PPCL Logo Process Plant Computing Limited

Tel: +44 1753 893090 |

LinkedIn Twitter



Register below for one of our live webinars that share new approaches to process improvement & optimisation. 


Or watch one of our Pre-Recorded Webinars here...


Avoiding Downtime through Event Forensics and Prediction

 

Value lost to downtime and degraded product quality during abnormal events and designed overcapacity to compensate are among the largest avoidable costs in any process plant. Process events come in the form of disturbances, faults, trips and excursions that cause downtime and lost production. Operational excellence requires reducing the frequency and impact of these events.

This webinar explores PPCL’s novel Geometric Process Control (GPC) technology for understanding the course and causes of these events and generating real-time operator alerts for early warning of developing future events.

The start is investigating events using historical data from the plant historian through C Visual Explorer (CVE). Building from individual events to discovering similarities between events, CVE allows investigating significantly larger datasets than most engineering applications, hundreds of variables across thousands of time points. This power lets us quickly see and explore data far beyond what we’d typically use.

PPCL’s online process monitoring tool, C Process Modeller (CPM), goes beyond traditional alerts by implicitly including the relationship between process variables and providing a sensitive detector for changes. By excluding events and event precursors from our model of normal operation, CPM provides a powerful low-cost method of building event prediction models that have been shown to provide hours or days of advance warning of process changes, giving plant personnel more time to react and mitigate the effects of disturbances and faults. We’ll look at examples including a gas turbine-driven generator and surge in a large propylene refrigeration compressor. CPM models can be created in just a few hours, bringing the benefits of condition monitoring to applications where it isn't currently economical. Process availability and efficiency will improve and facilitate the move toward predictive maintenance and operational excellence.

RESERVE A SPOT:







Can't make it? Click here to access our previously recorded webinars


Reducing Operator Workload with Better Alarms

 

The role of operator alarms is to alert the operator of abnormal process operation as early as possible, giving the most time to react, address the developing issue and keep producing product while the plant is in a degraded state, or at worst shut the unit down in the most sympathetic way. Good alarms produce the economic benefits of higher plant availability and more saleable product, while reducing operator load, giving operators the ability to react to abnormal situations.

Achieving these benefits requires well-designed and properly positioned alarms as well as a healthy process. Many organizations have reviewed these alarm limits repeatedly through rationalization exercises, never getting the real benefits they know are possible.

C Visual Explorer (CVE), PPCL's flagship software product, provides a graphical framework for bringing the entire process history into the context of alarm rationalization. CVE lets engineers, operators and all involved see directly the full richness of operation and the process envelope, relating this immediately to current and potential alarm limits. The range of conditions a process experiences, including variations in feed, ambient temperature, changing demand, different goals and normal operation variability are included, linking operations through from incoming material to product quality and economic KPIs. It has never been easier to explore years of process history for thousands of variables for rationalization and other applications.

In this webinar, we demonstrate how to take full advantage of this display and data, leveraging the parallel plot to give easy access to historic data in the context of desired plant operation, alarm and process limits for hundreds of variables. Bringing historical operation into the alarm process allows the immediate evaluation of alarm limits, taking less time and enabling continuous alarm review, and giving better limits that have been pre-tested as well as vital process understanding. We give examples on real systems and discuss some of the successes our users have had.


GPC Technology for Big Data and Predictive Analytics in Process Plants



The Problem: Process plants generate continuous data for thousands of variables at sub-minute frequencies. This is far beyond what process engineers can analyse with their conventional analysis tools. This data has enormous value, containing the records of plant operation and implicitly the relationships between process variables and quality variables. But only a tiny fraction of this data is used today. Many may consider the “Big Data” methods that are the current buzz, but these approaches are challenged by process plant data.

