Recent Advances in the Development of Automotive Catalytic Converters: A Systematic Review

Abstract

Despite the current boost in the use of electric vehicles to reduce the automotive sector’s footprint, combustion vehicles are and will be present in our cities in both the immediate and long term. In this sense, catalytic converters, which are exhaust gas post-treatment systems for vehicle emission control, are critical for complying with increasingly stringent environmental regulations. This work proposes a systematic review to identify the most relevant knowledge regarding the parameters (materials, geometries, and engine conditions), conditions (cold start, oxygen storage, and deactivation), and mathematical models to consider in the design of catalytic converters. The Scopus database contains 283 records related to this review’s objective. After applying the inclusion and exclusion criteria, 65 reports were retrieved for evaluation. A table was created to present the results and prepare this manuscript. The evaluation revealed that the following topics were active: the study of non-noble catalyst materials, as well as new substrate materials and geometries, for designing more compact and cost-effective catalytic converters; the development of strategies to improve conversion during cold starts; and the development of accurate and fast estimation models.

1. Introduction

Despite the significant efforts currently being made in the development of electric vehicles, most road transport is still powered by internal combustion engines, which primarily run on liquid fuels derived from petroleum and their blends with biofuels.

Many advantages of combustion engines are currently unattainable with electric engines, especially in the transport sector. Some examples are the high power-to-weight ratio and the high energy content of conventional fuels that enable a fast refill [1].

Even with complete combustion and consequently no emissions of polluting gases, CO${}_2$ emissions concern governments, as they contribute to global warming. However, recent studies based on life-cycle assessments demonstrate that until electricity generation becomes green, the CO${}_2$ emission rate of electric vehicles is comparable to that of internal combustion engine vehicles powered by gasoline and diesel [2]. Furthermore, a significant reduction in CO${}_2$ emissions from internal combustion engines can be achieved through the use of synthetic fuels [3], also called e-fuels, power-to-liquid (PtL) fuels, or power-to-gas (PtG) fuels.

E-fuels are synthetic hydrocarbons resulting from the combination of renewable hydrogen and CO${}_2$ captured either from concentrated sources or from the air. They are classified as low-carbon sustainable fuels, and it is believed that they will play a crucial role in the decarbonization of transport.

Synthetic fuels avoid some problems concerning the use of biomass-derived fuels, such as land use shortages and interference with food supplies [4]. Additional advantages of synthetic fuels are that they provide a variety of fuel sources, avoid dependence on fuel-exporting countries, allow for the improvement of the fuel energy density [5], and can replace fossil fuels without the need for modifications to the internal combustion engine and without a complete overhaul of the energy supply chain and economics [6].

There are also health challenges related to electric vehicles, such as the disposal of their batteries, which can create severe threats to the natural environment and public health [7]. In the case of internal combustion engines, despite the great advances in both engine efficiency and the technologies used to control exhaust pollution, some small amounts of undesirable emissions are released into the atmosphere, consisting mainly of carbon monoxide (CO), hydrocarbons (HC), and nitrogen oxides (NO${}_x$) [8]. In this regard, catalytic converters (CCs) are the main technological solution for satisfying the stringent emissions limits, as they can achieve very high conversion efficiencies, which range between 90 and 99.9% [9]. Catalytic converters are widely used in the automotive sector for reducing harmful gas emissions, but they are also employed in other industrial applications such as hydrogen production or methane autothermal reforming [10]. Automotive CCs perform two main functions: reduction and oxidation. The first consists of reducing NO${}_x$ to N${}_2$ + O${}_2$, and the second to oxidizing CO to CO${}_2$ and HC to CO${}_2$ + H${}_2$O.

Two-way catalytic converters only perform the oxidation reaction function. This kind of converter is used in diesel engines because their high NO${}_x$ emissions require specific techniques to eliminate them. These techniques include exhaust gas recirculation (EGR) [11] and selective catalytic reduction (SCR) systems [12]. SCR systems require an injection of urea that is thermally decomposed to produce ammonia, which is the reagent in the reduction reaction. Diesel oxidation catalysts (DOCs) are the most commonly used converters for diesel engines since they can efficiently oxidize CO and HC, as well as particulate matter (PM).

Three-way catalytic converters (TWCs) perform both oxidation and reduction functions simultaneously. TWC systems are suitable for gasoline engines, as they emit very low or negligible amounts of NO${}_x$ [13]. There are two substrates inside a TWC. The first is the reduction substrate, which acts on the exhaust gases, and the second performs the oxidation process.

Scientific reviews aid in gathering, understanding, and applying the overwhelming number of publications in specific fields, such as CCs. This work uses a systematic review methodology to collect scientific works on the topic of CCs that comply with specific eligibility criteria and answer a set of specific questions or review objectives.

The questions considered have been defined according to the population, intervention, comparator, and outcome (PICO) framework. This approach for formulating research questions is usually limited to clinical studies. However, as the work of Nishikawa-Pacher [14] claims, it can be extended to other disciplines like the present work.

Regarding the above PICO scheme, the following are considered:

• Population: Automotive catalytic converters (CCs);
• Intervention: Experimental tests and simulations;
• Comparison: Simulation vs. test-literature experimental data;
• Outcome: The responses to study and improve include the catalyst’s cost, cold-start performance and light-off temperature, CC efficiency, catalyst poisoning, emissions, the viability of alternative techniques, and simulation (agreement, computational time, integration, or implementation complexity).

Taking into account the above definitions, the research questions or review objectives include:

• What are the geometries, materials, and wash coats used in CCs, including both conventional and recent alternatives?
• What are the strategies for improving the cold-start performance of CCs?
• What is the role of oxygen storage?
• What parameters influence deactivation and to what extent?
• What are the main mathematical or simulation methods used to understand the performance of automotive CCs?

This review is organized as follows. Section 2 describes the eligibility criteria and the search, screening, and synthesis procedures. Section 3 and Section 4 present the synthesis results collected to answer the above-mentioned research questions. Finally, the main conclusions are drawn in Section 5.

2. Systematic Review Methodology

Review works aim to gain knowledge in specific scientific fields, as well as help identify challenges being faced. Scholars, professionals, and policymakers trust scientific literature to make decisions, but deepening a topic requires effort and time due to the continuously growing records (conferences, papers, proceedings, etc.). These shortcomings can be mitigated through adequate reviews.

To the best of the authors’ knowledge, literature or narrative reviews, as well as systematic ones, are the most common types used in engineering works. For interested readers, the work of Grant et al. [15], which focuses on health research, identified 14 methodologies by assessing the processes required in a review: search, appraisal, synthesis, and the analysis SALSA framework.

Narrative reviews describe a particular topic with the help of existing publications. This approach, which is conducted by experts, provides a chance to delete duplicates and assess the value of existing works. Nevertheless, there is no specific review procedure; therefore, this approach can inadvertently omit records, leading to bias.

On the other hand, systematic reviews gather existing investigations to answer specific questions following a detailed search, appraisal, and synthesis protocol [16]. Despite systematic reviews requiring more effort and time, these works are reproducible and provide precise answers, thus reducing the risk of bias due to their comprehensive nature.

2.1. PRISMA Framework

In the present review, we follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) framework. The proposed method (described in the following paragraphs) adapts the PRISMA statement, which focuses on studies about the effects of health interventions, to answer the engineering questions defined in Section 1, providing a reproducible and updateable review.

A transparent review process based on a checklist of 27 items is critical according to the PRISMA statement. The PRISMA checklist identifies the items to report in the introduction, methods, results, and discussion sections of a systematic review. Among the papers used to understand and apply the PRISMA approach, in this review, we consider the 27-item checklist and its expanded version [17], as well as the PRISMA 2020 explanation and elaboration document [18].

2.2. Method’s Main Elements

The PICO framework described in Section 1 helps define the records (title and abstract of a published study indexed in a database or website) to be included. Specifically, only records regarding the materials, manufacturing, and geometries of automotive CCs for emission control (Population); existing experimental and simulation approaches used to understand and improve their performance (Intervention); and conventional approaches (experimental and/or numerical) are included for comparison against novel strategies (Comparison), and only reports containing information about the selected PICO outcomes (catalyst’s cost, cold-start performance, CC efficiency, etc. (Outcomes)) are included.

The records of research papers, conference proceedings, and book chapters in the English language up to the present are eligible (eligibility criteria). Nevertheless, they must include quantitative or trend results regarding the outcomes indicated by the PICO framework (ineligible criteria).

Regarding the search procedure, the Scopus database was used to gather scientific records from 1946 (database inception) to the search date of 19 June 2023. We did not collect gray literature and did not examine references or cited works (no snowball or forward searching).

To define an adequate keyword combination to obtain records that met the eligibility criteria, we started with several tentative combinations, and the final one used in this work in Scopus pseudo-code was:

(TITLE (“Catalytic converter*”) AND KEY (“Catalyst” OR “material*” OR “manufactur*” OR “monolith” OR “substrate” OR “geometry”) AND TITLE-ABS-KEY (“cost*” OR “emission*” OR “agreement” OR “energy” OR “computation*” OR “time”) AND TITLE-ABS-KEY (automo*))

The search using the above combination resulted in 283 records, which were subjected to a screening procedure (Figure 1 shows the screening flow diagram). We applied a double-checking selection strategy. In the first step, we automatically selected 20% of the records (50 works) that accounted for 80% of the citations. Pareto’s method automatically identifies works that, according to most scholars, have made significant contributions to the topic (Figure 2). The same number of the most relevant papers, considering the Scopus relevance algorithm, were also selected. This statistical calculation relates to how well the search criteria align with the database records [19], as per PRISMA guidelines. Finally, we retrieved the last 50 records most recently published to incorporate updated information.

The resulting 150 records selected in the first step were then subjected to a second check, where three reviewers read and assessed the titles and abstracts. In this step, each reviewer screened 50 different records, checking the eligibility criteria. Disagreements were resolved by consensus. Additionally, one reviewer identified duplicate records. After the screening procedure, the final number of records to include was (after deleting duplicates) 26 (Pareto list) + 34 (most recently published) + 37 (Scopus relevance) = 97 records.

For the synthesis, a table with different fields was completed after reading each of the 90 reports retrieved (from 97 records, as the authors of this review did not have access to 7 of the reports). Figure 3 shows the fields considered and an example of one row. The authors decided on the list of fields by consensus after reading several reports, and the information obtained was discussed, similar to a narrative review, and presented in concise tables (comprising only critical fields) in Section 3 and Section 4. The eligibility criteria for including a report were similar to those used in the screening stage but were based on the collected information summarized in the synthesis table. Inclusion was determined by finding enough information in the synthesis table, including at least the paper’s objective, approach study, and main results, and checking whether this information related to the research questions.

The outcomes collected during the synthesis stage were expressed in mm and cells per square inch (cpsi) for the monolith geometry, as well as in percentage conversions for species emissions. International units were used for temperature, pressure, or other measurements when reported. Different abbreviations (summarized in the Abbreviations section) were used to simplify the summary tables generated during synthesis. This review does not present a statistical or meta-analysis of the report results due to the qualitative and quantitative nature of the data, and the variables related to the outcomes considered vary widely among the records.

Finally, the most probable bias risks are the omission of some records during the search procedure, as it did not consider forward or snowball searching, and the first step of the proposed double-checking procedure, where an automatic procedure was used as a reliability assessment tool for selecting records.

3. Experimental Studies—Critical Aspects in CC Design

Many factors affect a CC’s efficiency, such as the cold start, the CC’s capacity to store O${}_2$, catalyst deactivation, the catalyst’s material, and the internal geometry of the catalytic converter (CC). The study of these factors allows researchers to improve the efficiency of CCs. This section reviews the experimental methods analyzed to improve the efficiency of CCs.

3.1. Cold Start

The light-off temperature is the temperature at which the conversion efficiency reaches 50%. When the CC reaches that temperature, it is said to be activated, and the time that elapses until that moment is called the light-off time. During the light-off time, the efficiency of the CC is close to zero. The tailpipe-polluting gases emitted during this period are called cold-start emissions, which account for 60 to 80% of the total emissions [21]. Consequently, much of the research related to CCs is focused on reducing cold-start emissions.

Emissions during a cold start can be reduced by decreasing the light-off time, which lasts about 180 s. Active and passive methods can be applied to reach this target. Active methods comprise heating systems, such as electrical and fuel-burner systems, as well as air or hydrogen injection systems. Passive methods include the use of a pre-catalyst, thermal management, and placing the CC closer to the exhaust manifold. All of these systems must be appropriately designed to avoid excessively high temperatures that can lead to durability problems.

Electrical systems require 1 to 2 kW in 20 to 40 s to reduce the light-off time, and the alternator or battery must supply the electrical energy, which requires twice this energy from the fuel. Fuel-burner systems are faster in reaching light-off than electrical ones since they can deliver significantly more power (10 to 20 kW) but are more complex [22]. Bhaskar et al. [23] tested the influence of air supply, together with an electrically heated pre-catalyst, both placed before the main catalytic converter, on the final conversion efficiency. The pre-catalyst was heated using a band-type resistance heater (RHCC). It was composed of a stainless-steel substrate for a quick warm-up and was coated with copper oxide metal as the catalyst due to its low cost. The experimental results showed that this combination of strategies significantly reduced CO and HC emissions during cold starts. The best improvement was achieved by combining the RHCC and 90 L/min of air supply for a duration of 40 s from the engine’s cold start.

Injecting air into the exhaust manifold helps the oxidation reaction of CO and HC, which is an exothermic reaction that allows the temperature of the gases to rise before they reach the CC inlet. Injecting air and using pre-catalysts are methods supported by the high HC levels attributed to the rich operation required during the engine’s cold start [24]. However, the high temperatures reached require optimization of the CC so that it can endure the harsh environment.

Yun performed experimental tests to validate a 1D simulation of the entire exhaust system of a liquefied petroleum gas (LPG) vehicle equipped with a secondary air injection (SAI) system. The author found that, during the cold-start period, at the optimal operating condition of this supplementary system (1.6 g/s of air flow rate for 30 s), the light-off time was 6.9 s faster, leading to a decrease in HC and CO emissions of 29% and 50%, respectively, whereas NO emissions increased by approximately 12%. These results demonstrate that SAI systems facilitate warm-up during cold starts by encouraging exothermic oxidation [25].

Kovacev et al. [26] studied the effect of adding hydrogen on the light-off behavior of new substrate geometries obtained via additive manufacturing and experimentally demonstrated that the addition of hydrogen shifts the light-off curves of CO, total hydrocarbons (THC), and NO to lower temperatures, which leads to an improvement in cold-start emissions.

Passive methods can achieve the same improvements in the inlet gas temperature. This is the case when using pre-catalysts (close-coupled catalyst converters) placed before the main catalytic converter (also called underfloor catalytic converters) and close to the engine’s exhaust manifold to increase conversion efficiency through a fast light-off time. According to a study by Lee et al. [27], a significant reduction in the light-off time can be achieved using pre-catalysts if the ignition retard and misfire are controlled to avoid catalyst temperatures above 1050 ${}^{\circ }$C, as these temperatures cause drastic thermal aging in the CC, leading to its deactivation and, consequently, to lower conversion efficiency. The spark retard increases with the engine load. If the misfire or spark retard increases, the afterburn in the exhaust increases, resulting in a rapid rise in the exhaust gas temperature.

Mahadevan and Subramanian [21] tested several approaches to simultaneously reduce the light-off time, thermal degradation, and aging through the use of a Dynamic Catalytic Converter System (DCCS), consisting of a telescopic pipe, which enables regulating the length of the pipe so that it can be placed very close to the exhaust manifold during cold start of the vehicle. Once it attains the light-off temperature, it can be moved to the position of any other conventional catalyst to reduce damage from overheating. They tested the DCCS at different positions from the exhaust manifold and also its combination with a pre-catalyst (PC) and hot air injected at the exhaust manifold. They concluded that all methods tested resulted in a reduction in the light-off time; however, the lowest time to light-off (10 s for CO and 13 s for HC) was observed with a combination of the DCCS and preheated air delivered at 80 ${}^{\circ }$C and 20 L/min. The authors also observed that HC becomes oxidized at higher temperatures compared to CO, which is supported by the lower activation energy needed to oxidize CO to CO${}_2$ as it is a single-step reaction.

Improved passive methods are based on thermal management, which allows the storage or regulation of the amount of heat released by the exhaust gases so that the converter is above the light-off temperature for as long as possible. These methods include the use of refractory substrates or air around the converter itself. Some problems related to conventional insulation are that they do not provide more than 3 h of heat retention, and they can lead to exceeding temperature safety limits, resulting in thermal degradation and lower conversion efficiency. These problems can be overcome using variable-conductance insulation to regulate the heat released and the phase change material to increase heat storage [22].

Similar to the cold start is the cool-down process, which produces high tail-pipe emissions because the temperature of the CC falls below the light-off temperature. This problem is also critical for hybrid vehicles since the engine goes offline due to the hybrid nature of the vehicle. In this regard, Young et al. [28] proposed to include additional engine starts in the supervisory control strategy to prevent the CC from falling below the light-off temperature.

3.2. Substrate Design and Interaction with the Exhaust System

In this section, substrate design refers to the geometrical characteristics and the choice of material for the substrate.

The core of a CC is the catalyst support, also called the monolith or substrate, which is made of ceramic or metal. Ceramic monoliths usually have an internal honeycomb structure with square section channels. The geometrical characteristics of the honeycomb monolith significantly influence the efficiency of the converter. Extruded ceramic monoliths made of cordierite (2MgO·2Al${}_2$O${}_3$·5SiO${}_2$) are the most widely used, mainly because of their low manufacturing costs despite having a higher wall thickness (0.178 mm) that leads to lower open frontal areas (69%) compared to metallic substrates [29]. Metallic monoliths are used in high-performance vehicles [30] and have the great advantage of very high open frontal areas (91% open area) as a consequence of the ability to manufacture a very low wall thickness (0.050 mm), which provides a higher flow contact surface area while offering a lower resistance to flow (back pressure) [31]. Additional advantages of metallic substrates include their high thermal conductivity and low heat capacity, which allow very fast heating and a shorter light-off time.

Regarding new materials for substrates, Mahyon et al. [32] tested a new micro-structured ceramic hollow fiber substrate impregnated with less than 0.7 wt% of palladium (Pd) as the catalyst and observed many advantages such as high geometric surface areas, low pressure drop, and light-off temperatures of CO oxidation comparable to those of commercial counterparts. Gambarotta et al. [33] numerically compared a honeycomb substrate as a baseline to two open-cell foam-like structures with the aim of analyzing the effects of foam substrates on engine performance and overcoming some drawbacks related to honeycomb monoliths such as low mass transfer and poor flow homogenization. They concluded that foams, compared to honeycombs, warm up and cool down faster, which facilitates rapidly reaching light-off temperatures and increases conversion efficiency, higher mass transfer properties that allow for more compact converters, and higher pressure drops that result in a total fuel consumption increase lower than 0.20%.

Srinivasa and Venkateswara [34] tested clay marbles as a substrate for CCs, since their high porosity provides a good absorption capacity of the chemical coating and the spherical surface provides maximum gas contact area. At the same time, the voids between successive balls facilitate the passage of gases, causing minimum back pressure. The authors also demonstrated that hexagonal is the best geometry for the substrate channels since it results in a minimum pressure drop compared to triangular, square, and circular geometries.

The geometry of the whole monolith cross-section also influences the efficiency of the CC. For example, as the length of the substrate increases, the length of its channels and the residence time increase, leading to improved efficiency. An increase in the surface area contacting the exhaust flow also increases the efficiency of the CC. This contact area depends on the number of cells per square inch (cpsi). New geometries and materials for CC substrates are being investigated. Circular or nearly circular monolith contours have several advantages, such as promoting symmetric and uniform temperature distribution, uniform pressure distribution, and minimizing localized pressure peaks and deformation, which increase durability [35].

Regarding the influence of the internal geometry of the CC on its efficiency, factors influencing performance include the geometric configuration of the substrate channel, such as the hydraulic diameter, channel length, cell density, and wall thickness. It has been observed that higher cell densities, longer channels, thinner walls, and smaller hydraulic diameters can increase conversion efficiency, but longer channels produce a higher pressure drop [36]. The magnitude of the pressure drop and local friction coefficients also depends on the specific channel shape [10].

A study by Lapisa et al. [37] also showed that the geometric parameters significantly affect the conversion of HC and CO. They found that using folded perforated plates as a substrate resulted in good performance in terms of decreasing the HC (−93.6%) and CO (−0.56%) levels, as the zig-zag plate shape in their model increased the surface contact and intensity of the gas-catalyst collision.

Advances in internal geometries for the substrate are being investigated with the aid of additive manufacturing techniques. This is the case of diamond-based lattice substrates obtained through additive manufacturing, which have been compared to a conventional 400 cpsi honeycomb catalyst design, resulting in a significant improvement in the light-off temperatures for the CO, THC, and NO of the 3D-printed substrates [26].

The use of higher cell densities to achieve a faster light-off is one of the generally accepted geometrical approaches, despite the increase in the back pressure that ensues [38]. Brück et al. [39] performed experimental tests in metallic substrates and proposed low cell densities for underfloor catalytic converters and high cell densities for close-coupled catalysts to improve cold-start behavior. On the other hand, several studies have analyzed substrates with variable cell densities. A dual-zone monolith consisting of a high-cell-density core and a low-cell-density outer annulus did not improve the performance of the CC in terms of the light-off [38]. However, Yang et al. [40] proposed a new monolith structure with uniformly changing cell density and a new heater, which improved the flow field uniformity and NO conversion.

3.3. Catalyst Material

The low porosity of the substrates used in CCs (both ceramic and metallic) makes them inappropriate as catalyst supports. To overcome this problem, a thin layer of porous material, such as $\gamma$-Al${}_2$O${}_3$ (a type of alumina), is applied to the channel walls. This layer, called a wash coat, is used to disperse the catalyst materials, mainly precious metals that promote oxidation and reduction reactions with a very high conversion efficiency. Increasing the precious metal loading is a straightforward way to increase the efficiency of CCs, especially at low temperatures. However, the use of a high precious metal loading increases the price of a CC and favors sintering at high temperatures, leading to the deactivation of the CC [29]. The process used for the deposition of the catalyst on the substrate also affects the efficiency of a CC [41]. Additionally, wash-coat materials, such as ceria-based oxides, are added as oxygen storage agents [42].

The noble metals that usually function as catalyst materials are palladium and platinum, which enable the oxidation of HC and CO, and rhodium for NO${}_x$ reduction. The high cost of these metals, due to their scarcity, has prompted researchers to study alternative catalyst materials.

Using copper (CU) as an alternative low-cost catalyst material is one of the most investigated methods for emissions reduction. Dey and Chandra [43] reviewed the use of copper oxide catalysts for decreasing CO emissions and concluded that copper oxides are effective in CO oxidation. However, the crystal size significantly influences CO oxidation. The increase in temperature also promotes CO conversion rates unless it leads to the sintering of the catalyst. Venkatesan et al. [44] studied the influence of copper oxide catalyst on HC, NO${}_x$, CO, and smoke and observed maximum reductions at full load: 32% in HC and smoke, 61% in NO${}_x$, and 21% in CO when considering emissions without a CC as the baseline.

The combination of copper with other metals has also been studied. Chafidz et al. [45] tested a copper-zinc (Zn) metal as the catalyst, observing reductions in CO and HC emissions of up to 47.71% and 55.34%, respectively. Additionally, the activation energy of oxidation CO and HC was found to be 132.54 kJ/mol and 37.55 kJ/mol, respectively. The efficiency of the chrome-coated copper plate (Cu + Cr) as a catalyst was analyzed by [46], who found that the combination of these metals reduces CO and HC by 95.35% and 79.28%, respectively. A mixture of CU and manganese (Mn) oxides, (CuMnO${}_x$), is the best-known catalyst for CO oxidation at low temperatures [47].

Calcium titanium oxide (CaTiO${}_3$), referred to as perovskite, performs well as a low-cost catalyst due to its high activity and stability at different temperatures, which provide CO and HC conversion efficiencies of up to 77.89% and 68.50%, respectively [47]. Keav et al. [48] reviewed the main advantages of perovskite, highlighting its robust crystal structure and flexible oxygen content, which are helpful in redox reactions. In addition, its good oxidation activity can be facilitated by using noble metals. Calcium carbonate (CaCO${}_3$), referred to as limestone, was tested as a catalyst due to its low cost, good durability characteristics, and poison-resistant nature, showing potential for reducing HC, CO, and NO${}_x$ emissions [49].

Barium oxide (BaO) and magnesium oxide (MgO) were tested as catalysts, resulting in reductions in CO and HC emissions but not in NO${}_x$ emissions. Both catalysts achieved significant and similar reductions in HC emissions. CO emissions were also reduced, but the BaO catalyst had better conversion efficiency [50]. Barium chloride (BaCl${}_2$) has also shown potential to decrease HC and NO${}_x$ emissions, with reductions of 60.2% and 50%, respectively, as observed by [51].

It is also possible to find studies in which the monolith was manufactured using low-cost materials acting as catalysts. This was the case in a study by Ghofur et al. [52], who tested fly ash produced in the coal industry as a catalyst with two objectives: reducing environmental problems related to fly ash and decreasing exhaust emissions from engines. Based on a developed prototype of a CC combined with an air injection system, the authors observed reductions in HC and CO emissions of up to 48% and 45%, respectively. Zeolites have also been tested as substrates to act as catalysts. A study by Subramanian and Gnanasikamani [53] analyzed the conversion efficiency of two zeolites, ZSM 4A and ZSM 5A, over the exhaust gases of an engine fueled with a blend of diesel and pyrolysis oil from waste plastic. ZSM 5A performed better than ZSM 4A due to the amplification of the pore structure. For ZSM 5A, NO, HC, and CO emissions were reduced by 18%, 22%, and 10%, respectively, and for ZSM 4A, they were reduced by 12%, 16%, and 8%, respectively.

Although researchers have achieved significant improvements in emissions reduction through the use of alternative catalyst materials, more efforts are needed to reach the conversion efficiency levels achieved through the use of noble metals.

3.4. Oxygen Storage Capacity

The signal delay from the lambda sensor present in most current vehicles leads to oscillations in feed gas concentrations around a desired reference. Despite this being seen as a drawback, oscillation feeding has a practical advantage. Since the 1980s, several studies have revealed that control of the oscillating feed gas amplitude and frequency affects the efficiency of CCs. In this sense, Lie, Hoebink, and Marin [54] studied the influence of cycle CO exhaust concentration on the average CC conversion time. They observed an increment in the average conversion time below the light-off temperature and the opposite effect above it. This last fact highlights the use of a close stoichiometric ratio above the light-off temperature. An increment in the oscillation amplitude, as well as in the initial oxygen excess, prompted the observed results.

The oxygen storage capacity (OSC) explains the aforementioned behavior. The capacity of the CC to store and release O${}_2$ influences the TWC’s efficiency since the reduction process is more effective for rich mixtures (low amount of O${}_2$), in contrast to what occurs in the oxidation process, which is more efficient when the engine runs lean [55].

Cerium (Ce), which is usually present in the wash coat, is a key promoter of the OSC. It boosts the dynamic redox processes and is important for wash-coat stabilization and thermal resistance. While Ce provides oxygen storage capacity, the temperature and redox environment define the amount of oxygen stored. Koltsakis, Konstantinidis, and Stamatelos studied the OSC as a function of temperature and redox environment [56] and obtained the following results: in lean environments, the CC stores oxygen in the form of Ce oxides, and the oxygen capacity is reduced under stoichiometric conditions, whereas in rich environments, the temperature-storage capacity profile exhibits a convex shape (Ce oxidation increases the storage up to a maximum value, and after that, slower hydrocarbon oxidation consumes all oxygen, and there is no storage at all).

The development of an oxygen storage model concerning the oxygen storage from NO and the release from CO in a wash coat with Ce allowed Ohsawa et al. [57] to identify a larger amount of oxygen storage at the CC inlet, whereas a lower stoichiometry deviation at the outlet demonstrated the reduction effect of oxygen storage on air–fuel oscillations.

Zhang and Cheng experimentally investigated OSC deterioration due to aging and fuel sulfur [58]. Their study revealed that OSC decreased faster at the initial aging stage and by 10% for every 150 ppm of fuel sulfur. Additionally, the authors claimed that the loss of efficiency in the NO reduction in aged CCs is a consequence of the deactivation of precious metals and not of OSC inadequacy.

Investigations into improving the OSC using different wash coats have been conducted in parallel with investigations into the relationship between conversion efficiency and the interaction between the OSC and the air-fuel ratio. The interesting work by Kašpar, Fornasiero, and Hickey [29] reviewed studies on the advantages of CeO${}_2$–ZrO${}_2$ oxides (higher thermal stability and CC surface area) over conventional CeO${}_2$, explaining the reasons for their substitution since 1995. More recently, Khosh et al. [59] proposed a method for manufacturing nanorods of CeO${}_2$ doped with Zr that can be used in automotive CCs, and the experimental tests performed showed a significant increase in the specific surface area.

There is clear, solid knowledge about the OSC effect according to the experiments and simulations considered in this review. It is worth mentioning that the physical design of CCs, as well as the development of exhaust emissions control systems, have to consider this critical phenomenon.

3.5. Catalyst Deactivation

The causes of catalyst deactivation have also been investigated. Poisoning and high temperatures promote the loss of conversion or the deterioration of CCs. Automotive manufacturers are interested in the use of diagnostic tools to determine the deactivation of CCs. A simple approach consists of using thermocouples placed in the CC bricks to measure the exotherm (temperature rise due to chemical reactions) and correlate it with CCs’ steady-state HC conversion [60].

Typical studies on the deterioration of CCs conduct Rapid Aging Tests (RAT) to evaluate the failure modes of CCs. These tests can be performed in a vehicle, an engine test bench, or a laboratory. After vehicle aging, Beck, Sommers, and DiMaggio [61] studied the effect of thermal aging and poisoning (phosphorous and zinc) through the axial direction of coupled and underfloor CCs. They noticed an axial gradient in the light-off and warm-up for CO and HC conversion in the entrance of CCs. While poisoning influenced CO and NO${}_x$ activity, thermal deterioration affected the conversion of HC.

On the other hand, Poulopoulos and Philippopoulos [62] subjected a CC to laboratory thermal aging to evaluate its conversion performance after partial deactivation. Their experimental results indicated a larger influence on HC oxidation than on CO and an increase in the time required to start the conversion. More recently, Grane et al. [63] aged CCs in a CATAGEN Omega test reactor and used the changes in the light-off curves to determine the deactivation of the CCs (usually, aging moves light-off curves to higher temperatures [64]).

There are also environmental concerns about the emissions of noble metals due to the deterioration of CCs. A comparison between gasoline and diesel platinum-group element emissions (in soluble and particulate fractions) [65] revealed that unlike diesel engines, whose platinum emissions are at the same levels for both fresh and aged systems, aged gasoline engines produced lower noble metal emissions than fresh ones. On the other hand, aging increases the amount of soluble emissions, which are more dangerous than particulate ones.

The main characteristics of the studies cited in Section 3 are summarized in

Table 1. Main characteristics of reviewed experimental studies.

Ref.	Test Bench	Fuel	Species	Catalyst Mat.	Monolith Mat.	Monolith Dimensions and Structure	Operating Regime or Synthetic Flow
[21]	SI, 4-cylinder, Izusu	-	CO, HC	-	-	- Two pre-catalysts: 20% and 40% volume of the main CC + main CC	Cold start
[22]	Chrysler Dodge Neon 2.0 L, 4-cylinder	-	CO, HC, NO${}_x$	Nearly-fresh Pd, Loading 10.6 g/L	Cordierite	Two monoliths: D 90, L 152, 400 cpsi	FTP
[23]	Ambassador, SI, 4-cylinder	-	CO, HC	RHCC: Copper oxide (270 cm${}^3$)	RHCC: Metal	RHCC: fine-meshed stainless-steel substrate; Primary CC	1750 rpm under no load (20 s after cold start). The required load is applied at the same speed.
[25]	SI	LPG	CO, HC, NO${}_x$	Pt/Rh	Ceramic	MCC: 600 cpsi, WT 0.114; UCC: 400 cpsi, WT 0.140. Both CCs: D 118, L 136	Cold-start period in CVS-75 mode
[26]	CI, Diesel, 1-cylinder	Ultra-low-sulfur diesel	CO, HC, NO${}_x$	Catalytic carrier: gamma alumina Precious metal: Pd (2.5 wt%)	Al${}_2$O${}_3$ photopolymerizable ceramic slurry	Three substrates: - Diamond cell (similar to 400 cpsi); - Elongated diamond (similar to 400 cpsi); - Honeycomb (baseline) 400 cpsi.	One stationary condition representing low loads
[27]	SI, 1.0 L, 4-cylinder	-	-	-	-	CCC: meets Euro 3; UCC	From no load to full load conditions (1500–4000 rpm)
[28]	ICE hybrid electric, 2016 Chevrolet	E85 gasoline	CO, HC, NO${}_x$	-	Cordierite	1600 cpsi; oval (mm): 76.2 × 60.33, L 101.6, WT 0.1	US06 City drive cycle
[32]	Reactor	-	CO	Pd, 0.7 wt%	Micro-structured alumina hollow fibers	L 50; GSA: 40 (equivalent to 750 cpsi)	Preheated flow: 50 mL min${}^{-1}$ air; 50 mL min${}^{-1}$ (10% CO in 90% Ar)
[33]	TD, 1600 cm${}^3$, 4 in-line cylinders	-	CO, HC	-	Cordierite; Foams: Al${}_2$O${}_3$	Honeycomb: $ϵ$ = 63%, Dc = 1, SSA = 2700, 400 cpsi Open cell foam and Kelvin cell structures: $ϵ$ = 73%, Dc = 2, SSA = 1000 All CC: L150, D80	Simulation: EUDC Validation: steady-state operating conditions by the OEM
[34]	In-line OHV, CI, FSE; Swaraj Mazda Minibus	-	CO, HC	2 types: CeO${}_2$; ZnO${}_2$	3 types: - Marble clay; - Stainless steel; - Copper sheets.	Several typologies: - Balls; - Perforated disks; - Rolled sheets.	Idling condition (with and without each CC)
[35]	SI	-	-	-	Cordierite ceramic	Monolith contours: round, oval or elliptic, triangular, and wide oval or racetrack; Standard monolith: 400 cpsi, WT 0.0068${}^{\prime \prime }$	-
[36]	SI, 4-cylinders, Proton	-	CO, HC, NO${}_x$; TWC	Pd, Rh and Pt Pd > Rh > Pt	Cordierite	Two honeycomb UCC: - 400 and 600 cpsi; - Hydraulic diameter: 1.14 and 0.98 mm; - Designed for PROTON Wira 1.3 L and FIAT Punto Selecta 1.2 L.	Steady-state analysis (cold start not included)
[37]	Motorbike engine	-	CO, HC	Brass; (Cu 85%, Zn 15%)	Brass plate; WT: 0.5	Several typologies: - Set of perforated disks; - Set of folded and perforated disks; - Rolled brass plates.	3 engine speeds: 1700 (idle), 2700, and 3700 rpm
[39]	3.5 I, 4-cylinder	-	CO, HC, NO${}_x$	Pt/Rh (40 g/ft)	Metallic	5 diameters (400 cpsi): D 60, 70, 90, 105, 127; 5-cell density (D 90): 100, 200, 400, 500, 600 cpsi; 2 CC designs (400 cpsi): D 118 L 40, D 60 L 150; 2-foil WT: 0.05, 0.04	Bag I of the FTP and in a constant high load point
[40]	GDI	-	NO	Pt	Metallic	Two-zone substrate WT 1.7, L 100, D 100: - Middle: High cell density (500 cpsi, SSA 2995, $ϵ$ 72%); - Around: Low-cell density (400 cpsi, SSA 2728, $ϵ$ 75%); Heater system placed at the monolith inlet	6 steady-state operating conditions; (n 3000 rpm; Torque: from 11.0 Nm to 51.3 Nm)
[41]	Reactor	-	CO, HC, NO${}_x$	2 types: - Ag; - Pd.	2 types: - Alumina; - Ce–Zr.	-	Synthetic gas flow similar to automobile exhaust gases
[44]	CI; 1-cylinder, FSE	-	CO, HC, NO${}_x$, smoke	Copper oxide	Ceramic	Honeycomb Width: 150 mm; Spine: 40 mm	0%, 25%, 50%, 75%, and 100% of maximum rated load
[45]	Motorbike (Honda Supra Fit 2004)	-	CO, HC	Cu-Zn	Cu-Zn	CC: L 66, D 24 inner structure: 10 tubes	n: 2000 and 2500 rpm
[46]	SI, Toyota Kijang LSX 1.8 series, 4-cylinder	-	CO, HC	Chrome-coated copper (CU + CR)	Metallic	Honeycomb design	Stationary tests: idle conditions from 750 rpm to 5000 rpm with a range of 250 rpm
[49]	CI, FSE, 1-cylinder	-	CO, HC, NO${}_x$	Mullite (non-noble metal, based on material limestone)	Mullite	Perforated mullite plate	Variable engine running conditions
[50]	DI-Diesel, 1-cylinder, FSE, Kirloskar TV1	Pure diesel 3 blends: diesel + biodiesel + CeO${}_2$	CO, HC, NO${}_x$	2 nanocatalysts: - BaO; - MgO.	Metal	2 CC (tubes inserted in disks)	Brake Power: from o to 3.5 kW
[51]	CI	-	CO, HC, NO${}_x$	Blend: barium chloride + zirconium oxide	Metallic	Wire mesh: D 80, Dc 0.5	5 loads
[52]	SI, 4-cylinder, 1600 cm${}^3$, FSE, TOYOTA	-	CO, HC	Fly ash (activated using sulfuric acid)	Activated Fly ash + cement (adhesive)	Several lengths (L 50, 70, 90), D 50, Dc 2	Idle at 700, 1000, 1500, and 2000 rpm. Air injected at 0.05, 0.1, and 0.15 MPa.
[53]	Kirloskar 5.2 kW CI, 1-cylinder	100% Diesel; Blend: 50% pyrolysis oil + 50% Diesel	CO, HC, NO${}_x$	2 CC, each with a different Zeolite: - First CC: ZSM 4A; - Second CC: ZSM 5A.	Zeolite + Bentonite clay + White cement + Distilled water	CC casing: D 80, L 600 Structure with circular holes	Brake power: 0, 1.32, 2.6, 3.9, and 5.2 kW (loads: 0, 4, 8, 12, and 16 kg)
[57]	Flow reactor	-	CO, HC, NO${}_x$ (Simulated)	Pt/Rh	Ceramic	Crashed ceramic substance of Pt/Rh TWC, D 15	Inlet gas, 2 tests at 400 and 700 ${}^{\circ }$C: lean (CO: 2500 ppm; O${}_2$: 0.325%, 0.225%); rich (7500 ppm, O${}_2$: 0.175%, 0.255%)
[58]	Daimler-Chrysler 2.4 L, 4-cylinder	4 fuels: Gasoline with 7, 33, 266, and 500 ppm sulfur	CO, HC, NO${}_x$	Dual-brick system, 1.23 L per brick: - Front: Pd; - Rear: Pt + Rh.	Ceramic	Daimler-Chrysler 2001 Model, 2.4 L vehicles with ULEV designation	Steady-state conditions (1600 rpm and 0.5 bar intake pressure) with lambda variation
[59]	Reactor	-	CO, HC, NO${}_x$	Pd/Zr-doped CeO${}_2$, 4 Zr concentrations (5, 10, 15, and 20 mol%)	Cordierite	Honeycomb: L 20, D 10, 400 cpsi	Synthetic gas similar to gasoline exhaust gases: Propane (390 ppm) + CO (0.56 vol%) + NO${}_x$ (1002 ppm) + O${}_2$ (0.56 vol%) + CO${}_2$ (11 vol%) + Ar (balance), from 100 to 600 ${}^{\circ }$C.
[60]	3.8 L sequential. Port fuel injection	-	CO, HC, NO${}_x$	Pt/Rh or Pt/Pd/Rh; wash coat: alumina  ceria	Ceramic	UCC: 400 cpsi; front brick 81 × 145 oval, L 152; rear brick 81 × 145 oval, L 127	FTP
[61]	SI, 3.8 L	-	CO, HC, NO${}_x$	CCC: Pd, 75 g/ft${}^3$; UCC (two bricks): Pt 18.3, Rh 1.3 g/ft${}^3$	Cordierite ceramic	CCC: D 3.66${}^{\prime \prime }$ round, L 3${}^{\prime \prime }$; UCC: front brick 3.18${}^{\prime \prime }$ × 6.68${}^{\prime \prime }$ oval, L 6${}^{\prime \prime }$; rear brick: 3.18${}^{\prime \prime }$ × 6.68${}^{\prime \prime }$ oval, L 5${}^{\prime \prime }$	Reactor for lean and stoichiometric light-off activity and lean and stoichiometric warm-up activity at 600 ${}^{\circ }$C
[62]	IC, 4-cylinder OPEL 1.6 L	Unleaded gasoline with 11% w/w MTBE	CO, HC, MTBE, CH${}_4$, Ethylene	0.4% w/w Pt/Rh (5:1)	Ceramic (zirconia and alumina)	Made in 1997 according to Euro II: BET surface area: 8.29 m${}^2$ g${}^{-1}$; average pore radius: 47.40 A${}^{\circ }$	Engine operating cycle range n = 1000–2500 rpm and torque = 0.1–40 lbf ft; one experiment duration = 3900 s
[64]	Light-duty diesel truck, 1997, 2.5 L, DI/TCI, Euro II	2 fuels: low (4 ppm) and high (350 ppm) sulfur diesel fuel	PM	Pt-alumina zeolitic wash coat (5 mg/cm${}^3$ precious metal loading)	Cordierite	Two 2.5 L substrates in series with and without wash coats: - Contour: round; - 62 cells/cm${}^2$; - D 144, L 152; - WT 0.0017.	Steady-state speeds (40, 70, and 70 mph) and 3% grade. Transient effects from 40 to 70 mph. Wind speed from 25 to 70 mph. Air temperature 20 ${}^{\circ }$C. Relative humidity from 15% to 50%.
[65]	SI and CI	-	-	Four CC: SI: Front brick Pt/Rh (5/1); Rear Pd/Rh (9/1); SI: Single brick Pd/Rh (9/1) with CCC; Two CI CC: Single brick Pt (1.29 g)	-	Four CC manufactured in 1998 for: - Ford Fiesta van 1.3 L (Gasoline); - Ford Fiesta van 1.1 L (Gasoline); - Seat Ibiza 1.9 L (Diesel) 2 samples.	EUDC driving cycle for light-duty vehicles

. This summary includes the characteristics of the engine/vehicle or alternative testing system, the fuel used for powering the engine if it differs from regular diesel or gasoline fuel, the converted species analyzed in the study, the characteristics of the CC such as the dimensions and structure of monolith and catalyst materials, and finally, the characteristics of the engine/vehicle operating regime tested or the synthetic flow generated in the reactor. Neither the review studies nor the numerical studies validated through experimental data obtained from the literature have been included in Table 1.

4. Numerical Research to Improve CC Knowledge

Numerical simulations are utilized in research in the field of exhaust emissions after treatments because they offer a low-cost analysis of the influence of one or more parameters on the efficiency of catalytic converters and, consequently, on exhaust emissions. The physical and chemical phenomena involved in the performance of CCs can be described using mass, momentum, and energy transport, as well as chemical reaction models.

In general, conducting very accurate simulations is not easy because the complexities of the physical–chemical phenomena involved in the conversion process hinder their practical use in the design of CCs. Simplified models offer reasonable accuracy while preserving the models’ simplicity and computational efficiency.

The reports gathered in this review reveal intense research in developing fast estimation tools with reasonable accuracy. Revised models and simulations usually focus on conditions critical for the performance of CCs, such as cold starts (transient CCs and light-off behavior), conversion (chemical reactions and CC oxygen storage interaction), and flow distribution. Several explain coupled models that consider heat-mass transfer and the chemical reaction. In this sense, the reports summarized here complement the results presented in the previous Section 3.

The studied simulations exhibited differences, mainly in one or more of the following strategies:

• Equations or set of coupled equations to solve: Some authors focused on a specific phenomenon, such as the temperature distribution, and, therefore, solved one transport equation (heat, mass, or momentum). In contrast, other scholars solved coupled models to estimate the light-off and/or the conversion of CCs.
• Assumptions: The simplifications applied and the boundary conditions considered, as well as the methods used to determine the required equation coefficients.
• Temporal and spatial discretization: Compared to static models, dynamic models include temporal discretization. Spatial discretization is usually applied to the length of a CC (1D models) or a longitudinal section of a CC (2D models).
• System parameters and computed responses: Among the most studied parameters are the geometry of the CC, A/F flow rate, and engine operating conditions. The responses computed are temperature, conversion efficiency, and drop pressure, among others.

Table 2 lists the specific simulation works included in this review, with the descriptions focused on their mathematical and numerical features. Most works are related to estimating species conversion, mainly in the transient warm-up during the vehicle’s cold start, and use one-dimensional approximations. These studies address the flow maldistribution faced by simulations of the entire monolith (three-dimensional) using different approximations to reduce computational efforts, such as single-channel models, porous media, and alternative strategies. Nevertheless, they do not include realistic kinetic equations.

In the following paragraphs, we describe the details and specific results achieved using the models and simulations in relation to the research questions (Section 1). Starting with the topic of cold starts, Shen, Shamin, and Sengupta [66] used their model to estimate the light-off temperature of CCs during an FTP test and claimed that at least a substrate temperature of 600 ${}^{\circ }$C is required for reducing cold-start emissions. The simple CC model, proposed by Sanketi, Hedrick, and Kaga [67] to control emissions during cold starts, provided the CC conversion rate function of the air/fuel ratio and substrate temperature. Finally, Yun [25] developed a one-dimensional model of an exhaust system with SAI to identify the optimal operating conditions for reducing the light-off time and HC and CO emissions (Section 3.1).

Regarding the models that focused on species conversion, as shown in Table 2, the reviewed methods generally combined heat and mass transfer models with reduced reaction schemes. An example of coupled heat-mass transfer and chemical reaction is the solution presented by Psyllos and Philippopoulos [68]. Using this model, they noted that conversion depends on the volumetric rate according to the length of the CC, gas inlet temperature, and gas pollutant inlet concentration. On the other hand, for a Pe number greater than 250, there is negligible thermal conduction along the axial direction. Additionally, radiation influences the solid phase temperature; the maximum temperature is attained at 85% of the length of the CC. The model proposed by Aimard, Li, and Sorine [69] enables long-term simulations and can be used to design A/F controllers. It is worth mentioning the discussion by Koltsakis, Konstantinidis, and Stamatelos [56] on the limitations of approaches focused on emissions estimation during a complete driving cycle: a large amount of input data are required, as well as knowledge of the HC composition as a function of the engine temperature and operating condition and accuracy of the light-off and oxygen storage simulation for estimating CO and HC conversion. Besides estimating emissions in TWCs, the authors claimed that their model can be applied to the design of systems for improving the cold start of CCs, investigating the influence of flow distribution on the efficiency of CCs during steady-state operation, and evaluating the behavior of aged CCs. On the other hand, Koltsakis and Stamatelos [70] extended the previous approach using submodels for the main dynamic phenomena (oxygen storage and water–gas reaction) and concluded that compared to the experimental data, the extended model could explain the phenomena better than quasi-static ones. It could demonstrate that the air–fuel ratio oscillation improved CO and HC conversion in rich environments and NO conversion in lean environments, and that the air–fuel scan direction (from rich to lean environments or vice versa) influenced the conversion efficiency. Kumar et al. [71] described a reduced-order model designed for the control and diagnosis of TWCs with the ability to estimate oxidation in TWCs through two parameters: fractional oxygen storage and total oxygen storage capacity. The authors explained that this solution helps assess the age of TWCs and can result in adequate fuel control. Instead of modeling a CC as an independent component, Onorati, D’Errico, and Ferrari [72] integrated a 1D fluid dynamic engine model with a 1D thermal and chemical CC submodel. According to the authors, this solution enables the estimation of the gas composition in the exhaust system and the evaluation of the interaction between the engine and the CC, which can be helpful in exhaust system design. Regarding the chemical models, Pontikakis and Stamatelos [73] explained a method for determining the kinetic parameters of reduced reaction schemes using genetic algorithms. Unlike non-stochastic procedures, the proposed method provides a global optimum, leading to more accurate results. More recently, Bedi and Chauhan [74] presented a model for predicting methane oxidation and used it to compare two different catalysts and the effect of aging. The Pt/$\delta$-Al₂O₃ catalyst showed faster conversion and better behavior during aging compared to the CuO/$\delta$-Al${}_2$O${}_3$ catalyst.

The simulation of flow distribution is also a common topic due to the effect of maldistribution on the performance of CCs. The reviewed reports usually employ Computational Fluid Dynamics (CFD) models. For example, Jeong and Kim [75] used a CFD simulation to study the flow distribution in CCs considering pulsating exhaust gases, noting that maldistribution depends on the height and angle of the cones used to join the substrate to the exhaust system. Diffuser geometry was also studied using a CFD model by Khalil et al. [76], who found that an optimum inlet angle impacts the pressure drop and flow distribution. Martin et al. [77] studied the relationship between flow distribution simulations and measured the conversion and light-off. They concluded that the maldistribution produced by a small inlet pipe diameter reduces the conversion efficiency of CCs in some stages of the tested driving cycles (especially for NO${}_x$), and aging fosters this effect. On the other hand, they did not observe an acceleration in the light-off time. In addition to the external geometry, the internal shape of CCs influences their performance. In this sense, a study by Hayes et al. [38] presented a CFD model coupled with thermal-mass transfer and chemical schemes. The authors concluded that the substrate cell density affects the flow and temperature distribution, the resulting chemical reactions, and the pressure drop. Furthermore, an increase in the cell density increases the pressure drop but reduces maldistribution. Sun et al. [78] studied dual monoliths and also noted better uniformity for high-density cells when the gap width between substrates was increased. They also evaluated the effect of the length ratio of the front/rear monoliths on flow uniformity and conversion efficiency. The authors found that a longer front length improved uniformity and accelerated the light-off. Cornejo, Nikrityuk, and Hayes developed a series of reports [79,80,81,82] using various fast CFD models to understand flow distribution before and after the monolith. Reference [79] focused on a multiscale scheme that extends the turbulence decay in one channel to the converter scale and improves flow prediction after the monolith. In a related report [80], the substrate was modeled as a continuum (porous media), and the authors defined a Reynolds number, depending on the monolith’s wall thickness, to determine when the flow became turbulent (below a value of 100, the flow was mainly steady). This study was followed by two works considering multi-zone models [81,82]. The multi-zone solutions consisted of two and three permeability regions, respectively, and the authors claimed that these multi-zone models were in better agreement with the experimental results compared to the conventional permeability model (one single porous medium). Recently, using a CFD single-channel model, Cornejo et al. [10] explored the effect of cell geometry (cross-sections: square, hexagonal, and triangular) on pressure drop.

Table 2. Main characteristics of reviewed simulation studies.

	Cold Start	Conversion	Flow Distribution	Other Topics
Modeled	Exhaust system with SAI (LGPi engine) [25]; Model for cold-start control [67]; Transient performance of CCs [66]	Conversion in TWCs; Oxygen Storage [56,69,71]; Conversion in TWCs [70,73]; Performance of CCs [68]; Exhaust system (SI 4-cylinder engine) [72]; Oxidation of VOCs [74]	Turbulence [79,80,81]; Pressure drop [10,82]; Flow distribution [75,76,77,78]; Flow and temperature distribution [38]	CC mass transfer [83]; TWC temperature for CC control [84]; Temperature distribution in the exhaust system [85]; Influence of engine misfiring on thermal behavior of CC [86]
Energy Eq.	[25,66,67]	[56,68,69,70,71,72,73,74]	[38]	[84,85,86]
Mass Eq.	[25,66]	[56,68,69,70,71,72,73,74]	[10,38,76,77,78,79,80,81,82]	[83,86]
Moment Eq.	-	[72]	[10,38,76,77,78,79,80,81,82]
Chemical Eq.	Kinetic (13-step reaction mechanism); Oxygen storage (9-step reaction scheme) [66]; Empirical Wiebe profiles [67]; 7-step global chemical reactions (Langmuir–Hinshelwood) [25]	Oxygen storage model [56,69,71]; Langmuir–Hinshelwood [70,72,73]	First-order kinetic [38]; CO oxidation [78]	Reduced-order exothermic reaction kinetics [84]; CO, C${}_3$H${}_6$, H${}_2$ oxidation [86]
Behavior	Transient [25,66]	Transient [68,69,71,72,73,74]; Quasi-Steady [56] Dynamic [70]	-	Dynamic [84]; Quasi-steady [85]; Transient [86]
Dimension	1D [25]	1D [68,69,71,72,73,74]; 2D [56,70]	3D (Porous medium + turbulence BC after CC) [81]; 3D (Porous medium) [38,79,82]; 3D single channel [10,80]	1D [83,85,86]
Scheme	Finite differences [66]	Finite differences [72]	Finite volumes	Finite differences [86]
Software	AMEsim [25]	GASDYN [72]; CTRAN [73]; Matlab [74]	ANSYS Fluent [10,79,80,81,82]; STAR CCM [76]; COMSOL Multiphysic [38]; PHOENICS and STAR-CD [77]
Parameters	Engine conditions [66]; Temperature, A/F ratio [67]; SAI air-flow rate, supply time [25]	A/F ratio [69,70,72]; Engine conditions [69,73]	Length of turbulence zone Re [81]; Diffuser inlet angle [76]; Axial distance [77,78,80]; r/R [79]; Channel geometry [10]	Re · Sc [83]; Engine conditions [85]; Axial distance, misfiring, catalyst content [86]
Responses	Emissions [66,67]; Cumulative emissions [25,66]; Temperature [25]	Temperature [56,68,72]; Pressure [72]; Conversion [56,68,69,70,71,72,73,74]	Turbulence [79,80,81]; Pressure drop [10,76,82]; Flow uniformity index [76,78]	Sh [83]; Temperature [85,86]

Table 2. Main characteristics of reviewed simulation studies.

	Cold Start	Conversion	Flow Distribution	Other Topics
Modeled	Exhaust system with SAI (LGPi engine) [25]; Model for cold-start control [67]; Transient performance of CCs [66]	Conversion in TWCs; Oxygen Storage [56,69,71]; Conversion in TWCs [70,73]; Performance of CCs [68]; Exhaust system (SI 4-cylinder engine) [72]; Oxidation of VOCs [74]	Turbulence [79,80,81]; Pressure drop [10,82]; Flow distribution [75,76,77,78]; Flow and temperature distribution [38]	CC mass transfer [83]; TWC temperature for CC control [84]; Temperature distribution in the exhaust system [85]; Influence of engine misfiring on thermal behavior of CC [86]
Energy Eq.	[25,66,67]	[56,68,69,70,71,72,73,74]	[38]	[84,85,86]
Mass Eq.	[25,66]	[56,68,69,70,71,72,73,74]	[10,38,76,77,78,79,80,81,82]	[83,86]
Moment Eq.	-	[72]	[10,38,76,77,78,79,80,81,82]
Chemical Eq.	Kinetic (13-step reaction mechanism); Oxygen storage (9-step reaction scheme) [66]; Empirical Wiebe profiles [67]; 7-step global chemical reactions (Langmuir–Hinshelwood) [25]	Oxygen storage model [56,69,71]; Langmuir–Hinshelwood [70,72,73]	First-order kinetic [38]; CO oxidation [78]	Reduced-order exothermic reaction kinetics [84]; CO, C${}_3$H${}_6$, H${}_2$ oxidation [86]
Behavior	Transient [25,66]	Transient [68,69,71,72,73,74]; Quasi-Steady [56] Dynamic [70]	-	Dynamic [84]; Quasi-steady [85]; Transient [86]
Dimension	1D [25]	1D [68,69,71,72,73,74]; 2D [56,70]	3D (Porous medium + turbulence BC after CC) [81]; 3D (Porous medium) [38,79,82]; 3D single channel [10,80]	1D [83,85,86]
Scheme	Finite differences [66]	Finite differences [72]	Finite volumes	Finite differences [86]
Software	AMEsim [25]	GASDYN [72]; CTRAN [73]; Matlab [74]	ANSYS Fluent [10,79,80,81,82]; STAR CCM [76]; COMSOL Multiphysic [38]; PHOENICS and STAR-CD [77]
Parameters	Engine conditions [66]; Temperature, A/F ratio [67]; SAI air-flow rate, supply time [25]	A/F ratio [69,70,72]; Engine conditions [69,73]	Length of turbulence zone Re [81]; Diffuser inlet angle [76]; Axial distance [77,78,80]; r/R [79]; Channel geometry [10]	Re · Sc [83]; Engine conditions [85]; Axial distance, misfiring, catalyst content [86]
Responses	Emissions [66,67]; Cumulative emissions [25,66]; Temperature [25]	Temperature [56,68,72]; Pressure [72]; Conversion [56,68,69,70,71,72,73,74]	Turbulence [79,80,81]; Pressure drop [10,76,82]; Flow uniformity index [76,78]	Sh [83]; Temperature [85,86]

Regarding the “other topics” category (Table 2), at least three reviewed works deal with temperature estimations. Oh [86] modeled the thermal behavior of the CC during misfiring. The wall temperature is influenced by misfiring, and below 40%, the melting point of the CC is not attained. The solution by Shailer and Hayden [85] for estimating the thermal behavior of exhaust and after-treatment system components under steady and transient conditions can help design the layout and location of these elements. The authors developed correlations to describe the heat transfer process between the gas and walls. The model enables parametric studies, that is, the effect of operating conditions and dimensions on the temperatures. The recent work of Shah, Premchand, and Pedro [84] proposed a physics model that, through reduced schemes and gas–solid phase-energy balances, estimates the temperature of the CC in real time and, according to the authors, improves estimation accuracy compared to empirical methods. On the other hand, Santos and Costa [83] evaluated different mass transfer correlations and defined a characteristic Sherwood (Sh) number that depends on the chemical species to convert. According to the authors, chemical and internal diffusion limitations can explain the previous dependence and influence the performance of the CC.

Considering the above discussion, further efforts are critical for understanding the influence of factors different from conventional engine conditions that affect the performance of CCs. The internal geometry of a CC is a clear example of a factor that needs to be considered in future studies. In this sense, additive manufacturing could be critical in future investigations due to its ability to produce prototypes with complex shapes at affordable costs.

5. Conclusions

Much of the research related to CCs is focused on reducing cold-start emissions, as they account for up to 80% of total pollutant emissions. Active and passive methods are used to decrease cold-start emissions. Active methods include the use of electrical heaters, fuel burners, air injection, and hydrogen injection. Passive methods include the use of a pre-catalyst, thermal management, and placing the CC closer to the exhaust manifold. All of these systems must be appropriately designed to avoid excessively high temperatures that can lead to durability problems.

The geometrical characteristics and the material of the monolith significantly influence the efficiency of the CC. Some geometrical factors influencing the performance of the CC include the geometric configuration of the substrate channel, such as the hydraulic diameter, channel length, cell density, and wall thickness. The geometry of the whole monolith cross-section also influences the efficiency of the CC. Circular or nearly circular monolith contours have several advantages regarding durability, temperature, and uniform pressure distribution.

New ceramic hollow fiber and foam substrates provide advantages compared to regular ceramic and metallic substrates. Additive manufacturing technologies allow for the production of complex geometries for substrates, resulting in a significant improvement in the light-off temperatures for CO, THC, and NO compared to a conventional honeycomb catalyst design.

Although researchers have achieved significant improvements in emissions reduction through alternative catalyst materials, more effort is required to reach the conversion efficiency levels provided by noble metals.

Other topics studied relate to oxygen storage and catalyst deterioration. According to the reviewed works, the oxygen storage capacity (OSC) of some materials (mainly cerium oxides) present in the wash coating, together with the use of a lambda sensor for the determination of the O${}_2$ concentration, allows the engine control unit to vary the air–fuel ratio (lambda number, $\lambda$) to achieve optimal efficiency in the conversion of CO, HC, and NO${}_x$. The influence of the OSC and air–fuel ratio on the performance of a CC is known and critical in CC design. On the other hand, knowing the CC’s deactivation modes is critical for defining accurate and affordable diagnosis tools.

Finally, this review has revealed a large number of publications devoted to developing models with the aim of improving our understanding of the aforementioned topics. In general, these efforts, whose main characteristics are synthesized in Table 2, highlight the interest in improving estimations in a reasonable computational time.