A critical comparison of methods for benchmarking energy performance in WWTPs
Abstract
The energy efficiency of wastewater treatment plants (WWTPs) is a subject of growing interests by water utility and WWTPs operators. As consequence, benchmarking in recent years has become an important tool to assess the performance of these facilities and help identify best practices. This study compares various benchmarking methodology employed for the assessment of the energy efficiency in WWTPs. The literature review revealed three main benchmarking approaches: normalization, statistical techniques and programming techniques. Advantages and disadvantages were identified for each one. It was found additionally that no single Key Performance Indicators (KPIs) used to characterize the energy performance could be used universally. In general, while these methods can be used for comparison, the diagnosis of the energy performance remains an unsolved issue
Full text
A Critical Comparison Of Methods For Benchmarking Energy Performance In WWTPs Stefano Longo, Miguel Mauricio-Iglesias, Almudena Hospido Universidade de Santiago de Compostela, SPAIN 11 OCTOBER 2016
2 WHY ENERGY BENCHMARKING? Define energy consumption indicators Take representative samples Compare with other plants. Link the measurements with the plant operation Implement actions to improve the energy efficiency Monitor the effect of new actions DEFINE MEASURE ANALYSE IMPROVE CONTROL H2020 Coordination and support action ENERWATER. www.enerwater.eu
3 COMPARISON OF METHODS Index methods Data Envelopment Analysis Regression methods Pros - Easy to implement and understand - Statistical tests are readily available Cons - No scale effects - Exogenous variability cannot be accounted for - Need of composite index for multiple inputs/outputs Pros - Ready for multiple inputs/outputs - No need to specify a functional form - Useful for a given set (e.g. the WWTPs of an operator) - Can account for scale effects Cons - Can provide too many optima if many inputs/outputs - Sensitive to measurement error and outliers - Statistical tests are cumbersome - Finite sample effect - Exogenous variability is difficult to include - Homogeneous set of input/output Pros - Statistical tests are readily available - Composite result (the performance can be compensated) - Exogenous variability readily included - Provides effect and size of inputs and exogenous variables→diagnosis Cons - Composite result (the performance can be compensated) - Scale effects require the right functional form - Sensitive to outliers - Finite sample effect
4 INDEX METHODS - Set of 24 WWTPs from ENERWATER project. Different sizes and types. - Three outputs (flowrate, COD removed, solids treated) and one input (energy) - Ranking? - Weights? Efficiency ranking of a set of WWTPs according to a composite index
5 INDEX METHODS - A composite index with the following weights: •15% pumping •74% COD removal •11% sludge treatment - Weights are arbitrary but meaningful! Efficiency ranking of a set of WWTPs according to a composite index
6 DATA ENVELOPMENT ANALYSIS - Set of 24 WWTPs from ENERWATER project. Different sizes and types. - DEA allows to reconcile multiple outputs with different units -It is possible to rank the WWTP energy efficiency - Applying weights is difficult. Diagnosis is not clear Efficiency ranking of a set of WWTPs according to three output criteria Efficiency
7 REGRESSION METHODS - Set of 187 WWTPs with influent and effluent characteristics - Differences persist as we control for more covariates - No longer significant between Germany and Spain Does the country location impact the WWTP energy consumption? Germany France Spain No control -0.497*** (0.084) 0.614*** (0.157) 0.797***(0.110) Log(F) -0.425*** (0.059) 0.541*** (0.110) 0.680*** (0.077) Log(F) , CODinf -0.300*** (0.065) 0.609*** (0.107) 0.416*** (0.099) Log(F) , CODinf, PLF -0.270*** (0.063) 0.528*** (0.104) 0.378*** (0.095) Log(F) , CODinf, PLF, 2treat -0.133 (0.155) 0.840*** (0.114) 0.003 (0.235)
8 REGRESSION METHODS French WWTPs consume around 50% more energy than comparable German and Spanish WWTPs Does the country location impact the WWTP energy consumption?
9 CONCLUSIONS Simple index methods are very easy to understand and can be applied with flexibility(1. For multiple input/output evaluations they may require weights and/or data normalisation which may appear arbitrary. DEA is excellent for several input/outputs. It requires a homogeneous set of high quality data Regression methods are best suited for diagnosis and to cover the effect of exogenous variables. It requires a certain amount of data but it is less sensitive to outliers 1) Cabrera et al. 2016. Global Trends & Challenges in Water Science, Research and Management