Know-How of the Effective Use of Carbon Electrodes with a through Axial Hole in the Smelting of Silicon Metal

Abstract

This article describes elements of the know-how of using carbon electrodes produced using the technology of molding around a rod when smelting silicon metal. Application of our know-how will dramatically increase the competitiveness of silicon metal production. Experts’ concerns regarding the use of such electrodes were that such electrodes have a through axial hole. This significantly reduces the mechanical strength of such electrodes, which can presumably lead to problems associated with the breakage of the working side of the electrode, which is immersed in the smelting space of the furnace under the charge layer. Industrial testing of such electrodes was carried out in a 30 MVA furnace of “Tau-Ken Temir” LLP. During testing, we used an approach previously developed by our team for working with a furnace in the process of smelting silicon metal. In particular, we used an interval between top treatments of about 30 min and adhered to the principles of balanced smelting, i.e., provided a balance between the intensity of the uniform supply of the charge into the furnace and the current active electrical power. Industrial testing carried out over four weeks confirmed the stability of the operation of cheaper carbon electrodes with a through axial hole. The recovery of silicon into finished products was also improved to 88–89% and the specific energy consumption was reduced to 11.2–12.1 MWh/t of silicon metal from the initial value 14,752 MWh/t. Thus, we received additional evidence for the effectiveness of our approach in furnace operating compared to an approach based on the ultimate provision of gas and permeability of the furnace top due to excessively intense processing of the top and an uncontrolled, uneven supply of charge to the furnace.

1. Introduction

The metallurgical industry is an integral part of developed countries around the world. Silicon metal (MG-Si) is used to produce silumins (as an alloying additive) in the chemical industry for the production of silicones and is also used in the electronics and solar energy industries (in the production of solar panels, silicon chips, semiconductors). Silicon, as the basis of various materials, is widely used in other industries: construction, computer equipment, in the production of organosilicon compounds and fiber optic products, ferrous metallurgy, etc. [1,2,3,4,5,6].

Under industrial conditions, silicon is produced by the carbothermic method by melting a charge consisting of silica-containing raw ore materials and a mixture of carbonaceous reducing agents at a temperature of 2000–2200 °C in three-phase electric arc furnaces [7,8] according to the main reaction:

SiO₂ + 2C = Si + 2CO.

Silicon melted in electric arc furnaces always contains a small amount of impurity inclusions, mainly iron silicide with admixtures of aluminum and calcium. It may also contain oxygen-containing compounds in the form of slag inclusions (as products of the incomplete reduction of oxides of raw materials) and particles of oxycarbide and silicon carbide (as a result of incomplete reduction processes in the furnace) [9]. Thus, using X-ray phase spectral microanalysis (on a Superprobe JXA-8200 electron probe analyzer (Jeol Ltd., Tokyo, Japan). The authors analyzed silicon samples from JSC Kremniy (Russia) (Figure 1). As can be seen from the data presented in Figure 1 and

Table 1. EMPA results for silicon samples.

Probe Point (Figure 1b)	Si	Fe	Ti	Ca	Al	Zr	Total
1	58.2	0.0	0.0	41.3	0.5	0.0	100.0
2	37.3	0.0	0.0	26.6	36.1	0.0	100.0
3	34.5	34.6	0.0	7.4	23.5	0.0	100.0
4	52.4	44.5	0.0	0.0	3.1	0.0	100.0
5	34.3	27.0	23.0	6.1	5.8	3.8	100.0
6	34.5	33.2	22.1	0.0	1.5	8.7	100.0

, the inclusion in the sample under study is represented by intermetallic compounds of the following composition:

CaSi₂ (with a small amount of Al), AlFeSi₂ and AlFeSi₂ (with Ca admixture), FeSi₂ (with Al admixture), FeSi₂Ti (with Ca and Al admixtures) and FeSi₂Ti (with Zr admixture). The presence of zirconium in the impurity inclusion can be explained by its transition into the silicon melt (in an insignificant amount) from the brick lining of the SAF.

There is a persistent stereotype around the technology of producing silicon metal. It was not reflected in scientific works and still remains at the level of a myth, transmitted orally from older generations of metallurgists to the younger. This myth is that, when smelting silicon metal, it is necessary to achieve maximum gas permeability of the charge. For this purpose, it is necessary to carry out intensive processing of the top almost continuously. Experienced specialists know that this practice is flawed. There are indications of this in non-public literature. For example, Figure 2, taken from the manual for our furnace for smelting silicon metal, shows examples of correct and incorrect work on the top of the furnace.

