Acid whey treatment and conversion to single cell protein via aerobic yeast activated sludge

• Novel method of degrading acid whey wastewater.

• Described method will allow a significant reduction in the size of reactors needed to degrade this high strength industrial wastewater.

• Biomass produced by the method has excellent characteristics and should be suitable for use as livestock feed supplement.

• This research identified a specific species of yeast that was particularly suited to degradation of this material but has received little attention in past research.

ATC

Automatic Temperature Control

COD

Chemical Oxygen Demand

DNA

Deoxyribonucleic Aid

DO

Dissolved Oxygen

F/M

Food/Microorganism Ratio

FAO

Food and Agricultural Organization (United Nations)

GRAS

Generally Regarded As Safe

MLSS

Mixed Liquor Suspended Solids

PID

Proportional Integral Differential

RPM

Revolutions per Minute

SCP

Single Cell Protein

SMP

Soluble Microbial Products

TKN

Total Kjeldahl Nitrogen

TSS

Total Suspended Solids

USDA

United States Department of Agriculture

YNB

Yeast Nitrogen Base

The manufacture of dairy industry products such as yogurt and cheese yields liquid waste known as whey. There are two types: sweet whey and acid whey, whose typical physical and biochemical properties, as reported in the literature, are given in Table 1.

This paper focuses on the treatment of acid whey, which is problematic due to its very high COD compared to most food-processing wastewaters. Treatment of whey can be accomplished by:

• Aerobic treatment

• Anaerobic treatment

• Land application

• Microbial fermentation

• Reuse/mixing with other food products

• Single cell protein production

Aerobic wastewater treatment processes are uncommon because of their very large physical size (primarily tankage) and high energy demand relative to other processes. When used, aerobic treatment is generally added to polish effluent following other processes – for example, controlled feed to municipal waste treatment digesters, and anaerobic treatment (Chatzipaschali & Stamatis 2012; Prazeres et al. 2016).

Direct (untreated) land application is problematic due to the high organic load, which can cause changes in soil structure and degrade soil quality (Prazeres et al. 2016). Anaerobic treatment has the benefit of producing biogas, which has monetary value; the payback period after installation is typically between 5 and 16 years (Chatzipaschali & Stamatis 2012).

Single cell protein (SCP) production using whey as substrate has been accomplished using yeast. The most common genus used is Kluyveromyces (Ghaly et al. 2005; Yadav et al. 2014; Pescuma et al. 2015). Its advantage is that it already has a generally regarded as safe (GRAS) rating from the United States Department of Agriculture (USDA) and the product can be used as an animal feed supplement. A disadvantage is that it sometimes ferments lactose instead of turning it into biomass, which is undesirable in SCP production.

A variety of other microorganisms can also convert whey to useful byproducts, including: ethanol, butanol, hydrogen, lipids, polyhydroxyalkanoates, lactic acid, lactobionic acid, exopolysaccharides, galactooligosaccharides, ribonucleotides, ergosterol, propionic acid, and vitamin B₁₂ (Nemeth & Kaleta 2015; Pescuma et al. 2015).

The most common methods of whey wastewater treatment and disposal are anaerobic digestion and land application. The anaerobic digestion sludge does not degrade the soil but, as for most land application processes, disposal is linked to agricultural timing and regulatory regimes – for example, nutrient loading limits, and the timing of planting and harvesting.

SCP production is potentially interesting because it is relatively simple and the biomass yield is usable as animal feed or to produce yeast extract. Converting the whey to SCP also reduces transport, handling, and disposal costs, as well as creating potential revenue streams in the process residuals. Additional advantages are that yeasts thrive in the acidic conditions created as whey degrades and can tolerate the very high volumetric and F/M loading, as well as rapid changes in osmotic pressure. As a result, yeast treatment systems are:

• Significantly smaller than alternative processing (treatment) systems, thereby reducing capital costs, and

• Easy to operate.

In this study a mixed yeast culture was introduced and acclimated to degrade a simulated acid whey wastewater in a bench-scale reactor. The dominant species was Vanrija albida, which yielded more than 99% of the DNA in the sample analyzed. Conditions in the reactor were manipulated to suit, and thus select, species that can metabolize lactose aerobically (no fermentation) with limited, if any, micronutrient addition and without vitamin addition. Reactor operating conditions were varied to test high- and low- loading scenarios. The resultant biomass amino acid profile is compared to FAO guidelines for various livestock feed requirements, and operating parameters, oxygen uptake curves, nutrient requirements, cell yield, and effluent quality are evaluated.

