Tank reactions to mixing systems differ in operation based on the application. To assess this, a study of the Landia GasMix was conducted at a wastewater treatment plant that uses anaerobic digestion (AD) to stabilize the sludge following aerobic treatment. The mixing system was designed with all the mechanical components externally mounted for anaerobic digesters in municipal sewage treatment plants and in agricultural or industrial biogas facilities.

In addition, two smaller studies examined AD plants that primarily used agricultural wastes. One had a system operating on one of three AD reactors before the system was later fitted to all the reactors. The other one used the system in paired heat exchanger tanks. One of those tanks was missing part of the system, providing evidence of the relative effect of the missing component.

Mixing methods

The tested system uses a unique combination of gas and hydraulic mixing principles to mix with all the moving parts positioned outside the tank to provide easy maintenance as well as health and safety. During its more conventional hydraulic mixing, material is drawn from the bottom of the tank by a centrifugal pump. The pump impeller’s knife blades reduce the particle size with a chopping action. Then, the material is pumped back into the tank base through the jet nozzle (see Image 1). The nozzle tube is narrower than the tubing on the upstream side of the pump, so the material enters the tank at high velocity and causes vigorous mixing of the tank contents. In particular this affects the heavy materials, such as stones, that tend to accumulate at the bottom of the tank center by this design.

In the second mode of mixing, the centrifugal pump’s output is diverted upward to an injection nozzle located in the wall of the tank. Before the slurry from the centrifugal pump enters the nozzle, a pipe from the gas headspace in the top of the tank is connected at a right angle to the slurry pipe. This allows headspace gas, particularly biogas when the tank is used as a digestion tank, to be drawn into the slurry stream by the Venturi principle.

The gas-liquid stream passes through a plastic ring of precise shape and dimensions, causing a sudden drop in pressure of up to 3 bar before being injected at high velocity into the tank. The high velocity gas-liquid jet causes mixing of the tank at a higher level than the hydraulic mixing, and the gas bubbles reduce the digestate’s density. Because of this, low-density materials, such as plant fibers that often float to the surface, will be forced to sink into the tank.

The second mode is normally operated intermittently, and operation times can easily be adjusted or the system switched to only jet nozzle operation to suit substrate changes. No other mixing system is required.

Low dry matter substrate

Image 2. The mixing system at Viborg

Image 2. The mixing system at Viborg

The Energi Viborg Spildevand A/S plant in Viborg, Denmark, was the primary source of data in the study because of a long-term opportunity to collect data from two AD reactors with similar feeding and operational characteristics. The plant’s comprehensive supervisory control and data acquisition (SCADA) system controlled and monitored the processes during the analysis. The treatment plant sent primary and secondary sludge from the clarifiers to two parallel AD reactors. The sludge was then combined with fatty wastes collected from local restaurants and other waste streams.

Two continuously stirred tank reactors (CSTR) of 1,600 cubic meters (m3) each with a gas headspace of approximately 5 percent total volume operated at the mesophilic digestion temperature of 37°C was stored in a tank of variable volume up to 300 m3.

Both AD reactors were fitted with the mixing systems being studied, but only one of them had the nozzle and ring assembly. The reactors were also hydraulically mixed by passing the contents through the heat exchangers at a rate of 60 m3 per hour. The gas flow and gas composition from each reactor was measured and recorded.

However, the gas composition measured online was not accurate compared to a portable gas analyzer that was known to be accurate and was calibrated regularly. The portable analyzer was used to take measurements on four occasions, and measurements were taken every 15 minutes for both reactors for a period of two hours, which is the cycle time of feeding for both reactors.

A considerable amount of data was collected from the plant. The online measurements were:

  • Methane percentage in the biogas from each reactor, online and offline
  • Gas flow rate for each reactor
  • Methane flow for each reactor (calculated by the SCADA system using the above two values)
  • Carbon dioxide and other gases for each reactor (calculated by the SCADA system using the whole biogas minus the methane percentage)
  • Temperature at the top and bottom of each reactor
  • Digestate recirculation times, pause times and flow rates
  • Mixing operation and pause times
  • Mass of the different substrate inputs to each reactor

The offline laboratory measurements were:

  • pH
  • Suspended solids (SS)
  • Iron
  • Ammonia
  • Total phosphorous
  • Total nitrogen
  • Chemical oxygen demand (COD)
  • Soluble chemical oxygen demand (sCOD)
  • Total solids (TS), also known as dry matter
  • Organic matter in the TS, also known as
    volatile solids (VS)
  • Volatile fatty acids (VFA)
  • Alkalinity, which is the buffering capacity of the system, such as the ability to resist changes in pH

Of these measurements, only pH, COD, sCOD, TS, VFA and alkalinity were considered important and subjected to statistical analysis using a T-test, which provided a probability value that data sets from each reactor were significantly different. The data were compared as R1 versus R2, and also as each reactor compared to the same reactor but during a different experimental period.

During the entire experiment, pH, VFA, TS and alkalinity were not found to be significantly different when R1 and R2 were compared as paired data for each measurement period (p<0.01). However, during the period before the system being studied was installed, COD was found to be significantly higher in R2, although the means are quite similar, and sCOD was significantly higher in R1. Both COD and sCOD were not significantly different between R1 and R2 during the period of study.

