The intermediate thermal hydrolysis process: results from pilot testing and techno-economic assessment

Anaerobic digestion (AD) has been around for over 100 years and is currently the most widely used method of sludge treatment (Appels et al. 2008). It achieves the required pathogen kill to allow for sludge to be recycled to land (Mills 2015). Additionally, it has the added benefit of reducing the dry mass of sludge for disposal and producing a methane rich biogas that can be used as fuel in a combined heat power (CHP) plant (Gunaseelan 1997).

Although AD is widespread and effective sludge treatment technique for the water industry, it has limitations (Mills 2015). For this reason, several pre-treatment technologies have been explored to enhance AD performance. The thermal hydrolysis process (THP) is the most widespread of these processes. It is a 20 year old technology with CAMBI^® being the first company to successfully market it to the water companies. This technology uses a combination of high pressure, heat and sudden de-pressurisation prior to AD or flash that hydrolyses organic matter, bursting cell walls and solubilising extra polymeric substances. This makes the organic substrate more readily available for bacteria to digest during AD (Shana 2012), enhancing the efficiency of gas generation, increasing the quantity per unit mass of sludge and the speed at which it is produced.

In addition to the enhanced biogas yields, the main benefits of this pre-treatment can be summarised as follows (Pickworth et al. 2006; McNamara et al. 2012):

• Increased volatile solids destruction (VSD);

• Process allows increased throughput in existing assets reducing capital costs;

• Reduction in mass and enhanced dewatering characteristics reducing transport costs and increasing the quality of product for farmers

This technology has been adopted worldwide, with more than 40 full scale THP sites either in operation or construction that will process 800,000 tonnes of dry solids (TDS) per year (Cambi 2014).

However, there are some limitations to the current THP setting. The process itself has a high energy demand in the form of steam being injected in the reactor to reach 6.5 barg for 30 minutes, resulting in additional support fuel required (Mills 2011). Some options have been investigated to optimise the process both in terms of further increasing the gas yield and cutting down in energy demand. This paper looks into the techno-economic analysis of two optimised THP options: (1) surplus activated sludge (SAS) only THP and (2) Intermediate thermal hydrolysis process (ITHP), and compares them with the conventional THP + AD and conventional AD.

Furthermore, this paper explores in depth the ITHP configuration, which consists of a first stage of digestion, followed by THP and a second stage of digestion. At lab scale, this process showed an increase of 18% in biogas yield compared to that of THP-AD, with a VSD of 65–67% (Shana 2012). The steam consumption is also reduced due to a smaller throughput after the first stage digestion and additional heat is also available for steam production due to higher biogas yields. These results led to the construction of an ITHP pilot plant at Basingstoke sewage treatment works (STW). This paper shows the performance of the pilot plant during its optimisation. The objective of this project is to evaluate whether the process is operable at realistic scale as well as replicating the gas yield and improved dewaterability observed by Shana (2012). The results obtained from the implementation of ITHP to pilot scale are discussed here.

The plant is not connected to Basingstoke STW. The two 20 m³ imports tanks (T1 & T2) at the front end can receive sludge from different sites by truck delivery, which allows for more flexibility in terms of sludge origin, type, dry solids (DS) content, age, etc. Typically, T1 is used for primary sludge (PS) and T2 for SAS. In order to keep the sludge well mixed, each tank is fitted with recirculation (40 min/h) and air mixing (5 min/h). Each tank would receive a weekly delivery of PS and SAS respectively.

The two types of sludge can be mixed at a desired ratio into the blending tank (T3) on a daily basis based on their respective DS. As with T1 & T2, T3 is fitted with recirculation line and air mixing in order to keep a well-mixed sludge and avoid stratification. A macerator was been added to the design to break up potential rags or other screenings.

On an hourly basis, a controlled volume of sludge was added to the first digester (T4), based on the agreed organic loading rate (OLR) of 3 kgVS/m³/d and measured % DS and volatile solids (VS) of the feed sludge. Both digesters were kept at under mesophilic conditions between 38 and 39 °C, by the use of automatic valves connected to two separate heat exchangers, heated by a water boiler. The gas produced was measured with flowmeters and continuously transferred into a gas holder.

Twenty minutes after each feed, the same volume of sludge was pumped out of digester 1 and transferred into a holding or buffer tank (T6). The volume of each digester was kept constant by checking the flows of each volume in and out. They were mixed by continuous recirculation of sludge.

Three times a week, conventionally digested sludge was dewatered using a belt-press. The liquid polymer is first diluted with high pressure final effluent and mixed with the sludge from T6 into the pipe delivering the flocculated sludge to the belt press. The cake obtained in the hopper after the dewatering stage averaged 23% DS. The cake was then transferred into a hopper (T8) to be diluted with hot water to about 14.5% DS. The dilution in T8 was controlled by a pressure gage located in the recirculation loop. Controlling the Klampress to produce a cake at 14.5% was found to be complicated and this pre-dilution system allowed for a consistent DS prior to the THP. The diluted cake was then recirculated for about 30 min and the pressure during recirculation of sludge was used as a reference for DS.

