Escolar Documentos
Profissional Documentos
Cultura Documentos
Evaluation of methods for reducing the ash content of waste frying oil
processed to biofuel oil
Maxim Bril'kov a,b, Anne B. Falck-Ytter a, Knut O. Strætkvern a,⁎
a
Hedmark University College, Department of Natural Sciences and Technology, Norway
b
FMC Biopolymer AS, 5545 Vormedal, Norway
a r t i c l e i n f o a b s t r a c t
Article history: Waste frying oil (WFO) reclaimed from food processing and stripped from various impurities is considered a
Received 20 January 2015 valuable feedstock for making biofuels. Specifically, to be used as bio-oil in industrial burners the ash content
Received in revised form 10 March 2015 of WFO should not exceed maximum limits set for dust pollution from burners. Samples obtained from a WFO
Accepted 12 March 2015
processing plant were used in a study to investigate efficient means of reducing ash to acceptable levels, i.e.
Available online 29 March 2015
b 0.015%. Since the ash mainly exists as discrete particles dispersed in the oil, physical methods of separation
Keywords:
such as filtration, extraction, settling and centrifugation were tested. Benchmark analysis of the current full
Waste oil scale process employing settling and 400, 200 μm filtration steps showed that end products after heating at
Ash 600 °C, contained 0.02–0.01% ash, regardless of high (0.5%) or low (0.02%) initial ash. Experimental filtration
Filtration with 45 μm screen demonstrated that ash could be reduced 40–50% further. In aqueous extraction experiments
Aqueous extraction ash reduction was dependent on both water:oil ratio and the temperature, in a manner showing interaction of
Centrifugation the two factors. Extraction at a water ratio of 0.2 at 40 °C provided a consistent 60–70% reduction including
Iodine number the settling effect, and obtained with test volumes up to 1 l. High-speed centrifugation offered final ash levels
as low as 0.012%. However, extensively purified WFO retained a particle fraction below 10 μm. The study points
to possible improvements of steps in the current WFO line. Interestingly, any measures to reduce end product ash
might be offset by burner temperatures above 800 °C due to the observation that approximate 50% of the ash
being carbon based material combusted above this temperature.
© 2015 Elsevier B.V. All rights reserved.
1. Introduction Deep-fat frying is one of the most common methods for food
preparation which improves the taste of food [8]. The same oil or
1.1. Waste frying oil (WFO) as by-product fat is used several times, mainly for economic reasons. However,
continuously using the same oil or fat for frying will cause various
The annual generation of vast amounts of used deep-fat frying oil physical and chemical changes to the oil. Some physical changes ob-
from fast food caterings, restaurants and food processing industry is served in vegetable oil after frying are: increase in viscosity, increase
posing a challenge for proper reuse. While some are paid for their in specific heat, change in surface tension, and change in color [9].
WFO to be disposed off, others are charged a fee for the waste handling. Oils are also subjected to three types of reactions during frying,
In certain instances the WFO is mixed into pet food, make-up, livestock mainly thermolytic, oxidative and hydrolytic [10,11]. The combined
feed, or fertilizer, but still large volumes are dumped into sewers and effect of these chemical reactions on the cooking oil is that many un-
landfills. Considering the benefit of energy conservation, companies desirable compounds are formed during repetitive heating and fry-
and individuals can easily produce cheap biofuel from WFO [1,2], espe- ing thereby increasing the polar content (FFA, ketones, aldehydes)
cially the used vegetable oils have been considered as a promising op- [10–12].
tion. Several studies show the possibility of using WFO in the biofuel/
biodiesel production [3–7]. 1.2. Purification of WFO
http://dx.doi.org/10.1016/j.fuproc.2015.03.008
0378-3820/© 2015 Elsevier B.V. All rights reserved.
488 M. Bril'kov et al. / Fuel Processing Technology 134 (2015) 487–493
Fig. 1. The outline of a WFO to biofuel-oil production. Boldface numbers indicate the sam-
