Gray box modelling of MSW biodegradation; Revealing its dominant (bio)chemical mechanism

Gray box modelling of MSW biodegradation

Revealing its dominant (bio)chemical mechanism

André G. van Turnhout1,Timo J. Heimovaara1

Robbert Kleerebezem2

1_{Department of Geosciences and Engineering}

Delft University of Technology

2_{Department of Biotechnology}

Delft University of Technology

14th International Waste Management and Landfill Symposium, Sardinia 2013

Outline

1 Motivation

Control and Prediction of Emission Potential

2 Landfill Simulator Data

Experiments carried out by Roberto Valencia

3 Biogeochemical Process Model

Gray box approach

Model Verification: Literature parameter ranges and Bayesian Inference using measured data

4 Results

Model fit and parameter distributions

5 Conclusion

Outline

1 Motivation

Control and Prediction of Emission Potential

2 Landfill Simulator Data

Experiments carried out by Roberto Valencia

3 Biogeochemical Process Model

Gray box approach

Model Verification: Literature parameter ranges and Bayesian Inference using measured data

4 Results

Model fit and parameter distributions

5 Conclusion

Release from Aftercare: Emission Potential

Three pilot projects for Landfill Stabilization in the Netherlands.

I leachate recirculation;

I landfill Aeration.

Aim is to reduce Emission Potential to such a level that leachate and gas no longer pose a threat.

I concentration of potential contaminants in waste body;

I leaching process ;

I occurrence of preferential flow.

0 10 20 30 40 50 0 500 1000 1500 2000 2500 3000 3500 4000 time (years) concentration

Solid (mass/tot vol) D (mass/immob wat vol) C (mass/mob wat vol)

Release from Aftercare: Emission Potential

Three pilot projects for Landfill Stabilization in the Netherlands.

I leachate recirculation;

I landfill Aeration.

Aim is to reduce Emission Potential to such a level that leachate and gas no longer pose a threat.

I concentration of potential contaminants in waste body;

I leaching process ;

I occurrence of preferential flow.

0 10 20 30 40 50 0 500 1000 1500 2000 2500 3000 3500 4000 time (years) concentration

Solid (mass/tot vol) D (mass/immob wat vol) C (mass/mob wat vol)

Organic Matter as a Control of Emission Potential.

Contaminants present in Solid Waste.

I nitrogen;

I carbon source for Methane and CO2.

Source of substrate for micro-organisms present in waste body.

I biological degradation controls local pH and redox conditions;

I inorganic chemistry highly dependent on pH and redox;

Dissolved Organic Carbon enhances solubility of micro-contaminants due to complexation

I heavy metals (Cu, Cr , Pb, etc.)

I organic micro pollutants (i.e. polycyclic aromatic hydrocarbons, PAH)

Organic Matter as a Control of Emission Potential.

Contaminants present in Solid Waste.

I nitrogen;

I carbon source for Methane and CO2.

Source of substrate for micro-organisms present in waste body.

I biological degradation controls local pH and redox conditions;

I inorganic chemistry highly dependent on pH and redox; Dissolved Organic Carbon enhances solubility of micro-contaminants due to complexation

I heavy metals (Cu, Cr , Pb, etc.)

I organic micro pollutants (i.e. polycyclic aromatic hydrocarbons, PAH)

Organic Matter as a Control of Emission Potential.

Contaminants present in Solid Waste.

I nitrogen;

I carbon source for Methane and CO2.

Source of substrate for micro-organisms present in waste body.

I biological degradation controls local pH and redox conditions;

I inorganic chemistry highly dependent on pH and redox;

Dissolved Organic Carbon enhances solubility of micro-contaminants due to complexation

I heavy metals (Cu, Cr , Pb, etc.)

