The steel manufacturing route at the plant of study implements the Direct Reduction – Electric Arc Furnace (DR-EAF) route. The plant imports the iron ore pellets and processes them to manufacture a wide range of products including reinforcing bars, wire rods, steel billets, and heavy sections. The plant was commissioned in 2 phases with each producing about 1.6 million tons/year of directly reduced iron (DRI) and about 1.4 million tons/year of steel. Both phases use an external reformer to produce synthesis gas (CO and H₂) which is fed to the direct reduction shaft for reducing the iron ore into sponge iron. This is followed by secondary steelmaking in the EAF and ladle furnaces. The DRI technology utilized at the steel plant under study is ENERGIRON III, which is the product of a collaboration between the two companies, Tenova and Danieli. Tenova, headquartered in Italy, is a worldwide partner for sustainable, innovative, and reliable solutions in metals and mining industries, whereas Danieli is an Italian supplier for equipment and physical plants to the global metal industry. This technology features an external steam reformer which is placed off-line from the process gas recirculation circuit for the following reasons [19]:

• The reformer catalyst is not affected by contaminants coming from the pellets (sulfur and other poisons);

• The reformer catalyst is not affected by upset conditions in the process gas, such as the CO₂ and CH₄ content.

Figure 4.

ENERGIRON III direct reduction process and HYTEMP system [19]

The synthesis gas from the external reformer unit is first passed through an industrial humidifier that maintains it under specific humidity level preventing any static electricity build-ups and preserving its properties. After that, the synthesis gas is heated to reduction temperatures, using a fired heater, and then fed into the shaft furnace. The exhaust gas (top gas) leaves the furnace and passes through a top gas recuperator for heat recovery, followed by an H₂O removal stage where H₂O is separated and removed from the top gas, it can later be used to feed the fired heater. The top gas is then compressed and fed into a Vacuum Pressure Swing Adsorption (VPSA) process where CO₂ is removed, and the top gas is finally recycled back. Hot DRI (HDRI) exiting the furnace is delivered to the HYTEMP system, developed by Tenova's HYL technologies.

This process uses nitrogen as the transport gas, and utilizes pressure build up rather than velocity of the gas for the transport process. This is to avoid degradation of the DRI. When the HDRI reaches the EAF bins tower, the gas is separated and recycled back into the process.

Table 2 summarizes the operating parameters of the ENERGIRON direct reduction process and Table 3 contains the chemical analysis of DRI produced at the respective plant. Both these tables are used to validate the results of the simulation model produced.

The EAF technology employed at the plant of study is a DANARC module (manufactured by Danieli) with a 150-tons capacity and an average tap-to-tap (TTT) time of 46  min per batch of HDRI. It is mounted with 5 oxygen and 3 carbon injectors. The EAF is also mounted with a cooling unit to maintain the structure and refractories within safe operating limits. This is done using water circulated through pipes that form a panel. Table 4 summarizes the operating parameters of the EAF under study.

Given the batch nature of electric arc furnaces and their fluctuating gases, energy storage systems are required to output a continuous flow of thermal energy. A commonly considered approach for storing sensible heat is the use of solids [20], and High Temperature Concrete is consequently a viable alternative. The solid storage medium would have a number of pipes allowing heat transfer in between itself and the heat transfer fluid passing through. However, the creeping of concrete need to be taken into account as it results in slits in between the oil pipes and the concrete block that will degenerate the heat conduction. Although such issues are out of the scope of this current study, they are to be considered in future extensions of this work.

The second step of the proposed method is related to the process mapping. In this section, we present the mapping of the studied manufacturing plant, its equipment, requirements, processes parameters and its GHG emissions to get a more comprehensive understanding of the steel manufacturing processes.

The following schematic presented in Figure 5 gives a high-level overview of the plant studied.

Figure 5.

Overview of plant studied

This framework is then integrated into an ASPEN Plus model to simulate the flows in a steady state simulation. The results of the simulation help us identify the most energy intensive process. Simulation-based investigations would take place to evaluate the amount of energy that can be recovered by fully utilizing the waste heat exhausted by the EAF.

To achieve the set-out objectives of this study, we first simulate the DR-EAF steel-making route in the steady-state process simulation software, Aspen Plus ® V10. For a given process design and with an appropriate selection of thermodynamic models, Aspen Plus uses mathematical models to predict the performance of the process. This simulation is carried out to consolidate the mass and energy flow of the manufacturing process, and helps in identifying ways to improve its energy efficiency. Simulation of the existing plant includes the following major sub-processes: steam reforming of natural gas to produce synthesis gas (H₂ and CO), direct reduction for the reduction of iron ore pellets, and secondary steelmaking in the electric arc furnace (EAF). This analysis has helped in identifying the EAF as the most energy intensive.

Using a prior, built-in model [21] for steam reforming as a basis, we increase the capacity to produce enough synthesis gas for the scale of the steelmaking facility and expand further on the model to incorporate the remaining subsystems of the plant. To simulate the performance of the process, Aspen Plus requires inputs such as a list of all components taking place in the system, features of inlet streams (e.g. Composition, temperature, pressure, and flowrates), chemical reactions taking place, and appropriate unit models. The outputs are the characteristics of intermediate and final flows, such as chemical compositions of sponge iron, molten steel, slag etc., together with the energy consumption and calculated operating conditions of temperature and pressure resulting from the process. Figure 6 illustrates the process model developed. A step-to-step explanation of every process within the developed model of the plant studied is disclosed in the sections following Figure 6.

Figure 6.

ASPEN process model

The model developed helped to understand, trace and further comprehend the flows and comportments of the individual processes within the considered plant. Nevertheless, as there was a lack of actual data from the field, being able to have a reference frame was an absolute necessity. Hence, the main objective of developing such a model was to monitor both, the high energy consumption and process related emissions of each stage and cross reference them with the literature and resources used in further steps. Having this model as a reference assured that all values taken from literature are consistent and in line with what is expected.

The next sections present the description of each step in the manufacturing process of the studied plant, and as presented in Figure 5.

The third step of the proposed method is related to the investigation of WHR techniques that could be implement at the EAF process. Two WHR models were selected and their simulations were developed to evaluate their performance. The two scenario models are: an Oil Loop with an Organic Rankine Cycle (ORC), and a Thermal Energy Storage (TES) made of a concrete block with an ORC. Unlike the conventional water-steam power generation cycles, power plants that are based on ORC technology are less demanding in terms of auxiliaries, safety systems, maintenance and operating costs [22]. The models are simulated using two software: SimTech's IPSEpro and Siemens's STAR-CCM+.

Due to the unavailability of actual data from the field of the considered plant, the input signals were based on the off-gas temperature and flow rate profiles from a similar EAF and presented in Figure 7 [23].

Figure 7.

The reference cycle [23]

The EAF used here has a production capacity of 80-90 tons per TTT cycle. Although this EAF is relatively lesser in term of its capacity to the one installed in the studied plant, its average TTT time, mass flow and temperature are comparable to the 150 tons EAF considered.

The initial states were disregarded in order to avoid any irregularities and the considerations were set to include the profiles from t = 5.4  min onwards. This was acknowledged as the reference cycle to be considered in the characterization of the waste heat recovery process.

Simulation of the oil loop with an ORC

The first WHR model was simulated using IPSEpro, a software system for calculating heat balances and simulating processes. Since Aspen Plus only simulates in steady-state, IPSEpro was used to get a more comprehensive idea of the transient influence on the suggested model and gain a deeper understanding of the dynamic responses of its processes.

Figure 8 below illustrates the process model for an intermediate oil loop that according to [24] would be able to convert the exhaust heat fluctuations in temperature into fluctuations in mass flow rate. This is achievable by maintaining the oil temperature as constant while controlling its mass flow rate [24]. The heated oil then moves into the Organic Rankine Cycle (ORC) evaporator heat exchanger in order to help evaporate its working fluid. The oil used as a heat transfer fluid is Therminol VP-1, an ultrahigh-temperature vapor/liquid phase fluid that has been extensively researched and particularly recommended for such purposes [24].

Figure 8.

IPSEpro waste heat recovery Model 1

ORC are recommended as a valued technology for waste heat recovery at low and medium temperatures, to the purpose of producing electricity in some cases or combined heat and power (CHP) in others [24]. An ORC is made of four key components: an evaporator, where its working fluid is vaporized; a turbine, where the subsequent thermal energy of the fluid is converted into power; a condenser, where the fluid is condensed back into its initial liquid state; and a pump, where the condensed fluid is impulse to the high evaporation pressure [24]. There has been plentiful of organic fluids that could be used to operate an ORC. However, choosing the appropriate working fluid for a particular WHR application depends on several factors including heat source characteristics, performance, environmental, operational, and economic aspects which has been discussed extensively in the literature. Several studies concluded that the optimal selection of an organic fluid is based on a correlation between the heat source and the critical temperature of the working fluid [25]–[29]. In this study, several fluids have been examined from the literature including R245fa, R245ca, R236ea, Cyclohexane, Cyclopentane, Isopentane, and Toluene considering the operational constraints such as maintaining a minimum pinch temperature of 10 °C, cooling water outlet temperature, and turbine inlet vapor quality of 1.

Simulation of TES with an ORC

The second WHR model was also simulated using IPSEpro. However, the Thermal Energy Storage medium was first simulated in STAR-CCM+, a commercial Computational Fluid Dynamic (CFD) based simulation software, extensively used for the simulation of products operating under real-world conditions. However, in order to smoothen the electrical power output profile and get a more stable source of electricity, the gas/oil heat exchanger shown in Figure 8 is replaced with the Concrete Thermal Energy Storage (TES) block seen in Figure 9. It is stated in [30], that utilizing thermal storage units is essential when dealing with intermittent heat production technologies. In addition, and according to [31], the implementation of High Temperature Concrete as a storage medium has demonstrated less environmental impact, even though other TES alternatives such as Molten Salts and Phase Change Materials (PCM) have had a higher value of embodied energy. Furthermore, the cost of the TES material is directly proportional to the intricacy of its fabrication procedure and although this is beyond the scope of this study, it is worth stating that High Temperature concrete or ceramic is significantly less expensive in terms of the cost per ton than the other materials [31].

Figure 9.

Geometry scene of the sensible TES block [32]

The design of the solid block was taken from a previous investigation [32], redeveloped and scaled up to meet the higher temperature and mass flow profiles considered in this study. The reconfigured solid block is 8×8x6  m in dimensions of width, length and depth respectively with 13 gas pipes and 38 oil pipes.

With its transient temperature and mass flow profiles, the EAF exhaust gas enters the gas pipes, transferring the energy in its heat to the solid block that will help damping its fluctuations by getting the heat distributed across its volume. The block will further transfer the heat in a level and smooth manner to the counter flow heat transfer fluid (HTF), Therminol VP-1. The HTF mass flow rate is kept constant. The pipes distribution through the block can be seen more clearly in Figure 10.

Figure 10.

STAR-CCM+ mesh scene of the inlet surface

After passing through the block, the HTF moves into the ORC. This step is done using the same ORC process as the IPSEpro simulation developed for the previous model. The values of the oil temperature and mass flow are taken from the STAR-CCM+ simulation outcome and treated as an input profile for the ORC. Table 5 shows a summary of the CFD setup.

Table 5.

CFD Model Set-up and Boundary Conditions

Physics	Model
Dimensions	3D
Time	Implicit transient model
Energy	segregated flow and solid energy
Equation of state	Ideal gas
Turbulent	κ-ε, Reynolds-averaged Navier-Stokes, realizable κ-ε two-layers, exact wall distance, two layers all y+ wall treatment
Radiation	Participating medium and Surface to surface, gray
Fluid properties	Vary with temperature
Solid properties	Vary with temperature
Walls	No slip condition pipes, adiabatic walls
Mesh	Polyhedral – finite volume
Mesh cells	1.8 million
Prism	2 prism layers with 1.1 expansion rate

Derived from the ASPEN steady-state process model developed, the pie chart in Figure 11 illustrates the total electrical energy requirement of the steelmaking plant, with the EAF requiring most of it.

Figure 11.

Total electrical energy requirement

Figure 12 below is a Sankey diagram representing the energy flow in and out of the EAF for an operation with 100% Cold DRI (CDRI). A significant heat is lost in the form of exhaust gases from the EAF, reaching a value of 258 kWh per ton of liquid steel produced, which is almost 50% of the total electric energy required by the process. As such, recovering this thermal energy internally for preheating or power generation can substantially increase the energy efficiency of the process.

Figure 12.

Sankey diagram of energy flow in and out of EAF per ton of liquid steel (kWh/t)

Table 6 illustrates the working limits and conditions for the ORC. Among the fluids used in this study, both R245fa and Cyclopentane matched the environmental and operational conditions with the highest performance.

Upon keeping a constant oil temperature, the resulting fluctuations in the mass flows of the oil loop and ORC fluids can be seen in Figure 13 below. As expected, they react to the exhaust gas fluctuations and replicate its behavior in order to adapt to the temperature constrains assigned. However, it can be noticed that the effect in terms of mass flow fluctuations on the ORC was much higher than in the Oil loop. Also, it is worth noting that the impact of fluid selection on mass flow fluctuations in the cycle is significant when comparing the high mass flow rate and high range of fluctuations with R245fa as opposed to using Cyclopentane as a working fluid. Consequently, Cyclopentane was investigated in a second implementation as an alternative ORC working fluid. In [33], laboratory experiments were executed to establish the maximum operating temperature for Cyclopentane as an ORC working fluid and both have proven that it is stable at the temperatures in which the studied system is operated in.

Figure 13.

Simulation model's mass flows

Figure 14 displays the power output using both R245fa and Cyclopentane with mass variation. The power fluctuation ranges between 350 to 2400 kW with an average of 1404 kW for R245fa whereas it ranges between 460 to 3150 kW with an average of 1862kW for Cyclopentane. This high fluctuation of power due to the varying mass flow would negatively affect the efficiency of a conventional turbine thereby makes it difficult to rely on such a system. Furthermore, having an appropriate design point of the heat exchanger and a turbine to operate in an off-design mode is a major challenge that would make the system unstable and impractical due to the high range of fluctuations.

Figure 14.

Simulation model's generated power

By maintaining a constant oil mass flow, the resulting temperature profiles of the Concrete block and the HTF passing through it are presented in Figure 15. The CFD based block was simulated from a presumed ambient temperature and for around 4  days of physical time. In order to provide the required heat source for this model, the EAF exhaust gas reference cycle considered in the previous model was repeated and reconfigured as a continuous recurrent flow of heat from a plant that is running continuously for 4 consecutive days. The model was simulated for 4  days to be able to identify the time required to preheat the block after it starts operating.

Figure 15.

TES block resulting temperatures

It can be seen that the fluctuations in temperature no longer exist and instead the temperature profiles are increasing in a steady manner. However, it has been noticed that it took more than 2  days to heat the block to the point where the HTF passing through it can result in the ORC operating and generating electrical power.

The exhaust gas and HTF were entering the block at the average temperature of 700 °C and 70 °C respectively. This can be seen in Figure 16 along with the heat flow and distribution across the block. During the simulation time, the Block average temperature and the Oil outlet temperature reached a value of 261 °C and 347 °C respectively. It is also noticed from the gas outlet contour that the gas pipes were fairly distributed, and this could be deduced from the evenly temperature reduction of the pipes. However, the temperature gained by the block is concentrated in the middle of the block reducing in higher temperatures of the oil outlet. According to the behavior of the obtained values, and if the exhaust gas keeps flowing through the assigned block pipes, the temperatures are anticipated to keep rising in a smooth manner until reaching a steady state value.

Figure 16.

TES adjacent surfaces temperatures

Figure 17 shows a cross-section in the middle of the TES cutting through the gas pipes and the solid concrete block. The temperature gradient in the mid-section varies between 366 °C to around 437 °C. This shows a good indication on the depth of the block although a slight temperature reduction near the oil outlet side.

Figure 17.

TES middle cross-section

Since using Cyclopentane as the ORC working fluid in the Oil loop model has displayed better performance, it was straightforwardly considered as the working fluid for the ORC in the TES model. As can be seen in Figure 18, for the simulation duration in which the HTF was able to evaporate the ORC working fluid and generate electrical power, the mass flow increased steadily in response to the increase in the HTF temperature.

Figure 18.

Model's ORC mass flow

Consequently, the power generated at the Turbine outlet presented in Figure 19 had an analogous response as well. The outlet power kept increasing throughout reaching a value of 2108 kW at the end. Again, if the exhaust gas keeps flowing through the block for an extended time period, the Turbine outlet power is also anticipated to keep rising until reaching a steady state value that correspondence to the block average temperature and the HTF outlet temperature steady state values.

Figure 19.

Model's ORC generated power

In a continuous charge of the EAF with a production capacity of 80-90 tons per TTT cycle, the average power of the off-gas from the EAF is about 7786 kW per cycle. Some of this is captured in the cooling water used in the EAF, and the remaining majority is captured by the waste heat recovery system. This excess heat can be re-used for water heating onsite, exported to other industrial plants, or can be utilized for district heating systems. If its temperature is low, the waste heat can be upgraded with heat pumps to steam for use onsite as mentioned, whereas if it is high, it can be used to generate electricity onsite [34]. The latter was the assumption for this study. Table 7 summarizes the potential of energy recovery in an EAF process with the WHR options considered in this study.

Using an Oil Loop coupled with an ORC, the average output power achieved during the considered time within 1 heat cycle of the EAF is 1862kW, indicating that up to 24 percent of power in the off-gas can be recovered and used to generate electricity. However, the power profile was accompanied with concerning fluctuations that may cause a significant vibrant operation problems and possible ORC components corrosion.

On the other hand, using a TES coupled with an ORC, within a simulation period of 4  days, an output power of 2108 kW was achieved, stipulating 27 percent of power in the off-gas can be recovered and used to generate electricity. Nevertheless, the last-mentioned value is anticipated to rise and reach a much higher steady state value in an extended period of time making it more stable and reliable for use onsite. To improve this efficiency further, it is important to consider a more accurate design specification of the TES Concrete model as the dimensions of the one developed in this study were straightly scaled up and used to prove the effectiveness of the proposed WHR model.

Finally, although the 80-90 tons EAF contemplated in the transient simulation of the WHR models is relatively lesser in term of its capacity than the 150 tons EAF considered in the steady-state simulation of the steel manufacturing plant, their average TTT time, and more importantly mass flow and temperature profiles are comparable and therefore the proposed WHR model is applicable to be used for both.