Practitioners in the field of silicon metal smelting achieve positive results [10,11,12]. However, these results are highly dependent on the personality of the furnace operator. When the operator is replaced, the skill of working with the furnace is lost and the results become unstable. In our previous article [10], we examined the mechanism of instability in the operation of the furnace. External signs of this instability are a violation of the cycle of accumulation of smelting products—the collapse of the charge. These changes are clearly visible in the shape of the diagrams that we presented earlier. To summarize, too frequent processing of the top leads to sharp and frequent fluctuations in the current load. Forced processing of the top leads to a change in the internal structure of the smelting bath and its natural electrical resistance. In this case, fluctuations in the electric current occur much more frequently and more chaotically compared to fluctuations associated with natural changes in the internal structure of the smelting bath. Some authors pointed out the importance of changing the current mode in the process of smelting silicon metal [13,14,15,16], but it should be recognized that these changes are only signs, but not the causes of a malfunction of the furnace.

A hidden sign of problems associated with forced processing of the top is a gross violation of the internal structure of the furnace smelting bath, hidden from visual observation. We can directly observe this structure only when the furnace is suddenly turned off, cooled and, gradually, layer by layer, manually unloaded from the furnace of raw materials and smelting products through the top.

In accordance with the established concept, a gas cavity is formed inside the furnace around the electrode. A typical structure of a smelting bath when smelting silicon metal is shown in Figure 3 [1].

During normal operation of the furnace, there should be a gas cavity of significant volume around each electrode in the smelting bath. Unjustifiably enhanced processing of the top of the furnace causes an imbalance between the amount of energy that enters the furnace and the amount of charge that is forced into the furnace in this mode of processing of the top. The energy that enters the furnace is not enough for the necessary chemical reactions to fully occur, heating the raw materials to operating temperature and smelting the reaction products. Some authors have pointed out the need for energy balancing of the silicon metal smelting process [17]. However, these recommendations are not always used in production practice.

At the same time, imbalance in the operation of the furnace is not the only negative result of incorrect operation of the top. When the gas cavity is filled with unreacted charge during forced processing of the top, the electrode becomes mechanically clamped in this charge. Large furnaces are often equipped with a mechanism for rotating the smelting bath. As a result, the electrode experiences increased lateral pressure, which can lead to the destruction of the electrode, especially if the pressure occurs at the nipple joint between two pieces of the electrode. A similar type of electrode failure is shown in Figure 4.

Technologists who do not understand the real reasons why the destruction of electrodes occurs follow the path of using increasingly mechanically strong electrodes with a deep degree of graphitization. However, even in this case, breaks of electrodes in the furnace due to improper operation of the top are not uncommon. The idea of using carbon electrodes, which are 1.5–2.5 times less durable compared to graphite electrodes, is not acceptable to them, and the idea of using carbon electrodes with a through hole along the axis of the electrode is generally terrifying. Meanwhile, carbon electrodes are produced under significantly milder firing conditions than graphite electrodes. This results in reduced electrode production costs, reduced emissions of pollutants into the environment and a reduced carbon footprint, which makes production significantly greener. In this regard, carbon electrodes are 30–40% cheaper than graphite electrodes. If we take into account that the share of costs for electrodes in the cost of silicon metal is 15–25%; then, when replacing graphite electrodes with carbon ones, the reduction in production costs can be up to 10%. This is already a significant competitive advantage. It is generally accepted in the expert community that carbon electrodes are less durable when smelting silicon metal, on the grounds that the mechanical strength of carbon electrodes is lower than the corresponding indicators in graphite electrodes. However, in this article, we want to prove that mechanical properties are not the most important factor that determines durability. The competence of the specialist who uses the electrode is much more important. In skilled hands, a carbon electrode can be much more durable than a graphite electrode. All Chinese silicon metal manufacturers use carbon electrodes manufactured using advanced molding technology around a removable rod. Among other reasons, this also explains the high competitiveness of Chinese-made silicon metal. In this article, we offer the results of industrial testing of carbon electrodes with a through axial hole and elements of know-how, describing how to properly use such electrodes based on the principles of balanced smelting to increase the competitiveness of silicon metal production.

