Dynamical modeling of substrate and biomass effluents in up-flow anaerobic sludge blanket (UASB) biogas reactor
- 416 Downloads
Abstract
Organic liquid waste from food production industry is inevitable. High chemical oxygen demand (COD) contents in organic liquid waste could disrupt the water ecosystem. On the other hand, COD contents can be reduced and utilized to produce biogas by UASB reactor. However, there is a problem in operating UASB reactor, namely the high biomass content in methanogenic granule form, which is washed out with the effluent. The influent flow rate affects biomass content and the suitable flow rate is important for the particular UASB reactor. To investigate the matter, the estimation of Monod parameters is determined to study the kinetics of substrate (COD) and biomass (active methanogenic granule). In this work, simulations of lumped and distributed models are performed to observe the behavior of substrate and biomass inside the reactor. It is concluded that the suitable influent flow rate for UASB reactor is 150–175 m^{3}/h, and the washed out biomass content is relatively low (from 0.001393 to 0.4919 kg/m^{3}). The steady-state condition is achieved from 2027 to 2533 days, with high COD removal.
Keywords
Anaerobic process Biomass content Chemical oxygen demand UASB reactorIntroduction
There are several parameters to determine water quality, i.e., COD, BOD (Biological Oxygen demand), DO (Dissolved Oxygen) and total amount of solute. High COD content on organic liquid waste could disrupt water environment ecosystem, because high COD content water tend to have low oxygen content. After COD is degraded aerobically, it will produce carbon dioxide and sediment, and for anaerobic processes, methane gas is released and would deplete the Earth’s atmosphere [1].
UASB reactor (Upflow Anaerobic Sludge Blanket) is a type of widely used biogas reactors to treat organic liquid waste with high efficiency of 70–90% [2]. During the biogas production process, the UASB reactor utilizes methanogenic bacteria which form granules as a medium to decompose COD into methane and carbon dioxide. The methane gas is then collected and used as fuel for various industrial purposes.
One of the 10 UASB reactors in Indonesia is able to produce as much as 30,000 Nm^{3} methane gas per day from pineapple and tapioca liquid waste. The produced biogas is then used to heat cassava for tapioca production process and as fuel of combined heat and power plant [3]. Yet, the considerable amount of biomass content in the granule form is released along with effluent flow. The clean waste should contain the treated liquid waste (with low COD content) and inactive biomass. In this case, the biomass that is released with effluent is composed by the dead or inactive bacteria, which is lifted upward due to gas composition within the granule [4].
The high released biomass contents in effluent flow are often formed by the short hydraulic retention time (HRT) or high influent flow rate [5], the overcapacity of biomass inside the reactor [6], or the operation during the start-up of granular reactors which would unintentionally resulted in the reduction of process performance [7].
As a hypothesis, it is estimated that the amount of wasted biomass is caused by too high influent flow rate. In the previous study by Bolle et al., an experimental method was conducted to determine the relationship between flow rate and reactor height to avoid short circuit flow which resulted in non-treated substrate in liquid waste [8], but the effect on biomass concentration of effluent has not been explained yet.
In this research, the UASB reactor was modeled as multilevel CSTR. The Monod parameters for the kinetics of substrate and biomass were determined from the available data. The simplified model was investigated firstly for observing the behavior of substrate and biomass in the reactor, and then the research was extended by implementing distributed models to take the height of the reactor into account.
Methodology
Data parameters of biogas plant
This research requires data for influent substrate concentration \((S_{0} )\), effluent substrate concentration \((S_{\text{e}} )\), effluent biomass concentration \((X_{\text{e}} )\), biomass volume inside the reactor \((V_{\text{b}} )\), and influent flow rate \((Q)\). The data were collected in 2 years (January 2015–December 2016). The method of simplified linear regression model (SLRM) was applied to extract the Monod parameters. The important information of bacterial specific growth rate, saturation coefficient, bacterial decay coefficient, and yield coefficient can be obtained by utilizing the kinetics models since microbial growth is also an autocatalytic reaction.
