A Simple Standardization Method for the Biodiesel ... | Springer Link

2 downloads 265 Views 510KB Size Report
Aug 2, 2015 - Wal-Mart. Bentonville, AR. M. “ White Rose. 12/06/2014 19:13 N. White Rose Inc. Carteret, NJ. N. “ Carlini. Undesignated. Aldi Inc. Batavia, IL. O.
J Am Oil Chem Soc (2015) 92:1357–1363 DOI 10.1007/s11746-015-2695-3

ORIGINAL PAPER

A Simple Standardization Method for the Biodiesel Cold Soak Filtration Apparatus Michael J. Haas1 · Meredith R. Barr1 · John Phillips1 · Karen M. Wagner1 

Received: 24 March 2015 / Revised: 10 July 2015 / Accepted: 16 July 2015 / Published online: 2 August 2015 © AOCS (outside the USA) 2015

Abstract  Commercially available refined vegetable oils were investigated as calibration standards for the filtration device and protocol specified by ASTM D7501 for conducting the biodiesel cold soak filtration test. Filtration time was determined to be a function of the amount of vacuum applied during filtration, with an 8 % change in the filtration time of soybean oil occurring across the vacuum range specified by ASTM D7501. At a constant vacuum of 57 cm Hg the mean filtration time of 150 mL of soybean oil was independent of operator, device, and oil lot number. Mean filtration time was also largely independent of brand: the average of the mean filtration times of replicate samples of seven brands of soybean oil was 396 s with a minimum significant difference (MSD) of 28 s, and the filtration times of seven of eight brands of soybean oil tested fell within this MSD. Refined edible-grade corn, canola, peanut, safflower and sunflower oils gave reliable filtration times and would be suitable standards. Each oil exhibited a characteristic filtration time, all greater than that for soy oil. Filtration times were an approximately linear function of kinematic viscosities, as predicted by Darcy’s Law. Edible vegetable oils can serve as reliable, affordable, consistent and generally available materials for confirming the

Eastern Regional Research Center: Mention of brand or firm names is for the purposes of identification only and does not constitute endorsement by the U.S. Department of Agriculture over others of a similar nature not mentioned. USDA is an equal opportunity employer. * Michael J. Haas [email protected] 1



U.S. Department of Agriculture, Agricultural Research Service, Eastern Regional Research Center, Wyndmoor, PA 19038, USA

operability of the filtration device used in the biodiesel cold soak filtration test. Keywords  ASTM D6751 · ASTM D7501 · Biodiesel · Cold soak filtration test Abbreviations CSFT Cold soak filtration test

Introduction The development and application of quality specifications are essential to the reliability and thus the adoption of a commercial product. In the case of biodiesel, quality standards are specified by ASTM D6751 [1]. As with all useful standards this specification is flexible and dynamic, expanding in depth and breadth as field experiences indicate necessary new quality parameters. Soon after the implementation of statewide mandatory biodiesel addition to petroleum diesel fuel in the U.S. state of Minnesota in the fall of 2005 instances of engine inoperability due to fuel starvation resulting from clogged fuel filters were experienced. It was subsequently found that in many instances the filters were plugged by solids that had formed post-production in the biodiesel component of the fuels, that the solids had formed at temperatures above those at which they were expected based on the cloud point test applied to diesel fuel, and that they were composed largely of saturated monoacylglycerols and sterol glucosides [2, 3]. An empirically developed test, the cold soak filtration test (CSFT), was found to be a reliable predictor of the tendency of a biodiesel to form solids at temperatures above the cloud point. An official protocol for the CSFT that performed acceptably with biodiesels from

