Reverse Osmosis Desalination Plant in Durban, South Africa

Authors: Lauren Burke, Cindy Chen, Osman Jamil, Natalia Majewska

Instructors: Fengqi You, David Wegerer

March 11, 2016

==Executive Summary==
An imminent global shortage of water will greatly affect the lives of millions of people. The situation in South Africa is predicted to exacerbate in the upcoming decades. This growing need for fresh drinking water motivated the design of a reverse osmosis desalination plant in the suburbs of Durban on the eastern coast of South Africa. A reverse osmosis process was selected based on the water salinity in the area and recent technological advancements in membrane efficiency.

The process is designed to operate for 350 days a year, 24 hours a day to yield 132,000 m3 of water per day. Prior to reverse osmosis, the seawater is pretreated to prevent membrane fouling using a cascade of filters and high-pressure pumps. Reverse osmosis is carried out at 37 bar with two banks containing 900 and 4000 membranes, respectively. After reverse osmosis, the pressurized concentrate that contains the unwanted salts and ions is sent to an energy recovery device to pressurize the reverse osmosis feed. The deionized water from the reverse osmosis system is remineralized and disinfected before being sent to the consumer.

Based on a thorough economic analysis, it has been determined that the process is not feasible at this stage. The annual estimated revenue is $37.5 MM. In addition, the 20 year NPV of the plant is -$23.5 MM with an IRR of 2.9%. A sensitivity analysis indicated that an increase in the price of water could potentially increase profits. Therefore, the design may become profitable implemented with the aid of government subsidies or if a drastic increase in water scarcity increases the selling price of water.

__TOC__

=Introduction=

==Background==
Close to 1.2 billion people, approximately one fifth of the world’s population, reside in areas of physical water scarcity [1]. Although there is enough water in the world for a population of seven billion people, it is unevenly distributed, and many regions have been experiencing water shortages for more than a decade. Given the current rate of population growth and water consumption, many studies predict that this problem will be exacerbated in the years to come. Water desalination is a process used to purify water so that it meets drinking standards for consumption. The two main technologies used for desalination are thermal desalination and reverse osmosis. While thermal desalination is a robust method that scales well, it is typically preferable for inlets of very high salt concentrations and very low energy costs, such as the gulf coast region. Reverse osmosis plants are more prevalent elsewhere, making up 80% of all desalination plants in the world [2]. Given the salinity of the inlet flow for this process and advances in technology, reverse osmosis has been selected as the optimal choice for this project.

==Location Selection==
The Durban area in South Africa has been selected as the location of this project since it is currently experiencing approximately 25% water withdrawal as a percentage of total available water. This value is projected to increase in the upcoming years to 40% water withdrawal due to physical water scarcity [3,4] as can been seen in Figure 1.

[[File:MapABC.jpg | thumb | center | 400 px | alt=Alt text| Fig 1: Comparison of water scarcity between 1995 and 2025.]]

In addition to the increasing water scarcity in the next decade, this location was selected due to relatively low location costs and salinity levels as seen in Appendix A. South Africa has a location cost factor of 1.1, which is only 10 % more than the Gulf Coast basis. The country also has relatively low cost of labor and has a labor productivity factor of 1.3 [5]. To provide energy for this process, hydroelectric power from a plant located approximately 200 km away from the proposed location will be used, allowing for a more sustainable energy source at a stable price point. Five desalination plants already exist in the country; all located on the western coast, in suburban areas as shown in Appendix B. A plant located on the eastern side, specifically in Amanzimtoti, near the large metropolis of Durban, would supplement these existing facilities by increasing the amount of available drinking water and provide an economic boost to an area that has one of the largest percentages of agricultural households in the country.

==Preliminary Market Analysis and Process Selection==
There are currently five large scale water desalination plants in South Africa, run by Veolia Water Technologies, as shown in Appendix B [6]. Similar to 80% of desalination plants in the world, South Africa largely utilizes reverse osmosis technology. The 20% of facilities that utilize thermal desalination account for 50% of all water desalinated, but these plants are largely concentrated in the Gulf States due to the high salt concentration of the sea water present and the abundance of cheap fuel. Reverse osmosis technology continues to improve, increase in efficiency, and recently gained market popularity as a desalination technology [2]. Furthermore, the salt concentration in South Africa is average, so multi-stage flash distillation is not required to reach the required drinking water purity. Therefore, reverse osmosis has been selected for this project because at lower inlet salt concentrations, it is more economically viable.

==Process Feed and Composition==
The salinity of the ocean water around the coast of South Africa is approximately 35 ppt, at 35 g of salt per every 1000 g of water [7]. The seawater inlet of the proposed desalination plant will have the chemical composition shown in Appendix C [8]. In addition to inorganic species in seawater, the inlet will also include heterotrophic bacteria at a concentration of 2.01*109 cells/L [9] and viruses at a concentration of 50*109 viruses/L [10]. Typical limits of total dissolved solids (TDS) are at 1 g/L for drinking water [7]. The products of this desalination process will be lower than this limit, at approximately 0.75 g/L. Individual mineral components must be lower than the upper limits shown in Appendix C to prevent negative health effects [11]. Based on the population of the greater Durban area, the total water intake of the city is estimated to be 2.9 million m3 of water per day [8]. To satisfy the drinking needs of 3.5% of this population, which is the typical output of a desalination plant, the goal of this process is to produce at least 100,000 m3 of freshwater per day.

=Process Alternatives=

==Pretreatment==

Pretreating seawater is important to prevent particle, colloidal, organic, mineral, oxidant, and biological fouling. Particle fouling is caused by sand, clay, and suspended solids in seawater. This can be avoided by using filtration. Colloidal and organic fouling can be avoided by using a combination of coagulation and filtration, although the preferred method is ultrafiltration. To remove any final particulates, nanofiltration (5-micron) is required before the feed water enters the RO membranes [12]. A summary of these fouling types and methods to prevent them can be seen in Appendix D.

