Simulation and modeling of MEG (Monoethylene Glycol) regeneration for the estimation of energy and MEG losses

Accepted Manuscript Simulation and Modeling of MEG (Monoethylene Glycol) Regeneration for the Estimation of Energy and MEG Losses

Hyun Soo Son, Yoo Ri Kim, Sang Min Park, Michael Binns, Jin-Kuk Kim PII:

S0360-5442(18)30962-9

DOI:

10.1016/j.energy.2018.05.128

Reference:

EGY 12963

To appear in:

Energy

Received Date:

16 January 2018

Accepted Date:

19 May 2018

Please cite this article as: Hyun Soo Son, Yoo Ri Kim, Sang Min Park, Michael Binns, Jin-Kuk Kim, Simulation and Modeling of MEG (Monoethylene Glycol) Regeneration for the Estimation of Energy and MEG Losses, Energy (2018), doi: 10.1016/j.energy.2018.05.128

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Simulation and Modeling of MEG (Monoethylene Glycol) Regeneration for the Estimation of Energy and MEG Losses Hyun Soo Son a, Yoo Ri Kimb, Sang Min Park b, Michael Binnsc, Jin-Kuk Kima,*1 a

Department of Chemical Engineering, Hanyang University 222 Wangsimni-ro, Seongdong-gu, Seoul

04763, Republic of Korea b Chemical Engineering Research Department, Advanced Technology Institute, Hyundai Heavy Industries

Co., Std., 1000 Bangeojinsunhwan-doro Dong-gu, Ulsan 682-792, Republic of Korea c

Department of Chemical and Biochemical Engineering, Dongguk University-Seoul, 30 Pildong-ro 1-gil,

Jung-gu, Seoul, 04620, Republic of Korea

Abstract MEG (Monoethylene glycol) is a hydrate inhibitor used for the recovery of subsea oil and gas. For practical and economic reasons, it is necessary to extract and re-use the MEG through a regeneration unit which removes hydrocarbons, water and salts. The economic performance of regeneration process depends on MEG losses and the amount of heat and power required for the separation. Since recovering as pure MEG as possible without disturbances induced by salts is important to maintain process sustainability, this study focuses on the modeling and simulation of salt and water removal steps, including the prediction of hydrate inhibitors required for subsea condition. Also, design methods presented in this study systematically provide identification of appropriate configurations and operating conditions, with which the economic performance concerning MEG loss and energy consumption can be systematically evaluated. Models are developed in process simulators and validated with industrial data. Hydrate inhibitor recovery of 99.42% from the reclamation unit considered in this study is comparable to typical recoveries reported in commercial processes in the range of 99.4% ~ 99.5%. It is also found that energy used for the separation of MEG from water in re-concentration unit accounts for at least 60% of total energy consumption.

* Corresponding author: Tel) +82 2 2220 2331 Email) [email protected] 1

ACCEPTED MANUSCRIPT

Keywords: Flow assurance; Hydrate inhibition; MEG regeneration; Energy efficiency; Process design; Process modeling;

Highlights  Systematic operational analysis of MEG process for ensuring offshore flow assurance  Integrated process modeling for the design of MEG reclamation and re-concentration  Multi-period design framework for enhancing operability of MEG process

2

ACCEPTED MANUSCRIPT

1. Introduction

While world’s energy consumption is expected to rise constantly until 2040 [1], the production gap between onshore and offshore fields has been narrowed for recent oil and gas field development [2]. A recent trend in oil and gas industry is that share in energy production from harsh environment has been increasing [2]. However, the extraction of subsea oil and gas is challenging due to the harsh environment under which it is transported in pipelines causing undesirable effects such as corrosion and the generation of slugs, hydrates, and emulsions. These undesirable byproducts are the result of reactions involving specific components extracted from the reservoir, and their accumulation can lead to the failure of subsea operations. The most serious issue is the formation of hydrates which can lead to the reduction of product quality, blockage and pressure loss in pipelines [3]. Hydrate is a crystalline compound that consists of light hydrocarbons surrounded by water, which can cause plugging in equipment and, eventually, processing failure. Hydrate is easily formed under low temperature and high pressure conditions [4]. Unfortunately, it is difficult to control the temperature and pressure in a subsea pipeline and so, alternative strategies for preventing hydrate formation are desired [5]. The most attractive method to prevent hydrate formation is with the injection of hydrate inhibitors such as MEG (Monoethylene Glycol) and methanol which can reduce and prevent the formation of hydrates. Although methanol is cheaper than MEG, an economic comparison of these two has shown that MEG is more cost-effective due to the ease with which it can be recovered and re-used [6]. It has also been shown that higher concentrations of MEG together with a bicarbonate pH stabilizing agent can lower corrosion rates [7]. For these reasons, the injection of MEG is commonly used in industry. The typical strategy for injection and recovery of MEG is shown in Figure 1 where a “rich” MEG containing 50 – 60 wt % MEG and impurities (water, salts, hydrocarbons) is processed in the MEG regeneration unit giving “lean” MEG containing around 80 - 90 wt % MEG which is recycled and injected. The impurities in the rich MEG include salts and water which should be removed in the regeneration unit. Therefore, a typical MEG regeneration unit consists of salt separation (i.e., reclamation) and water separation (i.e., re-concentration). There are two main types of MEG regeneration unit called full stream 3

ACCEPTED MANUSCRIPT

and slip stream reclamation which differ in the way they handle the different salts. Full stream reclamation removes both monovalent and divalent salts simultaneously before re-concentration, while slip stream initially removes all divalent salts then later removes part or all of monovalent salts either downstream or upstream of the re-concentration. Full stream reclamation requires more energy, but it is effective for cases where there are high TDS (Total Dissolved Solids) levels while slip stream reclamation is practical for more modest TDS levels [8]. MEG, as an alternative to TEG (Triethylene Glycol), may be also used for the dehydration of natural gas. Contreras and Foucart compare MEG with TEG for dehydration processes by evaluating capital and operating cost and they provide technical guidelines for the selection of dehydrating agents, considering natural gas specifications and dew point control [9]. On the other hand, MEG may be utilized for both sweetening and dehydrating natural gas in the contact tower [10], in which the heat-integrated process is simulated and optimized with Aspen HYSYS, with consideration of different injection rate and concentration of MEG. Although not only the performance of dehydration and sweetening, but also economic feasibility is investigated, the main focus is limited to study on dehydration, without full consideration of MEG recovery. Economic evaluation can be also an effective way of evaluating technical advantages, which has been applied to MEG regeneration process. The capital cost and operating cost for hydrate inhibition by MEG and MeOH are studied [5], in which various hydrate inhibition systems applied in the industry are discussed. While LCC (Life Cycle Cost) method is used to compare MEG injection with MeOH injection [6], Kim et al. [11] carry out Aspen HYSYS simulation for subsea and topside processes and evaluate the hydrate inhibition both by MeOH and MEG in terms of NPC (Net Present Cost). From their analysis, the better economics for MEG than MeOH, based on NPC per inhibitor injection flowrate, the more injection required. On the other hand, Kim et al. [12] propose new slip stream MEG regeneration process having two-stage distillation columns, which is then compared with conventional process, in terms of capital cost, operating cost and salt precipitation. They conclude as the modified MEG regeneration process gives better economics than conventional process. Also, the simulation study is conducted with Aspen Plus and it is found that less salts precipitation in the column from the new process is expected than the conventional process when various feed conditions are applied. Exergy analysis is applied for the analysis of MEG regeneration processes, namely, traditional process, 4

