Skip to Content
Author's profile photo Johannes Papst

How to make the best use of sensor data built into machinery and devices beyond simply monitoring threshold values

I would like to share a story I got from  Nick Deacon:

Just imagine the following scenario where a customer calls a machinery or equipment company:

Customer:  “Why did you deliver this new pump component?”

Response:  “It is a replacement pump for machinery xyz installed in your plant.”

Customer:  “But there is no problem with the pump on that machine!”

Response: “Well…  there will be a problem if you don’t replace it very soon, as the current pump only has 40 more hours left before failure…”

Of course such a scenario requires much more than just reading sensor data (e.g. oil temperature) and have alarms set to trigger when certain thresholds or limits are hit.  As the proliferation of sensors on industrial equipment generates vast quantities of data, a very sophisticated analysis is now required to detect when maintenance issues are developing and where serious problems are likely to occur.

SAP offers this far more sophisticated type of analysis.  SAP has data management and analytic products which can analyse sensor and other industrial data in real time.  These innovative technology solutions can even be deployed remotely to analyse equipment running 24×7.

One aspect of this analysis is to detect where the rate of change for one sensor starts to deviate from the correlated norm for another set of sensors or where the values deviate from historical patterns.  To illustrate the concept without getting too complex, please take a look at the very simple example below, where a pump bearing oil temperature and pressure starts to deviate from the norm for the given RPM.

Delta A indicates a slight rise in oil temperature without a change in RPM.  On its own this data point may not be important, but when you add in Delta B with points out a corresponding slight decrease in pressure, this combination of factors may be a strong signal for a maintenance request or a warning of an even more severe issue.  This type of analysis requires much more sophistication than standard sensor reading, as it is done over a rolling time window rather than at a single point in time. 

Pump Sensor Metrics Example.png

Making use of such advanced analysis can lead to the following benefits:

  • Issues can be identified earlier, saving maintenance and downtime costs
  • Severity can be analyzed automatically to determine if the situation at hand will lead to a critical failure or is it a less urgent or general maintenance issue
  • Spare parts can be shipped and maintenance personnel can be dispatched proactively to the location where a machine needs to be serviced
  • Value-added spare parts and high-margin services can be up-sold and cross-sold to customers based on perpetual machine data
  • Manufacturers can maintain large databases of sensor data for analysis of products across their entire customer base, thus improving predictive analysis capabilities
  • Engineering, Construction, Maintenance, Repair, and Operations initiatives can be promoted
  • Remote monitoring anytime, anywhere when coupled with SAP mobile products

SAP has the following products in our Database and Technology Portfolio on our Real Time Data Platform:

  • SQL Anywhere – Delivers a small scale mobile database,  ideal for capturing local data and synchronizing back to a data center when network communications are available
  • ESP  Event Stream Processor – Analyzes and correlates streaming data from multiple sources in true real-time and can be integrated with HANA
  • SAP HANA – Provides in-depth predictive analysis capabilities
  • Sybase IQ – Supplies mass storage for extreme quantities of data

SAP provide the extra horsepower necessary to absorb, analyze, and  make the best possible use of the tremendous amount of sensor data built into industrial machinery, equipment, and devices.

Assigned Tags

      You must be Logged on to comment or reply to a post.
      Author's profile photo James Rickard
      James Rickard

      Johannes, thank you for an interesting article. Would it be possible to expand on the architecture of the solution that you proposed, this appears to be a list of technology that can assist in the total solution but is missing a view on the process?

      The list of technologies is interesting but does not completely support an end to end process. For example none of these technologies on it's own can raise the purchase order for the replacement part, how is the integration to plant maintenance/quality management achieved. Is it proposed that the pump is connected to ERP plant maintenance through MII for example or is it recording heartbeats synchronized to a HANA db via SQL anywhere?

      What is the mechanism for predictive analysis to raise the prospect of the failure on the back end system?


      Author's profile photo Johannes Papst
      Johannes Papst
      Blog Post Author


      Thanks for your feedback and thoughts!  Your comments are exactly on point. 

      The main idea of this post is to share how SAP can help handle an environment where technical data is used.  And yes, absolutely, one idea could be to use the combination of HANA + SQL Anywhere ,  we can use SQL Anywhere to locally store the data at the equipment source and sync back to  HANA in the datacentre when the equipment has  IP connectivity, e.g. WIFI or 3G, and joins the local WIFI network.   An alternative would be to use HANA + ESP (Event Stream Processor), ESP can be used to capture the data at source and provide initial analysis and then stream the data to HANA for advanced analytics. This avoids transmitting large quantities of data continuously and at the same time provides instant predictive analytical capabilities. In this instance a normal maintenance scenario can be recorded and processed  later whereas a serious equipment situation can be handled immediately.  The example in my blog shows how HANA offers capabilities to compute usage data above and beyond the typical averages, maximums, minimums, etc.  Thanks again for your excellent feedback.