Fuel recycling feedback control via real-time boron powder injection in EAST with full metal wall

1.   Introduction

In nuclear fusion devices, strong interactions between high-temperature plasma and plasma-facing materials (PFMs) lead to damage to wall materials and deterioration of plasma performance. The extreme flux of fuel particles colliding with the PFMs results in the retention of fuel within the wall materials through absorption, implantation, and co-deposition processes. Such a fuel particle recycling is closely related to the wall retention that involves primarily two processes: short-term retention and long-term retention [1]. The short-term retention is closely related to the recycling flux. The two main mechanisms of long-term fuel retention are identified to be implantation and co-deposition. These mechanisms contribute to the gradual increase of recycled fuel particles during long-pulse discharges. This phenomenon directly impacts plasma confinement and poses a significant challenge to the steady-state operation of long-pulse plasmas in future fusion devices. Therefore, developing effective countermeasures to control fuel recycling is crucial for the stable long-pulse discharges.

Low-Z material injection techniques, as a real-time wall conditioning and ELM-controlling method, have been widely utilized in fusion devices. Lithium and boron powders are the two primary types of low-Z materials used in modern devices. At EAST, lithium powder has been injected since 2015 to assist ELM-free (Edge Localized Modes-free) H-mode discharges with their duration of ~ 18 s (450 times the energy confinement time) [2]. In 2021, with lithium injection, EAST realized a steady-state long pulse lasting over 100 s, along with a core electron temperature exceeding 10 keV [3]. It is also found that the lithium injection has improved wall conditions, consequently decreasing global recycling on EAST experiments [4–7]. NSTX experiments have also demonstrated that lithium powder injection provides effects similar to pre-discharge lithium evaporation coating, consequently improving plasma performance [8]. Another kind of low-Z material, boron powder, has shown to be effective to increase core temperature and reduce turbulent fluctuations on LHD discharges [9]. It is also found that boron powder injection has lowered oxygen content and improved wall pumping capability in ASDEX Upgrade, consequently enhancing the wall conditioning effects obtained from glow discharge boronization [10]. Similar results are also reported on DIII-D [11]. Additionally, it is found on WEST that boron powder injection has been beneficial for improving wall conditions and enhancing plasma confinement in their L-mode discharges [12]. As a consequence of all these experimental results, decision has been made to use boron injection as a “fall-back” wall conditioning method on ITER, with the proposed full tungsten wall condition in case GDB (Glow Discharge Boronization) is not sufficient.

EAST is a fully superconducting diverted tokamak with D-shaped poloidal cross section. The major parameters include major radius R = 1.85 m, minor radius a = 0.45 m, toroidal magnetic field B_T = 2–3.5 T. EAST has been upgraded with ITER-like tungsten divertors. Currently, plasmas can be operated in both single null (SN) and double null (DN) divertor configurations with full metal wall conditions [13]. On EAST, some remarkable experiments, including ELM suppression and enhanced divertor radiation, have been achieved by the boron powder injection using the impurity powder dropper (IPD) designed by the Princeton Plasma Physics Laboratory (PPPL) [14]. Additionally, it is observed that fuel particle recycling decreases with an increase in the amount of boron powder injected, resulting in an increase in the short-term fuel retention. The fuel recycling has decreased by up to 80%, as evidenced by the divertor neutral pressure and D_α line emission [15]. However, the legacy control method on EAST can only inject boron powder at a preset flow rate controlled by a signal generator during a single shot, which cannot meet the requirement of recycling control in long-pulse discharges due to the gradual recycling increase in the later phase. This leads to a deterioration in plasma confinement and stability, including possible plasma disruptions that can severely affect long-pulse discharges [16, 17]. Moreover, surface pre-coating layer erosion and changes in the wall temperature during long-pulse discharges can induce variations in the wall recycling, which evolve with discharge duration and are challenging to predict. Hence, real-time control of the boron powder injection rate to regulate the wall outgassing becomes imperative. To this end, a system that can apply such a technique has recently been implemented on EAST to adjust the powder injection flow rate in real-time based on the level of fuel recycling. This system provides an active particle control technique to achieve stable and high-confined plasmas in long (1000 s time scale) discharges on EAST, which can also provide an important reference for similar boron powder application on ITER.

The details of the implemented fuel recycling feedback control system based on boron powder injection are described in this article, along with experimental results to demonstrate the system’s effectiveness. Section 2 introduces the design and composition of the recycling feedback control system installed on EAST. Section 3 presents the results of active control and feedback control experiments. Finally, a summary is provided in section 4.