The pre-heated cake was transferred into the reactor tank (T9), where it was thermally hydrolysed. High pressure steam from the steam boiler was injected in the reactor bringing the mix to about 165 °C and 6.5 barg. The sludge was kept under those conditions for 30 mins (set point can be changed if required) before being ‘flashed’ in a matter of seconds into the flash tank (T10). Once hydrolysed, the sludge was transferred into another holding tank (T11). The tank was insulated in order to keep as much heat as possible and also fitted with a dilution to bring the %DS down to around 7%.

The operating regime of the second digester was similar to the first digester, where a set volume of hydrolysed sludge was transferred to the second digester (T5) on an hourly basis. Once again, this volume of sludge going in the second stage digester was based on an OLR of 5 kgVS/m³/d and measured %DS and %VS of the feed sludge from T11. The same volume was also pumped out of the digester (20 min after the feed) and directed to the drain. The biogas produced is measured and transferred into the same gas holder.

The gas holder allowed a constant pressure in the digesters by the use of a floating roof. When the roof is at a specified high level, the biogas is diverted to a flare stack where it is burned. An additional gas flowmeter was installed between the gas holder and the flare stack to measure total biogas generation. This point in the gas line proved to give the best most stable gas readings.

The second stage dewatering was done with two different technologies: Klampress existing onsite used for the first stage dewatering and a pilot scale Bucher unit with a capacity of 200 L per batch to replicate what a full scale results. In both cases, an SNF high energy mixer was used to allow for dual polymer injection.

All the tanks listed in Table 1 were sampled three times a week on site for DS, VS, volatile fatty acids (VFA), alkalinity, pH, ammonia. If any of the parameters looked into showed a problem in the process, sampling would be done on a daily basis until the digesters showed signs of stability. This was also done during ramp up periods. Table 1 shows the basic sampling schedule in more detail:

Additionally, the biogas composition from each digester was measured with a gas analyser also at least three times a week.

Once a week, samples of each type of sludge are sent to Thames Water (TW) laboratories for comparison and ensuring data reliability.

Two different methods were used to record VFA are:

• (1) Titration: carried out three times a week, analysis of the four sludges previously specified is done onsite and on the same day of collecting samples. The methodology involves 2 hr pre-centrifugation at 3,000 rpm. The analysis is done on 5 ml of supernatant diluted in 50 ml of DI water. Duplicates were done for the feed sludge and triplicates for digested sludge samples.

• (2) Gas chromatography with Flame Ionization Detector: Done twice a week at Thames Water laboratory. This method allows for speciation of VFA and involves a pre-centrifugation of the samples.

The relative performance of the ITHP process was also compared against existing processes, three THP variants described below: conventional THP, SAS-only THP and ITHP along with conventional AD were modelled.

In a typical process, both sludge streams are thickened and combined before being heated to approximately 37 °C inside a mixed digester tank with hydraulic retention times (HRT) of 20 to 30 days. The VSD is approximately 40% which yields 350 m³/TDS of biogas and translates to a 30% dry mass reduction (Appels et al. 2008). The final digestate is then dewatered to a cake of around 20% DS and transported off site for agricultural land use (Suh & Rousseaux 2002).

The THP process requires steam at approximately 12 barg and unfortunately there is insufficient high grade heat from the CHP to meet all of the steam requirements (Kepp et al. 2000). However, the operational performance is closer 0.51–0.53 MWh/TDS (Merry & Oliver 2014).

This process configuration was trailed at length and it is the focus of the pilot work discuss in this paper. Effectively locates the THP in the middle of two digestion stages. The first stage of digestion is a medium rate conventional digester which will obtain biogas from the readily available organic matter, the digested sludge now with a reduced mass is dewatered before thermal hydrolysis which can now be two thirds the size of a conventional plant.

Following the process model a number of key parameters are carried forward into the OpEx module and combining these with unit cost assumptions the following costs/revenues are calculated:

Cost bases:

• Electricity use (MWh/d)

• Labour (FTE's)

• Polymer (kg/d)

• Digestate volume (m³/d)

• Maintenance (% of CapEx – explained later)

• Electricity generated (MWh/d)

• Electricity eligible for ROCs (MWh/d)

The output is a net OpEx position which can be used to compare processes and in combination with the CapEx, explained next, used in full financial analysis of each process.

Using cost data at various scales (S) and an exponent value of 0.6 (average value for similar installations) a series of k-values were calculated (Table 2).

Using and adapting the data in Table 3, the total CapEx for each scenario was obtained; with this, the economic feasibility of each process scenario was calculated. Table 3 summarises the financial situation and the resultant NPV with and without government incentives for a 100 TDS/day plant for each scenario. A discount factor of 8% was used and the life of the plant was assumed to be 20 years. All other assumptions can be found in Appendix A. The financial benefit that UK water companies exploit from the increasing the regulated capital value of the company asset base has not been factored into the analysis.

