Data Mining and Artificial Intelligence for Improved Industrial Energy Efficiency
Utilize your process data to your advantage
Your process data contains valuable knowledge that can be extracted using our data mining techniques. The knowledge gained might identify potential areas for energy savings as well as other process related improvement opportunities
Some of Our Past Clients
As Energy Consultants we are very driven to find new ways our clients can save energy. Over the past decade we have developed cutting edge techniques and innovative solutions that can greatly improve our client’s industrial energy efficiency. We want our clients to start utilizing the data and systems they already have, bringing them closer to “Industry 4.0” in a cost effective and progressive way. Data Mining and Artificial Intelligence combined with our experience as Energy Experts has resulted in numerous successes for our clients to date.
Develop artificial expert systems that contain the knowledge discovered during the data mining exercise along with that of domain experts to help advise operators on running process to optimal conditions
What is Data Mining?
Data mining is the process of finding anomalies, patterns, and correlations within large data sets to identify patterns in data, help to classify data and to predict future outcomes.
Why is Data Mining important?
So why is data mining important? You’ve seen the staggering numbers – the volume of data produced is doubling every two years. Unstructured data alone makes up 90 percent of the digital universe. But more information does not necessarily mean more knowledge.
With the falling cost of process sensors and data acquisition systems, many industrial processes such as that found in the chemical, pharmaceutical, cement and wood sectors collect high volumes of data, sometimes going back several months and even years. This data is often used only for product validation and quality purposes, but in fact, can also be used to uncover valuable insights into process and utility plant operations.
Conventional techniques used to analyze process data can be very time consuming and often require specialized process knowledge as well as statistical and mathematical modeling skills.
By examining your process data using our data mining techniques, we can uncover valuable knowledge and patterns. This will provide insights into how your process variables affect particular outcomes of interest such as energy usage, production rates, and product quality.
It may only take the discovery of one such insight to provide you with significant process cost savings. For example, in one plant the discovery of a link between the fall off in the production rate and humidity due to higher ambient humidity, resulted in the purchased on a dehumidifier and a subsequent ability to increase production by 10% and a reduction in specific energy usage of 20%.
Data Mining allows you to
- Sift through all the chaotic and repetitive noise in your process data to establish important trends, patterns, and rules that can help you save on energy and other operational costs
- Understand what is relevant and then make good use of that information to assess likely process outcomes in the future
- Make better and more informed decisions regarding your process operation – both operational (short-term) and strategic (long term)
The Most Commonly used Techniques in Data Mining
Artificial neural networks are non-linear predictive models that learn through training and resemble biological neural networks in structure.
- Decision Trees: Tree-shaped structures that represent sets of decisions. These decisions generate rules for the classification of a dataset. Specific decision tree methods include Classification and Regression Trees (CART) and Chi Square Automatic Interaction Detection (CHAID)
- Genetic Algorithms: Optimization techniques that use processes such as genetic combination, mutation, and natural selection in a design based on the concepts of evolution
- Nearest Neighbor Method: A technique that classifies each record in a dataset based on a combination of the classes of the k record(s) most similar to it in a historical dataset. Sometimes it is called the k-nearest neighbor technique
- Rule Induction: The extraction of useful if-then rules from data based on statistical significance
What Applications suit these New Energy Saving Tools?
Energy savings data mining has a wide variety of applications in different industries.
General improvements that are typically obtained using data mining techniques in the industrial sector include the following
- Ability to maximize the value of the data collected by your existing SCADA and other process related data acquisition systems
- Identify and repeat best operating practices
- Identify the best performance at least energy usage by optimum control and sequence of equipment
- Provide verification of energy savings before and after process and utility plant improvements
Chemical and Pharmaceutical Sectors
- Finding optimum process set-points for maximizing plant outputs
- Minimizing energy usage in process heating and cooling applications
- Maximizing the performance of reactor vessels for optimal yields
- Improving unit energy efficiency of CHP plants
- Increasing yield of higher value products
- Reducing plant emissions and minimizing fuel usage in incineration plants
- Reducing plant utility costs of compressed air, nitrogen and water costs by optimizing performance, sequencing, and maintenance issue identification
Power Generation Sector
In the operation of the power plant, it is vital that all components of the power generation cycle operate at maximum efficiency and capacity to ensure an economical return of investment. The use of data mining techniques using the large historical data sets that normally exist in these industries can uncover patterns in poor performance that once corrected can save on energy usage and other operational costs.