I organic micro pollutants (i.e. polycyclic aromatic hydrocarbons, PAH)

Outline

1 Motivation

Control and Prediction of Emission Potential 2 Landfill Simulator Data

Experiments carried out by Roberto Valencia

3 Biogeochemical Process Model

Gray box approach

Model Verification: Literature parameter ranges and Bayesian Inference using measured data

4 Results

Model fit and parameter distributions

5 Conclusion

Experimental Setup

Example of Experimental Data

0 200 400 600 800 0 20 40 60 t (d) CO 2 &CH 4 (m 3) Biogas 0 200 400 600 800 0 0.5 1 t (d) VFA l (M) VFA_l 0 200 400 600 800 4 6 8 10 t (d) pH pH 0 200 400 600 800 0 0.05 0.1 0.15 0.2 NH 4 +_&NH 3 (M) t (d) NH 4 &NH 3 (M) 0 200 400 600 800 0 0.2 0.4 0.6 0.8 pCH₄ t (d) pCH 4 (atm) 0 200 400 600 800 0 0.5 1 pCO₂ t (d) pCO 2 (atm)

Outline

1 Motivation

Control and Prediction of Emission Potential

2 Landfill Simulator Data

Experiments carried out by Roberto Valencia 3 Biogeochemical Process Model

Gray box approach

Model Verification: Literature parameter ranges and Bayesian Inference using measured data

4 Results

Model fit and parameter distributions

5 Conclusion

Predictive Modeling of Temporal Dynamics

Using “fundamental parameters”.

I _{parameters from literature;}

I _{based as much as possible on accepted theory (i.e.}

thermodynamics).

Not too complex.

I lumped process model.

Approach aimed towards providing insight.

I State variables are related to measurable quantities.

Conceptual Model

No transport limitations

I Continuous Stirred Tank Reactor (CSTR)

Hydrolysis

I rate limited

Microbial growth

I kinetic redox reaction

I multiple species

I multiple terminating electron acceptors (NO₃−,

SO₄2−, CO2)

Landfill gas production (CH4, CO2 and NH3) I equilibrium gas partitioning

Solubility control (equilibrium, acid-base)

Rate Equation

Rate dependent reactions: Matlab

dM_il dt = n X j=1 νi ,j · k_jmax · m Y a=1 fa ! · Cj· Vl− ϕl →gi dM_ig dt = ϕ l →g i − xi· Fout ϕl →g_i = kla·  C_il−xi· p Hi 

Multiphase geochemical equilibrium: Orchestra

Mi = Ki ms

Y

j=1,j6=i

Rate Equation

Rate dependent reactions: Matlab

dM_il dt = n X j=1 νi ,j · k_jmax · m Y a=1 fa ! · Cj· Vl− ϕl →gi dM_ig dt = ϕ l →g i − xi· Fout ϕl →g_i = kla·  C_il−xi· p Hi 

Multiphase geochemical equilibrium: Orchestra

Mi = Ki ms

Y

j=1,j6=i

Outline

1 Motivation

Control and Prediction of Emission Potential

2 Landfill Simulator Data

Experiments carried out by Roberto Valencia 3 Biogeochemical Process Model

Gray box approach

Model Verification: Literature parameter ranges and Bayesian Inference using measured data

4 Results

Model fit and parameter distributions

5 Conclusion

Verification: Bayesian Inference

Parameter optimization using Bayesian Inference.

I MCMC method;

I PDF of optimal parameter space; P(θ | D) = P(θ)P(D | θ)

P(D) Prior estimate

I bandwidth reported in literature

I extremely wide search space

Verification

I parameter within literature bandwidth: reliable

I parameter outside: parameter compensates model error?

Verification: Bayesian Inference

Parameter optimization using Bayesian Inference.

I MCMC method;

I PDF of optimal parameter space; P(θ | D) = P(θ)P(D | θ)

P(D)

Prior estimate

I bandwidth reported in literature

I extremely wide search space Verification

I parameter within literature bandwidth: reliable

I parameter outside: parameter compensates model error?

Verification: Bayesian Inference

Parameter optimization using Bayesian Inference.

I MCMC method;

I PDF of optimal parameter space; P(θ | D) = P(θ)P(D | θ)

P(D)

Prior estimate

I bandwidth reported in literature

I extremely wide search space

Verification

I parameter within literature bandwidth: reliable

I parameter outside: parameter compensates model error?

