VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE

IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE

CIMA SIMADES

MODELO PARA REDUCIR AGUAS RESIDUALES EN LA INDUSTRIA DE PROCESOS

-
A ROADMAP TO REDUCE WATER DISCHARGE FOR THE PROCESS INDUSTRY

Miguel Bagajewicz
University of Oklahoma, 100 E. Boyd St, T-335, Norman, OK 73019, USA
E-mail: [email protected]

ABSTRACT

This paper discusses concrete measures that the chemical processing industry that makes intensive use of
water can take towards the goal of eliminating polluted water. The roadmap towards zero liquid discharge
solutions is discussed. Several solution approaches are briefly outlined and examples are shown.

Keywords: wastewater minimization, wastewater reuse, pollution prevention

INTRODUCTION

Water is a key element for the chemical and petrochemical industry, in hydrometallurgy, in the iron and
steel industry, in the food and agricultural industries (sugar factories, dairy industries, breweries, in the
textile industry, in the pharmaceutical and electronic component industry and many others.

ROADMAP OF PROCESS SOLUTIONS

Until a few years ago, the problem of water treatment was considered as a set of sequential treatment
operations of a single wastewater stream coming from all unit operations. Such a system is depicted for three
water user processes (Pi) and three treatment units (Ti) in Fig. 1a. One way of obtaining improved designs is
the reuse of wastewater. This reduces the cost because the overall water intake is smaller (Fig. 1b). The next
step is to resort to distributed treatment, which consists of series/parallel arrangements of the wastewater
treatment units (Fig. 1c). Finally, treatment can be decentralized in such a way that some pollutants are
removed from wastewater before it is reused again. This is indicated in Fig. 1d by solid lines. Dashed lines
correspond to the concept of zero-liquid discharge.

As depicted in Fig. 1d, the concept of zero discharge applies alternatively to the total elimination of the
disposal of environmentally hazardous substances in liquid form, that is, zero "liquid" discharge. When using
these closed circuits, end-of-pipe regeneration does not have to be conducted to the full extent required for
disposal as water can be reused with higher level of contaminants. Additionally, the absence of a discharge
eliminates internal administrative costs associated with the enforcement of environmental protection agencies.

These zero discharge process solutions have never been attempted, neither in academic case studies, and
apparently nor in practice. They constitute the burning challenge for both academia and industry. There are
however, a few issues than can readily be pointed out. First, many units require steam, that is, pure fresh water
of boiler quality. Thus, in order for these zero discharge cycles to exist, wastewater cleanup should be
thorough, something that could be too expensive to be realistic. Second, unless an expensive total evaporation
step is included, some water make-up and consequently, some disposal should take place to avoid the
accumulation of certain species not being removed.

The analysis made in this section suggests that in reality the problem one wishes to solve is the one
depicted in Fig. 1d (including the dashed lines), which has all the previous alternatives embedded.
Diepolder (1992) and Goldblatt et al. (1993) discussed how realistic is this concept from the practical point
of view, addressing issues such as the disposal of solids, the possible sizable revenues obtained for the
selling of low grade salts and avoiding their disposal (close to a million dollar per year for a typical
refinery), or the costs of disposal of groundwater solids, etc.

553
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE
CIMA SIMADES

Pollution indices can be calculated through simulation (Hilaly and Sikdar, 1996). Several ways have been
proposed to reduce the generation of pollutants, which is in reality the root of the problem (El-Halwagi, 1997).
However, for many processes the reduction of the generation of pollutants is not possible. For example, the
major pollutants in refinery wastewater are part of the crude and are not generated in the plant.

P1 P1
Treatment Plant Treatment Plant
Disposal Disposal
P2 T1 T2 T3 P2 T1 T2 T3
Fresh Fresh
Water Water
P3 P3
Wastewater Wastewater

(a) (b)

T1
P1 Treatment Plant
P1 T1 Solid discharge
T1
P2 T2
Fresh P2 T2
Water Fresh
P3 T3 Water
P3 T3
Disposal Disposal
(c) (d)

Figure 1
Figure 1: Water Utilization Systems in Process Plants

OPTIMAL WASTEWATER REUSE

The search for optimal wastewater reuse solutions was addressed by industry itself more than twenty years
ago (Carnes et al., 1973; Skylov and Stenzel, 1974; Hospondarec and Thompson, 1974; Mishra et al.,
1975; Sane and Atkins, 1977; Anderson, 1977). Thee process systems community started to offer
systematic procedures as early as twenty years ago (Takama et al., 1980). At least one book is devoted
entirely to the problem (Mann and Liu, 1999) and another discusses it in some detail (Rossiter, 1995). In
addition, there are a few commercial software companies offering products related to water management.
(AspenWater, WaterPinch, Enviropro).

