SAM4 Automated Clogging Detection Feature: Revolutionizing Wastewater Maintenance

Find out how SAM4 assists water and wastewater companies detect pump clogging remotely with advanced analytics. Prevent pollution events, reduce maintenance costs, and optimize operational efficiency with automated clogging detection.

Automated and Remote Clogging Detection for Optimized Wastewater Maintenance

Resolving and preventing sewage pump blockages is a significant operational challenge for wastewater companies. Currently, thousands of on-site visits are required to preventively inspect and reactively unclog these pumps each year. Our customer expressed the need for a solution to reliably monitor pump clogging remotely to transform their maintenance strategies. With SAM4 Health already monitoring their pumps, we developed an automated clogging detection feature that tracks clogging events remotely.

Since it has been launched, SAM4 Health’s clogging detection has identified hundreds of clogging events, allowing our customers to make data-driven changes to their operational strategies. Alongside our customer’s usage, we continuously develope our clogging detection feature to match their evolving needs. On this page, we explain how SAM4 Health’s clogging detection works and show a real-world example from September 2022.

(Place Holder)

How SAM4 Health’s Clogging Detection Feature Works

SAM4 Health is part of our SAM4 industrial analytics platform, powered by electrical signature analysis (ESA). ESA, developed in the 1980s, is a suite of tools used to monitor the health of critical equipment from a distance. SAM4 brings ESA into the 21st century using AI and wireless sensors that capture high-frequency data from motor control cabinets. This data provides early warnings of asset health degradation, such as the start of a clogging event. (Read more about how ESA works in our ESA Explainer.)

Machine Learning in SAM4 Health to Detect Clogging

SAM4 Health uses machine learning to detect pump clogging by analyzing subtle changes in electrical signals (current and voltage). SAM4’s AI models identify the early stages of clogging and can even pinpoint which component is degrading. Both electrical and mechanical faults impact these signals, each producing a unique “signature.” SAM4’s AI is trained to recognize these distinct patterns, enabling precise and timely clog detection.

In addition, SAM4 Health’s clogging detection model analyzes multiple features simultaneously to determine not only when clogging is developing but also where it is occurring. With real-time monitoring, SAM4 provides alerts to customers, allowing them to intervene promptly and avoid more serious operational issues.

(Place Holder)

Technical Deep Dive: Inside SAM4 Health’s Clogging Detection Model

The data captured by SAM4’s current voltage sensors undergoes a feature extraction phase. For that basic signal processing calculates key metrics related to the mechanical and electrical health of both the pump and its motor. For example, one feature may isolate data components around the supply frequency, while another focuses on frequencies related to the motor’s rotational speed. Each of SAM4 Health’s fault detection models uses one or more of these features, depending on the fault it is designed to detect. These detection models are where the machine learning (or “artificial intelligence”) takes place, enabling precise and efficient identification of issues.

When a new pump is first connected to SAM4 Health, the system begins by collecting initial data to train SAM4’s detection models on what normal operating behavior looks like for that specific pump. In the case of the clogging detection model, this training helps define the key features of the pump’s performance when no clog is present. Once the model is fully trained, any deviations from these expected patterns will trigger SAM4 to generate an automated clogging alert, allowing for early detection and intervention.

Because the SAM4 clogging detection model analyzes multiple features simultaneously, it can accurately detect a developing clog in its exact location.

For example, an increase in noise around the pump’s supply frequency combined with a drop in load indicates a clog between the impeller and the motor. Whereas shifts in other indicators suggest a clog in the volute. SAM4’s advanced clogging model processes all incoming data to determine not only when clogging occurs but also its precise location, ensuring efficient and targeted maintenance.

(Place Holder)

Once the AI model flags a developing clog, SAM4 Health then looks at the status of the entire pumping station to determine when and what severity it should report to the customer. For example, if one pump out of three is clogging and the other two are operating normally, SAM4 will issue a low-priority notification. However, if another pump starts showing signs of trouble, SAM4 will send a high-priority alert. This automated process streamlines notification prioritization, enabling customers to plan the most efficient and effective maintenance response.

(Place Holder): SAM4 Health evaluates the entire pumping station to determine the timing and severity of its notifications. (SAM4 Energy, seen here, also provides its recommendations for the whole station)

Note that clogging is only one of many issues the second pump might encounter — all of SAM4 Health’s other detection models are also continuously running, simultaneously tracking other electrical and mechanical faults such as voltage imbalance, impeller damage and misalignment. Detecting these problems at an early stage, does not onlz reduce total maintenance cost and environmental impact, but also extends machine lifetime and raises efficiency for the long term.

Clogging Feature in Action: Early Alert for Developing Clogging Incident Across Two Pumps

At a sewage pumping station two co-located pumps operate in an alternating pattern to meet required demand and provide system redundancy. One night, one of the pumps (Pump 2), displayed characteristics of clogging. This was indicated by continuously high clogging scores since the start of the incident as can be seen in Figure 3.

(Place Holder): From the generated data, you can clearly see the start of a clogging incident during the night of 9-10 September 2022. Hereafter, the clogging score remained at the highest level of clogging indication (around 1.0) until a maintenance crew cleared the blockage on 20 September.

With one other functioning pump at the station, this did not trigger an emergency response. The situation was monitored for 48 hours to determine if the blockage would clear on its own or require intervention. With no improvement seen at the end of this monitoring period, an orange alert was generated and sent to the customer indicating a clogging incident at the pump. The alert stated that same-day action was not required and that the situation could be resolved in the following week without disrupting pump station operations.

Approximately one week later, data revealed signs of developing clogging in the other pump at the station (Pump 1). With both pumps clogged, an emergency response was necessary due to the increased risk of a potential pollution event. Consequently, a red alert notification was sent to the customer, indicating the urgency of the situation.

A maintenance team was promptly dispatched to the site and removed both pumps from the pumping station. They confirmed that both pumps were partially blocked, validating the clogging detections made by SAM4 Health’s automated clogging feature. Once the blockages were cleared and the pumps were reinstalled, the clogging scores and other operational metrics returned to healthy levels, as shown in Figure 4.

(Place Holder): The resolution of the clogging incident is clearly visible on September 20, 2022. After the maintenance crew removed the blockages, the consistently high clogging scores for Pump 2 and the developing clogging signs in Pump 1 dropped, returning to healthy clogging score levels.

The SAM4 Health automated clogging detection feature successfully pinpointed the developing issues. It provided continuous insight into the severity of the situation and alerted to when action needed to be taken. This enabled the customer to optimize maintenance planning and prevent a potential pollution event from occurring.