Quantillion Technologies Case Study

What is an AI workstation? The one type of computer your power users are pleading for

Power users like engineers, data scientists, and designers require extra computing capacity, but equipping them is not as simple as that.

Quantillion was provided loaned HP products from HP.

2

For years, manufacturers have battled to make complex manufacturing processes faster, more efficient, more cost-effective, and especially less impactful on the environment. The challenge may have gotten a bit easier with factory process simulation and change analysis technology from Quantillion Technologies using AI and an HP Z8 Fury G5 Workstation.

The Dutch business, headquartered in Amsterdam and comprising 11 data

scientists and AI experts, uses advanced technology to improve manufacturing operations. It has developed a solution that creates a digital twin of factory processes to test and analyze how changes in different parts of a process can transform operational efficiency. The digital twin is needed because it replicates a real factory process and can be used, for example, to simulate operations and change scenarios an hour, day, or year ahead.

HP Z helps target low-cost, energy-efficient, and climate- friendly factories of the future

3

Building an AI playground Dajt Mullaj, AI Engineer at Quantillion Technologies, explained why a digital twin is important. “If you only use actual data, AI only learns what happened in a specific situation. That is useful, but it cannot answer ‘what if’ types of questions. The digital twin enables AI to propose completely new changes and determine impacts across the whole factory. It is like giving AI a playground to try new and different strategies.”

The technology works by gathering real-time factory data, using it to develop a digital twin and then applies AI to model and test changes in the digital twin. Critically, the Quantillion solution models how change impacts the whole factory, not just individual parts. For example, how does improving one element of production impact processes before, after, and over time?

Once created, the digital twin generates its own data only using real data to validate the simulation. Quantillion needed the power of HP Z technology to handle the complexity and volume of data required to create the digital twin and AI modelling. For instance, a full training run that takes one week of operations involves 50 million decision points. To handle that, Quantillion used the HP Z8 Fury G5 Workstation with single Intel® Xeon® w9-3495X CPU running at 1.90Ghz with an NVIDA RTX™ 6000 Ada 48GB GPU, enabling them to run more simulations simultaneously.

“With the high- performance computing power of the HP Z8 Fury G5, we can test more scenarios and ways to improve factory operations faster. This means we can demonstrate the benefits of innovation, like decarbonization, to manufacturers sooner.”

Dajt Mullaj AI Engineer Quantillion Technologies

4

“We need the power of the HP Z8 Fury G5 for either fast training or training in parallel, so you can test multiple ideas at the same time.”

Dajt Mullaj AI Engineer Quantillion Technologies

HP Z enables game-changing performance

“The HP Z8 Fury is a great piece of technology,” commented Mullaj. “As an AI engineer, it allows me to try more options faster. One of the challenges of AI is waiting to iterate through ideas as AI takes time to process. We need the power of the HP Z8 Fury G5 for either fast training or training in parallel, so you can test multiple ideas at the same time.”

Preventing molten iron heat loss Quantillion has used its technology and the HP Z8 Fury G5 in a joint project with Tata Steel Netherlands to develop a Decision Support System to reduce CO2 emissions in steel production. The project aims to reduce temperature loss when transporting six-meter-long, torpedo-like containers each with 400 tons of molten pig iron between the blast furnace and steelmaking plant. Although on the same site, these torpedoes can take up to 20 minutes being moved by rail.

5

The HP Z8 Fury G5 was used to make a digital twin of the transportation process and then run millions of scenarios to determine the fastest and most efficient way of moving all the pig iron torpedoes between the two locations. The AI model simulates and

predicts the behavior of the pig iron during transport so that adjustments can be made in real time. Usually, torpedoes are used on a first-come, first-served basis but, as it turns out, this may not be the best way to optimize heat retention.

About Quantillion

Headquartered in Amsterdam in the Netherlands, Quantillion is on a mission to bring the power of autonomous decision-making to every machine, every worker, and every factory. Leading industries worldwide are increasingly focusing on automation as a value driver. Automation lowers operational expenses, streamlines operations, and affords the certainty needed to focus on core tasks.