2.   Experimental setup

2.1   Actuator for recycling feedback control

The IPD [14], shown in figure 1, is the actuator of the feedback control system. The IPD is installed on the top port of the EAST machine, as shown in figure 2. The IPD injects impurity powder from the high field side. During its operation, impurity powder stored in the primary reservoir gradually slides into the drop tube, driven by a sine-wave signal voltage at a specific frequency. To meet the requirements of injecting various types of powder during plasma discharges, the main chamber contains four primary reservoirs, each designed for lithium, boron, silicon powder, and lithium granules. In the performed feedback control experiments, the boron powders with a particle average size at ~ 150 μm are injected at about 10 m/s velocity driven by gravity, with a lag time of ~ 3 s due to the long distance from the piezoelectric actuator to the plasma. The flow rate of boron injection is adjustable, ranging from 0 to 200 mg/s.

Figure  1.  Schematic diagram of the actuator structure: (a) overall appearance of the impurity powder dropper, (b) top view of the main chamber, (c) side view of the main components.

DownLoad: Full-Size Img PowerPoint

Figure  2.  Poloidal cross section of EAST with D_α arrays and the location of IPD.

DownLoad: Full-Size Img PowerPoint

2.2   Diagnostics for the recycling feedback control

Typically, there are three methods for characterizing fuel recycling: global recycling coefficient, wall retention, and D_α emission. The global recycling coefficient, calculated by particle balance equation [18, 19], provides an overview of fuel recycling, while wall retention indicates the quantity of fuel particles retained on the wall, offering an indirect reference for fuel recycling. However, calculating the recycling coefficient and wall retention involves numerous physical parameters, requiring a considerable amount of real-time diagnostic signals, and some physical quantities cannot be obtained in real time. The emission of D_α is utilized to measure the ionization rate of neutrals resulting from the deuterium ions off the surface of plasma-facing components [20]. D_α line emission can partly reflect the current local recycling level. Therefore, the D_α line emission signal is chosen as an indicator of fuel recycling due to its high instantaneous nature. However, its use as a quantitative descriptor of recycling is limited, and further research is needed to improve its quality or to develop alternative descriptors.

The D_α emissions in upper divertor, mid-plane, and lower divertor regions are measured by the three arrays of the filterscope system in EAST machine [21]. Each array has 13 channels, as shown in figure 2. Based on the observations of D_α emission signals from the multiple channels, it is found that the D_α emission signal from the sixth channel (the blue line in figure 2) in the lower divertor region is relatively stable and reliable, and it shows a noticeable response to the increase in fuel recycling. In addition, use of the channel in the lower divertor region is the most logical choice as we plan our experiments in lower single null configuration. In future research, we will test the performance of this channel for double null and upper single null plasmas.

2.3   The feedback control system

The main components of the recycling feedback control are depicted in figure 3, including the filterscope system, the PCS (Plasma Control System), a signal converter, and the IPD. The filterscope system measures intensity of D_α emission in real time, serving as an indicator of the fuel recycling level in the feedback control system. The D_α emission signal is then transmitted to the PCS, where it undergoes initial filtering. Subsequently, deviation from the target D_α emission intensity is calculated based on the real-time measurements and filtering. Utilizing this deviation and applying the PID algorithm, the PCS calculates the feedback control voltage value and outputs the control signal. The signal converter then converts it to a sine-wave signal. Upon receiving the sine-wave signal, the IPD’s piezoelectric crystal begins to vibrate. Finally, the boron powder, covering the surface of piezoelectric crystal, is injected into the plasma.

Figure  3.  The flowchart of recycling feedback control on EAST.

DownLoad: Full-Size Img PowerPoint

If the D_α emission intensity exceeds the target level, the PCS increases its output voltage in response to the growing deviation, which in turn raises the boron powder flow rate. Consequently, this action enhances the mitigation effect on D_α emission. Subsequently, the D_α emission intensity gradually decreases, moving closer to the target level. Conversely, if the D_α emission falls below the target, the PCS sets the output voltage to its minimum, nearly halting the boron powder injection. Utilizing this method, the system achieves automatic adjustment of boron powder flow rate within a single discharge, thereby implementing fuel recycling feedback control.

