the energy balance in farm scale anaerobic digestion

8/13/2019 The Energy Balance in Farm Scale Anaerobic Digestion

1/8


2/8

In the present study the effects of temperature on themethane production rate and methane yield in the digestion of

straw and sugar beet tops were investigated. The energy output

was compared with the energy input to determine the

temperature at which the best energy balance could be

achieved. The experimental results were obtained in pilot-

scale reactors, whereas the energy balances were calculated

both for the pilot scale reactors and for theoretical examples of

farm-scale systems.

2. Materials and methods

2.1. Pilot-scale reactors

The study was performed in two identical reactor tanks, each with an active

volume of 1.8 m3 and a total volume of 2.2 m3, that were stirred continuously.

The tanks were constructed of stainless steel (6 mm) and were insulated by a

100 mm layer of mineral wool. They were equipped with eccentricscrewpumps

(AB TELFA, Goteborg) used for feeding and recirculation, and with stirrers

(Mamic OY, Finland), pH probes (MiniCHEM, TPS, Australia), temperature

regulation, slurry-level indicators and gas volume meters (Gallus, G1.6 1R,

Euromekanik AB, Sweden). The temperature of the one reactor (reactor 1) was

regulated by a heat exchanger, whereas the temperature of the other (reactor 2)

was regulated by a heating coil placed in the bottom and the lower part of the

walls of the reactor. Fig. 1 shows a schematic view of reactor 2.

2.2. Substrate

The substrate consisted of sugar beet tops that were ensiled on a bedding of

wheat straw and were stored in bunker silos. For the substrate from harvest year

1, sugar beet tops together with the straw, which was used as bedding, was used.

For the substrate from the harvest year 2 only sugar beet tops were used, since

the straw had decomposed during storage, so that the methane potential of the

straw was considered to be too low. The substrate was prepared once a week by

mixingbeet tops and straw (ifadded)with water to obtain a slurrywith a volatile

solids (VS) content of 4.25.6%. The average content of total solids (TS), VS,

lactic acid and volatile fatty acids (VFAs), respectively, for the substrate that

was prepared during months 1724 is shown in Table 1.

2.3. Operation of the pilot-scale reactors

The reactors were initially inoculated with digestate from an unheated

anaerobic digester fed with cow manure (Onnestad, Sweden). The reactors were

fed semicontinuously, either twice a week or once a day. During the startup

period (months 14) the operational temperature of both reactors was

20 2 8C. During months 514 the temperature in reactor 1 was slowly

decreased to 11 8C, whereas in reactor 2 it was gradually increased to

37 8C. During this period the reactors were fed 70 l of substrate twice a week,

resulting in an average organic loading rate (OLR) of 0.50.6 kg VS m3 day1

(where m3 refers to the active reactor volume) and a hydraulic retention time

(HRT) of 90 days (Table 2). The substrate from harvest year 1 was used until

month 11, and the substrate from harvest year 2 was used during the rest of the

experiment.

Until month 15, the amount of substrate fed was controlled by the level

indicator, and the amount of effluent was measured by timing the pump. During

month 15 an extra tank equipped with a weighing cell (shown in Fig. 1) was

installed, allowing the amount of substrate fed to be controlled by weight,

although the amount of effluent emptied from the reactor was still measured by

timing.

During months 1624 the operational temperatures were set to 15 8C and

30 8C for reactor 1 and 2, respectively. During month 16 the reactors were fed

70 l of substrate twice a week. During month 1724 the reactors were fed once a

day and OLRs from 1 to 4 kg VS m3 day1 was applied to determine the

maximum OLRs at 15 8C and 30 8C, respectively (Table 2). The initial OLR for

this period was 1 kg VS m3 day1 for both reactors. This was increased to 2.1,

but sincereactor 1 was overloaded at thisOLR,the amount was decreased to1.6

for this reactor, which was then run at this OLR for the rest of the experiment.

The OLR for reactor 2 was first increased to 3.3 kg VS m3 day1 and then to

4.1 kg VS m3 day1 (Table 2).

2.4. Monitoring the process

The data on temperature, pH, volume of gas produced and energy input for

heating(only forreactor2) wasmonitored on-lineusing thePLC system SLC5/05with RSLogix 500 as software (Rockwell Automation) for data collection and

Citect5(AuticSystemAB,Landskrona,Sweden)astheHumanMachineInterface

software. Thecontrolsystemwas alsoused to controlthe feedingand temperature.