After the seminal paper by the group of Prof. Umeda in Japan (Takama et al., 1980), several methods were
proposed to address the water/wastewater allocation problem. Using several simplifying assumptions, it is
possible to anticipate, graphically and algorithmically, the minimum freshwater usage for the case of only
one contaminant without designing the system (El-Halwagi and Manousiouthakis, 1989; Wang and Smith,
1994a; Dhole et al., 1996).

To actually design the system several methods are available: Wang and Smith (1994a, 1995) proposed
graphical procedures very useful for small problems. Later Savelski and Bagajewicz (1999a,b, 2000a,b),
as well as Gomez et al. (2000) proposed an algorithmic procedure that allows the design of this system by
hand. Moreover, it has been shown that the problem has several alternative solutions featuring the same
freshwater intake. To help in sorting out through these alternatives, a mixed integer linear programming

554
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE
CIMA SIMADES

model was proposed by Savelski and Bagajewicz (2000c). This model minimizes the number of
interconnections or simply the total connection cost (piping cost) and is capable of handling forbidden
connections.

Multicontaminant Systems

Wang and Smith (1994a) also attempted to address a targeting and network design procedure for the case
of multicontaminant systems. The method proved to have limitations as it fails to identify optimal
solutions. Liu (1999) presented a few interesting heuristic rules. Although some of them are incorrect, the
solution procedure has remarkable simplicity and provides good sub-optimal (and sometimes optimal)
solutions. Later, Doyle and Smith (1997) and more recently Alva-Argáez et al. (1998a,b, 1999), Benko et
al. (1999), as well as Huang et al. (1999) presented MINLP or NLP models. All these models provide
solutions that cannot guarantee global optimality. Finally, Alva-Argáez et al. (1999) discuss some
transshipment models based on several assumptions, some of them restrictive.

A method capable of obtaining globally optimal was presented recently (Savelski and Bagajewicz, 2000d;
Bagajewicz et al., 2000b, 2000c). The importance of this method stems from its ability of being able to
provide as many sub-optimal solutions as desired that may be attractive to the practical engineer. The
method is also useful for retrofit. In recent papers, Savelski et al.. (1999, 2000) show that grassroots
designs as well as retrofit designs can be performed applying this methodology using complete and
realistic capital and operating costs. Finally, the importance of simultaneous minimization of utility and
fresh water usage was addressed by Savelski et al. (1997) and more recently by Savulescu and Smith (1998).
Bagajewicz et al. (1999a, 2000a) proposed to solve the targeting problem by coupling the linear programming
targeting problem with a regular transshipment model for heat integration.

OPTIMAL WASTEWATER TREATMENT

Wang and Smith (1994b) approached the design of distributed effluent treatment using a graphical
procedure. The authors used the same cost functions for the effluent treatment units as those proposed by
Takama et al. (1980). A revised version of this procedure was presented by Kuo and Smith (1997, 1998).
Recently, Freitas et al. (1999) used a hierarchical design approach. They constructed a relational database
and an expert system to determine the best sequence of treatment processes. The method cannot of course,
guarantee any optimality.

Alva-Argáez et al. (1998a,b) modeled the entire water management problem by means of a superstructure,
which leads to a MINLP problem. Galán and Grossmann (1998) also solved the effluent-treatment
distribution problem using mathematical programming. Finally Huang et al., (1999) propose a similar NLP
approach. The solvability of these MINLP problems is at the heart of the challenge. These methods do not
guarantee global optimality and some present serious numerical challenges. In addition, with the exception
of the paper by Galán and Grossmann (1998), these methods are capable of addressing the simultaneous
allocation of wastewater and treatment. That is solving the problem presented in figure 1d without of
course the recycle. Rivas and Bagajewicz (2000) extended their tree search method to address the
decentralized treatment problem. (figure 1d, without dashed line recycle).

The application of all these techniques not only reduces water discharge, but it also reduces capital and
operating costs.

ECONOMICAL IMPACT

Industry will generate 120,000,000 tons of wastewater per year (estimate made by the water pollution
control federation in 1989). Monsanto (Wales, UK) reduced water usage by 30% and saved $11.5 million
capital in the wastewater treatment units and reduced operating expenditures by $1 million a year. Similar
levels of reduction of water use and discharge have been reported for the pulp and paper industry (savings
of $0.8 million with 2 yr payoff). Petrochemical complexes can also see substantial reduction of operating

555
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE
CIMA SIMADES

costs (one example shows $0.3 million/year reduction with only $50,000 investment). (Mann and Liu,
1999).

