Data is not always available and formatted in a way that facilitates interpretation and contributes to assertive decision-making. It is common to hear reports that the volume of information collected in the industry is growing, but this information is not categorized, hierarchized, and much less integrated – as illustrated by the article “Use of data in the manufacturing sector“, published by STRATUS Blog in 2023.
Despite the notable advantages of the digitalization of the industry, according to the Data Paradox study (2021) – carried out by the consulting firm Forrester and commissioned by Dell Technologies – only 66% of companies in the world use data, and only 21% of these know how to take advantage of it properly to generate results.
In Brazil, the scenario is not much different. In 2016, according to a study by the National Confederation of Industry (CNI), called Special Survey on Industry 4.0 in Brazil (2021), only 48% of industries used some digital technology at a certain stage of production. In 2021, this percentage rose to 69%. Most of these companies are still in the initial process, known as “data newbies”, and do not know very well how to deal with the large volume of information.
To identify the main variables of a piece of machinery, it is necessary to define which analysis should be carried out and verify that the objective is aligned with the strategic goals of the business. After that, it is necessary to understand what data is available in order to identify the relevant data.
However, factors such as the presence of machines of different technologies in the industrial park, low-resolution manual processes (with incomplete information) or time gaps between one data and another usually prevent a stable analysis.
Often, the most important data is not easily available, but that does not mean that it cannot be found. In these situations, the data scientist, using the right tools, can create the data that makes the most sense to be monitored – from the combination of two or more variables that can be extracted natively from the factory floor.
Creating Data from Data
One way to create the data that needs to be monitored is to use formulas with the available information to calculate the key indicator. In industry, improving the performance of a production line often means innovating: through different management practices, new technologies, efficient designs, and rethinking the use of inputs.
To perform productivity analysis, it is necessary to measure and compare performance indicators, such as production quantity and speed, among other pre-defined standards. In this way, it is possible to identify areas for improvement, implement corrective actions, and monitor the impact of these actions on the plant’s overall productivity and performance.
In this example, we will consider 2 machines that, for about 40 minutes, have been producing at the same time (same availability), and that obtained the following results:
Machine 1: 10,287 products
Machine 2: 11,120 products
When comparing the results, there is a difference of almost 8%, which will impact the result of the entire month if it is not resolved. When searching for the productivity data of these machines, the following information is obtained:
Analyzing graphs separately can make it difficult to identify problems and optimize the use of low-yield machines. Therefore, an effective technique is to combine both information into a single image. While this may not provide a clear conclusion, it does allow for yield comparison.
Data found: how to perform a relevant analysis
Within the productivity analysis, one of the indicators that can be monitored is the speed of the machines, as mentioned earlier. Since the production data is collected every 5 seconds, it is possible to calculate the speed every 30 seconds with the formula:
Speed=∆Production/∆Time
Even if the “speed” variable cannot be extracted directly from the machinery, it is possible to discover it and focus only on the most relevant data to understand the performance of the line. The answer to the question “which one should I look at” is often nebulous, precisely because the information sought cannot be collected directly from the equipment, but can be created. When comparing the data, it is evident in which period there was the greatest discrepancy and what can be done to improve performance.
In this comparison, it is clear that the speed of machine 1 is lower than that of machine 2. At no time can machine 1 reach the same speed level as machine 2, being about 6% below at peak performance moments.
With this information, it becomes evident that the work to improve the overall performance of the line must focus on increasing the production speed of machine 1, with a gain margin of more than 6%. Machine 1 still doesn’t operate as fast as machine 2, which is more efficient. With this, and with the discovery that it is possible to increase the speed of machine 1, the result will be greater throughput in less time.
This work needs to take into account the rated speeds of subsequent machines. In this case, a V-Graph can be a great aid tool.
Data that unlocks results
The use of data in industry is an increasingly present reality in Brazil and in the world. With a growing demand, digitalization allows factories to achieve greater production optimization, meeting needs and requirements in less time and in a more automated way.
To do this, it is necessary to monitor the right data, which is not always available natively. That doesn’t mean it can’t be found. On the contrary, with a team of data scientists, it is possible to calculate the variable that should be observed, either by comparing two variables or using formulas.
Although the example used deals with a performance analysis, this line of reasoning (data comparison) also works to monitor other areas, such as the use of raw materials, natural resources, maintenance, product quality, among others. This scenario is realized thanks to the range of possibilities allowed by data science and the analytical capacity of scientists, who use their experience, creativity, and market vision to find the most relevant metrics for the industry. Using methods like these transforms data into valuable information that unlocks results.
Discover how data science contributes to different applications:
Performance
Raw Material
Natural Resources
Quality
Research & Development