What kind of business requires data analysis

Forward-looking data analysis and quality assurance

Xiao Wei19. April 2017

Already used today: data analysis and quality assurance

Data analysis is already very popular, especially in the financial world and in retail. In fact, many believe that data analysis is an absolute must in today's business world. In the process industry, the amount of data, the data speed and the data variance have increased explosively in recent times. The question therefore arises as to how the efficient use of advanced data analyzes (e.g. statistical analyzes, predictive analyzes, big data analyzes, etc.) can reduce operational risks and improve business results. This topic is of great importance to us and our customers and requires the greatest attention here and now.

We are facing increasing low-cost competition from a number of emerging markets that are putting strong downward pressure on profit margins. Another challenge for the industry is the ever-increasing demands of consumers, which forces us to keep larger stocks on hand, allow for a shorter product life and offer higher quality levels.

Every business involves risks. Some of these risks, such as customer preferences or competitive forces, are extremely difficult to assess and manage. Other risks, on the other hand, such as operational risks, are easy to understand, identify and mitigate, which results in increased operational excellence, higher profit margins, competitive advantages and better customer relationships. Operational risks are directly related to five general aspects: product quality, efficiency, plant reliability, safety and human reliability. This article deals primarily with the question of how advanced data analysis can help to gain new insights into the topic of quality and compliance.

The challenge of quality assurance

Many companies have recently struggled with quality assurance issues, resulting in wasted raw materials, product recalls and even a loss of corporate image. The key questions frequently asked by our customers are “Why is it becoming more and more difficult to achieve our quality goals?” And “How can we find out in advance, i.e. before the product is delivered, that something is wrong with the product?”.

We identified two key quality assurance challenges: aging equipment and the increasing demand for higher quality and consistency.

The equipment in many companies is becoming increasingly obsolete and unreliable. It is therefore becoming more and more difficult to optimize the condition of the system and to achieve the quality goals that have been set. Another phenomenon is an unprecedented demand for high quality and consistency. This problem arises in both Western Europe and Japan, although these countries are already leading the way in the industry in terms of quality and consistency. To further increase the level of quality, companies have to go through their databases and check all data for the entire production plant and all processes. However, due to the numerous factors involved, the identification of all quality-influencing parameters is extremely complex and difficult. Most companies think they know all of the key variables that are responsible for the quality of their processes and products, but in our experience this is pure wishful thinking. This inevitably leads us to the following question: How can we reconcile aging and increasingly unreliable systems with customer demand for better quality and consistency when there are so many hidden quality parameters that have a strong influence on process and product quality?

Yokogawa's innovative solution for quality assurance

Based on its many years of experience, Yokogawa has developed an innovative solution for quality assurance: Process Data Analytics, an application software for the detection of quality and productivity losses in an early phase of the manufacturing process. This is made possible by analyzing process data, system status information, the operating history and other data.

This solution focuses on data processing and preparation and uses a structured method for identifying, solving and modeling product quality problems based on the 4-M method - man, material, method, machine (Figure 1).

A "tongue twister": The pattern recognition method used

In addition to this method, we have also developed our advanced data analysis software. This industry-leading solution uses the Mahalanobis-Taguchi (MT) method to analyze a wide variety of statistical variables. MT is a pattern recognition method that is widely used in quality control and quality management.

First, the standard deviation is determined from two or more data sets in the normal state. The compared data records are then recorded from the data in the comparison state. The basic principle is based on measuring the distance between the standard deviation and the compared data sets, taking into account the interactions between the data sets. The normality or anomaly of the compared data set is assessed with the help of this distance, the so-called Mahalanobis distance (Figure 2). Any deviation from normal conditions is recognized and a warning message is issued about a possible loss of quality.

Far more than just a factor

In the process industry, especially in the batch-oriented process industry, there is a high degree of anomalies and extremely complex relationships. Up to now there has not been such an intelligent and reliable early detection system for quality problems with which plant operators could specifically interrupt or adjust operational processes during production using quality detection in real time. This is where our pioneering software solution Process Data Analytics comes into play: Process Data Analytics analyzes production processes based on parameters such as temperature, pressure, flow rate, fill level and other process data as well as data on plant processes and device maintenance, which are provided by a PIMS (Process Information Management System). , DCS (distributed process control system) or PLC (programmable logic controller).