Big Data techniques focus on pulling subtle correlations from largely uncorrelated data, but chemical processes have extensive relationships due to balances and governing physical laws. Predictive Analytics provides generalized answers through simplifications; choosing a small subset of variables, processing or averaging data, and ignoring the fundamental complexities such as nonlinearity of the process. This can destroy much of the richness of the data and reinforce preconceptions. It can also be time-consuming and require a statistician to interpret the results.

The Solution: Geometric Process Control (GPC) is a visual technology: quick and easy to understand and implement by anyone familiar with the process. Hundreds of variables can be seen simultaneously on one graph to gain an overall understanding of your operation, target investigation, test hypotheses and quickly identify improvements.

PPCL presents a Geometric Process Control alternative to reach the goals sought from Big Data and Predictive Analytics that shows how to operate the process so that objectives are met, keeping their values at the desired targets.

In the webinar, delivered in May 2017, Dr Alan Mahoney, PPCL's Technical Director, demonstrated GPC using graphical tools to screen for key variables and eliminate the effects of uncontrolled variables.  He showed how to approach big datasets and explore them visually using the parallel plot, a unique graphical technique that puts axes parallel to each other rather than perpendicular, allowing the exploration of interactions between variables with orders of magnitude fewer graphs. By connecting historical data completely across the process from incoming conditions and initial processing conditions to final quality variables, KPIs and performance with the richness of years of data in a technique that can examine hundreds of variables, the parallel plot enables discoveries and exploration that are not possible with today’s techniques. He also demonstrated online GPC models for achieving quality targets and operational excellence.


In Pursuit of Operational Excellence

Perhaps once or twice in a person’s career something comes along that is not just an improvement on what was there before. Starting from different principles, it makes possible what was previously unachievable or even unimaginable and, in time, establishes new levels of operational excellence. Many engineers in the 1960s could not imagine that computers would ever be of use in operating their plants, while younger engineers today have difficulty understanding how a plant could be run without them. Fax machines and photocopiers in the 1980s replaced the telex, typewriters, carbon-paper and internal-mail carts of the 1970s and in turn were replaced by word processing and email in the 1990s. The telephones of the 1980s and face-to-face trade-shows and meetings of the 1990s are even now are being replaced by video conferencing and instant messaging.

All these massive changes came about without great drama through individual engineers recognising easier and better ways to do things and creating pressure on their organisations to provide the necessary facilities. And the process of change continues.

PPCL has developed Geometric Process Control (GPC) with its 1,000-variable graph and the first-ever general method of modelling the operating envelope of a process as a multi-dimensional geometric object. The graph brings remarkable analytical, optimisation and monitoring power to the engineer without requiring more than high-school mathematics, and leads to the reduction of operational variability which is at the heart of operational excellence.

Operating envelope models are used for fault detection and prediction, predictive alarming, operator guidance, compliance with KPI’s, target-setting, SPC replacement or “just” for helping the process operator to stay inside many sets of operating limits and KPI limits simultaneously. The appeal of these models for plant use is that they are equation-less, requiring no maths but only process knowledge from the user. This makes them very cheap in man-hours to develop and maintain.

Before being exposed to Geometric Process Control, many have struggled to understand how a mathematical model could possibly be “equation-less” or why a 1,000 variable graph has such a big advantage over the traditional methods of Predictive Analytics. Seeing is understanding with GPC, and many have had a Eureka moment in our demonstrations and webinars as the accuracy of our descriptions and the potential of GPC strike home. PPCL is working in industries across the world to improve the business efficiency of process plants. Why not see what we could do for you?



Event Forensics and Prediction 

 


Unscheduled downtime is one of the biggest threats to any process plant. Value lost to downtime and degraded product quality during abnormal events and the necessity of designed overcapacity is one of the largest preventable costs. Process events come in the form of disturbances, trips, and excursions that cause downtime and lost production. Achieving operational excellence requires reducing the frequency and impact of these events.

This webinar explores PPCL’s novel Geometric Process Control (GPC) technology for understanding the course and causes of these events and generating real-time operator alerts for early warning of future events.

