In the automotive industry, numerous data sources are generated during the various processes performed by manufacturers, suppliers, and service providers. The relevant data can be stored in various types of applications and databases, such as production management, quality management, shop floor data collection, and commercial systems. In the case of connected vehicles, the vehicles themselves can be a new source of data, so to speak, and thus, potential input for the analysis.
Automotive data analytics helps manufacturers to better understand the performance of their production, to detect defects and to anticipate potential problems with equipment and processes. Furthermore, companies benefit from better planning within their different units. The increasingly networked automotive industry is more digital than ever before. Thus, companies need data analytics to make decisions based on the data available.
Why Data Analytics Matters in the Automotive Industry
Automotive manufacturing includes production planning, parts inventory, dispatch, service, and customer contact activities such as sales, service and after-sales parts throughout the world.
Data analytics in automotive provides added value by allowing companies to analyze activities within an automotive company by means of data, instead of by periodic reports. It helps to better understand the way things are and, by identifying patterns and trends, enables companies to be prepared for the future.
Even with traditional reporting tools, a lot of value can be generated by combining data from different systems to create historical reports that help explain trends and even allow for the investigation of single instances to identify root causes.
Within an automotive manufacturer organization, the use of data analytics can highlight a number of issues such as production losses, defective products and potentially equipment failures before they occur. Outside of the automotive manufacturer, the same techniques can also provide insight into customer behavior, measuring performance of suppliers and managing inventory and commercial aspects of automotive businesses.
How Automotive Manufacturers Use Data Analytics
At first glance, the use of data in production seems counterintuitive. Automotive manufacturing involves a very large number of individual components and processes, which are executed by human workers. A massive amount of data is generated by production, but one must first analyze all the individual process steps before the root cause of a potential deficiency can be determined.
Production inefficiencies often have a cumulative effect. This means that even seemingly minor production setbacks, such as occasional short stoppages, longer than usual changeovers or production processes that run below their average cycle time, can collectively cause substantial losses of production capacity over time.
In manufacturing analytics for the automotive industry, production performance of vehicles can be measured by analyzing output against target output. Here we analyze all the operational measures of production like equipment utilization, cycle time, downtime, quality, etc. across different stages in the production of a vehicle and at different levels like line, machine, model and period.
However, understanding production loss is much more valuable than simply increasing production. By identifying areas that have the greatest impact on overall production, automotive manufacturers can implement targeted and effective improvement initiatives.
The same goes for quality issues. In the automotive industry there are hundreds of thousands of components and very complex production processes. There can be numerous causes for defects. Therefore, it is important to connect quality information with production and process information. This enables teams to analyze in detail whether defects occur more frequently with certain machines, materials, suppliers, production conditions or even with specific production series.
By connecting the quality defects with the production data, quality management is no longer only examining the defective products, but also tries to understand the circumstances that caused these defects.
The Role of Predictive Analytics in Automotive
One of the key trends in the use of analytics in the automotive sector is to move from analysis of historical information to predictive analytics to help plan for future outcomes.
Predictive analytics in the automotive industry predicts future events using historical and current data to identify patterns. In the automotive manufacturing environment, this may enable manufacturing to better plan and respond to future events, particularly with regard to maintenance of plant and equipment.
In this context, predictive analytics would utilize historical and current data from items such as machine readings, maintenance records and current operating conditions of plant and machinery to identify changing behavior that may indicate an increased risk of failure.
Predictive maintenance does not mean that every equipment failure can be predicted. The behavior of machines and components in the automotive industry varies greatly, and it is impossible to derive conclusions from available data 100% of the time.
To give another example, a machine that in the past always operated within a specific range of values starts to behave differently. This change of behavior can now be detected by analytics, and the relevant personnel can be notified in good time to start to prepare for a potential failure of this asset.
However, for other automotive problems with plenty of historical data, such as predictions on quality, production and even demand, it is worth developing predictive analytics in order to get ready and act on time.
From analyzing “What happened?” to predicting “What is likely to happen?”
Vehicle Data Analytics and Connected Vehicles
Today, data analysis extends far beyond the boundaries of the plant. With the increasing number of networked systems in the car, a large amount of vehicle data is available in the vehicle and is transmitted to the manufacturer’s servers.
Even when a car is on the road, lots of data is generated by all connected systems and sensors. Analyzing this so-called vehicle data from connected vehicles allows insights into the vehicle’s usage and behavior that would not be available if vehicles were operating in isolation as fully developed products.
To support the optimization of production and service in automotive companies, data collected from the vehicle can be analyzed. Different applications for usage-based services, for example, can help to detect problems or even patterns of the driver’s behavior. For other applications like maintenance and service, the data can be used to plan in advance.
More vehicle data does not automatically mean more value. First, there has to be sufficient reliable, relevant and accurate data available to answer business and operational questions. In addition, the various sources of vehicle and business data need to be connected to other relevant data without generating another silo of information.
Analytical capabilities are as important as the connected technology in order to create value from the generated data.
Big Data and the Automotive Industry
The data that is collected from the vast array of sources within the automotive industry is a key element for big data analytics. The business processes are distributed across a variety of locations, including production sites, suppliers, dealers, customers and service workshops.
Big data analytics in the automotive industry has far more value than just massive amounts of data for processing purposes. There is lots of potential in connecting different types of data and establishing links between them.
