By connectivity, we mean the "translation" of your machine data into an open protocol, such as OPC-UA, so that they can be "understood" by third-party systems. Newer plants are often equipped with OPC-UA from the factory and do not require additional connectivity. However, in reality, your manufacturing consists of various, partly older plants that do not use open protocols. For the MDE solution not to only capture new plants, the data from the older plants must be "translated" into an open protocol.

Once connectivity is established, we can read data from the control unit. This presents a challenge that many are not aware of: data preparation. Every plant generates thousands of data points, of which only a part is needed. Additionally, one must find the suitable data points in the data chaos. Furthermore, every plant names the same data points differently and uses different units. To compare values between plants and make them usable for the entire site, the data must be standardized and made comprehensible to humans. Often there is also a desire to link machine data with product data from the MES or ERP.

The prepared data should not only be used in the moment but also compared over longer periods (e.g. "What was my OEE three months ago?") and stored (e.g. for traceability). Therefore, we must store the data and make it centrally accessible.

Lastly, we come to the obvious part: the data usage for optimizing manufacturing, whether for calculating OEE or for production monitoring with shop dashboards. It is also important to have an open solution so that the data can not only be used by a closed system but can also be exported to third-party systems (e.g. PowerBI, Excel, Minitab). This gives you a solution that will still meet the requirements in ten years.

Can’t it be done more easily? Yes, with ENLYZE.

You realize: machine data acquisition is a larger project than you might have thought. For the implementation, you previously had to piece together various providers and wait years for a solution to be in place. 

ENLYZE implements this process from start to finish within 2 weeks, so you can focus on working with the data. Because your manufacturing does not need a large IT project, but added value from the data. Learn more about it here.

What MDE software providers are there and what can they do?

Once it is clear what is necessary for machine data acquisition, we can now categorize providers. This way, you can understand which part of MDE these providers cover and where you still have gaps.

We will take a look at five solutions that are very popular in the German-speaking area:

1. ENLYZE: With ENLYZE, you start working with your machine data from day 1. We implement the components of connectivity, data preparation, and storage within two weeks. You also receive turn-key applications so that you can immediately optimize your production with the data, e.g. for production monitoring, OEE management, or process optimization.

Our motto: Work WITH data, not FOR it. Here you can learn more about ENLYZE.

2. Kepware: Kepware is excellent for solving step 1 (connectivity). This way, your plants speak the same language (OPC-UA) and you can further process the data in steps 2-4 - however, you will need additional providers, which means more time and additional costs.

3. Cybus: Cybus offers solutions for steps 1-2, with their greatest strength being in data modeling. Cybus is an ideal solution for large companies with strong IT departments looking to digitize similar assets. Example: Porsche AG wants to digitize ten identical manufacturing lines. 

3. MES providers: Many MES providers claim they offer machine data acquisition. What exactly they mean by that is usually unclear. From our experience, connectivity must either be solved by the customer (i.e. by you) or through third-party providers, known as integrators. This means that most MES providers only start once all machine data is made available via OPC UA. These integrators then use solutions like Kepware to provide the relevant machine data via OPC UA. The problem: there is no consideration for the long-term usability of the data. If you want to use the same data in other systems, it quickly becomes clear that this is only possible with great effort.

4. OPC Router: OPC Router is a partial solution between steps 1 and 2. With this tool, you can connect two data sources in a graphical interface, e.g. mapping a data point on the OPC-UA server to your database. However, connectivity must already be resolved. Data preparation, storage, and usage cannot be resolved by this.

What hardware do I need for machine data acquisition?

Ideally, the software provider offers the hardware as a package because this significantly reduces integration times and all systems can communicate easily with each other. When hardware and software are purchased separately, it often leads to a provider puzzle that causes headaches and is difficult to roll out to other areas.

With our customers we only install our Edge Device, which securely transmits the data from the plants to ENLYZE. We provide the remaining hardware for you as part of our solution.

In general, the following additional hardware components are needed for machine data acquisition:

• Sensors: If the plant itself cannot measure certain parameters (e.g. the temperature of the hall), additional sensors are installed and used for monitoring and optimizing production.

• Data loggers, IoT gateways, or Edge Devices: These devices collect data from plants and sensors. They can also perform preliminary data processing to reduce the amount of data sent to centralized systems.

• Servers and storage solutions: To store and further analyze the data, you need server hardware. This can be on-premise or cloud-based.

Is there also freeware for machine data acquisition?

Due to the increasing popularity of open-source projects in automation, many free tools are now available, enabling IT-savvy individuals to capture machine data.

Here you should keep in mind that high internal costs arise for operation and integration, as you need qualified personnel with the know-how to operate these products. 

Below we briefly introduce some of our favorite projects that could be used in the development of your own machine data acquisition system. These are technological building blocks that need to be assembled, operated, and maintained. Work arises "in the gaps" between these blocks. 

Connectivity:

• PLC4x

• OPC UA clients in all conceivable languages

• Modbus clients in all conceivable languages

• Step7 for Siemens S7-200/300/400

• Beckhoff ADS in Python

Data collection:

• Your own scripts via Python or Bash

• NodeRED

Data broker - depending on setup

• Mosquitto (MQTT)

• RabbitMQ 

• Kafka

Data storage

• Postgres and TimescaleDB 

• InfluxDB 

• CockroachDB

United Manufacturing Hub (UMH):

If you are looking for a comprehensive open-source IIoT (Industrial Internet of Things) platform, United Manufacturing Hub (UMH) is a good place to start. 

UMH enables experts to integrate all data sources into a so-called Unified Name Space (UNS). By unifying the data, any application cases can be built on top of it - from real-time dashboards to advanced analytics and AI.

Data acquisition here, for example, occurs via NodeRED or OPC-UA, with storage in Timescale DB. As an open platform, the data can be redistributed to any number of systems via MQTT or visualized in Grafana.

However, we often advise medium-sized companies against building an in-house solution. Here, there is often the false assumption that the costs are significantly lower since the software "costs nothing" at first glance due to being open source. In our experience, however, at least a team of 4 to 5 experienced people is necessary for the setup and operation. Consequently, the ongoing costs can quickly add up to €30,000 to €40,000 per month. This fact leads to projects not often breaking out of a PoC stage, as the necessary internal resources are not provided. Because the setup of an IIoT platform cannot just be done on the side by an overworked IT department.