2.1. Chemicals pling points analyzed in this study, either taken from the process flow or holding tanks.
2.3. Centrifugation The initial test for screening of extraction conditions was carried out
as a two-level factorial design with center points. An oil sample (70 ml)
Centrifugation was carried out at 8 500 ×g for 30 min on samples was mixed with water by shaking for 1 min at different water:oil ratios
(60 ml) of the raw material and end product, at room temperature (0.05–0.25) in a 100 ml conical separating funnel. The funnel was placed
(22 °C) in a Beckman JA-20 rotor. The resulting supernatant from each vertically on a stand in a heating cabinet and held at different tempera-
sample was taken to ash analysis. tures (30, 40 and 50 °C) for 2.5–3 h until layers appeared visually well
M. Bril'kov et al. / Fuel Processing Technology 134 (2015) 487–493 489
separated. The water layer was drained and the oil fraction was ana- 3.1. Process batch characteristics
lyzed for its residual ash content. In scale-up extractions, 250 ml and
1000 ml oil were used in proportionally larger, straight-wall separation Each critical point sample from the production line was tested on ash
funnels and held for layer separation for 6 and 9 h, respectively. content, viscosity (Table 1) and iodine value (Table 2). The ash content
is reduced for both set of samples throughout the process, with end
2.6. Ash content determination products at 0.021% for high (batch 1) and 0.010% for low initial ash
(batch 2). Viscosity of the WFO decreases throughout the process as
Replicate samples were weighted (± 0.001 g) in tarred porcelain well, but only significantly for the batch 1, with a negative correlation
crucibles (i.d. 40 mm, height 8 mm) of known mass and then pre- to the ash content. Difference in viscosity between the batches is appar-
burned inside a hood over a gas burner. The crucibles were then trans- ent, however, it is not significant (F1,8 = 3.24, p b 0.05). The iodine value
ferred to a high temperature furnace (Carbolite CWF 11/5, Sheffield, UK) as a measure of unsaturated fatty acids in samples 2–5, is as expected far
and incubated for 4 h, typically at 600 °C. Some samples were heated above the mineral oil control, but remains almost the same for both
also at 800 and 975 °C. Crucibles were cooled in an exicator and weight- batches from the dewatering step and throughout. The oil composition
ed. Ash content (%) of the sample was calculated according to the equa- of a batch is considered largely unchanged, although in both batches,
tion: Ash content = [(m2 − m1) / mo] · 100%, where m1 was the mass the iodine value of filter reject (6) was markedly lower than in the sub-
(±0.001 g) of the empty crucible, m2 the mass of the crucible and the sequent filtrates (Table 2). Thus, a fraction of insoluble saturated fats
ash, and m0 the mass of sample. were removed. Between batches 1 and 2, on the other hand, a signifi-
cant difference (F1,8 = 8.82, p b 0.05) in oil composition was observed
2.7. Degree of unsaturated oils by iodine value signifying that batch 2 has a higher content of unsaturated fatty acids.
The content of unsaturated oils in WFO was determined by absorp- 3.2. Combustion temperature
tion of halogen, expressed as the iodine value (IV). The analysis was car-
ried out according to the ISO 3961 standard [23]. Replicate samples To observe the effect of oven temperature on the ash determination,
were analyzed, and mineral oil was used as a control. raw material oil and end product sample taken after the second filtra-
tion step were analyzed at three different temperatures (Fig. 2). The an-
2.8. Viscosity alyzed ash content in the crude WFO dropped by almost 50% when
increasing the combustion temperature from 600 to 800 °C, and an ad-
For viscosity measurements a Brookfield LVDV-1 Prime digital vis- ditional 10% drop was observed when increasing to 975 °C. For the
cometer (Brookfield Engineering Laboratories, Inc., Middleboro, USA) filtrated end product oil, already quite low in ash, the remaining ana-
was used. The test probe volume (about 400 ml) was taken in a lyzed ash also dropped with increasing temperature. This indicates
600 ml glass with 8.8 cm internal diameter. Measurements were obtain- that about 38–59% of the ash in the crude material is of carbon based
ed with a S62 spindle at 30 °C. material, which is readily removed at a higher combustion temperature.
ANOVA was used for testing of differences between batches, apply- Using high-speed centrifugation, WFO raw material and the end
ing a level of significance at p b 0.05. Modeling and statistical analysis products taken from the two different product batches were separated
of data from factorial testing of aqueous extraction (Design of Experi- into a liquid top phase and bottom phase, which consisted of solid par-
ments) was carried out using MODDE 10.1 from Umetrics. ticles, impurities and saturated solid fats. The top phase was analyzed on
ash content and compared to the original samples (Fig. 3). Centrifuga-
3. Results tion reduced the ash in raw material samples by 90–98% to below
0.015%, both for high and low initial ash WFO. Only minor effect of the
In the present study two sample sets of WFO processing from a com- centrifugation was observed on the two end products, suggesting that
mercial operation (Fig. 1) were obtained and characterized to obtain there is a fraction of ash particles (0.012% average) that was too small
benchmark values for their ash content. The production batches were to settle and remained buoyant in the medium.
both of high and low initial ash content thereby describing any likely
range of raw material variation at this facility. Physical methods of ash 3.4. Filtration
reduction (centrifugation, filtration, aqueous extraction) were then
tested at lab-scale, under the assumption that these are scalable separa- In filtration experiments when oil was passed through a 45 μm filter
tion methods. by a peristaltic pump, a significant decrease in ash content was observed
Table 1
Effect of the processing steps on two production batches of bio-oil from WFO differing in initial ash content. Ash content (mean ± SD) and viscosity determined in samples obtained from
the process line according to Fig. 1.
