Turning Manual Manufacturing Variability into Data-Driven Control – with Sebastian Dykas of Smith+Nephew

Modern industrial manufacturing has long operated under a persistent paradox: plant leaders typically know precisely where efficiency losses, downtime, and scrap accumulate, yet they frequently lack the real-time visibility required to intervene before those losses compound into massive financial liabilities. According to a foundational paper published by the National Institute of Standards and Technology (NIST), a critical barrier to achieving proactive manufacturing is the inherent delay between physical machining operations and subsequent quality assessments. This temporal gap prevents rapid feedback loops from informing active production lines.

The financial toll of these visibility and maintenance gaps is staggering. A landmark 2020 NIST report, utilizing adjusted 2016 economic values, estimated that U.S. discrete-manufacturing sectors bleed roughly $119.1 billion annually due to preventable maintenance issues and operational friction. When unaddressed equipment reliability, unexpected downtime, and sudden spikes in defect rates converge, they transition quickly from localized plant-floor headaches into severe enterprise-level output and revenue concerns.

To explore how advanced operational insights and emerging data methodologies can dismantle these long-standing industrial bottlenecks, Emerj’s Marilie Fouché recently sat down with Sebastian Dykas, Director of Manufacturing Engineering and Maintenance at Smith+Nephew, on the AI in Business Podcast. Dykas, a seasoned operations leader whose extensive resume includes leadership roles at Pfizer, Stryker, and Chrysler, unpacked practical frameworks for transforming manual manufacturing variability into data-driven control.

The Evolution of Industrial Quality Assurance: Moving Past Pass/Fail Metrics

For decades, traditional quality control has relied heavily on binary classification systems. As NIST’s authoritative statistical handbook notes, a fundamental distinction exists between rigid pass/fail classifications and continuous numerical measurements.

In many manufacturing environments, operators routinely check whether a machined part meets basic inspection criteria without retaining its actual, precise dimensions. A standard go/no-go gauge yields a simple acceptance or rejection result, but it completely conceals how a specific feature migrates toward its upper or lower specification limit across successive production runs. This missing historical context robs engineering teams of the opportunity to correct mechanical drift before cumulative wear transforms workable material into expensive scrap.

Dykas illustrates this phenomenon using the classic example of a drilled hole whose diameter gradually shifts as the cutting tool undergoes continuous thermal and mechanical wear. While a vigilant machine operator might eventually adjust the machine parameters to compensate, engineers are often left blind to the underlying wear pattern if neither the granular dimensions nor the subsequent machine adjustments are digitally recorded and preserved.

To combat this, advanced digital frameworks advocate for a three-step operational intervention: capturing exact dimensional metrics, analyzing the historical trajectory of tool wear, and implementing automated or semi-automated interventions before defects materialize. For instance, advanced production software can establish a strict tool-change interval—such as halting operations after precisely 80 pieces—preventing further use until the drill bit is replaced, thereby slashing error rates.

Bridging the Gap Between Measurement and Machine

The logistical hurdle on many legacy plant floors involves manual documentation. Machinists and quality inspectors frequently measure components using an array of precision instruments, manually record the readings onto paper logs, and eventually archive or scan those documents for compliance. While scanning preserves paperwork for auditing purposes, the raw measurements themselves remain trapped in analog silos, inaccessible for real-time trend analysis or automated machine feedback.

Dykas advocates for an integrated workflow where an operator measures a part, transmits the digital reading instantaneously, and the host machine autonomously adjusts its parameters while producing the next component in the queue. Technologies capable of supporting this closed-loop workflow—including wireless measuring instruments, software solutions for automated dimensional compensation, and tightly integrated inline inspection equipment—are already commercially mature.

For manufacturing leadership evaluating their digital readiness, Dykas suggests a straightforward operational litmus test: Can the metrics gathered from the immediately preceding part directly influence the production decisions of the very next part? By confirming that the machine registers its specific product context, production sequence timing, and inspection data, plants can successfully bridge the historical divide between quality assurance and active machining.

Connected Production and Maintenance Records for Faster Root-Cause Analysis

The operational dangers of delayed quality reporting become acutely visible when a scrap event develops silently over several days. In a typical delayed scenario, raw operational data only reaches an analytical software platform long after final inspection. By the time a new scrap category is flagged on a management dashboard, days of defective output have already accumulated, forcing engineering teams to work backward in a costly and time-consuming forensic investigation.

Compounding this issue is the traditional organizational firewall separating production monitoring from maintenance management. In many facilities, if an algorithm or floor supervisor notes that an operator’s shift output is falling behind schedule, an investigation might reveal faulty or degrading equipment. However, the resulting maintenance work order is typically logged in an entirely separate computerized maintenance management system (CMMS) that fails to communicate bidirectionally with the primary production schedule.