Biological fouling is caused by bacteria, viruses, microorganisms, and protozoa, which contribute to biofilm formation on the membrane surface. Chlorination during pretreatment kills these microorganisms to prevent biofouling on the membranes. Existing plants typically use sodium hypochlorite injections of 3 mg/L to achieve this. Other options for chlorination include the use of chlorine gas or chlorine dioxide, but sodium hypochlorite is preferred at this stage because it has the lowest operating cost. After killing bacteria, it is necessary to dechlorinate the water because chlorine oxidizes the membranes in the RO process. One possible method for dechlorination is the injection of sodium bisulfate, which results in harmful byproducts, including hydrochloric acid. To prevent this, dechlorination can be achieved using an activated carbon filter, which has been selected for this process. These filters have a very high probability of adsorbing chlorine, making them the optimal choice for dechlorination [12].

==Membrane Selection==
Water treatment processes use several types of membranes, including microfiltration, ultrafiltration, nanofiltration, and reverse osmosis. Microfiltration membranes have the largest pores, followed by ultrafiltration, nanofiltration and finally reverse osmosis [13]. RO membranes filter out monovalent and multivalent ions, viruses, bacteria, and suspended solids [14]. This makes RO membranes the ideal choice for the process, especially if the water is to be consumed.

RO membranes are either made of cellulose acetate or polysulfone coated with aromatic polyamides [13]. Cellulose-based membranes have NaCl rejection values of up to 99.5% at 2000 psig and are often made from acetylated cellulose. Thin Film Composite (TFC) membranes are based on aromatic polyamides, and their characteristics surpass those of acetylated cellulose membranes. They have an average rejection of 99.5% at only 225 psig of feed pressure, making the process more economical and energy efficient. The membranes are made from a highly permeable support of polysulfone and are coated with a cross-linked aromatic polyamide thin film, which allows for the high salt rejection percentage [13].

There are four types of modules for membrane arrangements: plate-and-frame, tubular, spiral wound, and hollow fiber. An illustration of these types can be seen in Appendix E. The plate-and-frame module is the most simple, containing two end plates, a flat membrane, and spacers. The tubular module is an annulus with the membrane lining the walls and solution flowing through the tube. The spiral wound option has a flat membrane in the form of a sheet that is wrapped around a perforated collection tube. Finally, hollow fiber modules are bundles of fibers in a pressure vessel [13].

TFC membranes in the spiral wound arrangement have been chosen for this process. TFC membranes reject small organics that could pass through cellulose-based membranes, withstand a larger pH range and can be used at higher temperatures. The spiral wound arrangement was chosen because it has a large surface area and is easier to clean. Hollow fiber membranes provide a higher surface area, but they also incur high maintenance and replacement costs.

==Energy Recovery==
There are two main methods for energy recovery: direct recovery of mechanical energy, and conversion to electricity. Direct recovery of the mechanical energy from the high pressure concentrate is the most common method of energy recovery [9]. The devices used to accomplish this can have up to a 98% efficiency. With this type of energy recovery, the high pressure from the concentrate can only be used to pressurize another stream of water. Alternatively, conversion of energy of the high pressure concentrate into electrical energy is much more versatile, but much less efficient. While energy capture from water can be as high as 95% as in hydroelectric power plants, the devices using that energy will likely have a much lower efficiency [10]. For example, a typical pump has an efficiency from 40% to 60%, resulting in much higher overall energy loss. Therefore, for the proposed design, since versatility is not as big of a concern as energy conservation, a direct recovery system will be used for the mechanical energy from the high pressure concentrate to pressurize the filtered water for reverse osmosis.

==Remineralization==
Reverse osmosis membranes cannot selectively remove certain species. Since typical drinking water contains some salts, it is necessary to reintroduce them before water distribution and use. There are four major methods used for water remineralization after reverse osmosis. These methods utilize a mixture of three smaller processes: mixing with saltwater, mixing with salts or gases, and percolation with salt-based filters. Typically, percolation yields the best water quality, but also requires the largest investment and has the highest operating costs [15]. Method overviews are given in Appendix F. All of the methods commonly used bring the water to an acceptable range for drinking water in South Africa. For the proposed design, the process of mixing the reverse osmosis product with brine will be used. While this slightly produces lower water quality than its alternatives, the process still meets all federal guidelines for drinking water and chlorine concentration will be kept below 100 mg/L to prevent corrosion.

==Disinfection==
Although the product stream should not contain any bacterial or viral materials due to the effectiveness of the filtration and RO steps, disinfection is typically done as a precaution. This ensures that water quality is maintained despite any upstream issues with membrane integrity. Chlorine gas, sodium hypochlorite, ozone, and UV light are options for disinfection [16]. UV treatment and ozone are environmentally friendly and have no chemical byproducts, but the costs tend to be much higher. Chlorine gas and sodium hypochlorite have low capital investment and operating costs, but the byproducts can be difficult to manage and are not environmentally safe. Chlorine dioxide treatment is highly effective, requires very low capital investment, and produces byproducts that are easier to manage. While the operating costs are slightly higher than that of chlorine gas and sodium hypochlorite treatment, chlorine dioxide treatment has the best combination of environmental friendliness, overall investment, and effectivity [16].

==Waste Treatment==
An essential factor of the proposed reverse osmosis plant is waste treatment. Outlet streams of desalination plants include hyper-saline solutions at densities higher than circulating ocean water. This difference in density will cause the brine to settle to the ocean floor, potentially resulting in negative effects on the fauna and flora [17]. Conventional brine disposal methods include surface water and sewer discharge, deep well injection, and evaporation ponds [18].

