c fgd pumper

Data analytics and pump control in a wastewater treatment plant

Abstract

This paper analyzes data from ten 2,610 kW centrifugal pumps in a large wastewater treatment facility in order to identify efficiency opportunities, evaluate efficiency metrics, and assess applicability of operating space analysis. Three efficiency interventions are evaluated: pump maintenance, ranking and utilizing the most efficient pumps, and reducing the number of pumps operating during low flow periods. Systemic energy savings of 4.3%, 809,000 kWh annually, were identified. Applying average and true weighted efficiency to datasets with hourly and 5 min sampling rates demonstrates the effects of efficiency metric selection. Operating space analysis is introduced as a method of intersecting pump and system spaces. It is used to evaluate how methods of pump control, such as variable speed drives, affect the performance of a system. Results show that specific energy and maximum efficiency occur at slightly different operating points in the case study example, thus illustrating the utility of specific energy in control optimizations. Operating space analysis may simplify data mining for pumped systems, improve control algorithms, and chart opportunities for next-generation control technologies and further research.

Introduction

Centrifugal pumps are ubiquitous in fluid systems such as clean water distribution, wastewater treatment, pumped hydro energy storage, building HVAC systems, petroleum extraction, mining, and crop irrigation. In these energy-intensive systems pumps are often major energy consumers. Globally, pumps consume hundreds of billions of kilowatt hours of electricity each year. In the US, pumps consume an estimated 6% of U.S. electricity, equivalent to 232 billion kWh annually. This paper begins with key concepts from literature on evaluating and improving pump system efficiency. Using this background, we evaluate data from ten, 2610 kW centrifugal pumps in a large New England wastewater treatment plant (WWTP) to understand the influence of pump design, selection, maintenance and operation on system efficiency. We quantitatively explore the efficiency impact of interventions and qualitatively present trade-offs to implementation. The methods and recommendations developed for this case study can be applied more broadly to other pumping systems. Those insights are extended into a presentation of operating space analysis, a methodology built on the traditional pump curve and system intersection. Operating space analysis enables researchers to analyze complex interactions between pump control elements and fluid systems, motivating new pump design and system operation research. Pump system efficiency is maximized by a combination of improved design, pump selection, and adaptive control, shown in Fig.1, which includes accounting for wear. Data collection and analysis provide a foundation to enable improvements at each stage, as well as closing the loop on the pump lifecycle. Physical dimensions, such as clearances to enhance robustness, in the design influence operational efficiency. Likewise, knowledge of the operational duty cycle and system curve informs pump selection and appropriate sizing, which is important for capital and operational efficiency. Operational flexibility depends on whether controls such as variable speed drives are installed with the pumps. Furthermore, pump monitoring and maintenance can substantially reduce energy consumption. Given the wide range of operating pressures, flows, and speeds, advanced operational and pump control can further reduce energy consumption. Real systems operate over a range of flow and pressure conditions. In contrast, centrifugal pump manufacturers typically specify a single design point for most efficient operation; the best efficiency point (BEP). This is an important issue because system fluctuation away from the BEP leads to energy losses and increased wear. Fig.2 shows traditional pump and system characteristics intersecting at the BEP. It also visualizes shaded ranges of variation about the characteristics. Energy losses can be reduced throughout the pump life cycle: These strategies lead to more efficient operation and improved equipment lifetimes. Although the BEP is considered a standard metric for pumps, it does not capture the efficiency of the system operating over a range of conditions over time. True weighted efficiency (TWE), defined in Eq.(1), is the energy-weighted average efficiency for a pump. It is the proportion of hydraulic energy output compared to total energy input, where P n is the power input, t n is the time interval, and η is the efficiency. T W E directly measures the proportion of input energy wasted and enables engineers to compare pumps by calculating energy savings while accounting for the pump design and the system variations. The Pump Energy Index (PEI), promoted by both the Hydraulic Institute and the U.S. Department of Energy (US DOE), utilizes a similar method, but compares a pump with an aggregate baseline for a pre-determined load profile. A detailed review of pump efficiency metrics can be found in Dahl’s 2018 paper. T W E is a particularly helpful metric because it allows for direct and robust calculation of energy savings. Eq.