Data Analytics in the Energy Industry

Usually you will be following one or both of these strategies:

• Reduction of emissions to reduce ecological footprint
• Staying within permit limits to avoid costly fines or shutdown.

No matter which way you look at it, it remains a challenge as our graph shows:

(Image © Carl Bozzuto, CIBO Consultant): The turquoise circle indicates the point of maximum efficiency. According to this graph a reduction of air for combustion to decrease NO_x is not a viable solution as it has the highly undesirable side effects of reducing efficiency and sharply increasing CO production. Conversely, using excess combustion air to reduce CO will lead to a drop-in efficiency and rising NOx emissions, also not a possibility. Data Analytics provides the tools to deal with a dilemma of this kind.

Using data analytics on your boiler’s or combustion turbine’s operational and CEMS data will help you get a complete picture with respect to emissions. It will also let you understand if there is an opportunity to improve or maintain emissions without touching the relationship shown in the graph, and normally there is. 

Typical questions around boiler and combustion turbine emissions answered by using data analytics:

• Am I missing any contributing factors to my plant’s emissions, which factors affect emissions positively / negatively?
• What effect does seasonality have on emissions?
• How far from or close to our certificate limits can we operate and what safety margins do we require?

Even in times of low fuel cost an increase in efficiency leads to welcome savings, depending on equipment type and operating modes, 2 – 6 % per year can be attained without any hardware modifications. As boilers and turbines are normally designed to run at optimum efficiency when loads are high it makes special sense to optimize low load operating modes as these are also very common.

(Image © Nate Verhanovitz, Michigan State University): As the graph shows it is possible to consistently run a more than 50 year old boiler, installed in 1965, originally coal-fired, retrofitted to burn natural gas, at an overall efficiency of more than 83 %. The orange dots indicate the most common operating modes. Drilldown capabilities allow analysis of which factors drive those operating modes and are therefore easily available for further analysis and optimization.

Data Analytics will let you see which operational factors (variables) have the biggest positive or negative influence on the efficiency of your operations. This provides your engineers and operators with clear guidance which of these need to be strengthened and which diminished. Multivariate analysis will also let you see cause-and-effect relationships you might not have been aware of. This promotes operator training and general knowledge about the plant’s operations.

Using monitoring and control based on the multivariate models that you produce during the analysis of your data ensures that your plant runs under optimal conditions at all times. Should deviations occur our control software will spot them way before they become noticeable to the human eye so that deviations can be corrected almost immediately.

Typical questions around boiler and turbine efficiency and cost savings answered by using data analytics:

• What are the factors/variables that have the strongest influence on the plant’s operations with respect to efficiency?
• How do these influences depend on the operating mode or load level the plant is running in?
• How can I make use of those to always run as efficient as possible?
• Are there causal relationships between factors I am not seeing that affect operations?

Operational upsets and unplanned downtime are big worries of the industry. Of course, maintenance schedules help a lot and are required by insurances or industry standards. An annual outage, most likely during a low load period is state of the art. But what to in between those periods, are there any possibilities to avoid any negative surprises through failing equipment? Analysis and monitoring of operations with multivariate data analytics methods is the solution here.

Equipment lifetime is a topic that ties in closely with maintenance. Can one take things one step further and estimate the end of service life for a piece of equipment in addition to maintenance requirements? Knowing when a capital investment will be required will help a lot with financial planning. Our graph shows that this is possible.

Simulated lifetime prediction of jet turbines. Right: prediction at 100 % lifetime (end of life), left: prediction at 60% of lifetime. At 60% of the known equipment lifetime from historical data already good estimations can be made, at short and medium lifetimes there is a very good fit at long lifetimes there is some scatter but the model rarely overestimates here.  So, historical condition data helps to judge the life expectancy of equipment.

Typical questions around boiler and turbine maintenance and lifetime prediction answered by using data analytics:

• Is full condition-based maintenance possible or what other strategies are there to optimize plant uptime?
• Which factors / variables relate to the condition of equipment, are we recording everything required for the purpose?
• How can I predict the remaining lifetime of my equipment?
• Are there possibilities to prolong the lifetime of my equipment?