Data analysis in production and logistics

Big Data Analytics in production, which means initially analyzing machine data for the optimization of production and logistics processes. The central task for companies is to integrate and analyze “dumb” raw data from machines and operating processes in such a way that they can be transformed into real control information. Software solutions examine the totality of existing data immediately after the time they are generated, and can recognize hidden patterns that indicate potential errors or potential for improvement. You’ll find the finest details, which also escape the eye of the most experienced expert.


> code>

In the area of ​​maintenance management and optimization, sensor data is particularly relevant. These machine-to-machine data (M2M), which are generated in buildings, devices or within the infrastructure, are used to capture states and influences. State-of-the-art production plants are constantly producing large quantities that can be automatically analyzed by equipment manufacturers or specialized service providers.


Thousands of parameters in view


If the software detects any inconsistencies in the data, it is examined whether it is relevant information or merely statistical outliers. If necessary, the responsible servicemanager will be alerted to take the necessary steps.


Benefits of Big Data Analytics


However, when the corresponding analytical software is used, the manufacturing sector still needs to catch up, as the study "Competition Factors Analytics" by SAS and the University of Potsdam shows. Only 37 percent of the manufacturing companies interviewed in the study evaluate their machine and sensor data at all. And where analytical software is used, this is usually only spontaneous, individual and not strategic.


Achillesfere data quality


Nearly all industrial companies surveyed use less than half of the company's data currently available for analytical purposes. If data are used at all, they are mainly derived from systems for enterprise resource planning (eg SAP) or customer relationship management. Sensor and machine data play only a subordinate role.


However, a significant application example of the advantages that Big Data Analytics can bring with regard to machine data is "Predictive Asset Maintenance". The predictive maintenance is not only used in the production, but also in the energy sector.


An example of this is the oil producer Shell Exploration and Production, which collects daily large quantities of data on pumping, drilling, and material consumption on the Perdido oil rig in the Gulf of Mexico and then automatically analyzes it.


SAS Predictive Asset Maintenance's big-data analytics software specifically looks for deviations from certain data patterns and gives the engineers information on the condition of the machines. For example, the system alerted when the likelihood of a pumping pump failing soon increases. A single maintenance engineer is therefore able to keep thousands of parameters in mind at the same time - and react quickly.


An example from the industry, where the analysis of machine data led to a significant improvement in processes, is Posco: The South Korean steel group had too much waste in hot-rolled steel strips. Conventional statistical methods could not identify the cause. The analysis of the physical processes using SAS software led to a reduction of the committee from 15 percent to 1.5 percent.


Elsewhere, the analytical software was able to uncover the trigger for profitability differences between different production plants - this was the starting point for optimization measures that brought in 1.2 million US dollars annually. The use of Analytics also helped to ease inventory. Following appropriate optimization measures, Posco is able to cover the customer's demand by as much as 60% of its inventory.


As the examples given show, the evaluation of machine data by using a powerful analytical solution offers numerous advantages. The most important ones are


Automated processes are only as good as the data on which they are based. Human errors, faulty sensors, inconsistent and outdated data influence the results. And in processes that run in secret, the errors (sources) are not immediately apparent. The machine-generated data must be matched with the expected values ​​and integrated into the monitoring process.


For example, a logistics company has integrated hundreds of sensors into the fleet of its fleet. If one of them reports too low tire pressure, this is necessary. There is a need for action in any case, whether the sensor correctly reports a damage or is simply defective.


Also interesting is


To ensure that this need for action is recognized at an early stage, it is crucial that the analysis is based on current, consistent and complete data. However, the analysis of data often fails in data quality. Some reasons


A high quality of the data is crucial for making informed decisions and for effective measures to be taken. So-called data governance enables data to be analyzed, improved and controlled.


Strategic approach required


When a company's strategy for improving data quality is set up, the first thing to clarify is: what is the meaning of "good" (internal and external) data? The answer: They must be consistent, up-to-date and complete in all systems. What is meant by "current" can differ from company to company and from industry to industry. For example, energy companies use data from smart meter and smart grid sensors in production and transmission facilities to provide accurate consumption estimates.


How many customers will use electricity at the same time? How will they use it? How will the weather affect demand? The more accurate the predictions for consumption, the better the energy supplier can be and the less compensation energy is necessary. An increase in the measurement density from 15 to 3 or 1 minute intervals can result in a significant improvement in data quality.


In the course of increasing digitization, the amount of machine and sensor data will continue to rise. This information offers a huge potential: whether it is to optimize the network utilization, for the early maintenance of machines, as a basis for decision-making in the lending business, for the planning of vehicle fleets or for the prediction of customer queries. However, exploiting this potential requires high data quality and high-performance analysis software.


Conclusion


Anyone who merely collects data, but does not prepare or evaluate it with Big Data Analytics, renounces an important decision-making basis. Companies that rely solely on their gut feeling and experience values ​​instead of making data-driven decisions are missing important business advantages.