(2) shows how to directly calculate energy savings, E s, from an intervention’s calculated T W E i, a baseline T W E b, and a baseline energy consumption E b. The results in Section2.2 use Eq.(2) to calculate the impact of different interventions for design, selection, and operation. Improving the performance of centrifugal pumps has driven innovation since well before the Machine of Marly was completed in 1684, many books on pump performance elaborate on great advances since then. Karassik in _The Pump Handbook_ extensively covers pump performance, design, selection, maintenance, and operation. Gülich presents efficient design methodologies to address the nuances of hydraulic, volumetric and mechanical efficiency. While pump design and control are mature areas, there are still significant opportunities to improve efficiency. Due to their ubiquity and large pumping requirements, wastewater plans are often used for case studies on energy efficiency. The National Renewable Energy Laboratory NREL published a 2012 report demonstrating the utility of intermittent process energy audits to identify systemic inefficiencies. Shankar etal. present a comprehensive review of pump efficiency enhancement opportunities, focused on selection and operational optimization. Torregrossa notes that accessible methodologies to analyze pump energy consumption are lacking. Longo etal. also addresses this gap in their 2019 paper. Improving the efficiency of WWTPs is of great importance but real-time data is lacking for better efficiency monitoring and control. If available, larger datasets could help enable more widespread implementation of Supervisory Control And Data Acquisition (SCADA) systems which could then evolve in conjunction with frequent digital data collection of flow, differential pressure, motor speed and power. This would enable real time decision making based on efficiency metrics and trends. The most basic methods of flow control in a pump system are adjusting valves to constrict flow or turning pumps on and off. The introduction of variable frequency drives (VFDs) in the past half century provide system control of either flow or efficiency. The question of how best to control a VFD pump in a real system is still an open topic in the field of optimization. Notably, there are examples of using data and optimization to improve wastewater plant operations. For example, Zhang etal. present the use of a neural network to estimate a pump system from data and then apply aiNet to solve a bi-objective optimization in order to optimize WWTP control. They state their system can theoretically lead to as much as 25% energy savings. Similarly, Torregrossa and Capitanescu share their application of heuristic optimization algorithms in one paper to reduce the energy consumption of a wastewater treatment plant, and fuzzy logic in another. Filipe etal. present model-free data predictive control using deep reinforcement learning and proximal policy optimization, which when used to control pumps in a Portuguese WWTP, led to a 16% decrease in energy consumption. These references show significant savings are possible with advanced control; however advanced control needs to be built upon a foundation of best practices for pump selection, sizing, and system configuration. A priori knowledge of pump and systems characteristics and how to best use them in conjunction with advanced control techniques would greatly help plant managers and practitioners select the best integrated pump and control schemes for their applications.

Materials and methods

Pump operating data were acquired from a large New England wastewater treatment plant effluent pump station. The pumps consume 19.5 million kWh annually, which is approximately 2.4% of the state’s annual electricity consumption. The system consists of ten Fairbanks Morse vertically mounted end suction centrifugal pumps with 42” 5-vane impellers. Each pump is driven by a 2610 kW (3500 HP) variable speed, synchronous three phase motor with variable frequency drive. Each pump was originally

Materials and methods

Operating space analysis extends the classical two dimensional pump model demonstrated in Fig.2 to a multidimensional analysis, incorporating control variables such as speed to construct intersecting operating and system surfaces. This method was motivated by the question: to what degree can VFDs, or other variable mechanisms, enable control over both flow and efficiency objectives? Typical practitioners strive to simultaneously control flow and maximize efficiency. However, VFDs alone are

Conclusion

While pump design, selection, and operation are mature fields, significant opportunities exist to improve performance by leveraging advances in data synthesis, system control, and operational optimization. The operating space analysis and WWTP case study presented here demonstrate how classical pump fundamentals with feature rich data analysis can yield significant energy savings. Three feasible interventions were quantitatively compared: pump maintenance, utilizing the most efficient pumps,