Johnson & Johnson · Pharmaceuticals & biotech
AI process control and digital twins to cut energy use at J&J Xi'an
23% reduction in energy consumption
Use case · Operations & production
Measure and model energy use by line and utility system, then cut waste through better scheduling, setpoints and load management.
Energy optimization applies analytics and machine learning to electricity, gas, steam, compressed air, chilled water and other utilities. It starts by metering consumption at the level of lines, machines and utility systems, then models how energy use relates to production, weather and equipment state. With that baseline, the system flags waste, recommends better setpoints and schedules, and helps shift flexible loads away from expensive tariff periods.
It matters because energy is often one of the largest controllable costs in a plant, yet it is usually tracked from a single monthly bill. Compressed air leaks, idle equipment left running, chillers and boilers operated inefficiently, and peak demand charges set by a single bad interval are common and largely invisible without granular data. Energy data also feeds carbon reporting that customers and regulators increasingly ask for.
Add submeters on the largest loads and utility systems, and pull existing data from building management systems, PLCs and utility interval meters. The goal is energy use per line, per system and per shift rather than per site.
Models relate consumption to output, product mix, ambient temperature and run state, so the team can see energy per unit and tell real waste from normal variation.
Anomaly detection highlights loads running during idle periods, rising baseline consumption from leaks or fouling, and equipment drifting from its efficient operating point.
Optimizers sequence compressors, chillers, boilers and pumps to meet demand at the lowest energy cost, and adjust setpoints such as header pressure and supply temperatures within operating limits.
Forecasts of plant load and tariff signals guide when to run flexible processes, pre-cool, or stagger equipment starts to avoid setting new peak demand.
Line-level metering shows equipment left running on breaks and weekends and leaks that raise base load, which a monthly utility bill hides entirely.
Forecasting and staggering large loads avoids short consumption spikes that can set the demand charge for an entire billing period.
Running the right combination of compressors or chillers at their efficient points, instead of everything at partial load, reduces energy for the same service.
Measured energy by line and product supports carbon footprints and customer sustainability questionnaires with data rather than estimates.
Each example is labeled by who reported the result. How we grade evidence.
Johnson & Johnson · Pharmaceuticals & biotech
23% reduction in energy consumption
National Academies of Sciences, Engineering, and Medicine · Report · 2024
A National Academies consensus study report, supported through DOE's Advanced Materials and Manufacturing Technologies Office, on options for a national plan for smart manufacturing technology development and deployment. It examines promising technologies, the research and resources needed to speed adoption industry-wide, and education and workforce needs.
Common sources are compressed air leaks and over-pressurized systems, equipment left running during idle periods, utility plants operated at inefficient partial loads, and peak demand spikes from many large motors starting together. Most of these are invisible from a site-level utility bill and only show up with line or system-level metering.
AI models relate energy use to production, weather and equipment state, which lets them flag consumption that is higher than it should be for current conditions. Optimization algorithms then sequence utility equipment and schedule flexible loads to meet demand at the lowest cost. The savings come from operational changes, not new equipment, which keeps the investment modest.
Energy normalization adjusts consumption for factors such as production volume, product mix and outdoor temperature. Without it, a busy month looks worse than a slow one even if the plant became more efficient. Normalized metrics like energy per unit are what make savings measurable and comparable.
Safety, health & sustainability
Collect energy, material and supplier data to calculate emissions by site and product and answer customer and regulatory requests.
Operations & production
Models recommend or automatically adjust process setpoints to keep quality on target while improving yield, throughput and energy use.
Operations & production
Automatically capture machine states, downtime reasons and output to measure OEE accurately and target the biggest losses.
Operations & production
Models learn normal process behavior across many sensors and flag unusual patterns before they become quality, safety or equipment problems.
Put AI to work
Genvo, our implementation partner, works with manufacturers to pick the right first project, connect the systems you already run and put AI agents and tools into daily use, with people approving what goes out.
Assessments and consultations are provided by Genvo. How our partnership works.