2.4   Signal converter

The IPD initiates the powder injection only upon receiving a specific sine-wave voltage. However, the PCS is not designed to generate sine-wave voltages. Consequently, a signal converter is designed to transform arbitrary wave signals into sine-wave signals, which can now be recognized by the IPD. The signal converter, with its overall structure illustrated in figure 4, is primarily composed of three parts: the signal input module, the signal conversion module, and the signal output module. The signal input module is equipped with an isolation amplifier to ensure the reliability and accuracy of input signals. Additionally, a voltage divider circuit is used to reduce the input voltage signals to a level suitable for the microcontroller’s input range. The signal conversion module, which is the heart of the signal converter, converts the input wave signal into a sine-wave signal of a specific frequency.

Figure  4.  Overall structure of the signal converter.

DownLoad: Full-Size Img PowerPoint

The signal output module is tasked with delivering the converted sine-wave signal. A filtering circuit is utilized to remove noise from the output analog signal. In this study, a bandpass filter is utilized to permit a specific range of frequencies, which is composed of a low-pass filter and a high-pass filter. Specifically, the high-pass filter is designed with a resistance of R_HP of 18 kΩ and a capacitance of C_HP of 0.1 μF, resulting in a cutoff frequency calculated by the equation ${F}_{\mathrm{H}\mathrm{P}}=\dfrac{1}{2{{\text{π}} }{R}_{\mathrm{H}\mathrm{P}}{C}_{\mathrm{H}\mathrm{P}}}$. Similarly, the low-pass filter is designed with a resistance of R_LP of 36 kΩ and a capacitance of C_LP of 1 nF, leading to a cutoff frequency using the same formula. By utilizing these parameters, we have established a frequency band ranging from 88 Hz to 4.5 kHz, satisfying the requirement of planned experiments.

Ultimately, the voltage signal, which has been adjusted in the signal input module, is amplified by the operational amplifier circuit before being transmitted. Consequently, the resulting sine-wave signal is then sent to the IPD system. Through the integration of the PCS and the signal converter, a sine-wave signal featuring variable amplitudes is generated within a single plasma discharge.

3.   Experiment results

3.1   Active fuel recycling control

To investigate how the varying control voltages from the PCS and their corresponding boron flow rates affect the D_α emission, a dedicated series of active feed-forward recycling control experiments at various flow rates is conducted before implementing the feedback control. Figure 5 illustrates the evolution of D_α emissions in four discharges, each with own distinct PCS control voltage: #125380 at 0 V, #125386 at 0.4 V, #125385 at 0.9 V, and #125382 at 1.2 V. The basic plasma parameters are: plasma current (I_p) ~ 300 kA, electron density (n_e) ~ 2.8×10¹⁹ m⁻³, LHW power ~ 1 MW, ECR power ~ 1.7 MW, and a lower single null configuration.

Figure  5.  Recycling active control with boron powder: (a) accumulated mass of injected boron powder, (b) B V emission from EUV diagnostic, (c) the plasma stored energy W_mhd, (d) the total radiation power P_rad,tot, (e)‒(h) D_α emission with different PCS feed-forward voltages.

DownLoad: Full-Size Img PowerPoint

The duration of boron injection ranges from 2 to 5 s in these discharges. Figure 5(a) illustrates the cumulative mass of boron powder for the four discharges, corresponding to their respective flow rates of 0 (black line), 2.05 (blue line), 13.02 (yellow line), and 44.18 mg/s (red line). Figure 5(b) depicts the variation in B V emission intensity with varying PCS output voltages. For the reference shot #125380 at 0 V (black line), the B V emission signal, measured in arbitrary units, stabilizes around the value of 10 due to the pre-discharge boronization. This value serves as the reference point for assessing the relative changes in B V emission under other discharge conditions. For the discharges with voltages of 0.4, 0.9, and 1.2 V, the maximum values of B V emission signal are approximately 20, 130, and 230, respectively. These values represent relative increases of approximately 100%, 1200%, and 2200% over the baseline value. These increases indicate that the amount of boron powder, effectively contributing to the B V emission signal, is augmented as the PCS output voltage is elevated. Figures 5(e)–(h) show the evolution of D_α emissions in the lower divertor region with different PCS output voltages. In reference discharge #125380, the D_α emission is ~ 0.35 at the 5 s mark, used as the baseline for comparison with other discharge conditions. In the discharges with voltages of 0.4, 0.9, and 1.2 V, the D_α emissions eventually decrease to 0.26, 0.11, and 0.01, respectively. These decreases correspond to relative reductions of about 25%, 68%, and 97% from the initial value observed in discharge #125380. This trend suggests that the D_α emission can be more effectively controlled by increasing the PCS output voltages.

In plasmas, intensity of D_α is indicative of the rate of neutral particle recycling. Some of these neutral particles are generated through interactions between the plasma and wall. Introduction of boron powder serves to capture fuel particles, thereby suppressing the release of fuel particles from the wall by chemical reactions or co-deposition [15]. At a very low boron powder injection voltage of 0.4 V, the resulting boron flow rate is minimal, leading to a negligible effect on D_α. This implies negligible impact on fuel recycling. However, as the voltage increases beyond 0.9 V, a significant reduction in the fuel recycling is evident, as demonstrated by a marked decrease in the D_α emission. As depicted in figure 5(g), there is a gradual increase in the D_α emission that correlates with the reduction in boron injection as shown in figure 5(b).

The sudden drop at about 7 s is caused by the cessation of ECRH1 and ECRH3. With the reduction in heating power, there is a diminished supply of fuel particles available for ionization, consequently leading to a reduction in the D_α emission level. In #125382, the disruption is attributed to an intense increase in radiation (figure 5(d)), caused by the excessive injection of boron powder.

Figure 6 illustrates the percentage changes in D_α emissions across the upper divertor, midplane, and lower divertor regions during the boron powder injection. Understanding the proportion of D_α reduction at various PCS output voltages is crucial for real-time adjustment in feedback control. The D_α emission intensity at the 5 s mark is selected as the baseline. Despite the varying PCS control voltages, consistent downward trends in the D_α emissions are observed across different positions at a given voltage. This indicates that local changes in D_α at the lower divertor region can reflect overall changes in D_α. Therefore, the adoption of a single channel of D_α measurement at the lower divertor region seems to provide a valid representation of the overall recycling process.

Figure  6.  Reduction in D_α from (a) upper divertor region, (b) midplane, and (c) lower divertor region in shots with 0 V (black line, #125380), 0.4 V (blue line, #125386), 0.9 V (yellow line, #125385) and 1.2 V (red line, #125380).

DownLoad: Full-Size Img PowerPoint

At 6 s in the upper divertor region, the D_α emission intensity is reduced by 2%, 4%, 32%, and 54% corresponding to the respective PCS voltages of 0, 0.4, 0.9, and 1.2 V. By 7 s, these reductions increase to 10%, 32%, 66%, and 92%, respectively. These variations indicate that the rate of D_α reduction also differs under different voltages, averaging about 16%, 33%, and 46% per second for the respective voltages of 0.4, 0.9, and 1.2 V. The fluctuations indicated by the error bars in the percent change of D_α are relatively minor compared to the overall percent change in D_α resulting from the injection of boron power. This observation suggests that the active feedback system maintains a high level of effectiveness, as the fluctuations do not significantly impact the system’s performance. Therefore, it is reasonable to speculate that the accumulative quantity of boron powder does impact the magnitude of the D_α reduction and the flow rate of boron powder can influence the speed of this reduction. These results provide important data support for selecting a threshold of the PCS control voltage in fuel recycling feedback control experiments at EAST.

Based on the results of the aforementioned experiments, it is observed that boron injection at 0.4 V has only a slight effect on D_α emission while boron injection at 1.2 V causes plasma disruption due to excessive boron powder. Therefore, the minimum and maximum PCS output voltages are set at 0.4 and 0.9 V respectively in the feedback control experiments.

3.2   Recycling feedback control

In the EAST 2023 spring campaign, successful fuel recycling feedback control has been achieved in shot #125392. Figure 7 illustrates the evolution of D_α emissions in this shot with its PCS feedback control compared to the reference shot #125393. The basic plasma parameters in these two discharges include: I_p ~ 300 kA, n_e ~ 2.8×10¹⁹ m⁻³, LHW ~ 1 MW, ECR ~ 1.7 MW, and a lower single null configuration. The target D_α is pre-programmed to 1.5 times the value measured at the 3 s mark. In shot #125392, the D_α target value is about 0.466. As shown in figure 7, the PCS adjusts the output voltage in response to the D_α emission exceeding the target value of 0.466 from 3 s to 7.5 s. Due to the long delay of boron injection, the B V emission starts to increase at 6.5 s. Once boron is introduced into the plasma, the D_α emission starts to decrease, ultimately reaching the target value by 7.5 s. As the D_α emission decreases from 6.5 to 7.5 s, the PCS output voltage is reduced from 0.9 to 0.4 V aligning with the real-time deviation from the target. Once the D_α value falls below the target after 7.5 s, the PCS maintains a minimum voltage of 0.4 V. Ultimately, the feedback control ends at 9 s. It should be noted that the B V emission remains at high concentration even after the feedback control stops, due to the delayed effect of injection. While these results indicate that the D_α emission has successfully been controlled through the newly-developed feedback control system, it is important to note that the prescribed D_α level is undershot because of the 3 s time lag. To address this limitation, future work will explore the implementation of a 3 s filter function to the actuator signal. Additionally, for instances of short-term plasma excursions, we will assess the feasibility of employing a gas feed system as a supplementary actuator.

