A PRACTICAL MODEL FOR FULL-SCALE OPTIMISATION OF THE ANAEROBIC - PowerPoint PPT Presentation

A PRACTICAL MODEL FOR FULL-SCALE OPTIMISATION OF THE ANAEROBIC DIGESTION PROCESS Stephen R Smith and Jin Liu Department of Civil and Environmental Engineering, Imperial College London Email s.r.smith@imperial.ac.uk

Background • Overall 80% of sewage sludge in the UK is treated using anaerobic digestion (AD) • Anaerobic digestion is a practical and cost-effective method for the stabilisation and treatment of residual sewage sludge • It is also a significant producer of renewable energy in form of biogas • There is still a lack of practical modelling tools available to guide operators and maximise the energy output

AD is a dynamic system affected by multiple factors: Hydraulic retention time Sludge composition (HRT) Temperature Sludge age Primary SAS ratio Dry solids (DS) Volatile fatty acid Volatile solids (VS) (VFA) Mixing efficiency Iron dosing Ammonia Organic loading rate Pretreatment Asset age

Data Summary Routinely collected parameters: • Temperature • DS • HRT Additional parameters collected at specific sites: • VS, sludge age, primary SAS ratio, VFA etc.

Biogas yield (m 3 /t DS) Data Summary 1,000 Conventional mesophilic anaerobic digestion (MAD) dataset 100 200 300 400 500 600 700 800 900 0 Site1 Site2 Site3 Site4 Site5 Site6 2013-2017 Site7 Site8 Site9 Site10 Site11 Company 1 Site12 Site13 Site14 Site15 Site16 Site17 Site18 Site19 Site20 Site21 Site22 Site23 Site24 Site25 Site26 Site27 Site28 Site29 Site30 Site31 2011-2016 Site32 Site33 Company 2 Site34 Site35 Site36 Site37 Site38 Site39 Site40 Site41 Site42 Site43 Site44 2014-2016 Site45 Company 3 Site46 Site47 Site48 Site49 Site50 Site51 Site52 Site53 Site54 Site55 Site56 2009-2016 Company 4 Site57 Site58 Site59 Site60 Site61 Site62 Site63 Site64 Site65 Site66

Multiple Regression Analysis Simple linear multiple regression: 700 𝑧 = 𝑐 1 𝑦 1 + 𝑐 2 𝑦 2 + … + 𝑐 𝑜 𝑦 𝑜 +c Biogas yield (m 3 /t DS) 600 Simple regression • Significant predictors are selected 500 line based on P< 0.05 400 However, the full-scale AD data are clustered 300 • Categorical factor: multi-level 200 regression 0.0 2.0 4.0 6.0 8.0 10.0 12.0 DS feed (%) • Centering approach: evaluating the Site12 Site14 THP Site 3 THP Site 6 relative changes

Multiple Regression Analysis Conventional MAD model: Biogas yield = 230.9 * (Ln(Temperature) - 3.6) + 136.2*(Ln(HRT) - 3.0) - 224.8 * (Ln(DS) - 1.5) + 75.5 * ((Ln(HRT) - 3.0) * (Ln(DS) - 1.5)) + site factor Variation explained: 0.46% 2.55% 5.31% 0.11% 42.42%

Biogas yield (m 3 /t DS) 200 400 600 Model Validation – Conventional MAD 0 08/2011 11/2011 02/2012 05/2012 Data involved in model generation 08/2012 11/2012 02/2013 05/2013 R 2 =0.65, P <0.001 08/2013 (2011 - 2016) 11/2013 Site 42 02/2014 05/2014 08/2014 11/2014 02/2015 05/2015 08/2015 11/2015 02/2016 05/2016 Biogas yield (m 3 /t DS) 200.00 400.00 600.00 0.00 04/2016 06/2016 08/2016 10/2016 Independent datasets 12/2016 (2016 - 2019) 02/2017 R 2 =0.59, P <0.001 04/2017 Site 38 06/2017 08/2017 10/2017 12/2017 02/2018 04/2018 06/2018 08/2018 10/2018 12/2018 02/2019

