Can OEMs Survive Digital Transformation and Hardware as a Service?

By Guest Blogger: Eitan Vesely

Category:
Industry Trends

If Industry 4.0 is to be defined as a revolution, we should consider a universal rule that applies to all revolutions: New elites replace the existing centers of power. How does this apply to today’s OEMs? OEMs are that is integrating production equipment (OT) with IT systems which are becoming smarter by using Machine Learning and AI.  The traditional OEM revenue stream was based on selling industrial equipment that was often bundled with O&M service agreements. This model was valid when product innovations were incremental, and the expected lifetime of industrial equipment was measured in decades.

Today, we are in a transitional period. OEMs recognize that they will need to adapt to a new model but that they must operate in the absence of a common definition for Industry 4.0 or consensus technology standards for IoT infrastructure.

(Re)Introducing Hardware as a Service

In 1962, Rolls-Royce launched the “Power-by-the-Hour” model. Instead of purchasing jet engines, customers leased the engines, while Rolls-Royce maintained responsibility for performance and maintenance.

A confluence of factors is driving the adoption of an updated version of the Rolls-Royce leasing model, known as Hardware as a Service or HaaS. First, the cost of embedding sensors within industrial machinery has drastically declined, as has the cost of storing, transporting and analyzing of the Big Data that these sensors generate. Second, until recently, the field of Machine Learning was mostly confined to academia. Today the discipline is attracting funding and talent. We are witnessing an injection of billions of dollars in corporate R&D and Venture-Capital-funded startups, as companies recognize the potential applications of Artificial Intelligence.

Traditional maintenance technologies, such as SCADA rule based monitoring systems, were partially effective and most sensor data was not analyzed. With advances in data science, Machine-Learning-based Predictive maintenance is the key enabler of HaaS. 

At a strategic level, HaaS is a win-win for both the OEM and the industrial plant. If the equipment is leased, the plant can focus on yield rate performance, while equipment health becomes the OEM’s responsibility. The OEM applies the Machine Learning algorithm to the sensor-generated data in real time so that evolving failures can be identified before unplanned shutdowns occur.

In multiple engagements we have with OEMs, we are witnessing today the establishment of remote monitoring control rooms from which the OEMs are monitoring their equipment at multiple customers’ sites. On top of providing this service, OEM’s engineers are now able to close the loop on their machines performance, identify design related issues and perfect the next generations. 

Potential Inhibitors of HaaS Implementation

The success of HaaS and digital transformation should not be taken for granted. Three significant hurdles must be overcome.

First, OEMs lack core competencies in Machine Learning and are likely to struggle to build internal capabilities. OEMs that manufacture equipment built for decades of use are likely to struggle in the dynamic, rapidly evolving field of Machine Learning. For instance, within the last two years, Automated Machine Learning (AutoML) has become one of the most important topics for data scientists. Moving from large-scale equipment production to specialized Machine Learning data science will not be easy.

The second hurdle is the nature of the relationship between OEMs and their customers. The current transactional model is relatively simplistic: The machinery’s owner is responsible for its health. With a leasing model, both the industrial plant and the OEM share responsibility for the asset’s performance. In theory, the interests of the OEM (as the owner) and the industrial plant (as the operator) are completely aligned. Downtime is the common enemy. The OEM is contractually obligated to maintain its equipment and may face penalties if it does not maintain performance levels. For the operator, in the case of unscheduled downtime, production is lost, thereby hurting yield rates and financial performance. 

The reality is more complex. Even with the best of intentions, the operational priorities of the OEM and the plant are unlikely to converge.   

According to Schneider Electric and Hartford Steam Boiler’s expert assessment, many contributing factors lead to equipment failure, including environmental issues (8% of failures), humidity (9% of failures) and improper operations (9% of failures). In a survey of 200 executives by Honeywell, 42% admitted to running equipment harder than they should.  

The third hurdle relates to organizational issues within both the OEM and the industrial plant:

  • HaaS affects most OEM business processes, and some of the current OEM employees likely lack the skills required for the new model. This includes sales and marketing, manufacturing, operations and finance.
  • A misalignment between the OEM and its industrial customers could slow the adoption of HaaS. A recent study by Emory University and Presenso, focusing on industrial plant O&M employees, indicates overall satisfaction with current Predictive Maintenance practices.  Attempts by OEM to assume responsibility for maintenance may face internal plant resistance.

Towards a Shared Risk Model 

Careful risk management is necessary for OEMs to survive the transition to HaaS. It would be a mistake to overhaul the contractual terms of Long-Term Service Agreements (LTSAs) overnight. OEMs cannot assume all aspects of O&M activities with their current resource capacity. The first step is for OEMs and industrial plants to recognize that their interests are aligned, and that reliability and maintenance programs should be built in collaboration. For instance, preventive maintenance which is typically time-based will likely be performed by the industrial plant. With HaaS, the OEM can provide insights, generated from Machine Learning based monitoring systems, that can be used to recalibrate the scheduling of these planned activities. Similarly, the OEM will need to take contractual responsibility for maintaining environmental standards within the guidelines of OEM standards.Adjusting to a New Industrial Plant

Vendor Ecosystem Development

The support of third parties is necessary to transition from a product to a services model.  It is impractical for OEMs to develop deep expertise in the rapidly evolving discipline of Big Data science. Therefore, OEMs will need to acquire or partner with Machine Learning vendors that can support the scaling of cognitive analytics coverage across the OEM’s industrial assets. Similarly, complementary Industry 4.0 technologies for inspection and maintenance, including drones, 3D printing, virtual and augmented realty should not be built in-house. 

Conclusion

It is too early to predict the exact contours of the post-Industry 4.0 landscape. The OEM that is seeking to navigate this period must recognize the risk involved in either waiting on the sidelines or over-investing in a new model that may not be sustainable. OEMs should approach digital transformation with caution and find ways to share the risk with both customers and partners.

About Your Guest Blogger:

Eitan Vesely is the CEO of Presenso, a Machine Learning Predictive Maintenance solutions provider. He was previously a hardware specialist and a support engineer for Applied Materials, where he specialized in software-hardware-mechanics interfaces and system overview. Mr. Vesely holds a BSc degree in mechanical engineering.

Engage with ARC Advisory Group