Since quality is influenced by various factors, the key to success lies in the implementation of this solution, which brings all employees across several departments together and enables unrestricted collaboration. Most data analytics companies in the market, however, simply collect data and do the data analysis in-house without considering the customer's operations, thereby disregarding critical factors and drawing incorrect conclusions.

Helping people help themselves

It is also common practice that customers are provided with the data analysis software, but they have to carry out the analysis themselves. This type of software is generally anything but user-friendly and far too complex to use for not highly specialized users. Therefore, the software is hardly used by the customer after purchase. When developing Process Data Analytics, Yokogawa relied on a combination of data analysis and knowledge of the customer's plant processes. We not only sell this unique software solution, but also work closely with our customers in our role as solution specialists and accompany them until they can master their quality problems independently.

Basically, we all know that efficient quality control is essential to building a successful business. We are all aware of the important role quality assurance plays in achieving cost reductions and protecting assets. But are these really the only advantages of consistently high quality? Many companies and even the market leaders in our industry will answer this question in the affirmative. Quality control functions are often seen as unnecessary overhead, and in some cases even fall victim to cost-cutting measures.

Quality control is all well and good, but not particularly attractive.

Is that correct? Let's look at the following theses:

Companies that are perceived as suppliers of higher quality products generate three times higher profits than other companies.

Businesses can increase their profits by almost 100% by increasing their customer loyalty by only 5%”.

These are the conclusions of C. D. Heagy in his study on intangible assets and their effects. In fact, the positive effects of quality assurance on sales / profit generation and the increase in customer loyalty are fundamentally underestimated or even ignored. For example, as part of our collaboration with a tire manufacturer, we analyzed its quality parameters and worked out an optimal solution concept together. Due to the widely varying customer demands, the tire manufacturer's quality goals are difficult to implement. In this case, with the help of the so-called Response Surface Method (RSM), we have created action surface plans to establish a causal relationship between quality and production conditions in a trade-off relationship. By carrying out our analysis, a production system was set up that enables flexible adaptation of the production conditions to the customers' changing quality requirements. With our analytical quality assurance solution, our customers can accelerate the sales phase and improve the competitiveness of their products, which ultimately leads to a significant increase in their sales and profits.

This is how it's done

Our Process Data Analytics consulting service is currently used by 120 companies based on the following general procedure.

Step 1: Detailed location analysis

  • Complete process analysis
  • Problem detection
  • Formulating a hypothesis based on experience
  • Data acquisition (e.g. records of process data, quality data, operation / production / maintenance, etc.)

Step 2: pre-process data

  • Data cleansing
  • Feature extraction from data (e.g. Figure 3)

Step 3: Statistical data processing and preparation

  • Choice of the appropriate method
  • Product quality indexing
  • Anomaly detection / early detection (e.g. Figure 4)

Step 4: root cause analysis

  • Investigation of the cause (s)
  • Determination of the cause (s)

Step 5: reporting

  • Recommendation of countermeasures
  • Preparation of a report

Summarized

Yokogawa's innovative software solution Process Data Analytics offers a combination of advanced data analysis and in-depth process knowledge with which our customers can ensure and continuously improve the quality of their products.

This solution is used to control and ensure product quality in the oil industry, the petrochemical industry, the pulp and paper industry, the iron and steel industry, the pharmaceutical industry, the food industry, the automotive industry, the glass and rubber industry, the electrical and electronics industry and other industries. Process data analytics includes:

  • comprehensive methods
  • a future-oriented analysis software
  • qualified data analysts.

Data analysis has changed the world. We are doing our best to improve the lives of our customers and society as a whole using these great technologies. Are you ready for a better future?

This post is a "Re" blog. The original article was published in English on the Advanced Solutions Blog. Read more exciting articles here.

[Total: 0 Average: 0/5]

The author

More information