Industrial IoT (IIoT)
Predictive Maintenance Becomes More Important Across Indiana Factories
Published by IOT Indiana Research & Editorial Team

- Five pressures raising the stakes
- 1. The maintenance labor squeeze
- 2. Older equipment, newer expectations
- 3. Customers who penalize late shipments
- 4. Cost pressure from materials and trade conditions
- 5. Energy and sustainability reporting
- What the unplanned downtime really costs
- Is your plant ready? A quick self-assessment
- The maturity ladder
- Measures that prove it is working
- A first project in 120 days
- Days 1 to 20: Choose and baseline
- Days 20 to 50: Instrument and connect
- Days 50 to 90: Learn normal
- Days 90 to 120: Act and report
- Illustrative example: a foundry fan
- Where programs go wrong
- Support in Indiana
- The bottom line
Predictive maintenance has been a talking point in manufacturing for more than a decade. What has changed in Indiana is the context around it. Plants are competing for technicians with some of the largest construction and manufacturing projects the state has seen, customers expect shorter and more reliable delivery windows, and a lot of installed equipment is old enough that its failures are no longer rare surprises. Those pressures turn predictive maintenance from a nice experiment into an operational necessity for many factories.
This article takes a pressure-first view. Instead of repeating how the sensors work, it asks what is pushing Indiana plants toward condition-based maintenance now, how to tell whether your plant is ready, which measures prove progress, and how to run a first project that your maintenance team will actually support.
Five pressures raising the stakes
1. The maintenance labor squeeze
Large investments in the state, including SK hynix's planned packaging plant in West Lafayette, Eli Lilly's building program in Lebanon and Meta's data center campus there as well, will all need electricians, instrument technicians, millwrights and controls specialists. Each of those projects has been tied in public reporting to a large construction workforce and to hundreds or thousands of permanent roles. The skilled trades people who keep a stamping plant or a foundry running are the same people those employers want. Existing plants should plan for higher wages and more turnover in skilled trades.
Predictive maintenance helps because it changes the work from reacting to emergencies at any hour to scheduling repairs in daylight. A smaller team can cover more equipment when most interventions are planned.
2. Older equipment, newer expectations
Many Indiana plants run machines that have been in service for decades. These assets still make money, but their failure patterns are wearing out in ways that do not announce themselves. Bearings, gearboxes, hydraulic components and motors tend to degrade gradually and then fail suddenly, which is exactly the pattern condition monitoring is designed to catch.
3. Customers who penalize late shipments
Automotive, appliance and aerospace customers can charge back for late deliveries or move volume to competitors. An unplanned breakdown on a bottleneck machine therefore costs more than repair parts and overtime. It can cost the next purchase order.
4. Cost pressure from materials and trade conditions
Tariff and trade changes have made some inputs more expensive and less predictable. When margins are thin, the waste from scrap, expedited freight and idle labor during a breakdown matters more. Preventing failures is one of the few cost levers that does not depend on bargaining with suppliers.
5. Energy and sustainability reporting
Failing equipment often uses more energy before it breaks. A motor with worn bearings draws more current, and a leaking compressed air system wastes power continuously. As customers ask for emissions data, monitoring that doubles as an energy tool gains value.
What the unplanned downtime really costs
Many plants underestimate downtime because they count only parts and repair labor. A fuller accounting includes more lines.
| Cost line | How to estimate it |
|---|---|
| Lost contribution margin | Hourly contribution margin of the stopped machine times hours stopped, adjusted for how much can be made up later |
| Repair labor and parts | Technician hours at fully loaded rates plus parts and emergency freight |
| Idle labor | Operators and support staff waiting, at loaded wage rates |
| Scrap and restart losses | Material lost during shutdown and restart, plus time to stabilize the process |
| Expedited shipping and overtime | Extra costs to meet commitments after the stop |
| Customer penalties and lost orders | Contractual chargebacks, plus the harder-to-measure risk to future business |
Run this calculation for your last three serious breakdowns. Most plant managers are surprised by the total, and the number becomes the business case for a pilot.
Is your plant ready? A quick self-assessment
Predictive programs fail more often from organizational gaps than from technology. Score your plant on each question: yes, partly or no.
- Do you know which five machines cost the most when they stop?
- Does your maintenance system record failures with a cause and a duration, not just "repaired"?
- Does someone own preventive maintenance schedules and review them at least quarterly?
- Do operators have an easy way to report abnormal sounds, vibration or heat?
- Can you get spare parts for critical equipment within a day?
- Is there a person who can interpret trend data, or a vendor contract that covers it?
If most answers are "no," start with basic reliability practices first: a clean failure history, a critical asset list and disciplined preventive tasks. Sensors added to chaos produce alarms nobody trusts.
