Article pubs.acs.org/IECR
Coupling a Chemical Reaction Engine with a Mass Flow Balance Process Simulation for Scaling Management in Papermaking Process Waters Patrick Huber,*,† Sylvie Nivelon,† Pascal Ottenio,† and Patrice Nortier‡ †
Centre Technique du Papier, BP 251, 38044 Grenoble Cedex 9, France Laboratoire de Génie des Procédés Papetiers (LGP2) UMR CNRS 5518, Grenoble INP-Pagora, 461 Rue de la Papeterie, 38400 Saint Martin d’Hères, France
‡
S Supporting Information *
ABSTRACT: Papermills manufacturing recycled board usually face severe calcium carbonate scaling problems. We present a mass flow balance simulation of this papermaking process (PS2000, G2-based), coupled with a chemical reaction engine (IPhreeqc). The simulation allows predicting the pH and calcite saturation index throughout the process. Developed chemistry modules can simulate anaerobic microbial activity in the process waters, together with local contact with the atmosphere. Also, chemistry models of the integrated wastewater treatment (with anaerobic and aerobic treatment steps) are developed. The coupled simulation accurately describes the scaling tendency of process waters. This makes it possible to study curative solutions to the scaling problems in recycled board mills. The effect of a biocide treatment is simulated and discussed. This is predicted to largely reduce scaling in the wastewater treatment and limit the amount of generated sludge.
■
INTRODUCTION Scaling is a major problem in paper mills, especially for those using recycled fibers as raw materials. The predominant scale deposits in those mills have been found to be calcium carbonate,1 as the recovered paper raw material contains calcium carbonate filler. Scaling is especially critical in the aerobic step of the integrated wastewater treatment plant.2 The papermaking industry has made considerable effort over the last decades to reduce fresh water consumption and environmental impact.3−5 However, this causes build-up of dissolved species in process loops, promotes anaerobic bacteria activity, and aggravates scaling problems.6 Computer simulations have proved very useful to design and optimize papermaking processes.7 Various applications dedicated to pulp and paper processes have been developed using some of the major simulation packages: ASPEN Plus,8,9 BALAS,10 CADSIM,11 PROSIM,12 PS2000 (G2-based),13 WinGEMS,14 and others. On the other hand, chemical speciation methods have been shown to accurately describe the scaling tendency of process waters, even in contaminated flows of closed circuit board mills.15 In this respect, various dedicated chemical softwares may be used: CHEMSHEET,16 FRENCHCREEK,17 OLI18,19 ORCHESTRA,20 PHREEQC,21 Visual MINTEQ,22 and others. Several bridges between process simulation softwares and chemical engines have been developed in order to take into account electrolyte chemistry problems (such as ASPEN-OLI23 or BALAS-CHEMSHEET for papermaking processes24). The first documented coupling of IPhreeqc modules25 with an external simulation (COMSOL Multiphysics) was reported in the field of geochemistry.26 Our research group has been developing the PS2000 process simulation dedicated to papermaking processes under the G2 environment (Gensym), because of its exceptional flexibility in © 2012 American Chemical Society
terms of the human−machine interface and object-oriented development of simulation models. This tool allows one to perform both static and dynamic process simulations.27 Many papermaking processes could be successfully simulated using the developed mass balance approach,28 with numerous applications for implementation studies of new process equipments in the industry and as a tool for investment decision-making. However, some shortcomings appeared when applying these methods to the simulation of scaling in papermills, so that we needed to develop additional chemistry modules, enabling to calculate dissolution of calcium carbonate and precipitation from supersaturated streams. Supersaturation is the key parameter that determines precipitation:29 the saturation ratio is defined as ratio of the ionic activity product (IAP) to the equilibrium constant (Ks). S=
[Ca 2 +][CO32 −] IAP = Ks Ks
where [ ] denotes activity. S < 1 describes an under-saturated solution, S = 1 refers to a solution in equilibrium with the solid, while S > 1 indicates a supersaturated solution, which may lead to precipitation, scaling in our case. In this work, we present the coupling of the PS2000 mass flow balance process simulation13 with the IPhreeqc chemical reaction engine25 that extends the capabilities of the process simulation and makes it possible to accurately take into account dissolution and precipitation of calcium carbonate. A simulation model of a recycled board mill is constructed and resolved Received: Revised: Accepted: Published: 421
