Aerospace, Automotive, Energy, failure analysis, Oil and Gas, Railway

Anticipating Failures to Prevent Damage

20/11/2026
Cetim Failure Analysis

Industrial failures do not occur by chance. Understanding the factors that contribute to them is key to preventing them more effectively.

In industry, a failure can begin with an almost imperceptible defect and ultimately lead to an accident or, more commonly, costly repairs and production losses. Corrosion alone is estimated to cost nearly $2.5 trillion annually worldwide, representing approximately 3.5% of global GDP.

Yet the tools available to design, calculate and inspect equipment have never been more advanced: numerical simulation, structural strength calculations and equipment inspections. So why do failures still occur?

According to Pascal Macquet, a court-appointed expert in mechanical engineering: “There is rarely a single cause or a single responsible party. When you take the time to investigate thoroughly, there are often multiple factors leading to damage, failure or an incident.”

Industrial Environments Are Conducive to Risk

Modern industry itself creates certain conditions that can contribute to failures, starting with the increasing complexity of systems. Today’s systems incorporate a growing number of components and materials, with interactions that can sometimes be difficult to fully understand. This complexity goes hand in hand with increasing specialization of expertise, which can fragment the overall understanding of a complete system. As a result, components that are individually compliant with specifications may, once assembled, give rise to an unexpected failure mode. For example, a vibration transmitted from one component to another may be enough to cause degradation.

Digital tools, now ubiquitous across industry, can also create blind spots when too much trust is placed in them. They enable faster work but can leave less time for the verification steps that once served as safeguards. Certain defects visible on 2D design drawings may, for example, be less apparent in the 3D representations now widely used.

These technological developments do not make failures inevitable. They do, however, highlight the need to reinforce sound engineering practices capable of compensating for these blind spots.

Preventing Failures Rather Than Dealing with Their Consequences

How can the risk of failure be reduced? The most advanced risk mitigation tools, starting with Failure Mode, Effects and Criticality Analysis (FMECA), have an important role to play when the level of criticality justifies their use. However, for most manufactured products, failure prevention relies less on the strict application of a formal methodology than on sound design practices.

Philippe Guégan, a specialist in new product design at Cetim, identifies several practices that remain underestimated today, beginning with the clear definition of requirements. “Poorly defining the requirements at the outset means discovering the problems only once the product design has already been frozen,” he explains.
This means understanding how the product will actually be used and under what operating conditions before validating the technical solution. A second key practice is to embrace iteration.
“Getting it right the first time is an illusion, both in design and in manufacturing,” emphasizes Philippe Guégan. A robust product may require revisiting the initial design to correct its weaknesses.
This approach should also encourage the pursuit of simplicity. The fewer the components, moving parts and different materials involved, the fewer the potential sources of failure.

Finally, reliability is built into the design itself through choices relating to material selection, dimensional tolerancing, manufacturing processes, inspectability and maintainability. Each component must be designed with its operating environment in mind, and the complete system must then be verified as a whole, since an assembled product may no longer behave according to the assumptions made at the outset.

Learning from Structural Weaknesses

Even when a structural weakness eventually emerges, it can also become a new starting point. The experience of Steriflow, a manufacturer of sterilization equipment for the food processing industry, provides a good example.

The company operated a 1,600 mm diameter rotary autoclave that had performed reliably for 30 years. To meet customer demand for greater processing capacity, Steriflow decided to increase the size of the drum, while the products being processed were also becoming heavier, particularly with the growing use of glass packaging. After a period of operation, the welded stainless-steel structure began to show signs of weaknesses.

According to Laurent Tisserand, Technical Director at Steriflow: “The problem came from the welds. The most straightforward solution was therefore to eliminate them.”

With support from Cetim, the company moved to a bolted structure, while maintaining the same requirements in terms of mechanical strength and service life. But this choice also delivered benefits that went far beyond solving the initial problem: manufacturing was simplified, maintenance was made easier, and production that had previously been carried out in-house could be outsourced. The new design also proved so economically attractive that Steriflow is considering applying the same principle to its 1,600 mm autoclave.
As Laurent Tisserand explains: “Ultimately, we moved from an empirical approach to design and development to an anticipatory approach.”

Mitigating Legal Liability

A common misconception persists: “It didn’t happen on my premises” does not mean “It isn’t my responsibility.”

In practice, a company remains the primary point of accountability to its customer, even when a defect originates from a supplier or subcontractor. It is then up to the company to seek recourse from the party responsible. This is why it is important to reflect in purchasing contracts the commitments made to customers, including delivery deadlines, warranties and limitations of liability.
Without such “back-to-back” agreements, a company may be required to compensate its customer for losses beyond what it is able to recover from its supplier.
In the event of litigation, responsibility is also rarely attributed to a single party. Court-appointed experts increasingly find that technical responsibility is distributed among several parties.

This is another reason why companies need to define their contractual commitments carefully and properly document their supply chain, thereby reducing the risk of having to compensate beyond their fair share of responsibility.

AI for Failure Prevention

The use of artificial intelligence in engineering design is becoming increasingly widespread. By reinforcing a systems-level perspective, AI makes it possible to explore a broader range of potential risks:

  • Explore more scenarios: load cases, extreme operating conditions, manufacturing tolerances, temperatures, corrosion and interactions between components.
  • Detect inconsistencies: by cross-referencing larger volumes of data, AI can identify interactions or contradictions between requirements.
  • Explore alternative solutions: generative design can be used to test numerous geometries and explore different trade-offs between performance, reliability and manufacturing constraints.

However, AI does not validate a design in place of the engineer. Human responsibility remains essential in defining assumptions and validating results, supported by reliable data and a sound understanding of the tool’s limitations

This article is an extract of Cetim Infos N°283
For more information contact Romain Barrellon
, in charge of the “Failure analysis activities“ at Cetim.