The methane content of the biogas produced was measured on-line (but not

continuously)using a methanedetectorbased on infraredlight absorption (Simrad

GD10 IR Gas Detector, Safetech HB, Vastra Frolunda, Sweden).

The alkalinity and the concentration of lactic acid and of VFAs (acetic acid,

propionic acid, butyric acid and valeric acid) were measured off-line. Sampling

and analysis were carried out as described by Bjornsson et al.[15], except that

the alkalinity was measured using a Scott Titroline titrator (Tillquist, Sundby-

berg, Sweden). VS and TS were analyzed as described in APHA[16]. From the

alkalinity that was measured, a was calculated as the difference between total

alkalinity and partial alkalinity divided by the partial alkalinity[17]. A value of

aof less than 1 indicates a stable process, whereas a value above 1 is a sign of

instability[18].

2.5. Energy consumption

2.5.1. Energy for heating

During operation of the pilot scale reactors, the energy consumed by the

heating of reactor 2 was monitored. The energy consumption at each loading

rate was plotted against the difference in temperature between the reactor and

I. Bohn et al. / Process Biochemistry 42 (2007) 576458

Fig. 1. Schematic view of the pilot-scale reactor (reactor 2) after installation of

the new feeding system, where a weighing cell is used to control the amount of

feeding. 1: heating coil in the reactor, 2: pump for feeding and circulation, 3:

pump for feeding, 4: cutting mixer for preparation of the substrate, 5: weighing

cell for control of the amount of slurry fed to the reactor.

Table 1

Average values and standard deviations (shown in brackets) for the TS, VS, lactic acid and VFA content of the feed slurry prepared for month 1724

TS (%) VS (%) Lactic acid (g l1) Acetic acid (g l1) Propionic acid (g l1) Butyric acid (g l1) Valeric acid (g l1)

6.4 (0.54) 4.7 (0.38) 2.9 (1.20) 4.0 (0.86) 2.3 (0.54) 1.7 (0.61) 1.4 (0.72)


3/8


4/8

other results was obtained with the substrate from the year-1

harvest. The yield increased with the temperature and the

highest yield obtained was 0.40 m3 CH4kg VS1 at 30 8C. At

11 8C, the yield was only one-third of that obtained at 30 8C. At

temperatures above 30 8C the yield decreased. During months

514, the operation of both reactors was stable, although some

accumulation of VFAs occurred. In reactor 2, the pH increased

with increasing temperature from pH 7.3 at 24 8C to 7.6 at

37 8C. At temperatures below 30 8C, the acetic acid concen-

tration was generally below 0.05 g l1, but it occasionally

increased up to a level of 0.20 g l

1

. The concentration ofpropionic acid varied between 0.03 and 0.22 g l1. At 30 8C

and higher, acetic acid and propionic acid were rarely detected,

and then usually at concentrations of less than 0.10 g l1. In

I. Bohn et al. / Process Biochemistry 42 (2007) 576460

Table 4

Temperatures in an unheated farm-scale reactor during the course of a year

Temperature (8C)

Soil

(at 1 m depth)aManure tankb Assumed

operational

temperature

January 4 79February 4 n.m.

March 5 n.m.

April 3 n.m.

May 9 n.m.

June 11 n.m. 11

July 15 1720 17

August 17 1720 18

September 14 1517 15

October 13 1214 13

November 8 n.m.

December 4 n.m.

a Monthly average as calculated from Nimmermark[23].b Measured data from an unheated digester tank for cow manure (Onnestad,

Sweden).