13

1358

J Am Oil Chem Soc (2015) 92:1357–1363

Fig. 1  Filtration apparatus employed. As per ASTM D7501-12. Reprinted, with permission, from ASTM D750112, Standard Test Method for Determination of Fuel Filter Blocking Potential of Biodiesel (B100) Blend Stock by Cold Soak Filtration Test (CSFT), copyright ASTM International, 100 Barr Harbor Drive, West Conshohocken, PA 19428. A copy of the complete standard may be obtained from ASTM, http://www.astm.org

a variety of feedstocks, including those with higher cloud and pour points than soybean oil biodiesel, was developed, approved as ASTM D7501 [4] and became a component of ASTM D6751. In combination with other advances, the implementation by the industry of technologies to ensure that biodiesel passed the CSFT has led to a steady improvement in the quality and reliability of biodiesel [5, 6]. The CSFT consists of three successive actions: (A) incubation of 300 mL of fuel at 4.5 °C, the ‘cold soak’; (B) a slow rewarming to 25 °C; and (C) filtration under defined conditions. The accumulation of fuel-borne solids on the filter reduces the flow rate, increasing the filtration time, which is the measured parameter. The original Standard called for a maximum acceptable filtration time of 360 s. Fuel meeting this specification is now termed grade ‘No. 2-B’. To further ensure low temperature operability a second grade, ‘No. 1-B’, intended for cold weather use and characterized by a maximum 200 s filtration time, was specified by ASTM D6751 in 2012. Longer filtration times are associated with increased risk of fuel filter plugging during real world use. The filtration step involves passage of the fuel sample, within a range of specified vacuums, through a specified filter medium in a defined physical setup (Fig. 1). Filtration rate will be affected not only by the presence of solids by also by system features such as the constancy of the vacuum, the accuracy of the vacuum gauge by which it is measured, the vacuum integrity of the several fittings and couplers in the system, the quality of the filter and of its seal to the filtration device, and by operator awareness and skill. Other factors, such as whether the low temperature incubation and the subsequent reheating are conducted properly, will also influence the filtration rate.

13

The integrity of the CSFT and the reliability of its results can dictate whether a batch or lot of biodiesel will enter commerce. Standardization or ‘control assay’ protocols are typically developed to assure the reliability of quantitative assays. However, we are not aware of such a protocol being described to date for the CSFT, and have sought to develop one. Our goal was to identify a readily available liquid that gave reproducible filtration times when filtered with the CSFT apparatus under the conditions specified by ASTM D7501. In initial work we tested constructed solutions whose flow rates during filtration were dictated by the presence of suspended solid crystals. However, this approach gave poor performance and reproducibility due to variations in the formation, morphology, stability and mixing of the crystals. The filtration rate of a solution is a function not only of its solids content but also of its viscosity, as described by Darcy’s Law [7]:

R = k P/U

(1)

where R is the filtration rate, P is pressure drop across the filter, U is viscosity, k is a constant. A homogeneous liquid could be a useful and desirable standardization solution since it lacks the difficulties noted above for solutions containing solids. We therefore explored the possibility that a liquid of appropriate viscosity could serve as an acceptable filtration standard in the CSFT. In initial studies automotive products such brake fluid, power steering fluid, antifreeze, and various viscosities of engine oil were tested. For various reasons all were deemed unacceptable. Aqueous dilutions of commercial edible corn syrup were also tested since disposal of an

J Am Oil Chem Soc (2015) 92:1357–1363

1359

aqueous standard would be straightforward and inexpensive. However, the reproducibility of filtration times was poor with corn syrups. Commercially available refined edible vegetable oils showed potential as filtration standards and were more thoroughly investigated, as described below.

Materials and Methods Materials Commercial edible refined vegetable oils (Table 1) were obtained from local food stores and used before their listed ‘Best By’ dates. Oils of the same type and brand were segregated by production date code or ‘Best By’ date (referred Table 1  Edible refined vegetable oils studied

to here as ‘lot’). Analysis of acylglycerol contents by HPLC [8] showed that all oils contained at least 97 % triacylglycerols, with the balance primarily diacylglycerols. Alkaline transesterification and GC analysis [9] indicated fatty acid compositions consistent with the labeled contents of each oil [10]. The sunflower oil was of the mid-oleic type, with oleic and linoleic acid contents of 64 and 26 %, respectively. The safflower oil was a high oleic variety, containing 6 % palmitic, 2 % stearic, 74 % oleic acid, 16 % linoleic acid, and 0.3 % linolenic acids. Biodiesel produced from soybean oil and meeting ASTM D6751-13 was generously provided by Renewable Energy Group Inc., Ames, IA. Glass microfiber filters (47 mm diam.) were Whatman Brand (GE Healthcare UK Limited, Little Chalfont, UK).

Designation Type

Brand

Lot no.

Producer/distributor

A

Soybean

9/22/2014 D9:58

B C D E F G H I J

“ “ “ “ “ “ “ “ “

Essential everyday “ “ “ “ “ “ “ “ America’s Choice

K



Foodhold USA

7/16/2015 1004374

L



Great Value

1/19/2015 12:19

M



White Rose

12/06/2014 19:13 N

N



Carlini

Undesignated

O



Crisco

12/03/2014 2338420 04:25

P



Wesson

Q

Peanut

LouAna

2130317000X12063 12/1/2014 Jun 15/2015 Z01:34

R

LouAna

May 25/2015 Z01:54

S

Safflower Canola

Crisco

Jun 25, 2015 3175420

T

Corn

Crisco

May 02/2015 3122420 00:42

Supervalu Inc., Eden Prairie, MN “ “ “ “ “ “ “ “ Pathmark Supermarkets The Great Atlantic and Pacific Tea Company, Montvale, NJ Giant Foodstores Carlisle, PA Wal-Mart Bentonville, AR White Rose Inc. Carteret, NJ Aldi Inc. Batavia, IL J.M. Smucker Co. Orrville, OH ConAgra Foods http://www.conagrafoods.com Ventura Foods, LLC Brea, CA Ventura Foods, LLC Brea, CA J.M. Smucker Co. Orrville, OH J.M. Smucker Co. Orrville, OH

U

Sunflower

Foodhold

06/11/15A CSC Lot 1003324 11:07A

Giant Foodstores Carlisle, PA

12/14/2014 D2:35 11/30/2014 8:35 N 12/05/2014 1:02 N 1/10/2015 2:53 N 12/26/2014 16:51 N 11/24/14 D5:40 1/1/15 D1:27 1/3/15 17:43 N 12/14/2014 10-15 N

13

1360

Filtration System and Use Vacuum filtration devices were assembled as specified by ASTM D7501-12 (Fig. 1). Vacuum was provided either by a laboratory benchtop vacuum pump or by house vacuum. The vacuum was measured with a recently calibrated VWR Model 23609-222 Traceable Pressure Meter (VWR International, http://www.vwr.com). Volumes of 150 mL of vegetable oil were chosen for study since these mimicked real world durations of use for the system, giving filtration times of 300–400 s, an amount of time comparable to that for which the system would be under vacuum for a slow-filtering biodiesel sample that failed the 360 s CSFT limit specified by ASTM D6751. Oil samples were poured into the filtration device with vacuum already applied. The greater viscosity of vegetable oils relative to biodiesel increases their tendency to cling to the vessel wall. Therefore to ensure consistent transfer of test material to the filter we followed the stipulation of ASTM D7501 that the sample container be drained for 10 s after the bulk of the liquid was poured onto the filter. This gave reliable and essentially complete sample transfer. Filtration time was recorded as the time from addition of the last drops of sample to the filtration device to the moment when portions of the filter became noticeably brighter upon going dry due to the passage of all the test liquid. An air bleed valve in the vacuum system allowed the operator to maintain the vacuum at the desired value throughout filtration. Work was conducted at room temperature (23 ± 1 °C). Unless otherwise noted, filtration times were determined in duplicate. Viscosity Measurement Kinematic viscosities of commercial vegetable oils were determined in duplicate in a controlled temperature oil bath held at 23 °C with a size 150 Cannon-Ubbelohde viscometer (Cannon Instrument Co., State College, PA) as per manufacturer’s instructions. Data Analysis Results are presented as the means ± standard deviations of replicate determinations. The statistical significance of differences in mean filtration times were assessed at p = 0.05 using the general linear models procedure (GLM) of the SAS/STAT statistical analysis package (SAS Institute, Inc., Cary, NC, USA).

Results and Discussion Validation of System Operation When subjected to the full cold soak protocol as per ASTM D7501-12, 300 mL of biodiesel had a filtration time of

13

J Am Oil Chem Soc (2015) 92:1357–1363

91 ± 5 s (4 replications) on the filtration systems employed here. This is comparable to the CSFT value of 82 s reported for this biodiesel by its manufacturer, suggesting that the test devices were functioning correctly. Relationship Between Applied Vacuum and Filtration Time Using commercial soybean oil (Table 1, sample A) we explored the effect on filtration time of various values of applied vacuum within the range specified by ASTM D7501 and found a substantial effect (Fig. 2). As would be expected, a higher vacuum caused faster filtration, resulting in a nearly linear negative relationship between filtration time and applied vacuum (Fig. 2). Over the range of vacuums tested the filtration times of soy oil fell by 8 % as vacuum increased. It is possible that the filtration times for biodiesel samples could exhibit a similar dependence on applied vacuum. ASTM D6751 specifies a single acceptable maximum filtration time, not a range of times. A fuel with a failing (i.e. long) filtration time under minimum acceptable vacuum may pass the test when filtered at a higher, though still acceptable, vacuum. It is unclear if such a fuel would subsequently cause filter blockage in the field. A tightening of the range of specified allowable vacuums may eliminate such difficulties should they occur. All subsequent filtration times reported here were collected at a vacuum of 22.5 ± 0.5 in. (57 ± 1 cm) Hg, the midpoint in the range specified by ASTM D7501. Dependence of Filtration Time on Operator and System Variation Two separate filtration units were constructed as per Fig. 1 and used by two different operators to conduct the filtration of a single lot of soybean oil (Table 1, ‘A’). Each operator filtered ten 150-mL aliquots of oil over one of the filtration units. Mean filtration times for data obtained by the two operators were not significantly different: 404 ± 6 and 404  ± 11 s. Exchanging the units between the two operators and conducting another series of filtrations gave similar results. Thus, refined soybean oil gave highly reproducible filtration times irrespective of operator and filtration apparatus. Lot‑Dependent Impacts on Soybean Oil Filtration Time To examine the effects on filtration rate of lot-based variations within a brand of soybean oil, duplicate samples from eight lots of one brand (Table 1, Samples B through I) were studied. The average of the mean observed filtration times for the eight lots was 394 s with a standard deviation of 4.9 s and a minimum significant difference (MSD) of 21 s. The difference between the largest and smallest average filtration times was 22 s, one second greater than the MSD,

J Am Oil Chem Soc (2015) 92:1357–1363

1361

Fig. 2  Relationship of the filtration time of 150 mL of oil A (soybean, Table 1) to applied vacuum. Data points are the means of replicate determinations

430 425 420

Filtraon Time (sec)

415 410 405 400 395

y = -9.342x + 620.72 R² = 0.8688

390 385 380 20.5

21

21.5

22

22.5

23

23.5

24

24.5

25

25.5

Applied Vacuum (inches Mercury)

Table 2  Mean filtration times of vegetable oils Oil type

Oil sample

Mean filtration time (seconds)

Soybean Peanut Safflower Canola Corn

Table 1, Samples B − I Table 1, Sample Q Table 1, Sample R Table 1, Sample S Table 1, Sample T

404 ± 11 525 ± 1 587 ± 18 452 ± 1 409 ± 1

Sunflower

Table 1, Sample U

475 ± 13

indicating a lot-based impact on filtration rate. However, for an assay such as the CSFT where operator judgment, experience and reaction time impact the measured filtration time, a one second difference is negligible. For all practical purposes it can be said that filtration time was independent of production lot for soybean oil. Brand Dependence of Soybean Oil Filtration Time Filtration times were measured on duplicate samples of seven brands (Table 1, Samples J–P) of edible refined soybean oil. The average of the resulting mean filtration times was 396 s with a minimum significant difference of 28 s. Filtration times for six of the seven brands, as well as for the brand studied above in the lot-to-lot work which was not included in this test, were within the MSD. An eighth brand (Table 1, Sample J), however, had a mean filtration time of only 311 s, nearly 100 s below that of the others tested. The reason for its much faster filtration is not known. With the exception of this sample it appears that refined soybean oils can serve as effective standards for

calibration of the CSFT filtration apparatus and will give comparable CSFT times irrespective of brand. The faster flowing brand may also serve well as a system check material, though giving smaller CSFT values than typical for soybean oil. Potential of Other Vegetable Oils to Serve as Filtration Standards Soybean oil is the least expensive vegetable oil and would seem to be the most probable choice for use as a filtration standard. Nonetheless, we explored the possibility that other affordable and widely available vegetable oils would exhibit filtration behaviors similar to that of soy oil, allowing the use of virtually any refined vegetable oil as the standard. Table 2 lists the mean filtration times of duplicate 150-mL aliquots of peanut, safflower, canola, corn and sunflower oils, as well as the average filtration time for 8 lots of one brand of soy oil. For each oil the filtration time was highly reproducible, with replicates differing by 3 % or less. Filtration times were generally unique to each oil, although the mean filtration time of corn oil was essentially identical to that of soy oil. Further studies may show that some or all of these other oils display sufficient uniformity of filtration time across brand and lot to serve as calibration fluids for the filtration protocol. Relationship between Oil Viscosity and Filtration Time Kinematic viscosities of the test oils were determined in duplicate and mean values were plotted against the corresponding mean filtration times (Fig. 3). In accord with Darcy’s Law (Eq. 1) a near linear relationship existed between

13

1362

filtration time and kinematic viscosity. The equation of the best-fit straight line joining the data points (Fig. 3, line A) was y = 8.61x − 111.9 (R2 = 0.82). Thus the differences seen above in the filtration times of the oils can be understood in light of differences in their kinematic viscosities. Further studies to more fully identify the relationship between filtration time and kinematic viscosity for commercial vegetable oils may be appropriate, however, since reasonable quadratic fits to the data points could also be made (R2 > 0.9, data not shown). One data point, for safflower oil, appears to deviate substantially from the best-fit line relating filtration times and viscosities for the tested oils (Fig. 3, line A). A best-fit line through only the other five data points is described by the relationship y  = 6.54x  + 17.11 and gives a much better fit to those points, with an R2 of 0.93 (Fig. 3, line B). It is unclear why the safflower oil sample diverged so noticeably from the filtration time—kinematic viscosity relationship established by the other oils. The sample was clear and bright, suggesting that it did not contain fine particles that would have increased its filtration time. GC and HPLC indicated a typical safflower oil composition. Further work may indicate whether this is a typical feature of safflower oil or perhaps a trait of only the particular sample studied here. Its aberrant performance suggests that safflower oil may not be a reliable filtration standard. Notwithstanding this observation, vegetable oils other than soy oil may generally be suitable for use as filtration standards, with the filtration time being unique for each oil.

