HOUSTON & SAN FRANCISCO, USA – Stuart Morstead, Arundo’s Chief Operating Officer, gave a keynote address at the recent Wells Fargo Energy Conference in San Francisco on the topic of machine learning and advanced data science in the energy industry. Like other heavy industries, major energy companies are coping with tectonic shifts in the data landscape, driven by reduced costs for data generation, storage and compute; near-ubiquitous connectivity; and a growing set of sophisticated machine learning and other analytical methods to apply to rapidly proliferating data sets. These trends are collectively driving the Industrial Internet of Things.
Since at least the 1980s, successive waves of global forces, new technology, and resulting best practices have permeated heavy industry with varying degrees of rapidity. The Industrial Internet of Things increases the transparency, auditability, and velocity of business data and decision-making. The resulting wave of digital transformation is already changing how heavy industries behave.
Industrial companies that adapt most quickly to this new environment are likely to significantly lead their slower competitors in a world that may look increasingly like the “winner takes all” landscape of some technology-driven markets. Industrial leaders are already investing heavily to create such separation.
Stuart discussed Arundo’s experience delivering material improvement in operational performance for leading companies in energy. This typically involves generating meaningful business insight from massive sets of time series data and other types of disparate, unstructured data sets.
Arundo has delivered large-scale data science use cases to a variety of industrial leaders, including independent and national oil companies, unconventional drillers, power utilities, and large maritime fleets. A few common threads typically repeat in all of these companies:
- Data science-driven solutions could improve revenue, profitability, safety, and other measurable sources of value related to physical assets and equipment.
- However, corporate software infrastructure and IT capabilities often do not support the deployment of large-scale data science models, even when internal data science teams exist.
- In many cases, scarce data science resources are focused on core business drivers (e.g., oil exploration), ignoring other critical sources of value (e.g., equipment maintenance).
- Business leaders are sometimes isolated from ongoing data science-driven insight by a lack of connectivity between data science models and existing operating dashboards and control systems.
In addition, many of these organizations face similar challenges in beginning their digital transformation:
- Multiple sources of data need to be analyzed in order to create a single source of truth.
- The volume of time series data far outstrips human capacity to monitor or analyze.
- The messiness of unstructured text or numerical data requires significant work to extract insights.
- Traditional analytical methods are less useful than machine learning techniques to generate meaningful insight from such disparate data sources.
For these reasons, large-scale data science has yet to fully take hold in many energy and other heavy industrial companies.
Nevertheless, many asset owners and operators are “sensoring up” their physical operations – even before finalizing the new business strategies, operating processes, and software tools required to realize value from new digital assets and data streams. Not surprisingly this is creating both anxiety and opportunities for stakeholders throughout the value chain.
One example Stuart discussed at the conference involved a European pump manufacturer seeking to bundle digital capabilities into its core product line. As larger industrial OEMs begin to bundle asset performance management and predictive maintenance solutions into their physical products, this pump manufacturer realized its potential disadvantages in future procurement cycles if it could not offer a digital solution. Arundo partnered with this company to create a best-of-breed condition-based monitoring system. The pump manufacturer is now introducing this solution into leading oil & gas companies globally.
Many OEMs supplying heavy industries face existential questions about how to incorporate software and data science expertise into their standard physical products. In markets where profit pools typically accrue to market leaders, investing to achieve technical leadership in areas such as software and data science is challenging for all but industry leaders; to do so outside of core business competencies comes with serious risks. In addition, enticing the best talent in software and data science to work for industrial equipment manufacturers – often in locations away from the urban innovation hubs where such talent is typically most attracted – is a significant challenge that compounds the issues facing industrial OEMs in this new data landscape.