Fig. 2. Methane yield obtained at different temperatures at an OLR of 0.5

0.6 kg VS m3 day1 and an HRT of 90 days. Results for 1118 8C were

obtained with reactor 1 and results for 2037 8C with reactor 2. Results for

11 8C, 12 8C and 3227 8C were obtained with the substrate from harvest year

2. The other results were obtained with the substrate from harvest year 1. The

error bars show the standard deviations;n = 2 for 18 8C, 22 8C, 24 8C, 26 8C,

27 8C, 32 8C, 35 8C and 37 8C; n= 3 for 12 8C and 28 8C; n= 4 for 13 8C,

25 8C, and 30 8C;n= 5 for 14 8C;n= 6 for 20 8C;n= 7 for 11 8C andn= 8 for

17 8

C. T

able5

Monitoredparametersofthebiologicalproces

s

OLR

(kgVSm3

day1)

pH

a

Aceticacid(gl1)

Propionica

cid(gl1)

Methaneyield

(m3

CH4

kgVS1)

Methaneproductionrate

(m3

CH4

m3

day1)

Minimum

Max

imum

Minimum

Maximum

Minimum

Maximum

Minimum

Maximum

Reactor1

0.5

7.4

7.5

0.2

7

0.3

2

0.0

7

0.2

8

0.0

9

0.2

0

0.2

3

0.1

1

1.0

7.4

7.5

0.3

1

0.3

9

0.3

7

0.7

2

0.2

5

0.7

7

0.1

7

0.1

7

1.6

7.1

7.2

0.5

9

1.0

1

2.3

0

3.1

1

1.0

1

2.2

8

0.1

4

0.2

2

2.1

7.2

7.5

b

0.4

6

0.9

1a

1.1

0

2.7

4

1.9

7

3.0

3

0.1

1

0.2

4

Reactor2

0.5

7.4

7.4

0.2

4

0.3

2

0

0.0

8

0

0.0

4

0.3

9

0.1

8

1.0

7.4

7.5

0.2

2

0.3

7

0

0.0

8

0

0.0

1

0.3

2

0.3

3

2.1

7.4

7.6

0.2

3

0.3

2

0.0

2

0.7

8

0.0

1

0.2

5

0.3

2

0.6

6

3.3

7.5

7.6

0.2

7

0.3

2

0.0

4

0.6

0

0.0

2

0.1

1

0.2

9

0.9

8

4.1

7.0

7.2

b

0.3

2

1.0

4a

1.4

4

3.1

6a

0.3

1

0.7

1a

0.2

4

0.9

7

a

Increasingconcentrationduringtheperiod.

b

Decreasingconcentrationduringtheperiod.


5/8

reactor 1, the pH remained between 7.2 and 7.5 at temperatures

of 1711 8C. At temperatures of 16 8C and above, the

concentrations of acetic acid and of propionic acid were below

0.10 g l1 and 0.13 g l1, respectively. At temperatures of

15 8C and below, the acetic acid and the propionic acid

concentrations were in the range of 0.040.35 g l1 and 0.10

0.38 g l1, respectively.

Table 5 shows the results of increasing the OLR at

operational temperatures of 15 8C and 30 8C, respectively,

obtained during months 1624. The reasons for higher values

being obtained in reactor 1 for acetic acid and aat an OLR of

1.6 kg VS m3 day1 than for an OLR of 2.1 kg VS m3 day1

was that the reactor was run first at the higher loading rate. The

optimal gas production rates were obtained at the highest

loading rates tested which were 4.1 kg VS m3 day1 for

reactor 2 (30 8C) and 2.1 kg VS m3 day1 for reactor 1

(15 8C). However, there were signs of instability at these

loading rates since acetic acid and propionic acid accumulated,

resulting in a decrease in pH and an increase in a (Table 5). The

optimal loading rates were therefore found to be1.6 kg VS m3 day1 for reactor 1 and 3.3 kg VS m3 day1

for reactor 2. For reactor 1, the methane yield decreased with

increasing OLR. For reactor 2, the methane yield decreased

slightly for OLRs higher than 0.5 kg VS m3 day1, and also

when the reactor was overloaded at 4.1 kg VS m3 day1, but

at loading rates of 13.3 kg VS m3 day1 the yield was fairly

constant.

3.2. Energy balances

The energy input and the energy output for the pilot scale

reactors are shown in Fig. 3. The energy input is dividedinto theenergy needed for feeding, heating and stirring, respectively.