Figure  7.  Results of fuel recycling feedback control in discharge #125392 (red line) and comparison with reference shot #125393 (black line, almost no boron): (a) PCS output voltages, (b) actual D_α and target D_α, (c) B V emission.

DownLoad: Full-Size Img PowerPoint

Figure 8 shows the evolutions of D_α and neutral pressure, both used as indicators of fuel recycling, in both lower and upper divertor regions in shot #125392 with boron powder feedback injection. The central W concentration (W UTA in figure 8(d)) gradually increased during the boron injection. This increase suggests that the introduction of boron atoms may lead to more W sputtering, possibly due to the greater mass of boron atoms compared to that of deuterium atoms. The plasma current, density, and plasma stored energy remain stable during the boron powder injection from 3 to 9 s, indicating that the flow rate of boron powder at this voltage range can maintain good plasma confinement performance. As the B V emission gradually increases from 6.5 to 9 s, the D_α emission concurrently decreases by 50% and 57% in the upper and lower divertor regions. In addition, the neutral pressure, measured by vacuum gauges in the upper and lower divertor regions, also decreases simultaneously. These results demonstrate that fuel recycling can be effectively controlled through the boron powder feedback injection, which eventually works to maintain good plasma confinement performance in long-pulse discharges.

Figure  8.  Evolutions of basic parameters in shot #125392 with recycling feedback control: (a) plasma density n_e and plasma current I_p; (b) plasma stored energy W_MHD; (c) total heating power P_heating; (d) boron line emission B V and core W emissions W UTA; (e) D_α emission in the lower and upper divertor regions; and (f) neutral pressure in the lower and upper divertor regions.

DownLoad: Full-Size Img PowerPoint

4.   Summary and discussion

A new fuel recycling feedback control system utilizing boron powder injection has been successfully developed in EAST with a full metal wall. This system comprises the filterscope system, the PCS, a signal converter, and the IPD. The D_α emission, measured by the filterscope system, serves as the indicator of the fuel recycling level. Relevant algorithms for feedback control are implemented within the PCS. Additionally, a signal converter has been designed and implemented to convert the PCS’s wave signals into sine-wave signals.

A series of dedicated active control and feedback control experiments has been carried out using this newly-developed system in EAST. The results of the active control experiments demonstrate that D_α can be more effectively controlled by increasing the PCS’s control voltages to inject more boron powder under the full metal wall condition. In the feedback control experiments, it has been demonstrated that fuel recycling is well controlled by the newly-developed feedback control system.

The performance of this system is influenced by many integrated and complex factors, including powder flow rate, injection location and dimensions, powder size, system response sequence and time, feedback system real-time response dynamics, etc. Previous experimental results [6, 22, 23] indicate that the flow rate of boron powder is a particularly significant factor in fuel recycling control. The system’s performance also depends on the response time, which is currently delayed by the use of gravitational injection method. To minimize this delay, we are exploring the feasibility of implementing a pneumatic method to facilitate rapid introduction of boron powder into plasmas. Furthermore, to refine the actuator’s response and prevent overshoots and undershoots, we are considering a convolution of the derived actuator signal with a moving 3 s filter function, instead of an instantaneous response. Additionally, a gas feed system is considered to serve as an auxiliary actuator for short-term plasma excursions.

Based on the study introduced in this article, exploring more suitable real-time representative quantities of fuel recycling level is found to be an essential step in the successful application of feedback control system because the D_α emission signal provides comprehensive information beyond just fuel recycling. Therefore, the preliminary research presented in this article is expected to serve as an important reference for real-time control of fuel recycling in future fusion devices, which will operate under improved plasma parameters and long-pulse discharge conditions.