EXAMPLE

Table 1 shows results for a case study that was presented by Savelski et al. (2000), where complete data of
the problem can be obtained. The example contains 8 refinery processes: caustic treating, Distillation,
Two Sweetening , Hydrotreating, and 2 desalters. Four contaminants are considered: Salts, Organics, H2S,
Ammonia. Finally costs include piping and pumps as well as all operating costs. The table shows that
operating costs are dominant and that retrofit solutions tend to minimize the freshwater to the same level of
a grassroots design.

The retrofit problem consists of the same constraints and a slightly modified objective function. In this
objective function, it is considered to a) putting new piping where there is none, and b) adding a pipe
and/or a pump where there is one. The piping costs are reduced tenfold with comparison to the grassroots
solution. The optimal solution whose cost is $2,048,600 differs from the first sub-optimal solution in only
$1,100 (0.05%), which can be considered negligible. This difference comes from piping costs.
The difference between centralized and decentralized treatments is not negligible and is illustrated in table
2 and figure 2.

CONCLUSIONS

Several methods are available to perform grassroots design and retrofit designs of water utilization systems in
process plants, leading to smaller water discharge. These methods are economically realistic and provide
several sub-optimal solutions for the practicing engineer. In particular, retrofit designs show that it is always
worthwhile to invest in piping and pumps to achieve maximum efficiency. Contrary to existing common
knowledge, these water management projects are profitable.

ACKNOWLEDGMENT: Funding from EPA grant 825328-01-0 is greatly appreciated.

REFERENCES

1. Alva-Argáez A., A. C. Kokossis and R. Smith. Automated Design of Industrial Water Networks.
AIChE Annual Meeting, paper 13f, Miami, (1998a).
2. Alva-Argáez A., A. C. Kokossis and R. Smith. Wastewater Minimisation of Industrial Systems using
and Integrated Approach. Comp. & Chem. Eng., Vol. 22, Suppl., pp. S741-S744, (1998b).
3. Alva-Argáez A., A. Vallianatos, A. C. Kokossis. A Multi-Contaminant Transshipment Model for mass
Exchange Networks and Wastewater Minimisation Problems. Comp. & Chem. Eng., Vol. 23, pp.
1439-1453, (1999).
4. Anderson D. Practical Aspects of Industrial Reuse. Int. Chem. Engng Symp. Series, 52, (1977).
5. Bagajewicz M., M. Savelski and H. Rodera. On The Effect of Heat Integration in the Design of Water
Utilization Systems in Refineries and Process Plants. AIChE Annual Meeting. Dallas, (1999a).
6. Bagajewicz M., H. Rodera and M. Savelski. Energy Efficient Water Utilization Systems In Refineries
and Process Plants. Submitted to Computers and Chem. Eng. (2000a).
7. Bagajewicz, M., M. Rivas and M. Savelski. A New Approach to the Design of Water Utilization
Systems with Multiple Contaminants in Process Plant. 1999 Annual AIChE Meeting. Submitted to
Computers and Chemical Engineering (2000).
8. Benko N., E. Rev, Z. Szitkai and Z. Fonyo. Optimal Water Use and Treatment Allocation. Comp. &
Chem. Eng., Vol. 23, Supp., pp. S589-S592, (1999).
9. Carnes, B. A., D.L. Ford and S. O. Brady. Treatment of Refinery Wastewaters for Reuse. Nat. Conf.
Complete Water Reuse, Washington D.C., (1973).
10. Diepolder P., Is 'Zero Discharge’ Realistic? Hydrocarbon Processing, October (1992).
11. Dhole V. R., N. Ramchandani, R. A. Tainsh and M. Wasilewski. Make your Process Water Pay for
itself. Chemical Engineering, January, pp. 100, (1996).