April 15, 2012 October 16, 2012 December 12, 2012 December 12, 2012 dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
Table 1. Summary of Process Water Analyses (“IC” = Inorganic Carbon, “VFA” = Volatile Fatty Acids, “S(6)” = sulfate) sampling point
pH
temp (°C)
Ca (mM)
IC (mM)
Mg (mM)
Na (mM)
S(6) (mM)
Cl (mM)
VFA (meq/L)
COD (mg/L)
pulper outler (1) dilution water to pulper (2) pulp to thickening (4) thickening filtrate (7) to papermachine (9) headbox (10) wire water (12) water to save-all (14) save-all filtrate (15) anaerobic treat. outlet (21) aerobic treat. outlet (22) fresh water (23)
6.64 6.40 6.63 6.54 6.42 6.33 6.47 6.49 6.48 7.08 7.90 7.35
34.2 32.0 34.9 35.9 34.0 35.3 34.6 34.5 35.7 34.9 26.5 17.1
21.2 17.8 22.4 21.3 24.7 22.9 21.9 21.1 21.0 18.7 3.4 2.5
4.3 2.2 3.2 2.4 4.4 3.3 1.6 1.9 2.9 41.2 14.3 4.8
1.1 0.9 1.1 1.1 1.1 1.1 1.0 1.0 1.0 0.9 0.8 0.3
11.8 9.3 10.8 10.6 10.6 10.0 9.8 9.4 10.1 10.3 10.1 0.4
3.5 2.3 3.0 3.4 3.2 3.9 2.7 2.7 3.2 2.6 0.8 0.3
4.7 3.2 4.0 4.1 4.1 4.2 4.5 4.2 4.5 7.0 3.9 0.9
35.6 22.8 31.6 29.6 32.0 27.6 30.4 32.0 29.6 3.6 2.4 4.8
6460 5153 N/A N/A N/A 6500 N/A N/A 5585 1868 129 N/A
Briefly, a flow sheet of the mill is constructed with dedicated unit operations specific for the process. Several suspended solids (cellulosic fibers, cellulosic fines, mineral filler, and impurities) and dissolved species (calcium, inorganic carbon, magnesium, chloride, sulfate, acetate and COD) are included in the simulation. The full mill model is equilibrated by resolving flow and mass balances around each node and unit operation for each simulated species. The mass balances are resolved iteratively, until the difference between two successive values of a species is lower than a preset tolerance. Separation coefficients are defined for some species on particular unit operations and adjusted to match measured concentration profiles. Also, some release coefficients are defined for dissolved species (for instance, we assume that repulping the recovered paper releases defined amount of COD, chloride, and sulfate in kg/T of paper). The fresh water is also considered as a source of dissolved species. Their concentration is adjusted against measured water quality. The temperature is calculated from a classical enthalpy balance. All parameters of the simulation are given in the Supporting Information. In the original implementation of the simulation, all species were considered as chemically inert. The coupling with a chemistry engine allows one to calculate the pH in each flow and to determine the extent of calcite dissolution or precipitation. The mineral filler species is split in two fractions: a “CaCO3_solid” species which is allowed to dissolve in or precipitate from the flow and an inert “other filler” fraction (clay, mostly). The mass balance is resolved for the suspended solids and total dissolved species concentrations (total Ca, total C (inorganic carbon)), total acetate (VFA), total Mg, total S(6) (sulfate), total Cl), corresponding to the master solution species in the chemical speciation). Also, the pH is now tracked in each flow as a “nonbalance” parameter. The principles of the coupling procedure between the process simulation software and the chemistry engine are illustrated in Figure 1. Several chemistry modules have been developed to describe scaling-related phenomena (see examples in a schematic process in Figure 2). (1) Fluxes mixing module: results of mixing several fluxes together are calculated by PHREEQC and reused in PS2000, and dissolved or precipitated amounts are adjusted between CaCO3 solid and dissolved species. The pH of the flux is calculated and followed in each point of the simulated process. (2) “Bacterial activity” module: simulates the volatile fatty acids (VFA) production under the form of acetic acid, then the flux characteristics after “acidification” by VFA and dissolution equilibrium with CaCO3 filler are calculated by PHREEQC and reused in PS2000. (3) Equilibrium with the
while taking into account chemical reactions that affect calcocarbonic equilibria. The simulation is calibrated against measured process water data. Finally, an example of a curative solution to scaling problems is studied by simulation.
