When considering starting an industrial monitoring and data analytics project, it is common for industrial companies to have one of the following problems (or a mixture of them):

• We do not collect any data, or we still collect the data by hand
  
• We collect a lot of data and don’t get value out of it or don’t use it
  
• We have data, but it is stored in different platforms or disconnected applications that function as information silos

Does this sound familiar?

The solution to these problems is to collect, centralize and properly monitor data in order to convert it into valuable information with which to make decisions. But… What is the best way to do this?

We tell you the keys to a successful industrial monitoring project.

1. Capturing Data

The first step is to collect the data, considering that it can be found in very different sources, either hardware (sensors, PLC, automatons…) or software (ERP, MES, Excel, CMMS, Databases, CSV…)

At this point it is key to take advantage of the data you already have. When we think about capturing data, we always think that we have to invest in PLCs, sensors and so on, but the truth is that it is not always necessary.
You can start with something basic, since your machines and systems (that MES, that ERP, those Excel, the operator’s notebooks…) usually have data. Besides, once you start working with some data, their aggregation generates new potentially useful and interesting data.

Our advice here is to start by building on what you already have and grow, which makes it easier to get started and reduces investment. In addition, the 80-20 rule really holds true: 20% of the data will provide 80% of the value, after which you can expand your data collection.

Ideally, all data should be collected automatically, but since this is not always possible, we can start by collecting data by hand. This is common for data from old machines, context data…

The Challenge of Capturing Data

Data collection in industrial monitoring can be challenging for a variety of reasons: From inconsistencies between systems such as ERP and MES (for which we recommend setting boundaries and filtering out anomalies) to collecting too much data and being swamped by it, without getting value out of it. 

Of course, another common case is when we want to extract information from old machinery or proprietary systems.

If you want to know more about it, we recommend you to take a look at our post on Problems of data projects in industry

Data Collection from Other Applications or Software

It is essential that we can read data from other applications or data sources that we already have in our company.

• Read data through an API: hay softwares y sistemas que ya están preparados para poder consultarles información, para lo cual nos habilitan una API con la que podemos extraer la información (datos, configuraciones, estados…) 

• Through the Database itself: in customized applications or other systems, it is sometimes necessary to access the DB directly to obtain the information. The use of views and users with limited permissions is recommended.

• Exchange of information through shared files: it is always preferable to use an API or a database to obtain the information. But there are more closed systems that do not provide this option and only allow exporting Excel or CSV, so this must be taken into account when capturing data.

Finally, there are systems that are not prepared to query them for information, but rather they are the ones that send data, so it is very interesting to have some mechanism for receiving data (own API or similar).

Direct Data Collection from Machines and Sensors (Industrial Protocols)

Protocols, as mentioned above, are another important aspect when it comes to industrial data monitoring. If the manufacturer of the machinery does not provide access to the data, or only provides it through its proprietary platform (after going through the cash register), this can create a problem.

It is therefore essential that the machinery works with standard protocols. In industry there are several free protocols that are commonly used, among them we must highlight:

• Modbus: the most traditional of the 3, it requires individually consulting each equipment.
  
• OPC-UA: probably one of the most widespread alternatives. It is usually accompanied by an OPC-Server that allows any type of protocol and driver to be translated into this standard, and serve them in a simpler way.
  
• MQTT: although it was created for the IoT world, it is currently one of the main modern alternatives to OPC-UA, with the advantage that, in this case, communication is always initiated by the equipment.

To find out more about these two protocols and their differences, we recommend that you take a look at our Comparison between MQTT and OPC-UA 

What if my machines do not have these protocols? Do I have an alternative? 

If the machine protocol is proprietary (manufacturers with their own buses, etc.), it is possible to use gateways for its conversion. Industrial gateways make it possible to add different forms of communication to an existing machine or application without having to do a major redesign.

But even sometimes we will not have this option, or it will be too costly. One solution to this problem is to incorporate sensors that, while not replicating every machine metric, will provide valuable information about production.

2. Obtaining Indicators

Once we have the data, it is necessary to work on it properly in order to extract value from it. To do this, the raw data must be taken, validated and processed. 

