and diffuse new ideas widely, provided evangelists and pioneers have insights from experiments, and war stories of successes or failures to share. There seems to be a dearth of such real-life insights, or at least folks aren’t talking much.

While the pioneers are willing to make small bets on experiments, many of them don’t seem to follow these up with larger investments focused on wider adoption. What seems to hold them back? Talk to the pioneers on why they are not pushing ahead, and typical responses (and what they mean by that) would be:

  • We don’t think the technology is ready yet (what we got is mostly eye-candy – yet to figure out any useful purpose)
  • While we see the promise, this isn’t a top priority yet (the benefits we have seen in the experiment aren’t substantial enough to change our investment priorities – no business case, yet)
  • This will prominently figure in our digital journey (in future) (we don’t think we have figured out how to use the technology beneficially in our business and the evangelists don’t seem to know either)
  • We are currently focused on laying the foundation for this (within our organisation) (we are stuck with internal challenges and need to sort these out before putting money in digital)

Current lack of enthusiasm in going beyond the one-off pilot can be seen as a reluctance to invest in the face of specific challenges. What are these challenges, and how can we try and address these?

Challenge #1: The Problem of (too) Big Data

Recently, when I was talking to the promoter of a startup offering smart manufacturing solutions, he was boasting that their pilot installation in a shop floor captured 34,000 datapoints when their competitor offered to capture a mere 1,000 datapoints. When I asked the promoter, what actionable insights for the shop floor manager were they able to generate from this larger dataset, he started thinking.

A data scientist in a shop floor is somewhat like a kid let loose in candyland. There are hundreds of assets in the field each spewing multiple datapoints every millisecond. Add high frequency data capture and one can end up with terabytes of data in short order. The ignored question most often is, “what data do we really need, and at what frequency, to generate actionable insights?” – called feature extraction. Before this question can be asked, one should ask, “what insights are meaningful and actionable in this production setting?” Both these questions are often ignored in the design of smart manufacturing initiatives. What’s needed first off is an unwavering focus on what’s meaningful and actionable for the shop floor manager.

Vendors and solution providers often try to impress clients with their ability to capture more data and then structure these in innovative ways. Does this really render smartness to the shop floor? Some say that artificial intelligence (AI) and machine learning (ML) are the answer – feed the terabytes of data to AI/ML algorithms and soon you will have smartness, just like how Google guesses what you want to search for. For that to work in the context of manufacturing, feature extraction – picking the small subset of data that is relevant for an asset (in other words, knowing what data is redundant) – is critical and needs to be done at the design stage itself. Feature extraction is a problem already solved in many industrial manufacturing contexts via the Historians and MES and there is no need to reinvent the wheel. The domain experts are those who can provide precise guidance on this. Unfortunately, in most situations, domain experts are either not brought in, or they are brought in too late (after the data scientists have run amok in candyland).

What’s currently smart in the shop floor are humans, using their wisdom gained from years of experience and experimentation – the domain experts. The best bet right now is for a hybrid approach of AI/ML built on the foundation of human expertise. This requires deep involvement of domain experts right from the early design stage in developing solutions for digital manufacturing. The involvement of domain experts is augmented by recent advances in low-code development environments, which should catalyze their involvement and contribution in addressing the challenges of data collection, structuring, analysis and insights. Domain experts have another critical role to play – but that’s in addressing yet another challenge.

Challenge #2: The Problem of Integration Responsibility

In recent years, an ecosystem of diverse players is taking shape in digital manufacturing. We have the big-tech firms such as Amazon, Microsoft, Google, Cisco, who are offering general purpose computing and storage solutions an IP address away (cloud computing). These players are offering technology platforms on which manufacturing companies can build their digital manufacturing infrastructure. Then we have industrial corporations like Siemens, ABB, GE, offering purpose-built technology platforms for digital manufacturing. These are not very different from the general-purpose ones offered by big-tech other than some effort to bring low-code environments and an ecosystem of partners with domain expertise into these purpose-built platforms. Then there are startups is various shapes and sizes that offer smart manufacturing solutions – many of them have limited experience in the relevant manufacturing domains. Most of their expertise is in the technology of data capture, data structuring and analysis. We saw what happens when these players take the lead in building solutions. There are also the consulting companies who have expertise in specific manufacturing domains as well as in digital technology – these players are offering solutions on paper. Bottomline for the manufacturing firm looking to invest in digital is that no single vendor is taking end-to-end responsibility for design, development and deployment of the digital manufacturing solution, the resultant benefits and return on investment.

