In high-mix environments, inspection systems must first be flexible – everything else builds on that foundation

High-mix pharmaceutical manufacturing demands inspection systems that can adapt quickly to changing products, formats and batch sizes without compromising performance. Flexible inspection platforms are therefore becoming a critical foundation for modern production environments. However, while mechanical flexibility enables frequent changeovers, engineering tasks such as recipe development and optimisation often remain tied to the production line. This article explores how combining flexible inspection systems with a digital engineering layer allows manufacturers to shift these activities upstream, improving stability, reducing iteration effort and enabling more predictable performance in increasingly variable production scenarios.

Pharmaceutical manufacturing is undergoing a structural shift. As personalised medicine, clinical pipelines and advanced therapies scale, the production paradigm moves from high-volume standardisation towards high-mix flexibility.

This shift introduces three compounding pressures:

• More product variants and frequent changeovers

• Smaller batch sizes with higher economic sensitivity

• Increasing requirements for traceability, documentation and audit readiness.

Visual inspection is one of the first processes to feel this pressure. It sits at the intersection of product quality, regulatory compliance and line performance. Any instability – whether mechanical or digital – directly impacts throughput, yield and release timelines.

This shift is particularly visible in high-value drug segments, where complex biologics and injectable therapies place additional pressure on inspection performance. In these contexts, even minor deviations can have significant economic implications and product loss must be minimised. At the same time, increasing demand and compressed timelines require faster ramp-up and stable operation from the outset, further intensifying the need for predictable inspection performance.

To manage this complexity effectively, inspection variability must be understood in two dimensions:

Differences in container types, formats and handling behaviour introduce mechanical complexity. Frequent format changes amplify the importance of robust handling concepts, fast and repeatable changeover procedures and stable inspection performance across formats.

Inspection outcomes depend on digital parameters such as recipe logic, thresholds, trigger points and defect classification. These elements are often developed and optimised directly on production equipment – where every iteration consumes valuable machine time and introduces operational risk.

While automated inspection systems have significantly improved consistency and throughput, many existing approaches still struggle in high-mix environments. Traditional rule-based systems are optimised for stable, repeatable conditions and can reach their limits when confronted with frequent product changes, varying material properties and evolving defect characteristics.

As variability increases, maintaining a consistent balance between detection performance and false reject rates becomes more complex. What works for one product or batch may not translate directly to the next, leading to increased tuning effort and reduced predictability during ramp-up.

A key shift in advanced inspection strategies is the relocation of engineering work from the production line into a controlled digital environment. A digital inspection twin designed to support engineering and preparation workflows enables simulation of inspection relevant machine behaviour, as well as offline development and optimisation of inspection recipes. The digital representation is intended to support engineering decisions and preparation activities.

This approach supports the creation of a ‘digital golden recipe’ that serves as a structured and traceable preparation baseline for subsequent implementation.

The shift towards digital engineering must be compatible with regulatory expectations, including life cycle-based validation, data integrity and risk-based approaches to software assurance. Rather than treating validation as a downstream documentation exercise, successful implementations integrate compliance into the operating model.

• Version-controlled changes with full traceability across the engineering life cycle

• Structured documentation supporting engineering decisions and test activities

• Risk-based validation aligned with system criticality and executed on the production system.

Traditionally, recipe development competes with production time and training requires real material and equipment availability.

Key characteristics include:

• Clear separation of development, test and production environments

With a digital twin, recipes are prepared and refined before deployment, enabling a more efficient and controlled validation process on the production system. Training becomes repeatable and rare defect scenarios can be analysed systematically. This allows machine time to be used primarily for production rather than iterative optimisation.

Another often underestimated aspect is the growing complexity of inspection knowledge. As systems become more advanced and product portfolios expand, maintaining expert-level understanding across operators and engineers becomes increasingly challenging. Standardised approaches to recipe development and the ability to train under controlled, repeatable conditions can therefore play a key role in improving consistency and reducing dependency on individual expertise.

Flexible inspection systems provide the operational foundation for high-mix performance. Digital engineering approaches build on this foundation, enabling more efficient preparation, improved consistency and greater control over inspection outcomes.

This can lead to:

• Faster and more reliable changeovers

• Higher process stability

• Support for the reduction of false rejects

• Shorter ramp-up phases

• Improved consistency across sites.

One of the central challenges in automated inspection is the trade-off between detection sensitivity and false reject rates. Increasing detection thresholds can improve defect identification but may also lead to higher rejection of good products, directly impacting yield and cost. In high-mix environments, this balance becomes even more critical, as each product variant may require different parameter settings. Improving decision quality, rather than simply detection capability, therefore becomes a key performance driver. Beyond operational improvements, inspection systems are also evolving in their role within the overall manufacturing environment.

Inspection systems are increasingly evolving beyond their traditional role as defect detection units. In modern pharmaceutical manufacturing, they are becoming an integral part of the overall data landscape, generating structured, high-frequency information about product quality, process stability and system performance.

This shift creates new opportunities, but also new challenges. While data availability is no longer the limiting factor, the ability to interpret and use this data effectively becomes critical. Inspection data must be contextualised, linked to upstream and downstream processes, and translated into actionable insights that can support decision-making. In high-mix environments, this becomes particularly important. Variability across products and batches generates a wide range of inspection outcomes, making it harder to distinguish between acceptable variation and true process deviations. A more integrated, data-driven approach to inspection can help improve transparency and enable more informed adjustments.

As a result, inspection is no longer just a quality gate at the end of the line. It becomes part of a continuous feedback loop that supports process optimisation, stability and long-term performance improvement.

A structured rollout includes defining target outcomes, establishing data foundations, validating a focused use case, and standardising and scaling. This approach ensures measurable benefits while maintaining regulatory alignment.

High-mix, low-volume manufacturing increases both physical and digital complexity in inspection processes. By combining flexible inspection systems with a digital engineering layer, manufacturers can shift critical activities upstream, improve robustness and create a more controlled implementation pathway. Variability becomes predictable and manageable, enabling stable and efficient inspection performance.

Robert Kibele is head of Product Management at Körber Pharma. He works in the fields of pharmaceutical inspection and manufacturing technologies, with a focus on advanced inspection processes and their integration into complex production environments. With a background in packaging and industrial engineering, he is dedicated to translating industry requirements and emerging technologies into robust, scalable solutions for pharmaceutical manufacturing.