Mixing tank with high dry matter

The second plant, Madsen Bioenergi I/S in Spøttrup, Denmark, treats approximately 285 tons of cattle and pig slurry and deep litter manure plus maize silage every day. The substrates are mixed in a 135-m3 mixing tank before being pumped through a chopper pump to reduce particle size and into two 25-m3 heat exchange tanks. The heat exchange (HE) tanks contain a spiral tube from the reactor outlets. Both HE use the Landia GasMix centrifugal pump with cutting knives. The only difference between them is that only one is fitted with the entire system as mentioned in the previous section.

From the HE units, the substrate is pumped to the main reactors — HE1  substrate to Reactor 1 (R1) and HE2 to Reactor 2 (R2) — and from the reactors to two secondary reactors, ET1 and ET2. Similarly, the digestate from R1 goes to ET1 and that from R2 to ET2. After the ET units, the digestate is combined in final storage tanks. All tanks are 4,600 m3 apart from the initial mixing tank and the HE tanks.

Samples were collected from the HE units on two separate occasions. On the second occasion, samples were taken every 15 minutes. In addition duplicate digestate samples from the different digestion tanks were taken, and the gas quality was measured directly from the tops of the tanks.

The HE samples were measured for viscosity, TS and VS, and the electrical consumption was measured during the heat exchange process. The plant did not measure biogas flow rates from the individual tanks. Instead, it measured only a combined gas output from the entire plant. This made measuring any changes in gas production that may have been caused by the nozzle impossible. However, measuring the removal of organic matter was possible, and this can be correlated to gas production as a mass balance.

Visual inspection of samples from each HE showed a noticeable difference between the two. The sample from the HE without the nozzle had a much coarser texture and, upon hand stirring, felt more viscous. Laboratory analysis of the samples showed a slightly higher dry matter in the sample without the nozzle (11.01 percent compared to 10.36 percent, possibly because of a small sampling error as both HEs had the same source), but viscosity in the HE without the nozzle was 46 percent greater than the HE with the nozzle, 9.61 centipose (cP) compared to 6.6 cP.

The decreased viscosity improves the efficiency of pumping and mixing in the downstream process tanks. The plant operator discovered that mixing in the process line following the HE needed to be reduced. The reduction in mixing time was 12.5 percent, which provided a savings in electrical consumption of 210 kilowatt-hours per day.

High dry matter substrate

Image 3. The mixing system installed at LBT Agro K/S

Image 3. The mixing system installed at LBT Agro K/S

The third plant (see Image 3), LBT Agro K/S, treats chicken manure, other liquid manures, deep litter, glycerol and agricultural wastes. The high solids input is pre-treated with a chain flail. The plant operates thermophilically with a digestion temperature of 50.5°C. Daily input is approximately 250 tons, and the dry matter in the reactors is high for the CSTR design, at 12 to 14 percent. The plant also uses a gas scrubbing system to upgrade the biogas to natural gas quality before grid injection.

The plant uses three main reactors of 1,500-m3 working volume followed by a storage tank for both biogas and digestate, with reactors R13, R14 and R15. Initially, the system under study was fitted to one reactor only, R15. However, this reactor was reduced in active volume to 1,300 m3 because of the size and dry matter content. Mean data from the period during which the system was fitted only to R15 showed a 10.8 percent increase in biogas productivity above R14 (4.64 m3 per day compared to 4.19 m3 per day), accounting for the slightly reduced volume of R15 at that time. Data for R13 during this period gave a mean of only 3.48 m3 per day because of maintenance operations, so it was not compared directly.

The plant operator reported a mean raw biogas methane percentage of 63 percent after the conversion on all three reactors compared to a mean of 61 percent before any units were fitted. However, this was not tested by Aarhus University.

Conclusion

The studied mixing system increased methane production by nearly 11 percent in a digester treating agricultural residues at Madsen Bioenergi. Reduction of volatile solids was increased by 11 percent as well in a digester treating agricultural residues. However, in a digester treating sewage sludge and waste grease, the methane production was increased by a smaller amount because it had a greater benefit when used with manures and biomass substrates.

Additionally, viscosity was decreased by more than 31 percent when the system was used in a heat exchange tank treating agricultural residues prior to anaerobic digestion.

The reduced viscosity input from the heat exchange tank allowed for a 12.5 percent reduction in downstream reactor mixing times, equivalent to a savings of 210 kWh per day in electrical consumption.

This article is an edited extract from a major study by Alastair James Ward, Department of Engineering – Manure Technology and Biogas at Aarhus University in Denmark. A copy of the full study is available from Landia Inc., at info@landiainc.com or 919-466-0603. For more information about the Landia GasMix system that was assessed in this study, visit the website.

Ward’s research area is biogas technology with particular focus on monitoring and pre-treatment technologies, multivariate analysis and process analytical technologies. His monitoring work has involved the application of Near Infrared Spectroscopy for online measurement of key process parameters in the anaerobic digestion process, macro nutrient determination in plant material and for estimation of biogas potentials of a variety of substrates. Ward’s pre-treatment research has focused mainly on increasing the biogas potential or the rate of biogas production of lignocellulosic feedstocks such as plants and animal manures. Methods used have included extreme thermophilic, thermo baric, microwave, oxidation, extrudation and enzymatic methods.