2. Materials and Methods

This article is the fourth in a series of works aimed at improving the technology of smelting silicon metal [10,18]. We continued our work in the industrial conditions of the silicon plant of “Tau-Ken Temir” LLP using equipment that we used before [10].

Table 2. Electrical parameters of a single-phase furnace transformer for a 30 MVA furnace.

Tap	Transformer
	Primary/HV			Secondary/LV
	Phase Current, A	Line Voltage, V	Power, kVA	Phase Current, A	Voltage, V	Power, kVA	(C3) 1
1	202.5	35,000	7088	48.80	145	7076	13.2
2	210.1	35,000	7354	48.80	150	7340	12.9
3	217.6	35,000	7616	48.80	156	7603	12.6
4	225.2	35,000	7882	48.80	161	7871	12.3
5	232.8	35,000	8148	48.80	167	8135	12.1
6	240.3	35,000	8411	48.80	172	8398	11.8
7	247.9	35,000	8677	48.80	178	8662	11.6
8	255.5	35,000	8943	48.80	183	8926	11.3
9	263.0	35,000	9205	48.80	188	9189	11.1
10	270.6	35,000	9471	48.80	194	9457	10.9
11	278.1	35,000	9734	48.80	199	9721	10.7
12	285.7	35,000	10,000	48.80	205	9984	10.5
13	285.7	35,000	10,000	47.62	210	10,000	10.3
14	285.7	35,000	10,000	46.44	215	10,003	10.0
15	285.7	35,000	10,000	45.28	221	9998	9.8
16	285.7	35,000	10,000	44.20	226	10,002	9.5
17	285.7	35,000	10,000	43.17	232	10,002	9.3
18	285.7	35,000	10,000	42.18	237	10,001	9.1
19	285.7	35,000	10,000	41.24	243	10,001	8.9
20	285.7	35,000	10,000	40.34	248	10,000	8.7
21	285.7	35,000	10,000	39.47	253	9998	8.5
22	285.7	35,000	10,000	38.65	259	10,003	8.3
23	285.7	35,000	10,000	37.86	264	10,003	8.2
24	285.7	35,000	10,000	37.09	270	9999	8.0
25	285.7	35,000	10,000	36.36	275	9999	7.8

shows the electrical parameters of the furnace transformers. The furnace uses three single-phase transformers, one transformer for each electrode in the furnace. The transformer has 25 voltage taps. C₃ is the Westley coefficient, which determined by the formula:

$${С}_3={\displaystyle \frac{I}{{\left(\frac{P}{100}\right)}^{2/3}}},$$

where I—the current strength, kA; P—the power, kVA.

(C₃) 1, (C₃) 2, (C₃) 3 are three options for calculating the Westley coefficient. So, the Westly factor value is given for each tap on the low voltage side—(C₃) 1—obtained using the formula presented in the literature [19,20].

In accordance with the data presented by the author [19], very good performance of silicon smelting in the metal is achieved at a Westly factor value of 9.8, which in our case corresponds to the 15th voltage tap. It may not be entirely correct to use the nominal value of the electrical power of the furnace transformer when calculating the Westly factor. However, in fact, it is voltage taps 13–17 that are working in accordance with the manual of our furnace.

Table 3. Electrical parameters of the furnace.

