Posted: 03 Nov. 2020

How AIoT and modern IoT platforms can optimize industrial asset maintenance

Authors: Leonard Michalas and Dejan Boberic

Using Artificial Intelligence of Things (AIoT) to retrieve sensor data and PTC Thingworx to predict faults: a predictive maintenance use case at the Deloitte Digital Factory.  

To showcase the possibilities provided by Artificial Intelligence of Things (AIoT), Edge Connectivity, and the development on basis of IoT platforms like PTC Thingworx, a predictive maintenance use case was set up around an electric compressor at the Deloitte Digital Factory.

Traditionally manufacturers have been using a schedule-based approach to determine when an asset on the shop floor requires maintenance. The older an asset is, the more maintenance activities have been planned and performed for the asset. However, studies show that only 18% of assets fail due to age. 82% of failures occur randomly or due to other factors. This shows that the schedule-based approach is inefficient since a large number of assets are maintained unrelated to actual demand.

This inefficiency can be significantly reduced by using IoT and Data Science to identify potential defects before they actually occur and only then plan and execute maintenance activities. This approach of planning maintenance orders is also commonly known under the term Predictive Maintenance.

AIoT & Edge Connectivity

Recent developments in sensor technology and microchips have achieved high performance and rich features at a low price point for IoT edge devices. Devices cheaper than 10 EUR are capable of executing complex algorithms for signal processing and machine learning in real-time. For this case, we decided to retrofit our compressor with a set of non-intrusive sensors, i.e. sensors that do not require integrity-compromising mounting or installation on the asset. As a result of this approach, it has to be considered that sensor values may be less meaningful or require more complex signal processing to achieve the same significance than more intrusive approaches. Those, on the other hand, can void existing regulatory approvals and require expensive and time-consuming re-approval processes.

Our goal was to create a Digital Twin of our asset, an electric compressor, without interfering with the integrity of the setup. We chose to install three temperature sensors, one microphone, and two cameras that allow us to digitize the values of the analog gauges. This AIoT setup enables us to create a Digital Twin of the compressor with the following properties:

·       Temperature in °C, at three critical measuring points 1-3

·       Sound level in dB

·       Pressure in bar, at inflow and outflow of tanks

The data is collected and pre-processed in state-of-the-art edge devices that are capable to connect securely and reliably to most modern IoT platforms, in this case, PTC Thingworx.


IoT platforms

A lack of interface and communication standards leads to a gap between IoT edge devices and business applications. An IoT platform closes this gap by acting as a middleware, which mediates between the two ends. However, modern IoT platforms go one step further by adding functionality to the hardware and application layer. For this reason, capabilities for edge data processing or complex data analytic algorithms can also belong to the functionality set of an IoT platform.

The market for potentially usable IoT platforms is large and selecting the right one for a company or application is not always easy. However, in this case, the IoT platform PTC Thingworx was chosen because of its lightweight management of IoT devices and extensive connectivity options. An in-depth comparison of different IoT platforms and a decision matrix for deciding when which IoT platform fits best for your company’s needs is part of Deloitte's IoT platform offering.

Data sent from our edge device is processed and persisted in PTC Thingworx. The Thingworx Analytics module is used to determine if the incoming data indicates an upcoming asset failure.

If the asset is about to fail a message is sent to Deloitte's Cloud4M module which is responsible for creating a maintenance ticket and providing data to an AR application, which supports the service technician. Cloud4M is Deloitte’s accelerator to extend Salesforce’s core cloud components with a manufacturing-ready data model, useful AppExchange integrations, and a predefined integration framework.

The AR application visualizes the Digital Twin of our asset as a 3D model and provides the service technician with information about the service case, the concrete maintenance task, as well as step-by-step instructions on how to resolve the issue, e.g. cleaning or changing a filter. In case of persisting issues live remote support can be requested and offsite experts can provide additional remote guided support.

If you would like to explore this solution and get further insights into how IoT improves the efficiency of your asset management approach, please reach out to us.


Andreas Staffen

Andreas Staffen

Partner | Technology Strategy & Transformation

Andreas Staffen verantwortet das Offering IoT and IT Architecture (Smart Manufacturing) für Deutschland und gestaltet die Digitalisierung der Supply Chain seit 2004. Dabei begleitet er deutsche, europäische und globale Unternehmen bei der erfolgreichen Umsetzung schlanker und integrierter IT Architekturen für die Entwicklung und Produktion. Durch die Umsetzung des Industrie 4.0 Gedanken in der Deloitte Digital Factory werden die Auswirkungen auf die Geschäftsmodelle unserer Kunden erlebbar und die weitere Gestaltung einfacher realisierbar.