The maturity ladder
Predictive maintenance is the top of a ladder, and plants gain a lot from each rung.
- Run to failure. Appropriate only for cheap, non-critical equipment with spares on the shelf.
- Calendar-based preventive. Tasks happen at fixed intervals. Simple, but wasteful when equipment is serviced too soon, and unreliable when it fails between visits.
- Route-based condition checks. Technicians take vibration, temperature or thermal readings with handheld tools on a schedule. Low cost and a good training ground.
- Continuous monitoring on critical assets. Fixed sensors send data to a gateway, with alerts when readings cross limits.
- Predictive analytics. Models estimate remaining useful life or classify fault types, and work orders are generated automatically.
Most Indiana plants should aim for rung three across many assets and rung four on the few that matter most. Rung five pays off when there is enough history of failures to learn from, which is often not the case for a first-year program.
Measures that prove it is working
- Unplanned downtime hours on monitored assets, compared with the prior year.
- Planned to unplanned work ratio. A healthy shift moves toward planned work, a common target being a large majority of tasks.
- Mean time between failures (MTBF) for monitored assets, which should rise.
- Mean time to repair (MTTR), which should fall as faults are caught earlier and parts are staged.
- Emergency purchase orders and expedited freight on maintenance parts.
- Alert accuracy. The share of alerts that led to a real finding, tracked honestly. If it is low, thresholds need work.
A first project in 120 days
Days 1 to 20: Choose and baseline
Pick three to five machines with high downtime cost and a failure history. Pull two years of records and count failures and hours. Without a baseline, you cannot prove improvement.
Days 20 to 50: Instrument and connect
Fit vibration and temperature sensors on rotating equipment, current sensors on motors and pressure or flow sensors on hydraulic or pneumatic systems as appropriate. Send data through a local gateway to a historian or cloud service. Name every signal consistently from the start.
Days 50 to 90: Learn normal
Let the system gather data across product mixes, shifts and seasons. Set conservative alert limits. Review every alert with a technician, and record whether it was a true finding, a false alarm or an unclear case.
Days 90 to 120: Act and report
Convert useful alerts into work orders. Report results monthly to production and finance, including avoided downtime estimates with the reasoning shown. Decide whether to extend, adjust or stop.
Illustrative example: a foundry fan
The figures here are invented to show the method. A foundry has one large dust collection fan whose failure forces the casting line to stop within an hour for air quality reasons. A bearing failure happens roughly every 18 months, costing about 14 hours of downtime at $4,000 per hour in lost margin and idle labor, plus $9,000 in repairs: about $65,000 per event.
The plant mounts two vibration sensors and a temperature probe for about $4,500 installed. Three weeks before the next expected failure, vibration at the bearing frequency rises steadily. The team schedules a swap during a Saturday maintenance window, costing $3,000 in parts and four hours of labor.
The avoided loss is roughly $60,000, though the plant should log it as an estimate, since no one can be certain the bearing would have failed on schedule. Several such catches in a year, across a handful of critical assets, make the program's value visible in plain dollars.
Where programs go wrong
- Too many sensors too soon. Dozens of alerts per day teach people to ignore the system.
- No owner. Someone has to review alerts, adjust limits and push work orders through. Assign a person, not a committee.
- Weak connection to the maintenance system. If findings stay in a dashboard, they do not become repairs.
- Ignoring operator knowledge. Experienced operators often sense problems long before sensors do. Their observations are a data source and a source of credibility.
- Treating it as an IT project. It is a maintenance and reliability program that uses IT.
Support in Indiana
Purdue Manufacturing Extension Partnership advisers work with Indiana manufacturers on reliability and technology adoption. Conexus Indiana and the Next Level Manufacturing Institute offer training and networking. Utility programs can help with energy-related monitoring. State grant programs change over time, and the Manufacturing Readiness Grants were paused in 2025, so confirm what is currently available before relying on it.
The bottom line
Predictive maintenance matters more across Indiana because the cost of surprises has gone up while the supply of people who fix them has gone down. The best response is modest and disciplined: know your critical assets, build a clean failure history, monitor a few machines well and prove the value in dollars. Plants that do that will find that their technicians are better equipped, their customers see fewer surprises and their next investment decision is easier to justify.
Frequently asked questions
Related articles
Industrial IoT (IIoT)
Industrial IoT (IIoT)
Cummins and the Connected Factory: What Industrial IoT Means for Indiana Manufacturing
Industrial IoT (IIoT)
Indiana's Pharmaceutical Manufacturing Boom Creates New IIoT Opportunities
Industrial IoT (IIoT)