Surface water discharge directly disposes hyper-saline solutions to surface water. This method has low capital, operating, and manufacturing costs. Although South Africa has less stringent policies on waste eliminations into surface water, ethical considerations should still be taken into account to eliminate negative environmental effects. Sewer discharge directly disposes brine outlet solutions into the sanitary sewer system. This method is cost effective when there are sewer systems nearby the plant. However, this is not the case for our location [18]. Deep well injection disposes brine into porous subsurface rock formations. While this method may be more environmentally friendly, it requires extremely high capital costs. Site evaluation would require time and money, and an extensive scale up analysis would be required to ensure that this method would be feasible. Evaporation ponds utilize solar energy to reduce the water content in brine solutions. This method incurs low costs and is reliable, as few pieces of equipment are required. However, depending on the amount of brine removed from the plant, large amounts of land may be needed to contain the effluents.

The most common technique for brine disposal used by water desalination plants is to discharge back into the ocean [19]. This is legal in South Africa as evidenced by the largest desalination plant in the country at Mossel Bay [20]. According to regulations, a permit must be acquired to discharge large amounts of waste into the ocean [21], but the precedent from Mossel Bay demonstrates that South Africa will grant such a permit to a water desalination plant. In addition, research of nearby protected marine environments was conducted, and it was concluded that the proposed plant location near Amanzimtoti is at least 25 km from the nearest Marine Protected Area [22]. Therefore, direct brine disposal is the chosen alternative for the process.

=Process Overview=
Raw seawater is taken from the ocean and pumped to the pretreatment facility. Sodium hypochlorite is added to remove particulates and biological organisms that would cause fouling in the RO membranes. Pretreatment involves a series of filters to remove the majority of the particulates [13] and an activated carbon filter to dechlorinate the pretreated water before it is pumped to high pressure. Reverse osmosis occurs in two banks of spiral wound membrane modules. The high pressure concentrate exits the reverse osmosis system and is sent to an energy recovery device. This converts mechanical energy from the pressurized concentrate into mechanical energy that pumps the low pressure filtered seawater. The low pressure concentrate is then returned to the ocean. The purified product from reverse osmosis is sent to posttreatment where it is remineralized by mixing with a slight percentage of pressurized concentrate in order to restore the necessary minerals and to meet drinking water standards. It is then disinfected with chlorine dioxide and sent to the consumer. See Figure 2 for a block flow diagram of this process.
[[File: Figure2RHP.png | thumb | center | 400px | alt=Alt text | Fig 2: Block flow diagram of the proposed desalination process.]]

=Process Flow Diagram=
See Appendix G for the full process flow diagram. The process has been divided into two sections to aid in organization. The pretreatment section is represented by the 100 series of equipment while the reverse osmosis and posttreatment section is represented by the 200 series. A detailed layout of the reverse osmosis modules is provided in Figure G3 in Appendix G.

In the pretreatment step, a cascade of filters is implemented in the following order: macrofiltration, microfiltration, activated carbon, ultrafiltration and nanofiltration. Three pumps are utilized to account for the pressure drops across these filters. After pretreatment, the clarified seawater stream is split and sent to a pump and an energy recovery device to pressurize the reverse osmosis feed. After going through both reverse osmosis banks, the deionized water is remineralized with brine and disinfected with chlorine dioxide. The brine is used in the energy recovery device, then returned to the sea.

Control systems were added to this process to maintain safe operations and a standard level of product quality. For pretreatment, a fail-close system measures the level of seawater in V-100 and regulates the amount of sodium hypochlorite needed to disinfect the seawater. In the 200 series, two control loops were implemented, including a fail-close pressure control loop that maintains the necessary pressure by monitoring the concentrate entering C-200. A fail-open level controller measures the amount of purified water stored before post-treatment. The valve can increase the inlet flow rate of the final storage tank if the liquid level of purified water exceeds the set-point. This by no means encompasses all the control that should be implemented in the process and only serves as a representation of the necessity of these systems.

=Mass and Energy Balances=
The feed stream is assumed to be at the average seawater temperature of around 25℃ and at 1 atm [23]. The inlet water stream is estimated to have 500 ug/L of particulate matter. Flow rates, conditions, and compositions of each stream can be found in the stream table in Appendix H.

Pressure drops calculated for each pump and filtration step to estimate the utility requirements in the process can be found in Appendix I. The Ergun equation was used to calculate the pressure drops across each level of macrofiltration, which removes all particulate matter. Imperfect packing was assumed, with a void fraction of 0.4. The particle diameter was based off of the Wentworth Grain Size Chart [24], which can be found in Appendix J. The diameter of the filtration tank is 3.5 m because having a larger cross-sectional area decreases velocity. The pressure drops for the microfiltration, ultrafiltration, and nanofiltration modules were calculated using the Darcy equation and transmembrane pressure relationships from literature [25].

Granular activated carbon (GAC) filtration will be used for the dechlorination of the pretreated water. According to the US EPA, surface loading rates for GAC filters range between 2 to 10 gpm per square foot [26]. Using this typical flow rate, the pressure drop of flow through the activated carbon filter was estimated to be 0.6 psi per foot of bed depth found in Appendix K [27]. The activated carbon filter is assumed to effectively remove all of the hypochlorite from the stream.

The amount of brine used to remineralize the permeate ensures that the chlorine levels of the drinking water outlet remain below the upper limit set by South African regulations, as stated in Appendix C. The required concentrations of active chlorine in pretreatment and posttreatment to effectively disinfect the streams are 3 mg/L and 0.2 mg/L, respectively [28, 29].

=Design Constraints=
Sustainability has been a strong motivator throughout the design process. An energy recycle system has been implemented to combat this issue. This allows for cheaper production cost and lower energy requirements. Further sustainability efforts could include utilizing renewable energy sources such as solar, wind, and nuclear power. Environmental and ethical considerations are involved with releasing brine back into the sea. Although there are no laws controlling the release of concentrate back into the ocean, the process has been designed to carefully consider the impact and location of the release. The process avoids dangerous effects of this action by being located in an area far from declared marine protected areas. In addition, the brine will be released 500 m away from shore, decreasing its impact on coastal marine life.