ACCEPTED MANUSCRIPT

slip stream process and full stream process, with which inefficient equipment units are identified [13]. On the other hand, Park [14] simulates both full and slip stream reclamation MEG regeneration units using the NRTL (Non-Random Two Liquid) property method, in which optimal operating conditions of pretreatment are presented. In parallel with process simulation, experiments for the mixture of MEG-H2ONaCl-CaCO3 are carried out to analyze the relationship between pH and deposition of divalent salts. Teixeira et al. [15] propose the method for recovering THIs (Thermodynamic Hydrate Inhibitors) which are lost from the vapor outlet of high pressure separator in the conventional process. This process is called as SS THI RU (Supersonic Separator Thermodynamic Hydrate Inhibitor Recovery Unit), which consists of a static swirling device, laval nozzle and diffuser. Different process schemes for three THIs, i.e., methanol, ethanol, and MEG are simulated with Aspen HYSYS, which shows that SS THI RU reduces the loss of THI in the vapor phase by at least 79% without compromising capability of inhibit hydrate formation. However, since the main interest in subsea flow assurance is to control the level of salts in the process so that any potential operating problems in the equipment caused from the formation of fouling, emulsion and foaming, can be reduced. For this reason, a number of studies have been focused on understanding of thermodynamic behavior for MEG regeneration process. Bikkina et al. [16] test a wide range of corrosion inhibitors and provide selection guidelines under harsh process conditions of MEG regeneration. AlHarroni et al. [17] discuss the impact of the recovered MEG on hydrate formation under various operating conditions, especially during start-up and clean-up phases, and it is found that the efficiency of MEG is gradually deteriorated over the time, due to continuous and repeated use and recovery of MEG [18]. Also, equilibrium conditions of hydrate formation at the presence of MEG are experimentally examined [19]. Otherwise, the component interactions in the system are investigated as a holistic manner. Wang et al. build a mixed solvent framework by combining Helgeson-Kirkham-Flowers EOS (Equation of State) with Pitzer-Debye-Hückel equation to calculate thermodynamic and transport properties in MEG-H2O-inorganic salts-gases mixture [20]. Alternatively, salt behavior is predicted by expressing solubility as an empirical function of molality and activity coefficient in Sandengen’s model, which reflects the influence of MEG and salts [21]. To our best knowledge, it has not been discussed with regard to detailed analysis on operating parameters of MEG regeneration process and their impacts on process efficiency, specifically, energy and 5

ACCEPTED MANUSCRIPT

MEG losses. The goal of this study is, hence, to develop systematic procedure, with which cost-effective and energy-efficient process design is obtained for MEG regeneration units, while ensuring to recover as pure MEG as possible for its repeated and continuous use. The design method allows systematic evaluation of process performance in the context of energy consumption and MEG loss. Consideration of possible process configurations and its technical impacts on the process performance has been made, while the most appropriate and practical conditions for key design and operating variables under offshore production environment has been identified, leading to effective suppression of hydrate formation. This study includes the estimation of MEG injection rate required for inhibiting hydrate formation in subsea production, based on well production conditions and subsea production plan, as well as process modeling and design of MEG recovery unit in the offshore topside platform. A series of unit operations are modeled with the commercial simulator, subject to design specifications for the recovery and the product purity of MEG. In this study, the distillation column is to recover 80 wt% of MEG as a bottom product. For the design of a reclamation unit, the most appropriate selection of precipitating agent is considered as salt removal is heavily dependent on the types of salts. Also, colligative property is accommodated in the prediction of equilibrium operating conditions for the reclamation unit, while economic trade-off between operation and capital cost costs is investigated through sensitivity analysis. Techno-economic dependence of temperature on energy consumption and MEG losses of MEG recovery units is also evaluated. With these investigations and analyses, MEG recovery process can be designed to avoid any potential shutdown without having salts upset conditions, while energy consumption is effectively minimized in a holistic manner. In particular, the amount of MEG loss is estimated with modeling and design methodology presented in this work. This is one of novel elements in the current work because the amount of MEG loss is assumed or pre-specified in previous modeling and design studies. The MEG loss rate calculated is compared with public data for MEG recovery units of CCR technology or CAMERON, with which soundness in the modeling is validated. Another novelty in this study is to consider the presence of salts and to evaluate its impact on energy consumption and MEG loss, which has not been explicitly addressed in previous researches. Furthermore, computer programming based on active X is made to link different simulators and to exchange process design information effectively, which overcomes the limitation in accessibility between different simulation database. 6

ACCEPTED MANUSCRIPT

The applicability of process design methodology presented in this work is demonstrated through the case study by considering variation in feed conditions for different production phases. With consideration of different wellhead conditions and production rate, wellhead operating temperature and MEG injection rate required are effectively estimated, and operating characteristics of production field can be predicted in confidence. Such understanding improves operating flexibility for the design of the MEG regeneration process and allows cost-effective management of energy.

7

ACCEPTED MANUSCRIPT

2. Process modeling

The MEG regeneration unit considered in this study consists of re-concentration and reclamation sections. The re-concentration is designed based on typical industrial data, while the reclamation is assumed to start the operation later so that the specific characteristics of well regarding salt production can be systematically reflected in the design for ensuring flexibility in operation. Section 2.1 describes the modeling and validation of the re-concentration, while sections 2.2 and 2.3 describe the details of reclamation modeling and the integration of the two separation steps.