Estimation of bacterial reaction kinetics with simple linear regression method
By utilizing kinetic models, the important information of bacteria can be obtained, such as bacterial specific growth rate, saturation coefficient, bacterial decay coefficient, and yield coefficient. To obtain the kinetic model parameters of the UASB reactor, a simple linear regression method can be applied as also implemented by Matangue et al. [9] and Bhunia and Ghangrekar [10].
Simulation of kinetic reaction with UASB reactor: the case of the segmented model
Segmented model validation
Preliminary model parameters
No. | Parameter | Symbol | Value | Unit |
---|---|---|---|---|
1 | Reactor volume | \(V\) | 2040.88 | m^{3} |
2 | Influent substrate concentration | \(S_{0}\) | 11 | kg/m^{3} |
3 | Yield constant | \(Y_{i}\) | 0.28047 | kg VSS/kg COD |
4 | Decay constant | \(K_{\text{d}}\) | 0.00070 | h^{−1} |
5 | Maximum specific growth rate | \(\mu_{ \text{max} }\) | 0.001078 | h^{−1} |
6 | Monod constant | \(K_{\text{S}}\) | 0.93 | kg/m^{3} |
7 | Initial biomass concentration | \(X_{0}\) | 167.68 | kg/m^{3} |
8 | Washout constant | \(w_{i}\) | 0.0002 | – |
9 | Flow rate | \(Q\) | 153 | m^{3}/h |
10 | Cross-sectional area of UASB reactor | A | 510.33 | m^{2} |
Main model simulation (bacteria and substrate concentration model on UASB reactor as a function of reactor height and time)
The initial substrate concentration is S_{0}. As the substrate enters the reactor, the substrate concentration becomes the concentration at 0 m (z = 0 m), reduced by the mass transport factor due to dispersion. At the top (z = 4.8 m), the concentration gradient is equal to zero. In this case, the above considerations are formulated for the initial and boundary conditions as follows:
Since the model is a system of partial differential equations, solving the simulation with numerical method is necessary, i.e. the system was then represented and solved by the simulink program.
Main model parameters
No. | Parameter | Symbol | Value | Unit |
---|---|---|---|---|
1 | Reactor height | H | 4.8 | M |
2 | Influent substrate concentration | \(S_{0}\) | 11 | kg/m^{3} |
3 | Yield coefficient | \(Y\) | 0.28047 | kg VSS/kg COD |
4 | Decay constant | \(K_{\text{d}}\) | 0.00070 | h^{−1} |
5 | Maximum Specific growth rate | \(\mu_{ \text{max} }\) | 0.001078 | h^{−1} |
6 | Monod constant | \(K_{\text{S}}\) | 0.93 | kg/m^{3} |
7 | Initial biomass concentration | \(X_{0}\) | 167.68 | kg/m^{3} |
8 | Washout constant | \(w\) | 0.0002 | – |
9 | Up-flow rate | \(q\) | 0.299871 | m/h |
10 | Dispersion constant | D_{(z)} | \(3.10^{ - 5} {\text{e}}^{ - 0.981z}\) | m^{2}/s |
Applying various influent flow rate on preliminary model and main model
Preliminary model parameter variation
No. | Q (m^{3}/h) | \(v_{\text{up}}\) (m/h) | D (m^{2}/h) | Pe | N |
---|---|---|---|---|---|
1 | 50 | 0.0980 | 0.016 | 29.39908 | 16 |
2 | 75 | 0.1470 | 0.026 | 27.13761 | 15 |
3 | 100 | 0.1960 | 0.035 | 26.87916 | 15 |
4 | 125 | 0.2450 | 0.045 | 26.13252 | 15 |
5 | 150 | 0.2940 | 0.056 | 25.19921 | 14 |
6 | 175 | 0.3430 | 0.066 | 24.94468 | 14 |
7 | 200 | 0.3920 | 0.077 | 24.4356 | 14 |
8 | 225 | 0.4410 | 0.087 | 24.33028 | 14 |
9 | 250 | 0.4900 | 0.0981 | 23.97479 | 13 |
Main model parameter variation
No. | Q (m^{3}/h) | \(v_{\text{up}}\) (m/h) | D(z) (m^{2}/s) |
---|---|---|---|
1 | 50 | 0.097997 | \(9.10^{ - 6} {\text{e}}^{ - 0.981z}\) |
2 | 75 | 0.146995 | \(1.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
3 | 100 | 0.195994 | \(2.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
