Multivariate Statistical Process Monitoring of the SB Waste Water by dfhrf555fcg


									                              CPACT Confidential Proprietary. Photocopying not permitted. Copy requests through CPACT (Newcastle)

       Multivariate Statistical Process Monitoring of the SB Waste Water Treatment Plant§

The study demonstrated the applicability of multivariate statistical methods for process performance
monitoring and process output modelling to the SB waste water treatment plant (WWTP), a two stage
activated sludge process for COD removal and removal of ammonia.

Dynamic PLS and dynamic Multi-block PLS modelling methodologies were used to generate
statistical models of nominal process operating conditions and to model and predict important quality
variables such as settling velocity index (SVI) and the Stage Effluent and the Final Effluent conditions.
Modelling, prediction and performance monitoring of the SVI of the stages is intended to facilitate
decision making and overall control of the WWTP.

The dynamic multi-block prediction of Stage 1 SVI and Effluent conditions is markedly improved over
    Title :
    c :\y an ni s\s b\ mb _ res ul ts \c2 d0 _1 9 .ep s
    C re ato r:
    MA TL A B , Th e Math w orks , In c.
                                                                      the dynamic PLS approach. Predictions of Stage 2 and
    P re vie w :
    This E P S pi ctu re w as n ot sa ve d
    w ith a p rev ie w in clu de d in it.
    C omme nt:
    This E P S pi ctu re w ill p rin t to a
    P os tS c rip t p rin te r, b ut n ot to
    o the r t yp es o f p rin te rs.
                                                                      Final Effluent conditions were also encouraging and
                                                                      superior to the predictions from the standard dynamic PLS
                                                                      models. The results of the dynamic multi-block PLS
                                                                      predictions of the final effluent NH3 levels are shown for
                                                                      both nominal and validation data sets.

 For both Stages the dynamic PLS models provided encouraging one-step ahead predictions of the
         SVI and some other Stage Effluent variables.                             The prediction of other output variables was
         unsatisfactory. MSPC SPE plots correctly identified most of the non-conforming data points.
         Individual Stage dynamic PCA models provide a day-in-advance assessment of plant performance.
 The dynamic multi-block PLS models show improvements for Stage SVI and Final Effluent
         conditions. Stage 1 and 2 Effluent conditions can also be modelled. MSPC SPE plots for the overall
         multi-block model identify the conditions and effluents (RAS and WAS flows and Stage Effluent)
         from the Stage 1 clarifiers as responsible for most of the faulty operating conditions.
 The modelling has been encouraging despite limitations of the data that includes missing data on
         important variables, non-measured and/or unmeasured, but important, variables and the combining
         of spot and composite samples. The results can be used as a feedback to process engineers to
         enhance process knowledge, improve faulty data identification, and identify important variables to
         be additionally monitored and pinpoint the underlying cause of a specific abnormal operation.

    Sargantanis, I, E.B. Martin and A.J. Morris, “ Multivariate Statistical Process Monitoring of an Industrial
     Waste Water Treatment Plant”, to be submitted to Water Research.

To top