The energy needed for heating the substrate and the digester

tank was estimated from the monitored values and compared to

the calculated values. The monitored values, however, were

found to be unreasonably high, since the energy needed for

heating the substrate was 1.44 times as high as the calculated

value and the energy needed for heating the tank was 5.5 times

as high as the calculated value. Because, as taken up in the

discussion, the heat losses from the pilot-scale reactors

appeared unnecessarily high, and the calculated values were

considered to be more realistic, the latter values were used. The

energy for feeding includes preparation of the substrate slurry,

pumping of the substrate and the effluent, and recirculation of

the reactor slurry before and after feeding. Since for OLRs at 1

4 kg VS m3 day1 feeding occurred once a day, the energy

input for preparing and feeding the substrate only decreased

slightly for an increase in OLR. The energy needed for heating

and for stirring per tonne of VS increased with a decrease in

OLR since HRT increased with a decrease in OLR. Because of

the high energy consumption that stirring and heating required,

the energy balance at an operational temperature of 30 8C is

negative for OLRs of 1 and 0.5 kg VS m3 day1, whereas at

an operational temperature of 15 8C it is negative at all OLRs.

Fig. 4shows the energy balance calculated for a theoretical

farm-scale system in which the reactors are similar to those

used in the pilot-scale experiment but are scaled up and have

improved insulation. For this scenario all the energy balances

are positive with energy inputs ranging from 1926% of the

energy produced.

For an operational temperature of 30 8C, the best net energy

production of 10 GJ tonne (t) VS1

is obtained at0.5 kg VS m3 day1. At OLRs of 13.1 kg VS m3 day1

net energy production is in the range of 8.39.2 GJ tVS1. At

these OLRs the energy needed for heating and stirring per tonne

of VS decreases to only a slight extent since most of the energy

needed for heating is required for heating the substrate. At the

same time the yield is fairly constant.

For an operational temperature of 15 8C both the energy

input and the energy output decreases proportionally with the

HRT, resulting in an energy need of 2023% of the energy

produced for all OLRs. Since the methane yield is highest for

the lowest OLR (0.5 kg VS m3 day1), the energy balance is

best for this OLR, although the net energy output is only 65% ofthat obtained at 30 8C at the same loading rate.

Fig. 5 shows the energy production each month and the

energy consumption, which is constant, for a simple farm-scale

system run at ambient temperatures and at an OLR of

0.5 kg VS m3 day1. The process is assumed to be active

during June through October. The energy output increases with

increasing slurry temperature. The best energy balance is

obtained in August when there is a net energy production of

9.3 GJ tVS1. The average net energy production is

5.8 GJ tVS1. The energy consumption is 1331% of the

energy produced with an average of 21%.

I. Bohn et al. / Process Biochemistry 42 (2007) 5764 61

Fig. 3. The measured energy production and the calculated energyconsumption for the pilot-scale reactors at different OLRs at 15 8C (A) and 30 8C (B)shown as the

energy production ( ) and the energy input for pretreatment and feeding of the substrate (&), for heating ( ) and for stirring (&).


6/8


7/8

may lead to crust formation, however [21], a matter which

should be borne in mind.

4.2. Energy balances

Although the energy balance of a pilot scale reactor cannot be

expected to adequately reflect conditions present in a full-scale

reactor, it was calculated because empirical values were

available. Since the values used for heating and for stirring

were found to be unrealistically high for such a pilot scale system

they were adjusted to a level appropriate for a farm-scale system.

The power consumption of the stirrer for a farm scale reactor

with 450 m3 sludge volume has found earlier to be between

1700 W and 4500 W resulting in a maximum energy

consumption of 10 W per m3 of reactor volume [21]. For the

pilot scale reactors in the present study, the corresponding

energy consumption was 60 W m3. The high energy input for

stirring in this pilot-scale reactor can be explained by the

difficulties in obtaining equipment with a reasonable level of

energy consumption for a pilot-scale reactor of the sizeinvolved. The energy required for pumping was 1 MJ m3

substrate, which is similar to what has been observed for a farm-

scale system[21].

The monitored values for the energy consumption for

heating was compared with the calculated energy needs. The

energy needed to heat the substrate was found to be 1.44 times

as high as the theoretical value for it. This can be thought to be

due to heat being lost by the hot water pipes leading to the

reactor. This appears likely since the boiler supplying the hot

water was located at a considerable distance from the reactor,

and also since pipes were poorly insulated. If this is assumed to

be the case, part of the high loss of heat from the reactordetermined in monitoring must have been caused by a loss of

heat from the pipes. This would make the loss of heat from the

reactor about 3.8 times as high as that which was calculated.