4 | 125 | 0.244992 | \(2.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
5 | 150 | 0.293991 | \(3.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
6 | 175 | 0.342989 | \(4.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
7 | 200 | 0.391988 | \(4.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
8 | 225 | 0.440986 | \(5.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
9 | 250 | 0.489985 | \(5.10^{ - 5} {\text{e}}^{ - 0.981z}\) |
Results and discussion
Based on the governing equations, the calculation and simulation were conducted to estimate the reaction parameters. The models of substrate and biomass concentration are functions of time, or height and time.
Monod parameter estimation
The Monod parameters for bacteria are estimated using Linear Regression Method based on the laboratory data. The data are initial bacterial concentration, initial substrate concentration, substrate concentration of effluent, reactor volume and influent flow rate for 99 weeks from the start-up. The results are depicted as follows.
In this case, the mean value of \(\mu_{ \text{max} }\) is then 0.001071.
The segmented model validation
A validation of the simplified model is carried out to ensure that the model represents the condition of the biogas plant. The applied estimated parameters will produce the time dependent substrate of reactor 1 until reactor 14 as follows.
Steady condition is reached after the reactor operates for 50,770 h or 2115 days. Under this condition, the bacteria in reactor 1 are able to degrade COD by 57.51% and the concentration becomes 4.674 kg/m^{3}. In overall, the COD is degraded by 97.81% with effluent substrate concentration of 0.2409 kg/m^{3}.
The concentration of biomass refers to the concentration of active microorganisms. It can be observed that the growth rate of organism in reactor 1 is the fastest compared to other reactors, due to high substrate concentration. In this case, the plant data show that a COD removal is of 7.521–10.910 kg/m^{3} with 95% confidence level and 1.946 of standard deviation; whereas the simulation result is 5.122–14.504 kg/m^{3}. The average ratio of plant data with simulation result is 1.17 which shows the validity of the model.
The distributed model validation
Most of the substrate is degraded at the bottom of the reactor. The figure shows that at the 80-day period, the existence of concentration gradient is observed at the height of 3.5 m. For the next operation time, the concentration gradient becomes higher. Up to the 400-day period, most of the substrate has been degraded before reaching the height of 1 m.
Figure 8 shows that most of the biomass is at the bottom of the reactor. After the reactor operates for 400 days, the increase of biomass concentration from 167.68 to 2460 kg/m^{3} is observed.
On the observed full-scale reactor, weekly measurement of biomass concentration is carried out for 7 sampling points which are representing the conditions of biomass concentration at some certain reactor heights, namely, 0, 0.3, 0.6, 0.9, 1.2, 2, 3, and 4 m. For each height, a sample of 1 L—which composed of active methanogenic granule which considered as biomass, inactive granule and substrate—are taken, then for each sample, flushing is carried out to remove floating methanogenic graule which is considered inactive. Then, the measurement of biomass/sludge weight and volume was conducted by following the SV60 method (heating samples with a temperature of 100 °C for 60 min). Thus, the biomass concentration was obtained for each sampling point.
