An AI-driven digital twin has fundamentally altered the way new products are developed in the automotive sector
Your product might be a game changer. But if it takes two years longer than your competitors to bring it to market, that’s not considered a win in the chemical process industries (CPI).
In today’s world, customers want the right formulation for their products now — and oftentimes before they know they need it. An increased speed-to-market approach, combined with an evolving list of new regulations and sustainability standards, has driven companies like global specialty chemical manufacturer Lubrizol Corp. (Wickliffe, Ohio; www.lubrizol.com) to rethink the process of product development and to ask how digital technology can be applied to address myriad challenges.
The need for speed
A historic challenge came forward about three decades ago when Lubrizol’s customers — namely, original equipment manufacturers (OEMs) in the automotive sector — started to face increasing competitive pressure to enhance performance and fuel efficiency, as well as to meet more stringent global emissions regulations. Entire vehicle designs needed to be revamped — from aerodynamics to engine structures — to meet these standards. And as a critical player in the supply chain, Lubrizol needed to find a way to speed up the development of new engine oils, fuel additives and lubricants to keep pace, all while enhancing product quality, reducing costs for customers and emphasizing sustainability benefits.
Necessity being the mother of invention, Lubrizol created its Q.LIFE platform to address these issues, and along the way discovered groundbreaking formulations and created new research and development paradigms.
Digital twins and AI models
Lubrizol’s Q.LIFE platform is a digital twin, or virtual replica, that simulates vehicle engine performance and internal friction characteristics over time. As a result, engineers and designers can understand how different formulations of engine oils and fuel additives impact a vehicle’s performance and lifespan. Simply put, virtual replicas take the place of physical objects, processes or systems. We often think of the process or asset as a proverbial “black box” — a system where we can observe data inputs going in and outputs coming out, but where the internal workings are not fully understood or are too complex to understand completely using first principles. With Q.LIFE, teams can analyze the inputs and outputs, as well as the relationships they have to each other. A statistical or mathematical model is then developed, which essentially represents how the system behaves as a function of various measurable inputs. The objective is to use the data to create a virtual representation of the black box, combine it with domain expertise and first principles, allowing users to analyze and understand its performance.
Prior to Q.LIFE, Lubrizol’s engineers would have to conduct complex experimental designs and test combinations of variables on real engines to create a new fluid formulation (Figure 1). Customers had to wait weeks to see results. Furthermore, each test would come with a large cost and generate significant emissions.
FIGURE 1. In the past, researchers had to conduct numerous manual experiments in order to test new products like engine oils
Now, engineers test formulations virtually. A trained engineer can digitally modify upwards of 50 characteristics of a base formulation’s chemical nature and select the engine for testing. Using advanced statistical models and machine-learning algorithms, the platform models the behavior of different engines to predict the formulation’s impact on fuel efficiency, engine wear, chemical composition of oil and, ultimately, its performance over time. This saves hours of time, hundreds of dollars and tons of emissions.
Importantly, AI models and the tools that rely on them are only as good as the data that inform them. Quality data are not manufactured overnight. Lubrizol spent almost three decades making sure the datasets it collects are consistent across its global footprint. This allows teams to achieve a high level of model accuracy and ensure that models become better and smarter as they learn from data over time.
The competitive advantage
The immediate benefits of using digital twins within Q.LIFE in Lubrizol’s research and development process have been numerous.
First, Lubrizol and its partners have realized millions of dollars in cost savings. Engine tests can cost millions of dollars per year. Q.LIFE cuts costs by orders of magnitude, helping to provide users with cost-effective formulations.
Second, Q.LIFE has increased speed-to-market capabilities, as companies can adapt quickly to changing needs and develop new products faster than ever, cutting down the typical product development cycle from years to months.
Third, digital twins have improved workforce efficiency. Instead of spending time conducting rounds of experiments, Lubrizol’s chemical engineers and data scientists can now focus on the creative problem solving that leads to industry breakthroughs.
Most importantly, Q.LIFE testing has helped both Lubrizol and its customers reach their sustainability goals. Testing formulations virtually allows engineers to identify the formulations that are most likely to succeed. These are the formulations that move forward to real engine testing, minimizing fuel consumption and reducing emissions across the R&D process. In addition, Lubrizol’s customers are continually seeking products with a sustainable lifecycle analysis (LCA). Engine oils developed through virtual testing have greater end-to-end operational sustainability, which reduces the carbon footprint for Lubrizol and for its downstream customers.
Breakthroughs on the horizon
Digital twin correlations on variables like engine wear and tear, emissions and fuel efficiency yielded surprising first-, second- and third-order effects. Such findings have resulted in simplified formulations. Tests can reveal how products can accomplish the same goals with fewer formulation components, reducing complexity and cost. That means more cost savings in procurement, as well as simplified formulations and operations.
Lately, as new regulations are introduced, some formulations and testing are out of the current models’ scope. This creates a challenge, because new testing can be both capital- and time-intensive. To combat this, Lubrizol has been continuously creating new models in Q.LIFE. Instead of starting from scratch, engineers can extrapolate findings from existing data and as new data are collected. Then, they can apply proprietary algorithms to facilitate transfer learning, which is a machine-learning technique where a model trained on one task is reused and adapted for a different (but related) task. This is just one example of how AI is helping accelerate new product development.
Lubrizol is also using digital twins in its manufacturing processes. For example, when manufacturing engineered plastics, the manufacturing and extrusion process can be quite complex. Modeling not only the first-principle variables, but also various other inputs in the manufacturing and scaleup process helps Lubrizol better control its operations, predict manufacturing yield and create a high-quality product. Such digital twins provide sustainable and highly efficient operations.
Now, when confronted with a time-consuming, labor-intensive problem, we often ask ourselves, “Can you make a digital twin?” Automating certain R&D processes has driven sustainability gains, improved shareholder returns and strengthened value proposition to industry partners, helping foster sustainability, creativity and efficiency. ■
Author
Abhishek Shrivastava is Lubrizol’s managing director – India, Middle East and Africa (IMEA). In this role, he is responsible for leading Lubrizol’s IMEA leadership team to deliver differential growth, based on a local-for-local approach. He brings nearly 20 years of industry experience, including previously serving as Lubrizol’s vice president, Innovation and Decision Science. In this role, he was responsible for leading a team focused on the development and oversight of innovation growth platforms, sustainable product development and digital innovation tools. Prior to Lubrizol, Shrivastava held a variety of leadership roles at Dow Chemical and DuPont related to business transformation, new business development, R&D, manufacturing technology, quality and technical services and site leadership. He has a bachelor of technology degree in chemical engineering from the Indian Institute of Technology (IIT), Kharagpur, India, and a Ph.D. in chemical engineering from the University of Minnesota.