Outline

1 Motivation

Control and Prediction of Emission Potential

2 Landfill Simulator Data

Experiments carried out by Roberto Valencia

3 Biogeochemical Process Model

Gray box approach

Model Verification: Literature parameter ranges and Bayesian Inference using measured data

4 Results

Model fit and parameter distributions

5 Conclusion

Model Fit & Parameter Distributions

0 500 1000 0 20 40 60 t (d) CO 2 &CH 4 (m 3) Biogas 0 500 1000 0 0.5 1 t (d) VFA l (M) VFA_l 0 500 1000 4 6 8 10 t (d) pH pH 0 500 1000 0 0.05 0.1 0.15 0.2 NH 4 +_&NH 3 (M) t (d) NH 4 &NH 3 (M) 0 500 1000 0 0.2 0.4 0.6 0.8 pCH₄ t (d) pCH 4 (atm) 0 500 1000 0 0.5 1 pCO₂ t (d) pCO 2 (atm) 0.040 0.06 0.08 100 200 300 kmax hyd d−1 0 0.5 1 0 200 400 kmax meth d−1 0 2 4 0 100 200 kmax sulph d−1 4 6 8 x 10−3 0 100 200 300 kmax NH 4 d−1 0 0.2 0.4 0 100 200 300 K_s,meth mol L−1 0 5 10 0 100 200 300 K_s,sulph mol L−1 0 0.05 0 200 400 C_{i,methanogens} mol L−1 0 0.5 1 0 200 400 C_{i,sulphate reducers} mol L−1

Model Fit & Parameter Distributions

0 200 400 0 10 20 30 40 t (d) CO 2 &CH 4 (m 3) Biogas 0 200 400 0 0.2 0.4 0.6 0.8 t (d) VFA l (M) VFA_l 0 200 400 5 6 7 8 t (d) pH pH 0 200 400 0 0.05 0.1 0.15 0.2 NH 4 +_&NH 3 (M) t (d) NH 4 &NH 3 (M) 0 200 400 0 0.2 0.4 0.6 0.8 pCH₄ t (d) pCH 4 (atm) 0 200 400 0 0.5 1 pCO₂ t (d) pCO 2 (atm) 0.04 0.06 0.080 100 200 300 kmax hyd d−1 0.5 1 1.5 0 200 400 kmax meth d−1 0 2 4 0 100 200 300 kmax sulph d−1 0 0.005 0.01 0 100 200 300 kmax NH 4 d−1 0 0.1 0.2 0 100 200 300 K_s,meth mol L−1 0 0.5 1 x 10−3 0 200 400 K_s,sulph mol L−1 0.01 0.02 0.030 100 200 300 C_{i,methanogens} mol L−1 0 0.5 1 0 100 200 300 C_{i,sulphate reducers} mol L−1

Correlation between Parameters

0 0.5 1 Ci,sulphate reducers 0 0.020.04 Ci,methanogens 0 5 10 Ks,sulph 0 0.2 0.4 Ks,meth 4 6 8 x 10−3 kmaxNH 4 0 2 4 kmax sulph 0 0.5 1 kmax meth 0.04 0.06 0.08 0 0.5 1 kmax hyd Ci,sulphate reducers 0 0.02 0.04 Ci,methanogens 0 5 10 Ks,sulph 0 0.2 0.4 Ks,meth 4 6 8 x 10−3 k max NH 4 0 2 4 k max sulph 0 0.5 1 k max meth 0.04 0.06 0.08 k max hyd

Summary

“Gray box” Bio-geochemical model is able to describe measured multi-parameter dynamics in Column Experiments of Roberto Valencia.

Limited amount of parameter fitting;

Parameters that fall outside literature range are:

I _{highly correlated with each other (inhibition constants);} I not based on direct measurements (sulphate dynamics);

I indication of mass-transport limitations in model.

Outlook

I _{Test approach in systems where mass-transport is}

included.

I Model will allow prediction of future development of emission potential.

Compounds in Model

Rate Dependent Reactions and Inhibition

Geochemical Equilibria