The data of full-scale UASB reactor for 240 days
Reactor height (h) | Biomass concentration (kg/m^{3}) |
---|---|
0 | 931.5 |
0.3 | 931.5 |
0.6 | 931.5 |
0.9 | 879.75 |
1.2 | 905.625 |
2 | 310.5 |
3 | 284.625 |
4 | 33.12 |
The data of full-scale UASB reactor for 320 days
Reactor height (h) | Biomass concentration (kg/m^{3}) |
---|---|
0 | 1037 |
0.3 | 1037 |
0.6 | 1037 |
0.9 | 1037 |
1.2 | 725.9 |
2 | 114.07 |
3 | 93.33 |
4 | 31.11 |
After the validation of simplified and distributed model, the apparent deviation is due to the different model and assumption for each model. In this case, the distributed model should be more accurate than the simplified model.
The results with influent flow variations
After the validation, the variation of the influent flow rate can be predicted with a calculable accuracy. In this research, 9 variations are performed on the segmented model simulation. The simulation was conducted by the parameters of reactor volume, influent substrate concentration, yield constant, decay constant, maximum specific growth rate, Monod constant, initial biomass concentration, washout constant, and cross-sectional area of UASB reactor as shown in Table 1. Then, the results are tabulated as follows.
In the segmented model, different COD concentrations of the effluent are obtained along with different HRT. It shows that the COD concentration of the effluent increases proportionally with the influent flow rate. Lower values of HRT cause the substrate leaving the reactor before it is completely degraded. A storage pool is required to hold the substrate before the liquid waste is released into the environment. The pool keeps the liquid waste for 3 months to further remove the COD. Therefore, higher storage pool is required for higher influent flow rate.
For the steady-state operation, higher influent flow rate resulted in lower COD concentration. This is due to higher influent flow rate, which more substrate is processed by methanogenic granule. This is different from the transient operation in which the biomass is not fully functional yet.
Conclusions
The Upflow Anaerobic Sludge Blanket (UASB) biogas reactor is modeled and investigated in this research. Two types of model which are lumped/simplified and distributed model are implemented. The linear regression analysis is applied to estimate the Monod parameters. The results show that the influent flow rate affects the concentration of released biomass. It is found that influent flow rate is proportional to the released biomass concentration. However, the opposite result happens in the transient condition which the leaving substrate from reactor is not completely degraded. The models also predict that the suitable influent flow rate is from 150 to 175 m^{3}/h, where fairly few bacteria concentration is released (0.001393–0.4919 kg/m^{3}). The range of the obtained flow rate will require settling time of 2027–2533 days.
Flow rate variations
No. | Influent flow rate (m^{3}/h) | Upflow velocity (m/h) | HRT (h) | COD at HRT (kg/m^{3}) | COD removal when the first HRT reached (%) | Transient time (day) | Substrate concentration at reactor 1 for steady-state operation (kg/m^{3}) | COD removal at reactor 1 for steady-state operation (%) | Substrate concentration at reactor 14 for steady-state operation (kg/m^{3}) | COD removal at reactor 14 for steady-state operation | Biomass concentration at reactor 1 for steady-state operation (kg/m^{3}) | Biomass concentration at reactor 14 for steady-state operation (kg/m^{3}) |