2.1 Model building and its validation at re-concentration

The MEG regeneration unit removes water and any remaining hydrocarbons to recover MEG. This is modeled with a commercial simulator, Unisim Design®. Figure 2 and Table 1 show a snapshot of PFD (Process Flow Diagram) and feed conditions, respectively. The Peng Robinson EOS and NRTL property methods are used to estimate thermodynamic behavior and to predict VLE (Vapor-Liquid Equilibrium). Separation of the remaining hydrocarbons is modeled with the Peng Robinson EOS, while the separation of water and MEG is predicted with NRTL method, according to the selection mechanism of AspenTech [22]. As seen in Table 1, this unit has two feeds including considerable amounts of heavy hydrocarbons. These are present because the targeted gas well does not fully separate liquid hydrocarbons, which are mixed in the aqueous phase with the MEG. There are various methods to calculate the required amounts of hydrate inhibitors to prevent hydrate formation. A simple method, is to use Hammerschmidt equation given here as eq 1. This equation is known to be accurate when a maximum 30 wt % of MEG or methanol are present. Alternatively, NielsenBucklin equation as shown in eq 2 can calculate the amount of hydrate inhibitor for the mixtures containing up to 80 mol % of inhibitor, particularly for methanol [23]. Hammerschmidt equation is used to calculate the amount of MEG for hydrate prevention in this study, rather than Nielsen-Bucklin equation. Nielsen-Bucklin equation is adequate when methanol is used as an inhibitor. Hammerschmidt equation is an empirical correlation which is based on experimental data [24] and, furthermore, has been validated 8

ACCEPTED MANUSCRIPT

with experimental data [25]. The accuracy and applicability of Hammerschmidt equation have been widely reported in both academic [26] and industrial communities [27]. ∆𝑇 =

𝐾H𝑊

(1)

𝑀(100 ‒ 𝑊)

where ΔT is the temperature depression (°C), M is the molecular weight of the inhibitor (g/mol), W is the inhibitor concentration in percentage mass and KH is a constant which is 1297 for ethylene glycol. ΔT = -72 ln(1-xM)

(2)

where ΔT is temperature depression (°C) and xM is the mole fraction of methanol. In both eq 1 and eq 2, temperature depression is required for the prediction of inhibitor concentration which is correlated with hydrate formation conditions. For prediction of hydrate formation conditions, the method of Katz can be applied [28]. The effect of salts on temperature depression (i.e., reinforcement of hydrate inhibition with salts) can be considered further for predicting temperature depression as described by østergarrd et al. [29]. The process configuration shown in Figure 2 is designed to obtain an MEG product stream by removing slugs, liquid hydrocarbons, and water from feed stream. More specifically, slugs (formed during pipeline transportation) are removed in the slug vessel (SEP-1). The remaining hydrocarbons are separated in a rich MEG flash drum (SEP-2) followed by re-concentration, where water is removed from MEG in a distillation column. The condenser and reflux units are modeled separately from the main column to adjust and fix the condenser temperature at E-3. There are two heaters labelled as E-1 and E-2 which are required for preheating rich MEG stream in addition. The outlet stream temperatures and pressures are set at the specified values as shown in Table 2. Flash separators (i.e., SEP-1, SEP-2 and SEP-3) are modeled at the same pressure as the inlet stream. The distillation column is simulated using 26 stages in total with the feed being added on the 18th stage. The reflux ratio for the column is adjusted to satisfy product concentration (i.e., 80 wt % MEG) which is achieved with a value of 0.56. The results of process modeling for re-concentration process in Unisim Design® shown in Table 3 are compared with the typical industrial plant data. As can be seen from this table, most of the key indicators including mass and energy balances show good agreements between model predictions and industrial data. The temperature of lean MEG (S3) and reboiler duty are slightly under-predicted which might be a result 9

ACCEPTED MANUSCRIPT

of some inherent uncertainty in the VLE calculation as the column is modeled using a tray type in Unisim Design®, where distillation column on offshore FPSO (Floating Production Storage Offloading) are commonly equipped with packed type columns to prevent flooding caused by the continuous movement at sea. 2.2 Modeling of reclamation

2.2.1 Process configuration

Extension of the MEG regeneration unit to account for salts produced in the reservoir requires the addition of a reclamation process. The generation of salts is a typical phenomenon for subsea reservoirs, and feed A is assumed to have TDS as shown in Table 4, where maximum values from the whole period of operation are applied not to underestimate the impact of salts on equipment such as fouling and scaling. It is also assumed that there are only three major species for feed A, sodium, calcium and chloride, where the quantities are adjusted to maintain electroneutrality by calculating charge balances for the mixture. Sodium ion and chloride ion are taken as sodium chloride and calcium ion and chloride ion as calcium chloride. The absolute quantities of salts are calculated based on the produced water as shown in Table 4. It is not difficult to distinguish condensed water from produced water [25],but taking produced water as solvent for sizing purpose gives a better design margin by over-estimating the amount of salts to be dealt with. As the reclamation to be designed should be additionally introduced to the existing re-concentration, a full stream reclamation is considered here where both monovalent and divalent salts are removed. Monovalent salts are condensed at the bottom of a flash separator by vaporizing solvent where feed is mixed with hot recycled flow as a way of providing heat indirectly. The system configuration, for example, the location of outlet nozzle for liquid and solid stream, can be varied, which is case-dependent and feedspecific [30]. The reclamation process simulated in this study is based on configuration given in Figure 3, while there is also a possibility to improve product purity by adding a spinner to the conventional nozzle [30]. It is assumed that both monovalent and divalent salts are separated in a single flash separator so that the 10

ACCEPTED MANUSCRIPT

number of pieces of equipment can be reduced. However, if operating conditions favoring deposition of monovalent salts are different to those of divalent salts, one of them can cause malfunction due to accumulation of salts at any location in the liquid loop. This is an important issue especially when both monovalent and divalent salts removal are integrated as in this study. Therefore, operating conditions for the system in Figure 3 are settled based on the removal of monovalent salts, while precipitating agent is injected for removing divalent salts. Such method for removing divalent salts has been discussed and proven by various works including Fernandez’s [31]. In this study, it is also assumed that the entire recycled flow saturated with salts constantly removes salts through a solid removal equipment, i.e., a centrifuge. Saturated conditions for the recycled flow may not be achieved in the operation. For such cases, the degree of reheating for the recycling stream should be adjusted in order to ensure the saturated conditions. As the current study presents process design of MEG recovery units, process conditions required for recycling flow to be at the saturated conditions are thermodynamically determined. In addition to the design guidelines mentioned above, a pump is required to connect the reclamation to re-concentration. This is because the re-concentration is operated at atmospheric pressure while reclamation typically operates at vacuum pressures to prevent MEG degradation, where flash separator can also be accompanied with to prevent vapor flowing.