- Finding problems that reduce overall generation output
- Finding problems that reduce overall generation fuel efficiency including problems associated with the steam boiler, turbine, and other plant
- Reducing power station emissions such as chimney NOx and dust
- Examine process performance potential to reduce energy usage and increase output by uncovering process running conditions that represent best and poor performance
- Improving the energy efficiency of the main process and utility plants
- Increasing production yield
- Reducing plant emissions
- Increased plant throughput
- Reducing plant utility costs of compressed air, nitrogen and water costs
Examples of Our Work
Large Pizza Plant Europe
Data Mining of large data sets containing:
- Energy usage
- Production rate
- Rejects rates
- Refrigeration plant Energy usage
- Ambient Temperature
Results: We identified large energy saving opportunities as well as production bottleneck solution
Data Mining of Large Gas Turbine Power Generation Systems
The exercise collected over one year’s process data consisting of 50 parameters collected every 15 minutes.
These parameters included:
- Ambient air temperature
- 30 different gas turbine parameters
- Filter pressures
- Gas input and power output
- Heat rates
Results: Identification of opportunities to improve overall plant efficiency and generation output worth 3 million dollars per year.
Data Mining of Large Oil Fired Boiler in a Media Factory
The oil boiler was used for supplying steam to an absorption chiller. The boiler process parameters were monitored every hour and the data made available for analysis.
The parameters collected included:
- Steam flow
- Combustion air flow
- Condensate temperature
- Air flow ratios
- Condensate return
- Chilled water inlet and outlet temperatures
- Ambient conditions
Results: The variable that mostly affected the outcome of interest (oil usage and efficiency) were identified Various scenarios for improvements were then modeled using predictive neural networks. Low-cost savings in the order of €100,000 were identified in the running cost of the boiler and chiller with near instant payback periods.
Potential Savings Using Data Mining
Using advanced data mining and intelligent application, typical operational savings between 5% to 10% are possible in the low-cost to medium-cost investment category.
Intelligent Industrial Systems
Artificial Intelligence is a general term used to describe various forms of advanced computational and control techniques associated with the analysis of large data sets. These techniques mimic the natural reasoning process of the human brain, the basis of which is the artificial neural network, a computer simulation program originally designed by computer scientist over 30 years ago and which has advanced considerably ever since.
Because neural networks analyze data more wisely than conventional techniques, they can provide insights into problem domains previously not possible to obtain using existing methods which may require vast computational power and manpower to do so. The net result of utilizing artificial intelligent techniques in industrial environments will be an improvement in the factors that influence complex industrial processes.
By using artificial intelligent techniques it is, therefore, possible to discover for example patterns in a factory production system that may cause production bottleneck, or an increase in process variability resulting in reduced product quality or increase energy usage and cost.
Additionally, data mining can extract patterns and knowledge from historical data to find improvements and discover how various process variables such as temperature, pressure and humidity usage affect an outcome of interest – such as finished product quality, or specific energy usage
Artificial Intelligence gives companies the power to “Prevent” problems not just fix them, to “Analyse” data not just depend on experience. Artificial intelligence computation methods allow you to focus on the “Process” and not just the product, to provide for “Intelligent” localized process control and optimization and not just process monitoring. Therefore by using “Intelligent Data Analysis” and not just seat of pants management an improvement to the overall company direction will be possible.
Artificial Intelligence is the technology that we’ve all been waiting for, just like spreadsheets were to engineers over 30 years ago. What would they do without spreadsheets today! - It’s unthinkable, isn’t it?
Indeed it is important that today’s engineers learn and use these artificial intelligent techniques in their profession if their competitors are not to gain a technological edge.
The falling cost of data has encouraged all of us to collect more data than we can analyze. Imagine if you could extract hidden patterns from data containing 300 variables and 800,000 time-stamped data sets. The valuable knowledge extracted from this data using artificial intelligent techniques may contain the clues to past poor performance of your processes or even better the conditions the allowed for past best performance and without paying a large band of statisticians to do it assuming that you have the budget for their fees for several months.
The Manual Method "Alternative"
If you hate statistics, then neural networks may be for you. When process engineers are faced with complex multi-variable problems they have a number of options:
- Use rule of thumb and manual intervention – but expect that large errors can result
- Wade through a mass of statistical and non-linear programming techniques – if you remember how to do them and hope that the process interactions you’re trying to figure out are all linearly related to one another and there only a few parameters that really matter
- Try Neural Networks
A neural network takes its design from the current understanding of how the nervous system of living things works. It can learn and won’t forget, it can remember and it can adjust to new situations unlike the present “dumb” control systems around today.
The use of Artificial Intelligent techniques in Industry will provide a new era in process improvements that will result in big reductions in process variability and hence product quality. Improving process variability will reduce reject rates, cut costs and increase market share – the key to a successful business.