---|---|---|---|---|---|---|---|---|---|---|---|---|
1 | 50 | 0.098 | 48.98 | 0.00 | 100.00 | 935 | 2.362 | 78.53 | 0.4758 | 95.67 | 1236.00 | 9.460E−05 |
2 | 75 | 0.147 | 32.65 | 0.063 | 99.42 | 1063 | 2.738 | 75.11 | 0.3906 | 96.45 | 1598.00 | 3.074E−04 |
3 | 100 | 0.196 | 24.49 | 0.897 | 91.85 | 1273 | 3.228 | 70.65 | 0.3217 | 97.08 | 1904.00 | 5.757E−04 |
4 | 125 | 0.245 | 19.59 | 2.386 | 78.31 | 1615 | 4.012 | 63.53 | 0.2670 | 97.57 | 2072.00 | 6.684E−04 |
5 | 150 | 0.294 | 16.33 | 3.677 | 66.58 | 2027 | 4.560 | 58.55 | 0.2488 | 97.74 | 2089.00 | 1.393E−03 |
6 | 175 | 0.343 | 13.99 | 4.641 | 57.81 | 2533 | 5.659 | 48.55 | 0.2221 | 97.98 | 1955.00 | 4.919E−03 |
7 | 200 | 0.392 | 12.25 | 5.386 | 51.04 | 3270 | 7.043 | 35.97 | 0.2018 | 98.17 | 1609.00 | 1.600E−02 |
8 | 225 | 0.441 | 10.88 | 5.996 | 45.49 | 3362 | 8.413 | 23.52 | 0.1712 | 98.44 | 1161.00 | 4.867E−02 |
9 | 250 | 0.490 | 9.796 | 6.494 | 40.97 | 3440 | 8.724 | 20.69 | 0.1711 | 98.44 | 1050.00 | 1.743E−01 |
Notes
References
- 1.Chetty S, Pillay K (2015) Application of the DIY carbon footprint calculator to a wastewater treatment works. Water SA 41(2):263–272CrossRefGoogle Scholar
- 2.Wagiman (2007) Identifikasi potensi produksi biogas dari limbah cair tahu dengan reaktor up-flow anaerobic sludge blanket (UASB). Bioteknologi 4(2):41–45Google Scholar
- 3.Karthikeyan K, Kandasamy J (2009) Upflow anaerobic sludge blanket (UASB) reactor in wastewater treatment. Water Wastewater Treat Technol 2:80–198Google Scholar
- 4.Rodriguez-Gomez R (2011) Upflow anaerobic sludge blanket reactor: modelling. Thesis, Royal Institut of Technology, Stockholm, SwedenGoogle Scholar
- 5.Moharram MA, Abdelhalim HS, Rozaik EH (2016) Anaerobic up flow fluidized bed reactor performance as a primary treatment unit in domestic wastewater treatment. HBRC J 12(1):99–105CrossRefGoogle Scholar
- 6.Van Lier JB, Vashi A, van der Lubbe J, Heffernan B (2010) Anaerobic sewage treatment using UASB reactors: Engineering and operational aspects. Environmental Anaerobic Technology: Applications and New Developments, pp 59–89Google Scholar
- 7.Szabo E, Hermansson M, Modin O, Persson M, Britt-Marey W (2016) Effects of wash-out dynamics on nitrifying bacteria in aerobic granular sludge during start-up at gradually decreased settling time. Water 8(5):172CrossRefGoogle Scholar
- 8.Bolle WL, van Breugel J, van Eybergen GG, Kossen NW, Zoetemeyer RJ (1986) Modeling the liquid flow in up-flow anaerobic sludge blanket reactors. Biotechnol Bioeng 28(11):1615–1620CrossRefGoogle Scholar
- 9.Matangue MTA, Campos CMM (2011) Determination of kinetic parameters of an upflow anaerobic sludge blanket reactor (UASB), treating swine wastewater. Cienc Agrotecnol 35(6):1204–1210CrossRefGoogle Scholar
- 10.Bhunia P, Ghangrekar MM (2008) Analysis, evaluation, and optimization of kinetic parameters for performance appraisal and design of UASB reactors. Bioresour Technol 99(7):2132–2140CrossRefGoogle Scholar
- 11.Rodriguez-Gomez R, Moreno L, Liu L (2013) A model to predict the behavior of UASB reactors. Int J Environ Res 7(3):605–614Google Scholar
- 12.Pena MR, Mara DD, Avella GP (2006) Dispersion and treatment performance analysis of an UASB reactor under different hydraulic loading rates. Water Res 40:445–452CrossRefGoogle Scholar
- 13.Pavlostathis SG, Giraldo-Gomez E (1991) Kinetics of anaerobic treatment. Water Sci Technol 24(8):35–59CrossRefGoogle Scholar
Copyright information
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.