A systematic selection and decision matrix for energy-efficient intensified distillation technologies

3.1 Overview of intensified distillation technologies

Intensified distillation technologies have the potential to enhance the process performance in terms of energy requirement and capital cost reduction. However, each technology has its advantages and disadvantages. Table 1 lists seven intensified distillation techniques considered in this study, including intensification targets, features, and the summarized main key advantages and disadvantages of each technology [3], 28]. Some technologies (e.g., cyclic distillation, HiGee) might be less well known than others, but they are all applied at industrial scale (TRL9). The readers are guided to relevant review papers for more information about cyclic distillation [29] or HiGee distillation [30], 31]. Binary and ternary separations are defined based on the number of products instead of the number of components.

Table 1:

Features and performances of intensified distillation technologies.

Type	Key feature	Remarks on design and performance
Ternary		Dividing wall column (DWC) + Low re-mixing effects + Highly purified side products + CAPEX reduction (>25 %) + Reduced energy demands (20–40 %)   - Fewer degrees of freedom on pressure   - Complex column structure and control
Ternary		Thermally coupled distillation (TCD) + Low re-mixing effects + Highly purified side products + Reduced energy demands   - Operation complexity   - Complex control
Binary		Heat pumps assisted distillation (HPAD) + Upgrading energy level + Reduced energy demands (40–80 %)   - High CapEx of large heat pumps   - High CapEx of large temperature lifts
Binary		Heat integrated distillation columns (HIDiC) + High energy savings (60–90 %)   - Vacuum column not applicable   - Operation and design complexity   - Complex integrated process and control
Binary		Cyclic distillation (CyDist) + No liquid remixing or flooding on the stage + High product purities + High mass transfer efficiency + Reduced energy demands (∼35 %) + Adjustable residence time   - Operation complexity   - Vacuum operation not applicable   - Complex column structure and control
Binary/Ternary		Membrane assisted distillation (MAD) + Size reduction + Reduced energy demands + Facilitated separation process + Enhanced product purities   - Higher equipment cost   - Membrane limitations (short lifetime)
Binary		HiGee + High mass transfer efficiency + Reduced energy demands + Size reduction (6–8 times)   - Difficult to scale up (use numbering up)   - Rotational device needing more energy   - Aversion to long-periodic operation

3.2 Distillation sequence efficiency

Distillation sequence efficiency (DSE) is a parameter used to evaluate the energy efficiency of each alternative distillation sequence [25]. It enables advantages derived from process intensification for a given feed composition to be quantified, and does not rely on heuristics. Equations (1) and (2) illustrate the calculation of the DSE value.

where η denotes Carnot efficiency, T _B is the temperature (Kelvin) of the bottom product (reboiler), T _D is the temperature (in K) of the distillate product (condenser). x _i is the liquid composition of component i, C represents distillation columns.

The higher the DSE number, the higher the efficiency of the distillation sequence; while the lower the DSE number, the more advisable process intensification technologies are. Therefore it is a useful tool to gain an idea of the energy efficiency of the process, and whether there is intensification potential for the process.

To evaluate the reliability of this criteria, we calculated the DSE for the optimization results from our previous work [26] about the zeotropic multicomponent C2–C5+ separation using simple columns, and apply the DSE calculation on these 14 possible sequences. As shown in Figure 1, the optimization results show that sequences S1, S2, S4, S6, S7 are the most economically attractive sequence family in terms of total operating cost, followed by S5 and S3 - the direct sequences are more energy efficient in this case. Based on the DSE numbers calculated for the 14 sequences, the above-mentioned sequences are all pre-selected within the first sequence family (S1, S2, S3, S4, S5, S6, S7), with a clear gap compared to the rest of the sequences in terms of DSE numbers. Note that the fast calculation is a simplified screening criteria assuming that the products are at 100 % purity and recovery, and the change of operating conditions for example column pressures are not considered, yet it requires only the feed composition and their normal boiling points which are very simple to use. Therefore, helping reduce the second half of the design or search space (S8–S14) safely by calculating the DSE numbers dramatically saves on design and computational input, and showing that the DSE numbers is a reliable pre-selection tool.

Figure 1:

DSE evaluation based on the optimization results (a) Li et al. [26], and (b) DSE calculation.

3.3 High-level questions and potential solutions

Before evaluating process intensification possibilities, the limitations of the processes need to be identified. In addition to the restrictions of phase equilibrium (e.g., low relative volatility) or mass transfer (high viscosity), six high-level questions are made to identify the special components in the separation which might make the separation difficult, as shown in Figure 2, and six corresponding solutions are provided in order to overcome the separation difficulties. This step is performed prior to using the process intensification (PI) matrix.

