The system collects PLC data from casting machines into daily files and sends email alerts for a metal casting factory in Thailand

How the brief came about

The brief for this case was to store the operating values of metal casting machines so they could be reviewed when a machine had a problem. The casting machines' PLCs (the controllers that run the machines) already held set values and measured values, but the factory's team could not extract the operating log (Data Log) from the machines to store it.

Without historical data, the team had to look for causes through observation at the machine, which is difficult when a problem occurs only occasionally. Data recorded at set intervals makes it possible to look back and see which values departed from normal during the period when the problem occurred. The customer therefore wanted the PLC values recorded continuously.

Requirements and constraints

The system had to store PLC values from the 2 casting machines on a server in the factory and give an alert when data transmission failed. The main requirements were as follows:

• Read 52 values per machine, covering both set values and measured values
• Install a server running nimitFlow at the customer's factory
• Send machine data over the factory Wi-Fi
• Record data as files that can be opened later when investigating a cause
• Give an alert when a gateway goes down or cannot be reached

The constraint found during the project was the network connection around the machines. We resolved this by changing where the network equipment was installed.

Why we read values directly from the PLC instead of pulling the Data Log from the casting machine

We read values directly from the PLC because the factory's team could not extract the operating log from the casting machines, but the values needed were already in the PLC. The Industrial IoT Gateway therefore reads values from data addresses in the PLC one by one, according to the list of 52 values defined in the system analysis document.

The system records data as CSV files (tabular data files), in line with the requirement to open historical data when a problem occurs. The scope of this work does not include a real-time display. CSV files can be opened in a spreadsheet program, and keeping 1 file per day for each machine makes it straightforward to find the file for the period when a problem occurred.

1. 1

   Gateways read values from the PLC

   Each machine has 1 gateway that reads values from the PLC, both set values and measured values, such as injection speed (Shot Velocity), hold-phase pressure after injection (Injection Hold Pressure), mould temperature (Die Temp) and time per casting cycle (Cycle Time).

2. 2

   Data sent over Wi-Fi to the server

   The gateway sends data over the factory Wi-Fi to the server with nimitFlow installed in the factory, so the data stays within the factory's own network.

3. 3

   Recording to one CSV file per day

   The first Workflow contacts the machine at a set interval, gathers the data and writes it to a CSV file, 1 file per day, in separate folders for each machine. When a full day's data has been collected, the system sends an email notification.

4. 4

   Monitoring connections and sending alerts

   The second Workflow checks the connection to each machine's gateway. The third checks that nimitFlow is still running. If contact fails, the system emails the administrator.

Deliverables

We delivered hardware, software and documentation for the 2 casting machines as follows:

• 2 industrial IoT gateways, 1 connected to the PLC of each casting machine
• A server with nimitFlow installed in the factory
• 3 Workflows: daily CSV file recording, gateway connection checks and checks that nimitFlow is still running
• Remote server administration through a VPN (an encrypted network connection channel) and Remote Desktop (controlling a computer remotely)
• User manual explaining gateway configuration, how to change the file storage location, email recipients and reading interval, and how to get a Workflow running again when it stops

Confirmed results from testing and work acceptance

The casting machine data collection system installed by Inspektion passed testing of the 52 PLC values per machine on each of the 2 machines. The Integration Test also covered CSV file recording and all 3 sets of alerts, and it passed on every item.

The customer's representative signed the acceptance document (Exit Criteria) in August 2025. This case has no measured business figures, so we report only results backed by documents. What was delivered is casting machine operating data recorded as daily files, available to review when investigating the cause of a machine problem.

What we learned

This project offers 3 principles for collecting data from machines.

• A data collection system must also monitor itself. If the system stops without notice, the team finds out only when they need the data and discover it is missing. That is why this project separated the Workflow that checks the gateway and the Workflow that checks nimitFlow from the Workflow that records data.
• The network must be checked at the actual machine location. In this project, connection problems arose during the work and were resolved by changing where the network equipment was installed. A signal survey at the machine should therefore be carried out before installation.
• The list of values to collect is both the work specification and the acceptance criteria. The list of 52 values per machine was used to configure the gateways and to carry out acceptance item by item. This list should be agreed before work starts, so that both parties accept the work against the same criteria.

How this approach can apply to your factory

If your machines are controlled by PLCs but you still cannot extract their operating data to store it, you can use the same approach as this case: read values from the PLC through a gateway, record them as daily files on a server in your factory, and give an alert when the connection fails. The parts that need adapting are the list of values, the connection to the PLC and the on-site network.

Besides the items in the table, you should prepare 2 things before work starts.

• Define the server location and file storage location from the start
• Plan the procedure for restarting the system after a power cut or a server shutdown