Sampling pointsa Sample type High initial ash (batch 1) Low initial ash (batch 2)
Table 2 0.6
Iodine values (g/100 g, mean ± SD) during processing of two production batches of biofu-
el-oil from raw material with high initial (batch 1) and low initial (batch 2) ash content. 0.02
Sampling points Sample type Batch 1 Batch 2
0.5
1 WFO input 80.24 ± 2.95 85.74 ± 0.46
2 Dewatered oil 69.79 ± 8.62 79.51 ± 0.44
3 1st filtrate 72.51 ± 2.04 78.14 ± 1.55 0.01
4 2nd filtrate 71.01 ± 0.90 77.39 ± 0.78
0.4
5 End product 73.55 ± 3.14 81.21 ± 0.37
6 Reject 1st filter 63.01 ± 0.14 70.42 ± 0.14
1.0
Ash content (%)
600°C 0.06
0.791
0.8 800°C
975°C
Ash content (%)
0.6 0.04
0.413
0.4 0.328
0.02
0.2 0.133
0.1 0.0825
0.00
0.0 2nd process filtrate (4) End product holding tank (5)
WFO input 2nd filtrate
Fig. 4. Effect on the ash content in WFO process samples after filtration through a lab-scale
Fig. 2. Effect of the oven combustion temperature on the measured ash content. Bars indi- filter (50 mm diam.). Bars indicate the percent ash in two process samples (points 4 and
cate percent ash in two oil samples (points 1 and 4) taken from the WFO-process. 5) from low-ash WFO (batch 2) when filtered through a 45 μm filter.
M. Bril'kov et al. / Fuel Processing Technology 134 (2015) 487–493 491
A B
C D
Fig. 5. Effect on the ash content from extraction analyzed in a factorial design of water ratios and holding temperatures. The water:oil ratio (x-axis in all plots) was tested at a low and a high
level using 70 ml oil (2), and holding for 3 h at high and low temperatures (30, 50 °C). The W:O ratio was slightly shifted downwards between the two experiments. Contour plots on the
left show the ash reduction in (A) high initial ash WFO (W:O = 0.125–0.25), and in (C) low initial ash WFO (W:O = 0.05–0.2). Boxed numbers on isocurves are percent ash reduction from
the initial feed. Corresponding line plots to the right show the factor interaction on the ash content, for (B) high initial ash WFO, and (D) low initial ash WFO. Plots differ on the ash content
scale by a factor of 5. Error bars where visible, represent replicate ash analysis.
The analysis of the oil from critical points in the production line
confirmed the expected reduction to a low ash level, apparently re-
0.04
gardless of how much initial ash was present. This could be affirma-
tive for the overall process already being robust. However, based on
an arbitrary testing over 15-month period, the final ash content had
0.02 shown to vary in the range 0.01–0.05% (E. Lundby, personal commu-
nication). Thus, further knowledge on the nature and behavior of the
ash was mandated.
For the WFO of high initial ash (batch 1) the viscosity correlated with
0.00 the ash content, suggesting that ash affects oil viscosity and may pass to-
250 ml, 1000 ml, 45 µm 10 µm gether with removal of saturated (solid) fats by filtration and settling.
6 hrs 9 hrs filtrate filtrate This suggestion was supported by the difference in iodine value be-
Oil samples from water extraction (1:5), tween the raw material and the end product within a batch. Also be-
and after filtration of the 1000 ml raffinate. tween the batches, the difference in iodine value should reflect a
higher amount of polyunsaturated fatty acids in batch 2. This difference
Fig. 6. Scale-up of water extraction (W:O 1:5) and subsequent filtration of the raffinate.
Bars indicate percent ash in the oil phase after extraction with 50 and 200 ml of water,
in oil composition is also evident from the initial viscosities of the WFOs
and held at 40 °C, and after filtration of the 1000 ml batch oil phase through a labscale since melting points of unsaturated oils are much lower than for satu-
filter. rated oils causing the decreased viscosity [24].