One way to break this logjam is for one or the other vendor to take on the responsibility of solution integration. Ideally placed are the industrial corporations, but they are focused on building an ecosystem of partners and don’t seem to be building up the needed momentum to break the logjam. This opens the business opportunity for digital system integration for manufacturing and logistics. To be successful in this requires a combination of capabilities in the specific manufacturing domain and in digital technology. Those players who are able to build and meld these capabilities will most likely win the race to become the preferred partners for digital transformation in manufacturing.

Challenge #3: The Problem of Organising for Digital Transformation

The organisation that works best in the current context – before digital transformation – is likely not suitable for driving digital transformation successfully or standing to benefit from it. The logic for organisational incompatibility is that the business strategy that leverages digital is likely to be very different from the current business strategy and the organisation needs to be realigned for the new digitally focused business strategy.

Initiatives to drive digital transformation in manufacturing can originate from the top – with the board’s and CEO’s mandate probably along with a broader agenda for the whole company to go digital, or from the bottom – pushed by manufacturing or operations or other functions. In either case, the tension among functional heads to stake claims on digital is quite common as going digital will be associated with increased budgets, visibility and prestige if success follows. At the same time, senior managers will also be mindful of the downside risk of failure of digital initiatives that they sponsor. Caught in this, even top-down initiatives to digitally transform manufacturing suffer from competing sponsorships by multiple CxOs (the chief of manufacturing or operations, the chief of IT, the chief of digital etc.) coupled with high risk aversion – a sure recipe for low innovativeness and a high chance of not achieving anything worth mentioning.

In the case of bottom-up driven initiatives, a sponsoring CxO is most likely supported by a preferred vendor but faces the risk of lack of collaboration by other CxOs who compete with the backing of other vendors. This problem is further aggravated by what we discussed in Challenge #2. Without a vendor taking end-to-end responsibility, CxOs may not be keen to sponsor the investment proposals to their CEO and board. For bottom-up proposals for investments in pilots, the bar for the business case will be lower and review of the business case will likely be less rigourous. The game changes totally when it comes to a full-fledged investment proposal. The board or its investment committee will insist on a business case that clears the corporate rate of return benchmark, it will review the business case thoroughly and it will also ask the sponsoring CxO to be accountable for the outcomes. Clearly, none of the CxOs is likely to take on that accountability, as:

  • There is no guarantee that the sponsoring CxO’s peers will collaborate for success
  • The sponsoring CxO faces all the downside risk of failure but will have to share the glory of success with peers who are seen to collaborate
  • The sponsoring CxO is not cushioned from the risk of execution failure by any single vendor who takes on end-to-end responsibility

Typical organisational solutions such as cross-functional teams and so on are not likely to work here as these are more suitable for solving coordination issues, not accountability. Some organisations have tried putting the responsibility to a newly created C-level role such as Chief Digital (Transformation) Officer. Unfortunately, such roles are not endowed with P&L responsibilities and related authorities, making it difficult for these C-level officers to drive transformation decisively. What is most likely to yield results is an organisational solution that is tailor-made for a given company, that ensures that whoever holds the responsibility for digital transformation also holds P&L responsibility for the whole digital business case.

This is another way of saying that there isn’t much sense in looking at the business case for the digital transformation of manufacturing alone. It must be positioned in the broader context of the digital transformation of the business. This drives the alignment between business strategy and organization. In sum, before embarking on the digital transformation of manufacturing (or other functions), the company first ought to update its business strategy to reflect going digital, including in manufacturing. That done, the organizational problem will be easier to handle. In this process, if some deeply held and long-followed conventions are to be overturned, so be it.

Challenge #4: The Problem of Readiness for Execution

Strategy professionals often use the cliché, 1% strategy and 99% execution, inspired by a quote attributed to a famous inventor. The vagaries of execution are well known to managers and need no elaboration.

In the context of the digital transformation of manufacturing, a rather nuanced challenge crops up. We talked about having meaningful and actionable insights in Challenge #1. While emphasizing the role of domain experts in gaining these insights, we pointed to yet another critical role they have to play. That’s in helping the organisation solve the problem of readiness for execution.

