The supply pump is one of the central components of a CIP plant. It ensures media circulation through the entire system. If the supply pump fails, the entire plant process will come to a halt.
As the pump is operated at various speeds depending on the cleaning step, static monitoring is difficult.
The aim is to implement AI (Artificial Intelligence) based pump monitoring in order to detect and report any deviations of the operating state in time. Maintenance works can be carried out in real time.
Process monitoring of the pump, including vibration and speed monitoring, is not performed, or only to a very limited extent– at best for static process variables. No monitoring or visualisation systems such as moneo are installed. Accordingly, arising damage to the pump is not signalled in time.
Using suitable IO-Link sensors, the critical process variables of the pump are to be recorded.
The data recording of the normal condition is used to create a model that enables monitoring irrespective of the operating status and thus identification of deviations from the normal condition (anomalies).
moneo|RTM, including the DataScienceToolbox and the SmartLimitWatcher function, is installed centrally on a server. The IO-Link masters are connected to the server via an internal VLAN. The sensors used are each connected to an IO-Link master.
moneo|RTM records and visualises the data. The SmartLimitWatcher function of the DataScienceToolbox is used to analyse the recorded data and calculate a corresponding model. After a teach-in phase, this model takes over the monitoring of the pump and reports any deviations from the normal condition.
Pumps can be operated in different states (e.g. load or no-load). Different thresholds are permissible in each of these states. The SmartLimitWatcher function of the DataScienceToolbox can set thresholds dynamically. If process values are outside a specific confidence band, warnings or alarms will be issued as with static thresholds.
In order to monitor the flow rate (target variable), the SmartLimitWatcher is used. Support variables (speed, pump pressure, vibration data) are used for this purpose. They describe the flow characteristics in different operating states. For example, with increasing flow, the speed and the pump pressure increase as well.
Data recording has increased transparency, resulting in optimisation potentials. The higher plant uptime has improved the overall process. Integrated alarm management ensures fast reaction to changing process parameters, optimising maintenance. All measures increase process and product quality. moneo|RTM ensures detailed process visualisation.
The plant was digitised successfully without any changes or interventions to the existing PLC or software.
Get the big picture on the moneo dashboard.
The dashboard provides the user with an overview of the relevant process values for this plant.
The analysis function can be used to access historical data and compare different process values. The diagram shows a typical characteristic curve for start-up ①, operation ② and stop ③.
It can be observed that the speed and pressure curves are almost identical. The flow in the system is trailing slightly, which is normal due to the inertia of the medium.
Various parameters of the pump can be monitored statically, as they are independent of the operating status. For example, in this case, the motor temperature must not exceed 50°C. This is easily achieved by setting static warning and alarm thresholds.
In this use case, the SmartLimitWatcher is used to monitor the flow of the pump (target variable).
Using the support variables (pump pressure, speed and acceleration values), a model is calculated which creates a confidence band around the process value. It defines the threshold values for the flow, taking into account different operating states.
The sensitivity, and thus, the width of the confidence band, can be adjusted for the lower and upper warning and alarm thresholds via parameters (inactive, low, medium and high). This makes it possible to hide any false warnings or alarms.
This function can be used to easily define what should happen after a warning or alarm has been triggered, e.g.:
In addition to the process values of the sensors, moneo also records the operating hours of the pump. This function can be implemented quickly and easily using the “Operating hours counter” template.
A data source ② describing the operating state is required. In the example below, the speed is used and the following thresholds ③ are set: