Optimising Starch Transformation in Aquafeed Extrusion for Enhanced Water Stability and Performance (Oreochromis mossambicus)

Robert T. Strathman^*1 and Garrick Yan²

¹ President of Famsun USA Design & Engineering, Inc, USA

² Manager of Famsun Group’s Global Applications Center, USA

*corresponding author : 9225 Indian Creek PKWY, 32 Corporate Woods, Ste 820, Overland Park, KS, USA, 66210 Email: Rob.Strathman@Famsun-USA.com

ORIGINAL RESEARCH article
Volume 2 – 2024 |

Aquaf. J. 2024

Received: 2 November 2023 /  Accepted: 4 December 2023 / Published: 20 January 2023| doi.org/10.61985/AJ2401

DOWNLOAD ARTICLEAbstract

The physical quality of aquafeed pellets is paramount for the health and performance of aquatic species in aquaculture systems. Starch, a small, but necessary component of aquafeed, plays a crucial role in creating durable and water-stable pellets. This paper investigates the complex relationship between starch transformation during extrusion cooking and its influence on water stability.

Through a series of experiments involving various extrusion parameters, including three melt moisture levels and three screw speeds, the study reveals that the key to producing water-stable aquafeed lies in optimizing starch gelatinization and minimizing dextrinization. High screw speeds, in conjunction with moderate to high melt moisture levels, facilitate starch gelatinization while avoiding dextrinization, resulting in pellets with superior water stability. Conversely, low melt moisture levels at any screw speed or any melt moisture at low screw speeds lead to poor water stability due to increased dextrinization or suboptimal gelatinization levels.

The research also underscores the critical role of the glass transition of starch. Starch in its semi-crystalline state is found to be more susceptible to shear degradation during extrusion. Therefore, rapidly elevating the raw material’s temperature above its glass transition temperature is crucial for achieving water-stable aquafeed.

1. Introduction

The well-being and performance of fish and shrimp within confined aquaculture systems, including flow-through, pond, and recirculated systems, are intricately linked to the physical quality of the feed they consume (Nardi et al., 2021). Ideally, aquafeed pellets should exhibit resilience against impacts and attrition during handling and maintain structural integrity even when submerged in water for durations exceeding the feeding habits of the target aquatic species. Suboptimal pellets, characterized by the presence of fines, breakage, and rapid disintegration, not only compromise the nutritional value but also leave behind residual nutrients. These leftover nutrients, if allowed to accumulate, can trigger detrimental effects such as oxygen depletion, increased carbon dioxide levels, and the conversion of nutrients to Total Ammonia Nitrogen. These ecological disturbances can adversely affect the aquatic ecosystem, leading to reduced health, diminished feed conversion efficiency, and stunted growth performance.

While various ingredients contribute to aquafeed quality, starch and functional proteins stand out as the two key components responsible for producing physically robust and water-stable pellets. The functionality of starch in relation to the production of durable pellets has been widely explored (Samuelsen & Oterhals, 2016; Samuelsen et al., 2017; Samuelsen et al., 2018). However, our comprehension of the underlying mechanisms governing water stability remains somewhat limited. This investigation is specifically focused on the treatment of starch during the extrusion process and its profound implications for the water stability of aquafeed.

1.1 Extrusion Cooking and the Significance of Starch

Extrusion cooking is a continuous, multivariant process that subjects raw ingredients to thermomechanical conditions, facilitating physical and chemical transformations. This process yields a highly viscous, plasticized melt capable of being extruded through a pressurized barrel and die plate to form pellets. Extrusion relies on a myriad of process parameters and the unique properties of starch to impart diverse physical attributes to the end-product. These characteristics include density, cell structure, hardness, pellet durability, and water stability.

In essence, starch gelatinization (SG) is a reaction observed in moisture-rich thermal operations, involving the rupture of an insoluble membrane that encapsulates thousands of tightly packed starch polymer chains. Heated water infiltrates these granules, initiating solubilization of the chains, melting of native crystalline structures, and subsequent swelling. This process generates an internal pressure sufficient to rupture the granules, releasing the starch chains and allowing a gelatinous paste to form (Lai and Kokini, 1991). However, it’s important to note that SG during extrusion often differs from conventional aqueous cooking. In extrusion, where water is typically limited, many starch granules lack adequate moisture to complete the swell-rupture mechanism. Consequently, mechanical shear becomes instrumental in disrupting and releasing the starch chains necessary for gel formation, which is critical for creating the internal pellet matrix and its binding properties.

Starch transformation during gelatinization also entails a glass transition process. This transition involves a rheological shift of biopolymers, like starch and protein, from their native crystalline state to a molten, viscoelastic state (Slade & Levine, 2001). The critical glass transition (Tg) and melt transition (Tm) temperatures mark the stages at which polymeric materials shift from a semi-crystalline, glassy structure to a rubbery state, Tg, and eventually to a molten, viscous fluid, Tm. These temperatures can be manipulated with the addition of plasticizers, with water being the plasticizer in the case of aquafeed. By introducing higher levels of plasticizer, the Tg and Tm decline. These critical temperatures also vary depending on ingredient composition.

Studies examining the effects of extrusion on pure starch have revealed that under low moisture conditions, chain scission, often referred to as dextrinization, can occur. Dextrinization entails the breaking of starch’s linear or branched chains of glucose monomers, resulting in hydrophilic and highly water-soluble fragments (Colonna and Mercier, 1983; Colonna et al., 1984; and Gomez and Aguilera, 1984). However, Barron et al. (2000, 2002) have proposed that the fragmentation of starch granules and chains during extrusion is primarily attributed to interparticle friction, especially in low-temperature extrusion scenarios.

Further investigation by Liu et al. (2010) and Li et al. (2013) studied the specific mechanisms of starch chain degradation during extrusion. Their research revealed that the relatively inflexible shorter branches of the amylopectin polymer chain were more susceptible to shear degradation compared to the longer linear amylose counterpart. Furthermore, they found that starch was most vulnerable to chain scission when in a semi-crystalline granular form (<Tg), and less susceptible when it transitioned into a molten state (>Tm). These researchers concluded that molten starch provides chain flexibility and mobility, thereby enhancing resistance to shear degradation. The insights gained from extruding pure starch may also prove relevant in deciphering the factors governing water stability in extruded aquafeed.

• 2. Materials and Methods

2.1 Test Treatments

A sinking aquafeed diet comprising soybean meal, fish meal, corn gluten meal, whole wheat, and wheat gluten was subjected to fine grinding and subsequent extrusion. The extrusion process was carried out using an H66 twin screw extruder with a Length/Diameter ratio of 20:1. The screw profile used for all treatments is shown in Figure 1.

This extruder was equipped with a two-pass preconditioner, a 75-kW main motor, a variable frequency drive (VFD), a pressure-based Density Control System (P-DCS), and a die with three 5.0 mm holes (Famsun Group, LTD. Yangzhou City, Jiangsu Province, CHN).

The aquafeed, containing 15.6% starch, underwent extrusion under varying melt moisture levels (23.0%, 27.5%, and 32.0%) and three screw speeds (250, 375, and 500 rpm). A constant dry mix feed rate of 300 kg/hr was maintained for all treatments. The average retention time in the preconditioner was measured at 3.5 minutes for treatment #1, serving as an approximation for all other treatments. Steam and water mass flow rates to the preconditioner were pre-set and adjusted to meet the desired melt moisture targets and maintain a 92°C preconditioner discharge temperature. No additional steam or water was injected into the barrel. The P-DCS was maintained at a constant pressure of 0.15 MPa for all treatments.

After achieving a stable operational state, defined as a minimum of 15 minutes of continuous operation with consistent mass flow rates (±0.5%), stable extruder motor current (±5%), and a steady density profile, a three-kilogram sample was collected at the extruder outlet for each treatment. These samples were immediately spread thinly on trays and allowed to air-cool overnight, followed by further drying to approximately 9% moisture content in an air oven.

Specific Thermal Energy (STE), Specific Mechanical Energy (SME), Total Specific Energy (TSE), and Melt Moisture (MM) were estimated using the following equations:

Equation 1:  STE:  q_te = (Q_dm + Q_ss + Q_ws) / ṁ_s

Equation 2:  SME:  q_me = P_e / ṁ_s / 3600

Equation 3:  TSE:  q_e = q_te + q_me

Equation 4: MM: X_mm = ((ṁ_ss + ṁ_ws + (ṁ_dm * X_dm))/ṁ_s

Where:

–           q_te:  STE, kJ/kg

–           q_me: SME, kJ/kg

–           q_e:   TSE, kJ/kg

–           Q_dm: Energy from the dry mix, kJ/h

–           Q_ss:   Steam energy to the system, kJ/h

–           Q_ws:  Water energy to the system, kJ/h

–           P_e:    Extruder motor power, kW

–           X_mm: Moisture content of the melt, %

–           X_dm:  Moisture content of the dry mix, %

–           ṁ_ss: Mass flow rate of steam to the system, kg/h

–           ṁ_ws: Mass flow rate of water to the system, kg/h

–           ṁ_dm: Mass flow rate of dry mix to the system, kg/h

–           ṁ_s:   Total mass flow rate of all materials through the system, kg/h

The uncoated aquafeed pellets were evaluated for starch degradation by-products (i.e., reducing sugars) using D-glucose and Dextrose Equivalent assays to quantify the extent of chain scission caused by the various treatments. Changes in water solubility, as indicated by an RVA’s Cold-Water Viscosity, were also assessed as an indirect measure of hydrophilic components. The degree of gelatinization was assessed using a modified enzymatic method and the RVA’s Hot-Water Viscosity.

2.2 Test Methods and Analysis

Luff-Schoorl Dextrose Equivalent (DE) Method: The DE Method was employed to measure the total reducing power of the water-soluble portion of each sample relative to a dextrose standard, expressed as a percentage on a dry matter basis. The DE value inversely correlates with molecular weight, reflecting the degree of hydrolysis. This test was conducted in triplicate.

D-glucose and Total Starch Content: D-glucose levels were quantified using Megazyme’s GOPOD and k-testa-100A Assay Kits, reported as a percentage of total starch content. These tests were conducted in triplicate.

Degree of Starch Gelatinization: A modified enzymatic method was utilized, measuring glucose released colorimetrically at 420 nm and expressing it as a percentage of gelatinized starch relative to the total starch content. This test was conducted in triplicate.

Rapid Visco Analyzer (RVA): The RVA, specifically the Starch Master-2 from Perten Instruments BA, Hagersten, Sweden, was employed to assess starch gelatinization and dextrinization in the post-extruded pellets. Analysis parameters included a constant impeller speed of 160 rpm, a solid-liquid ratio of 5g sample and 23 ml water, and a heating-cooling cycle of 50°C to 95°C to 50°C.

Water Stability: The Dissolution Rate Method was utilized to evaluate water stability. The method quantified the loss of pellet mass during a 30-minute immersion in static water at 30°C. Initially, 50g samples are weighed, placed in hanging wire mesh colanders, and then submerged. The samples were lifted and gently agitated for 30 seconds every 10 minutes, followed by drying and calculation of the loss of mass. The method converts the initial sample weight to a dry matter basis and then fully dries the soaked samples in an air oven to account for moisture absorption variances. The percentage of pellet mass lost during the bath is then calculated per Equation 5:

Treatment	Melt Moisture	Screw Speed	Precond-itioner Temp	1STE	2SME	3TSE	4TMF	Dextrose Equivalent	D-Glucose	5CWV @100 Sec	6HWV @350 Sec	Gelatin-ization	Bulk Density	Dissolution Rate
#	%	rpm	oC	kJ/kg	kJ/kg	kJ/kg	kg/hr	%	%	cP	cP	%	g/L	% Loss
3	32.0	500	92.4	235	127	362	402	7.18±0.01	2.84±0.12	64	235	86.6	650	6.3
4	27.5	500	92.6	223	152	375	377	7.02±0.19	2.72±0.52	84	255	91.7	652	6.5
9	23.0	500	92.6	213	183	396	355	7.92±0.11	5.07±0.22	137	240	91.2	591	67.2
2	32.0	375	91.1	235	108	343	402	7.06±0.33	2.55±0.28	85	337	88.2	669	28.5
5	27.5	375	92.9	223	110	333	377	6.96±0.11	3.41±0.11	82	307	90.4	657	12.2
8	23.0	375	92.4	213	148	361	355	7.63±0.14	4.94±0.84	141	259	90.6	633	59.2
1	32.0	250	91.5	235	79	314	402	6.92±0.17	2.34±0.09	87	375	87.0	679	30.6
6	27.5	250	93.0	223	91	314	377	7.04±0.03	3.30±0.11	112	453	87.2	681	35.3
7	23.0	250	92.8	213	111	324	355	7.32±0.15	4.65±0.10	90	393	90.1	677	29.6
1: Specific Thermal Energy, 2: Specific Mechanical Energy, 3: Total Specific Energy, 4: Total Mass Flow Rate, 5: Cold Water Viscosity, 6: Hot Water Viscosity

Regression Analysis (SRA) was used when multiple factor correlations were not established; statistical significance set was to p<0.05. Statistical analyses and modeling were conducted using Minitab Software, version 21.3.1 (Minitab, LLC., State College, PA, USA). 

2.3 Statistical Analysis

Two extrusion variables (screw speed and melt moisture) were examined at three levels (high, medium, and low) to investigate their impact on physicochemical properties of aquafeed pellets. Stepwise Multiple Regression Analysis (SMRA) was employed to assess the primary factors influencing water stability. Statistical significance for ANOVA analysis of each overall model was set to p<0.10. Simple Regression Analysis (SRA) was used when multiple factor correlations were not established; statistical significance set was to p<0.05. Statistical analyses and modeling were conducted using Minitab Software, version 21.3.1 (Minitab, LLC., State College, PA, USA). 

• 3. Results and Discussion

3.1 Extrusion Data Overview

Table 1 presents a comprehensive overview of the extrusion parameters and product responses for the nine treatments. Notably, as melt moisture levels decreased from 32.0% to 23.0%, a slight reduction in thermal energy levels was observed. This decrease in Specific Thermal Energy (STE) was primarily due to a reduction in water flow, resulting in less steam required to maintain the target preconditioner temperature. Overall, STE levels remained within a narrow range of 213 to 235 kJ/kg.

Furthermore, a clear linear relationship emerged between screw speed and Specific Mechanical Energy (SME), with SME levels showing a steady increase as screw speed increased (correlation coefficient r=0.798, p=0.010). Given the limited range in STE levels and the linear correlation between screw speed and SME, it’s not unexpected that this relationship also extended to Total Specific Energy (TSE) (correlation coefficient r=0.915, p=0.001).

The effects of melt moisture and screw speed on SME and TSE were modeled, revealing strong correlations (SME: R2=95.47%, p<0.001, and TSE: R2=93.50%, p<0.001), as depicted in the representative model in Figure 2. Higher screw speeds led to increased SME and TSE in a positive linear fashion, while the addition of water diluted the applied energy.

The bulk density, driven by expansion of the extrudate after the rapid vaporization of water upon exiting the die (Fan et al., 1994), was investigated. Bulk density was found to be positively correlated with SME (R2=93.12, p=0.002) and TSE (R2=92.86%, p=0.003). As expected, SMRA confirmed that bulk density increased linearly as energy levels decreased. At high energy input levels, bulk density increased linearly with rising melt moisture, while at low energy levels, density decreased slightly with increasing melt moisture, as depicted in the representative model shown in Figure 3.

The incorporation of a Pressure-based Density Control System (P-DCS) equally restricted expansion across all treatments, given it operated at a constant pressure, ensuring all treatments resulted in the production of sinking pellets.

The Parallel Coordinates Plot in Figure 4 demonstrates the impact of melt moisture and screw speed on water stability and its relationship with two of extrusion’s dependent variables: SME and TSE. Treatments 3 and 4 exhibited the highest water stability, with 6.29% and 6.50% loss of mass during the water bath, respectively. These treatments were extruded at 500 rpm with melt moisture levels of 32.0% and 27.5%, respectively. Conversely, treatments 8 and 9 resulted in the least water-stable pellets, processed with 23.5% melt moisture and screw speeds of 375 and 500 rpm, respectively. In general, high moisture levels and moderate to high energy inputs positively influenced water stability, while low moisture, low SME, and low TSE levels resulted in poor water stability.

3.2 RVA Data Overview

Raw starch granules are insoluble in cold water and typically exhibit a low initial Cold-Water Viscosity (CWV), as observed in the raw material sample in Figure 5. However, extruded products generally display higher CWV’s due to increased water solubility resulting from starch gelatinization and dextrinization. Fully gelatinized starch is approximately 2% water soluble in 25°C water, whereas a dextrinized starch with a DE value of 9 is extremely water-soluble: approximately 40% soluble at 25°C, leading to elevated CWV.

The RVA data for the nine treatments exhibited a wide range of CWVs, indicating varying degrees of chain scission. Treatment 3, extruded at 500 rpm with 32% moisture, displayed the lowest soluble starch content. Conversely, treatments 8 and 9, processed at 375 and 500 rpm with 23% moisture, respectively, exhibited elevated CWVs, suggesting these conditions caused chain scission, resulting in pellets containing degraded, highly water-soluble, starch fragments.

Hot-Water Viscosity (HWV), an indicator of starch gelatinization, is known to reach an elevated peak when high levels of raw starch are present, such as the case of the raw material. Among the nine treatments, the HWV was observed to decrease with increasing TSE (melt temperature), reflecting a reduction of intact starch granules, and reduced overall swelling during the RVA procedure. Treatments 3, 4, and 9, each extruded at 500 rpm, exhibited a high degree of starch gelatinization, as evidenced by their low HWV in Figure 5. Conversely, treatments 1, 6, and 7, processed at 250 rpm, contained the least gelatinized starch, resulting in elevated HWVs. These results indicated that higher screw speeds, associated with increased SME and TSE, promoted greater gelatinization.

3.3 Dextrinization

Chain scission, or dextrinization, occurred during low-moisture extrusion, as indicated by the significant quadratic correlations (R2=94.63%, p<0.001 and R2=76.49, p=0.013) between the two water soluble fragment indicators, D-glucose and DE, and melt moisture respectively, as illustrated in the representative SRA model shown in Figure 6.

Both DE and CWV levels were influenced by melt moisture and screw speed, as confirmed by the representative SMRA model in Figure 7 (DE: R2=89.21%, p<0.007 and CWV: R2=74.74%, p=0.059). The combination of high melt moisture and high screw speed resulted in low DE and CWV, indicating minimal shear degradation under these extrusion conditions. However, chain scission accelerated linearly when operating at high melt moisture and subsequently reducing the screw speed. This was attributed to the reduction in shear rate, which lowered viscous dissipation and mechanical energy input, causing the material to drop below its glass transition temperature (Tg), where it is most susceptible to shear degradation. Conversely, chain scission decelerated when operating at the low melt moisture and reducing the screw speed, suggesting that material already below its Tg was less likely to degrade further as the shear rate declined.

3.4 Gelatinization

The modified enzymatic starch gelatinization (SG) assay yielded high SG values ranging from 86.6% to 91.7% across all treatments, exhibiting limited correlation with extrusion parameters. This finding was consistent with previous research (Whalen et al., 1997), which suggested that the modified enzymatic SG assay was less sensitive than the RVA in measuring starch status. However, HWV, the starch gelatinization indicator, was found to be highly correlated with total energy input (R2=78.12%, p=0.002) as seen in Figure 8, consistent with previous reports (Caldwell et al., 2000) associating increasing total energy input with rising starch gelatinization levels.

3.5 Water Stability

The representative regression model in Figure 8 confirmed that starch gelatinization and dextrinization were the primary factors influencing aquafeed water stability (Water Stability vs. D-Glucose and HWV: R2=97.06, p<0.003 and Water Stability vs. DE and HWV: R2=91.13, p=0.005). Water stability improved linearly with increasing SG levels, indicated by the reduction in HWV. Additionally, water stability improved quadratically as chain scission decreased, as evidenced by declining D-glucose levels.

The impact of melt moisture and extrusion parameters: Screw Speed, SME, and TSE, on water stability were modeled (SS: R2=83.39%, p=0.074, SME: R2=82.55%, p=0.024, and TSE: R2=74.89%, p= 0.058) as shown in the representative model in Figure 10. These models validated that high screw speeds and high energy levels, combined with moderate to high melt moisture, optimized gelatinization and minimized dextrinization, resulting in water-stable pellets.

• 4. Conclusions

4.1 Implications for Aquafeed Production

The results of this study offer valuable insights into optimizing aquafeed production processes. To enhance the water stability of aquafeed pellets, it is advisable to extrude at higher screw speeds along with moderate to high melt moisture levels. These conditions promote starch gelatinization while minimizing chain scission, leading to the production of pellets with superior water stability. It is worth noting that sinking feeds can benefit significantly from the use of a density control system, which effectively counterbalances the increased energy input and the likelihood for expansion resulting from these optimal conditions. Furthermore, the extrusion process can benefit from a throttling valve to allow independent management of Specific Mechanical Energy without necessitating adjustments to other parameters. For instance, to prevent the dilution of energy levels due to added moisture, the valve can be adjusted to increase total energy input resulting in higher gelatinization levels.

4.2 Implications of Starch Transformation

The glass (Tg) and melt (Tm) transition temperatures of the raw material are pivotal factors that significantly influence the behavior of starch during the extrusion process. When the material remains below its Tg, it possesses a glassy, solid-like state. Conversely, exceeding its Tg initiates a transformation into a rubbery and pliable material, eventually transitioning into a viscous and elastic melt upon surpassing its Tm. A key finding of this study lies in the impact of extrusion temperatures, facilitated either by rapid elevation or the introduction of water to lower the Tg and Tm, which provides starch chains and their branches with increased mobility, rendering them more resilient against shear degradation.

This phenomenon is of paramount importance to producing sinking aquafeed, especially regarding the practice of operating at lower screw speeds to minimize pellet expansion. This common practice results in lower extrusion temperatures, which, in turn, heightens the risk of chain scission, lowers gelatinization levels, and ultimately compromises water stability.

In summary, this study underscores the critical role of extrusion parameters and the intricate transformation of starch in shaping the physicochemical properties of aquafeed pellets. Such comprehension equips the aquafeed industry with the knowledge needed to produce high-quality, water-stable pellets, thereby making significant contributions to the sustainable growth of aquaculture.

5. References

1. Barron C., Bouchet B., Della Valle G., Gallant D.J., and Planchot V. (2000) “Microscopical Study of the Destructuring of Waxy Maize and Smooth Pea Starches by Shear and Heat at Low Hydration”. Journal of Cereal Science 33 (2001) 289–300.
2. Barron C., Della Valle G., Colonna P., and Vergnes B. (2002) “Energy Balance of Low Hydrated Starches Transition Under Shear”. Journal of Food Sci, 67(4), pp. 1426-1437.
3. Caldwell E.F., Fast R.B., Levolella J., Lauhoff C., Levine H., Miller R.C., Slade L., Strahm B.S., Whalen P.J. (2000) “Cooking of Ready-to-Eat Breakfast Cereals”. Cereal Foods World. 45(6), pp. 244-252.
4. Colonna P., and Mercier C. (1983) “Macromolecular modifications of manioc starch components by extrusion-cooking with and without lipids”. Carbohydrate Polymers. 3(2), pp.87-108.
5. Colonna P., Doublier J.L., Melcion J.P., de Monredon F., and Mercier C. (1984) Extrusion cooking and drum drying of wheat starch. I. Physical and macromolecular modifications”. Cereal Chemistry. 61, pp. 538-543.
6. Fan J., Mitchell J.R. and Blanshard JMV, (1994). “A computer simulation of the dynamics of bubble growth and shrinkage during extrudate expansion”. J. Food Engineering. 23(3), PP. 337–356.
7. Gomez M.H., and Aguilera J. M. 1984. A physicochemical model for extrusion of corn starch. J. Food Science. 49(1), pp. 40-43.
8. Lai L.S. and Kokini J.L. (1991) “Physicochemical changes and rheological properties of starch during extrusion”. Biotechnol Prog; 7(3), pp. 251–266.
9. Li M., Hasjim J., Xie F., Halley P.J., Gilbert R.G. (2013). “Shear degradation of molecular, crystalline, and granular structures of starch during extrusion”. J. of Starch-Stärke. 66(7-8), pp.595–605. Available at: https://doi.org/10.1002/star.201470071
10. Liu W.C., Halley P.J., & Gilbert R.G. (2010). “Mechanism of Degradation of Starch, a Highly Branched Polymer, during Extrusion. The University of Queensland”. Published American Chemical Society. Macromolecules. 43(6), pp.2855–2864.
11. Nardi G., Beckman G., Manley C., & Backman S. (2021). “RAS Roundtable: Feeding Strategies to Maintain Optimal Water Quality and Fish Performance”. Innovasea, Skretting NA, and RAStech Webinar. Available at: https://www.youtube.com/watch?v=ZDoRAB2dCjA
12. Samuelsen T.A., Mjos, S.A., & Oterhals A. (2014). “Influence of type of raw material on fishmeal physicochemical properties the extrusion process, starch gelatinization, and physical quality of fish feed”. Aquaculture. Nutrition, 20(4), pp. 410-420.
13. Samuelsen T.A. and Oterhals A. (2016). “Water-soluble protein level in fishmeal affects extrusion behavior, phase transitions and physical quality of feed”.  Aquaculture Nutrition, 22(1), pp. 120-133.
14. Samuelsen T.A., Oterhals A., Kousoulaki K. (2017). “High lipid microalgae (Schizochytrium sp.) inclusion as a sustainable source of n-3 long-chain PUFA in fish feed-Effects on extrusion process and physical pellet quality”. J. Animal Feed Sci and Tech, 236(1), pp. 14-28.
15. Slade L. and Levine H. (2001). “Food Polymer Science approach to rheology and state transitions in cereal polymer processing and product quality”. AACC Annual Meeting. Charlotte, NC. Available at: https://www.researchgate.net/publication/301725364_Food_polymer_science_approach_to_rheology_and_state_transitions_in_cereal_polymer_processing_and_product_quality
16. Whalen P.J., Bason M.L., Booth R.I., Walker C.E., & Williams P.J. (1997). “Measurement of Extrusion Effects by Viscosity Profile using the Rapid ViscoAnalyzer”. Cereal Foods World. 42(6), pp. 469-475.