Fluke Reliability specialist Nicola Giannini on GMP HVAC for life sciences, done right

We’re proud to feature this interview with Nicola Giannini, Sales and Customer Care Manager at Fluke Reliability. Nicola helps teams shift from route-based checks to continuous, mixed-signal monitoring — allowing earlier action on high-impact HVAC, utilities, and process equipment, especially in pharmaceutical manufacturing environments.

Nicola Giannini did not set out to work in pharma. Trained as a nuclear engineer at the University of Pisa in Italy, he began his career at ENEA (Italian National Agency for New Technologies, Energy and Sustainable Economic Development), the country’s national research center, studying safety and vibration phenomena for nuclear power. Learning how to treat small signals as precursors to big consequences translated naturally to the controlled environments of life sciences manufacturers.

A family nudge pulled him from research to the plant floor. Nicola took a job at a maintenance firm, helping major life sciences manufacturers — GSK, Pfizer, Abbott (now AbbVie), Catalent, Haleon, Novartis, Roche —improve equipment reliability and uptime. “My father worked in heavy industry and pointed me to a maintenance company that specialized in predictive maintenance services,” he says. “In a small company they need you immediately so it’s a rapid-fire learning curve: you see an instrument once, you quickly learn it, and off you go! That quick pace pushed me deep into vibration, thermography, ultrasound, oil analysis, and NDT, increasing my knowledge at a great speed.”

He often worked on HVAC and site utilities, the systems that quietly govern product integrity. Across pharma the critical machines are often the same: cleanroom HVAC (air-handling units and belt-driven fans) and utilities like chillers and compressors. “On those we ran vibration analysis, thermography, ultrasound, and oil checks, especially in sterile vaccine areas where conditions must stay in spec.”

Building a Program, Not Just Fixing Assets

The real test came when diagnostics had to scale from a single asset to a whole site. Nicola spent a year embedded at a large pharmaceutical plant specializing in the production of meningitis vaccines that was responsible for producing meningococcal (meningitis) vaccines. Here, he worked with the reliability lead to put predictive maintenance into practice across bulk, primary, secondary, and packaging functions.

“I helped implement the philosophy of predictive maintenance across the newly formed reliability team,” Nicola explains. “We did root cause analysis and FMEA. We built a criticality ranking. We read the preventive maintenance plans together with the emergency work orders and looked for patterns.”

Evidence came first, not assumptions. “If a failure kept happening — like a belt every three months — we asked what predictive task could help. And when a bearing was replaced after a vibration call, we brought it back, cut it, and checked the actual failure to confirm the diagnosis.” Those case files turned into business justification. “We estimated that an unplanned stop can run to tens of thousands of euros once you add four hours for the change and ten hours for re-clean/sterilize,” Nicola says. “If you stop early, it’s roughly three hours instead.”

“If you can identify an anomaly early through the right predictive approach,” he continues, “you can organize the work — from the team to spare parts and logistics — and schedule the shutdown at a convenient time for production needs, significantly reducing the impact of the activity.” These approaches ensure maximum optimization in an environment like the pharmaceutical industry, where the preventive maintenance philosophy is strong. Instead of performing preventive changes, teams can shift to condition-based interventions guided by predictive analysis, significantly optimizing maintenance costs.

The Belt That Kept Running — and Why Vibration Monitoring Wasn’t Enough

One case made the value of this approach clear: a belt-driven cleanroom fan that looked “fine” on vibration — until it wasn’t. “There were three belts. One broke; the other two kept running. The motor still ran, the fan still ran — vibration increased a little but not much, so there was no big alarm”, he explains. “The vibration alarm threshold was set high and didn’t trip. The motor tries to hit the target with two belts, so you see the current absorbed increase. Even though the current absorption information was present, it wasn’t used in a predictive setting”

The fix was surprisingly simple: combine motor current and speed data from the PLC with vibration trends and alarms. Once they did, belt-related stoppages disappeared.

And the stakes were real for sterile operations. “If an HVAC unit stops in a vaccine area, you have to recheck what you’ve stored,” Nicola says. “A stock of vaccine could be between eight hundred thousand and one million euro. That’s not just downtime; that’s product risk.”

A Stepwise Blueprint that Fits Life Sciences

Nicola’s adoption path fits pharma budgeting and change control: learn offline, prove value, then move high-risk assets online using the same hardware.

“Many times for budget reasons, we start with an offline approach — sensors and route measurements every two or three months—and show results. We document maintenance hours avoided, re-clean and sterilize time, QA steps. The next year they can move the highest-risk units online. The sensors are the same, so you don’t waste money starting offline. Online is more powerful: you can get an email at night if there’s a problem, and with current and speed together with vibration you get earlier, cleaner alarms.”

This also tackles a common pain point: data silos. Bringing PLC/BMS tags and condition data into a single view speeds decisions and makes alerts more actionable. With that foundation in place, Nicola extends the same playbook to the utilities that drive uptime and the integrity of materials used to make products.

For instance, chillers and compressors are frequent sources of hidden loss, so they’re early candidates for mixed-signal monitoring alongside air handling units. And on product-contact equipment, the focus shifts from uptime to quality risk. “It’s important there is no rouging, no corrosion, no small cracks. We did specific jobs with partners to analyze those problems, because they can become a problem for the final customer.”

With the blueprint in hand, the work shifts from proof to practice — choosing first-mover assets, unifying PLC/BMS tags with condition data, and tuning alerts to how teams actually respond.

Current Deployments and Early Wins

Across life sciences sites, Nicola partners with maintenance, QA, and validation departments to move high-impact assets from routes to continuous, mixed-signal monitoring. The first candidates are picked by criticality — sterile-area air-handling units, chilled-water primary pumps, and compressed air. And today, thanks to the powerful capabilities of products such as Azima, with its artificial intelligence, and VibGuard, which has been a benchmark in online monitoring systems for years, small divergences surface before they snowball. Teams can then track maintenance recommendations as they flow directly into eMaint work order software. This approach is already shaping ongoing projects at various sites. “We’re implementing systems on HVAC units at sites. That’s the topic of discussion everywhere — pick the right assets, correlate the right signals, make the economics visible, then scale.”

What Teams Are Seeing

• Fewer belt-drive surprises from vibration + current/speed correlation, and in general, the right combination of information based on asset monitoring
• Faster calls from a single view of PLC/BMS tags and condition data
• A reuse-the-sensors path from routes to online (no stranded hardware)
• Cleaner business cases tied to a reduced need for re-cleaning, sterilizing, and other QA steps

Ready to see how our specialists can set you up for real results? Contact Nicola at [email protected] or +1390699313071.