556
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE
CIMA SIMADES

12. Doyle, S. J. and R. Smith. Targeting Water Reuse with Multiple Contaminants. Trans IChemE, 75,
Part B, No. 3, pp. 181-189, (1997).
13. El-Halwagi M. M. Pollution Prevention through Process Integration, Academic Press , (1997).
14. El-Halwagi M. M. and V. Manousiouthakis Mass Exchanger Networks. AIChE Journal, 35, 8, pp. 1233,
(1989).
15. El-Halwagi M. M. and V. Manousiouthakis Automatic Synthesis of Mass Exchanger Networks with
Single Component Targets. CES, 45, 9, pp. 2813, (1990).
16. Freitas I. S. F., R. A. R. Boaventura, C. A. V. Costa. Conceptual Design of Industrial Wastewater
Treatment Processes: A hierarchical Approach Procedure. Proceedings of the 2nd Conference on
Process Integration, Modeling and Optimization for Energy Savings and Pollution Reduction.
(PRESS'99). Budapest, Hungary, May, (1999).
17. Galán B. and I.E. Grossmann. Optimal Design of Distributed Wastewater Treatment Networks. Ind.
Eng. Chem. Res., 37, pp. 4036, (1998).
18. Goldblatt M.E., K. S. Eble and J. E. Feathers. Zero Discharge: What, Why, and How. CEP, April. p.
22, (1993).
19. Gómez J., M. Savelski and M. Bagajewicz. On a Systematic Design Procedure for Water Utilization
Systems In Refineries and Process Plants. Submitted to Chem. Eng. Comm. (2000).
20. Hilaly A. K. and S. K. Sikdar. Process Simulation Tools for Pollution Prevention. Chemical
Engineering, January, pp. 98, (1996).
21. Hospondarec R. W. and S. J. Thompson. Oil-steam System for Water Reuse. Proceeding of the
Workshop (AIChE), Vol. 7, (1974).
22. Huang C-H., C-T. Chang, H-C. Ling and C-C. Chang. A Mathematical Programming Model for Water
Usage and Treatment Network Design. Ind. Eng. Chem. Res., 38, pp. 2666, (1999).
23. Kuo W. C. J. and R. Smith, Effluent Treatment System Design. CES, 52, 23, pp. 4273, (1997).
24. Kuo W-C. J. and R. Smith, Designing for the Interactions between Water-Use and Effluent Treatment.
Trans IChemE, 76, Part A, pp.287-301, (1998).
25. Liu Z. Wastewater Minimization using a Heuristic Procedure. AIChE Spring Meeting, Houston, Session
43, (1999).
26. Mann J. G. and Y. A. Liu. Industrial Water Reuse and Wastewater Minimization. Mc Graw Hill (1999).
27. Mishra P.N., L. T. Fan and L. E. Erickson. Application of Mathematical Optimization Techniques in
Computer Aided Design of Wastewater Treatment Systems. Water-1974 (II). AIChE Symp. Series, 71,
145, (1975).
28. Rivas M. and M. Bagajewicz. A Robust Method To Allocate Distributed Treatment Unit In Water
Utilization Systems In Process Plants. AIChE Annual Meeting, Los Angeles, (2000)
29. Rossiter A. P. Waste Minimization Through Process Design. Mc Graw Hill (1995).
30. Sane M., U. S. Atkins. Industrial Water Management. Int. Chem. Eng. Symp. Series, 52, (1977).
31. Savelski M., S. Lingareddy and M. Bagajewicz. Design and Retrofit of Water Utilization Systems and
Zero Discharge Cycles in Refineries and Process Plants. Paper 188g. AIChE Annual Meeting, Los
Angeles, (1997).
32. Savelski M. and M. Bagajewicz. A New Algorithmic Design Procedure for the Design of Water
Utilization Systems in Refineries and Process Plants. Proceedings of PRESS 99 Meeting. Budapest,
June (1999a).
33. Savelski M. and M. Bagajewicz. Watersave. A New Approach To the Design of Water Utilization
Systems in Refineries and Process Plant. Proceedings of the 2nd International Conference on Refining
Processes, AIChE Meeting, Houston, (1999b).
34. Savelski M. and M. Bagajewicz. On the Optimality Conditions of Water Utilization Systems in
Process Plants with Single Contaminants. To appear. Chemical Engineering Science. (2000a).
35. Savelski M. and M. Bagajewicz. Algorithmic Procedure to Design Water Utilization Systems in
Refineries and Process Plants. Submitted to Chemical Engineering Science. (2000b).
36. Savelski M. and M. Bagajewicz. On The Use of Linear Models for the Design of Water Utilization
Systems in Refineries and Process Plants. Annual AIChE Meeting, Dallas, Submitted to Chemical
Engineering Research and Design. (2000c).
37. Savelski M. and M. Bagajewicz. On the Necessary Conditions of Optimality of Water Utilization
Systems in Process Plants with Multiple Contaminants. Submitted Chemical Engineering Science.
(2000d).