11

ACCEPTED MANUSCRIPT

2.2.2 Salt precipitation mechanism

Precipitating agent is added into the flash separator for the purpose of balancing alkalinity, thereby depositing calcium ions. Direct injection of carbonate is a straightforward method to balance the quantities of CO2 and Ca2+. While sodium carbonate was often used as a precipitating agent [32], sodium bicarbonate is selected in this study. This is because sodium bicarbonate is a weaker base than sodium carbonate; thus, it can produce a more modest amount of carbonate than sodium carbonate. Conversion of sodium bicarbonate into carbonate includes dissolution and equilibrium reactions as given below. NaHCO3(s) ⇆ Na+ + HCO3-

(3)

H2O + HCO3- ⇆ CO32- + H3O+

(4)

The other expected reactions involving salts are as follows: CaCl2(s) ⇆ Ca2+ + 2Cl-

(5)

NaCl(s) ⇆ Na+ + Cl-

(6)

CaCO3(s) ⇆ Ca2+ + CO32-

(7)

2H2O + CO2 ⇆ HCO3- + H3O+

(8)

Salt morphology is also important in that reaction thermodynamics are different. In particular, the calcium carbonate scale is polymorphous which takes the form among aragonite, vaterite, and calcite depending on the operating conditions, and such different structures influence the stability as well as solubility [33]. Since the operation of reclamation is assumed to be at steady state, CaCO3 in this study is in the most stable form as calcite.

12

ACCEPTED MANUSCRIPT

2.2.3 Operating conditions

The equilibrium states of flashing are calculated with Raoult’s law and Antoine equation as shown in eq 9 and eq 10 [34]. Colligative properties (eq 11) are also taken into account to reflect the influence of salts on equilibrium, where nonvolatile solutes decrease the vapor pressure because they do not participate in the equilibrium reaction [35]. yiP = xiPisat ln 𝑃

sat 𝑖

(9) 𝐵

(10)

= 𝐴 ‒ 𝑡+𝐶

P = xsolventP° (11) where yi is mole fraction of component i in vapor phase, xi is mole fraction of component i in liquid phase, Pisat is vapor pressure of pure component i, t is temperature in Celsius and A, B, and C are Antoine parameters, P is vapor pressure of solution, xsolvent is mole fraction of solvent, and P° is vapor pressure of pure solvent. To satisfy DOF (Degree of Freedom) analysis while applying eqs 9, 10, and 11, a range of potential operating pressures are screened as listed in Table 5 which also shows calculated equilibrium temperature at each pressure. However, as reclamation in MEG regeneration unit is generally operated under the vacuum to avoid thermal degradation of MEG, the maximum operating pressure is approximately 180 kPa which corresponds to the degradation temperature of MEG at around 163 °C [36]. The selection of operating pressure is based on engineering judgment as it brings a trade-off between capital cost and operating cost. For example, at low operating pressures, the vacuum pump for a vacuum system is expensive [37] but MEG is less liable to be thermally decomposed. As maintaining MEG quality is usually considered more important because of its reuse despite the capital cost, 35 kPa is selected as the design pressure for reclamation in this work. The operating temperature for the recycle heater is estimated such that equilibrium conditions in the flash separator are satisfied. Pressure drop in the recycle heater is assumed to be 20 kPa, recycle pump discharge pressure set at 200 kPa, and no pressure drop in the centrifuge.

13

ACCEPTED MANUSCRIPT

2.2.4 Process simulation

The modeling of reclamation is realized with the aid of commercial simulator, Aspen Plus® V8.4. ELECNRTL is applied which has been shown to accurately predict salt behavior including the effects of CO2 partial pressure, temperature, pressure, and NaCl concentration on CaCO3 solubility in a mixture of water, sodium, and calcium except for cases with high concentrations of NaCl [38]. The solubility of calcium carbonate is also updated within the simulator based on the work of Plummer et al. [39] to consider the effect of CO2 on CaCO3 as calcite. Separation prediction requires the relationship between liquid and solid in discharge stream of centrifuge along with operating conditions specified as in Section 2.2.3. Important parameters for designing centrifuges are ‘fractions of solids to solid outlet’ and ‘liquid load of solid outlet’ which are set to 0.9 and 0.1111, respectively, according to the industrial suggestion [40]. For the simulation results of the design case whose operating conditions are determined in the section 2.2.3, 639 kW and 489 kW are consumed for recycle heater and recycle pump each. Furthermore, a range of recycle heater temperatures at the same pressure are tested to analyze the effect of energy on salt separation as a sensitivity analysis, and the results are shown in Table 6. The results suggest that MEG recovery is slightly improved at the expense of heat energy while electricity consumption follows similar tendency to MEG recovery. It is because larger amount of recycle flows at lower temperature. It is also demonstrated that MEG is recovered as much as 99.42% for reclamation, while overall MEG recovery reaches 99.89% because feed B is not processed in the reclamation where most of MEG is removed. Thus, 99.42% of recovery is comparable to the industrial performance reported, such as 99.4% from CCR Technology Ltd. [41] and 99.5% from CAMERON [42]. However, if carryover which is considered as a significant source of MEG loss [43] is to be taken into account, MEG is to be lost as a larger quantity.

14

ACCEPTED MANUSCRIPT

2.3 Integrated modeling of reclamation and re-concentration

As described above, this study aims to design a reclamation unit in addition to existing re-concentration unit for controlling salts in the future. Two distinct process simulators are used for modeling each of these separations because of their different databases and models. However, a model framework should combine and connect re-concentration with reclamation for the holistic and system-wide investigation of the overall MEG regeneration unit to investigate economics in terms of both energy consumption and MEG loss. The integration is realized as shown in Figure 4, which schematically describes the linking between reclamation in Aspen Plus® and re-concentration in Unisim Design®. As an excel application, Aspen Plus Simulation Workbook (ASW) extracts product stream properties from Aspen Plus®, and it is connected to Matlab® which transfers the information to Unisim Design®. Matlab® with strategically programmed code connects Excel® and Unisim Design® based on active X.

15

ACCEPTED MANUSCRIPT

3. Case study

Seven cases are studied to demonstrate the applicability of the suggested design framework for the regeneration of MEG. These cases are classified based on the changes in hydrocarbon production at both low FWHT (Flowing Wellhead Temperature) and high FWHT. FWHT is a barometer for predicting wellbore temperature gradient [44] which is influenced by heat loss in the subsea system [45]; hence, it has an impact on operating conditions at offshore processing platform as well as the required amount of MEG injection to depress hydrate. Process streams are defined to reflect the various stages of well production with time, consequently considering different amounts of MEG injected. The hydrocarbons are assumed to be produced from Manifold A and Manifold B whose conditions such as production rate, temperature, and pressure changing over time are given in Table 7. It is typical that the production of resources increases until it reaches a maximum rate and then shows decline, thereby production modes are divided into early stage, maximum stage, and later stage of production for each FWHT. During the early stage of production, production in Manifold A is operated at HP (High Pressure) mode, including Case 1a for low FWHT and Case 1b for high FWHT. The maximum amounts of hydrocarbons are produced in the middle period when both Manifold A and Manifold B are in operation, which are characterized as Case 2a and Case 2b. Hydrocarbon resources are produced from Manifold B only at the later stage of the production which operates at LP (Low Pressure) mode, and it includes Case 3a for low FWHT and Cases 3b and 4b for high FWHT. As critical parameters for economics, energy consumption and MEG loss are analyzed in the case study. Energy consumption includes both heat duty and pump power, while MEG losses are quantified with unit price of $1,250/ton [46].

16

ACCEPTED MANUSCRIPT

3.1 Energy consumption

The calculated heat energy in simulation is given in Table 8. Heat energy related to the re-concentration accounts for typically 60 ~ 70% of overall heat energy in most of the cases which is attributed to heat energy utilized in the reboiler. For this reason, cases where there is a greater fraction of MEG in the feed require less heat energy because of relative ease of separating MEG from the water/MEG mixture. The difference of heat energy between low FWHT and high FWHT, which is more than 15%, can be explained by different MEG injection rates, due to different prediction of subsea temperature. The recycle pump in the reclamation unit accounts for almost 99% of the overall pump power as shown in Table 9. As the pump power depends on mass flow rate and generated head, it is proportional to the recycled flow rate for Cases 1 to 4. Pump power per unit feed also increases as the summation of salt mole fraction becomes lower. The slight difference in pump power per unit feed among Case 1b, 3b, and 4b is attributed to the relatively high concentration of heavy hydrocarbons in Case 3b and 4b, which have low heat of vaporization so that larger recycle flow is required.

3.2 MEG loss

Most of MEG loss occurs during reclamation, and the detailed simulation results of these losses are shown in Table 10. The main source of MEG loss comes from the centrifuge during which some liquid is discharged proportionally to the amount of salts removed. Therefore, MEG loss cost per unit feed in reclamation increases with the summation of mole fractions of MEG and water. In other words, it is expected to have more discharge of MEG from the centrifuge when the concentration of aqueous phase is high. On the other hand, MEG loss during re-concentration, in which most of MEG is lost as distillate, has shown no considerable difference across the different cases.

17

ACCEPTED MANUSCRIPT

4. Conclusions

Suppressing the hydrate formation in offshore system is of interest to maintain system integrity. Among the various solutions for preventing hydrates formation, injection followed by recovery of MEG as a hydrate inhibitor is considered effective and economically sound. In this research, a process modeling framework for MEG regeneration unit is proposed and tested. Re-concentration currently in operation shows generally good agreement between the simulation results and industrial data. On the other hand, reclamation is systematically designed including setting up a configuration, operating condition, etc. in case of salts production later, which is validated with industrial standard indexes. It is shown that the salts can be removed without possible MEG degradation. Unlike the previous researches which had assumed the MEG recovery, the recovery of MEG is calculated in this study. The 99.42% of MEG recovery in the reclamation is predicted through process simulation, which is comparable to 99.5% of CAMERON technology and 99.4% of CCR technology. This implies that the process considered is sustainable and environment-friendly due to minimal MEG loss to the environment. From the viewpoint of energy consumption, heat consumed in the reboiler of the distillation column is 2.3 times more than heat required for a recycle heater in the reclamation whereas most of the electricity is required in the reclamation for pumping the recycle stream back to the separator. Furthermore, novel strategy is applied to link the modeling of reclamation and re-concentration to assess the efficiency in a holistic manner, thereby making a more versatile platform. A case study has been investigated considering various offshore conditions for MEG regeneration and recovery in practices to demonstrate the applicability of the developed modeling framework. Different stages of production leading to seven production cases are analyzed in terms of energy consumption and MEG losses, from which MEG and salts concentration are found to be critical parameters for heat energy and power, respectively, while aqueous phase (MEG) concentration is important for calculating MEG loss cost per feed. In other words, power consumption and MEG loss are influenced by the design of a reclamation unit while the design of re-concentration process plays a critical role in heat consumption. As shown in the case study, specifically, at least 60% of overall heat is consumed to separate MEG from water, whereas electricity is mainly required for recycling hot stream in the reclamation. The process design framework proposed here is expected to provide better conceptual understanding of the MEG regeneration unit and 18

ACCEPTED MANUSCRIPT

design guidelines leading to cost-effective recovery of MEG in offshore oil and gas processing in a sustainable manner. Acknowledgement

This research was respectfully supported by Engineering Development Research Center (EDRC) funded by the Ministry of Trade, Industry & Energy (MOTIE) (No. N0000990) and by the World Class 300 Project (No. S2305678) of the Small and Medium Business Administration (Korea).

19

ACCEPTED MANUSCRIPT

Nomenclature ASW DOF EOS FPSO FWHT HP LCC LP MEG NPC NRTL PFD SS THI RU TDS TEG THI VLE

Aspen Plus Simulation Workbook Degree of Freedom Equation of State Floating Production Storage Offloading Flowing Wellhead Temperature High Pressure Life Cycle Cost Low Pressure Monoethylene Glycol Net Present Cost Non-Random Two Liquid Process Flow Diagram Supersonic Separator Thermodynamic Hydrate Inhibitor Recovery Unit Total Dissolved Solids Triethylene Glycol Thermodynamic Hydrate Inhibitor Vapor-Liquid Equilibrium

20

ACCEPTED MANUSCRIPT

References [1] U.S. Energy Information Administration. International Energy Outlook 2017. [place unknown]: U.S. Energy Information Administration; 2017 Sep. 76p. Report No.: DOE/EIA-0484(2017). [2] World Ocean Review. World ocean review 3, marine resources- opportunities and risks. Hamburg (DE): maribus gGmbH; 2014. [3] Sloan ED. A changing hydrate paradigm—from apprehension to avoidance to risk management. Fluid Phase Equilib. 2005 Feb;228–229:67-74. [4] Veil JA, Puder MG, Elcock D, Redweik RJ. A white paper describing produced water from production of crude oil, natural gas, and coal bed methane. Lemont (IL): Argonne National Lab.; 2004. 87p. Report No.: ANL/EA/RP-112631. [5] Brustad S, Løken KP, Waalmann JG. Hydrate prevention using MEG instead of MeOH: impact of experience from major Norwegian developments on technology selection for injection and recovery of MEG. In: Society of Petroleum Engineers, editors. 2005 Offshore Technology Conference; 2005 May 25; Texas, U.S.A. [6] Kim CS. Feasibility study and economic evaluation on application of MEG regeneration unit for oil FPSO [master's thesis]. Daejeon (KR): Korea Advanced Institute of Science and Technology; 2014. [7] Ekawati D. Effect of temperature, bicarbonate, and MEG concentration on pre-corroded carbon steels [master's thesis]. Stavanger (NO): University of Stavanger; 2011. [8] Addicks L, inventor; Aker Process Stystema As, assignee. Method for regeneration and reclamation of mono ethylene glycol using a vacuum slip stream. World patent WO 2010080038A1. 2010 Jul 15. Norwegian. [9] Contreras MAV, Foucart N. The MEG (Monoethylene) injection gas dehydration process evaluation for the Margarita field development. In: Society of Petroleum Engineers, editors. 2007 SPE Latin American and Caribbean Petroleum Engineering Conference; 2007 Apr 15-18; Buenos Aires, Argentina. [10] Alnili F, Barifcani A. Simulation study of sweetening and dehydration of natural gas stream using 21

ACCEPTED MANUSCRIPT

MEG solutions. Can J Chem Eng. 2018 Feb 12;9999:1-7. [11] Kim H, Yoo W, Lim Y, Seo Y. Economic evaluation of MEG injection and regeneration process for oil FPSO. J Petrol Sci Eng. 2018 May;164:417-426. [12] Kim H, Lim Y, Seo Y, Ko M. Life cycle cost analysis of MEG regeneration process incorporating a modified slip stream concept. Chem Eng Sci. 2017 Jul 20;166:181-192. [13] Teixeira AM, Arinelli LdO, Medeiros JLd, Araujo OdQF. Exergy analysis of monoethylene glycol recovery process for hydrate inhibition in offshore natural gas fields. J Nat Gas Sci Eng. 2016 Sep;35:798813. [14] Park SY. Salt management strategy on the design of MEG (Mono Ethylene Glycol) regeneration process for LNG FPSO [master's thesis]. Daejoen (KR): Korea Advanced Institute of Science and Technology; 2014. [15] Teixeira AM, Arinelli LdO, Medeiros JLd, Araujo OdQF, Recovery of thermodynamic hydrate inhibitors methanol, ethanol and MEG with supersonic separators in offshore natural gas processing. J Nat Gas Sci Eng. 2018 Apr;52:166-186. [16] Bikkina C, Radhakrishnan N, Jaiswal S, Harrington RM, Charlesworth M. Development of MEG regeneration unit compatible corrosion inhibitor for wet gas systems. In: Society of Petroleum Engineers, editors. SPE Asia Pacific oil and gas conference and exhibition; 2012 Oct 22-24; Perth, Australia. [17] AlHarroni K, Gubner R, Iglauer S, Pack D, Barifcani A. Influence of regenerated monoethylene glycol on natural gas hydrate formation. Energ Fuel. 2017 Oct 17;31:12914-12931. [18] Alef K, Smith C, Iglauer S, Gubner R, Barifcani A. The effect of regenerated MEG on hydrate inhibition performance over multiple regeneration cycles. Fuel. 2018 Jun 15;222:638-647. [19] Burgass R, Chapoy A, Li X. Gas hydrate equilibria in the presence of monoethylene glycol, sodium chloride and sodium bromide at pressures up to 150 MPa. J Chem Thermodyn. 2018 Mar;118:193-197. [20] Wang P, Kosinski JJ, Anderko A, Springer RD, Lencka MM, Liu J. Ethylene glycol and its mixtures with water and electrolytes: thermodynamic and transport properties. Ind Eng Chem Res. 2013 Oct 22

ACCEPTED MANUSCRIPT

14;52(45):15968-15987. [21] Sandengen K. Prediction of mineral scale formation in wet gas condensate pipelines and in MEG (Mono Ethylene Glycol) regeneration plants [dissertation]. Trondheim (NO): Norwegian University of Science and Technology; 2006. [22] AspenTech. EAP201 Aspen Plus: physical properties for process engineers. Bedford (UK): Aspen Technology; 2014. [23] Carroll J. Natural gas hydrates: A guide for engineers. Oxford (UK): Gulf Professional Publishing; 2014. [24] Sloan ED, Carolyn AK. Clathrate hydrate of natural gases. New York: CRC Press; 2008. [25] Gas Processors Suppliers Association. GPSA Engineering data book. Tulsa (OK): Gas Processors Suppliers Association; 2012. [26] Maekawa T. Equilibrium conditions of propane hydrates in aqueous solution of alcohols, glycols, and glycerol. J Chem Eng Data. 2008 Oct 31;53:2838-2843. [27] Herath D, Khan F, Yang M. Risk-based winterization to prevent hydrate formation in northern harsh environment. Ocean Eng. 2016 Jun 1;119:208-216. [28] Katz DL. Prediction of conditions for hydrate formation in natural gases. Trans AIME. 1945;160(1):140-149. [29] Østergaard KK, Masoudi R, Tohidi B, Danesh A, Todd AC. A general correlation for predicting the suppression of hydrate dissociation temperature in the presence of thermodynamic inhibitors. J Pet Sci Eng. 2005 Jul 30;48(1–2):70-80. [30] Perry RH, Green DW, Maloney JO. Perry's chemical engineers' handbook. New York: McGraw-Hill; 1999. [31] Fernandez LEC, inventor; Cameron International Corporation, assignee. Process scheme to improve divalent metal salts removal from mono ethylene glycol (MEG). United States patent US 2013/0118989. 2013 May 16. 23

ACCEPTED MANUSCRIPT

[32] Baraka-Lokmane S, Hurtevent C, Seiersten M, Flaten E, Farrell M, Bradshaw O, et al. Technical challenges and solutions in a closed loop MEG regeneration system for gas field offshore. Southampton (UK): WIT Press; 2013. [33] Krossholm CK. Crystal growth kinetics of calcium carbonate particles in natural gas production [master's thesis]. Trondheim (NO): Norwegian University of Science and Technology; 2012. [34] Kassa B. Glycol injection and processing [Internet]. 2008 Oct 16 [cited 2018 Apr 19]. [35] Hadzija BW. A lecture on colligative properties in an undergraduate curriculum. Am J Pharm Educ. 1995;59(2):191-195. [36] Christensen DL. Gas dehydration: thermodynamic simulation of water/glycol mixture [master's thesis]. Aalborg (DK): Aalborg University Esbjerg; 2009. [37] Loh HP (National Energy Technology Laboratory, Pittsburg, PA), Lyons J, White CW (EG&G Technical Services, Inc., Morgantown, WV). Process equipment cost estimation. Final report. U.S.: US Department of Energy: 2002. 78p. Report No.: DOE/NETL-2002/1169. [38] Coto B, Martos C, Pena JL, Rodríguez R, Pastor G. Effects in the solubility of CaCO3: experimental study and model description. Fluid Phase Equilib. 2012 Jun 25;324:1-7. [39] Plummer LN, Busenberg E. The solubilities of calcite, aragonite and vaterite in CO2-H2O solutions between 0 and 90°C, and an evaluation of the aqueous model for the system CaCO3-CO2-H2O. Geochim Cosmochim Acta. 1982;46(6):1011-1040. [40] Jariwala A. Discussing latest innovations in gas pre-treatment for FLNG. 7th Annual LNG TECH GLOBAL SUMMIT 2012; 2012 Dec 3-5; Rotterdam, Netherlands. [41] Trofimuk T, Ayres S, Esquier J. CCR MEG reclaiming technology: from mobile to the largest reclaiming unit in the world. In: Brazilian Petroleum, Gas and Biofuels Institute - IBP, editors. Rio Oil & Gas Expo and Conference 2014; 2014 Sep 15-18; Rio de Janeiro, Brazil. [42] CAMERON. PUREMEG: MEG reclamation and regeneration technology . Houston (TX): 24

ACCEPTED MANUSCRIPT

CAMERON; 2015. [43] Society of Petroleum Engieers. MEG regeneration technical meeting. Victoria & Tasmania section; 2009 Jun; Port Campbell, Australia. [44] Mohammed IY. Modeling and simulation of multiphase system in production well using WellFloR software. Int J Curr Eng Tech. 2014 Dec;4(6):4057-4062. [45] Okuno M, Makishma K. Temperature insulation design for deepwater gas well test, Abadi Field. In: Indonesian Petroleum Association, editors. 33rd Annual Convention & Exhibition; 2009 May 5-7; Jakarta, Indonesia. [46] Wang J. Asia MEG hits 44-month high, may extend gains on tight supply [Internet]. ICIS News; 2011 [cited Apr 19].

25

ACCEPTED MANUSCRIPT

List of Figures Figure 1. Schematic diagram of MEG recovery and re-injection Figure 2. Re-concentration in MEG regeneration unit Figure 3. Flowsheet of reclamation process in MEG regeneration unit Figure 4. Methodology for linking Aspen Plus® and Unisim Design®

26

ACCEPTED MANUSCRIPT

Figure 1. Schematic diagram of MEG recovery and re-injection

27

ACCEPTED MANUSCRIPT

Figure 2. Re-concentration in MEG regeneration unit

28

ACCEPTED MANUSCRIPT

Figure 3. Flowsheet of reclamation process in MEG regeneration unit

29

ACCEPTED MANUSCRIPT

Figure 4. Methodology for linking Aspen Plus® and Unisim Design®

30

ACCEPTED MANUSCRIPT

List of Tables Table 1. Feed conditions of MEG regeneration unit Table 2. Operating conditions for heat exchangers and pressure changers Table 3. Comparison of simulation results and industrial process data Table 4. The composition of formation water Table 5. Relationship between operating temperature and pressure for the reclamation Table 6. Modeling results for different operating temperatures at 35 kPa Table 7. Well conditions for the case study Table 8. Heating duties for the case study Table 9. Pump power for the case study Table 10. Costs of MEG losses for the case study

31

ACCEPTED MANUSCRIPT

Table 1. Feed conditions of MEG regeneration unit

Temperature (°C) Pressure (barg) Mass flow (kg/h) Heat flow (kW)

Feed A 14 7 2393 -7294

Feed B -13.8 7 4225 -11,349

Components mole fractions CO2 0.0001 CH4 0.0043 n-C6H14 0.0001 C6H12 0.0000 C6H6 0.0000 n-C7H16 0.0003 C7H14 0.0001 n-C8H18 0.0014 C8H10 0.0003 n-C9H20 0.0037 n-C10H22 0.0039 n-C11H24 0.0048 n-C12H26 0.0065 n-C13H28 0.0053 n-C14H30 0.0040 n-C15H32 0.0027 n-C16H34 0.0014 n-C17H36 0.0014 n-C18H38 0.0014 H2O 0.8734 MEG 0.0848

0.0009 0.0087 0.0001 0.0001 0.0001 0.0004 0.0001 0.0018 0.0004 0.0042 0.0037 0.0031 0.0024 0.0008 0.0002 0.0001 0.0000 0.0000 0.0000 0.6125 0.3604

32

ACCEPTED MANUSCRIPT

Table 2. Operating conditions for heat exchangers and pressure changers

E-1 E-2 E-3 Valve 1 Valve 2 Valve 3 Valve 4 P-1

Outlet temperature (°C)

Pressure drop or difference (kPa)

25 70 90

20 50 10 300 100 183 761 772

33

ACCEPTED MANUSCRIPT

Table 3. Comparison of simulation results and industrial process data

S1 Temperature (°C) S1 Mass flow (kg/h) S1 MEG mole fraction S2 Temperature (°C) S2 Mass flow (kg/h) S2 MEG mole fraction S3 Temperature (°C) S3 Mass flow (kg/h) S3 MEG mole fraction Preheater (E1) duty (kW) Preheater (E2) duty (kW) Reboiler duty (kW) Condenser duty (kW) Pump power (kW)