Expansion of Advanced AD Thermal hydrolysis process (THP) Increases: • Loading rate • Biogas yield • VS reduction • Pathogen kill • Dewaterability

Combined Conventional-THP MAD Model Development Apply conventional MAD model into THP datasets Observed biogas yield 800 y = 0.95x + 22.44 R 2 = 0.72 P <0.001 600 (m 3 /t DS) THP site 1 THP site 2 400 THP site 3 THP site 4 200 THP site 5 THP site 6 0 0 200 400 600 800 1000 Predicted biogas yield (m 3 /t DS)

Combined Conventional-THP MAD Model Development Conventional MAD model : Biogas yield = 230.9 * (Ln(Temperature) - 3.6) + 136.2*(Ln(HRT) - 3.0) - 224.8 * (Ln(DS) - 1.5) + 75.5 * ((Ln(HRT) - 3.0) * (Ln(DS) - 1.5)) + site factor Combined conventional-THP MAD model : Biogas yield = 265.3 * (Ln(Temperature) - 3.6) + 133.7 * (Ln(HRT) - 3.0) - 216.4 * (Ln(DS) - 1.5) + 61.7*((Ln(HRT) - 3.0) * (Ln(DS) - 1.5)) + site factor

Combined Conventional-THP MAD Model Impact of HRT, DS and temperature on performance

Optimisation Strategies Net energy balance - changing HRT m 3 day = Biogas volume 1 − Biogas volume 2 Net daily biogas gas Τ 100 × Digester volume DS × BY − Digester volume = DS 100 × Digester volume × BY × HRT 2 − HRT 1 × BY + x = × BY + x HRT HRT 2 HRT 1 × HRT 2 1 • Biogas volume 1 (m 3 /day) = the volume of biogas produced when HRT is equal to HRT 1 • Biogas volume 2 (m 3 /day) = the volume of biogas produced when HRT is equal to HRT 2 • BY (m 3 /t DS) = the biogas yield when HRT is equal to HRT 1 • x (m 3 /t DS) = the relative change of biogas yield when HRT is changed from HRT 1 to HRT 2

Optimisation Strategies Net energy balance - changing HRT • • Digester volume: 2000 m 3 Digester volume: 2000 m 3 Increasing daily • • Feed Volume: 100 m 3 /day Feed Volume: 133.3 m 3 /day feed volume to 133.3 m 3 /day • • HRT: 20 days HRT: 15 days • • DS: 5% DS: 5% • • Total sludge feed: 5.0 t DS/ day Total sludge feed: 6.7 t DS/ day 19.4% increase in biogas volume • • BY: 400 m 3 /t DS BY: 358 m 3 /t DS • • Biogas produced: 2000 m 3 / day Biogas produced: 2387 m 3 / day

Optimisation Strategies Net energy balance - changing temperature and DS 6 Net Energy out from 1 t wet sludge (kWh) 4 2 0 31 to 33 33 to 35 35 to 37 37 to 39 39 to 41 -2 Temperature increase ( o C) 2.7% DS 3.0% DS 4.5% DS 7.9% DS Energy required to heat 1 t wet sludge is 2.3 kWh. 2.7% DS is the lower 5% percentile range value of monthly average operational data for conventional MAD sites; 3.0% is the break point sludge feed DS for a positive net energy balance for MAD; 4.5% and 7.9% DS are the mean values of monthly average operational data for conventional and THP MAD sites, respectively.

Conclusions • The first time that simple operational models of the AD process have been developed based on full-scale operational data • Important to balance the three key operational parameters (temperature, HRT, and DS) to optimise the energy balance of the process

Acknowledgements • This research is sponsored by Anglian Water Services Ltd, Severn Trent Plc, Thames Water Utilities Limited, United Utilities Group Plc and Yorkshire Water Services Ltd • The views expressed in the presentation are those of the authors and do not necessarily represent the companies supporting the research