The most pressing ethical consideration is the required level of water quality. By selecting a location outside of the United States, different regulations and standards could be applied to the purified drinking water that will be sent to consumers. The drinking water produced in this process will not only meet the drinking water standards of South Africa but will also exceed those limits to provide a healthier and cleaner water alternative allowing for a better drinking water experience.

Throughout the plant, safety considerations heavily impact the specifics of the process. Multiple control loops were implemented to ensure effective and safe production. The chloride concentration is near its maximum possible level. If the chloride ion concentration increases, it could cause increased corrosion. Therefore, to ensure a longer lifetime for the process, all pipes in the system are to be made with polyethylene. An increase in the corrosiveness of the water could cause metals ions to contaminate the water with certain types of piping as is currently being seen in Flint, Michigan. Therefore, chloride ion concentration will be carefully monitored to prevent negative health effects.

=Equipment Sizing and Trade-offs=
Equipment was sized to ensure that capital costs are not unreasonably high while satisfying the design basis requirements. The macrofiltration tank is filled with large amounts of sand with different levels of coarsity. The tank was designed to be as wide as possible at 3.5 meters to minimize the pressure drop and pumping requirements. For the amount of water being filtered daily, roughly 50 kg of particulate matter will accumulate. At an approximate density of 3,000 kg/m3, this would result in an accumulation of 16.7 liters of particulate matter per day. Thus, with a total height of 2 meters, there should be enough space to accommodate both sand and accumulated particulate matter.

Filters are sold as compact modules containing a large surface area. Increasing the area of the filters decreases the pumping requirements of the system, but increases the capital cost of purchasing the required filters. For each filtration step, many of these units will be used in parallel to meet the filtration surface area requirements, shown in Appendix L. Care was also taken to ensure that the pressure drop across the filters did not exceed product specifications.

The system incorporates three storage tanks. The first tank has a inlet flow rate of 7080 m3/hr, and the second and third tanks each have inlet flows of 5670 m3/hr. Because these flow rates are high, the tanks cannot hold water for long periods of time. The storage capacity for this facility was chosen to be 1.5 hours’ worth of flow. With two parallel units at each step to ensure repairs can be made efficiently, the total storage capacity of this facility will be about 64,000 m3 [30].

Dead end filters were chosen for all of the filtration steps. This increases the capital cost required as the cleaning process for dead end filtration necessitates the purchase of an excess of membranes. Filters will be active for 60 seconds and subsequently cleaned for 5 seconds while the flow is rerouted to another set of membranes. During cleaning, air is sent through the filters in the direction opposite the flow. Purchasing an excess of filters ensures that the required surface area is always available even though roughly 1/13th of all filters will be undergoing cleaning at any moment [31]. The benefit of dead end filtration is that it results in a higher yield, and therefore lower pumping requirements overall.

=Economic Analysis=
The current capital cost for the project is $22.4 MM, with an OSBL capital cost of $8.96 MM, as predicted by the Aspen Process Economic Analyzer. This capital cost includes the equipment cost, $18.4 MM, the membrane costs for pretreatment and reverse osmosis, $2.1 MM, and piping, land, and facility costs totaling $1.9 MM. Detailed breakdown of equipment and membrane costing can be seen in Appendix M. Equipment costs have been estimated using fiberglass reinforced plastic for construction of tanks and using polyethylene for piping to reduce corrosion [32]. Periodic backwashing is required to remove any particulates that gather during filtration; this cost has been accounted for in maintenance, which was estimated to be 7% of the ISBL investment [33, 34].

The main revenue of producing approximately 132,000 m3 per day of desalinated water sold at $0.81 per metric ton is $37.5 MM [35]. Utility costs for this design are $13.7 MM per year. This value arises from the requirement of 20,209 KW of electricity purchased at $0.08 per KWh [36]. To decrease the environmental impact from the large utility requirement, electricity will be obtained from the Ingula Pumped Storage Scheme [37], a hydropower plant located approximately 250 km away from our proposed location.

Raw material costs for this project are relatively low, approximately $0.04 MM. This is mainly due to the fact that the seawater is obtained less than 500 m away from the coast and does not require transport or processing. The sodium hypochlorite for chlorination during pre-treatment can be purchased at $108 per metric ton [38], and the chlorine dioxide can be purchased at approximately $2200 per metric ton [39].

Land cost was estimated using online property listings by finding different vacant lots near Amanzimtoti. The cost per acre was estimated to be R550,000 or $35,000. The required land area is approximately 8 acres [40], resulting in total land costs of approximately $300,000. The predicted salary and overheads are $0.8 MM, which include 3 shifts of 5 operators at a salary of R150,000 or $9,500 per operator, the average salary for a plant operator in South Africa [41].

Using typical industrial economic relationships, the OSBL was estimated to be 40% of the ISBL, the engineering costs to be 30% ISBL plus OSBL, the contingency charge to be 10% of the ISBL plus OSBL, and working capital to be 20% of the total fixed capital cost. The construction schedule models plant building during the first two years at 40% and 60% investment, respectively. Full production is set to begin immediately upon plant completion. The depreciation rate chosen was the 5 year MACRS method because it yields better return than the straight-line model. The tax rate was selected to be 35%, reflecting the standard industrial tax rate in South Africa [42].

The net present value at 10 years is predicted to be -$34.2 MM and increases to -$23.5 MM at 20 years with an internal return rate is 2.9% at 20 years. As of now, the process is not economically feasible. However, as described by the sensitivity analysis below, the process may become economically viable if utility costs and product revenue change favorably.

=Optimization and Sensitivity Analysis=
A cost optimization was performed to determine how reverse osmosis modules could be organized to minimize costs. As the number of modules increases, the capital and operating costs decrease as seen in Figure 3. The module costs are trivial in comparison to the capital costs of the pumps. As the number of modules approaches 900, utility requirements approach a minimum value, but the number of modules needed in the second bank begins to increase exponentially, causing a spike in the capital costs. As a result, the optimal balance between operating costs and utility costs was found to be 900 modules in the first bank and 4000 modules in the second bank.