The acid whey mixture was designed to mimic the profile presented in Table 1. The feedstock was composed of 40 grams of lactose, 10 of whey protein blend, and nutrient made up to 1 liter with distilled water. Two dosing schemes were used to create ‘high’ and ‘low’ loading conditions. In high load condition, reactor mixed liquor was discharged until 200 ml of mixed liquor per liter of reactor volume remained. 800 ml of acid whey per liter of reactor volume was added per batch. In the low loading condition, reactor mixed liquor was discharged until 500 ml of mixed liquor per liter of reactor volume remained. 500 ml of acid whey per liter of reactor volume was added. A diagram of the waste and fill scheme is shown in Figure 1. The whey protein blend used was Optimum Nutrition 100% Gold Standard Whey (chocolate). Reactor loading conditions are presented in Table 2.

The nutrient formula was developed over the course of the research. The initial version came from Frigon & Liu (2016), the final formula is provided in Table 3.

Initially, lactose-consuming, non-fermenting yeasts were investigated. Fellomyces penicillatus was chosen as the best potential candidate (Kurtzman et al. 2011), after evaluation, although other choices were/are possible.

A pellet of freeze-dried yeast cells (ATCC# 32128, Manassas, VA, USA) was grown in a fermenter (Major Science, FS-02 series, Saratoga, CA, USA) using the standard yeast nitrogen base (YNB) with glucose as a carbon source (Kurtzman et al. 2011). Once the culture was enriched, a mixed culture of freeze-dried yeast developed during previous high strength wastewater research (Frigon & Liu 2016) was added to enhance the ability of the biomass to consume wastewater constituents and enhance contaminant removal (Moeini et al. 2004).

The biomass was acclimated from YNB/glucose to the nutrient/acid whey formula over a week. The qualitative oxygen uptake curves revealed that the biomass was slightly inhibited and analysis showed that all phosphorus in the system had been used, the phosphorus concentration used was doubled. Two weeks after the reactor was started, yeast extract was removed from the nutrient formula to check its effect on reactor performance. Adaptation to the change took about 2 days but the biomass remained viable. After observing very high nitrogen effluent concentrations, nitrogen was removed from the nutrient formula (except the very small amount associated with the molybdenum micronutrient).

Once reactor operation was stable with the mixed culture and modified nutrient formula – after about two weeks – the system was considered mature and testing began. Stable operation was identified by consistent effluent COD levels, moderate amounts of residual phosphorus and nitrogen, no apparent inhibition in oxygen uptake, and stable biomass composition under microscopic examination.

The reactor was first operated at the high loading rate and tests were performed for that condition. After those tests were complete, the reactor was operated at the low loading rate condition for two weeks, until effluent conditions and reactor operation were stable. The low loading tests were then conducted.

The fermenter was operated in batch mode. The 3-liter reaction vessel was operated at 2 liters for all tests and reaction progress was tracked using the mixing speed. Vessel mixing speed is directly related to oxygen uptake rate because higher mixing speeds increase the oxygen transfer rate into the bulk liquid. During high-load tests, the reactor was fed every 36 hours, during low-load tests, every 20 hours.

The fermenter controlled DO, mixer RPM, pH, and temperature, and recorded them once per minute. Table 4 shows the fermenter set points. The fermenter's pH PID control kept the pH within +/ − 0.1 of the set point. Air was fed into the fermenter at a constant rate of 4 L/min, the flow being controlled by a rotameter. The oxygen transfer rate was regulated by a mechanical mixer with automated speed control and DO was near constant. Minimum mixing speed was set at 250 rpm but, when the DO concentration moved outside the set points, the mixing speed was automatically increased/decreased as needed to meet oxygen demand.

Oxygen uptake curves are generally developed using a respirometer or similar instrument to quantify the amount of oxygen consumed by the biomass. This could not be done with the fermenter used in this study. The fermenter could vary the mixing speed to meet a DO set point. Recording changes in mixing speed enables a qualitative measure of oxygen uptake rate to be obtained.