■
METHODS Pulp samples were taken throughout a recycling mill located in southern France, producing linerboard from 100% recycled board. Briefly, the mill circuits comprise a pulp preparation loop and a paper machine separated by a thickening stage. Effluent is treated in an integrated two stage wastewater treatment plant (WWTP) consisting of an anaerobic step followed by an aerobic step. The mill circuits layout are best described from the corresponding simulation flow sheet (Figures 3−5). The mill has a low specific fresh water consumption of around 5 m3 per ton of produced paper. The physicochemical characterization of the samples included pH, temperature, chemical oxygen demand (COD), and measurement of the concentration of main ionic species, including organic acids. The pH and the temperature were recorded in situ in the samples (WTW multi 340i). The suspensions were then immediately prefiltered on glass fiber paper (Millipore AP20, pore size = 2 μm). The volatile fatty acid (VFA) concentration was measured by acid−base titration of the filtrate (Degrémont method, acidification to pH = 3.5 with sulphuric acid, then boiling for 3 min to strip the inorganic carbon, followed by cooling and addition of NaOH up to pH = 4, the amount of VFA (meq/L) is then equal to the required amount of NaOH (meq/L) to dose the filtrate from pH = 4 to pH = 7). The filtrate is then passed through a syringe filter (Chromafil Xtra RC-45/25, regenerated cellulose, pore size = 0.45 μm) and used for all further analyses. Sodium, magnesium, and calcium concentrations were measured by inductively coupled plasmaatomic emission spectroscopy (ICP-AES; Jobin-Yvon JY2000). Total inorganic carbon was performed by infrared absorption in the gas-phase after acidic outgassing (TOC, Shimadzu). The chloride concentration was measured using Merck KGaA Chloride 114753 test, and the sulfate concentration was measured using Merck KGaA 1.10019.0001 test. COD was measured with COD cell tests (WTW). A summary of the analyses results is given in Table 1. It should be mentioned that appropriate process water analyses for scaling tendency assessment may be time-consuming and difficult to perform as a routine test in papermills. The PS2000 process simulation was developed in the G2 environment. The methods have been presented in Ruiz et al.13 422
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
program contains all functions necessary for the chemistry modules in an executable form. Within the foreign image, all chemical methods are accessed through the IPhreeqc.dll (Dynamic Link Library).25 In the following, PHREEQC keywords are written in capital letters, while all IPhreeqc method names are written in bold letters. For each developed chemistry module, a function written in the C language describes the sequence of chemical calculations to be performed (see below for details). A PHREEQC input is built by passing preformatted strings to the AccumulateLine buffer (based on G2 current process parameters value). After execution of the PHREEQC input with the RunAccumulated method, the results are retrieved with the GetSelectedOutputValue method, in a table format defined by the SELECTED_OUTPUT block of PHREEQC. All functions corresponding to the developed chemistry modules are gathered into a “phreeqc.c” file. This file is then linked into the foreign image, into which G2 can make foreign function calls, via TCP/IP protocol. During G2 start-up sequence, the foreign image is linked by using the “connect to foreign image” grammar. Fluxes Mixing Module. By default, a mixing script for two flows is executed at each node and each unit operation. G2 passes all current dissolved solids concentration, pH, and temperature of the two flows to PHREEQC. The mixture is calculated using the MIX keyword in PHREEQC (with the mixture ratio calculated from the current simulation flows). The amount of solid calcite contributed by the two flows is then equilibrated with the mixture, using the EQUILIBRIUM_PHASES keyword (with target saturation index for calcite set to zero). The resulting total dissolved species concentration and dissolved or precipitated calcite amount are passed back to G2 for updating all current dissolved species and CaCO3_solid concentrations. The calculated pH is also passed back to G2. The separation coefficients are applied afterward. The
Figure 1. Principles of the coupling procedure between the process simulation software and the chemistry engine.
atmosphere module (full or partial): log(P CO2(g)) is an adjustable parameter that allows one to adjust the aeration level; the amount of CO2 exchanged by stripping is then calculated. The CaCO3 amount possibly precipitated is calculated and updated in PS2000. (4) Anaerobic treatment module: COD is converted to acetic acid with an adjustable yield (simulating acetogenesis), and then acetic acid is converted to dissolved CO2 with an adjustable yield (simulating methanization). (5) Aerobic treatment module: residual COD is converted to dissolved CO2 with an adjustable yield. CO2 is then stripped in a subsequent aeration step with an adjustable target (P CO2(g)) that leads to CaCO3 precipitation (with adjustable target SI-calcite). A detailed description of the chemistry modules follows. These modules are represented by the PHREEQCI “droplet” logos in the flow sheet. The “fluxes mixing module” is implemented by default at each node and is not graphically represented in the process flow sheet. General Coupling Method. The coupling procedure between the process simulation and the chemical reaction engine makes use of a G2 “foreign image”. This external
Figure 2. Summary description of the chemistry modules embedded in the process simulation (schematic process, see Figures 3−5 for a detailed description of the process). 423
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
Figure 3. Process simulation screenshot (1/3: pulp preparation area).