Simulation 25.01 5742 0.2510 109.2 2948 0.0001 121.9 3843 0.5372 161 245 2115 1887 0.88

34

Process data 25.09 5742 0.2513 109.3 2945 0.0001 135.6 3847 0.5373 156 243 2428 1904 1.1

ACCEPTED MANUSCRIPT

Table 4. The composition of formation water Concentration (mg/L) Sodium Potassium Calcium Magnesium Strontium Chloride Sulphate Bicarbonate

12,020 185 4900 440 580 24,030 860 3310

Amount of salts estimated Sodium chloride 38 kg/h Calcium chloride 17 kg/h

35

ACCEPTED MANUSCRIPT

Table 5. Relationship between operating temperature and pressure for the reclamation Pressure (kPa)

15

35

55

75

95

…

180

Temperature (°C)

104.2

122

132.6

140.4

146.6

…

164.9

36

ACCEPTED MANUSCRIPT

Table 6. Modeling results for different operating temperatures at 35 kPa Heater outlet temperature (°C) 113 116 119 122 130 140

MEG recovery (%) 99.43 99.42 99.41 99.42 99.60 99.74

Recycle heater duty (kW) 244 344 469 639 1075 1165

37

Recycle pump power (kW) 806 731 632 489 93 27

Table 7. Well conditions for the case study Manifold A

Case 1a Case 2a Case 3a

Mole composition (%) CO2 0.23 H2O 3.64 H2S 0.0002 CH4 95.81

Manifold B

Temperature (°C)

Pressure (bar)

Flow rate (MMSCFD)

13.88

82.50

99.80

14.26

82.50

49.12

Mole composition (%) CO2 0.23 H2O 3.55 H2S 0.0001 CH4 95.90

Temperature (°C)

Pressure (bar)

Flow rate (MMSCFD)

Amount of salts (kg/h) 144.0

13.63

82.52

108.00

223.0

14.56

41.81

20.00

59.0

(a) Low FWHT Manifold A Mole composition (%) Case 1b Case 2b Case 3b Case 4b

CO2 0.23 H2O 3.64 H2S 0.0002 CH4 95.81

Manifold B

Temperature (°C)

Pressure (bar)

Flow rate (MMSCFD)

14.02

82.50

99.80

14.33

82.50

40.56

Mole composition (%) CO2 0.23 H2O 3.55 H2S 0.0001 CH4 95.90

(b) High FWHT

38

Temperature (°C)

Pressure (bar)

Flow rate (MMSCFD)

Amount of Salts (kg/h) 144.0

13.69

82.50

105.05

207.0

12.65

21.00

97.12

137.0

12.18

21.00

105.74

149.0

ACCEPTED MANUSCRIPT

Table 8. Heating duties for the case study

Overall heat duty [W/(kg/h)feed] Overall heat duty [W/(kg/h)MEG] Re-concentration heat duty [W/(kg/h)feed] Re-concentration heat duty [W/(kg/h)MEG] Mass fraction of MEG in feed

Overall heat duty [W/(kg/h)feed] Overall heat duty [W/(kg/h)MEG] Re-concentration heat duty [W/(kg/h)feed] Re-concentration heat duty [W/(kg/h)MEG] Mass fraction of MEG in feed

Case 1a

Case 2a

Case 3a

888.4

933.5

1003.5

2218.0

3130.5

3658.5

531.6

642.9

667.5

1327.1

2155.9

2433.6

0.4006 (a) Low FWHT

0.2982

0.2743

Case 1b

Case 2b

Case 3b

Case 4b

1200.0

954.4

1199.1

1218.2

9357.0

4198.3

10,777.8

11,629.6

829.1

720.7

846.0

852.8

6464.8

3170.4

7603.8

8141.3

0.1282 0.2273 (b) High FWHT

0.1113

0.1047

39

ACCEPTED MANUSCRIPT

Table 9. Pump power for the case study

Overall pump power [W/(kg/h)feed] Overall pump power [W/(kg/h)feed] Recycle pump power in reclamation [W/(kg/h)feed] Recycle pump power in reclamation [W/(kg/h)MEG] Sum of salt mole fraction in feed

Overall pump power [W/(kg/h)feed] Overall pump power [W/(kg/h)feed] Recycle pump power in reclamation [W/(kg/h)feed] Recycle pump power in reclamation [W/(kg/h)MEG] Sum of salt mole fraction in feed

Case 1a

Case 2a

Case 3a

154.5

248.5

203.0

385.7

833.5

740.1

153.7

247.8

202.0

383.8

830.8

736.5

0.0131

0.0134

0.0174 (a) Low FWHT Case 1b

Case 2b

Case 3b

Case 4b

200.2

311.6

230.3

220.1

1560.9

1370.7

2069.8

2101.0

199.2

310.8

229.3

219.1

1553.3

1367.2

2060.8

2091.4

0.0153 0.0141 (b) High FWHT

0.01546

0.01552

40

ACCEPTED MANUSCRIPT

Table 10. Costs of MEG losses for the case study

Overall MEG loss cost [$/(kg/h)feed] Overall MEG loss cost [$/(kg/h)MEG] MEG loss cost in reclamation [$/(kg/h)feed] MEG loss cost in reclamation [$/(kg/h)MEG] MEG loss cost in re-concentration [$/(kg/h)feed] MEG loss cost in re-concentration [$/(kg/h)MEG] Mole fraction of MEG+H2O in feed

Overall MEG loss cost [$/(kg/h)feed] Overall MEG loss cost [$/(kg/h)MEG] MEG loss cost in reclamation [$/(kg/h)feed] MEG loss cost in reclamation [$/(kg/h)MEG] MEG loss cost in re-concentration [$/(kg/h)feed] MEG loss cost in re-concentration [$/(kg/h)MEG] Mole fraction of MEG+H2O in feed

Case 1a

Case 2a

Case 3a

23.1

26.6

25.7

57.6

89.1

93.8

21.5

24.8

23.5

53.6

83.0

85.6

1.6

1.8

2.3

4.1

6.0

8.2

0.9798

0.9792

0.9766 (a) Low FWHT Case 1b

Case 2b

Case 3b

Case 4b

28.8

28.5

26.2

25.8

224.5

125.4

235.5

246.2

26.4

26.8

24

23.2

206.2

118.0

215.4

221.2

2.3

1.7

2.2

2.6

18.3

7.4

20.1

25.0

0.9758

0.9756

0.9764 0.9781 (b) High FWHT

41