Renewable production of ammonia and nitric acid

Funding information Bosch-Forschungsstiftung Abstract Decarbonization of the power sector offers ammonia industry an opportunity to reduce its CO2 emissions through sector coupling. Extending from our previous work, we propose a power-based process for ammonia and nitric acid production. The coupling of nitric acid production facilitates highly efficient heat integration between steam electrolysis and the rest of the process. We investigate the economic performance of the production complex through a model-based dynamic optimization approach, considering scenarios with or without incorporation of intermittent wind power as well as deployment of battery storage. In all cases, the wind power integration proves to be economic with a peak-to-base load ratio of up to 2.3. The new process reduces primary energy consumption by more than 13% compared to conventional technologies. However, it is only economically competitive with help of either a low-cost battery storage or a higher carbon price on fossil fuels. The results also confirm the importance of considering process dynamics during process design.


| INTRODUCTION
The rapidly growing share of renewable power generation offers manufacturers opportunities to reduce their carbon dioxide emissions by electrification. 1,2 Within the chemical industry, various molecules and their electrochemical production routes are being investigated. 3 Through sector coupling and flexible operation, industrial processes are capable of providing ancillary services in power markets. 4,5 Thus, they can act as large-scale buffers to help increase capability of the electricity grid to incorporate renewables. 6,7 Today's ammonia production exhibits a high carbon footprint, 8 thus motivating research on sustainable production methods. 9,10 Electrified ammonia production dates back to the 1920s, where the Haber-Bosch process was coupled with a large-scale hydro-powered alkaline water electrolysis. 11 This two-step synthesis route remains a promising option. 9 Research works on this option can be grouped on unit operations, processes and supply chains: (a) on the unit operation level, Cheema and Krewer 12 use steady-state simulations to demonstrate a wide operating range of a three-bed autothermic Haber-Bosch reactor system. Reese et al. 13 show technical feasibility of a lab-scale Haber-Bosch process driven by intermittent wind power. Millet, 14 Petipas et al. 15 and Mougin 16 investigate transient operability of electrolyzers, such as proton exchange membrane (PEM) electrolyzer 14 and solid oxide electrolyzer (SOE), 15,16 and report no additional degradation on the stack level; (b) on the process level, we 17 propose an ammoniabased energy storage process involving the usage of a reversible hightemperature solid-oxide fuel cell. Chen et al. 18 present a novel design of an ammonia synthesis system for a solar thermochemical energy storage system. The new design enables direct production of supercritical steam up to 650 C. Demirhan et al. 10 apply a superstructure framework to optimally configure and dimension ammonia production processes based on renewable resources and investigate the options of PEM and alkaline electrolyzers. Allman et al. 19,20 propose to embed operation information into a design problem in order to estimate operating costs under power dynamics; (c) on the supply chain level, recent works analyze the potential of incorporating distributed wind/ solar-powered ammonia plants in the countrywide fertilizer industry. [21][22][23] Economic advantages have been identified in regions with strong ammonia demand and high availability of wind power.
Investigation on heat integration of renewable ammonia production is rare. Our previous work 17 shows that conducting heat integration between a Haber-Bosch process and a high-temperature SOE drastically enhances efficiency. Furthermore, few publications consider process dynamics and their influence on design, especially when intermittent renewables are considered as feeds. The integration of process dynamics helps account for losses of operating efficiency and hence improves the accuracy of cost estimation, as we for instance showed for concentrated solar power. 24 With the explicit consideration of plant controllers, the flexibility of the designed process can be evaluated in a more realistic way. 25 We propose a power-based acid production complex for ammonia and nitric acid. Heat integration between Haber-Bosch process, SOE, and nitric acid production improves the efficiency as well as the stack utilization of the electrolyzer. We analyze the economic performance of this process through dynamic optimization under variable power inputs as well as different operation strategies. In the following, we first briefly describe the process concept and the models used for the analysis. Next, we introduce case studies and the associated optimization problem formulations. Finally, we present the results of the cases studies and analyze the impacts.

