Preventing disruptions 
with PYTHIA Prevention

Whether a disruptive event occurs infrequently or regularly, whether it is a material-related production interruption or wear-related machine maintenance – the key is to understand, predict, and ultimately prevent disruptive events. By doing so, productivity can be increased, breakdowns can be avoided, and quality can be assured.

Preventing disruptions

Production processes are hard to control, especially complex ones with thousands of parameters. With PYTHIA Prevention, you can achieve that in great speed and quality.

Despite the vast amount of data, process disruptions occur that are hard to understand and even harder to prevent. Those disruptions damage productivity in two ways: To clean up and restart can already be very cost-intensive. Furthermore, running with reduced speed to have an additional safety margin reduces the outcome.

In many cases, predicting those disruptions is theoretically possible. However, due to the sheer complexity and amount of data, it is almost impossible to create a solution that continually monitors the data streams and efficiently warns about disruptions, until now.

PYTHIA Prevention automatically analyses the historical data of the process and creates a prediction model for disruptions. Combined with your domain knowledge, PYTHIA Prevention recommends the right counter-measure at the right time to avoid disruptive events.

PYTHIA Prevention helps to avoid disruptive events. Combining cutting-edge automated data analytics and forecasting with your domain expertise creates an unbeatable solution for preventing disruption.

How it works

First, you provide the unfiltered, raw history data of your processes (can be up to 1 TB). An in advance exclusion of individual signals or parts of the data is neither necessary nor recommended. You provide a rule, how the disruptive target event can be identified within the data, e.g., ‘sensor_318 is zero’.

Next, PYTHIA performs an automated root cause analysis and searches for the different process states that lead to the defined disruption (classification). As a result, for each class, the root cause indicators are listed by their relevance. Also, a prediction model is created, which predicts the occurrence of these classes.

Now, your domain knowledge is included: For every class, your experts decide which counter-measures to take to bring the process back into a safe condition. As the last step, the live detection is implemented into your live process data streams to be able to perform a live detection and prediction.

PYTHIA Prevention now warns the operator of the imminent occurrence of a disruption and recommends the right counter-measure to prevent it in time.

Applications

predictive maintenance – predictive intervention – quality Assurance – prescriptive maintenance
Paper making – Preventing paper breaks

Process interruptions in the continuous process of industrial paper manufacturing are called ‘sheet breaks’. These breaks can occur at any time without any warning. The sheet or paper web breaks inside the machine. Subsequently, there are various cleaning and retreading steps to take before production can be continued. Sheet breaks have a significant negative impact on a paper machine’s productivity.

With PYTHIA Prevention, those breaks can be continually predicted and, in many cases, prevented.

Turbines – Preventing vibration crisis

If the vibrations of a turbine axis exceed a certain maximum level, a vibration crisis occurs. If uncontained, such an event could lead to a major breakdown.

With the PYTHIA Prevention technology, these disruptive events can be predicted in advance, and the operator is being told in time which counter-measures to take to reduce or even avoid the vibration crisis.

Mining – Predicting impurities during flotation

The best possible quality is a fundamental requirement in the production of steel plates because material defects can quickly lead to significant problems. A decisive factor is to prevent the occurrence of surface cracks that can occur during the hot rolling process. Many different factors influence the appearance of cracks. Therefore, process monitoring is of central importance, which takes the right counter-measures at the right time.

PYTHIA Prevention automatically analysis and classifies the different reasons for surface cracks and helps to prevent them by recommending the right counter-measures at the right time.

Wind farms – Predictive maintenance

The maintenance of wind turbines is a great challenge. Often the turbines are located in remote places that are difficult to access. This brings two significant disadvantages: If maintenance is performed like replacing a part, it is costly and should never be done without necessity. However, if a wind turbine fails, this often causes downtimes of a week or longer before the repair can be done. Therefore, it is crucial for the profitability of wind parks to know as well as possible when which turbine needs which kind of maintenance.

With PYTHIA Prevention, a model can be easily trained for each wind turbine predicting its current status and when it needs maintenance.

General – Predictive maintenance

In almost every production environment, wear and tear is a crucial challenge. Because if a module fails during productive operation, the damage is often much more expensive than the timely maintenance of this module. However, to know when a module needs to be serviced, a prediction model is required in order to determine the current module status based on a wide range of parameters and predict when the module will need to be serviced.

PYTHIA Prevention does exactly that enabling to switch from corrective maintenance to predictive maintenance.

These are only a few examples of PYTHIA Prevention applications. In fact, the potential industrial applications are countless.

General applicability

PYTHIA Prevention can be used in many industrial applications. If you are wondering whether it can be used in your particular case as well, focus on the following criteria:

  • There exists an disruptive event, that occurs frequently or regularly
  • There exists no obvious simple solution to live predict and prevent the disruption
  • There exists a system for live process data acquisition including history data that can be accessed
  • The acquired data contains enough information to predict the disruption (this can be checked in a pre-analysis)

If your process environment matches those criteria, there is a good chance that PYTHIA Prevention can be applied to your use case.

Why PYTHIA? Find it out!