Performances of Transcritical Power Cycles with CO2-Based Mixtures for the Waste Heat Recovery of ICE

In the waste heat recovery of the internal combustion engine (ICE), the transcritical CO2 power cycle still faces the high operation pressure and difficulty in condensation. To overcome these challenges, CO2 is mixed with organic fluids to form zeotropic mixtures. Thus, in this work, five organic fluids, namely R290, R600a, R600, R601a, and R601, are mixed with CO2. Mixture performance in the waste heat recovery of ICE is evaluated, based on two transcritical power cycles, namely the recuperative cycle and split cycle. The results show that the split cycle always has better performance than the recuperative cycle. Under design conditions, CO2/R290(0.3/0.7) has the best performance in the split cycle. The corresponding net work and cycle efficiency are respectively 21.05 kW and 20.44%. Furthermore, effects of key parameters such as turbine inlet temperature, turbine inlet pressure, and split ratio on the cycle performance are studied. With the increase of turbine inlet temperature, the net works of the recuperative cycle and split cycle firstly increase and then decrease. There exist peak values of net work in both cycles. Meanwhile, the net work of the split cycle firstly increases and then decreases with the increase of the split ratio. Thereafter, with the target of maximizing net work, these key parameters are optimized at different mass fractions of CO2. The optimization results show that CO2/R600 obtains the highest net work of 27.43 kW at the CO2 mass fraction 0.9 in the split cycle.


Background
Recently, with the rapid development of economy, the demand for energy in various industries has exploded. Although there exist many types of energy sources, such as fossil energy, solar energy, and geothermal energy, fossil energy is still dominant. So far, CO 2 produced by fossil fuel combustion has caused serious global warming, which attracts the attention of all countries in the world. Thus, it is urgent to improve the utilization efficiency of fossil fuel. As one of the major application fields of fossil fuel combustion, internal combustion engines (ICEs) consume a large amount of fossil oil every year. However, only 30-40% of the energy is converted into useful work [1], and a large amount of fuel heat is released through the exhaust gas and coolant of the vehicle. In general, the waste gas temperature of ICE can be up to 400-700 • C and the temperature of engine coolant is about 80-90 • C. Therefore, the recovery of high temperature exhaust heat can greatly improve the ICE efficiency and reduce fuel consumption. So far, various thermodynamic cycles have been proposed for efficiently recovering waste heat from ICE. Among these proposed cycles, the most representative technology is the organic Rankine cycle (ORC) [2][3][4], which has the advantages of simple structure and easy maintenance.

Transcritical Power Cycle with CO 2
To address the above issues, many scholars have used CO 2 to replace the organic fluid in the field of waste heat recovery. Compared with the organic fluid, CO 2 has low GWP and exhibits high stability, with being non-flammable and non-toxic. Furthermore, because of the low critical temperature (31.1 • C), CO 2 is preferred to be applied in the transcritical Rankine cycle to recover the ICE waste heat. Compared to the ORC, the CO 2 power cycle has obvious advantages: (1) the high density of supercritical CO 2 state can greatly improve the compactness of heat exchangers, thus resulting in a smaller system footprint [10]; (2) the CO 2 power cycle can be started up quickly, and has a fast response speed [11]. This cycle is more suitable to recover the ICE waste heat. In view of these advantages, extensive research has yet been carried out to investigate the CO 2 power cycle performance in the waste heat recovery. For instance, Shi et al. [12] established an experimental system of the transcritical CO 2 power cycle to recover the diesel engines waste heat. The effects of the pressure ratio, engine speed, and pump speed on system performance were experimentally conducted. The experimental results indicated that the net work (2.05 kW) can be expected to be obtained with 1300 rpm of engine and 80 rpm of pump, and the expected thermal efficiency is 4.3% with 1300 rpm of engine and 70 rpm of pump. Li et al. [13] investigated a transcritical CO 2 power cycle with a preheater and recuperator to recover the engine waste heat, and analyzed in detail the temperature disturbance during the heating process. The results illustrated that the reduction of temperature interference can increase the cycle thermal efficiency. Meanwhile, the optimized net work increases from 14.7 kW to 19.0 kW. In addition, a transcritical CO 2 cycle test to recover low-grade waste heat was developed by Li et al. [14]. The effect of working fluid mass flow on the system performance was investigated. It was found that with the increase of working fluid mass flow, the overall system pressure increases and output power decreases. In terms of the performance comparison between CO 2 and organic fluid, Baik et al. [15] optimized the net works of transcritical CO 2 and R125 cycles to recover the low-grade heat. The results showed that

Cycle Layout
In this study, two transcritical cycles, namely the recuperative cycle and split cycle, are considered. The corresponding layouts are respectively shown in Figures 1 and 2. It can be observed that besides the four basic components (working fluid pump, heater, turbine, and condenser), a recuperator is added at the pump outlet to recover the energy of the outlet fluid from the turbine in the recuperative cycle. This happens so that the cycle efficiency (η th ) and exergy efficiency (η ex ) can be increased. For the recuperative cycle, the detailed process description can be referred to in these pieces of literature [28,29]. The corresponding T-s diagram is provided in Figure 3 under design conditions. Figure 3 shows that the exhaust gas outlet temperature is still as high as 253 • C. This is because there exists a large amount of recuperative heat, thus resulting in a high temperature of the working fluid entering the heater. For the ICE, a turbocharged technology is used, and a six-cylinder engine is considered to produce waste gas. The inhaled fresh air is firstly compressed and exchanges heat with the jacket water, and then enters the cylinder to mix with the gasoline. After combustion, the generated waste gas is exhausted. The discharged gas enters the turbine to provide work for the air compressor. As shown in Figures 1 and 2, the working fluid of the power cycle is directly condensed by the air, to get the lowest condensation temperature. As for the other two condensation methods, the readers can refer to the literature [27].  As for the split cycle, Heater2 is added at the outlet of Heater1 to further extract the waste heat. After being compressed in the pump, the flow is separated into two streams. One stream enters into the recuperator to conduct a heat exchange with the high-temperature fluid at the turbine outlet. Another enters Heater2 to recover the waste heat. Thereafter, two flows are mixed at the inlet of Heater1 to continue absorbing exhaust gas heat. The remaining processes of this cycle are similar to those of the recuperative cycle. Figure 4 presents the corresponding T-s diagram of the split cycle under design conditions. The figure shows that the split cycle can make the waste gas temperature decrease to 139 • C.