Benefits of Artificial Intelligence
These are some of the benefits that can be obtained by using artificial intelligence and advanced computational techniques in your business.
- Increase Yield by using data mining techniques to uncover the factors that affect process output that may cause production bottlenecks, poor quality, and a reduction in production output
- Reduce energy usage through process optimization of the set process points, bandwidth, and operational schedules along with intelligent control to produce stable and more responsive processes and associated systems
- Reduced environmental impact by achieving a better understanding of process the parameters that contribute to emissions excursions
- Increase process plant reliability through a better understanding of the factors that affect process plant utilization
- Reduce manpower requirements by using intelligent controls
- Increased plant safety by utilizing the ability of a smarter process plant control system that can utilize various artificial intelligent techniques to learn from experience and adapt itself to ensure that the past events that were detrimental to the process performance are reported before they happen or eliminated
- Improved product quality through a better understanding of the variables that affect quality along with their individual contributions to quality or lack of it.
- Understand the factors that affect machine downtime and overall equipment effectiveness
Examples of Using Artificial Intelligence in Industry
70 of the Fortune top 100 companies use Artificial Intelligence.
Artificial Intelligence or AI is not just for the Ph.D.’s of this life, it’s a technology that everyone can utilize to his or her benefits. Below are a few examples of the type of industries that are using AI to remain on top. Remember that 70 of the Fortune top 100 companies use Artificial Intelligence. They try to keep this a secret – and for a good reason, it is considered a competitive advantage to using artificial intelligent techniques. Here are some examples:
- Power station optimization of power output
- Self Turning PID Loops to improve control system performance
- Expert system to model and validate process data
- Data mining of distillation column operational variables
- Improve control of a refrigeration plant to optimize the compressor COP using data mining of operational variables
- Steam System Optimization (Boilers and Turbines) to continuously optimize air/fuel ratios and hence reduce running cost
- Prediction of product quality using neural networks in glass manufacture
- Using a soft sensor developed by data mining of past process variables to determine the product quality, should a critical variable be changed
- Furnace scheduling in an aluminum rolling mill using expert systems to minimize energy cost
- Optimization routines using genetic algorithms to establish the most economical mix of ingredients in fiberboard manufacture
- Optimize Air Compressor installation to reduce electrical running costs
Summary of the Possible Uses of AI in Industry
- Data Mining for process, product improvements and management information systems
- Intelligent sensors
- Supervisory Control
- Advanced Control
- Monitoring and Diagnostic Analysis
- Process modelling and simulation
- Knowledge-based systems for product improvements and management information systems
Comparison of Advantages Using Artificial Intelligence to Improve Business Performance
|Behavioral Style||Traditional Approach||Artificial Intelligence Approach|
|Problem Solving||Expert Based||Based on Data / Systems|
|Reasoning||Experience Based||Statistically Based|
|Outlook||Short Term||Long Term|
|Decision Making||Intuition||Data Based|
|Direction||Seat of Pants||Benchmarking / Metrics|
|Control||Centralised||Localised with AI|
|Improvements||Dumb Automation||Continuous Optimisation|
Using Artificial Intelligence to become a World Class Manufacturer
Your company’s survival depends upon you continuously growing your business. We all know that growth in business means growth in sales of your products. Growth in product sales is largely determined by customer satisfaction. Customer satisfaction is governed by your product quality, price and delivery times. Quality, price, and delivery are not just controlled by chance but by the ability of your process to produce a high-quality product, free from defects, at the lowest possible cost and cycle times. Only a process that is free from variations will be free from defects.
Simply put, the quality of our products and services are a reflection of how capable our processes really are. To measure product quality is to measure process quality. Process quality is dependent upon the control we have over it.
Artificial Intelligence will help your company reduce process variations and hence reduce defects, reduce operational costs and cycle times. It will help you to identify potential problems, provide intelligent control and meaningful management information when you need it.
Artificial Intelligence allows us to work smarter, not harder. This translates into making fewer mistakes in all that we do. From the way we manufacture our products to the way we organize our delivery. As we “data mine” we discover and eliminate harmful sources of variation, our quality goes up and our defects rates go down, we keep customers and grow our business.
The table below shows a relative means of benchmarking different organization’s processes and systems, based on defects per unit where a unit is any task, product or physical item.
|Process Capability||Defect Per Million Opportunity|
|2||308,500 (Worst Eastern)|
|3||66,800 (Average Eastern)|
|4||6,200 (Average Western)|
|5||2300 (Good Western)|
|6||3.4 (World Class)|