For example, production data can be connected with maintenance data to analyze how the performance of equipment during production affects output. Quality problems can be analyzed in conjunction with production data to determine the root cause of recurring problems. Supplier information can be analyzed in conjunction with quality and delivery data to better evaluate suppliers.
Production and inventory data are also combined to make correct material requirements planning (MRP) decisions for future production and in order not to run out of stock.
Different systems have different data formats and use different definitions for essentially identical things. They have different field names. Historical data, where it exists, can be incomplete. Massive amounts of poor-quality data render even the best decision-support models ineffective.
Much poor-quality data is generated in the industry. This will only make decision-making harder. Advanced analytics tools can create sophisticated models to analyze the data. However, the results are only as good as the data used for the analysis.
Analytics Across the Automotive Value Chain
An even more important area of application of business analytics in the automotive industry is all business processes and companies within the value network of the automotive manufacturer, e.g. suppliers, dealers, customers, service and logistics providers.
Business analytics, therefore, extends across the automotive value chain: from supplier to end customer. It helps manufacturers and their partners to manage their operations and their distribution and sales networks.
In the case of a manufacturer, analytics would include production data but also data about how demand for vehicles is changing, the performance of the suppliers, where there is too much inventory and why customers are not buying as many vehicles as expected.
For example, while increasing the plant’s production to reach higher levels of equipment utilization seems to offer more operational efficiency, it might actually create more inventory that does not meet demand. Moreover, while reducing suppliers’ costs seems to create no other operational problems and even has the potential to save more costs elsewhere, poor quality or late delivery from suppliers can cause many more problems in other parts of the company.
Analyzing the information for production and business operations will reveal these trade-offs.
All of this can be supported by customer analytics, examining customer information in automotive, determining purchasing behavior, preferences and even behavior in customer-facing activities.
Analyzing data to make decisions at every point in the business where there is data is the true value of automotive analytics, and it is not limited to the shop floor.
Moving From Reporting to More Advanced Analytics
Different levels of analytics answer different questions.
Descriptive analytics describes what has happened in the past and can be supported by typical reports, such as production reports, quality dashboards and key performance indicators (KPIs).
Diagnostic analytics examines why events have occurred. Analyzing production, such as comparing the performance of different production lines, can help with root-cause analysis of events or trends within different products or time periods.
Predictive analytics analyzes historical and current information to make predictions about future events. Examples include potential equipment failures, quality failures or changes in production requirements in the automotive industry.
Finally, prescriptive analytics assesses potential alternatives within given business constraints. This could be used to assess different production schedules or even to plan the optimal distribution of production resources.
It is important to understand that a company does not necessarily need to move to the most advanced form of analytics within the automotive industry. Many simple and reliable dashboards exist within the automotive industry today. It is key to establish solid foundations within analytics for automotive companies.
The foundation matters.
Challenges of Using Data Analytics in Automotive
There is significant complexity in data distribution in automotive manufacturing across multiple plants, functions, applications and technology generations. For example, a manufacturer may have modern connected machines running in parallel with legacy equipment and systems that have not been designed to exchange data in an automated fashion.
Different data have been collected in different systems within a plant and even within different plants. Moreover, much of this data is not accurate, and there are often lots of gaps within historical data.
Can the analytics then be used in decision-making? Even the best model only has value when it is actually used by others. Data scientists often develop complex models, but these are of little value if there is no understanding of what the results mean.
Implementing analytics to support the business therefore requires more than technology. A thorough understanding of data management, data integration, domain knowledge of the automotive industry and process knowledge of how decisions are made in the organization are also required.
Making Data Analytics Part of Automotive Decision-Making
Analytics will never be a better judge than the automotive professional themselves, but in the end, it will provide them with even better evidence to make their decisions.
An experienced production manager would recognize that a particular production line behaves differently under certain circumstances. The same maintenance engineer would recognize early signs of wear on specific equipment parts, and analytics could then identify similar patterns of wear on other, comparable equipment parts.
Combining domain knowledge and data analysis is essential to interpret the resulting information in the specific context of decisions made on the shop floor or throughout the organization as a whole.
There are plenty of questions that can be transformed into data science projects, but it does not mean that one should. A better approach would be to identify the business problem first, then establish the information needed for that problem and develop the necessary analytical solution to support decision-making by individuals on the shop floor and in the organization.
Automotive companies are already producing large amounts of data in their daily business, and the big opportunity is to join this data with the people and decisions that are to be influenced.
Another sector example could be a manufacturer with recurring equipment downtime, for which they first analyze production and maintenance data for this specific issue. A different organization could start analyzing data for its quality losses, production bottlenecks or even inventory issues. However, once the data foundation and the related analytics approach for one relevant problem have been set up, this can also be used for other problems within the organization.
Kepler Advanced Analytics is a top-tier data science service provider that specializes in data engineering, AI, advanced analytics, business analytics and operational analytics. We help companies generate value from data by translating complex data into meaningful insights that organizations can put to practical use in driving business decisions.
Future growth in the automotive and manufacturing industry will be created by connected production sites and vast amounts of data being generated within them. While big data volumes are key for future success, the company that will reap the most benefits from this opportunity is not necessarily the one with the biggest volume of data, but the one that best connects this data to the decisions made in production and beyond: for quality, maintenance, the supply chain, customers and future development.