Q&A: Applying AI and Digital Twins to Improve Machine Design and Manufacturing Processes

Machine design engineers continue to face growing challenges—from supply chain disruptions to rising operational costs—that demand practical, technology-driven solutions. Drawing on more than 15 years of experience, Dominik Fry, a licensed professional engineer and leader in energy and industrial fluid systems, shares his insights on how integrating AI with engineering simulation tools can improve predictive maintenance, optimize manufacturing processes and increase system flexibility.

Machine Design reached out to Fry for some actionable strategies to help readers meet today's manufacturing demands with greater efficiency and resilience.

Editor’s Note: This interview may have been edited for style and clarity.

Machine Design: How can machine design engineers best leverage AI technologies to improve predictive maintenance and process optimization in manufacturing environments?

Dominik Fry: Machine design engineers can unlock the full potential of AI to enhance predictive maintenance and process optimization by adopting the following strategies:

MD: What are the most effective strategies for overcoming integration challenges of AI and new digital tools into existing manufacturing tech stacks, especially given legacy infrastructure?

DF: Overcoming the integration challenges of AI and new digital tools starts with simplification. With  87% of manufacturing leaders are actively streamlining their tech stacks, machine design engineers should consolidate platforms to reduce complexity and eliminate data silos to establish a strong foundation for AI deployment.

The key mindset is adopting a standard-first approach by normalizing data tags and units for seamless integration with existing systems like Computerized Maintenance Management System (CMMS) and Supervisory Control and Data Acquisition (SCADA). This approach helps facilitate easier adoption of new tools across the organization. Finally, a phased development strategy is equally important. Validating AI recommendations in simulation before live implementation will minimize operational risk and prevent potential errors, ensuring integration with legacy infrastructure enhances performance without disrupting ongoing operations.

MD: How does PIPE‑FLO software enhance the design and simulation of fluid systems compared to traditional methods, and how can it be integrated into the early design stages?

DF: PIPE‑FLO enhances fluid system design and simulation by providing engineering-grade accuracy far beyond traditional spreadsheets. It delivers precise hydraulic and thermal calculations with up to 17-decimal precision, enabling reliable lifecycle modeling that traditional ad‑hoc methods cannot match.

Additionally, pump selection is streamlined, as engineers can size and choose pumps based on actual system conditions using performance data from PUMP-FLO, one of the world’s largest manufacturer-verified pump catalogs. The interface of built-in visualization also allows engineers to visually map systems, observe flow and pressure behavior in real time, and identify potential design issues before they appear in the field.

The key function of PIPE-FLO is its ability to ensure design-to-digital-twin continuity from the earliest stages of the design process. For example, engineers can develop early-stage Flo-Sheet models, import CAD designs, and maintain a consistent model throughout the system’s lifecycle, creating a single source of truth.

Machine design engineers can also simulate “what-if” scenarios, such as pump logic, valve behavior or operating schedules, using the OverTime model and time-based thermal analysis. This allows teams to evaluate system performance under varying operating conditions before making design or procurement decisions.

MD: What role does real-time data visibility play in managing supply chain disruptions, and how should machine design teams account for these dynamics in system design?

DF: The role of real-time data visibility plays a huge role in managing supply chain disruption, enabling continuous scenario updates. Data can be used to rerun performance models whenever tariffs or speculations shift in real time. For example, PIPE‑FLO models updated with sensor tags help teams re-evaluate flows and pressures when suppliers change materials or tariffs affect availability, preventing last-minute redesigns. 

With 85% of manufacturers changing supply strategies, machine design teams should account for these dynamics by designing for operational agility, not just nominal conditions to account for these shifting dynamics. Build systems that tolerate component substitutions and fluctuating speculations, and validate them in digital twins before procurement. The goal is to ensure systems can accommodate variability without compromising reliability.

MD: What are some best practices for designing machine components with cost optimization in mind amid rising operational costs and supplier changes

DF: Cost optimization starts by treating energy as a first-order design variable. Teams can ensure long-term operational efficiency by evaluating system power costs early to optimize for full lifecycle energy expenses, not just upfront purchase price.

Designing for interchangeability is also critical. By validating multiple component options, engineers can maintain acceptable performance windows even when switching suppliers, protecting schedules and budgets from disruption.

Finally, it’s important to quantify risk premiums. This can be done by leveraging market insights and predicted changes to include sensitivity margins in TCO and NPV models. This will help guide teams to resilient choices rather than lowest‑bid procurement.

MD: How can machine design engineers contribute to simplifying tech stacks and data management protocols to reduce production timeline delays?

DF: Machine design engineers can play a key role by establishing a single system of record, using platforms like PIPE‑FLO as the authoritative source for “as-calculated,” “as-built” and “as-operated” data across design and operations. Consolidating tools is equally important to streamline collaboration and workflows while improving predictability and ensuring on-time delivery. As a result, data shows that tool consolidation is a prerequisite for on-time delivery.

Additionally, automating data handoffs ensure models remain current without manual intervention. By standardizing tag dictionaries and automating imports and exports, engineers eliminate schedule delays caused by data drift, keeping projects on track and aligned from design through operations.

MD: What training or skill development programs would you recommend to close the AI knowledge gap among machine design engineers and manufacturing teams?

DF: Closing the AI knowledge gap starts with role-based upskilling. Operators, designers and data engineers should pair domain expertise, such as hydraulics, thermals or mechanical systems, with data literacy, including feature engineering and anomaly detection. This ensures that AI is applied effectively on top of validated physics models, which then functions as a trusted digital twin.

With AI embedded into these digital twins, operators and engineers gain a complete, intelligent view of system behavior, enabling them to communicate more effectively.

Training should be simulation-first and highly hands-on. Programs like FLO‑Master Academy provide fast, accessible instruction and certification in PIPE‑FLO, building the modeling skills necessary to support AI-ready workflows. This helps provide teams with practical experience that translates directly into operational impact.

Overall, teams must foster environments with accessible learning opportunities, promoting engagement with resources that develop multidisciplinary expertise across mechanics, controls, and IIoT.

MD: How is the adoption of AI and automation impacting quality control processes specific to mechanical system design and manufacturing?

DF: AI and automation are transforming quality control from inspection to prediction, allowing early-stage detection of process drift through digital twin references to reduce scrap and rework for teams.

In mechanical system design and manufacturing, simulation-driven models increasingly define acceptance thresholds for commissioning and factory acceptance testing (FAT). AI systems continuously monitor production data and flag real-time deviations against these model-based standards.

Additionally, in the face of supply chain volatility, AI-assisted quality control can evaluate alternate parts and materials against a common baseline model. This ensures functional equivalence and performance compliance, allowing manufacturing lines to remain operational despite part substitutions.

MD: Given geopolitical‑ and tariff‑related uncertainties, how should machine design engineers approach material selection and supplier diversification to maintain flexibility and resilience?

DF: Machine design engineers should prioritize multi-source equivalency by developing models that evaluate similar materials and components across different regions. In an effort to maintain profit margins, 52% of global manufacturers have actively reduced their reliance on suppliers in high-tariff regions, signaling a major shift toward regionalization and supply chain diversification.

I also recommend regionalizing where warranted by lifecycle costs by using TCO models that account not just for unit price, but tariff volatility, compliance costs and lead‑time risk.

To further enhance these strategies, machine design engineers should maintain an up-to-date digital twin to continuously validate performance with alternative materials or vendors as market and supply conditions change. This effectively serves as an engineering “shock absorber,” safeguarding both operational performance and supply chain resilience.