| PROCESS DESIGN
As shown in Figure 1a, an ammonia production process is obtained from modification of the closed-loop ammonia-based energy storage process from our previous work. 17 In the following, we will denote this as the standard concept. The exothermic ammonia synthesis drives steam generation through the heat exchanger HE-1. The SOE is operated above the thermoneutral voltage (i.e., in the exothermic mode) in order to raise up the temperature of the product stream. This enables the upstream heat exchange in HE-2/3.
Based on the standard concept, we extend the renewable ammonia production with coproduction of nitric acid, see Figure 1b. While the coproduction is common industrial practice, it has not been considered for power-based concepts and here, it serves a special purpose. Namely, a part of the high-temperature heat released from the ammonia burner, which is the first stage of the nitric acid production process, is used for supporting the operation of the SOE through the superheater HE-6. There are several advantages of doing so. First of all, the SOE can be operated in a different operating mode, that is, the thermal-neutral or endothermic mode, with a higher utilization factor of the stack. 26 Hence, the costly investment for the SOE is reduced.
Besides this, compared to the case, in which the SOE is operated at the exothermic mode (see e.g., Wang et al. 17 ), the size of the superheater is also expected to be smaller because of an increasing temperature difference at both ends. Moreover, the mass flow rate of the hot stream is now available as an additional control degree of freedom for SOE operation. Similar to a standard nitric acid production process concept utilizes by-products from the ASU. The higher oxygen content of the by-product stream compared to that of the air favors the yield of nitric acid. 27 Within process blocks, heat integration measures are adopted from the literature. 17,27 The detailed process flow sheet is presented in the Supporting Information.

| Case definition
An ammonia/nitric production complex built in the north of Germany is studied. We consider three scenarios of the power supply. In the first one, the production complex is operated at steady state at an optimal point with grid electricity. In the second one, the grid power only provides a constant base load. In addition, the production complex absorbs wind power from an optimally sized wind farm. The wind power supply is a product of the installed peak power and a unit wind power curve. The latter is synthesized from historical wind speed data 35 and includes four representative day profiles. The load of the production complex tracks the total power supply dynamically. The last scenario is an extension of the second one. A battery storage is deployed in order to smoothen the dynamics of the power supply due to wind power. The battery has a self-discharging rate and the charging and discharging processes have an equal half-cycle-efficiency. For more information about the synthesis of wind power curve, please refer to the Supporting Information.
In all three scenarios, the specific production cost is minimized while satisfying an emission-reduction target compared to production based on natural gas (NG) of 1.7 t CO 2 /t NH 3.  41 The financial incentives for the CO 2 emission reduction are considered in the framework of the EU Emissions Trading System (ETS), which sets a cap on the maximum level of emissions for heavy energy-using installations and establishes an installation-level market for emission permits. 42 Currently, the European ammonia industry receives nearly full free allowances. 43 The trading of these allowances is assumed to be an extra revenue for the renewable production complex. The technical and economic parameters are released in the Supporting Information.

| Control scheme
The control scheme serves two purposes: first, the production rate is adjusted so that the power demand of the production complex tracks the power supply and secondly, system states are regulated within the feasible operating range.
For the first purpose, the load tracking is initiated by the SOE because of its high power consumption. The steam feed rate F c/in,E−8 of the SOE is adjusted according to with the production power demand _ W dmd , the set point of total power supply _ W set supp and the load gradient k, which is derived from the maximum ramp rate max j dj avg,E− 8 dt j of the average current density of the stack: where A cell,E−8 represents the total functional surface area of the SOE, F the Faraday constant and u E−8 the steam utilization factor. We select a conservative value of 0.025 A/cm 2 min −1 for the maximum ramp rate to avoid thermal damage because of frequent load changes. 44 The production rate of nitrogen is adapted to the hydrogen production rate according to the stoichiometry of the ammonia synthesis reaction. To be noticed, for a sufficiently large load gradient, the deviation between the set point and the real power supply becomes negligible. This assumption is implemented in this work.
For the second purpose, P-and PI-controllers are implemented: the level and pressure control of the steam boiler, the temperature control of the cryogenic distillation column that is responsible for the nitrogen gas quality, the temperature control of the first ammonia synthesis reactor that helps increase the loop stability, 32

| Optimization problem
The specific production cost C prod per ammonia-equivalent is set as with the time window T, the capital expenditure C capex , the operating expenses C opex , the net ammonia production rate of the production complex _ m prod,NH3 , the production rate of nitric acid in the dry mass

