Long before artificial intelligence took off, manufacturers were already running a range of software to collect data and simulate production lines. The tools span products from industrial engineering heavyweights such as Aveva and Siemens to in-house programs built by a plant's own engineers. As the volume of data and the number of tools have grown, so has the need to link them on a single platform where operations can be monitored and managed in real time at a glance. Simacro has built a digital twin solution to do that, and is now knocking on the door of the global market.
"Manufacturing processes have to run with great precision, according to the principles of physics and chemistry," Yoon Jung-ho, chief executive of Simacro, said in an interview with The Seoul Economic Daily at the company's office in Seoul's Gangnam district on the 16th. "We built a digital twin solution that grafts AI onto conventional models grounded in those principles, so it can analyze a process accurately and propose ways to improve it."
ProcessModel V, the digital twin solution Yoon developed, targets petrochemicals, steel, oil refining, food and pharmaceuticals. Manufacturing divides into two types by production method. In discrete manufacturing, as in automobiles and semiconductors, parts with a physical structure are assembled into a product. In process manufacturing, as in chemicals, biotechnology and food, products are made by harnessing the chemical properties of materials. Unlike discrete manufacturing, process manufacturing hinges on precisely controlling conditions such as temperature, pressure, flow rate and concentration. Another trait unique to it is that a finished product is hard to return to its original raw material state.
Yoon founded the company around digital twins for process manufacturing because he saw an opening there. "When people talk about digital transformation on the factory floor, they usually picture physical processes," he said. "Chemical and biotech processes need an AI model that sets guardrails based on first principles."
Many manufacturers agree such a tool is needed, but actually building and using a digital twin is not easy. "Collecting industrial data from various sensors, turning it into a simulation and analyzing it with AI have all been run separately," Yoon said. "Collecting data did not mean you could explain or predict an anomaly, and the older approach also fell short on visualization."
Simacro's digital twin combines physics-based simulation with real-time data and AI analysis to stand up a digital twin quickly. It also offers intuitive data visualization and automated insights. Yoon points in particular to "model ops," the ability to link a company's existing tools on one platform, as what sets the product apart. Model ops refers to an environment in which assorted models and software are connected to data so they can be validated, managed, run and improved. "There are plenty of companies offering solutions, from firms like Aveva and Siemens that support physics-based offline processes to Palantir Technologies, which is built on online data," Yoon said. "Simacro plays the role of tying them together in the middle."
Development of the feature began after a customer raised the need and came to Simacro. Manufacturers typically have several engineers assigned to each piece of software. Their proficiency varies from engineer to engineer, and managers had trouble viewing the status of every tool in one place. Simacro spent three years building a feature to solve that. "We provide 100% support in our solution for AspenTech software, which most manufacturers use," Yoon said. "Models a company has developed itself can also be linked after a proof of concept."
That engineering is translating into cost savings on the factory floor. A Japanese chemical company cut part of its production costs using Simacro's software and became a paying customer after a proof of concept.
Simacro is also working with the largest energy company in the United States, which is looking for ways to cut carbon emissions. Because of the nature of the business, securing data from each device and piece of equipment and estimating chemical reactions is complex work. Simacro links the relevant data and various models on its own platform so that analysis can run automatically. "Through our solution, this energy company can now obtain data on how it should design equipment to reduce carbon emissions," Yoon said. "Lately we have been discussing using our service to view crude oil reserve estimation models and crude extraction models in an integrated way."
Simacro has also applied to the Ministry of SMEs and Startups' Global TIPS program and is awaiting the result. The program identifies promising technology startups capable of expanding overseas and supports their research and development and commercialization. If selected, the company plans to validate the commercial viability of its solution together with the global energy company ExxonMobil.
At home, it has signed formal contracts with CJ CheilJedang and Samsung C&T after proofs of concept. Simacro was able to land major clients at home and abroad because Yoon, drawing on deep expertise in the field, moved quickly to develop the technology. A chemical engineering major in college, Yoon spent more than a decade at AspenTech developing modeling software from 2000 to 2012. He later worked at the French engineering firm Technip and the Saudi Arabian petrochemical company SABIC.
With demand for AI transformation running high among manufacturers, Yoon sees now as a prime opportunity to expand. "Demand for digital twin solutions is high in the Southeast Asian market, and we are running proofs of concept with Middle Eastern companies as well," he said. "There is demand from mid-sized Korean companies too, but we still have no marketing or sales team, so we are weighing how to handle it."
Yoon founded the company in the United States in 2018 and set up a Korean entity in 2020. It has raised 3.5 billion won ($2.5 million) to date, with backers including Bluepoint Partners and IBX Partners. Late this year it will seek additional funding, led by global investors, in a round expected to total 15 billion won ($10.8 million).
Yoon plans to spend the money on advancing the AI features. "We want to use open source to build a large language model specialized for manufacturing," he said. "We are also developing an AI agent with global enterprise sales capabilities." He added: "In manufacturing AI transformation, you cannot ignore the legacy players, and it pays more to lay the groundwork so their solutions can be applied. I hope our product becomes the standard for production processes."