Figure 4. Process simulation screenshot (2/3: papermachine area).
saturation index for calcite and log(P CO2(g)) are also calculated and passed to G2 but used only for inspection. “Bacterial Activity” Module. In order to simulate the release of volatile fatty acids (VFA) from anaerobic microbial
activity, a source of acetic acid was implemented (acetic acid is added to the solution, using the REACTION keyword, then the resulting solution is equilibrated with the incoming CaCO3_solid amount using the EQUILIBRIUM_PHASES key424
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
Figure 5. Process simulation screenshot (3/3: wastewater treatment area).
word). The added acetic acid flow rate (in kg/h) can be adjusted locally in the process (by clicking on the brown “droplet” modules labeled as “vfa-x” in Figures 3 and 4). Equilibrium with Atmosphere Module. Precipitation in such a process has been shown to occur because of CO2 stripping upon contact with air.15 Therefore “aeration” modules were developed (represented as blue “droplet” modules in Figures 3 and 4). In these modules, the solution is simultaneously equilibrated with incoming CaCO3_solid amount and the CO2(g) phase (using the EQUILIBRIUM_PHASES keyword; to ensure that atmospheric CO2 would not dissolve in the solution in case the target log(P CO2(g)) would be higher than that in the incoming flow, the initial reservoir of CO2(g) is set to zero; if necessary, precipitation can be prevented here by using the “dissolve” switch for the calcite equilibrium in PHREEQC). The target saturation index for calcite is set to zero. The target log(P CO2(g)) can be adjusted to the measured value (resulting from speciation of sampled process water). Remember that full saturation in CO2 corresponds to log10(P CO2(g)) = 0, while equilibrium with atmospheric CO2 corresponds to log10(P CO2(g)) = −3.4. The precipitated amount of calcite is then updated in G2, together with dissolved species concentrations. Anaerobic Treatment Module. Finally, chemistry modules were developed to simulate the scaling behavior of effluents from the wastewater treatment plant. The first module simulates the anaerobic water treatment (red “droplet” logo in Figure 5), through a sequence of acetogenesis and methanization steps.30 Acetogenesis is simulated by partial conversion of the COD concentration tracked by G2 to acetic acid (using the REACTION keyword). The flow is equilibrated with residual CaCO3_solid amount (using the EQUILIBRIUM_PHASES keyword), which is usually fully dissolved in these conditions. Then the methanization step is simulated by partial conversion of the total acetic acid to dissolved CO2, using the REACTION keyword, with a negative amount of acetic acid, (which is the way PHREEQC removes chemicals from a solution), and a positive amount of CO2(g) (note that the production of methane has no impact on calcium chemistry and was not simulated here). The acetogenesis yield, methanization yield, and VFA to CO2 molar conversion coefficient (describing the stoichiometry of the methanization path) can be adjusted by clicking on the module logo.
Aerobic Treatment Module. The second water treatment module simulates the aerobic water treatment step (orange “droplet” logo in Figure 5). Consumption of the residual COD by the activated sludge is simulated by partial conversion of the COD concentration tracked by G2 to dissolved CO2 (using the REACTION keyword). Then aeration is simulated by simultaneously equilibrating the solution with the CO2(g) phase and calcite amount (using the EQUILIBRIUM_PHASES keyword; both CO2 stripping target level and target level for saturation index with respect to calcite can be adjusted by the user). Other Chemistry Procedures. By default, only dissolution equilibrium with calcite is allowed throughout the process (so that supersaturated fluxes occur, as observed in some parts of the process). This is achieved by using the “dissolve” switch in the EQUILIBRIUM_PHASES block. Precipitation is allowed only in aeration modules, and in the aerobic water treatment module. Accurate pH calculation requires that the solutions are electro-neutral. Instead of resolving an additional mass balance for charge imbalance in G2 (which would be computationally inefficient), the ionic balance is corrected by adding a dummy cation or anion in the speciation. Inert species “Cation+” and “Anion−” were added to the thermodynamic database. In each speciation calculation, a preliminary speciation is actually run only to identify the sign of charge imbalance, and then the speciation is rerun with charge compensation by either the dummy cation or anion (using the “charge” switch in the SOLUTION block). The electroneutrality correction is performed in each speciation of each chemistry module. This correction is found to have only minor impact on the results, as the pH is imposed on all sources flows to the process. Time of Analysis. Thanks to the efficient numerical methods of IPhreeqc, the computation of a full mill model, coupling chemistry and mass flow balances, is relatively fast. In this model with about 100 unit operations, resolving a change of parameter value (such as fresh water pH) usually takes less than 1 min on a standard PC. Themodynamic Database. The PHREEQC.DAT database21 has been modified to include the major VFA species (i.e., acetate, propionate, butyrate, and lactate), with corresponding dissociation constants and complexation reactions with calcium, sodium, and magnesium (taken from the MINTEQ.V4.DAT database). All considered species, reactions, reaction constants, 425
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
microbial activity was suspected, i.e., pulp and process water storage tanks with significant retention time. We adjusted the corresponding dosages of acetic acid (Table 2) to try to match the observed pH in the pulp preparation and papermachine loops. It is possible to explain the measured pH profiles throughout the mill to a good approximation. Value at selected control points (represented as numbered yellow circles in the flow sheets) are shown in Figure 7.