| First scenario: Pure grid power supply
Without incorporation of nitric acid production, the power-based ammonia production is not profitable, even if the sales of CO 2 allowances for 13 million $/year are taken into account. The power-based process reduces primary energy consumption by 12% compared to that of conventional best-available technologies. When incorporating the nitric acid production, the RoI improves by 3 percentage points. This is caused by two factors: first, the capital expenditure on the additional components for nitric acid production is compensated by the cost saving on the SOE, the size of which is decreasing by 30% due to higher power density achieved under the thermal-neutral operation mode; and second, as the SOE is operated at a higher stack temperature around the thermal-neutral point, the energy efficiency is improved. In both process concepts, the heat transfer between the ammonia synthesis and the steam generation is on a similar order, but at a different temperature level because of the difference in operating pressure of the steam boiler. This explains the higher operating pressure of the ammonia synthesis loop in the production complex. Table 2 illustrates the different operating conditions between two process concepts.
The production complex enjoys limited economies of scale under variation of the CO 2 saving target, as shown in Table 3. The main reason is the large share of operating expenses to total costs. Moreover, the most expensive component, the SOE, exhibits a linear cost function of the size.

| Second scenario: Mixed power supply without deployment of battery
The plant is operated dynamically with a peak-to-base load ratio of Namely, the deterioration of operating efficiency of gas compressors at extreme loads is intentionally exploited to increase power consumption of the ammonia block at the peaks, as shown in Figure 2.
More wind power can therefore be integrated and this lowers the operating costs. As a chain effect of the larger synthesis gas flow, the production complex tends to produce less nitric acid, because the synthesis gas releases more heat in the steam boiler and less heat input from the nitric acid plant is needed. Another example that demonstrates the impact of dynamic operation on the selection of the operating parameters is the increasing consumption of the pure oxygen from the SOE and the decreasing absorption of the oxygen-rich by-product from the ASU. This is attributed to different time constants between the unit operations.  Return on investment (%) from 1.7 in the second scenario to 2.3. The lower power cost because of increased wind power consumption compensates the investment on battery and the specific production cost is slightly reduced. As illustrated in Figure 3a, the load peaks of the production complex are smoothened and extreme operating conditions are avoided. One evidence is the increasing operating pressure of the ammonia synthesis loop. The battery is mainly used for shifting the energy within 2 days, as shown in Figure 3b. This could be a result of the chosen wind power curve. As the daily data are selected with high standard variance, consecutive windless days are not considered.
The optimization result is sensitive to the battery cost. As the cost of battery rises from 200 to 300 $/kWh, the optimal battery size shrinks dramatically to 20 MWh/20 MW. At a cost of 400 $/kWh, which is a conservative cost estimation for the future, 48 deployment of battery is no longer profitable.

| Discussion
In order to stay competitive against conventional productions with an average profit margin of 10%, 49   While potential plant investors may consult the grid operator to select a site with best opportunities of wind integration and grid supports, the grid planners may acquire a better foresight on the growing load demand through electrification of the chemical processes.
The plant operation in this work has been simplified by assuming a perfect foresight of exogenous inputs, for example, wind availability.
In real life, the plant operator needs to exploit economic incentives from the intermittent supply and at the same time meet the demand, while coordinate up-and down-stream processes and manage the inventory in between. To achieve this, we expect three essential technology advances: first, a cross-sector, real-time capable information structure that bridges the production, the management and the external players (e.g., grid), as proposed by Backx et al. 53  More open questions and challenges on this topic are reviewed by Daoutidis et al. 56 and Mitsos et al. 7 At the end, it is the plant manager who needs to welcome intentional dynamics and enable the changes.
Besides operation, there are also challenges with respect to process design methods. Based on optimal operation with fixed design, Sass and Mitsos 57 report limitations of quasi-steady operation and underline the importance of applying suitable ramp constraints. In a design optimization problem, it is, however, often difficult to define such realistic ramp constraints a priori. The sizing of the SOE in the second scenario gives a good example: to avoid swings between the endothermic and exothermic operating mode, positive temperature spread over stack is formulated as an operating constraint. Figure 4 shows that the temperature spread drops to zero always at those time

| CONCLUSION
We proposed a renewable ammonia/nitric acid production complex with comprehensive heat integration measures. We investigated optimal designs in various power supply scenarios with explicit consideration of process dynamics and a simple control scheme. We obtained the following major findings: First, the incorporation of the nitric acid production improves the operating condition of the SOE. The stack utilization increases by 30% and the specific production cost is reduced by 3%. The cost savings become even more significant at a higher stack price in the near future. Second, the proposed plant dem-