The start is investigating events using historical data from the plant historian through C Visual Explorer (CVE). Building from individual events to discovering similarities between events, CVE allows investigating significantly larger datasets than most engineering applications, hundreds of variables across thousands of time points. This power lets us quickly see and explore data far beyond what we’d typically use.

PPCL’s online process monitoring tool, C Process Modeller (CPM), goes beyond traditional alerts by implicitly including the relationship between process variables and providing a sensitive detector for changes. By excluding events and event precursors from our model of normal operation, CPM provides a powerful low-cost method of building event prediction models that have been shown to provide hours or days of advance warning of process changes, giving plant personnel more time to react and mitigate the effects of disturbances and faults. We look at examples including a gas turbine-driven generator and surge in a large propylene refrigeration compressor. CPM models can be created in just a few hours, bringing the benefits of condition monitoring to applications where it isn't currently economical. Process availability and efficiency will improve and facilitate the move toward predictive maintenance and operational excellence.


Visualizing the Process Operating Envelope and Revolutionizing Alarm Rationalization

 

Operator alarms should sit at the boundary of normal operation, alerting operators as soon as possible to developing situations that will affect process integrity, product quality and time online. The operator can then take the best action to reducing the disturbance impact, continuing operation with minimal degradation while the cause is properly addressed.

Achieving these benefits requires alarms properly positioned. At many installations these same limits have been reviewed repeatedly for years through repeated rationalization exercises, never getting the real improvement desired.

C Visual Explorer (CVE) provides the framework to take full advantage of process history in the context of alarms, letting engineers, operators and all involved to directly see the full richness or operation, the process envelope, and how this relates to current and proposed alarm limits. The full richness of operation that a process experiences with variations in feed, ambient temperature, changing demand, different goals, and normal operation variation are all there, linked all the way through from incoming material specifications through the process to qualities and economic KPIs. It has never been easier to explore years of process history for thousands of variables for rationalization or any other application.

In this webinar we demonstrate how to take full advantage of this display and data, leveraging the parallel plot to give easy access to historic data in the context of desired plant operation, alarm and process limits for hundreds of variables. Bringing historical operation into the alarm process allows the immediate evaluation of alarm limits, taking less time and enabling continuous alarm review, giving better limits that have been pre-tested as well as vital process understanding. We give examples on real systems and discuss some of the successes our users have had.



Event Forensics and Prediction 

 


The cost of abnormal process events has a major impact on the running of every process plant. Lost production due to downtime, urgent maintenance, degraded product during events and the necessity of planned overcapacity to deal with these events is one of the largest preventable costs of doing business. These events come in the form of disturbances, process events, trips, and other events that cause downtime and lost production. A large part of achieving operational excellence is reducing the frequency and impact of these events.

As a number of causes are responsible for these events, a broad approach is required for investigating, generating understanding, and mitigating the effect of future events. In this webinar Dr Alan Mahoney, PPCL's Technical and Operations Director, looks at using PPCL’s novel Geometric Process Control (GPC) technology for three applications of this strategy.

The start is investigating events using historical data from the plant historian through C Visual Explorer (CVE). We look at both individual events and expand to multiple similar events, demonstrating how to find commonality of cause and potential precursor signatures. CVE enables the consideration of significantly larger datasets than most engineers are used to, up to hundreds of variables at hundreds of thousands of time points. This lets us powerfully and quickly consider data far beyond that which we’d typically use.

Dr Mahoney then turns to PPCL’s online process monitoring tool, C Process Modeller (CPM). CPM goes beyond traditional alarming by recognizing the relationship between process variables and providing a sensitive detector for when this changes. By excluding events and event precursors from our model of normal operation, CPM provides a powerful, low-cost method of building event prediction models that have been shown in some instances, to provide hours or days of advance warning of changing process conditions, giving plant personnel much more time to react and mitigate the effects of disturbances and faults. We look at examples including a gas turbine-driven generator and compressor surge in a large propylene refrigeration compressor.