VFA remained fairly stable for both digesters at an average of 370 ± 70 mg/L in digester 1 and 1,448 ± 478 mg/L in digester 2 with slight rises in both digesters during ramp up periods. Alkalinity also remained very stable for both digesters with average values of 3,461 ± 257 mg/L and 4,878 ± 620 mg/L in digesters 1 and 2 respectively. VFA/Alk ratios were within acceptable levels for digester 1 at 0.11 ± 0.02. Digester 2 showed a VFA/Alk ratio of 0.32 ± 0.13. Ammonium stayed below 3,000 mg/L in both digesters, with an average of 854 ± 96 mg/L in digester 1 and 1,678 ± 246 mg/L in digester 2. The pH remained at 7.4 ± 0.2 for digester 1 and slightly higher, 7.8 ± 0.3, in digester 2. Therefore, all stability parameters indicate good digestion taking place.

Rag blockages have been a big issue in the receiving tanks (T1, T2 and T3), blocking the feed pumps with the consequent stops in feed. The PS:SAS ratio in the feed was also difficult to control. Heat exchanger leaks also had an impact on the digesters, with a considerable drop in temperature. During this period, the average total gas yield was 330 ± 23 m³/tDS and a VSD of 45 ± 3%.

Since October 2014, low temperatures (70 °C) at 6.5 barg in the THP due to poor steam-sludge mixing in the reactor during batches was observed, which resulted in incomplete hydrolysis. Additionally, poor mixing in digester 2 was identified as a possible cause of low VSD in the second phase. A lithium test in this digester indicated a 20% reduction in the active working volume (WV).

After November 2014, DS in the feed sludge to the THP was dropped a few % points which led to an increase in temperature of the hydrolysis reactor to 158 °C. A venting system was added to bleed off the inert gases from the reactor which rose the mean temperature to 161 °C during batches. Figure 4 depicts the temperature and pressure in the THP reactor during a given batch. It shows the improvement in temperature of the reactor with the changes made:

From January 2015, the site saw a clear improvement in gas production, averaging 505 ± 81 m³/TDS from March to October 2015. The lower points registered around day 160 coincide with a low VS% feed that the site received during those days. The efficiency of gas production dropped and quickly picked up after day 179 when a higher VS% feed was received. Some higher values of SGP around day 150 coincide with a higher PS % fed into digester 1 due to issues with WAS feeding.

VSD was calculated with a Mass Balance method and clearly correlates with the SGP. The period of March–October 2015 saw an average of 64 ± 9% VSD.

The dewatering trials were carried out on two different dewatering units:

• (1) Klampress belt press – The unit available onsite, used to dewater the 1st stage digested sludge, was used. The results showed DS of 34%.

• (2) A pilot Bucher Press was trialled for 3 weeks, achieving 42%DS with different SNF's polymers and a maximum of 44% DS with Floplam HIB.

The sludge from digester 2 was held in a buffer tank prior to dewatering and hence, the temperature of the sludge at the time of dewatering had drop to 12 °C. Lab scale work has shown that the same sludge dewatered at lower temperatures give lower DS in the cake. Therefore, the results obtained in both presses could see an improvement at temperatures closer to 35 °C.

It can be seen that the ITHP sludge reached 40%DS after 65 min whereas the THP sludge needed almost double that time.

Table 3 displays the technical performance of the 4 processes modelled, ITHP performance derived from the full scale pilot results displayed in this paper.

It can be seen in Table 4 that THP, SAS only-THP and ITHP all have advantages over the conventional AD, also shown in previous studies (Mills et al. 2014). THP and SAS only THP are similar in performance although SAS-only THP doesn't require support fuel, and reduction in performance is small 70 kWh_e/TDS compared with a 280 kWh_t/TDS in support fuel. The ITHP process is very impressive showing a clear step change over conventional THP using no support fuel and producing 10% more biogas than THP and achieving a net efficiency of 17%. However, it does require greater digestion capacity, which in many cases may exist already at sites with historically conservative digestion sizing.

The results for a typical plant at 100 TDS/day has been summarised in Table 4.

This analysis is for a ‘green field’ site has shown a clear progression that follows the technical performance from conventional AD, to THP and then on to the second generation THP configurations like ITHP. The best performer is the ITHP configuration which has a <6 year payback and £ > 6 M NPV improvement on the nearest rival. THP and SAS-only THP have an identical IRR but SAS-only THP has a better NPV and payback, due to its reduced CapEx. If existing digestion capacity is sufficient, SAS-only THP or the ITHP payback will be further improved.

The current results from the pilot plant show that the second stage digester is very stable, being resilient to shock loads and sudden changes in temperature. The average process performance is similar to that seen in the lab work, with an average of 505 ± 81 m³/TDS of biogas yield and 64 ± 9% VSD over a stable period, making ITHP more efficient in terms of gas generation than conventional THP. The final cake product achieved 44%DS with a Bucher press. The high rate of water release would also allow for higher throughputs on the dewatering units compared to conventional THP. ITHP is therefore viable as a full scale option for Thames Water which currently holds a total of 9 sludge centres with THP.

The analysis shows that ITHP this is financially superior to the other processes on the market assuming a green field site. However, ‘greenfield’ sites are quite rare so to conclude conventional THP still remains an excellent solution when existing AD volume or land area is extremely constrained. SAS only THP offers an improved alternative where digestion capacity is marginal and capital is limiting.