Fate of resin acids in pulp mill secondary treatment systems

Water Research 36 (2002) 2878–2882

Fate of resin acids in pulp mill secondary treatment systems Stephen P. Makris, Sujit Banerjee* Institute of Paper Science and Technology, 500 Tenth Street NW, Atlanta, GA 30318, USA Received 1 December 2000; accepted 1 September 2001

Abstract Profiles of resin and fatty acids (RFAs), COD, and aquatic toxicity were measured across the secondary treatment systems of three pulp mills. The RFAs sorb to suspended solids, principally fiber, and are partially removed through settling. An activated sludge system is more efficient in removing RFAs than an aerated stabilization basin (ASB) because of its higher solids level. Dehydroabietic acid accounts for a significant fraction of the effluent toxicity in the two ASBs studied. The microorganisms in an ASB are unable to respond rapidly to an RFA spill, and effluent toxicity can be elevated for a prolonged period because of hydraulic backmix. The applicability of several laboratory studies to field situations is assessed. r 2002 Elsevier Science Ltd. All rights reserved. Keywords: Sorption; Toxicity; Resin; Acid; Fate; Treatment

1. Introduction Although many of the constituents that enter secondary treatment are toxic to aquatic organisms, a sizable fraction of these are removed during treatment [1]. Resin acids occur naturally in softwood and have been repeatedly implicated as contributors to effluent toxicity [2–6]. Of these, dehydroabietic acid (DHA) is of particular concern because it can be anaerobically reduced to retene [4], which is toxic to aquatic organisms. Reports on the mechanism of resin acid removal are conflicting. Zender et al. [7] attribute the removal of DHA under field conditions to biodegradation, and DHA has been shown to be biodegradable in the laboratory [4]. Yet, laboratory-scale work shows that even microorganisms acclimated to resin acids were unable to degrade a shock load because of a long lag period [8,9]. DHA and other resin acids are hydrophobic species capable of sorption to fiber and biomass present in the treatment system. Hence, sorption of DHA to suspended solids and their subsequent settling must play some role. Our principal objective in this study is to *Corresponding author. Tel.: +1-404-894-9709; fax: +1404-894-4778. E-mail address: [email protected] (S. Banerjee).

study the importance of sorption/settling of DHA through studies on full-scale systems and to determine its implications on effluent toxicity.

2. Experimental The bulk of the study was conducted at Mill A, which is located in the southeastern US. Its pulp production consists of 20% bleached hardwood, 30% bleached softwood, and 50% unbleached softwood. Its (4
106 m3) aerated stabilization basin (ASB) contains four reactors with a total retention of 30–40 days. Samples were taken on three occasions in 1999, and once in 2000. The first reactor was not aerated during 1999. The second and third reactors housed six and twelve 30 kW aerators, respectively. In early 2000, 900 kW of aeration was added to the first reactor and 270 kW was removed from the third; the samples collected in 2000 reflect this change. Grab samples were collected from the inlet and outlet of the first reactor, the outlets of the second and third, and from the final effluent. These samples are designated 1-in, 1-out, 2-out, 3-out, and final, respectively. Some of the samples were filtered through a Whatman 934AH glass fiber filter to determine the total suspended solids (TSS). The difference between the

0043-1354/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved. PII: S 0 0 4 3 - 1 3 5 4 ( 0 1 ) 0 0 4 9 1 - 2

S.P. Makris, S. Banerjee / Water Research 36 (2002) 2878–2882

3. Results and discussion The long hydraulic residence time associated with the Mill A lagoon makes the concentration profiles along the lagoon difficult to interpret, since the input to the lagoon is not necessarily constant. The concentration of a compound at a given location reflects not just the various degradative and dissipative pathways, but also a varying input load. Fig. 1 compares filtered and unfiltered RFA concentrations. The difference is linear with TSS as shown in Fig. 2, which confirms that the insoluble RFA is bound to TSS. As expected, the TSS decreases progressively across the lagoon; e.g., for the December 1999 sampling it dropped from 82 ppm at 1out to 26 ppm at the final effluent. The Fig. 2 plot only includes values between 1-in and 2-out, since the TSS is low beyond that point and the differences in RFA are small. Hence, settling must be a removal mechanism for the RFAs bound to the TSS. Biodegradation may, of course, occur after settling, but this is outside the scope of our study. The inlet sample was taken from the inflow to the treatment system and the solids are mainly wood fiber. Consider the solids : water distribution at locations 1-out and 2-out in Fig. 1. Values for the (dimensionless) fiber : water distribution coefficient, Kd ; were identical at the two locations at 7000 for the December 1999 samples; the corresponding values for March 2000 were 15,200 and 15,600, respectively. If only resin acids are considered, the 1-out value for the December sampling was 6500. The partitioning of resin acids between inactivated aerobic biomass and water follows a linear isotherm [17], with a much lower Kd of 300–1100. Hence, our values are far too high to reflect simple partitioning to biomass. Hydrophobic compounds sorb