Tap	Furnace
	Phase Current, kA	Voltage, V	Power, kVA	(C3) 2	Power Factor	Active Power, kW	(C3) 3
1	84,524	84	21,228	11.0	0.70	14,860	14.0
2	84,524	87	22,019	10.8	0.70	15,413	13.6
3	84,524	90	22,809	10.5	0.70	15,966	13.3
4	84,524	93	23,614	10.3	0.70	16,530	13.0
5	84,524	96	24,405	10.0	0.70	17,083	12.7
6	84,524	99	25,195	9.8	0.70	17,637	12.5
7	84,524	102	25,986	9.6	0.70	18,190	12.2
8	84,524	106	26,777	9.4	0.70	18,744	12.0
9	84,524	109	27,567	9.3	0.70	19,297	11.7
10	84,524	112	28,372	9.1	0.70	19,861	11.5
11	84,524	115	29,163	8.9	0.70	20,414	11.3
12	84,524	118	29,953	8.8	0.70	20,967	11.1
13	82,480	121	30,001	8.5	0.70	21,000	10.8
14	80,436	124	30,010	8.3	0.70	21,007	10.6
15	78,427	127	29,993	8.1	0.70	20,995	10.3
16	76,557	131	30,007	7.9	0.70	21,005	10.1
17	74,773	134	30,007	7.7	0.70	21,005	9.8
18	73,058	137	30,003	7.6	0.70	21,002	9.6
19	71,430	140	30,002	7.4	0.70	21,001	9.4
20	69,871	143	30,001	7.2	0.70	21,001	9.2
21	68,364	146	29,993	7.1	0.70	20,995	9.0
22	66,944	149	30,008	6.9	0.70	21,006	8.8
23	65,575	153	30,008	6.8	0.70	21,005	8.6
24	64,242	156	29,998	6.7	0.70	20,999	8.4
25	62,977	159	29,997	6.5	0.70	20,998	8.3

presents the electrical parameters of a furnace made on the basis of the above furnace transformers when connecting the windings on the low voltage side according to a triangle diagram. Table 3 calculates the phase current and voltage based on the fact that, when connected in a triangle, the current strength will increase by (3)^0.5 times, and the phase voltage will decrease by the same proportion. The rated power is calculated, in the usual way, as the product of the phase current by the voltage and the number of transformers on the furnace (3 pieces). Based on the furnace power rating and phase current calculated by the Westly factor—(C₃) 2. In this way, the calculated Westly factor reaches a value of 9.8 at the 6th tap of the furnace transformers. (C₃) 3 uses for calculating the phase current value and the active power value.

This doesnt make any sense, because taps 1 to 12 are used only when heating the furnace and cannot be used for continuous operation due to lack of power. Table 2 also calculates the estimated active power (furnace load) as the product of the rated power and the estimated power factor—0.7, which is taken both from our own practice of working on the furnace and from recommendations in the literature [19]. Based on the estimated active power and phase current of the furnace, the values of (C₃) 3 were calculated. Based on the calculations, (C₃) 3 reaches a value of 9.8 at the 17th tap, which is also quite close to reality.

Ultimately, we accepted that we would work on the 15th tap of the furnace transformer, which is most familiar to plant personnel. We would use automatic control of the furnace, not by impedance but by the current load at the level of 73 kA (the limit value at this tap was 78.427 kA), which gave an active power of 19.5 MW (the limit value was 21 MW at a power factor of 0.7).

Since the furnace electrodes were the center of our testing, we will discuss them in more detail. Silicon plant “Tau-Ken Temir” LLP has worked throughout its existence without observing the principles of balanced smelting. Therefore, we used the most expensive diameter range, as well as the most durable and high-quality graphite electrodes with an HP qualification level from the El6 Group (Russia) in production.

Table 4. Weight and dimensional parameters of carbon and graphite electrodes from the El6 Group company.

Nominal Diameter		Nominal Length		Approximate Weight per Piece
mm	in	mm	in	kg
1272	50	2700	106	4900
		2900	114	5300

and

Table 5. Technical characteristics of carbon and graphite electrodes from the El6 Group.

Options	Unit	Brand
		MKM-A	MKM-S	MCGM	MGM
Specific electrical resistance, no more	µOhm	35	30	25	15
Breaking force, no less	MPa	3	3	4	3.5
Bending strength, min	MPa	6	6	8.5	7
Total porosity, no more	%	20	20	24	28
Thermal conductivity, no less	W/(m × K)	10	11	20	80
Thermal expansion coefficient, (t 20–520 °C), no more	×10−6 K−1 at 20 °C	3.6	3.5	3.5	3.4
Apparent density, no less	g/cm3	1.58	1.56	1.58	1.59
Ash content, no more	%	2.5	2.5	1.6	1.5

present the technical characteristics of these electrodes.

From Table 5 it can be seen that the El6 Group company produces both carbon electrodes (grades MKM-A, MKM-S) and electrodes of a higher level of graphitization (grades MCGM, MGM) with properties similar to electrodes of the HP qualification. However, in both cases, the company uses continuous molding technology, i.e., the electrode body is solid without any holes. Silicon plant “Tau-Ken Temir” LLP has traditionally used electrodes of the most expensive MGM brand.

