Data sharing in industrial ecosystems (part 3 of 4)
Data sharing in industrial ecosystems (continuation)
by Prof. Dr. Boris Otto, Prof. Dr. Nikolaus Mohr, Matthias Roggendorf, PhD and Tobias Guggenberger
Key challenges in exchanging data in industrial ecosystems
Data ecosystems obviously have clear benefits, so what is holding companies back from taking full advantage of them? The reasons are manifold. Some involve organizational, technical, or legal barriers or restrictions, while others are fueled by typical ways of thinking and experience-based concerns. They can be divided into four major categories:
Challenges related to culture and mindset:
New value opportunities depend on sharing data in larger industrial ecosystems that may involve parties with limited trust in one another. Participants might be able to analyze shared data to derive confidential business information. Considering every eventuality is difficult. For example, it seems reasonable to share welding position data so that another party can optimize production pathways, but a competitor could potentially use that data to analyze the resulting shapes – and determine in advance what new products the first company is planning. Typically, however, data in isolation rarely creates a competitive advantage.
Another major reason that companies do not engage in data sharing models is the perceived lack of control when data leaves their premises. For example, software licensing has reached a high level of maturity (in terms of both contracting and technical measures) but data sovereignty is still hard to guarantee and implement.
The need to identify enough value to justify investment can also be an obstacle. Some organizations have difficulty breaking out of existing thinking patterns, while others lack the capabilities to run the kind of structured assessment needed to drive engagement in data ecosystems. Often, this requires a collaborative give-and-take governance model at odds with traditional buyer-supplier-relationships.
Challenges related to the need for ontologies:
Data can only be shared if it can be easily interpreted and quickly integrated with other data sources. It must be unambiguously understood by all ecosystem members. Thus, a common language or ontology is required all parties agree on. In some industries, such as banking or healthcare, such data languages were established decades ago to enable interoperability and the leveraging of standard software. In the industrial sector, however, individual ontologies and standards exist – such as eCl@ss and the Asset Administration Shell, to name just two examples. What is needed though is a comprehensive architecture of standards which includes different types of data such as master data, reference data, manufacturing and supply chain event data etc. This architecture stack cannot be developed by an individual ecosystem member, but requires consensus within a community of practice.
An industrial data ecosystem promising a significant competitive advantage could be a strong impetus for companies to pragmatically define standards for sharing data beyond single-purpose cases and allow other parties to interpret it correctly. Generic ontologies for exchanging sensor and robotics data that can be easily extended are on the rise due to advances in connected devices and the need for interoperability in heterogeneous environments. However, industrial environments are often a mix of older systems and new IoT-based sensors, making it even harder to reach agreement on common standards.
Technological challenges to designing and mastering the required platform and services:
Even when companies understand the value of sharing data in industrial ecosystems, they often lack the skills to make their data available in an effective and efficient way. The challenge starts with internal technical interfaces for supplying data and continues with developing an environment to process and integrate data from different sources and run models using advanced analytics methods. If companies decide to build new products based on such data, they also need the capabilities to productionize the resulting models and ensure effective commercial and technical operations, including guaranteeing service levels and scaling to potentially millions of customers. These capabilities typically do not exist in engineering-focused industrial clients, so they need to be systematically built either as new entities or with external help.
Challenges in effectively managing data to make timely, automated access to it possible:
The last challenge relates to the internal capabilities required to provide data from internal applications that fits its intended purpose, is of sufficient quality, and is available in a timely fashion. A lot of data sits in siloes and is not even accessible. Other data may be used for its primary purpose (for example, on the shop floor) but not for any others. Also, many organizations, especially in the industrial space, have limited capabilities for managing their data assets strategically. As a result, they have trouble distinguishing between data that can be easily shared and strategic data that poses significant opportunities but entails the risk of exposing intellectual property to untrusted entities. Before these companies participate in a data ecosystem, the need to establish basic data governance principles and systematically develop their employees’ data literacy.
So far, only niche solutions in highly selective segments have been developed (see boxes for examples). In many cases these data sharing ecosystems were created by a dominant player with the power to mandate that other market participants join the network.
Lufthansa’s AVIATAR is a platform for the aviation industry. It integrates data from sources on airline operations, aircraft, maintenance systems, and more to create a comprehensive tool for fleet management. As it works independently of organizational or technical boundaries, it can be used to optimize all operations of any imaginable configuration of fleets.
AVIATAR focuses on digitizing the value chain, from predictive maintenance to automated fulfillment and on to complete digital maintenance, repairs, and operations. Add-ons can provide complementary features such as layover monitoring, layover sourcing, and staffing. Finally, marketplaces for loaning and exchanging components and tools support ecosystem-wide optimization.
365FarmNet is an innovative software platform for manufacturer-independent farm management. The service integrates multiple partners and data sources to create a comprehensive management tool. Its modular architecture combines free-to-use functions (such as data management, weather analytics, documentation, and basic planning tools) with complementary value-added services (for telematics, route optimization, soil sampling, and more). The open interface makes it possible to integrate intelligent apps from a variety of agricultural suppliers, such as machinery manufacturers or pesticide and fertilizer producers, creating a data ecosystem around the platform. The approach demonstrates that bringing an integrated value proposition to customers requires sharing data across sectors.