Thermodynamic Modeling
To compare the thermodynamic performance of CO2 mixtures in the two cycles, mathematical models are established. Before giving the used eq following assumptions are applied to simplify the modeling.
1. Without considering the variation of the system kinetic and potential e 2. The system operates stably and without considering the heat loss an drop of pipes and components; 3. The counter-flow heat exchanger is considered in the simulation; 4. The compression and expansion processes are characterized by isen ciencies.
According to the above assumptions, energy equations of each com listed in Table 1 for the two cycles.  For the ICE, a turbocharged technology is used, and a six-cylinder engine is considered to produce waste gas. The inhaled fresh air is firstly compressed and exchanges heat with the jacket water, and then enters the cylinder to mix with the gasoline. After combustion, the generated waste gas is exhausted. The discharged gas enters the turbine to provide work for the air compressor. As shown in Figures 1 and 2, the working fluid of the power cycle is directly condensed by the air, to get the lowest condensation temperature. As for the other two condensation methods, the readers can refer to the literature [27].

Thermodynamic Modeling
To compare the thermodynamic performance of CO 2 mixtures in the considered two cycles, mathematical models are established. Before giving the used equations, the following assumptions are applied to simplify the modeling.
1. Without considering the variation of the system kinetic and potential energy; 2. The system operates stably and without considering the heat loss and pressure drop of pipes and components; 3. The counter-flow heat exchanger is considered in the simulation; 4. The compression and expansion processes are characterized by isentropic efficiencies. According to the above assumptions, energy equations of each component are listed in Table 1 for the two cycles.

Components
Recuperative Cycle Split Cycle The corresponding net work and cycle efficiency are calculated by: Split cycle (2) In addition to the cycle efficiency, the waste heat recovery efficiency (η r ) can be expressed as below: Before establishing the exergy equations, the specific exergy value at each steady point (i) is calculated as [30,31]: where the ambient temperature (T 0 ) is set to be 298.15 K. The exergy destruction [30,31] of each component is presented in Table 2. Table 2. Exergy destruction of each component for the recuperative cycle and the split cycle.

Recuperative Cycle Split Cycle
Pump In Table 2, T average -5 is the average temperature of the cooling medium. As for the exergy efficiency, it can be obtained by: Based on the listed equations, the MATLAB platform is used for code development. The calculation flows for recuperative and split cycles are respectively given in   Figure 5. Thermodynamic calculation routine of the recuperative cycle.   Figure 6. Thermodynamic calculation routine of the split cycle.

GA Optimization
In this study, the system parameters of the two cycles are optimized by GA, due to its simplicity, versatility, and suitability for parallel processing [30,31]. Under the given ICE gas conditions, the maximum Wnet is considered as the optimization objective. For the recuperative cycle, T3 and Ph are the optimization cycle variables. However, for the

GA Optimization
In this study, the system parameters of the two cycles are optimized by GA, due to its simplicity, versatility, and suitability for parallel processing [30,31]. Under the given ICE gas conditions, the maximum W net is considered as the optimization objective. For the recuperative cycle, T 3 and P h are the optimization cycle variables. However, for the split cycle, SR is added as the third variable for performance optimization. Figure 7 gives the flowchart of the GA. The GA encodes the randomly generated chromosomes of the initial population firstly. Then the chromosome is decoded to obtain the corresponding system parameters. According to these parameters, thermodynamic calculations are conducted. The fitness function for each chromosome corresponds to the output work. Afterwards, the temperature differences in recuperator and heaters are checked to guarantee the fact that the calculated temperature difference meets the required minimum value. If not, all chromosomes will be replaced by the next population, which are propagated by choice, intersection, and mutation operators. Thereafter, W net is obtained by recalculating the new generation. The optimization will finish when the genetic calculation is iterated to the maximum number of generations. Finally, the optimal parameters in the final population are substituted into the thermodynamic calculations to obtain the system efficiency and other parameters. In addition, the population size is programmed to 100 and the maximum number of generations is stated as 80.
Entropy 2021, 23, x 10 of 29 split cycle, SR is added as the third variable for performance optimization. Figure 7 gives the flowchart of the GA. The GA encodes the randomly generated chromosomes of the initial population firstly. Then the chromosome is decoded to obtain the corresponding system parameters. According to these parameters, thermodynamic calculations are conducted. The fitness function for each chromosome corresponds to the output work. Afterwards, the temperature differences in recuperator and heaters are checked to guarantee the fact that the calculated temperature difference meets the required minimum value. If not, all chromosomes will be replaced by the next population, which are propagated by choice, intersection, and mutation operators. Thereafter, Wnet is obtained by recalculating the new generation. The optimization will finish when the genetic calculation is iterated to the maximum number of generations. Finally, the optimal parameters in the final population are substituted into the thermodynamic calculations to obtain the system efficiency and other parameters. In addition, the population size is programmed to 100 and the maximum number of generations is stated as 80.

Cycle conditions and Organic Fluid Selections
The working conditions of a six-cylinder engine are enumerated in Table 3. The outlet temperature and mass flow rate of the exhaust gas are 466 ℃ and 0.278 kg/s, respectively [23]. As for the gas components, due to the low contents of nitrogen oxides and

Cycle conditions and Organic Fluid Selections
The working conditions of a six-cylinder engine are enumerated in Table 3. The outlet temperature and mass flow rate of the exhaust gas are 466 • C and 0.278 kg/s, respectively [23]. As for the gas components, due to the low contents of nitrogen oxides and sulfides, only CO 2 , O 2 , H 2 O, and N 2 are considered. The corresponding mass fractions are 9.1%, 9.3%, 7.4%, and 74.2%, respectively. The CO 2 -based mixture is formed by CO 2 and an organic fluid. The mixture properties are strongly dependent on the selection of the organic fluid. Thus, selecting a suitable fluid is crucial to obtain a better cycle performance of the CO 2 -based mixture. In general, environmental properties have to be considered in the fluid selection. The selected fluid should be non-toxic, non-corrosiveness and stability, and have zero ozone depletion potential (ODP) and a GWP less than 150 [32].
Considering that the flue gas temperature is as high as 466 • C, the traditional refrigerant is easy to decompose, so a high temperature working fluid is selected. Based on the above standards, five kinds of organic working fluids, namely R600, R600a, R601, R601a, and R290, are selected from the organic fluid library [33]. The physical parameters of these organic working fluids and CO 2 are summarized in Table 4. Table 4. Thermodynamic properties of the selected working fluids.

