Digitalization in the process industries: Humans as success factor in production

Publish date:

Classic technologies of digitization create sustainable process efficiency in production. But what are the challenges of using data-driven digital methods in the process industries?

Goal: Sustainable process efficiency in production

Process industries (chemical, pharmaceutical, industrial biotechnology, paper, cement, glass, food, and water) have made significant progress in sustainability and productivity in recent years. In the chemical industry, for example, resource and energy efficiency as well as process and occupational safety have improved steadily and remarkably through the development and application of better production processes. This was achieved primarily through better chemical and technical understanding and implementation at the production level. In both fields, classical technologies of digitization such as (soft) sensors, modeling and simulation, advanced process control, etc. play an important role.

Current situation: Room for improvement

Due to the very long lifecycles of typical chemical plants, on the other hand, it is still common to find that production processes are based on old, partly no longer fully understood knowledge and a great deal of experience. Typically, much data is recorded and stored, but the latter in scattered silos and without internal and mutual connection, thus difficult to access for a systematic evaluation.

If a change in the raw materials, the product specification or due to regulatory changes is required in such plants, then this is cumbersome to implement within a reasonable development effort and often does not lead to an optimal operation.

Also, an increase in plant capacity, e.g. through debottlenecking or new investment, is often associated with considerable development effort, because a detailed process and plant understanding is missing or the bases for the scale-up are often no longer available.

Even where extensive know-how has been built up for recent plants in process development, including possibly process models, a consistent transfer of this knowledge over plant design, construction and commissioning to production is not always given. Finally, even successfully transferring process knowledge from development to production level often lacks the validation of that knowledge with the help of widely available production data. However, this is not primarily a matter of technology, because in recent years big data analytics and numerous cloud-based offerings have developed an agile method tool set that enables application specific implementation in very short sprint cycles in just a few weeks.

In the field of methodology, the classical-deterministic modeling and simulation of thermodynamics, kinetics and processes (e.g., cycle times etc.) has been established for chemical process development. Even if the motto “No model, no plant” is not consistently maintained, modeling the critical process steps (reactors, thermal separation) is common practice. The same methodology is also used for overcoming capacity bottlenecks (debottlenecking).

However, only in parts of the current production an improvement in the key parameters is realized by systematic analysis of existing data and use of the insights gained. Even the lowest levels of the automation pyramid – the control of individual process parameters – show, according to a study [1], a surprisingly little knowledge-based status: one third of the control circuits is not in operation at all, another third is operated in the basic settings of the control parameters and only the last third has adapted control parameters.


What are the reasons that delay the more consistent application of data-driven digital methods – as is common in other industries – in the process industry?

As with many change processes, it is largely the people who decide – willfully or subconsciously – on success or failure. It must be recognizable to them how digital transformation addresses their concrete questions on the shop floor. There is no one-fits-all strategy, every question requires individual approaches and solutions.

The White Paper “Professions 4.0 – How Chemists and Engineers Work in Digital Chemistry” by the Association for Chemistry and Business, VCW (GDCh Section) identifies two major challenges:

“Many chemists and engineers want to be involved wherever possible and keep control in detail. Future competency-based job profiles and work content – along with ‘sharing’ tasks and ‘handing over’ individual responsibilities to other key competences – are what make up most chemists and engineers. Emerging professions such as rather IT-oriented pure data scientists, more balanced chemistry computer scientists and primarily chemical and business oriented value chain managers as well as stakeholder managers in the environment of large projects are currently known as a concept, however, the belief in them as new colleagues with specialist skills is almost completely missing.”

In addition, there are barriers in the grown IT and process landscapes. Acquisitions, project-driven expansions, migrations or uncontrolled conversions as well as departmental along with country-specific data separations and long-term vendor contracts lead to missing interfaces, missing standardizations, data and knowledge silos as well as uncoordinated master data. IT thus puts the brake on innovation as consolidation projects can uncover unplanned risks very late and make calculations difficult. The successful fit/gap analysis in the context of a viable enterprise architecture is often skipped and any risks for the short-term success of the project are ignored.

Solution approaches

A generic flow of digitization projects includes:

  • Collection of use cases
  • Data acquisition/storage
  • Data evaluation
  • Data analysis
  • Contextualization
  • Prioritization
  • Architectural planning
  • Measures definition/implementation
  • Data integration and scaling.

The holistic view of the value chains in the ecosystem, which unites partners, customers as well as operations and which develops agilely for suitable architectural cuts, inevitably leads to increased transparency, increased reactivity and better predictions. The central success factor of digitization projects in the process industry has been the contextualization of correlations (identified initially purely empirical from data analyses) with physical-chemical-engineering laws

At the beginning there is the problem definition -> which question does operations currently concern: quality, yield, capacity, cost, flexibility, process tolerance, logistics, etc.? Often a quantification will be necessary -> what is the specific size of the problem and what is the potential for improvement?

The analysis of the data can be done by deterministic (modeling, simulation) or statistical (correlation analysis, etc.) methods. While the deterministic methods require a basic understanding of the underlying laws, statistical methods provide purely empirical correlations that need to be put into context. Precisely this step requires interdisciplinary work, in which the participating chemists and engineers have to give up part of their control and competence. Experience has shown that this happens naturally where the achieved or potential improvements are obvious. Success is collectivized and the motivation and understanding as well as interest in change processes increase enormously. Successful implementations from related industries are being adapted currently to the chemical process landscape with growing interest. For example, predictive analyses from the automotive environment have already arrived alongside smart supply chain implementations for yield optimization in many areas of chemistry.

In addition to significantly reduced project risks, short implementation times have the advantage that evidence of increased added value can be shown just a few weeks after they have been implemented. The classical development cycles are eliminated, the generated savings or the income from new data-driven business models help to steer the overall budget in a controllable framework according to value.

The VCW study also focuses on the future: “Universities and colleges are required to train future-oriented people. Without swiftly coordinated and implemented changes among all those affected, today’s preparatory and study periods mean that even in 2030, hardly any PhD graduates in chemistry and engineering will be available with the skills that digitized chemistry needs already today, but in full by 2025 at the latest.”

Many thanks to the authors Oliver Lade, Volker Brendel and Goetz Wehberg.

[1] VanDoren, V.: Advances in control loop optimization. Software takes users from simple tuning to plant-wide optimization. Control Engineering May 2008.

Contact Us

If you would like to know more about the theme of this blog or discuss what we can do for your business, please contact Oliver Lade

Related Posts

Capgemini Invent

Advancing consumer segmentation during COVID-19 and beyond

Tanya Anand
Date icon July 14, 2020

In order to deliver meaningful personalized brand engagement, brands need to better...

Capgemini Invent

Risk culture: from reactive to preemptive

Erekle Tolordava
Date icon July 13, 2020

Organizations with a preemptive risk culture are better equipped to mitigate, minimize, and...

Capgemini Invent

Why cost reduction is just one piece in the post-COVID-19 cost resilience strategy

Etienne Piollet
Date icon July 10, 2020

The far-reaching economic impacts of the 2020 coronavirus will extend long into next year and...