[[File:Figure3RHP.PNG |thumb|center|400px|alt=Alt text|Fig 3: Cost optimization as a function of number of Bank 1 RO modules]]

A sensitivity analysis was performed to elucidate the effect of four major variables on the NPV at the end of 20 years: selling price of water, maintenance cost, operating cost, and capital cost. Each variable was modified between 75% and 125% of its present value to determine its effect on the NPV. Overall, the price of water has the largest impact, followed by the operating costs, capital costs, and the maintenance costs, as seen in Figure 4.

[[File:Figure4RHP.PNG |thumb|center|400px|alt=Alt text|Fig 4: Sensitivity analysis showing changes in NPV as prices and costs change]]

Although the NPV is currently negative, the sensitivity analysis shows a path towards having a positive NPV. If the price of water increases by 15% and the cost of energy decreases by 15%, the NPV reaches a positive value. These are values that could potentially be reached through negotiations with various government agencies and energy providers. It may be possible to obtain electricity at a lower price due to the construction of a 3,000 MW hydroelectric dam in 2017. Additionally, due to the rising water scarcity in Africa, an increase in the price of water can be expected in the future.

=Conclusion=
The proposed process meets the design requirements of producing 100,000 m3 per day of clean drinking water. However, due to high operating costs and significant capital costs, the design is not economically viable, due to a 20 year NPV of -$23.5 MM. A sensitivity analysis showed that the price of water is most important for determining profitability. Currently, the selling price of water is too low for the plant to be profitable. If the plant were to be constructed, contracts could be negotiated between suppliers and electricity providers in order to provide a consistent lower price. Therefore, it is not recommended to implement the plant design unless there are significant changes in the market for drinking water in South Africa.

=References=

=Appendices=

==Appendix A==

etc.

Desalination - Team B

2016-03-11T22:46:02Z

2016-03-11T06:33:21Z

Cindyxchen:

Desalination - Team B

2016-03-11T06:32:14Z

Cindyxchen:

Team B Final Report

Authors: Lauren Burke, Cindy Chen, Osman Jamil, Natalia Majewska

Instructors: Fengqi You, David Wegerer

March 11, 2016

==Executive Summary==

__TOC__

==Introduction==

=Background=

=Location Selection=

=Preliminary Market Analysis and Process Selection=

=Process Feed and Composition=

==Process Alternatives==

=Pretreatment=

=Membrane Selection=

=Energy Recovery=

=Remineralization=

=Disinfection=

=Waste Treatment=

==Process Overview==

==Process Flow Diagram==

==Mass and Energy Balances==

==Equipment Sizing and Trade-offs==

==Technical Analysis==

Desalination - Team B

2016-03-11T06:28:18Z

Cindyxchen:

Team B Final Report

Authors: Lauren Burke, Cindy Chen, Osman Jamil, Natalia Majewska

Instructors: Fengqi You, David Wegerer

March 11, 2016

__TOC__
==Executive Summary==

==

Desalination - Team B

2016-03-11T06:27:30Z

Cindyxchen: Created page with "Team B Final Report Authors: Lauren Burke, Cindy Chen, Osman Jamil, Natalia Majewska Instructors: Fengqi You, David Wegerer March 11, 2016"

Team B Final Report
Authors: Lauren Burke, Cindy Chen, Osman Jamil, Natalia Majewska
Instructors: Fengqi You, David Wegerer
March 11, 2016

Heat exchanger

2016-02-19T21:08:52Z

Cindyxchen: /* Linear Programming */

Author: Alex Valdes [2015], Cindy Chen [2016] 

Stewards: Jian Gong, and Fengqi You

 

==Introduction==
Heat exchangers are necessary process units that are part of any detailed process flow diagram. Process streams commonly interact through heat exchangers in order to save money on heating and cooling utilities. Furthermore, the surface area of the heat exchanger is proportional to the amount of heat that can be transferred and is the most indicative cost component of a heat exchanger (Wilcox, 2009). Therefore, all of the commercial simulators include models for heaters, coolers, heat exchangers, fired heaters,and air coolers (Towler and Sinnott, 2013). Typically, the only inputs necessary for heat exchanger models to converge are properly specified inlet streams (flow rate, temperature, pressure, composition), the pressure drop of flow pathways, and the outlet temperatures or the duty.

=Aspen HYSYS V8.0=

Model simulators such as HYSYS are extremely useful for engineers to quickly estimate capital costs and utility requirements.

===Model Equations===
HYSYS uses the following equations for adiabatic steady state heat exchangers (Wilcox, 2009):

<math>Q = Nps(Hps,in-Hps,out)</math> (heat transferred from process stream)

<math>Q = Nus(Hus,out-Hus,in)</math> (heat transferred to utility stream)

<math>Q = UAF(dTavg)</math> (rate of heat transfer)

where: Q is the rate of heat exchange (e.g., in kJ/h), Ni is the flowrate of stream i (e.g, in kmol/h), Hi is the specific enthalpy of stream i (kJ/kmol), U is the overall heat transfer coefficient (kJ/m2.K), A is the heat exchange area (m2), F is the correction factor for the deviation from cocurrent or countercurrent flow, dTavg is the average temperature difference between the streams for true cocurrent or countercurrent flow.

===Heuristics===
Before performing simulation, it is important to have an idea of practical considerations and parameter values. A comprehensive list of Heuristics in Chemical Engineering (Walas, 1990) is found at http://people.clarkson.edu/~wwilcox/Design/heurist.pdf. For shell and tube heat exchangers:
*Tube side is for corrosive, fouling, scaling, and high pressure fluids.
*Shell side is for viscous and condensing fluids.
*Pressure drops are 1.5 psi for boiling and 3-9 psi for other services.
*Water inlet temperature is 90°F, maximum outlet 120°F.

