Blessing or curse?
Horia Orenstein (SAS) on the opportunities of sensor-generated data in the process industry
A smartphone detecting your movements, a bracelet measuring your heart rate or blood pressure – if these devices are able to capture information, it’s thanks to the in-built sensors generating the necessary data. The so-called Internet of Things is booming, but to what extent is the industry ready for its multiple implications? Horia Orenstein, Business Development Manager Oil & Gas at SAS, clears a path through the jungle of sensors and smart meters.
Mr Orenstein, how do manufacturers see the tsunami of sensor-generated data?
Horia Orenstein: “Let me start by saying that my experience with sensor-generated data mainly stems from the petrochemical industry. In such a process-driven environment, sensors and meters are both a blessing and a curse: a blessing because they give a 360° view of the process at any one time, and a curse because they’re expensive and – more importantly – there are not enough means to transform the data into actionable information.”
How can data analytics help here?
Horia Orenstein: “When you consider a process, it’s of the utmost importance to be able to detect and even predict possible deviations and process disruptions, such as trips. And it’s just as crucial to understand the main contributors to the deviations leading to critical disruptions! You would need some kind of smart meter or sensor that can ‘look’ inside the processing units. You have to constantly monitor these parameters and predict the risk of a failure. That’s where we apply analytics. A more complete set of measurable parameters helps to ensure better measurement of process efficiency in real time, in both technical and economic terms, and also to forecast the life expectancy of equipment, processing units and entire plants. One key point here is that when deviations from expected values occur, the most significant contributors are reported together with the deviations, so that operators, engineers and managers understand the problem instantaneously and can take appropriate corrective action.”
What can SAS do to make this process more predictable?
Horia Orenstein: “We give our clients the ability to control this process end to end and predict any non-standard modes of operation. However, predicting a failure that will occur within the next few minutes is a trivial matter. Using analytics, we enable our clients to detect an upcoming failure in plenty of time to rectify it. Unlike current practices, where anomalies are overlooked.”
Why is this so important?
Horia Orenstein: “Many critical anomalies lead to off-spec quality product, reduced output, poor process efficiency and even worse… production stoppages and potential repairs demand investment. Last but by no means least, it’s important to flexibly control effective transition between the necessary modes of operation defined above. The overall result is not only predictability of technical performance, but also significantly improved economic efficiency and safety. And it works: thanks to more predictable processes, ConocoPhillips has managed to reduce the number of unplanned shutdowns by 80% and increase process efficiency by 20%; what’s more, no accidents have been recorded. We also see the same kind of figures with industrial clients operating with continuous processes.”
You have more than 19 years’ experience in data analytics. Can you share some best practices with us?
Horia Orenstein: “Certainly. My first piece of advice would be to draw up an objective and result-oriented roadmap. You should also get management buy-in for deploying staff and creating an internal strategic initiative for change. Thirdly, the innovation project is not an experiment outside normal operations; it’s the transformation of daily operations. Plan not only for implementation but also for operationalisation, i.e. handover from project to operations and shadow operations until the new practice becomes standard. And finally, collaboration between the client’s staff (management, engineers and operators) and the analysts (SAS analysts in this case) is mandatory for success.”
Want to make the most of sensor-generated data? Here are five recommendations from Horia Orenstein:
1. Start by transforming the available data into meaningful information. Make use of the available data in your manufacturing execution system (MES) and other systems to detect plant-wide anomalies, because 70% of disruptions and incidents are the result of unknown anomalies.
2. Identify the sources of inefficiency and the safety hazards. What hidden potential is available in your plant and processes? What economic effect does this have? This process, as well as the previous one of detecting chains of anomalies, is converted by SAS into predictive models.
3. Develop a new set of control procedures triggered by prediction. Organise your operations around these new procedures.
4. Forecast process conditions and constraints more accurately and adjust operations to these scenarios with new modes of operation.
What’s our take on this?
“Although the use of sensors in controlling processes is not new, the advancement in sensor technology has opened new and interesting frontiers”, professor Behzad Samii comments. “For example, smart electricity meters provide real time access to the consumption data. This data, however, has no added value for the service company unless it is used to segment consumers and trigger differentiated service. Preventive data analytics can help grid managers to avoid black-outs and other major service disruptions. Beyond that, dynamic pricing for example, can provide healthier financial returns, which, in turn, challenge service providers to diversify their supply base.”
Read the interview with Professor Behzad Samii on the smart meter project in partnership with Eandis
In July 2013 SAS conducted a survey among the senior readership of IndustryWeek magazine in order to find out to what extent they use and develop product-embedded sensors. The main findings are compiled in the Sensor Data and Business Analytics Survey