1. Q1: Are there any components at risk of solidification (e.g., freezing) during the separation? If so, high pressure distillation column is needed in order to increase the temperature of the column to avoid the solidification. In this case crystallization is also a good choice for this separation as compared to distillation if the solidification temperature is higher than the standard room temperature, as no cooling will be needed.

2. Q2: Are there any non-condensable components? Condensing (near) non-condensable components results in low condenser temperature, which may require refrigerant as the cooling utility. Partial condensers could be used to enable the non-condensable components to be removed in the vapor phase without being condensed. Non-condensable gases can be separated by membrane separation.

3. Q3: Are there any thermo-sensitive components, e.g., thermal denaturation, polymerizing or decomposing? Vacuum column is suggested for the separation of thermo-sensitive components in order to lower the temperature of the column. Alternatively, two-column configurations can be applied. The feed is fed into a normal pressure column to separate the non-thermo-sensitive component, and the thermos-sensitive component is fed into the second column which in this case is a smaller vacuum column, and the second distillate can be recycled back to the feed of the first column.

4. Q4: Is there a large amount of light component in the feed? Evaporation of a large amount of the light component results in high reboiler duties in the column, therefore, separating the light component by membranes before feeding the mixture into the column can reduce the energy requirement. Evaporation can also be used to replace distillation when there is a large amount of light component in the feed.

5. Q5: Is there a large amount of heavy component in the feed? Similarly, condensing a large amount of heavy component results in high condenser duties of the column. (Steam) stripping, adsorption or preconcentration columns can be used in this case.

6. Q6: Are there any azeotropic mixtures in the feed? Azeotropes need to be identified before process design or intensification. If it is a heterogenous azeotrope, a heterogeneous azeotropic distillation configuration can be applied. If it is a homogeneous azeotrope, membrane separation can be considered. If the azeotrope is pressure sensitive, pressure swing distillation can be used to separate the azeotrope effectively. By adding additional solvent, extractive distillation or azeotropic distillation can be used for separating homogeneous azeotropes, and additional solvent recovery is needed.

7. Q7: The number of outlet streams need to be identified beforehand. DWC, thermal coupled distillation columns, and distillation sequences are typically used for ternary or multicomponent mixture separations. Whereas other technologies (e.g., cyclic distillation, HiGee) are more commonly applied to binary separations.

Figure 2:

High-level questions and potential solutions.

3.4 Process intensification matrix

After identifying all special components by going through the high-level questions, a PI matrix has been developed to find out the intensification potentials of the separation system, and to match different intensified distillation technologies based on the evaluation criteria defined in this study, as shown in Figure 3. The columns represent the intensified distillation technologies, and the rows list the criteria relevant to phase equilibrium and mass transfer limitation of a separation task:

1. Phase equilibrium limitation identification: separations are categorized as azeotropic mixture and very close boiling point mixture separation (∆Tb≤10 °C); close boiling point mixture separation (10 °C<∆Tb<20 °C); and (near)ideal mixture separation (∆Tb≥20 °C) based on the normal boiling point difference (∆Tb) and relative volatilities (RV). High minimum reflux ratio, high heat of vaporization, high recovery or high purity products can also lead to high energy requirements even for (near-)ideal mixture separations.

2. Mass transfer limitation identification: Mass transfer could be limited by high viscosity (dimensionless correlating to Reynolds number), which causes difficulties by creating turbulence and achieving high gas/liquid interface; low liquid phase diffusion coefficient leads to inefficient vapor liquid mass transfer.

Figure 3:

Process intensification matrix for advanced distillation technologies.

In addition to the above mentioned criteria, vacuum operation is also an important criterion to be considered, as for example, it is very widely used in the separation of biochemicals.

These criteria are presented in an intensification matrix for advanced distillation technologies. The check mark indicates the specific technology is recommended when meeting the criteria; the exclamation mark represents the technologies which can be considered, but furthers are needed; while a cross mark represents technologies not suitable for use, according to that specific criterion. No mark indicates that the criteria is not relevant to the decision of the specific technology. Taking the first column as an example, if ∆Tb≤10 °C, conventional column could be used to achieve the separation requirement, however, high reflux ratio and energy requirements might result. HPAD and HIDiC are recommended, because the temperature lift between the top and the bottom of the column in this case will be small, cyclic distillation is not applicable in this case because the pressure drop will be too large if cyclic distillation is used for very difficult separations (∆Tb≤10 °C). All the marks for the matrix are pre-filled and for a separation task the users need to go through the intensified technologies (row by row) to find the right recommendations. Taking the first row as an example, HPAD is highly recommended if ∆Tb is lower than 10 °C or is between 10 °C and 20 °C, or if there is a component with high heat of vaporization to be separated. If ∆Tb is higher than 20 °C, further of the coefficient of performance (COP) of heat pumps is needed for HPAD. For low relative volatilities (between 1.01 and 1.15), the mixture could have close boiling points (e.g., 50 °C and 55 °C) when the temperature is not very high, and HPAD is not applicable in this case, or there might be still over 20 °C difference in boiling points (e.g., 200 °C and 220 °C), which HPAD is applicable. So, further checks are needed if the mixture has low relative volatility. For vacuum operation conditions, different heat pump configurations can be considered [3]. The rest of the criteria are not relevant to the decision making of HPAD.