More rules of thumb for various types of heat exchangers, as well as many other pieces of process equipment, are contained in the source above.

===Types of Heat Transfer Operations in HYSYS===
Although shell and tube heat exchangers are widely used in practice, and most commonly taught to engineers first learning heat transfer, HYSYS contains various heat transfer units:

Table 1: HYSYS Heat Transfer Simulation Operations
{| class="wikitable"
|-
!Unit
!Heat Exchange Between
!Input(s)
!Output(s)
|-
|Cooler
|Hot process stream and utility (energy)
|Flow rate, composition of inlet stream. For inlet and outlet stream TWO out of THREE: temperature, pressure, vapor phase fraction
|Duty, Q
|-
|Heater
|Cold process stream and utility (energy)
|Same as cooler
|Duty, Q
|-
|Heat Exchanger
|Two process streams, shell and tube geometry
|Same as cooler for first process stream. For second process stream: composition, and TWO out of THREE: temperature, pressure, vapor phase fraction for inlet and outlet
|Second process stream flow rate; Duty, Q; Heat Transfer Coefficient, U
|-
|LNG Exchanger
|Multiple process streams, plate or plate-fin geometry
|Same as shell and tube Heat Exchanger
|Same as shell and tube Heat Exchanger
|-
|Air Cooler
|Process stream and ideal air mixture
|Same as cooler for process stream.
|Total air flow; Fan rating; Heat transfer coefficient, U
|-
|Fired Heater (Furnace)
|Combustion fuel gas and process stream(s)
|Same as cooler for process stream. Air/fuel ratio
|Flow rate of fuel/air; Duty, Q; Heat transfer coefficient, U
|}

==Tutorial for Heat Exchanger==
====Input Properties====
* Open Aspen HYSYS and create a New Case under File menu.
* Create a component list by adding all components present in the process.
* Select a thermodynamic fluid package that is applicable to the process see Property Package wiki article for more details on options or go to http://people.clarkson.edu/~wwilcox/Design/thermodl.htm

====Enter Simulation====
* In the model palette, there are a few options for heating/cooling units. Use a Heater or a Cooler to change the temperature of one process stream using a utility. Use a Heat Exchanger for two process streams exchanging heat, thereby changing the temperature of each. The following steps are for a Heat Exchanger, but the steps are similar for a Heater or Cooler.
*Click on the exchanger in the Flowsheet window. In the Design Tab under Connections, create Tube and Shell Side Inlet and Outlet streams (four streams). Choosing which fluid goes tube side and which goes shell depends on many factors, but some rules of thumb include putting high pressure and/or corrosive fluids tube side (Sloley, 2013).
* To get a feel for how HYSYS interprets the situation, open the Specs section still in the Design Tab (image to the right). It can be seen that HYSYS needs five pieces of information (Degrees of Freedom =5) about the inlet and outlet streams. If the process is overspecified, HYSYS should give a message indicating the type of error, or may simply say no solution. After understanding the information in this table, enter four inlet parameters and one outlet parameter (the temperature of one of the outlets).
*At this point, HYSYS should say that there is an unknown Delta P. Specify (guess) the inlet and outlet pressures, the exchanger should then converge. Good initial guesses for pressure drop are between 0.3-0.7 bar or 30-70 kPa (Towler and Sinnott, 2013)
[[File:fig1.jpg|700px]]

Figure 1. Heat Exchanger in Flowsheet and Design Specs Page

====Customizing Heat Exchanger====
*Once the exchanger has converged and the desired outlet temperature of one of the process streams has been achieved, various parameters can modified and more detailed analyses on the exchanger can be looked at.
*Under the Design Tab, under Parameters, choose one of the five potential heat exchanger models (most common choice is rigorous shell and tube).
Troubleshooting note: sometimes the rigorous heat exchanger model won't converge at first, even if the system is properly specified. Sometimes when this happens, switching to Simple End Point model and then switching back to the rigorous model will make it converge.
*At this point, utilize a neat feature of HYSYS in the Rigorous Model Section of the Parameters page: Size Rigorous Shell&Tube. Clicking on this will have HYSYS automatically run cases to determine optimal parameters of pressure drop and heat exchanger area (there is also an option to automatically run this feature). Once this feature converges, very useful information about the required overall heat transfer coefficient, pressure drop, heat transfer area, and other physical parameters is readily viewable under the Rating Tab.
*Also under the Rating tab, physical features of the heat exchanger such as the number of shell/tube passes can be modified (real heat exchangers typically have more than one or two tube passes per shell). These features of course change the necessary heat transfer coefficient (favorably), pressure drop (unfavorably), and unspecified outlet stream conditions.

[[File:fig2.jpg|700px]]

Figure 2: Rating Tab containing physical/geometric parameters of exchanger

===Common Issues===
There a few common problems that arise when using heat exchangers in these programs.

When heat exchangers are used with streams that go to earlier stages of the process, an information loop occurs and the program is less likely to converge. Many times the process design requires a later process stream, such as the bottoms of a distillation column, to heat an earlier stream, such as the feed to the same column (Towler and Sinnott, 2012).

Another problem arises if the specifications of the heat exchanger are impossible, but the model still converges with physically unreasonable results - such as a temperature cross.

To avoid these problems, it is good practice to use utility heaters and coolers instead of heat exchangers to get an idea of the required heat load and parameters of an exchanger. The heaters and coolers are also useful for obtaining initial guesses of outlet temperatures and pressure drops. After that information is obtained, the designer will have a much better chance of simulating a heat exchanger that will converge with meaningful results (Wilxcox, 2009).

=Heat Exchanger Networks=
A typical plant requires the use of many individual heat exchanger units. This combined process heating and cooling of a plant could be accomplished through the use of utilities, but this will be very costly in term of capital and utility costs. Heat exchanger networks utilize multiple heat exchanger units to optimize the process design. Effective incorporation of heat exchanger networks will minimize the use of utilities, minimize the number of heat exchangers, and minimize costs associated with the design.