Fig. 3  Correlation of filtration time and kinematic viscosity at 23 °C for various commercial refined vegetable oils. Reading from the left, the data points are for: soybean, corn, canola, sunflower, safflower, and peanut oils. Values are the means of replicate determinations of filtration time and viscosity, except for soybean oil whose filtration time value is the average of the means of replicate determinations for samples O, K, L of Table 1. A Linear fit to all data points R2 = 0.82. B Linear fit to all points except safflower oil. R2 = 0.93

13

J Am Oil Chem Soc (2015) 92:1357–1363

The standardization protocol described here does not require the time-consuming ‘cold soak’ and warmup steps of ASTM D7501, and the vegetable oils used as test fluids are inexpensive, widely available, consistent over time and space, nontoxic and can be readily disposed. Furthermore an attractive additional feature of the use of vegetable oils is that they do not introduce a new (and petroleum-based, in the case of, e.g., the use of isoparaffins) waste stream to the operation but rather can be recovered and transesterified to yield biodiesel if desired. This would enhance the profitability of the operation rather than introducing new disposal costs. The viscosities of vegetable oils are such that acceptable durations of filtration were achieved with 150 ml of the sample rather than the 300-ml volume used in testing biodiesel, which reduces reagent cost and liquid handling and disposal. These oils cannot, however, be reliably recovered and reused in repeated standardizations: the filtration times of reused samples were as much as 25 % lower than obtained on first filtration (data not shown).

Conclusions We have determined that within the vacuum range specified by ASTM D7501 for measuring the cold soak filtration time of biodiesel the filtration time of a sample is dependent on the degree of applied vacuum. Substantial variations in filtration time were observed across the range of vacuum values specified by the ASTM Standard. Within this vacuum range the filtration time of commercial refined edible soybean oil over a properly functioning filtration device was constant across variations in unit, operator, brand and lot. Soybean oil thus is an acceptable reference material for validation of correct functioning of a filtration device intended for use in ASTM D7501. Other refined vegetable oils may also be suitable reference filtration standards, though it is notable that they are generally more expensive than soybean oil. Commercially available edible vegetable oils represent readily and widely available calibration fluids that are affordable, consistent in properties and quality over time and space, and whose disposal, particularly if converted to biodiesel following their use, does not impose burdens such as the introduction of novel new waste streams on a production or testing facility.

References 1. American Society for Testing and Materials (2012) Standard specification for biodiesel fuel blend stock (B100) for middle distillate fuels, designation D6751-12. ASTM, West Conshohocken 2. Pfalzgraf LM, Lee I, Foster J, Poppe G (2007) Effect of minor components in soy biodiesel on cloud point and filterability.

J Am Oil Chem Soc (2015) 92:1357–1363 Biorenewable Resources No. 4; a special supplement to inform. AOCS Press, Champaign, IL. pp 17–21 3. Chupka GM, Fouts L, McCormick RL (2012) Effect of low-level impurities on low-temperature performance properties of biodiesel. Energy Env. Sci 5:8734–8742 4. American Society for Testing and Materials (2012) Standard test method for determination of fuel filter blocking potential of biodiesel (B100) blend stock by cold soak filtration test (CSFT), designation D7501-12. ASTM, West Conshohocken 5. Alleman TL, McCormick RL, Deutch S (2007) 2006 B100 quality survey results: milestone report. Golden, CO: National Renewable Energy Laboratory, Golden, CO. Technical Report NREL TP-540-41549 6. Alleman TL, Fouts L, Chupka G (2013) Quality parameters and chemical analysis for biodiesel produced in the United States

1363

7. 8. 9. 10.

in 2011. National Renewable Energy Laboratory, Golden, CO. Technical Report NREL/TP-5400-57662 Anonymous. (2014) Darcy’s Law. http://hlsweb.dmu.ac.uk/ahs/ elearning/RITA/Filtration/Darcy.html. Accessed 15 Jul 2014 Haas MJ, Cichowicz DJ, Jun W, Scott K (1995) The enzymatic hydrolysis of triglyceride-phosphoglyceride mixtures in an organic solvent. J Am Oil Chem Soc 74:519–525 Juneja VK, Foglia TA, Marmer BS (1998) Heat resistance and fatty acid composition of Listeria monocytogenes: effect of pH, acidulant and growth temperature. J Food Protect 61:683–687 Hammond EW (1993) Chromatography for the analysis of lipids. CRC Press, Boca Raton, p 174. ISBN:0-8493-4255-4

13