and their temperature dependence laws are listed in Huber et al.15 All saturation indices (SI) are defined as SI = log10(S), where S is the supersaturation, so that an equilibrium situation corresponds to SI = 0, and under- and supersaturated to SI < 0 and SI > 0, respectively.
■
RESULTS AND DISCUSSION The developed coupled simulation allows one to accurately describe scaling related phenomena, as shown in the validation profiles (Figures 6−10). Dissolution and precipitation of
Figure 7. Comparison of measured and simulated pH throughout the process. Figure 6. Comparison of measured and simulated temperature throughout the process.
In order to check the simplification hypothesis that replacing a real organic acids mix (produced from anaerobic activity) with acetic acid only would lead to similar results in terms of scaling tendency, a batch dissolution simulation was performed with PHREEQC. Calcite was supposed to be dissolved in pure water with a total organic acid concentration of 30 meq/L, made up of either 100% acetic acid or the organic acid molar distribution given in ref 32 (i.e., 70% acetic acid, 12% propionic acid, 7% butyric acid, and 11% lactic acid). Calculations show that the dissolution equilibrium with calcite releases almost the same amount of calcium (760.4 mg/L for the organic acids mix vs 759.6 mg/L for equivalent acetic acid) and leads to the same pH = 6.165. The only sources of dissolved calcium and inorganic carbon are the fresh water and the dissolution of CaCO3 filler with incoming raw material. The only sinks for dissolved calcium and inorganic carbon are precipitation of CaCO3 and CO2 stripping zones (i.e., zones where pulp come into contact with air such as in the aerobic water treatment step). With these hypotheses, and adjustment of acetic acid sources only, the calculated concentrations profiles of dissolved calcium, inorganic carbon, and acetate are in good agreement with the measured values throughout the process (Figure 9). This demonstrates that organic acids released by anaerobic microbial activity in stagnant zones of the process are responsible for partial dissolution of the calcite filler. This phenomenon explains why calcium concentration is usually so high in closed
calcium carbonate are well simulated by solving liquid− solid−gas phase chemical equilibria for process flows, with a limited number of adjustable parameters. Previous speciation calculations on similar process waters31 showed a large deficit in anionic species. We proposed that this could be attributed to organic acids produced from microbial activity (it was shown previously that taking into account dissociated organic acid species as acetate largely improves the calculated ionic balance of the process water, see ref 15). Indeed, in a recycled board mill with low fresh water consumption, the closure of the circuits tends to concentrate dissolved species and nutrients in the process loops (such as dissolved starch released from the raw material). These are ideal conditions for anaerobic bacterial development, so that concentrations of volatile fatty acids (VFA) are usually quite high (typically from 1 to 5 g/L), with acetic acid accounting for more than 50% of total VFA.32,33 VFA are weak acids, responsible for the acidification of processing water and presumably cause partial dissolution of the calcite fraction of filler. We initially started with the process simulation equilibrated for suspended solids and dissolved species other than calcium, inorganic carbon, and acetate. Then we added several acetic acid sources to the process, at locations where anaerobic
Table 2. Dosages of Acetic Acid in Each Volatile Fatty Acids Source (VFA-x) in Reference Conditions
acetic acid (kg/h)
VFA-1
VFA-2
VFA-3
VFA-4
VFA-5
VFA-6
VFA-7
sewer collector tank
dump tank
disk filter filtrate tank
main pulp storage tower
white water tank
save-all tank
save-all filtrate tank
40
60
30
145
60
40
35
426
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
circuits board mills (in the range of 800 to 1000 mg/L here). The inorganic carbon concentration is much lower than what could be expected from a stoichiometric dissolution of calcium carbonate. This is explained by CO2 stripping in aerated areas of the process. The pH in the circuits finally results from both CaCO3 dissolution equilibrium with VFA and local exchanges with the atmospheric CO2. The equilibrium pH in the pulp preparation area is slightly acidic (between 6.4 and 6.6) but is largely buffered by CaCO3 dissolution. If the raw material did not contain CaCO3 filler at all, the simulation shows that the equilibrium pH would be around 3.8 throughout pulp preparation and papermachine area (with unchanged acetic acid sources). The proposed simulation of the water treatment steps allows one to reliably predict the scaling behavior of the effluents of the integrated wastewater treatment plant (see control points 15, 21, and 22 in Figures 8−10). Detailed discussion of the Figure 10. Comparison of measured and simulated log(P CO2(g)) throughout the process (“measured log(P CO2(g)”) results from chemical speciation of sampled processed water).
