Manufacturing technology

He's shedding light on companies' hidden costs – and saving them millions

DTU researcher aims to help manufacturing companies become more competitive and make better decisions by using artificial intelligence to find the invisible costs their financial systems miss.

When companies such as Terma need to develop a new radar consisting of thousands of components, AIMO can give insights into the actual costs by creating an entirely new product versus reusing components from existing products. Photo: Terma

Friday 12 December 2025

Sole Bugge Møller

If water starts leaking from a pipe, it might not be noticed immediately. Similarly, many manufacturing companies have numerous hidden costs that slowly accumulate and can lead to significant losses if not detected and addressed. A DTU researcher aims to plug these small leaks using artificial intelligence.

Most manufacturing companies use ERP systems (Enterprise Resource Planning) that provide an overview of everything from finances to inventory, supply chains, procurement, and sales, so they know the cost of most decisions in monetary terms. However, ERP systems are static and have many blind spots. They only account for direct costs and not the extra work required to produce one product over another, or the additional maintenance or storage space it will entail.

"So they don't get the real picture of what it will cost to make one decision over another," says Carsten Keinicke Fjord Christensen.

He is a postdoc at DTU Construct, specializing in modularization, which is based on the idea that reusing components in multiple places simplifies many processes and saves time and money. In his research project AIMO, Carsten Keinicke Fjord Christensen and his team of researchers have developed a new solution that uses artificial intelligence to analyze company data, predict how changes will affect the entire organization, and help make better decisions.

"It's about making fundamentally better decisions in the company and making them smarter about trade-offs and how decisions affect their colleagues' tasks. Ultimately, they become more productive and get more out of their resources," says Carsten Keinicke Fjord Christensen.

An hour saved can waste others' time

Carsten Keinicke Fjord Christensen believes that most companies think of office hours as fixed – they have already accounted for the cost of paying employees and therefore think less about what a new process will cost in work hours. When a company decides whether to produce product A or B, the ERP system looks at direct costs like materials and wages but doesn't account for the work hours needed to develop and maintain the two products.

For example, an engineer might save an hour by designing a new component for a product instead of finding an existing one. However, colleagues in the organization will spend several  hours figuring out everything from how to manufacture the component, what materials to use, where to store it, and how to implement it in the final product.

"We would rather use our skilled people to develop something new for the customer instead of having them deal with unnecessary complexity," says Carsten Keinicke Fjord Christensen.

Oops, this feature requires cookies

To view the content, you must change your cookie consent to allow functionality and targeting cookies

Qualifying what we can't comprehend

Carsten Keinicke Fjord Christensen has tested the AI model with several companies, including defense technology company Terma.

Terma manufactures radar systems that easily consist of up to 15,000 components. AIMO examines each nut and bolt when estimating the cost of making a new product or changes to a product, which is useful for Martin Elkjær, who is responsible for the development and maintenance of Terma's radar solutions.

"We are constantly expanding Terma's product range, so it's relevant to look at how we best approach the task and create the most value: where components most often become obsolete, so we need to redesign an existing product without improving it in the customer's eyes. AIMO can qualify something based on many data points that we otherwise can't comprehend," he says.

Can save millions

The collaboration with AIMO has shed light on some internal processes where different departments need to handle individual components multiple times, which is very time-consuming. Terma can potentially create huge savings by using AIMO.

Carsten Keinicke Fjord Christensen had the AI model analyze a specific example where Terma’s existing ERP system would recommend developing a certain type of radar over another. However, when AIMO included costs for maintenance, development, planning, etc., it would be much cheaper to go with the alternative. In fact, Terma could save a significant amount over the product's lifetime.

"There is potential for significant savings by making hidden costs visible, including friction in the process. Even if the model isn't entirely accurate, it fundamentally makes our work more efficient and helps improve our competitiveness," says Martin Elkjær.

Terma has not yet decided whether to proceed with the AIMO model after the end of the research project, and Martin Elkjær acknowledges that it requires fundamental trust in the AI model and the output it generates.

"I can certainly see the value in AIMO being an input in the decision-making process, but it will never stand alone," he says.
When the AIMO research project concludes later this year, Carsten Keinicke Fjord Christensen hopes to turn it into a startup. He is currently working with DTU Skylab to realize its potential, and the AI model also needs further training and optimization.

Contact

Carsten Keinicke Fjord Christensen

Carsten Keinicke Fjord Christensen AC TAP Department of Civil and Mechanical Engineering

Innovation and product development Operations analysis Systems analysis Artificial intelligence (AI)