To dismantle these organizational and digital silos, Dykas outlines a systematic roadmap for leaders to evaluate their internal handoffs:

• Audit the exact time elapsed between the generation of a physical defect and its identification by engineering staff.
• Evaluate whether equipment maintenance histories are accessible to the personnel investigating production shortfalls.
• Integrate production tracking platforms with enterprise CMMS architecture to ensure cross-functional visibility.

By moving from reactive forensic investigations to a connected workflow, plants can intercept mechanical degradation much closer to the point of origin. Furthermore, leveraging unified production and equipment records allows facilities to transition away from rigid, calendar-based preventive maintenance schedules toward dynamic, usage-based maintenance protocols driven by actual machine cycles and operational stress.

Navigating the Balance Between Software Integration and Digital Complexity

While the benefits of an interconnected digital plant are profound, Dykas issues a critical word of caution regarding the implementation of universal digital architectures. Commissioning a monolithic, plant-wide software overhaul from scratch can quickly become prohibitively expensive, disrupt ongoing operations, and bog down internal IT resources.

Instead, Dykas recommends partnering with specialized software vendors and nurturing internal engineering capabilities to build targeted, modular applications tailored to specific manufacturing environments. The core challenge, as Dykas emphasizes, lies in integration:

“They can become very quick at creating the tools that can be very specific for certain types of manufacturing environments that will give you the data you need. The difficulty of that is: how do you integrate all those things together to understand what’s actually going on in the whole plant? How do you integrate the CMMS? How do you integrate all this other software that you have so you can get the whole story?”

Solving Persistent Scrap Through Strategic Process Redesign

Not all manufacturing variation can be engineered away simply by collecting more data or tightening software controls. In many instances, operations rely heavily on manual finishing techniques—such as an operator using an abrasive belt to smooth a component’s surface after primary machining. Because the final quality of the part depends directly on human craftsmanship and technique, results naturally vary from operator to operator and shift to shift.

In such scenarios, aggressive data collection merely exposes and quantifies the underlying variation without making the manual process inherently repeatable. Dykas offers a clear rule of thumb for industrial leaders: improve measurement and process controls where variability can be systematically managed, but actively investigate fundamental process redesign where variability remains permanently baked into a manual operation.

To illustrate the process-redesign path, Dykas points to advanced capital investments, such as upgrading to multi-axis milling machinery capable of producing complex geometries in a single setup, thereby eliminating subsequent manual finishing steps entirely. While such equipment investments require significant capital outlay—hypothetically running into millions of dollars—they fundamentally alter the production sequence, removing a notorious source of human-induced variation rather than merely monitoring its recurring failures.

Structuring the Capital Investment Decision

When evaluating whether to replace variable manual processes with advanced automated machinery, leadership teams should structure their financial and operational comparison around several core variables:

• Capital expenditure versus the recurring labor costs and material waste associated with manual operations.
• The potential reduction in total cycle time and its impact on overall facility throughput.
• The retraining and upskilling requirements for engineering and operational personnel.
• The long-term impact on product consistency, scrap reduction, and regulatory compliance.

Dykas advises dividing these complex investigatory responsibilities across specialized internal teams. Continuous-improvement specialists should remain focused on stabilizing and optimizing existing operational workflows. Sustainment engineering teams should concentrate on monitoring and refining stable, machine-based processes. Meanwhile, advanced-manufacturing engineers should collaborate closely with operations leadership and external equipment builders to explore alternative production methodologies that eliminate persistent scrap points altogether.

Protecting Engineering Time for Strategic Innovation

A frequent operational pitfall in modern manufacturing facilities is the over-allocation of engineering talent to daily firefighting. When skilled manufacturing engineers are consumed entirely by emergency downtime response and immediate floor-level troubleshooting, they have virtually no bandwidth to evaluate systemic process controls or collaborate with external technology vendors on innovative alternatives.

Dykas emphasizes that plant leadership must actively protect investigation time for their engineering personnel. By providing engineers with the autonomy and structured time needed to design long-term process strategies, facilities can develop robust modernization roadmaps while floor operators continue meeting immediate daily production quotas.

Ultimately, achieving a resilient, data-driven manufacturing operation requires a balanced strategy. Leaders must leverage digital monitoring and measurement feedback to stabilize processes that can be controlled, while maintaining the courage to invest in fundamental equipment redesigns to eliminate operations that generate persistent scrap. By working closely with experienced equipment builders and industry partners across sectors like medical devices and aerospace, modern manufacturers can successfully navigate the transition from manual variability to total digital control.