Fluids
Molecular Mass (g/mol)  Figure 8 shows the critical temperatures of five mixtures at different CO 2 fractions. It is found that the critical temperature decreases with the increase of the CO 2 mass fraction. At the same CO 2 fraction, the critical temperatures of five mixtures satisfy the order: R601 > R601a > R600 > R600a > R290. As for the critical pressure, Figure 9 presents the variations of these mixtures. The critical pressure firstly increases and then decreases with the increase of the CO 2 mass fraction. When the CO 2 fraction is larger than 0.2, the critical pressure satisfies the order: R601 > R601a > R600 > R600a > R290. In addition, it should be mentioned that the selected organic fluid is flammable. However, when CO 2 is used as an addition to mix with the organic fluid, the flammability will be decreased with the increase of the CO 2 fraction. According to Zabetakis's research [19], at a CO 2 mass fraction larger than 0.3, the mixture has exceeded the combustible range of R290. Although no similar data is available for other working fluids, the flammability of these five organic fluids is similar. Thus, in this study, it is thought that the five CO 2 -based mixtures are non-flammable, when the CO 2 mass faction is no less than 0.3.   Figure 10 illustrates the temperature glide variation trend with CO2 ma at the bubble temperature of 30 ℃. It is obvious that when the CO2 mass creases, the temperature glide of mixture first increases and then decreases. considered mixtures, CO2/R290 has the lowest temperature glide, while CO the largest value. Moreover, when the CO2 mass fraction is around 0.3, t glides of these five mixtures reach the maximum value.     Figure 10 illustrates the temperature glide variation trend with CO2 ma at the bubble temperature of 30 ℃. It is obvious that when the CO2 mass creases, the temperature glide of mixture first increases and then decreases. considered mixtures, CO2/R290 has the lowest temperature glide, while CO the largest value. Moreover, when the CO2 mass fraction is around 0.3, t glides of these five mixtures reach the maximum value.  Figure 10 illustrates the temperature glide variation trend with CO 2 mass fractions at the bubble temperature of 30 • C. It is obvious that when the CO 2 mass fraction increases, the temperature glide of mixture first increases and then decreases. Among the considered mixtures, CO 2 /R290 has the lowest temperature glide, while CO 2 /R601 has the largest value. Moreover, when the CO 2 mass fraction is around 0.3, temperature glides of these five mixtures reach the maximum value. In addition to the employed CO2-based mixtures, standard design condi so required for the thermodynamic analysis of these two transcritcial system in Table 5, in the thermodynamic calculation, T1, T3, and Ph are set to be 30 and 13MPa, respectively. For the split cycle, SR is assumed to be 0.7 under t sign condition. Furthermore, isentropic efficiencies of turbine (ηt) and pump to 0.7 and 0.8, respectively. Pinch point temperature differences (PPTDs) of th tor, Heater1, and Heater2 are all set at 15 ℃. Meanwhile, to reduce the mixtu bility, the CO2 mass fraction is set at 0.3.
To further study the thermodynamic performance of the two cycles, analyses are conducted on T3, Ph, and T1. Meanwhile, the effect of SR on the p of the split cycle is also discussed. As listed in Table 5, T3 varies from 270 ℃ an interval of 5 ℃ and Ph changes from 8.5 MPa to 20MPa. In addition, the range of T1 is 30-90 ℃. In the split cycle, SR varies from 0 to 1. After that, sys eters are optimized at various CO2 mass fractions (0.3-1) for the two cycles.  In addition to the employed CO 2 -based mixtures, standard design conditions are also required for the thermodynamic analysis of these two transcritcial systems. As listed in Table 5, in the thermodynamic calculation, T 1 , T 3 , and P h are set to be 30 • C, 375 • C, and 13 MPa, respectively. For the split cycle, SR is assumed to be 0.7 under the basic design condition. Furthermore, isentropic efficiencies of turbine (η t ) and pump (η p ) are set to 0.7 and 0.8, respectively. Pinch point temperature differences (PPTDs) of the recuperator, Heater1, and Heater2 are all set at 15 • C. Meanwhile, to reduce the mixture flammability, the CO 2 mass fraction is set at 0.3. Table 5. Standard design conditions of the two transcritical systems.

Design Parameters Set Value Range of Variation
Condensation bubble temperature ( • C) 30 30-90 Turbine inlet temperature ( • C) 375 270-390 Turbine inlet pressure (MPa) 13 8.5-20 To further study the thermodynamic performance of the two cycles, sensitivity analyses are conducted on T 3 , P h , and T 1 . Meanwhile, the effect of SR on the performance of the split cycle is also discussed. As listed in Table 5, T 3 varies from 270 • C to 390 • C at an interval of 5 • C and P h changes from 8.5 MPa to 20 MPa. In addition, the considered range of T 1 is 30-90 • C. In the split cycle, SR varies from 0 to 1. After that, system parameters are optimized at various CO 2 mass fractions (0.3-1) for the two cycles.

Results and Discussion
According to the conditions provided in Table 5, the recuperative cycle and the split cycle performance are derived from the established models. A synthesis of these two systems' performances from the perspective of energy and exergy is compared. Furthermore, the effects of T 3 , P h , T 1 , and SR on the system performance are analyzed. On this basis, these parameters of the recuperative cycle and the split cycle are optimized with the maximum W net as the objective. Furthermore, at different CO 2 mass fractions, performance of these two cycles are optimized and compared. The detailed results and discussions are presented in the following subsections.

Cycle Analysis and Performance Comparison
Under set operating conditions, mixture characteristics at various state points are achieved for the recuperative and the split cycle, as listed in Tables A1 and A2 of the Appendix A for CO 2 /R290(0.3/0.7). The corresponding T-s diagrams for the recuperative cycle and the split cycle are respectively presented in Figures 3 and 4. Based on these state properties, energy performance of five CO 2 -based mixtures in the recuperative cycle and the split cycle are sequentially provided in Tables 6 and 7. From Table 6, it is observed that CO 2 /R600a(0.3/0.7) has the lowest turbine outlet pressure 2.88 MPa. While the highest W net is achieved by CO 2 /R290(0.3/0.7) and the lowest W net is obtained by CO 2 /R601(0.3/0.7). In terms of η th , CO 2 /R290(0.3/0.7) has the highest η th 23.74%, followed by CO 2 /R600a(0.3/ 0.7), CO 2 /R600(0.3/0.7), CO 2 /R601a(0.3/0.7) and CO 2 /R601(0.3/0.7). Thus, overall, CO 2 /R290 (0.3/0.7) has the best thermodynamic performance. For the split cycle, W net and η th of CO 2 /R290(0.3/0.7) are the highest, followed by CO 2 /R600a(0.3/0.7), CO 2 /R600(0.3/0.7), and CO 2 /R601a(0.3/0.7). It should be noted that CO 2 /R601(0.3/0.7) has the largest recuperated heat 131.49 kW, but the worst thermodynamic performance. In terms of thermodynamic performance, CO 2 /R290(0.3/0.7) is still the first choice for the split cycle. In comparison to the recuperative cycle, the split cycle possesses a higher W net , as shown in Table 7. This phenomenon is attributed by a higher mass flow rate in the split cycle. Meanwhile, the exhaust gas outlet temperature of the split cycle is lower than that of the recuperative cycle. For the considered five mixtures, the exhaust gas outlet temperature of the recuperative cycle ranges from 253.35 • C to 283.92 • C, while the split cycle has a range of 139-143.5 • C. It means that the introduction of Heater2 can deeply recover the heat from the exhaust gas for the split cycle. Thus, by constructing advanced cycle structures, the output work can be increased and fuel combustion of ICE can be reduced. Tables 8 and 9 respectively gives the exergy parameters of five mixtures in the recuperative cycle and split cycle under design conditions. For the exergy performance of each component, it is obvious that in the recuperative cycle, the irreversibility of the recuperator is much larger than those of other components. However, in the split cycle, the largest irreversibility is obtained by the condenser, except CO 2 /R290(0.3/0.7). As for the total irreversibility, compared with the recuperative cycle, the split cycle has a higher value, due to the addition of Heater2. For the considered five mixtures, the irreversibility of the split cycle ranges from 27.29 kW to 33.06 kW, while the recuperative cycle has a range of 18.9 kW-19.52 kW. In terms of η ex , the value of the split cycle is lower than that of the recuperative cycle for each mixture. Among the five mixtures, CO 2 /R290(0.3/0.7) has the largest efficiency in the two cycles. The corresponding values are 45.22% and 43.55%, respectively.  Figure 11 illustrates the changes of W net with T 3 for the two cycles. The T 3 is investigated in the range of 270 • C to 390 • C. From Figure 11a, it is found that the W net of the recuperative cycle tends to increase and then decrease as T 3 increases. This phenomenon can be interpreted based on the fact that with the increase of T 3 , the mass flow of mixtures gradually decreases. Under a lower T 3 , the variation of W net is mainly affected by the enthalpy difference; while under a higher T 3 , the change of W net depends on the mass flow rate. As for the considered five mixtures, CO 2 /R290(0.3/0.7) has the largest W net in the recuperative cycle, followed by CO 2 /R600a(0.3/0.7), CO 2 /R600(0.3/0.7), CO 2 /R601a(0.3/0.7) and CO 2 /R601(0.3/ 0.7). As for the split cycle, similar trends of these mixtures are observed, and W net satisfies the order: CO 2 /R601(0.3/0.7) > CO 2 /R601a(0.3/0.7) > CO 2 /R600(0.3/0.7) > CO 2 /R600a(0.3/0.7) > CO 2 / R290(0.3/0.7). Meanwhile, for the same mixture, the split cycle has a larger W net than that of the recuperative cycle. Furthermore, in these two cycles, when the organic fluids are isomers, the corresponding CO 2 mixtures perform a similar performance, such as CO 2 /R600a(0.3/0.7) and CO 2 /R600(0.3/0.7). tures gradually decreases. Under a lower T3, the variation of Wnet is mainly affected by the enthalpy difference; while under a higher T3, the change of Wnet depends on the mass flow rate. As for the considered five mixtures, CO2/R290(0.3/0.7) has the largest Wnet in the recuperative cycle, followed by CO2/R600a(0.3/0.7), CO2/R600(0.3/0.7), CO2/R601a(0.3/0.7) and CO2/R601(0.3/ 0.7). As for the split cycle, similar trends of these mixtures are observed, and Wnet satisfies the order: CO2/R601(0.3/0.7) > CO2/R601a(0.3/0.7) > CO2/R600(0.3/0.7) > CO2/R600a(0.3/0.7) > CO2 / R290(0.3/0.7). Meanwhile, for the same mixture, the split cycle has a larger Wnet than that of the recuperative cycle. Furthermore, in these two cycles, when the organic fluids are isomers, the corresponding CO2 mixtures perform a similar performance, such as CO2/R600a(0.3/0.7) and CO2/R600(0.3/0.7). Split cycle (Y-axis range: 12kW~28kW) The variations of ηth with T3 are shown in Figure 12. The figure presents that ηth naturally increases with the increase of T3 in the two cycles. Although the Wnet of the split cycle is greater than that of the recuperative cycle, the split ηth is significantly lower than that of the recuperative cycle. This is due to the fact that the added Heater2 absorbs more heat from the exhaust gas in the split cycle. Among the considered five mixtures, CO2/R290(0.3/0.7) shows the best ηth, followed by CO2/R600a(0.3/0.7), CO2/R600(0.3/0.7), CO2/R601a(0.3/0.7), and CO2/R601(0.3/0.7). The effects of T3 on ηr are presented in Figure  13. It can be seen that the heat ηr decreases as T3 increases. Meanwhile, being different with efficiency curves of the recuperative cycle, the curves of the split cycle among the five mixtures are more closed with each other.  The variations of η th with T 3 are shown in Figure 12. The figure presents that η th naturally increases with the increase of T 3 in the two cycles. Although the W net of the split cycle is greater than that of the recuperative cycle, the split η th is significantly lower than that of the recuperative cycle. This is due to the fact that the added Heater2 absorbs more heat from the exhaust gas in the split cycle. Among the considered five mixtures, CO 2 /R290(0.3/0.7) shows the best η th , followed by CO 2 /R600a(0.3/0.7), CO 2 /R600(0.3/0.7), CO 2 /R601a(0.3/0.7), and CO 2 /R601(0.3/0.7). The effects of T 3 on η r are presented in Figure 13. It can be seen that the heat η r decreases as T 3 increases. Meanwhile, being different with efficiency curves of the recuperative cycle, the curves of the split cycle among the five mixtures are more closed with each other. flow rate. As for the considered five mixtures, CO2/R290(0.3/0.7) has the largest Wnet in the recuperative cycle, followed by CO2/R600a(0.3/0.7), CO2/R600(0.3/0.7), CO2/R601a(0.3/0.7) and CO2/R601(0.3/ 0.7). As for the split cycle, similar trends of these mixtures are observed, and Wnet satisfies the order: CO2/R601(0.3/0.7) > CO2/R601a(0.3/0.7) > CO2/R600(0.3/0.7) > CO2/R600a(0.3/0.7) > CO2 / R290(0.3/0.7). Meanwhile, for the same mixture, the split cycle has a larger Wnet than that of the recuperative cycle. Furthermore, in these two cycles, when the organic fluids are isomers, the corresponding CO2 mixtures perform a similar performance, such as CO2/R600a(0.3/0.7) and CO2/R600(0.3/0.7). The variations of ηth with T3 are shown in Figure 12. The figure presents that ηth naturally increases with the increase of T3 in the two cycles. Although the Wnet of the split cycle is greater than that of the recuperative cycle, the split ηth is significantly lower than that of the recuperative cycle. This is due to the fact that the added Heater2 absorbs more heat from the exhaust gas in the split cycle. Among the considered five mixtures, CO2/R290(0.3/0.7) shows the best ηth, followed by CO2/R600a(0.3/0.7), CO2/R600(0.3/0.7), CO2/R601a(0.3/0.7), and CO2/R601(0.3/0.7). The effects of T3 on ηr are presented in Figure  13. It can be seen that the heat ηr decreases as T3 increases. Meanwhile, being different with efficiency curves of the recuperative cycle, the curves of the split cycle among the five mixtures are more closed with each other.

Effect of Turbine Inlet Pressure
In this section, the effect of Ph on the system performance is revealed by varying the pressure settings from 8.5MPa to 20MPa in the step of 0.5MPa. Figure 14 shows the effects of Ph on Wnet in the recuperative cycle and the split cycle. It can be observed from Figure 14a that in the recuperative cycle, Wnet increases as Ph increases, while the growth speed gradually slows down at around 14MPa. By comparison, the Wnet of the five mixtures is ordered by CO2/R290(0.3/0.7) > CO2/R600a(0.3/0.7) > CO2/R600(0.3/0.7) > CO2/R601a(0.3/0.7) > CO2/R60 1(0.3/0.7). As for the split cycle, when the pressure gradually increases, Wnet increases first and then decreases, as illustrated in Figure 14b. For the performance comparison, the split cycle has a larger work than the recuperative cycle for the same mixture. Taking CO2/R600a(0.3/0.7) as an example, when Ph is 16MPa, the Wnet of the split cycle is 18.5kW and that of the recuperative cycle is 15.42kW. This is mainly due to a large mass flow rate of the mixture in the split cycle. The variations of ηth with Ph are shown in Figure 15. It can be seen that with the increase of Ph, ηth first increases and then remains stable in the recuperative cycle and the split cycle. Different from the comparison of Wnet, the ηth of the recuperative cycle is higher than that of the split cycle. This can be explained by the absorption of more waste heat in the split cycle.

Effect of Turbine Inlet Pressure
In this section, the effect of P h on the system performance is revealed by varying the pressure settings from 8.5 MPa to 20 MPa in the step of 0.5MPa. Figure 14 shows the effects of P h on W net in the recuperative cycle and the split cycle. It can be observed from Figure 14a that in the recuperative cycle, W net increases as P h increases, while the growth speed gradually slows down at around 14 MPa. By comparison, the W net of the five mixtures is ordered by CO 2 /R290(0.3/0.7) > CO 2 /R600a(0.3/0.7) > CO 2 /R600(0.3/0.7) > CO 2 /R601a(0.3/0.7) > CO 2 /R60 1(0.3/0.7). As for the split cycle, when the pressure gradually increases, W net increases first and then decreases, as illustrated in Figure 14b. For the performance comparison, the split cycle has a larger work than the recuperative cycle for the same mixture. Taking CO 2 /R600a(0.3/0.7) as an example, when P h is 16 MPa, the W net of the split cycle is 18.5 kW and that of the recuperative cycle is 15.42 kW. This is mainly due to a large mass flow rate of the mixture in the split cycle. The variations of η th with P h are shown in Figure 15. It can be seen that with the increase of P h , η th first increases and then remains stable in the recuperative cycle and the split cycle. Different from the comparison of W net , the η th of the recuperative cycle is higher than that of the split cycle. This can be explained by the absorption of more waste heat in the split cycle.

Effect of Turbine Inlet Pressure
In this section, the effect of Ph on the system performance is revealed by varying the pressure settings from 8.5MPa to 20MPa in the step of 0.5MPa. Figure 14 shows the effects of Ph on Wnet in the recuperative cycle and the split cycle. It can be observed from Figure 14a that in the recuperative cycle, Wnet increases as Ph increases, while the growth speed gradually slows down at around 14MPa. By comparison, the Wnet of the five mixtures is ordered by CO2/R290(0.3/0.7) > CO2/R600a(0.3/0.7) > CO2/R600(0.3/0.7) > CO2/R601a(0.3/0.7) > CO2/R60 1(0.3/0.7). As for the split cycle, when the pressure gradually increases, Wnet increases first and then decreases, as illustrated in Figure 14b. For the performance comparison, the split cycle has a larger work than the recuperative cycle for the same mixture. Taking CO2/R600a(0.3/0.7) as an example, when Ph is 16MPa, the Wnet of the split cycle is 18.5kW and that of the recuperative cycle is 15.42kW. This is mainly due to a large mass flow rate of the mixture in the split cycle. The variations of ηth with Ph are shown in Figure 15. It can be seen that with the increase of Ph, ηth first increases and then remains stable in the recuperative cycle and the split cycle. Different from the comparison of Wnet, the ηth of the recuperative cycle is higher than that of the split cycle. This can be explained by the absorption of more waste heat in the split cycle.    Figure 16b. This can be explained by the decrease of mass flow rate in the split cycle. Even so, compared with the recuperative cycle, ηr of the split cycle is still higher, and this increase of efficiency is due to the addition of Heater2, which increases the total heat absorption of the system.

Effect of Condensation Temperature
In general, there are three different cooling methods for fluid condensation in ICE [27]. The obtained T1 ranges from 30 ℃ to 90 ℃. Thus, the temperature effects on the performance of the recuperative cycle and the split cycle are analyzed. Special care should be given CO2/R290(0.3/0.7). Due to the limit of critical temperature, the maximum T1 is set to be 70 ℃ to guarantee the subcritical condensation of the mixture. Figure 17 presents the variations of the Wnet with T1 in the two cycles. It is clear that the Wnet of the recuperative cycle and the split cycle naturally decrease as T1 increases. Taking  Figure 16a,b show η r curves of two cycles at different P h . It is clearly seen from Figure 16a that in the recuperative cycle, η r increases rapidly with the increase of P h . This is because the heat absorbed by Heater1 from the exhaust gas gradually increases with the increase of P h . On the contrary, η r of the split cycle decreases with the increase of P h , as shown in Figure 16b. This can be explained by the decrease of mass flow rate in the split cycle. Even so, compared with the recuperative cycle, η r of the split cycle is still higher, and this increase of efficiency is due to the addition of Heater2, which increases the total heat absorption of the system.   Figure 16b. This can be explained by the decrease of mass flow rate in the split cycle. Even so, compared with the recuperative cycle, ηr of the split cycle is still higher, and this increase of efficiency is due to the addition of Heater2, which increases the total heat absorption of the system.

Effect of Condensation Temperature
In general, there are three different cooling methods for fluid condensation in ICE [27]. The obtained T1 ranges from 30 ℃ to 90 ℃. Thus, the temperature effects on the performance of the recuperative cycle and the split cycle are analyzed. Special care should be given CO2/R290(0.3/0.7). Due to the limit of critical temperature, the maximum T1 is set to be 70 ℃ to guarantee the subcritical condensation of the mixture. Figure 17 presents the variations of the Wnet with T1 in the two cycles. It is clear that the Wnet of the recuperative cycle and the split cycle naturally decrease as T1 increases. Taking CO2/R290(0.3/0.7) as an example, the Wnet of the split cycle decreases about 4kW and that of the recuperative cycle decreases about 3kW for every 20 ℃ rises of T1. Figure 18 shows that ηth of two cycles decrease with the increase of T1. It is worth

Effect of Condensation Temperature
In general, there are three different cooling methods for fluid condensation in ICE [27]. The obtained T 1 ranges from 30 • C to 90 • C. Thus, the temperature effects on the performance of the recuperative cycle and the split cycle are analyzed. Special care should be given CO 2 /R290(0.3/0.7). Due to the limit of critical temperature, the maximum T 1 is set to be 70 • C to guarantee the subcritical condensation of the mixture. Figure 17 presents the variations of the W net with T 1 in the two cycles. It is clear that the W net of the recuperative cycle and the split cycle naturally decrease as T 1 increases. Taking CO 2 /R290(0.3/0.7) as an example, the W net of the split cycle decreases about 4 kW and that of the recuperative cycle decreases about 3 kW for every 20 • C rises of T 1 . ered five mixtures in the split cycle, CO2/R290(0.3/0.7) has the largest ηth and CO2/R601(0.3/0.7) has the lowest efficiency when the T1 range is 30-70 ℃. In addition, for every 20 °C increase of T1, ηth of CO2/R290(0.3/0.7) in the split cycle decreases about 4%, while that of recuperative cycle is decreased by about 3%.   The influences of T1 on ηr for the recuperative cycle and the split cycle are illustrated in Figure 19. It can be found that ηr decreases with the increase of T1 in the recuperative cycle from Figure 19a. It can be explained by the fact that the increase of T1 will enlarge the recuperated heat, thus decreasing the absorbed heat in Heater1. When T1 increases 20℃, ηr of CO2/R290(0.3/0.7) in the recuperative cycle decreases about 4%. However, for the split cycle, the increase of T1 affects ηr slightly for different mixture fluids, as shown in Figure 19b. This is because that although the heat of Heater1 decreases, the heat absorbed in Heater2 continues to increase. Therefore, the total heat absorption of the split cycle varies little. Taking CO2/R290(0.3/0.7) as an example, T1 increases from 30 ℃ to 70 ℃, and ηr decreases from 59% to 58%.  Figure 18 shows that η th of two cycles decrease with the increase of T 1 . It is worth noting that in the recuperative cycle, when T 1 is lower than 55 • C, the W net of the considered mixtures satisfy the following order:    The influences of T1 on ηr for the recuperative cycle and the split cycle are illustrated in Figure 19. It can be found that ηr decreases with the increase of T1 in the recuperative cycle from Figure 19a. It can be explained by the fact that the increase of T1 will enlarge the recuperated heat, thus decreasing the absorbed heat in Heater1. When T1 increases 20℃, ηr of CO2/R290(0.3/0.7) in the recuperative cycle decreases about 4%. However, for the split cycle, the increase of T1 affects ηr slightly for different mixture fluids, as shown in Figure 19b. This is because that although the heat of Heater1 decreases, the heat absorbed in Heater2 continues to increase. Therefore, the total heat absorption of the split cycle varies little. Taking CO2/R290(0.3/0.7) as an example, T1 increases from 30 ℃ to 70 ℃, and ηr decreases from 59% to 58%. The influences of T 1 on η r for the recuperative cycle and the split cycle are illustrated in Figure 19. It can be found that η r decreases with the increase of T 1 in the recuperative cycle from Figure 19a. It can be explained by the fact that the increase of T 1 will enlarge the recuperated heat, thus decreasing the absorbed heat in Heater1. When T 1 increases 20 • C, η r of CO 2 /R290(0.3/0.7) in the recuperative cycle decreases about 4%. However, for the split cycle, the increase of T 1 affects η r slightly for different mixture fluids, as shown in Figure 19b. This is because that although the heat of Heater1 decreases, the heat absorbed in Heater2 continues to increase. Therefore, the total heat absorption of the split cycle varies little. Taking CO 2/ R290(0.3/0.7) as an example, T 1 increases from 30 • C to 70 • C, and η r decreases from 59% to 58%.  Figure 19. Effect of condensation temperature on recovery efficiency: (a) recuperative cycle, (b) split cycle.

Effect of Split Ratio
In the above analysis, SR is set to 0.7 for the split cycle. However, in practical engineering, the SR can be manually controlled. When the value is set to 1, it represents the recuperative cycle. Meanwhile, at a SR of 0, it means that the split cycle becomes a basic transcritical cycle. Therefore, the effect of SR on the split cycle performance is worth investigating. Figure 20 shows the variation curves of Wnet in split ratio within the range of 0-1. It is obvious from the figure that with the increase of SR, Wnet firstly increases and then decreases. There exists a maximal value for each mixture. This is because that at a small SR, the split cycle is close to a simple transcritical cycle, thus producing a lower Wnet. Meanwhile, when SR is close to 1, the split cycle is more similar to the recuperative cycle, and the advantage of the split is weakened. It should be noted that when SR is larger than 0. As for the ηth in Figure 21, when SR gradually increases, ηth of five mixtures increas- Figure 19. Effect of condensation temperature on recovery efficiency: (a) recuperative cycle, (b) split cycle.

Effect of Split Ratio
In the above analysis, SR is set to 0.7 for the split cycle. However, in practical engineering, the SR can be manually controlled. When the value is set to 1, it represents the recuperative cycle. Meanwhile, at a SR of 0, it means that the split cycle becomes a basic transcritical cycle. Therefore, the effect of SR on the split cycle performance is worth investigating. Figure 20 shows the variation curves of W net in split ratio within the range of 0-1. It is obvious from the figure that with the increase of SR, W net firstly increases and then decreases. There exists a maximal value for each mixture. This is because that at a small SR, the split cycle is close to a simple transcritical cycle, thus producing a lower W net . Meanwhile, when SR is close to 1, the split cycle is more similar to the recuperative cycle, and the advantage of the split is weakened. It should be noted that when SR is larger than 0.8, there exists a little rise of W net . Similarly, for the five mixtures, W net satisfies the order:  (a) (b) Figure 19. Effect of condensation temperature on recovery efficiency: (a) recupe split cycle.