Operating models can be created, with experience, in just a few hours, bringing the benefits of condition monitoring to applications where it isn't currently economical. Process availability and efficiency will improve and facilitate the move toward predictive maintenance and operational excellence.


Get your Alarm Limits Right the First Time!

 

Operator alarms should sit on the boundary of normal operation, alerting operators as soon as possible to developing process disturbances, issues impacting quality and situations that will result in significant downtime if not addressed. The operator can then take action to minimise the process disturbance, keeping everything running as economically as possible while the situation or fault is permanently addressed.

To achieve this ideal situation, alarms must be properly positioned. The personnel at many plants visit and review the same limits repeatedly for years through repeated rationalisation exercises, never getting the real improvement they want.

PPCL’s award-winning software C Visual Explorer (CVE) provides the first framework for taking advantage of process history in the context of alarm rationalisation, allowing the full visualisation of the richness of operation that a process experiences with variations in feed, ambient temperature, changing demand, different goals, and normal operation variability. Exploring years of process history for thousands of variables has become efficient and straightforward even in the context of a rationalisation meeting.

In this webinar Dr Alan Mahoney, PPCL's Technical Director, demonstrated this new method, which leverages the parallel plot to give easy access to historic data in the context of desired plant operation as well as alarm and process limits for hundreds of variables at once. Bringing process history into the process allows alarm limits to be evaluated immediately as well as allowing continuous alarm review, finally breaking the cycle of periodic alarm rationalisation. He gave examples on real systems and discussed some of the successes our users have had.


Big Data & Predictive Analytics - GPC is a Better Solution for Process Improvement and Monitoring





The Problem:
Modern process plants are highly instrumented with digital control systems, and recording this data is inexpensive. Process plant data - thousands of points for decades at sub-minute frequencies - dwarfs conventional analysis and graphs. Data of this scale has enormous value as it contains the records of plant operation and physical relationships, yet only a tiny fraction of this data is used today, or indeed can be used with traditional methods. With the advent of “Big Data” methods, many are considering the application of these techniques to process data. However, Big Data approaches that have produced results in other fields are challenged by process plant data.

Big Data techniques are targeted at identifying small correlations among largely uncorrelated data, but chemical processes have extensive intrinsic correlations due to balances and governing physical laws. Predictive analytics provides generalized answers through simplifications; choosing a small subset of variables, processing or averaging data, and ignoring the fundamental complexities such as nonlinearity of the process. This can destroy much of the richness of the data and reinforce preconceptions. It can also be time-consuming and require a statistician to interpret the results.

The Solution: Geometric Process Control (GPC) is a visual technology: quick and easy to understand and implement by anyone familiar with the process. One thousand or more variables can be seen simultaneously on one graph to gain an overall understanding of your operation, target investigation, test hypotheses and quickly identify improvements.

PPCL presents a Geometric Process Control alternative to reach the goals normally desired through Predictive Analytics that shows how to operate the process so that the quality (and even other) objectives are met, not just predicting, but keeping their values at the desired targets.

In this webinar Dr Alan Mahoney, PPCL's Technical Director, demonstrated GPC using graphical tools to screen for key variables and eliminate the effects of uncontrolled variables.  He showed how to approach big datasets and explore them visually using the parallel plot, a unique graphical technique that puts axes parallel to each other rather than perpendicular, allowing the exploration of interactions between variables with far fewer graphs. By connecting historical data completely across the process from incoming conditions and initial processing conditions to final quality variables, KPIs and performance with the richness of years-worth of data in a technique that can examine hundreds or thousands of variables simultaneously, the parallel plot enables discoveries and exploration that would not be possible with today’s techniques.

See for yourself why GPC models cost so little and get your process working perfectly more of the time! 


 


PPCL © Copyright 2011-2017. All Rights Reserved. | Terms of use | Privacy Statement | P.O. Box 43, Gerrards Cross, Buckinghamshire, SL9 8UX. UK.