What is important and desirable when processing data is to achieve these two objectives:  

• To have real-time information available in order to act as quickly as possible. 
  
• That the person for whom the information is intended can have it already processed to help them in their decision making. 

The objective of this phase is to obtain valuable information, And what do we mean by valuable information? Information that allows us to make better decisions. And let’s not forget that everyone makes decisions on a day-to-day basis: operators, maintenance, management, IT… But each of them needs to see different things.

The generation of indicators is done precisely to provide the user with quality information. This allows us to do things like: 

• Automatic reporting for production meetings or monthly meetings.
  
• Analyzing and classifying incidents, allowing for comparisons between shutdowns and analyzing them based on their quantity and impact.  
  
• Generating KPis, such as OEE and other relevant metrics. These indicators are essential tools for the modern manufacturing industry. Their use, analysis and interpretation are essential to improve the efficiency, productivity and profitability of companies. But to be effective, we must avoid defining complex KPIs that no one understands, as this would be a sure path to frustration.
  If you wish to know more about how to generate the best KPIs, you can check out The Best KPI for Industry 
  
  To create an effective KPI that is well tailored to your needs, consider the following:
  – Be clear about your objectives, KPIs need well-defined targets.
  – You must be able to measure your progress objectively.
  – KPI data should be collected in a clear procedure, with no room for interpretation.
  – Different KPIs will have different reporting frequencies (not everything needs real time).
  – Once we have the basic KPIs we can go deeper into other KPIs like lean.
  
  On the other hand, to avoid KPIs that do not bring real value to your business, keep in mind: 
  – You must select KPIs aligned with your objectives and goals, choose those that really drive your results.
  – Too many KPIs can make you lose focus, use the SMART (specific, measurable, achievable, relevant, time-bound) approach to decide what is crucial. 
  – Focus your KPIs on areas where you can make significant changes, not on aspects outside your control.
  – Creating KPIs should be a collaborative effort involving those who will use them, which also ensures that they are relevant and realistic.
  – Not every problem needs a KPI, so ask yourself if they really address the problem effectively. An example of a useful KPI in production would be the one resulting from cross-referencing consumption vs. reference produced, to help in the planning of our production line.
  
  
• Generating intelligent alarms, automatic responses via software (such as ticketing tools) or hardware (instructions to a PLC), warnings, notifications…
  With alarms, it is essential to avoid information overload. To do this, we must distinguish between levels of severity and ensure that notifications reach the right people, on the right device, at the right time.

3. Visualization and Industrial Grafana

The aim of all the above-mentioned steps is to generate a dashboard that presents, in a simple and very visual way, the information that really adds value to processes and decision-making. The function of industrial monitoring dashboards is, therefore, that the person in charge can know at a glance how things are going and can make decisions quickly and effectively (and then go to see the detail where necessary). 

This is precisely the concept that we like so much about Your plant at a glance. To learn more about the importance of visualisation in industrial monitoring systems, you can read our blog The importance of data visibility in your factory. 

In our experience, the simple fact of having a good visualization system, without making any further improvements, either in processes or otherwise, achieves the following results:

• 15% improvement in productivity just by bringing information closer to the operators, as the visualisation of goals and achievements helps to increase their motivation. 

• It improves reactivity to problems by 25% thanks to personalized alerts, informing the person responsible only when necessary (and not when irrelevant micro-stops, scheduled stops, etc. occur). 

• Reduces by 40% the time required to create reports for meetings by saving us from using Excel, cross-referencing data by hand, printing them..    

It is true that there are different options when it comes to generating dashboards, but from our experience we recommend using Grafana in your industrial monitoring.  

Why? Grafana offers advantages such as simplicity and flexibility when designing and creating dashboards, and above all a great adaptability that many other visualisation platforms lack. This adaptability allows each user of the dashboards to perceive them as tailor-made for their specific needs.  

In addition, dashboards made in Grafana are so aesthetically appealing that you will get that WOW! effect that we all like to achieve when we show what we do, whether to our clients, auditors or our own colleagues. 

If you wish to learn how to get the best out of your Industrial Grafana, in our blog on the best Grafana plugins for industry we tell you how to do it. Don’t miss it! 

And if you want to see a practical example, just go to our demo 👇