generates a tremendous amount of inorganic carbon (from methanization of produced VFAs), which causes a dramatic increase of the saturation index with respect to calcite and CO2(g). Over the subsequent aerobic treatment step, CO2 stripping causes extensive precipitation in the aeration basin (the activated sludge actually consists of more than 60% CaCO3, as measured in the biological sludge samples15). The simulation shows that 187.1 kg/h of CaCO3 are precipitated in the aerobic treatment step in reference conditions. Precipitation also occurs in several other process locations, with much lower precipitated amount. A proposed curative solution to dissolution and reprecipitation of CaCO3 in such a board mill could be to limit anaerobic activity in the process water by appropriate biocide treatment. As a first application of the developed coupled simulation, we estimate the effect of such a biocide treatment by reducing acetic acid sources. The dosage is reduced in the same proportion for each of the seven acetic acid sources. Interestingly, reduction of VFA production directly reduces dissolved calcium concentration (inorganic carbon concentration is not much affected because of CO2 stripping in
Figure 8. Comparison of measured and simulated saturation index with respect to calcite throughout the process (“measured SI” results from chemical speciation of sampled processed water).
simulation of water treatment steps is beyond the scope of this paper. Our intent here is to calculate the amount of CaCO3 precipitated on the activated sludge. Briefly, the anaerobic step
Figure 9. Comparison of measured and simulated values for total calcium concentration [Ca], total inorganic carbon [IC], and total volatile fatty acid [VFA] throughout the process (note that VFA is simulated as equivalent acetic acid). 427
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
dissolution of the calcium carbonate filler coming with the raw material. Also, the simulated CO2 stripping is shown to explain the observed precipitation problems. The simulation allows studying the impact of curative treatments. Reduction of organic acids release (which may be achieved by appropriate biocide treatment) is predicted to greatly reduce scaling problems and lower sludge discharge from the water treatment steps.
aeration zones). Thus, there is a large potential to limit this calcium source by appropriate biocide treatment. Reducing VFA production is also predicted to largely limit further reprecipitation in the aerobic water treatment. The simulation shows that precipitation in the aerobic water treatment could be totally inhibited by reducing VFA production by 80% in the pulp storage tanks. As already mentioned, the biological sludge typically consists of more than half of reprecipitated calcium carbonate so that appropriate management of anaerobic activity in the process water could greatly help reducing the amount of biological sludges to be discharged and handled by the industry. An other interesting perspective to reduce scaling problems could be to recycle the biotreated effluent to the process (at same fresh water consumption). This makes use of the wastewater treatment as a kidney to deconcentrate the process waters. Work is currently in progress to evaluate this alternative water management strategy. In the present state, the model should be able to predict the effect of several variations to the process; however, much more work is needed to have a fully predictive model. For instance, the model should predict the impact of raw material variation (in terms of calcite filler content). Also, the impact of variable fresh water quality (hardness) on scaling tendency or interest for a softening treatment could be evaluated. However, it is much more difficult to predict the evolution of bacterial activity due to process variations. Although the model can predict the consequences of enhanced or reduced bacterial activity (through adjustment of VFA sources), it cannot predict the impact that furnishes variations or reduced fresh water consumption may have on bacterial activity itself. This requires additional biochemical modeling, which is not implemented yet. The developed simulation method is very versatile, as it allows one to quickly reproduce mill circuit configurations, with dedicated models of most papermaking unit operations being available in our library. As already mentioned, the time required for resolving full mill models is relatively fast (of the order of 1 min in the present case study). Also, thanks to the open database of PHREEQC, new species and their associated chemical reactions can quickly be added to the thermodynamic database and then tracked in the process simulation. Here, the study was focused on calcium carbonate scaling, but the same method could easily be applied to simultaneously model other types of scales, such as calcium oxalate or barium sulfate, which is also causing problems in other pulping and papermaking processes. Also injection of other types of chemicals could be simulated as well (such as caustic soda to neutralize VFA from anaerobic activity). It could also be possible to take advantage of the PHREEQC capability to resolve kinetic rates to model the dissolution and growth rate of calcium carbonate, together with CO2 stripping kinetics upon aeration (see our ongoing work on this topic34). Also the coupling with PHREEQC will allow one to take into account ionic exchange phenomena with cellulosic fibers, which is important in situations with lower dissolved calcium concentration.
■
ASSOCIATED CONTENT
S Supporting Information *
Parameters used for the process simulation, including species source values (fresh water quality, raw material contaminants, chemicals injection), separation coefficients for each species on each unit operation, pH and temperature values for the source fluxes, specific heat values, and parameters of the various chemical modules. This material is available free of charge via the Internet at http://pubs.acs.org.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Phone: +33.(0) 4.76.15.40.51. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS This work was supported by Centre Technique du Papier (CTP) and CTPi members. Participants to the SCALE project are acknowledged. Thanks are due to E. Fourest and J. Ruiz (CTP) for useful discussions.
■
REFERENCES
(1) Bulow, C.; Pingen, G.; Hamm, U. Complete water system closure: special attention had to be paid to the calcium carbonate problem, Kappa Zulpich Paper. Pulp Pap. Int. 2003, 45, 14−17. (2) Nivelon, S.; Pichon, M.; Piollet, A. In 84th Annual Meeting Technical Section, Paptac, Montreal, Canada, 1998; Vol. A, pp 123− 126. (3) Badar, T. A.; Hodge, G.; Wiegand, P. In 1994 International Environmental Conference, Portland, OR, April 17−20, 1994; Vol. 1, pp 33−48. (4) Barton, D. A.; Lagace, P.; Stuart, P. R.; Miner, R. In 1996 Recycling Symposium, New Orleans, LA, March 3−6, 1996; pp 221− 228. (5) Habets, L. H. A.; Deschildre, A.; Knelissen, H. J.; Arrieta, J. In 2000 International Environmental Conference and Exhibit, Denver, CO, May 6−10, 2000; Vol. 2, pp 833−839. (6) Hamm, U.; Schabel, S. In 8th International Water Association Symposium on Forest Industry Wastewaters, Vitoria, Brazil, April 9−12, 2006. (7) Dahlquist, E. Process Simulation for Pulp and Paper Industries: Current Practice and Future Trend. Chem. Prod. Process Model. 2008, 3. (8) Huang, H. J.; Ramaswamy, S.; Al-Dajani, W. W.; Tschirner, U. Process modeling and analysis of pulp mill-based integrated biorefinery with hemicellulose pre-extraction for ethanol production: A comparative study. Bioresour. Technol. 2010, 101, 624−631. (9) Olsson, J.; Zacchi, G. Simulation of the condensate treatment process in a kraft pulp mill. Chem. Eng. Technol. 2001, 24, 195−203. (10) Ruohonen, P.; Ahtila, P. Analysis of a mechanical pulp and paper mill using advanced composite curves. Appl. Therm. Eng. 2010, 30, 649−657. (11) Laperriere, L.; Wasik, L. In 2001 Engineering/Finishing & Converting Conference Proceedings, 2002; Vol. 1.