557
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE
CIMA SIMADES

38. Savelski M., M. Rivas and M. Bagajewicz. A New Approach to the Design of Water Utilization
Systems with Multiple Contaminants in Process Plant. Annual AIChE Meeting, Dallas. (1999).
39. Bagajewicz M. M. Rivas and M. Savelski A Robust Method To Obtain Optimal And Sub-Optimal
Design And Retrofit Solutions Of Water Utilization Systems With Multiple Contaminants In Process
Plants . Proceedings of PSE 2000, Keystone, Colorado., (2000).
40. Savulescu L. E. and R. Smith. Simultaneous Energy and Water Minimisation. AIChE Annual
Meeting, paper 41f, Miami, (1998).
41. Skylov V. and R.A. Stenzel. Reuse of Wastewaters-Possibilities and Problems. Proceeding of the
Workshop (AIChE), Vol. 7, 1974.
42. Takama N., T. Kuriyama, K. Shiroko and T. Umeda. Optimal Water Allocation in a Petroleum Refinery.
Comp. & Chem. Eng. 4, 251-258, (1980).
43. Wang Y. P. and R, Smith. Wastewater Minimization. Chemical Engineering Science, 49, 7, pp. 981,
(1994a).
44. Wang Y. P. and R, Smith. Design of Distributed Effluent Treatment Systems. Chemical Engineering
Science, 49, 18, pp. 3127, (1994b).

558
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
CIMA SIMADES IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE

Table 1. Economic Comparison. Retrofit Design

Original Without Optimal Sub- Retrofit Sub-
Design Water Grassroots optimal Solution optimal
reuse Solution Grassroots Retrofit
Solution Solution
Freshwater Flow (ton/hr) 178.99 191.17 162.59 162.59 162.59 162.59
Freshwater & Treatment $ 2,255,300 $ 2,408,700 $ 2,048,600 $ 2,048,600 $ 2,048,600 $ 2,048,600
Piping and Pumping Cost $ 33,300 $ 243,100 $ 337,600 $ 342,500 $ 59,500 $ 60,600
Total Cost $ 2,288,600 $ 2,651,800 $ 2,385,900 $ 2,391,100 $ 2,108,100 $ 2,109,200

Table 2. Economic Comparison - Centralized Treatment vs. Decentralized Treatment Scheme

Centralized Treatment Decentralized Optimal Decentralized Sub-Opt.
Design Solution Solution # 1
Freshwater Flow (ton/hr) 119.33 41.94 42.45
Freshwater & Treatment Cost $ 1,503,600 $ 528,400 $ 534,800
Piping and Pumping Cost $ 230,700 $ 247,300 $ 252,700
Decentralized Treatment Cost - $ 68,000 $ 66,400
Total Cost $ 1,734,300 $ 843,700 $ 853,900

Freshwater Wastewater
Header Header
Fw1= 2.40
Caustic
(D = 1”) Treating
F1,7= 2.4
(D = 1”)
F2,out = 13.59
Fw2= 25.00
Distillation
(D = 3”)

Fw3= 8.57
Amines
(D = 2”) Sweetening
F3,4= 0.59
(D = 1”)

Fw4= 9.75 F4,out = 10.34

Sweetening
(D = 2”) (Merox I)
F3,5= 0.74
(D = 1”)
Fw5= 12.19 Sweetening F5,out = 12.93

(D = 2”) (Merox II)

Fw6= 25.0
(D = 3”) Hydrotreating

F3,7 = 6.05 F6,7 = 25.00

(D = 2”) (D = 3”)
Fw7= 50.76 F7,out = 84.21
Desalter I
(D = 4”)

F2,8 = 11.41 F 3,8 = 1.20

(D = 2”) (D = 1”)
Fw8= 28.91 F8,out = 41.52
Desalter II
(D = 3”)

Figure 2. Water reuse optimal solution (Centralized treatment)

559
 VII CONGRESO INTERAMERICANO SOBRE EL MEDIO AMBIENTE
CIMA SIMADES IV SEMINARIO INTERNACIONAL DEL MEDIO AMBIENTE Y DESARROLLO SOSTENIBLE

Freshwater
Header Reusewater
Header

FT,1= 2.67 Caustic F1,out = 2.67

D =1” Treating
Fw2= 25.00 F2,T = 23.01 Primary
Distillation
D = 3” D = 3” Treatment Unit
F6,T = 57.94
Fw3= 8.57 Sweetening D = 4”
D = 2” (Amines) FT,4 = 5.51
FT,5 = 23.01
D = 3” D = 2”
Fw4= 4.53 Sweetening F4,out = 10.04
D = 2” (Merox)
F3,5 = 1.28

Fw5= 3.84 D = 1” FT,6 = 49.76

Hydrotreating D = 4”
D = 1”
F2,6 = 1.99 F3,6 = 7.29
F5,6 = 28.13 F6,out = 29.23
D = 1” D = 2”
D = 3” Desalter

Figure 3: Decentralized treatment Optimal Solution for a refinery problem

560