With this intensification matrix, both the relevance (i.e., whether the criteria have an influence on the selection of a specific technique), and recommendations of the PI techniques are suggested.

3.5 Description of the overall approach

The overall approach of the decision making procedure is illustrated in Figure 4. For binary separations, high level questions are considered first in order to find the special components which may cause high energy requirements of the separation. Three sets of data are first extracted from the separation task, including the basis of the tasks, thermodynamics data and fluid dynamics data. The basic data shows the feed composition, key components and separation requirements; thermodynamic data includes the vapor pressure of each component, relative volatilities, heat of vaporization and the pressure sensitivity analysis (via y-x diagram) for the separation mixture; and fluid dynamics includes the liquid viscosity, diffusion coefficient in liquids, and the estimated vapor flow rate if needed. As discussed before, there are seven high-level questions to go through. If the answer to any of the questions is YES, the suggested separation technologies are provided accordingly, until the answer to all of the questions is NO, then go to the next step to the PI matrix.

Figure 4:

Fast-screening approach for the intensified distillation processes.

For ternary and multicomponent mixture separations, DSE evaluation is carried out first for the pre-screening of the distillation sequence alternatives. The intermediate results (i.e., η values) provide additional useful information. For instance, the Carnot efficiency of each column indicates when a heat pump or enhanced distillation would be useful. The recovery efficiency for each particular compound allows an easy comparison between alternatives, considering variations in the feed composition. After the DSE evaluation, the procedure is then the same as for binary separation. Finally, the recommended PI technologies are listed and further design and intensification can be carried out based on the recommendations.

Overall, this methodology consists of analyzing the given separation task in terms of selection criteria (e.g., volatile difference between key components, the type of separation tasks, feed and product flow, product specifications at different target purity levels, operating pressure, reboiler duty and its temperature level, reflux ratio, heat of vaporization). Subsequently, the methodology relates the selection criteria to process bottlenecks and promising intensified distillation techniques, through a connection matrix, to effectively address the identified bottlenecks. Finally, potential solutions are compared against the task specifications.

4.1 Binary zeotropic mixtures separation – methanol/water

Methanol-water separation is a large scale separation with the capacity of 1,100 ktpy feed. Feed consists of 69.81 mol% methanol and 30.19 mol% water [32], normal boiling point (N.B.P.) difference of 35.5 °C (N.B.P water 100 °C, methanol 64.5 °C), relative volatility (RV) is 2.98, heat of vaporization of methanol 1,273.4 kJ/kg and water 2,265.6 kJ/kg. The desired methanol product purity is 99.99 mol%, with a 99.98 % recovery, requiring a minimum reflux ratio (RRmin) of 0.87 (under atmospheric pressure).

4.2 Binary azeotropic mixtures separation – (1) ethanol/water, (2) acetonitrile/water

The ethanol/water separation is needed when producing food grade alcohol or producing hydrated ethyl alcohol fuel. For example, when producing hydrated ethyl alcohol fuel, the capacity of the design is 700,000 l/day of hydrated ethyl alcohol fuel, with a minimum top concentration of 92.6 wt% ethanol from an initial ethanol/water mixture of 8 wt% ethanol. Binary azeotrope (95.63 wt % ethanol) can also be formed between ethanol and water. The normal boiling point difference of ethanol and water is 22 °C. Heat of vaporization of ethanol 851.1 kJ/kg and water 2,265.6 kJ/kg.

Acetonitrile (ACN) recovery from wastewater is an important process, as ACN is an essential solvent in the pharmaceutical industry with an excellent ability to dissolve a wide range of chemicals and its strong C–C bonding makes it difficult to hydrolyze for use as a reaction solvent. The widely investigated waste ACN effluent contains 500 kmol/h feed, with the composition of 20 mol% ACN and 80 mol% water. The product specifications of ACN and water are set at 99.9 mol%. Acetonitrile and water forms an azeotropic mixture having a boiling point of 76.5 °C and containing about 82.2 wt % ACN at atmospheric pressure.

4.3 Ternary zeotropic mixtures separation – benzene (B)/toluene (T)/xylene (X)

For the near ideal mixture separation of BTX, the feed consists of 33.3 mol% benzene, 33.4 mol% toluene and 33.3 mol% m-xylene [33], ΔTb = 59.0 °C (N.B.P.: benzene 80.1 °C; toluene 110.7 °C and xylene 138.4 °C), RV_B/T=2.39, RV_T/X=2.19. Heat of vaporization: benzene 395.9 kJ/kg; toluene 365.1 kJ/kg; xylene 347.0 kJ/kg. The purities of BTX products are 99.0 mol%, and the recoveries are 99.8 %, 98.0 %, and 99.1 %, respectively. The RRmin of B/TX separation is ∼1.90, and 1.92 for T/X separation. The viscosity values (expressed in cP at 20 °C) are 0.652, 0.590 and 0.620, respectively.

5.1 Binary zeotropic mixtures separation – methanol/water

As indicated in Figure 4, this binary separation goes to the pre-question step. As for methanol water separation, there is no risk of solidification, non-condensable components or thermosensitive components, the ratio of the amount of light component and heavy component in this case is not large, and no azeotropes are formed. The next step is to go to the PI matrix. The boiling point difference is within the range of 20–60 °C, conventional distillation column is applicable, the application of HPAD and HIDiC requires a further of the COP, while cyclic distillation and HiGee distillation are recommended. Water has high heat of vaporization, therefore HPAD, HIDiC and membrane assisted distillation are recommended. Methanol products require high purity, so cyclic distillation is recommended.

These steps are summarized in Figure 5. In summary, conventional distillation columns can achieve the separation target of this task, yet the energy requirement is not optimized. As for the intensification techniques, bying the COP, heat pumps and HIDiC (an internally heat integrated distillation column) are recommended for methanol/water separation, while cyclic distillation and membrane assisted distillation are also promising. These recommendations are also in line with the reports from literature on different intensification techniques for the separation of methanol/water, e.g., HPAD [32], MAD [34], and cyclic distillation [35]. However, note that although HiGee distillation is also recommended and there is research supporting this [36], HiGee is limited by the difficulty of manufacturing large-size high-speed rotators. The annual production capacity per HiGee unit is usually smaller than 10 ktons, and in practice the annual production capacity is less than 5 ktons in most cases [31].

Figure 5:

Case study on the methanol/water separation.

5.2 Binary azeotropic mixtures separation – (1) ethanol/water, (2) acetonitrile/water

For ethanol/water separation, there is a binary azeotrope (95.63 wt % ethanol) that must be taken into account. For Q6, the answer is yes, thus membrane separation, extractive distillation, azeotropic distillation and pressure swing distillation could possibly break the azeotrope. However, based on the y-x diagram, ethanol-water is not a pressure sensitive mixture, so pressure swing distillation is not recommended in this case. Due to the additional solvent that may be needed (leading to a ternary system) E-DWC and A-DWC are also recommended. HPAD and HIDiC can also be considered upon furthers, as using different entrainers lead to different temperatures thus the COP is different. Luo et al. [37] and Ponce et al. [38] reported two designs of using HPAD and HIDiC for separating ethanol and water.

For acetonitrile/water separation, assume that the industrial wastewater contains 20 mol% ACN. According to Q5 and R5, a preconcentration column is first needed to remove part of the large amount of water as heavy component. Figure 6 illustrates the use of the developed methodology. Acetonitrile and water forms an azeotropic mixture having a boiling point of 76.5 °C and containing approximately 82.2 wt% acetonitrile under atmospheric pressure. PSD, MAD, ED are recommended for acetonitrile/water separation, and HPAD could be applied, while the detailed configuration needs to beed based on the specific temperature lift. According to Liu et al. [39]; molecular sieve membrane assisted distillation is built for a 10 kt/a acetonitrile separation plant in Shandong Huihai Company, China. Zhai et al. [40] developed a heat-pump assisted extractive-azeotropic dividing wall column (DWC) for the acetonitrile/water separation.

Figure 6:

Case studies on (a) ethanol/water and (b) acetonitrile/water separation.

5.3 Ternary zeotropic mixtures separation – benzene (B)/toluene (T)/xylene (X)

Since this is a ternary separation, DSE evaluation is carried out first. Figure 7 shows the η of each stream calculated based on equation (1), and the DSE is over 35 % which means that the separation is relatively easy. The DSE for direct sequence (38.33 %) is higher than that of indirect sequence (37.42 %), meaning that direct sequence is more favorable than indirect sequence for the BTX separation.

Figure 7:

DSE evaluation of direct sequence (left) and indirect sequence (right).

The next steps go to high-level questions and the PI matrix, Figure 8 shows the results for the BTX separation. According to the evaluation procedure, the use of dividing wall column is recommended for BTX separation, and bying the coefficient of performance, heat pumps could also be applied additionally or alternatively. These recommendations are supported with the studies on DWC [33], HIDiC [41], and heat pump assisted DWC [42].

Figure 8:

Case study on the benzene/toluene/xylene separation. Note: *Assume compressor efficiency 0.7, minimum temperature difference 5 °C, pressure drop 50 kPa. **AHP, absorption heat pump; VR, vapor recompression.