■
CONCLUSIONS The coupling of a chemical reaction engine (IPhreeqc) with a mass flow balance process simulation (PS2000, G2-based) developed in this work allows one to accurately predict the scaling behavior of process waters in a closed circuit recycled board mill. Simulations prove that organic acids released by anaerobic microbial activity are responsible for partial 428
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
Industrial & Engineering Chemistry Research
Article
(12) Beneventi, D.; Zeno, E.; Nortier, P.; Carré, B.; Dorris, G. Optimization and management of flotation deinking banks by process simulation. Ind. Eng. Chem. Res. 2009, 48, 3964−3972. (13) Ruiz, J.; Ottenio, P.; Carré, B. In Symposium Simulation & Process Control, Munich, Germany, December 2003. (14) Ravnjak, D.; Ilic, G.; Moze, A. Designing water reuse in a paper mill by means of computer modelling. Chem. Biochem. Eng. Q. 2004, 18, 13−20. (15) Huber, P.; Nivelon, S.; Nortier, P. Scaling risk assessment in a closed circuit recycled board mill by speciation methods. Tappi J. 2012, 11, 53−61. (16) Koukkari, P.; Penttilä, K.; Hack, K.; Petersen, S. CHEMSHEET−An Efficient Worksheet Tool for Thermodynamic Process Simulation. Microstruct., Mech. Prop. Processes−Comput. Simul. Model. 2000, 323−330. (17) Ferguson, R. J. In Corrosion 2002, NACE International, Denver, CO, 2002. (18) Wang, P.; Anderko, A.; Springer, R. D.; Kosinski, J. J.; Lencka, M. M. Modeling chemical and phase equilibria in geochemical systems using a speciation-based model. J. Geochem. Explor. 2010, 106, 219− 225. (19) Anderko, A.; Wang, P.; Rafal, M. Electrolyte solutions: from thermodynamic and transport property models to the simulation of industrial processes. Fluid Phase Equilib. 2002, 194, 123−142. (20) Meeussen, J. C. L. ORCHESTRA: An object-oriented framework for implementing chemical equilibrium models. Environ. Sci. Technol. 2003, 37, 1175−1182. (21) Parkhurst, D. L.; Appelo, C. A. J. User’s guide to PHREEQC (version 2)- A computer program for speciation, batch-reaction, onedimensional transport, and inverse geochemical calculations 1999. (22) Gustaffson, J. Visual Minteq, ver. 2.50; Department of Land and Water Resources Engineering: Stockholm, Sweeden, 2006. (23) Gorensek, M. B.; Staser, J. A.; Stanford, T. G.; Weidner, J. W. A thermodynamic analysis of the SO2/H2SO4 system in SO2depolarized electrolysis. Int. J. Hydrogen Energy 2009, 34, 6089−6095. (24) Kalliola, A.; Kukkämäki, E.; Pajarre, R.; Koukkari, P.; Hakala, J. In Papermaking Research Symposium, Kuopio, Finland, June 1−4, 2009. (25) Charlton, S. R.; Parkhurst, D. L. Modules based on the geochemical model PHREEQC for use in scripting and programming languages. Comput. Geosci. 2011, 37, 1653−1663. (26) Wissmeier, L.; Barry, D. A. Simulation tool for variably saturated flow with comprehensive geochemical reactions in two- and threedimensional domains. Environ. Model. Software 2011, 26, 210−218. (27) Ruiz, J.; Ottenio, P. In Control Systems, Risto Ritala Tampere University of Technology: Tampere, Finland, 2006; pp 79−84. (28) Carré, B.; Ruiz, J.; Ottenio, P.; Brun, J. In 9th PTS-CTP Deinking Symposium, Munich, Germany, May 9−12, 2000. (29) Mersmann, A. Crystallization Technology Handbook, 2nd ed., revised and expanded; Marcel Dekker: New York, 2001. (30) Habets, L. H. A.; Hooimeijer, A.; Knelissen, H. J. In line biological process water treatment for zero discharge operation at recycled fibre board mills. Pulp Pap. Can. 1997, 98, 504−507. (31) Clément, S.; Gouiller, A.; Ottenio, P.; Nivelon, S.; Huber, P.; Nortier, P. Speciation and supersaturation model in papermaking streams. Process Saf. Environ Prot. 2011, 89, 67−73. (32) Jung, W. K.; Kutzner, H. J. Microbiologic problems associated with closed process water systems in the paper industry. Eur. J. Appl. Microbiol. Biotechnol. 1978, 5, 215−224. (33) Amat, A. M.; Arques, A.; Miranda, M. A.; López, F. Use of ozone and/or UV in the treatment of effluents from board paper industry. Chemosphere 2005, 60, 1111−1117. (34) Huber, P. Kinetics of CO2 stripping and its effect on the saturation state of CaCO3 upon aeration of a CaCO3-CO2-H2O system: application to scaling in the papermaking process. Ind. Eng. Chem. Res. 2011, 50, 13655−13661.
429
dx.doi.org/10.1021/ie300984y | Ind. Eng. Chem. Res. 2013, 52, 421−429