The considerable loss of heat from the reactor tank can be

explained in terms of thermal bridges, such as at the points

where the stirrer and the level indicator were located and the

loss of heat through the reactor pipes, which were less well

insulated than the rest of the reactor was. Also, the heat transfer

from the heating coil to the slurry has shown to decrease

severely due to sedimentation when the heating coils are placed

in the bottom of the reactor[31]. A similar problem may be

adding to the excess loss of heat in the pilot scale reactors.

The energy needed by different mesophilic farm-scalesystems for heating, expressed as a percentage of the total

energy production, has been determined by estimation or

calculation to be 1112% for an HRT of 20 days[32], 15% for

an HRT of 47 days and an ORL 2.6 kg VS/m3 day[21]and 30%

for an OLR of 3 kg VS/m3 day[33]. These values are similar to

those obtained by calculation in the present study, but lower

than the values monitored for the pilot scale reactors. In an

earlier example of this sort, however, the amount of the

methane produced that was used internally for heating the

process was shown to be as high as 50%[12]. Both this and the

results of the present study suggest that reactors may not always

be as well insulated as they are often assumed to be.

Since the methane yield was lower at 15 8C than at 30 8C,

the former temperature provided a poorer energy balance at

each of the OLRs than the latter temperature did, even though

less energy was needed for heating. The difference in energy

balance increased as the loading rate increased. The amount of

energy needed in the form of electricity (preparing and feeding)

should be borne in mind here particularly, since 2.2 J of primary

energy in the form of natural gas have been found to be needed

to produce 1 J of electricity[9]. In contrast, energy for heating

can be obtained directly from the biogas produced. Applying

these considerations to the energy balances makes the balance

for the 30 8C even more advantageous to the 15 8C than shown

inFig. 4.

At an operating temperature of 30 8C the energy balance was

found to be nearly constant at OLRs of 13.1 kg VS m3 day1.

However, since the methane production rate increased linearly

when OLR was within this interval, running the reactor of an

OLR of 3.1 kg VS m3 day1 would be utilizingthe reactor at its

optimal capacity.

Because of the poorer energy balance obtained at 15 8C, abiogas plant with a well-insulated digester tank should always

be run at 30 8C when used for the degradation of crop residues.

The low level of subsidization provided in Sweden for the

production and use of biogas and the small profit margins

attainable in agriculture generally, may keep farmers from

investing in the building of such tanks. However, digestion of

crop residues at ambient temperature could be carried out in a

rebuilt manure storage tank without heating. Such a tank could

provide a net energy yield 60% of that attainable with use of a

well isolated reactor run at 30 8C and 0.5 kg VS m3 day1.

Rebuilding manure tanks to serve as digesters has been done at

low cost and with good results[12,14,34]. Such a system wouldhave to be run at a lower OLR however, and gas production

would only be possible during the warm months of the year. For

digesting the 85.7 tVS assumed to be available on the farm, a

reactor volume of about 1170 m3 would be needed. The system

would only be useful in producing energy for the farm if there

were some way of using the gas produced during the summer

months. One possible application would be to use the gas as

vehicle fuel, although this would require refining it, which is

expensive if carried out on a farm-scale [7]. The biogas could

also be used for heating stables or (as is done in Italy [14]) in

connection with on-farm diary-production.

5. Conclusions

The methane yield of sugar beet tops and straw was found to

decrease with decreasing temperature, but remained stable down

to 11 8C at a HRT of 90 days. The highest yield was obtained at

30 8C. The optimal loading rates were found to be 1.6 and

3.3 kg VS m3 day1 at operational temperatures of 15 8C and

30 8C, respectively. Energy balances calculated theoretically for

a farm-scale systemshowed that a process runat 30 8C resulted in

a better energy balance than one run at 15 8C, due to of the lower

yield obtained at 15 8C. The best net energy output was found to

be obtained at an operational temperature of 30 8C and a loading

rate of 3.1 kg VS m3

day1

. An alternative low-cost solution,

I. Bohn et al. / Process Biochemistry 42 (2007) 5764 63


8/8

the energy balance in farm scale anaerobic digestion

Documents