Jul 08, 2019

A Steady Stream of Insights

Advanced analytics help water & wastewater utilities avoid fines

This dashboard report depicts a filter backwash cycle summary for a drinking water treatment plant. The municipality was concerned with improving its sand filter consistency, and advanced analytics like those shown were used to develop an efficient program to achieve that goal.
This dashboard report depicts a filter backwash cycle summary for a drinking water treatment plant. The municipality was concerned with improving its sand filter consistency, and advanced analytics like those shown were used to develop an efficient program to achieve that goal.

In water and wastewater treatment plants and facilities, managers have a flood of data to inform their decisions. Supervisory control and data acquisition, distributed control and laboratory information management systems (LIMS) all have databases that produce an endless supply of process data that can identify situations requiring action or optimizations to improve operations.

Turning process data into insights in a timely fashion has been a challenge for treatment plant personnel, many of whom have traditionally performed process data analytics in spreadsheets. It is difficult to extract data from disparate systems and import it into spreadsheets, but the lack of reusability often is a greater problem. If an engineer builds an Excel model to predict when flow will go out of range, for example, he or she will have to rebuild it to examine a different data set. 

Now, advanced analytics applications offer a change for the better. McKinsey & Co. views advanced analytics as a means to apply mathematical tools and statistics to data for better business assessments and practices.

 

Intersection of Flow Data & Lab Results

Seeking continuous process improvement, water treatment plant managers review a variety of process data types, with wet chemistry or lab results stored on LIMS being a high priority. With lab results—such as alkalinity, hardness, total organic carbons and biological oxygen demand—typically stored every shift or every few hours, wet chemistry data volumes are high. 

Plant personnel traditionally have exported these data types into spreadsheets, a time-consuming process restricted to predefined time ranges. Advanced analytics make it possible to analyze historian data such as flow rates, pressures, temperatures and turbidity. This is then overlaid with data from LIMS databases. Once engineers perform calculations, they can adjust the time range and update the calculations for additional time periods, as well as easily publish or share insights. 

It is much faster and easier to calculate things, such as 30-day averages, perform root cause analysis, and create predictive analytics on asset failure or maintenance. With an advanced analytics application such as Seeq, this type of analytics can take minutes to conduct, rather than hours or days with physical paper and spreadsheets.

In addition, sharing and publishing insights from engineers enable the operations team to access charts and calculations in near real-time. If more details are needed, users can click into a Seeq document to see the underlying calculations and data because the analysis results are connected directly to the source data. These capabilities enable plant personnel to make faster, data-based decisions.

 

Real World Advanced Analytics Application

Water treatment monitoring and reporting. A municipal water provider needed to improve sand filter consistency and boost performance of its fleet of filters in its water treatment plant. As such, it needed to identify and monitor for poor filter performance while prioritizing filter maintenance. At the same time, it needed to reduce the total amount of clean water used in the backwash process.

The operations team had used a large, complex spreadsheet developed many years ago to track filter performance. The team was interested in filter backwash cycles, which require seasonal changes to the run time between backwashes and also to the backwash rates. It needed to compare multiple filters to identify any mechanical problems with the filters, and also needed to gauge required media replacement.  Because the spreadsheets were complex and infrequently updated, the backwash cycles were only adjusted a couple times per year. 

If the run lengths are not adjusted properly, the solids can become embedded deep in the filter, ultimately requiring filter shut down and media replacement, which is an expensive and time-consuming proposition.

The engineers created an analysis of the filters that enabled operators to:

  • Contextualize data through the use of operating mode conditions;
  • Develop monitoring metrics—such as unit filter run volume—and quantify the costs of each backwash;
  • Visualize each backwash cycle overlaid on subsequent cycles and also compare them to other filters;
  • Deploy models across the filter fleet using asset structures; and
  • Create daily and monthly reporting dashboards, which update automatically, to monitor and adjust filter performance before a shut-down became necessary.

 

Long-term improvements included:  

  • Eliminating monthly Excel reporting;
  • Increasing filter performance consistency; and
  • Increasing the ratio of production-to-backwash across filters.

 

The platform saved the municipality time and money, so it could optimize general processes and increase plant throughout for processing clean water.

Advanced analytics for waste-blockage modeling. Municipalities use wastewater treatment pumps or lift stations to keep waste flowing through the system using increased sewage pressure. Municipalities monitor flow rates to determine if there is too much wastewater at a station, and to see if pumps are cycling too frequently. 

When the process goes awry, raw sewage can spill out of manholes or other apertures. A field team can be dispatched to clean it up, but the utility is not always the first to know about the spill. In fact, the first notification of a wastewater spill often comes from a member of the public, hours and sometimes days after the spill. This can intensify the public health, environmental impacts and cost of clean-up efforts, including fines that might run $10,000 per event. Following a sewage spill at an environmentally significant site, one municipality sought a way to reduce the likelihood and impact of spills occurring in the future. 

The utility conducted a proof of concept project using Seeq to develop an online sewer blockage detection system. The team created a blockage model based on data from a recent spill with one goal in mind: detect the blockage earlier and faster than previous methods.

Powered by advanced analytics, the model identifies blockages by detecting the absence of normal fill and pump behavior in near real time. The absence of pump runs or extended fill time during peak times signifies an abnormality and possible blockage. The blockage model gave the utility a chance to head off spillages, safeguard the environment and improve public perception.

The model built on the application detected the blockage 13 hours before the spillage. This provided field teams with hours to deal with potential blockages in the sewer network. In addition, engineers reduced the amount of time spent working with spreadsheets, and the utility made process improvements to boost operating efficiencies based on this analysis. 

Advanced analytics for industrial wastewater treatment plant monitoring. Wastewater treatment plants remove contaminants, prevent permit exceedance and perform other critical functions. The source of wastewater can be highly variable, and the wastewater treatment plant’s operation normally is reactive. This can cause high fluctuations in treatment plant operation. 

Using Seeq, wastewater plant personnel can monitor upstream operations. Anomalies can provide early warning, guiding required treatment of incoming wastewater. For example, the quality of the water coming into the plant can be used to adjust the frequency of boiler or cooling tower blowdown.

Monitoring equipment in upstream units gives an early indication of an impending problem. Trouble with one unit might point to a pH excursion, for example. The utility might be able to clear inventory from its tanks to store the wastewater and bleed it into the system gradually, averting issues.

Advanced analytics insights allow plant personnel to:

  • Predict water quality using sensors that already exist;
  • Optimize chemical usage;
  • Alert maintenance crews of blockages to prevent spill events;
  • Protect the environment from sewage spill events or excursions; and
  • Document and continuously improve predictive models.

 

Conclusion

Water operations rise or fall based on asset performance. And every pump, blower, chemical feed system, instrument and sensor is highly dependent on changing flow parameters. In this environment, tracking and analyzing interrelated influences is critical. Advanced analytics offers accelerated, shareable insights to meet these goals, and represents a quantum leap forward over traditional spreadsheet-based approaches.

About the author

Michael Risse is the chief marketing officer and vice president at Seeq Corp. Risse can be reached at [email protected] 

expand_less