492 M. Bril'kov et al. / Fuel Processing Technology 134 (2015) 487–493
Ash after the second filtration (4) and in the end product sample A recommended method for removing impurities from WFO is by
(5) should in fact be the same, because the oil after filtration goes steaming the oil using saturated water vapor at 95 °C, thereby allowing
straight into a storage tank as a final product. The significant difference the water-soluble impurities to hydrate and dissolve in the water phase
between the samples can be explained as time dependent settling of (approx. 5 w/w%), the swelling causing the particles to sediment easier
particulates during storage, since end products can be stored in the [28]. The temperature range for water extraction applied in the current
tank over various periods of time. This clearly indicates that the ash study was moderate in comparison, but for technical reasons it was not
level can be reduced further. increased further. However, the significant ash reduction observed at
50 °C and at a low water ratio (Fig. 5) can be interpreted as the effect
4.2. Ash properties of denser hydrated ash particles precipitating at a faster rate than ash
at lower temperatures.
Physical dependent experiments such as centrifugation, settling
and filtration demonstrated that most of the ash exists as solid parti-
4.4. Conclusion
cles, having mass and could be removed by gravitational and size de-
pendent methods of separation. Apparently, a minor fraction of the
In conclusion, examination of the current process revealed that two
ash in WFO is intrinsic to pure oils, but most come from the contact
stage filtration and time-dependent settling removed up to 96% of the
with immersed food items at point of use. During frying, spices and
initial ash, but still not sufficient to obtain a predictable output indepen-
salt added to the food will also partly remain in the used oils as ash
dent of the WFO input. A third filtration stage of 45 μm or below is likely
[13]. The significant organic carbon content still remaining in the
to remove a substantial fraction of the remaining ash as an alternative to
ash after 600 °C was illustrated by combustion at 800–1000 °C.
a lengthy settling in the end product. Water is already arbitrarily intro-
Therefore, given the fact that combustion temperature of industrial
duced early in the process line during washing of containers; however, a
burners is higher than 800 °C [25], it may be concluded that the actu-
controlled step of aqueous extraction might provide more efficient ash
al ash level through the chimney most likely will be lower than indi-
removal prior to filtration. As a polishing step to obtain consistent ash
cated in the analysis.
levels close to or below 0.015%, high speed centrifugation is suggested.
Sufficiently high combustion temperature in the burner might compen-
4.3. Effect of physical separation methods
sate for any of these measures.
[17] D.C. Boffito, C. Pirola, F. Galli, F. Di Michele, C.L. Bianchi, Free fatty acids esterification [23] ISO 3961, Animal and Vegetable Fats and Oils — Determination of Iodine Value, In-
of waste cooking oil and its mixtures with rapeseed oil and diesel, Fuel 108 (2013) ternational Organization for Standardization, Geneva, Switzerland, 2009.
612–619. [24] H.D. Belitz, W. Grosch, P. Schieberle, Food Chemistry, Springer, Verlag Berlin Heidel-
[18] Bernard RD, Gardned JG, Ueki J, inventors; Filtercop Inc., assignee. Cooking Oil Filter. berg, 2009.
United States Patent US 4988440. 1991 Jan 29. [25] C.E.J. Baukal, Industrial Burners Handbook, CRC Press LLC, 2004.
[19] Gyann J, inventor; GyCor International Ltd., assignee. Method of filtering spent [26] P. Doran, Unit operations, Bioprocess Engineering Principles, 2nd ed.Academic Press,
cooking oil. United States Patent US 4764384. 1988 Aug 16. Waltham, MA, USA, 2013, pp. 460–466.
[20] Hoover EF, inventor; Wise Potato Chip Company, assignee. Method of repurifying [27] S.S. Gujral, A. Popli, Introduction to Diels Alder reaction its mechanism and resent
cooking oils used in deep-fat frying operations. United States Patent US 3231390. advantages: a review, Indo Am. J. Pharm. Res. 3 (4) (2013) 1–24.
1966 Jan 25. [28] J. Cvengroš, Z. Cvengrošová, Used frying oils and fats and their utilization in the
[21] Matsumura M, Murakami S, inventors; Matsumura M, Murakami S, assignee. United production of methyl esters of higher fatty acids, Biomass Bioenergy 27 (2004)
States Patent US 6364917 B1. 2002 Apr 2. 173–181.
[22] Norwegian Environmental Agency [Internet]. Pollution regulations, Section 7,
Chapter 27: pollution from combustion plants fired with clean fuels; [cited 2014
Nov 18]Available from: https://lovdata.no/dokument/SF/forskrift/2004-06-01-931/.