Saving 10 kilotons of CO2 Using the HP Z8 Fury G5, Quantillion was able to quadruple processing performance. It ran up to 13 parallel simulations in three days instead of three or four in a week1 when compared to an existing computer built in-house – representing a performance improvement of over 200%2. The HP Z8 Fury G5 enabled the Decision Support System to test millions of different scenarios in just two hours and showed that process optimization could save several degrees of temperature loss. As a result, Tata Steel reduced pig iron temperature loss substantially. For every million tons of steel Tata Steel produces a year, reducing heat loss achieves a CO2 saving of five to ten kilotons3.

As well as reducing energy needed to reheat pig iron and making operations more efficient, hotter pig iron enables Tata Steel to use more recycled steel in its manufacturing process. One of the other key benefits of the Quantillion solution is enabling manufacturers to make significant improvements to factory operations and impact on climate change almost immediately, without the need for a huge capital investment.

6

7

Minimizing climate impact “Having the power of the HP Z8 Fury G5 is critical,” said Mullaj. “To achieve meaningful results, we must train models using complete and multiple circumstances and each of those needs extensive training. If we train the model using a specific amount of pig iron torpedoes but in real life there are more or less torpedoes, the model breaks down. The more extensively and comprehensively you can train a model, the more it is able to accurately predict the outcome of unknown or unforeseen events.”

Industry Factory simulation

Objective Help manufacturers reduce the impact of heavy industry processes like steelmaking on climate change by using digital technology to create factory simulations that can be used to test and analyze both future and real-time operational changes.

Approach Use the high-performance HP Z8 Fury G5 Workstation to develop a digital twin of a factory process, then apply AI modelling to run thousands of different scenarios on the digital twin to analyze the impact of change and how it can improve efficiency, cost-effectiveness, and sustainable manufacturing.

7

IT matters • Hardware: HP Z8 Fury G5 Workstation

with single Intel Xeon w9-3495X CPU, 1.90Ghz with an NVIDA RTX™ 6000 Ada 48GB

• Software:

• Windows 11 Pro • JavaScript to develop in-house

simulation program • Python for AI modelling

• VS Code and Unity Editor applications

Business matters • Improves sustainable manufacturing

industry-wide

• Helps Tata Steel cut five to ten kilotons of CO2 per one million tons of steel

• Speeds up factory simulation building by over 200%

• Two hours to test millions of simulated scenarios

Quantillion plans to develop a product that manufacturers can use to optimize their own factory operations, reduce costs, and make processes less harmful to the environment.

“Many of the processes in heavy industries like steelmaking are still very traditional and difficult to change,” noted Mullaj. “With the high-performance computing power of the HP Z8 Fury G5, we can test more scenarios and ways to improve factory operations. This means we can demonstrate the benefits of innovation, like decarbonization, to manufacturers sooner.”

8

What is an AI workstation? The one type of computer your power users are pleading for

Power users like engineers, data scientists, and designers require extra computing capacity, but equipping them is not as simple as that.

HP Z Workstations & Solutions for product designers & engineers From diffusion modeling and 3D modeling to real-time simulations and more, tackle your most complex projects with powerful AI workstations built to handle the most demanding design and engineering workflows.

1 Quantillion ran same modelling process on an existing device vs. HP Z8 Fury G5 to compare processing speed. 2 Calculated by counting % difference between 4 and 13 simulations - actual is 225%. 3 Quantillion/Tata Steel webinar on Decarbonization of Steel Production, SIM-PAC 2023.

© Copyright 2025 HP Development Company, L.P. The information contained herein is subject to change without notice. The only warranties for HP products and services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed as constituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein. 4AA8-4852ENW, June 2025

Learn moreLearn more about HP Z high-performance AI workstations

https://www.youtube.com/watch?v=2eETAgQJKvs https://www.hp.com/us-en/workstations/industries/data-science.html