COD, phosphorus, TKN, and total nitrogen tests were all performed using a DRB200 reactor and DR3900 spectrophotometer (Hach, Loveland, CO, USA). All tests were performed in duplicate, and effluent samples were centrifuged at 4,000 rpm for 10 minutes and passed through a 0.45 micron filter prior to analysis.

TSS tests were performed using Standard Method 2540 D (APHA 2012).

Specific gravity was tested using an automatic temperature control (ATC) salinity refractometer (ADE Advanced Optics BCBI9177).

Microscopic examination was performed with a standard light microscope.

Amino acid and DNA analyses were performed on the biomass. 50 ml aliquots were taken at the end of the reaction time, washed three times with distilled water, and sent to MtoZ Biolabs (Boston, MA, USA) for analyses.

The specific gravity of the filter liquid in the bioreactor at the beginning of the high-load batch reaction was 1.037, somewhat higher than sea water (1.023); by the end, it was 1.008. The equivalent values for the low-load condition were 1.020 and, again, 1.008 – see Figure 2.

COD loading for the reactor was high both volumetrically and by mass. Table 5 presents the loading conditions during this work, as well as typical aerobic and anaerobic reactor loadings. Volumetric loading rates were calculated on the basis of 36-hour reaction time for the high-load condition and 20-hour reaction time for low-load, the times being derived from the oxygen uptake curves.

COD removal in the high- and low- load reactors was 97 and 85% respectively, with corresponding final effluent COD concentrations of 4,835 and 4,720 mg/L.

Figure 3 shows changes in COD, nutrient, and MLSS concentrations vs. time, and their high- and low- load influent and effluent concentrations are given in Table 6.

The biomass amino acid profile was evaluated under both high- and low-load conditions. Table 7 presents the amino acid profiles for both high- and low-load conditions, as well as the ideal profiles for feed for various types of livestock. The livestock-related profiles are based on the lysine requirements according to FAO guidance (lysine requirement = 100) (Miller 2003).

The system's operation enabled very selective growth of particular yeast species, which is important if the species is cultivated for resale/reuse. In this case, it was possible to maintain a consistent culture for a sustained period under varying loading and operating conditions. Table 8 shows the proportional breakdown of the nine most abundant species determined by the DNA analysis.

The reactor was operated under two loading conditions to test changes in reaction time, SCP yield, and final effluent quality.

High specific gravity results were returned during the early part of batch operation. High specific gravity correlates to high osmotic pressure in cells in the water. High osmotic pressure is important for two reasons:

• 1.

  The osmotic pressure in the reactor changes significantly over the reaction period. This is due to dissolved solids removal as the yeast consumes whey, lactose and other solutes. Even moderate salt concentration/osmotic pressure changes over short periods (rapid) have been shown to inhibit bacterial systems (Kargi & Dincer 2000).

• 2.

  High osmotic pressure favors yeast cell growth kinetics. Dan et al. (2003) showed that when salt concentrations exceeded 25 g/l (specific gravity of ∼1.020), yeasts grew faster than bacteria.

Reactor loading conditions are important because they affect the reactor size. High-load reactors are smaller than low-load reactors, which reduces their capital and operating costs, and space requirements. As the yeast can withstand high loading, full-scale reactors can be significantly smaller than other biological treatment systems.

Run times exceeding the time taken to consume the primary substrate, as determined by mixing speed (aeration rate), did not lead to significant additional COD removal. The lower removal efficiency of the low-load reactor may be deceptive as the residual COD in both reactors probably comprised soluble microbial products (SMP) produced by the biomass (enzymes, etc) while consuming the acid whey. As such, the COD residual reflects biomass activity rather than remaining un-consumed substrate. After yeast separation from the mixed liquor for reuse, the reactor effluent is likely to require polishing, perhaps in a bacterial system, to remove further COD prior to discharge.

Nitrogen in the system came entirely from the acid whey. The high-load condition removed nitrogen somewhat better than the low-load condition, which reflects the greater biomass production. The high nitrogen content means that no additional nitrogen needs to be added for effective treatment, although some secondary treatment may be needed to remove nitrogen remaining in the effluent.

In early tests, observation of the mixing speed curves indicated inhibition due to inadequate phosphorus content and its concentration was doubled later in the testing. This indicates a need for phosphorus addition to achieve optimal treatment of acid whey when using yeast.

SCP yields for the high- and low- load conditions were 0.51 and 0.39 g-biomass/g-COD-consumed, respectively. The high-load condition is thus superior for SCP production. Higher yield under high-load conditions was also observed by Ghaly et al. (2005). The MLSS concentration fell slightly in both reactors over time, when the consumable substrate was used. This type of mass loss has been observed previously in yeast-related studies (Frigon & Liu 2016) and might have been caused by consumption of stored reserves and/or some degree of apoptosis.

The reactor was operated at low pH to minimize bacterial growth. Initially, reactor pH was 4.5, under which condition the yeast grew well but bacterial growth was also significant. This was apparent from both the activity in the reactor, as observed by mixing speed, and from microscopic inspection of the biomass. When reactor pH was maintained at about 3.5, there was no noticeable bacterial activity. No attempt was made to produce a sterile culture other than operational changes that favor yeast growth:

• low pH

• high initial volumetric and mass loading

• batch treatment

Yeast cells are large, so the health of the biomass and bacterial infection can be determined readily by microscopic inspection. Figure 4(a) shows yeast biomass with considerable bacterial contamination (during early reactor operation). Figure 4(b) shows healthy yeast biomass with little bacterial contamination.

Reactor mixing speed (rpm) was an effective measure of biomass activity. Reactor mixing speed dropped as oxygen use declined. Once mixing speed declined to the lowest set point, the reaction was essentially finished. In these tests, mixing speed was used but at full-scale the blower capacity needed to maintain some DO could be used in a similar manner to determine when the batch reaction was finished. Yeast cells prefer high loading and die/lyse when starved, so it is important to monitor the reaction as part of the operation and stop the batch once there is no more reaction. Cell biomass lysis results in loss of valuable biomass and increases the effluent's COD.

Figure 5 shows the mixing speed plot from a ‘high-load’ batch. It is similar to a respirometry curve. Although it is not as precise, it provided qualitative data about how the reactor was operating. It was easy to tell, for instance, when bacterial contamination was present, because bacterial oxygen uptake is much higher than for the yeast biomass. The maximum mixing speed for the yeast biomass was approximately 700 rpm but, when there was bacterial contamination, the maximum rose to approximately 1,100. Dan (2001) noted that air diffuses more rapidly into yeast biomass and suggested that that this might be due to yeast floc properties, although he did not identify any specific mechanism or reason for this. The reduction in energy used for yeast system aeration is likely to reduce operating costs compared to aerobic bacterial systems.

High loadings, both volumetric- and mass- based, are needed for yeast to remain dominant in biological reactors. This type of loading scheme generally can only be accomplished by batch reactor. Some researchers have tried to operate continuously fed reactors, but COD removal efficiency and yield were poor (Yadav et al. 2014). In this study, high- and low- load batch schemes offered high COD removal and similar effluent quality. The advantage of the low-load scheme was a reaction time of 20 hours compared to 36, but the biomass yield and amino acid profile were better under the high-load scheme.

The biomass produced during whey degradation is potentially valuable as an animal fodder supplement. The high- and low- load condition amino acid profiles are close to ideal for both pigs and chickens, and look promising for various fish species (only salmonid-related data are presented here). The profiles indicate that the yeast cell proteins are well suited to meeting the nutritional needs of many animals. For the yeast used in this research to be used as fodder, it would need certification as GRAS. A number of yeast species can degrade lactose, however, and might be viable alternative candidates for acid whey degradation. Another potential market for the biomass is yeast extract production. Yeast extract has high value at either lab or industrial quality and the volume that could be produced is quite high due to the strength of the wastewater.

The DNA results show that Vanrija albida was the dominant yeast in the culture, comprising over 99%. The species has not been investigated in the literature except as a passing note as being found in soil (Buzzini et al. 2017). The Mycobank database identifies Sporobolomyces albidus as a synonym, further tracking of which in Kurtzman et al. (2011) leads to Cryptococcus ramirezgomezianus as another synonym.

• Using yeast to degrade acid whey from dairy manufacturing is viable.

• Yeast reactor loading can be much higher than that of bacterial reactors on both volumetric and F/M basis, reducing reactor size significantly.

• Biomass produced by yeast has a good amino acid profile and may be suitable in livestock feed.

• Vanrija albida was the dominant species under the reactor conditions in this study and may deserve attention in relation to wastewater treatment.