After much doubt and debate, it was decided to purchase an experimental batch of carbon electrodes with a through axial hole in the amount of 110 tons from Hebei Orient Carbon Co., Ltd. (Xiaopingying, Linzhang County, Handan City, Hebei Province, China) for testing. The technical parameters of these electrodes are presented in

Table 6. Electrical characteristics of carbon electrodes from Hebei Orient Carbon Co., Ltd.

Diameter, mm	Brand S		Brand G
	Permissible Current Strength, A	Current Density, (A/cm2)	Permissible Current Strength, A	Current Density, (A/cm2)
1272	73,600	5.8	85,100	6.7

and

Table 7. Technical characteristics of carbon electrodes from Hebei Orient Carbon Co., Ltd.

Type	Brand S	Brand G
Apparent density, no less than, g/cm3	1.58	1.58
Specific electrical resistance, no more than, µOhm	35	30
Ash content, no more than, %	4.5	4.0
Bending strength, no less than, MPa	6.5	6.5
Thermal expansion coefficient, (t 20–520 °C), no more than, ×10−6 K−1 at 20 °C	4.5	4.0

.

The cheapest brand S electrodes were used. Figure 5 shows photographs of electrodes from both companies. The figure on the right clearly shows a hole in the end part of the electrode, which is closed to a depth of about 500 mm with a plug made of crushed graphitization waste with the addition of a binding component. These plugs are inserted on both sides of the hole only to prevent any foreign objects from entering the electrode. These plugs do not mechanically reinforce the electrode structure because the hole is not completely filled with the mixture, and the mixture itself has low mechanical strength. In the figure on the left, a continuous homogeneous surface is visible at the end of the electrode. Figure 6 shows a pilot batch of carbon electrodes in the warehouse of a silicon plant that are prepared for testing.

The charge materials for smelting silicon metal were traditional in chemical and granulometric composition as well as other parameters, such as quartz, low-ash coal, charcoal, and wood chips. Therefore, we do not provide a separate description for them here. There were proposals to use a non-standard carbonaceous reducing agent called Rexyl, which is described in the literature [21]. However, to ensure the purity of the experiment, it was decided to postpone the Rexyl test for the future and not to combine two tests in one period.

To achieve balanced smelting of the silicon metal, we used the approach we described previously [10]. However, since balanced smelting is the basis of our know-how for working with through hole electrodes, we will briefly repeat it again. We set and maintained active power during the entire testing period (furnace load) 19.5 MW. We expected that the specific power consumption would be 12.5 MWh per 1 ton of silicon metal in the form of a solid product after tapping, casting and crystallization in the mold. This would mean that the furnace’s productivity would be 19.5/12.5 = 1.56 t/h or the need for quartz 1.56 × 60/28 = 3.3428 t/h, subject to the 100% extraction of silicon into the finished product. Then, we had to decide for ourselves what kind of silicon recovery we would focus on. Considering that, at “Tau-Ken Temir” LLP, silicon recovery is very poor, we started the first week of testing with a conservative level: 75%. Thus, the loading intensity for quartz should have been 3.3428/0.75 = 4.457 t/h. In the following weeks of testing, we strived to consistently reduce assumptions of the specific power consumption and increase the silicon recovery in the finished product. Unlike the first experience of carrying out balanced smelting in 2017, we abandoned the concept of excess charge because it strongly depends on the values of specific power consumption and silicon recovery conventionally accepted in calculations. Instead, we used absolute values of the hourly consumption of quartz for smelting.