Effect of Split Ratio
In the above analysis, SR is set to 0.7 for the split cycle. However, in neering, the SR can be manually controlled. When the value is set to 1, it recuperative cycle. Meanwhile, at a SR of 0, it means that the split cycle be transcritical cycle. Therefore, the effect of SR on the split cycle performan vestigating. Figure 20 shows the variation curves of Wnet in split ratio within the r is obvious from the figure that with the increase of SR, Wnet firstly increase creases. There exists a maximal value for each mixture. This is because that the split cycle is close to a simple transcritical cycle, thus producing a low while, when SR is close to 1, the split cycle is more similar to the recupera the advantage of the split is weakened. It should be noted that when As for the ηth in Figure 21, when SR gradually increases, ηth of five mi es continuously. This is because a higher ratio led to a larger recuperated heat absorbed in Heater1 and Heater2 continuously decreases. It can be when SR is 0, ηth is lower than 10%. While as SR is 1, ηth is higher than 2  As for the η th in Figure 21, when SR gradually increases, η th of five mixtures increases continuously. This is because a higher ratio led to a larger recuperated heat. The total heat absorbed in Heater1 and Heater2 continuously decreases. It can be observed that when SR is 0, η th is lower than 10%. While as SR is 1, η th is higher than 20%. Figure 22 shows the variations of η r under different SR. As SR increases, η r at first barely changes and then decreases considerably. More working fluid enters Heater2 when SR is relatively small, so η r is high. After SR increases to a certain extent, the cycle is close to the recuperative cycle, so η r is reduced.
Entropy 2021, 23, x tively small, so ηr is high. After SR increases to a certain extent, the cycle recuperative cycle, so ηr is reduced. Split ratio Figure 22. Effect of split ratio on recovery efficiency in the split cycle.

Parametric Optimization
Based on the above analysis, it can be concluded that T3, Ph, T1, and S ent effects on cycle performance. However, due to the fact that T1 is dete cooling method, T1 here is set to 30 °C. For Wnet optimization of the recupe and Ph are set in the range of 240-400 °C and 8.5MPa-20MPa, respectiv split cycle, SR ranges from 0 to 1. On this basis, the optimization results fo tive cycle and the split cycle are provided in Tables 10 and 11, respectively  Split ratio Figure 22. Effect of split ratio on recovery efficiency in the split cycle.

Parametric Optimization
Based on the above analysis, it can be concluded that T3, Ph, T1, and S ent effects on cycle performance. However, due to the fact that T1 is dete cooling method, T1 here is set to 30 °C. For Wnet optimization of the recuper and Ph are set in the range of 240-400 °C and 8.5MPa-20MPa, respective split cycle, SR ranges from 0 to 1. On this basis, the optimization results for tive cycle and the split cycle are provided in Tables 10 and 11, respectively.

Parametric Optimization
Based on the above analysis, it can be concluded that T 3 , P h , T 1 , and SR have different effects on cycle performance. However, due to the fact that T 1 is determined by the cooling method, T 1 here is set to 30 • C. For W net optimization of the recuperative cycle, T 3 and P h are set in the range of 240-400 • C and 8.5 MPa-20 MPa, respectively. As for the split cycle, SR ranges from 0 to 1. On this basis, the optimization results for the recuperative cycle and the split cycle are provided in Tables 10 and 11, respectively. For the recuperative cycle, Table 10 shows that the optimized W net is substantially increased, compared to W net under the basic operating condition. For instance, W net of CO 2 /R290(0.3/0.7) under basic operating conditions is 16.11 kW, while the optimized W net is 18.33 kW. The increase of W net reaches up to 13.78%. Among the five mixtures, a W net of CO 2 /R290(0.3/0.7) is the largest and that of CO 2 /R601(0.3/0.7) is the smallest. As for the split cycle, performance is also greatly improved. From Table 11, for CO 2 /R290(0.3/0.7), compared to W net of 21.05 kW at basic operating conditions, the optimized W net of 23.37 kW increases 11.02%. Compared with the optimization results of the recuperative cycle, the split cycle has a larger W net and a higher η r for the considered mixtures.

Performance Comparison at Different CO 2 Mass Fractions
Considering that the mixture properties change considerably with the CO 2 mass fraction, performance of the two cycles are further obtained at different fractions. To avoid the mixture flammability, CO 2 mass fraction ranges from 0.3 to 1.0. In this range, performance comparisons are respectively conducted under design and optimization conditions. The details are presented in the following subsections. Figure 23 presents the variations of W net with the CO 2 mass fraction for the recuperative and split cycles under design conditions. In Figure 23a for the recuperative cycle, it is clear that except CO 2 /R290, with the increase of CO 2 mass fraction, the W net of the other four mixtures first increase and then decrease. This phenomenon can be explained by the temperature glides. At a lower CO 2 mass fraction, the temperature glide of the five mixtures is large and the W net is relatively small. However, with the increase of a CO 2 mass fraction, the temperature glide gradually decreases and the W net of the mixtures gradually increases. It should be noted that the W net of the five mixtures are equal with each other at the CO 2 mass fraction 0.65. When CO 2 mass fraction is lower than 0.65, CO 2 /R290 exhibits the highest W net , followed by CO 2 /R600a, CO 2 /R600, CO 2 /R601a, and CO 2 /R601. However, when the CO 2 mass fraction is greater than 0.65, a small difference exists for the W net of each mixture. As for the split cycle, the variation trend of W net is similar to that in the recuperative cycle. Being different with the recuperative cycle, the split cycle has an equal W net for different mixtures at the CO 2 fraction 0.7. When the fraction is less than 0.7, the highest work is still obtained by CO 2 /R290. While, the value of net work of CO 2 /R290 is the lowest when the fraction is higher than 0.7. In terms of the W net comparison between the two cycles, the split cycle always has a larger work than that of the recuperative cycle.

of 29
CO2/R290 exhibits the highest Wnet, followed by CO2/R600a, CO2/R600, CO2/R601a, and CO2/R601. However, when the CO2 mass fraction is greater than 0.65, a small difference exists for the Wnet of each mixture. As for the split cycle, the variation trend of Wnet is similar to that in the recuperative cycle. Being different with the recuperative cycle, the split cycle has an equal Wnet for different mixtures at the CO2 fraction 0.7. When the fraction is less than 0.7, the highest work is still obtained by CO2/R290. While, the value of net work of CO2/R290 is the lowest when the fraction is higher than 0.7. In terms of the Wnet comparison between the two cycles, the split cycle always has a larger work than that of the recuperative cycle. At different CO2 mass fractions, ηex of the two cycles are provided in Figure 24. It can be observed that ηex has similar curves with Wnet, in terms of the variation trends and curve distributions. For the recuperative cycle, the minimum ηex 34.46% is obtained by pure CO2, while the maximum efficiencies of these mixtures are obtained at different fractions. Among the five working fluids, Figure 24a depicts that when the mass fraction of CO2 is 0.3, CO2/R290 has the highest ηex of 45.22%. It should be noted that the equal efficiencies of different mixtures are obtained at the CO2 mass fraction 0.5. For the split cycle in Figure 24b At different CO 2 mass fractions, η ex of the two cycles are provided in Figure 24. It can be observed that η ex has similar curves with W net , in terms of the variation trends and curve distributions. For the recuperative cycle, the minimum η ex 34.46% is obtained by pure CO 2 , while the maximum efficiencies of these mixtures are obtained at different fractions. Among the five working fluids, Figure 24a depicts that when the mass fraction of CO 2 is 0.3, CO 2 /R290 has the highest η ex of 45.22%. It should be noted that the equal efficiencies of different mixtures are obtained at the CO 2 mass fraction 0.5. For the split cycle in Figure 24b,  ficiencies of different mixtures are obtained at the CO2 mass fraction 0.5. For the split cy-cle in Figure 24b, the maximum efficiency 43.55% among the five mixtures is obtained by CO2/R290 (0.3/0.7). Meanwhile, mixtures have the minimum efficiency at different mass fractions. For the considered five mixtures, the lowest efficiency are 39.85%, 38.69%, 36.65%, 32.72% and 31.45% for CO2/R290(0.95/0.05), CO2/R600a(0.35/ 0.65), CO2/R600(0.3/0.7), CO2/R601a (0.3/0.7) and CO2/R601(0.3/0.7), respectively.    Figure 25 shows the variation trends of the component irreversibility with the CO 2 mass fraction. For the two cycles, with the mass fraction of CO 2 increases, the irreversibility of the recuperator increases significantly. When the mass fraction of CO 2 is low, the mixture has a better thermal match in the recuperator. However, opposite variations are observed for turbine in the recuperative cycle. Furthermore, for the split cycle in Figure 25b, with the increase of CO 2 mass fraction, the irreversibility in Heater2 increases. Due to this, the total irreversibility of the split cycle is always larger than that of the recuperative cycle.
24 of 29 Figure 25 shows the variation trends of the component irreversibility with the CO2 mass fraction. For the two cycles, with the mass fraction of CO2 increases, the irreversibility of the recuperator increases significantly. When the mass fraction of CO2 is low, the mixture has a better thermal match in the recuperator. However, opposite variations are observed for turbine in the recuperative cycle. Furthermore, for the split cycle in Figure 25b, with the increase of CO2 mass fraction, the irreversibility in Heater2 increases. Due to this, the total irreversibility of the split cycle is always larger than that of the recuperative cycle.

Optimization Conditions
To further compare the system performance, the maximum Wnet at different CO2 mass fractions is obtained by parametric optimization. Similarly, in the optimization process, T1 is set to 30℃. Considering the fact that the employed mixture has the highest critical pressure of 10.64MPa for CO2/R601, the Ph ranges from 11MPa to 20MPa. For T3, the range is 240°C-400°C. Furthermore, in the split cycle, SR varies from 0 to 1. Figure 26 shows the optimal Wnet of recuperative and split cycles at different CO2 mass fractions. It can be noted that the Wnet of two cycles increases with the increase of the CO2 mass fraction. Among the considered five mixtures, when the mass fraction of CO2 is lower than 0.6, the Wnet of CO2/R290 is the largest in the split cycle. Meanwhile, it

Optimization Conditions
To further compare the system performance, the maximum W net at different CO 2 mass fractions is obtained by parametric optimization. Similarly, in the optimization process, T 1 is set to 30 • C. Considering the fact that the employed mixture has the highest critical pressure of 10.64 MPa for CO 2 /R601, the P h ranges from 11 MPa to 20 MPa. For T 3 , the range is 240 • C-400 • C. Furthermore, in the split cycle, SR varies from 0 to 1. Figure 26 shows the optimal W net of recuperative and split cycles at different CO 2 mass fractions. It can be noted that the W net of two cycles increases with the increase of the CO 2 mass fraction. Among the considered five mixtures, when the mass fraction of CO 2 is lower than 0.6, the W net of CO 2 /R290 is the largest in the split cycle. Meanwhile, it is clear that at the CO 2 mass fraction of 0.9, CO 2 /R600 obtains the maximum W net . For the recuperative cycle, as the CO 2 mass fraction is lower than 0.7, the W net of five mixtures satisfies the order: CO 2 /R290 > CO 2 /R600a > CO 2 /R600 > CO 2 /R601a > CO 2 /R601. Through the performance comparison, the W net of the split cycle is greater than that of the recuperative cycle at different CO 2 mass fractions for five mixtures. Moreover, the specific optimization parameters of the recuperative cycle and the split cycle are listed in Tables 12  and 13, respectively. 23

Conclusions
In this paper, two transcritical systems, namely the recuperative cycle and the split cycle, are employed to recover the waste heat of ICE. Five CO 2 -based mixtures namely R290, R600a, R600, R601a, and R601 are used. Based on the thermodynamic calculations, cycle performance is analyzed and compared. Key parameters including turbine inlet temperature, turbine inlet pressure, condensation temperature, and split ratio are investigated to reveal the effects on the system performance. On this basis, these key parameters are optimized to achieve the maximum net work in two cycles. Meanwhile, cycle performance of mixtures at different CO 2 mass fractions are compared under design and optimization conditions. From the above results, the main conclusions are drawn as follows: Under design conditions, in the recuperative cycle, the maximum cycle efficiency of 23.74% is obtained by CO 2 /R290(0.3/0.7). In addition, the largest net work of 21.05 kW in the split cycle is gained by CO 2 /R290(0.3/0.7). Compared with the recuperative cycle, the net work and recovery efficiency of the split cycle are much higher.
With the increase of turbine inlet temperature, the net works of the two cycles tend to increase firstly and then decrease, while the cycle efficiency continuously increases and recovery efficiency decreases greatly. Furthermore, as the turbine inlet pressure increases, the net work of the recuperative cycle gradually increases, whereas the net work of the split cycle firstly increases and then decreases. For the effect of the split ratio, as the split ratio increases, the net work of the split cycle first increases and then decreases.
Under design conditions, as the CO 2 mass fraction increases, the net work of mixtures except CO 2 /R290 shows a trend of first increasing and then decreasing in the recuperative cycle and the split cycle. However, under optimization conditions, the maximum net work increases continuously. The maximum net work of the split cycle is obtained by CO 2 /R600(0.9/0.1) and is 27.43 kW. Funding: This work was funded by the National Nature Science Foundation of China, grant number 52106037 and the Provincial Nature Science Foundation of Hunan, grant number 2021JJ40755. In addi-tion, the work was also supported by the program "Researches on key technology to optimization and dynamic simulation of supercritical CO 2 power cycle" from China Three Gorges Corporation, grant number 202003024 and the foundation from key laboratory of renewable energy power generation and conversion in Fujian province, grant number KLIF-202106.