==Model Equations and Heuristics==
Heat exchanger networks are designed through insightful use of pinch analysis. The pinch point divides the temperature range into two regions. Pinch point analysis is a tabular procedure that is a convenient alternative to the construction and analysis of hot and cold composite curves. The temperature at which the temperature difference between the hot and cold stream is the smallest is the pinch point temperature. Determining the smallest Δ<math>T_{min}</math> can be useful in calculating the optimal size of the heat exchanger (Marechal, 2005). The four steps of pinch point analysis are:
::1. Divide the temperature range into intervals and shift the cold temperature scale.
::2. Make a heat balance in each interval.
::3. Cascade the heat surplus and deficit through the intervals.
::4. Add heat so that no deficit is cascaded by adding the magnitude of the largest heat deficit from step 3 to each heat surplus and deficit. The position with a sum of 0 is the pinch position. The minimum heating and cooling utilities can also be determined in this step (Bagajewicz, 2008).
 Detailed calculations and examples on obtaining the pinch point can be found on the [https://processdesign.mccormick.northwestern.edu/index.php/Pinch_analysis pinch analysis] page.

A valid heat exchanger network system must follow appropriate heuristics. Heating utility can only be used above the pinch, while cooling utility can only be used below it. Additionally, heat from a hot stream with temperature above the pinch must not be transferred to a cold stream with temperature below the pinch (Towler and Sinnot, 2012).

==Minimum Energy Recovery Targets==
The primary objective of building heat exchanger networks is to maximize the energy recovery from the hot process streams to the cold process streams. Thus, before starting the synthesis of a heat exchanger network, minimum energy recovery (MER) targets must be determined. There are three methods to doing so:
::1. the temperature-interval method,
::2. a graphical method using composite heating and cooling curves, and
::3. linear programming (Seider, 2004).

 The temperature-interval and composite curve methods have been outlined in [https://processdesign.mccormick.northwestern.edu/index.php/Pinch_analysis pinch analysis] page. A linear programming example is discussed below.

===Linear Programming===
Consider a system where two cold streams, C1 and C2, are to be heated and two hot streams, H1 and H2, are to be cooled without phase change. The conditions and properties of these streams are listed below.

[[File:Streams24.png|450px|center|Stream information table (Seider, 2004)]]

Preliminary pinch analysis calculations lead to the following enthalphy changes in each interval.

[[File:Enthalpy_table.PNG|450px|center|Anaylsis Results (Seider, 2004)]]

[[File:Cascade_hen.PNG|225px|thumb|right|Figure 3. Cascade of temperature intervals, energy balances, and residuals (Seider, 2004)]]
 
The linear program can be formulated as:
 
::Minimize <math>Q_{steam}</math>
::with respect to (w.r.t.): <math>Q_{steam}</math>
::Subject to (s.t.):
::<math>\begin{align}
Q_{steam} - R_1 + 30 & = 0\\
R_1 - R_2 + 2.5 & = 0\\
R_2 - R_3 - 82.5 & = 0\\
R_3 - R_4 + 75 & = 0\\
R_4 - Q_{cw} - 15 & = 0\\
Q_{steam}, Q_{cw},R_1, R_2,R_3,R_4 & \ge 0 \\
\end{align}
</math>

 
The variables in this linear program are defined as follows:
::<math>Q_{steam}</math> is the heat entering from a hot utility
::<math>R_i</math> is the residual from interval i
::<math>Q_{cw}</math> is the cold utility duty
where all energy requirements should be multiplied by 10,000 Btu/hr.
 
This program will minimize the heat entering from a hot utility. Figure 3 shows the energy flows between each interval of the system and the values of the variables calculated in the initial and final pass.
 

This linear program set up could be solved using the General Algebraic Modeling System (GAMS), as shown below.

[[File: Gams_problem.png|375px|center|GAMS problem input (Seider, 2004)]]

The GAMS Program first names the variables used in the program, stating the Z is the objective variable that will be minimized and the other variables are "positive variables," satisfying the final condition stated in Equation LP.6 that all utilities and residual values must be greater than zero. The linear program model is named HEAT and is solved using LP (signifying linear program algorithms) to minimize the Z variable.

[[File: Gams.PNG|375px|center|GAMS problem input (Seider, 2004)]]

From this result, <math>R_3 = 0</math>, signifying that the pinch is located at interval 3, as seen in Figure 3. Additionally, this result reveals that the hot utility requirement, <math>Q_{steam}</math>, is 500,000 Btu/hr and the cold utility requirement, <math>Q_{cw}</math>, is 600,000 Btu/hr. However, this simulation only accounts for sensible heat changes. Additional variables, such as heat of reaction, heat of mixing, and even variable specific heat over time, should be considered in a more representative model.

This example is taken from Process Design Pinciples: Synthesis, Analysis, and Evaluation by Seider, Seader, and Lewin.

==Tutorial for Heat Exchanger Network==
Heat exchanger networks can be modeled using Aspen Energy Analyzer. Below is a tutorial provided by Aspen Technology (Aspen, 2011).

===Input Properties===
* Open Aspen Energy Analyzer
* Verify the units for your simulation through Tools >> Preferences >> Variables tab >> Units page
* Create a heat integration (HI) case under Features menu
* Enter process stream data in pop-up window (as shown below)

[[File: process_streams.JPG|400px|center|Process streams window (Aspen, 2011)]]

* Once the inlet and outlet temperatures are entered, Aspen will determine the stream type as hot or cold.
* Input the enthalpy or flow heat capacity information for each stream. Stream segmentation is useful for streams that change phase or have non-linear variations in enthalpy as the temperature changes. After entering the information for the final segment, the process stream is complete and the window would appear as shown below.