ppm

20

Mar 00

10

0 20

ppm

filtered and unfiltered resin and fatty acid concentrations represents the amount bound to the TSS. Mill B is also located in southeastern US and makes only bleached product from softwood. Samples were collected once in 2000. The 1.4
106 m3 lagoon receives 2130 kW of tapered aeration and is curtained into three zones, with 80% of the aerators being located in the first zone. Samples were taken at the inlet and outlet of the treatment system, as well as at six locations across the system based on the distribution of aeration power, i.e. 80% of the samples were collected from the first zone, with the remainder split between the second and third zones. Mill C is located in New Augusta, MS, and makes bleached pulp. Its fiberline swings between hardwood and softwood production as described in [10]. It uses an activated sludge system (AST) with a retention time of 17 h. Eight samples were collected semiweekly in 1994. Chemical oxygen demand (COD) and TSS were measured in the field according to standard procedure [11]. Samples collected for toxicity measurements were cooled to 41C and shipped on ice overnight. Resin acid samples were collected and stored in borosilicate glass bottles with Teflon-lined caps at pH 10 and maintained at 41C until tested. The resin and fatty acids determined were abietic, dehydroabietic, neoabietic, pimaric, isopimaric, sandracopimaric, palustric, oleic, and stearic acid, and are collectively designated as ‘‘RFA’’. Aqueous samples were extracted with diethyl ether and the RFAs determined by gc-fid [12]. Microtox analyses were performed following the whole effluent toxicity (WET) protocol [13], with a Microtox Model 500 Analyzer obtained from AZUR Environmental. Microtox assays have been correlated to both acute [14] and chronic toxicity [15]. Ceriodaphnia dubia 7-day survival and reproduction assays [16] were run in the final effluent of Mill A (amended on occasion with DHA) by Law Engineering and Environmental Services, Kennesaw, GA. Microtoxs toxicity results are reported as the EC50, or the effective sample concentration that causes a 50% reduction in light output. Chronic toxicity (C. dubia) is presented as the lowest observable effective concentration, or LOEC. The LOEC is the lowest concentration tested that statistically differs from the control with respect to the average number of neonates reproduced per organism. Dehydroabietic acid (DHA) dose–response curves were generated for both the acute and chronic toxicity protocols. Laboratory studies were performed to establish the relationship between DHA and Microtoxs acute toxicity in the wastewater matrix. Filtered, treated, final effluent was spiked with 30 ppm DHA and 2:1 serial dilutions were made at pH 8. The test was done in triplicate using the WET protocol [13].

2879

Dec 99

10

0 1-in

1-out

2-out

3-out

final

Fig. 1. Filtered (triangles) and unfiltered (circles) RFA concentrations.

S.P. Makris, S. Banerjee / Water Research 36 (2002) 2878–2882

2880

The level of aeration was increased before the March 2000 sampling, and this leads to an earlier drop in COD as compared to the December result. The toxicity is also reduced earlier in March, reflecting the removal of some of the toxicants. These results confirm that many of the compounds that induce acute toxicity are treated. However, the decrease in chronic toxicity is not as pronounced. Profiles of DHA, RFA, and COD (all filtered) collected on two occasions are presented in Fig. 4. The March results are unexceptional; all three constituents decrease in Reactor 1. The December results, however, show a DHA/RFA spike at the outfall of Reactor 1. This originates from an earlier spill, as confirmed by the mill. The COD decreases smoothly and the profile resembles those (not shown) collected on other occasions in 1999. Hence, the COD and RFAs must be removed through different mechanisms, at least in a spill situation. Since RFAs are among the more recalcitrant COD constituents, they are not degraded as rapidly and appear as a broad pulse. This possibility was suggested earlier by Werker and Hall [8,9], who found in laboratory-scale work that microorganisms acclimated to resin acids were unable to degrade a shock load because of a long lag period. Hence, if a spike clears the front end of the lagoon where biological activity is most intense, it could travel through the system relatively unaffected. Fig. 4 provides full-scale confirmation of Werker and Hall’s laboratory findings. The Microtox EC50 of DHA-spiked final effluent (Mill A) was 3.7 ppm with an a ¼ 0:5 significance range of 3.0–4.1 ppm. In the absence of a spill, the effluent DHA averaged about 30% of this value as shown in Fig. 5 for the May 1999, September 1999, and March 2000 samplings. If RFAs are spilled into the lagoon, their concentrations will be increased until the spike is washed out. If plug flow conditions apply, then a high RFA concentration would be found in the effluent for a