Imagine that there are actionable insights, but the shop floor manager cannot take the action as the current shop floor configuration does not permit those actions. For instance, say the production run needs to be changed four times on a given day based on real-time insights, but the time needed for one change is several hours so it’s not possible to implement – while the actionable insight was there, it couldn’t be acted upon. The source of the problem is legacy. The way the shop floor is configured, the way shop floor assets are (not) instrumented and (not) connected, the way data is collected and structured, and finally the extent to which the shop floor is designed and equipped for agility. Real-time insights can be leveraged only if actions leveraging the insights can be taken near-real-time. Thus a shop floor designed to leverage digital ought to be agile. Current configurations may not have adopted agility as a key design philosophy. To leverage digital, a radical rethink of the shop floor design would be needed. Associated with that is the remaining useful life of the not-so- agile shop floor and its assets. Domain experts have a critical role to play in understanding how the current shop floor configuration – the legacy – aids/ hinders leveraging of digital and how it can be reconfigured. They can guide the digital transformation team in making the shop floor more ready for digital transformation. For that however, they need to be involved right from the beginning of the digital transformation journey. Then there will be the tough calls on benefits of agility Vs. cost of building agility into the shop floor.

Challenge #5: The Problem of (Not Thinking About) Business Models

A business model provides a succinct articulation of how a firm seeks to create, deliver and capture value. Digital manufacturing, along with the digital transformation in other parts of the firm-level value chain, will radically redefine how the value chain is best configured to create, deliver and capture value. As we saw in Challenge #3, the business case for digital manufacturing is not stand-alone and needs to be embedded in the broader context of firm- wide digital transformation – across multiple functional and process areas, backed by an updated business strategy that leverages digital. This means paradigm shifts in what is produced and delivered in customers – the product or service and how things are done – the business processes. Evangelists talk of new business models such as moving from products to services (buying industrial equipment spares Vs. equipment availability as a service), moving from procurement of assets and licenses to subscriptions (setting up own IT infrastructure Vs. cloud computing and storage service), moving from in-house resources to participating in multi-sided platforms (own R&D Vs. open innovation market place), and so on. Yet, these isolated use cases do not deliver substantial benefits unless they are weaved into the tapestry of an updated business strategy that leverages digital across the firm-level value chain. This requires conscious prioritization of digital by the CEO and the board in terms of strategic direction for the firm. Even with the sponsorship of the board and CEO, coming up with an updated business strategy that leverages digital it is not an easy task. That requires reimagining the business in the context of what’s possible with digital.

Challenge #6: Reimagining Digital Manufacturing

Stanford University economist Paul David, in his 1989 seminar paper titled “Computer and Dynamo: The Modern Productivity Paradox in a Not-too Distant Mirror”, narrates how manufacturers struggled to reimagine the shop floor when US manufacturing industry moved from steam engines to electrical motors as the prime mover. In the steam engine era, shop floors were spatially designed in three dimensions (more like a multi-storied building) with a single shaft running through it that transmitted mechanical energy from a steam engine to a large number of production equipment, all connected to the shaft through belts and pulleys, gears and crankshafts. Early implementations of electrical motors replaced a massive steam engine with a massive electrical motor of similar capacity. The gains promised from electrification were not to be seen. It took three decades and some serious reimagination before designers figured out that the shop floor is freed of the yoke of a single shaft. That led to flat shop floor designs with each production equipment driven by smaller rated electrical motors sitting right on the equipment itself. The benefits of electrification of the shop floor then started emerging.

The first level of reimagining the shop floor with digital, primarily discrete implementation, is already done. The shop floor of today has been reimagined beyond recognition compared to the shop floors of the analog era that one would have seen a few decades ago. The journey for the next level – real-time coordination between machine and machine, is underway (e.g., cobots). The final level – real-time coordination between man and machine – what goes under the rubric of cyber-physical systems, has seen limited applications in the shop floor so far. The combination of (1) the ability of shop floors to communicate with other distant shop-floors, warehouses and logistics infrastructure real- time, (2) the really low-cost computing and storage that enables technologies such as artificial intelligence and machine learning, and (3) the potential for mobility in all these, opens up new possibilities of cyber-physical systems configurations that will bring a paradigm shift to manufacturing design and implementation and how it links to the rest of the business, customers, suppliers and the market. Digital manufacturing needs to be reimagined again, now, in light of these advances and all their ramifications for business as a whole.

Among the many stakeholders who want a piece of action in the digital transformation of manufacturing, the ones who can reimagine manufacturing not only in novel ways that radically improves quality and efficiency (the low hanging fruits), but innovate new ways of doing things within manufacturing and how it connects to the rest of the business ecosystem (the real trophy) will most likely emerge winners in this game.

About the contributor

Prof. Sai Prakash Iyer is an Adjunct Professor of Strategic Management at IIM Udaipur. A keen follower of all things digital, his teaching and consulting focuses on the business impact of digital technologies and its implications for business strategy, competitive dynamics and organizations.




Leave a Reply

Your email address will not be published. Required fields are marked *

twenty − six =