[[File: enthalpy_sim.JPG|350px|center|Segment data window (Aspen, 2011)]]

* Enter stream name, first inlet temperature value and target outlet temperautre value. Only the outlet temperature values and heat load/enthalpy values need to be added, as the simulation calculated the inlet temperature values. The process streams tab will appear as shown below once complete. The red arrows represent hot streams that need to be cooled and the blue arrows represent cold streams that need to be heated.

[[File: process_streams_2.JPG|400px|center|Complete process streams window (Aspen, 2011)]]

* Enter utility stream data in the utility streams tab. In the Name column, scroll through to find the utilities needed for the simulation. Examples include: cooling water, LP steam generation, air, fired heat, etc. Aspen will report the default costs associated with each utility, which will be used to calculate operating costs for the system.

===Enter Simulation===
* Examine the targets of the case. From HI Case view, select the Open Targets View icon that appears at the bottom of the view for all tabs. This window will display the energy targets, pinch temperatures, minimum number of units required to build the network, and preliminary cost analysis.

[[File: case_targets.JPG|400px|center|Case targets window (Aspen, 2011)]]

* Calculate the DTmin of the simulation. In the Targets window: Range Targets tab >> Calculate button >> Plots page. If a better approximation for optimal DTmin value is needed, click the Clear Calculations button to clear the plot before adjusting the DTmin range. To adjust the range, click the DTmin Range button located next to the Calculate button. A optimized DTmin range would resemble the figure shown below.

[[File: dtmin.JPG|400px|center|Range targets window (Aspen, 2011)]]

===Modify Heat Exchanger Network Diagrams===
* The heat exchanger network design can be viewed by clicking the Open HEN Grid Diagram icon in the HI Case view.
* Heat exchanger networks may include splitters, heat exchangers, and coolers.

* Add a splitter to the network: open palette view located in the bottom right corner of the Grid Diagram tab. Right click and hold the Add Split icon. Drag the cursor over the stream to be split until a bulls eye icon appears. Once the mouse is released, a solid blue dot will appear on the stream to represent the splitter. Click the blue dot once to expand the slitter.
* Add a heat exchanger: right click and hold the Add Heat Exchanger icon from the Design Tools palette. Once the mouse is released, a solid red dot will appear on the desired stream. Click and hold the red dot to the new stream. The heat exchanger will appear once the mouse is released. Double click either end of the heat exchanger to open the Heat Exchanger Editor view. On the Data tab, click the Tied checkbox by the appropriate streams and enter other temperatures as needed.

[[File: HX2.jpg|500px|center|Heat exchanger windows (Aspen, 2011)]]

* Adjust the split ratio: open HEN Diagram Properties View and click the New button. Input Split Ratio into the Name field. Select Edit >> Annotations tab >> Streams group >> Segments drop-down list >> Split Fraction. Double-click on either end of the heat exchanger to observe the changes in the hot stream inlet temperature. Double-click on either end of the splitter to open the splitter view. Adjust the split ratio accordingly.

[[File: Split_ratio.JPG|300px|center|Split ratio windows (Aspen, 2011)]]

Once the heat exchanger network is complete, the status bar on the Grid Diagram tab will appear green.

==Future Developments==
While heat exchanger networks try to optimize the energy costs for heating and cooling utility, capital costs are often difficult to calculate and incorporate in the analysis. Around the pinch point, the hot and cold streams are most constrained. This leads to the requirement of large heat exchangers to effectively transfer heat between hot and cold streams. A design for maximum heat recovery and minimum number of exchangers will typically include a loop in the network.

[[File:Hen_loop.jpg|500px|thumb|Example of loop in network design (Towler and Sinnot, 2012)|center]]

As seen in the figure above, a loop may cross the pinch point. These loops can be broken by transferring heat across the pinch point. This, however, violates the heuristic that heat should not be transferred across the pinch point and at least one violation of the specific Δ<math>T_{min}</math> will also occur. To restore Δ<math>T_{min}</math>, heat can be shifted along a path, which increases the energy consumption of the process. From this dilemma, it is clear that there are pros and cons to the attempt to minimize any heat exchanger network.

Analysis with the incorporation of capital costs can be done by assessing the heat exchanger network as a [https://en.wikipedia.org/wiki/Linear_programming#Integer_unknowns mixed integer non-linear programming] (MINLP) problem. MINLP problems are typically solved using numerical solvers. But large scale MINLP problems, such as the ones that may be generated for a process plant, may be difficult to solve using available numerical algorithms and solvers (Biegler, 1999).

=References=
# G.P. Towler, R. Sinnott. ''Chemical Engineering Design: Principles, Practice and Economics of Plant and Process Design''. Elsevier, 2012.
# Sloley. "Shell-and-Tube Heat Exchanger: Pick the Right Side". Chemical Processing. October, 2013
# Wilcox. "Sizing of heat exchangers using HYSYS or UniSim". 2009. http://people.clarkson.edu/~wwilcox/Design/hxsizing.htm date accessed: March 1, 2015
# S.M. Walas. (1990). ''Chemical Process Equipment - Selection and Design''. Elsevier
# W.D. Seider, J.D. Seader, D.R. Lewin, ''Process Design Principles: Synthesis, Analysis, and Evaluation'', Wiley: New York, 2004.
# Aspen Technology, Inc. Aspen Energy Analyzer: Tutorial Guide. (2011)
# M.J. Bagajewicz. ''Heat Integration'' Chemical Engineering PanAmerican Collaboration. 2008.
# F. Maréchal, D. Favrat. ''Combined Exergy and Pinch Analysis for Optimal Energy Conversion Technologies Integration''. 2005.
# L.T. Biegler, I.E. Grossmann, A.W. Westerberg. ''Systematic Methods of Chemical Process Design'', Prentice Hall: New Jersey, 1999.

File:Gams problem.png

2016-02-19T21:08:12Z

Cindyxchen: