The Problem: Machine Readings Needed Product-Specific Checks
Machines record thousands of readings every shift, but those readings sat in historians that nobody watched in real time. Problems were often found only after a stop or a quality defect.
Nobody watching the data
Temperature, pressure and vibration readings were stored but not checked as they happened.
Limits change with the product
A setting that is right for one product can be wrong for another, so one fixed limit per machine wasn't enough.
False alarms at start-up
Readings taken while a machine is starting up look out of range and create noise.
No clear next step
When a reading was wrong, there was no standard list of what to check or fix first.
Several historian types
Different lines stored their data in different historians, so one tool had to read them all.
The Solution: Machine Alarms Linked to Product Limits
ACM turns stored machine readings into alarms the team can act on. Each reading is compared with the limits for the product running at that moment, and every alarm comes with suggested actions.
Plant overview in red and green
Every machine shows green when it's fine and red with an alarm count when it isn't; one click shows the details.
Limits per product
Each reading has its own low and high limits, warning levels and target value for every product.
Knows what's running
A running signal and a product signal tell ACM if the machine is on and which product it is making.
Start-up delay
A wait time after start-up stops false alarms while the machine settles.
Guided fix actions
Each high or low alarm is linked to up to four actions, such as replace filter or restart machine.
Acknowledge and complete
Teams acknowledge an alarm, tick the actions taken, add notes and mark it completed; repeats of the same alarm are grouped.
Trends with limit lines
See a reading's history with its limits drawn on the chart, filtered by product, for up to four readings at once.
Daily alarm email
Users subscribe in Preferences to a summary of the previous day's alarms, sent every morning at 08:00.
Bulk setup from spreadsheets
Tags, machines, readings, limits and actions can be uploaded from CSV templates that are checked before saving.
How It Works: From Historian Readings to Alarm Actions
1
Connect the historians
One or more Wonderware, Proficy, Perf or InfluxDB historians are added in the admin area.
2
Describe the plant
Departments, lines, machines and machine parts are set up, by hand or by spreadsheet upload.
3
Set the limits
For each reading and product, the team enters the low and high limits, warning levels and target.
4
Link the actions
Each high or low alarm is linked to the actions that usually fix it.
5
Monitor automatically
ACM reads the data on a schedule, checks it against the limits and raises alarms.
6
Act and close
The team sees the red machine, opens the alarm, takes the actions, adds a note and marks it complete.
ThingWorx holds the screens, settings and cached data; each source system stays the official record for its own data.
1. ACM screens
Summary Viewer, Alarm Actions, Equipment Analytics, Parameter View, Alarms by Equipment, Preferences
2. Admin suite
Set up historians, tags, readings, products, plant structure, limits and actions
3. Alarm logic
Compares readings with product limits, skips start-up time and groups repeat alarms
4. Database and aggregation
SQL Server tables and scheduled jobs store and summarise alarm results
5. Plant historians
Wonderware, Proficy, Perf and InfluxDB
ThingWorx services built
Service
Purpose
Historian connectors
Read tag data from several historians, even several of the same type
Active spec check
Compares each reading with the limits for the product being run
Action trees
Link each high or low alarm to up to four fix actions
Aggregator
Summarises alarm results on a schedule for fast screens
Mail service
Sends the daily 08:00 alarm summary to subscribed users
Mass upload
Validates CSV files and loads plant setup in bulk
Design decisions
Limits are stored per reading and per product, so a product change never triggers wrong alarms
Every name in the setup must be unique, which keeps alarms and reports clear
Uploads are checked first: duplicates fail, and tags must exist in the historian
Chart data can be compressed, trading detail for speed on long time ranges
Separate installation and operation checklists confirm each site is set up correctly
Local-language names sit next to English names for every item
Problems seen early
Fewer false alarms
Clear next steps
A record of every fix
Morning summary
Fast rollout
This solution fits if you need to…
Set up automatic condition monitoring in ThingWorx
Raise machine alarms from Proficy, Wonderware or InfluxDB historian data
Use different alarm limits for each product on the same machine
Give operators guided actions when a machine alarm fires
Send a daily machine alarm summary email
Automatic condition monitoring uses software to check machine readings such as temperature, pressure or vibration against allowed ranges and raise alarms when limits are crossed. In the ACM app that PSPTechHub worked on, historian data is read on a schedule and compared with the limits for the product running. This is threshold-based monitoring, not a documented AI failure-prediction system.
Yes. ACM reads tags from Wonderware, Proficy, Perf and InfluxDB historians, and can connect to several historians at the same time. Historian connections are configured in the admin area, and uploaded tags must exist in the selected historian before setup is saved.
Each reading has its own low and high limits, warning levels and target for every product. A product signal identifies which product is running, and a running signal identifies whether the machine is on. ACM uses that context to check the reading against the applicable product limits rather than one fixed limit for every product.
Each reading has a configurable delay time, so ACM waits after the machine starts before checking its values. This lets the machine settle before limits are evaluated. Product-specific limits provide additional context; the case study does not quantify the reduction in false alarms.
The machine turns red on the overview, and the team opens the alarm to see up to four suggested actions. Operators acknowledge it, tick the actions taken, add notes and mark it complete. Repeated occurrences of the same alarm are grouped, and acknowledgements, actions and notes remain associated with the alarm.
Yes. Users subscribe in Preferences and receive a summary of the previous day's alarms every morning at 08:00 local time. This documented email feature is a daily summary, alongside on-screen machine alarms, rather than a claim that every alarm generates an immediate email.
Tags, machines, readings, limits and actions can be uploaded from CSV templates. Files are checked for duplicates and missing historian tags before saving. Installation and operation checklists support site setup, while the admin area also allows plant structure and monitoring configuration to be entered manually.