Testing began with the gradual replacement of graphite electrodes with carbon ones. At a certain point, the expansion of electrode columns began with carbon electrodes instead of graphite ones. The total length of the electrode column was about 22 m. With a length of one piece of electrode of 2.7 m, the number of pieces per column is 22/2.7 = 8.1 pcs. The weight of one 2.7-m-long piece of electrode was 4.9 tons (1.814 t/m). Thus, the weight of one electrode column was 4.9 × 8 = 39.2 tons (three columns 39.2 × 3 = 117.6 tons). The specific consumption of the electrode was about 0.1 t/t of silicon metal. With a productivity of 1.56 t/h (37.44 t/day) of silicon metal, the daily consumption of the electrode would be 37.44 × 0.1 = 3.744 tons. The gradual, natural replacement of graphite electrodes with carbon electrodes took 117.6/3.744 = 31.4 days. Accordingly, the available supply of 110 tons of carbon electrodes was enough to test within 4 weeks. About a week before the carbon electrodes reached the charge level in the furnace, a control testing period began. During the control week of testing, we did not interfere with the existing operating process. During the transition period, from the moment when the carbon electrodes touched the charge until the graphite electrodes were completely replaced by coal electrodes, a balanced smelting mode was gradually introduced in the furnace. At the same time, we practiced balanced smelting skills with personnel. The transition period was also about a week. After this, the testing itself began.

Conditions not related to ensuring balanced smelting during testing of the carbon electrodes were as follows: We gave the consumption of reducing agents in the charge based on the carbon factor of 1.05. At the same time, the share of carbon contributed by charcoal, low-ash coal and wood chips was 70, 25 and 5%, respectively. For critical disturbances in the operation of the furnace, we used a supply of borate flux. This adjustment had to be used only once during the entire testing period, when, due to the inexperience of one furnace operator, the principles of balanced smelting was violated. This did not lead to significant deviations of the final average values of the technological indicators from the norm.

As before, the most important factor in determining the success of testing is the endurance and stress resistance of the furnace operator. Forced processing of the top and loading of the charge beyond the calculated limit should not be allowed, despite subjective feelings in this regard. The experience of balanced smelting gained in 2017 was very useful, so the new test was much easier for the staff.

3. Results

The test results are presented in

Table 8. Initial parameters of balanced smelting.

Test Week	Average Active Power on the Furnace (Furnace Load), MW	Accepted Values for Calculation		Estimated Need for Quartz, t/Hour
		Specific Power Consumption, MWh/t	Silicon Recovery, %
control	19.3	12.5	75	4.457
1	19.6	12.0	80	4.353
2	19.4	11.5	85	4.275
3	19.4	11.0	90	4.221
4	19.5	10.5	95	4.189

and

Table 9. Test results.

Test Week	Spent During Testing			Technical Test Results
	Quartz, Tons	Electric Energy, MW	Electrode, Tons	Silicon Metal Produced, Tons	Silicon Recovery, %	Specific Power Consumption, MWh/mt	Specific Consumption of Electrode, t/t of Silicon Metal
control	748.8	3275	25	222	63.53	14,752	0.113
1	731.3	3271	26	269	78.82	12,160	0.097
2	718.2	3277	27	282	84.14	11,621	0.096
3	709.1	3271	28	289	87.33	11,318	0.097
4	703.8	3273	28	292	88.91	11,209	0.096

. In the control period, as usual at “Tau-Ken Temir” LLP, loading mode is characterized by great instability. The gas cavities around the working side of the electrodes are small. The furnace is not working well, even forcibly gaining a power of 19.5 MW due to the regular collapse of thermally unprepared charges at the bottom of the furnace. After the transition period was completed, furnace performance improved significantly. The top of the furnace became rigid and, visually, its gas permeability decreased.

However, despite this, the charge came off only on its own, without forced processing. In fact, we have reduced the hourly charge load, ensuring the stability of the charge supply to the furnace while maintaining active power (furnace load) at the level of 19.5 MW. Thanks to this alone, a reduction in the specific consumption of electrical energy and the electrode per ton of silicon metal began, and silicon recovery steadily grew. After receiving the results, we plotted the dependence of silicon recovery on the hourly consumption of quartz at a furnace load of about 19.5 MW (Figure 7) and the specific power consumption of silicon recovery (Figure 8).

As can be seen from Figure 7, reducing the quartz supply significantly increased the recovery of silicon in the final product, but this effect was weakened when the quartz consumption decreased to less than 4.275 t/h. Further reduction in consumption no longer significantly increased the efficiency of smelting silicon metal. This may be due to the fact that the size of furnace load had not reached its optimal value and needed to be increased to increase the smelting efficiency. This should be the subject of further research in the future.

The dependence in Figure 8 is compared with a similar dependence that T.E. Magnussen theoretically established in his article [19] (Figure 9). As can be seen from a comparison of Figure 8 and Figure 9, our data are in good agreement with T.E. