Content uploaded by Eero Inkeri
Author content
All content in this area was uploaded by Eero Inkeri on Mar 29, 2016
Content may be subject to copyright.
INTEGRATION OF POWER-TO-GAS –PROCESS TO WASTEWATER
TREATMENT PLANT WITH BIOGAS PRODUCTION
Eero Inkeria, Teemu Sihvonenb, Hannu Karjunena, Tero Tynjäläa, Matti Tähtinenb, Robert Weissc
aLappeenranta University of Technology, Skinnarilankatu 34, 53851 Lappeenranta,
bVTT Technical Research Centre of Finland Ltd, Koivurannantie 1, 40400 Jyväskylä, Finland
cVTT Technical Research Centre of Finland, Biologinkuja 5, Espoo, P.O. Box 1000, FI-02044 VTT, Finland
1. Introduction
Increasing the use of renewable energy leads to demand of energy storages. IEA technology roadmap
estimates 310 GW demand for additional grid-connected energy storage for year 2050 due to increas-
ing share of intermittent renewable energy. There are several types of storages, e.g. frequency regula-
tion, arbitrage and seasonal, as well as several technologies to offer these storages. The best solution
depends from the specific system [1]. PtG (power-to-gas) process could produce CH4 by making H2
with electricity and synthesizing it with CO2 to be used for long-term electricity storage and transport
fuel production. End-use and infrastructure are well developed for CH4, but currently PtG is under
development phase and not considered as fully mature nor feasible technology at large scale. [2]
Because of high CO2 content and resulting low capture cost, biogas production could be
one of the cheapest CO2 source for PtG-process [3]. Biogas is used as transport fuel and fuel for elec-
tricity and heat production. It is made from organic materials such as wastewater sludge and agricul-
tural waste by anaerobic digestion. If organic material is in anaerobic conditions with specialized bac-
teria, fermentative metabolic process will decompose organic material and produce biogas. Resulting
biogas consist mainly of CH4 and CO2 [4]. First MW-scale PtG-plant made by Audi utilizes CO2 from
biogas plant nearby [5]. A case study shows 2.8 TWh biogas potential for southern Finland [6].
Large WWTPs (wastewater treatment plant) commonly produce biogas as a part of the
wastewater treatment process. Treatment process aims to purify wastewater so that the quality of the
treated wastewater would not cause negative impacts to the receiving water body. This requires re-
moval of suspended solids, organic material, nutrients and pathogens. Removal system includes phys-
ical, chemical and biological processes like settling and degrading organic material and nitrogen bio-
logically [7]. After the treatment process, effluent is clean enough, but produced sludge is biologically
instable, contains pathogens and has large volume to be treated. In order to create stable and hygienic
end product, anaerobic digestion can be used. [4]
Purpose of this work is to study the WWTP-integrated PtG-process, considering aspects
like operational dynamics, benefits from heat integrations and buffer storage requirements. Modeling
tools are used for simulations at different time scales.
2. Modelling of the integrated system
2. 1 System design
PtG was chosen to be integrated with WWTP because it seems to have certain unique economical and
technical benefits: O2 (oxygen) from electrolysis can be blended to air that is used for aeration in
activated sludge system. Increasing the O2 content decreases the required volume flow rate through
compressors and also the compressor driver power, and required volume of the reactor [8]. If ambient
temperature is low, heating of digester is required [4]. This heat can be provided by waste heats from
electrolysis, methanation and compressors. Biogas is produced through the whole year and offers fairly
constant and cheap carbon dioxide source for methanation. If integration seems feasible, there are
many possible locations around the world, since wastewater treatment plants with biogas production
are very common.
Suomenoja WWTP plant at Espoo covers catchment area for 310 000 people and treats
average water inflow of 100 000 m3/d. Plant has also two anaerobic digesters (2 x 6 000 m3), from
which biogas (10 000 m3/d) is upgraded with water scrubber and compressed into gas grid. This plant
was chosen as basis for a case study because of availability of measurement data. The system studied
in this paper is presented in Figure 1.
Figure 1. Block diagram of the integrated system including main components as well as heat and material flows.
It is build based on Suomenoja plant with some modifications. Alkaline electrolysis is used to produce
H2 and O2. Latter is directly fed to activated sludge process by aeration compressors. Interim storage
is used to provide steady H2 and CO2 flow to methanation. Interim storage has three pressure levels,
with highest pressure of 300 bars. It is also possible to feed H2 directly to gas grid if storage is full.
WWTP has three integration aspects, it consumes O2 in aeration process and produces sludge into
digester, which has to be heated. Digester produces biogas, from which CO2 is captured to buffer
storage. Carbon capture is done with amine scrubber instead of water scrubber that is used for biogas
upgrade at the plant. Water scrubber dilutes captured CO2 with air in regeneration step, so it can’t be
used as it is. H2 and CO2 are fed into fixed bed catalyst methanation reactor, and produced CH4 is dried
and compressed to gas grid together with upgraded CH4 from biogas. Heat integrations are done so
that all cooling flows to compressors and methanation are collected and used to provide regeneration
heat for the amine scrubber. Cooling flow from electrolysis is used for heating of digester.
2. 2 Modeling procedure
Modeling procedure includes dynamical simulations with in-house Seasonal model and Apros process
simulation software. Seasonal model solves mass and energy conservation equations for different pro-
cesses by ideal stoichiometry at one hour time step. Gas compression and storages are modelled using
ideal gas assumption and polytropic compression with constant average heat capacity. Market envi-
ronment that system works on is derived from Nord pool market hourly data history for 2014. Seasonal
model is used for preparing the input time series and for year-long simulations. It does not calculate
WWTP or anaerobic digestion, but uses measured data as boundary conditions for these. Also initial
system configuration design for Apros model, for example capacities of electrolyser, methanation and
interim storage, is designed with Seasonal model
More accurate Apros model solves pressure-flow continuity equations to calculate flows,
temperatures, concentration and pressures of the system with sub-second time step for one week long
simulation periods. The model includes full PtG-process from electricity grid connection to gas net-
work connection with automation control. Detailed description of the Apros model is presented in [9],
[10]. Additional components implemented to Apros are ASM3 and ADM1, which are used to model
WWTP and anaerobic digestion. Detailed description of these models are presented in [11], [12].
Results from week-long simulations by Apros model are used to study physical, second-
to-minute time scale operational dynamics and heat integrations. The model is validated against meas-
ured data from Suomenoja WWTP. Seasonal model boundary conditions are initially set similar to
Apros model, so that accuracy of the Seasonal model can be evaluated. After evaluation, the parame-
ters of seasonal model are adjusted so the produced results agree with Apros model results. Then the
validated seasonal model is used for year-long simulations in order to study hour-to-week time scale
operational dynamics.
2. 3 Inputs and parameters
Input time series for Apros are prepared with Seasonal model. It is assumed that the need to store
energy can be determined based on electricity price, so that when there is surplus of electricity, spot
prices will fall down. Currently the demand for storages is not evident, but this method is still used in
order to give market-related variation for storage requirement. Amount of storage required is deter-
mined by fixed spot price limit, i.e. electrolyser is operating if electricity price is under specified limit.
Electrolyser is also offering FCR-N frequency control service. Required electrolyser load for fre-
quency control is determined by measured grid frequency (year 2014 in Finland), droop curve and
minimum part load of electrolyser. Resulting power input time series of one week time period for
Apros with one minute interval is presented in Figure 2 together with hourly averaged values used in
Seasonal model.
Figure 2. Input time series have one minute time step for Apros model and one hour for Seasonal model.
Input time series for WWTP are based on measurements from Suomenoja plant in Espoo. Some values
are measured at the plant laboratory with three day interval: influent COD, nitrogen and flow rate.
Other measurements are taken with one hour interval, including temperature at the aeration reactor,
aeration power, heating demand for anaerobic digester and biogas production from digester.
Alkaline electrolysis used in the study has a maximum power of 3 MW, minimum part
load of 20 %, operating pressure of 1.0 bar and efficiency of 63 %. Buffer storages try to maintain the
feed flow into the methanation reactor so that reactor pressure would stay at 6.3 bar. High and inter-
mediate pressure H2 storages have maximum pressure of 300 and 20 bar, for volumes of 25 and 15
m3, respectively. Methanation capacity is determined so that feed flow of H2 to reactor is 31 % of
maximum H2 production from electrolysis. This enables stable operation of methanation with inter-
mittent H2 production.
3. Results
3. 1 Results from Apros simulations
In the Apros simulations, the feed flow to the methanation reactor was kept constant. Interim H2 stor-
age was used to stabilize fluctuating H2 flow from the electrolysis to the methanation as presented in
[9]. H2 feed flow to the methanation stays mostly stable during the simulated week leading to stable
CH4 production also stable, as presented in Figure 3. In fast transients during electrolysis shut-down
or start-up, valves and compressors in interim storage are not able to do full stabilization causing fluc-
tuations in methanation feed flow. H2 to CO2 ratio in methanation changes, which leads to incomplete
methanation reaction. Incomplete reaction causes pressure fluctuation in methanation reactor in the
range of 5.3 – 8.0 bar.
On wastewater treatment side of the model it was considered that biogas production and
heating of digester are the most important measures for this study. Thus ASM3 and ADM1 models
were taken as they are parametrised in IWA studies[11], [12]. Sludge from ASM3 side has different
parameters what ADM1 uses. Parameter conversion was made as presented in [13]. The amount of
produced biogas was adjusted to meet the Suomenoja measurement by tuning the amount of inflow to
the digestion. Feed flow to digester was then kept constant during the simulations. Control logics in
WWTP were taken as presented in IWA bench model [11], [12]. As the control logic is not the same
than in the Suomenoja WWT-plant, the simulated results does not fully agree with measured values
during dynamic changes.
Heating demand of the digester (Figure 3) is first covered by heat from cooling flows of
electrolyser. If electrolyser is not operating, heat from methanation and compressors is used also. Heat
is also needed for amine scrubber (MEA), for which there is often not enough waste heat left. Addi-
tional heat has to be provided for example by combusting natural gas. More detailed distribution of
heat flows is presented in Figure 8.
Figure 3. At the left, total available heating capacity of the PtG system has a base load from biogas compression and methanation, and
large peaks because of electrolyser and H2 compression. Heating demand is satisfied only partly. At the right, mass flow rate of CH4
from methanation reactor is fairly stable with some peaks due to fast transients.
3. 2 Validation of Seasonal model
Seasonal model is compared with Apros simulations in order to adjust the parameters for more accurate
results for year-long analysis. Results that are compared include production of gases, pressure of the
storages, power of compressors and heat flows. Validation results for hydrogen production and pres-
sure of the high pressure H2 storage are presented in Figure 4. Coarser time discretization and the
simplified process description of the Seasonal model flattens out the peaks produced by the Apros
model. Despite small deviations, the average results agree well and justify the use of the Seasonal
model for fast year-long analysis. It was assumed that similar behavior can be expected for longer time
periods.
Figure 4. Results from Seasonal model are compared with results from Apros model. At the left is the pressure of high pressure H2
storage and at right is the total compression power.
3. 3 Results from Seasonal model
When same setup as in Apros is used, the problem is to utilize all H2 and CO2 with very limited flexi-
bility of the methanation. Methanation was not able to operate through the year, because H2 storage
run empty several times. Large share of the produced hydrogen has to be compressed directly to gas
grid when buffer storage is full and only one third of the captured CO2 is converted to CH4. Better
operation would require larger buffer storages or more flexibility to the system. Flexibility could be
introduced by enabling shut-downs and start-ups for methanation, or making part-load operation pos-
sible.
Effect of O2 addition to the WWTP aeration process was found to be relatively small.
Aeration requires O2 about 9500 kg/h and 3 MW electrolyser produces O2 only about 450 kg/h, which
is 4.7 % of the demand. Therefore all O2 could be fed to aeration, but no big savings in aeration power
can be expected.
If methanation has to operate all the time at full capacity, the capacity has to be chosen
so that it matches to annual average production of H2 i.e. full load hours of electrolyser. The effect of
enabling part load capacity of 50 % for methanation was also studied. Third studied option was cyclic
operation with the ability to shut down and start up the methanation, with minimum uptime of 6 h and
no possibility for part load. The relation between full load hours and different options for methanation
capacities is presented in Figure 5 together with resulting spot price limits to operate electrolyser.
Figure 5. Methanation capacities for constant operation with 100 % part load minimum, constant operation with 50 % part load minimum
and cyclic operation with 100 % part load minimum are presented at the left figure. The relation between electrolyser full load hours
and resulting electricity spot price limit at 2014 in Finland is presented at the right figure.
For 3 MW electrolyser, if methanation is operated constantly with full capacity, buffer storages as
large as 1400 m3 would be required to stabilize H2 flow to methanation. Requirement is highly de-
pending on full load hours and maximum power of electrolyser, as presented in Figure 6. Very high
and very low full load hours leads to smaller H2 storages than in the middle range. CO2 source is rather
constant but also limited in quantity for higher methanation capacities. When methanation capacity is
low, there is usually overproduction of CO2 and only very little storage is required. By increasing
methanation capacity and CH4 production, increases also demand for CO2 storage. If methanation
operates as cyclic, constant storage of 12 m3 is enough to satisfy minimum 6 h operation cycle.
Figure 6. Case with 3 MW electrolyser. Constant operation of methanation with no possibility to part load requires very large buffer
storages for H2, while enabling 50 % part load lowers the storage demand significantly. Cyclic operation of methanation requires only
15 m3 high pressure storage for H2. Demand for CO2 storage tends to increase with increasing full load hours, because CO2 production
is limited.
Enabling 50 % part load for methanation makes constant operation possible with smaller buffer H2
storage, but drawback was that there are more often times when storage is full. Therefore more H2 is
fed directly to gas grid, which may not be feasible due to limit for H2 content in the grid. If there is no
part load possibility in methanation, H2 storage becomes so large that only little amount of H2 goes to
gas grid. Cyclic operation of methanation is so flexible that almost no H2 have to be compressed to gas
grid. Relation of full load hours to H2 compressed to the gas grid and resulting CH4 production is
presented in Figure 7.
Figure 7. With constant methanation, smaller H2 storage size increases the amount of H2 that has to be compressed directly to gas grid,
because storage gets full more often. With cyclic operation there is no this problem. Resulting production of CH4 is presented at right.
Heating demand of the digester and amine scrubber are fairly stable through the year, with measured
total demand for digester 5700 MWh/a. That is the upper limit for saved heat compared to the conven-
tional setup of Suomenoja WWTP. Amine scrubber requires heat about 1860 MWh/a. With low oper-
ation time, PtG-system can replace only part of the heat demand, and additional heat for amine scrub-
ber has to be provided by combustion of natural gas. Net sum of replaced and additional heat for
different full load hours is presented in Figure 8. There is almost no difference between constant and
cyclic operation of methanation. Also an example about distribution of total waste heat flow provided
by Apros simulation is presented in Figure 8. Average heat flows from methanation, all compressors
and electrolyser are 0.15 MW, 0.13 MW and 0.52 MW, respectively. Average values are highly de-
pendent of full load hours of the PtG-system.
Figure 8. At left is heat balance between savings in digester heating and additional heat for amine scrubber. Additional heat for both is
provided by combusting natural gas. Distribution of the total waste heat flow from PtG-system is presented at right.
4. Discussion
In this study, PtG integration to WWTP was studied by modelling. Studied process was not optimized
and economics of the system was not considered. Therefore detailed analysis of the system feasibility
cannot be made. However, some general conclusions about the system performance and operation can
be drawn based on simulations. O2 mass flow is relatively small for continuous feed into aeration, so
there might be better use for O2, for example selling it in bottles or usage of O2 storage that could be
used for peak loads at WWTP. Heat integrations seems to have good value by replacing natural gas
heating of digester. On the other hand, CO2 capture with amine scrubber requires additional heat. It
might be better to use alternative CO2 capture method, such as pressure swing adsorption (PSA), that
would require less heat and more electricity instead. Simple heat storage could be used also for deliv-
ering heat more uniformly. System operation and profitability depends strongly on market environment
and especially on prices of electricity and frequency regulation services. Historical market data used
in the study does not reflect the full potential of the process in the future, when the intermittent renew-
able energy production is expected to have significantly larger market share.
5. Conclusions
If WWTP integrated with biogas plant treats wastewater from 300 000 people, maximum amount of
CH4 that can be produced from biogas CO2 content is about 13 800 MWh/a. In order to produce that
amount of CH4, 3 MW electrolyser could be used. There is a possibility to save 5700 MWh of natural
gas that is used to heat the digester with waste heat from PtG-system. Intermediate H2 and CO2 storages
can even out the fluctuations of gas sources and maintain constant methanation, but resulting storage
sizes might be too large to be feasible.
Acknowledgment
We acknowledge Dr Ulf Jeppsson and his colleagues at IEA, Lund University, Sweden for their kind
support providing Matlab files of ADM1 and ASM3. The authors also gratefully acknowledge the
public co-financing of Tekes, the Finnish Funding Agency for Innovation, for the 'Neo-Carbon Energy'
project under the number 40101/14, and the staff at Suomenoja wastewater treatment plant and Hel-
sinki Region Environmental Services Authority HSY about valuable information and measurement
data for this study.
References
[1] IEA, “Technology Roadmap: Energy storage,” p. 64, 2014.
[2] M. Lehner, R. Tichler, and M. Koppe, Power-to-Gas : Technology and Business Models. 2014.
[3] G. Reiter and J. Lindorfer, “Evaluating CO2 sources for power-to-gas applications – A case
study for Austria,” J. CO2 Util., vol. 10, pp. 40–49, 2015.
[4] A. Haandel and J. Lubbe, Handbook of biological wastewater treatment: Design and
optimisation of activated sludge systems., no. l. 2007.
[5] M. Götz, J. Lefebvre, F. Mörs, A. McDaniel Koch, F. Graf, S. Bajohr, R. Reimert, and T. Kolb,
“Renewable Power-to-Gas: A technological and economic review,” Renew. Energy, no.
AUGUST, 2015.
[6] J. Höhn, E. Lehtonen, S. Rasi, and J. Rintala, “A Geographical Information System ( GIS )
based methodology for determination of potential biomasses and sites for biogas plants in
southern Finland,” vol. 113, pp. 1–10, 2014.
[7] E. Metcalf and H. Eddy, Wastewater engineering: treatment and reuse. 2003.
[8] EPA, “EPA (1973) Oxygen Activated Sludge Wastewater Treatment Systems.” 1973.
[9] Savolainen, J.; Kannari, L.; Pennanen, J.; Tähtinen, M.; Sihvonen, T.; Pasonen, R.; Weiss R.,
“Operation of a PtG plant under power scheduling,” 10th Int. Renew. Energy Storage (IRES
2016) Int. Renew. Energy Storage Conf. Düsseldorf., p. 1, 2016.
[10] Weiss, R.; Savolainen, J.; Peltoniemi, P.; Inkeri E., “Optimal scheduling of a P2G plant in
dynamic power , regulation and gas markets,” 10th Int. Renew. Energy Storage (IRES 2016)
Int. Renew. Energy Storage Conf. Düsseldorf., p. 2.
[11] C. Rosen and U. Jeppsson, “Aspects on ADM1 Implementation within the BSM2 Framework,”
Tech. Rep., pp. 1–37, 2006.
[12] U. Jeppsson, “A General Description of the Activated Sludge Model No . 1 ( ASM1 ),”
Components, vol. 1, no. 1, pp. 1–16, 1996.
[13] C. M. Lopez-Vazquez, M. Mithaiwala, M. S. Moussa, M. C. M. van Loosdrecht, and D.
Brdjanovic, “Coupling ASM3 and ADM1 for wastewater treatment process optimisation and
biogas production in a developing country: case-study Surat, India,” J. Water, Sanit. Hyg. Dev.,
vol. 3, no. 1, p. 12, 2013.