strongly to the lignin in pulp fiber and fines [18,19], and it is likely that the resin acids are principally bound to fiber and fiber-derived fines rather than to microbial biomass at 1-out. The fibers progressively settle out, and the resin acids in the unfiltered and filtered samples eventually converge. This is not to minimize the importance of sorption to biomass. However, if fiber constitutes a significant fraction of the initial TSS, then the resin acids will substantially associate to and settle out with the fiber. COD and toxicity profiles of filtered samples are provided in Fig. 3 and show that the toxicity decreases just after the bulk of the COD is removed. The chronic and acute toxicity profiles correlate only to the extent that both decrease just after the first reactor for the December sampling and within the first reactor for the March episode, which benefited from increased aeration.

8

∆RFA (ppm)

6

4

2

0 20

40

60

80

100

TSS (ppm) Fig. 2. Relationship between bound RFA and TSS.

1000 900

ppm

720

COD

700

560

600 Microtox

100 75 50

C. dubia

25 0 1-in

1-out

2-out

3-out

final

COD

800

640

LOEC or EC50 (%)

LOEC or EC50 (%)

ppm

800

Microtox

100 75

C. dubia

50 25 0 1-in

1-out

2-out

3-out

final

Fig. 3. COD and toxicity profiles for filtered samples collected in December 1999 (left) and March 2000 (right). EC50 and LOEC apply to the Microtox and C. dubia, respectively.

S.P. Makris, S. Banerjee / Water Research 36 (2002) 2878–2882 800

1000 COD

ppm

ppm

750 700 650

800 700

16

16

12

ppm

RFA

8 4 0 1-in

1-out

2-out

COD

900

600

ppm

2881

12

RFA

8

DHA

4

3-out

0 1-in

final

DHA 1-out

2-out

3-out

final

Fig. 4. COD, RFA, and DHA profiles for filtered samples collected in December 1999 (left) and March 2000 (right).

ppm

20

Mar 00

10

ppm

0 20

a

Dec 99

ppm

Mill A Mill B a

10 0 8

Sep 99

4 0 8

ppm

Table 1 Comparison of DHA and RFA levels (ppm) across Mills A and B

May 99

4 0 1-in

1-out

2-out

3-out

final

Fig. 5. Total (unfiltered) RFA (triangles) and DHA (circles) profiles collected at various periods.

short period. Mixing will lower this concentration but will lengthen the duration of elevated RFA levels in the effluent. The EC50 for reproduction to C. dubia measured in Mill A effluent was 2 ppm. A higher LC25 value of 6.6 ppm was reported earlier [20], but this test was run in well water where the contribution of other toxicants was absent. Hence, the final effluent DHA value is just below the LC50 threshold in the absence of a spill, and may exceed it for a prolonged period under upset conditions. Roughly similar results were obtained for Mill B and are compared to those from Mill A in Table 1. A direct

DHAin

DHAout

RFAin

RFAout

3.0 8.0

1.7 2.4

9.6 39

4.5 7.1

Averaged over four sampling episodes.

comparison of samples taken within the lagoon across the two mills is difficult, since the geometry and mixing characteristics are very different. Hence, only the influent and effluent results are considered. Although the influent values were higher (on average) than those of Mill A, the final effluent concentrations were similar. As with Mill A, the DHA/RFA ratio increased from influent (0.21) to effluent (0.34), confirming the relative recalcitrance of DHA. A similar recalcitrance has been reported in a full-scale study [7]. Mill C alternated between softwood and hardwood production; the RFAs in the influent changed concomitantly. Over the eight sampling occasions, the unfiltered DHA and RFA influent concentrations were 472 and 30720 ppm, respectively; the corresponding effluent values were 0.270.2 and 171 ppm, respectively. The relatively high uncertainty results from the fiberline swing between hardwood and softwood. As before, the DHA/RFA ratio increased from the influent to the effluent, in this case, from 0.13 to 0.19. DHA decreases by 95% across the treatment system, as compared to 59% and 70% for Mills A and B, respectively. Although possible, it is unlikely that this increase is entirely due to improved biological action, since the efficiencies of ASTs and ASBs are similar for other constituents [10]. The suspended solids levels in an AST are much higher than those in an ASB, and sorption and settling is expected to play a larger role in an AST. These results are consistent with the results of an earlier field study

2882

S.P. Makris, S. Banerjee / Water Research 36 (2002) 2878–2882

[21] where the degradation of radiolabeled oleic acid in an AST and an ASB was compared. Sorption was found to play a more important role for the AST, which suggests that resin acid toxicity should be lower in an AST.

4. Conclusions Our major conclusion is that DHA can be responsible for a significant fraction of the chronic effluent toxicity in ASBs. Its presence in the effluent of an AST should be smaller. It is only partially removed biologically, and association with fiber or biomass and subsequent settling represents an important removal pathway. If an RFA spill is large enough to traverse the front end of an ASB, which is the region of highest biological activity, then the spike can move through the lagoon since both biological activity and the TSS level will progressively decrease through the treatment system. The spike will broaden as it moves due to mixing in the lagoon and could elevate effluent toxicity until it washes out.

Acknowledgements This study was partially funded by the member companies of the Institute of Paper Science and Technology.

References [1] Morck R, Jansson MB, Dahlman O. Resinous compounds in effluents from pulp mills. In: Back EL, Allen LH, editors. Pitch Control, Wood Resin and Deresination. Atlanta GA: Tappi Press, 2000. [2] Leach JM, Mueller JC, Walden CC. Identification and removal of toxic materials from kraft and groundwood pulp mill effluents. Process Biochem 1976;10(1):7–10. [3] Leach JM, Thakore AN. Toxic constituents in mechanical pulping effluents. Tappi J 1976;59(2):129–32. [4] Liss SM, Bicho PA, Saddler JN. Microbiology and biodegradation of resin acids in pulp mill effluents: a minireview. Can J Microbiol 1997;75:599–611. [5] Peng G, Roberts JC. Solubility and toxicity of resin acids. Water Res 2000;34(10):2779–85.

[6] Zanella A. Effect of pH on acute toxicity of dehydroabietic acid and chlorinated dehydroabietic acid to fish and Daphnia. Bull Environ Contam Toxicol 1983;30:133–40. [7] Zender JA, Stuthridge TR, Langdon AG, Wilkins AL, Mackie KL, MacFarlane PN. Removal and transformation of resin acids during secondary treatment at a New Zealand bleached kraft pulp and paper mill. Water Sci Technol 1994;29(5,6):105–21. [8] Werker AG, Hall ER. Limitations for biological removal of resin acids from pulp mill effluent. Water Sci Technol 1999;40(11,12):281–8. [9] Werker AG, Hall ER. The fate of a resin acid shock load in a biological system. Pulp Pap Can 2000;101(1):45–9. [10] Kemeny TE, Banerjee S. Relationships among effluent constituents in a bleached kraft pulp mill. Water Res 1997;1:1589–94. [11] Clesceri LS, Eaton AD, Greenberg AE. Standard Methods for Examination of Water & Wastewater. 20th ed., New York: American Public Health Association, 1999. [12] National Council for Air and Stream Improvement. Procedures for the analysis of resin and fatty acids in pulp mill effluents. Technical Bulletin No. 501, 1989. [13] Azur Environmental. Whole Effluent Toxicity Protocol, 1998. [14] Aarsand R, Berg O, Broch-Due A, Hoel H, Moen BA. Toxicity characterization of TMP process streams by different methods. Nord Pulp Pap Res J 1996;11:192–200. [15] Backman CJ, Firth BK. Comparison of Microtox testing with Rainbow Trout (Acute) and Ceriodaphnia (Chronic) bioassays in mill waste waters. Tappi J 1990;73(12):169–74. [16] Horning WB, Weber CI. Ceriodaphnia Survival and Reproduction Test Method 1002.0. Methods for Measuring the Acute Toxicity of Effluents to Freshwater and Marine Organisms, EPA Technical Report No. EPA/600/ 4-85/014, 1985. p. 58–75. [17] Hall ER, Liver SF. Interaction of resin acids with aerobic and anaerobic biomassFII. Partitioning on biosolids. Water Res 1996;30(3):672–8. [18] Severtson SJ, Banerjee S. Sorption of chlorophenols to wood pulp. Environ Sci Technol 1996;30:1961–9. [19] Severtson SJ, Banerjee S. Dual reactive domain model for sorption of aqueous organics by wood fiber. J Colloid Interfacial Sci 2001;236:362–8. [20] O’Connor BI, Kovacs TG, Voss RH. The effect of wood species composition on the toxicity of simulated mechanical pulping effluents. Environ Toxicol Chem 1992;11:1259–70. [21] Williams CL, Mahmood T, Corcoran H, Zaltzmann ME, Banerjee S. In-situ measurement of local biodegradation during secondary treatment. Application to bleached pulp mill chloro-organics. Environ Sci Technol 1997;31:3288–92.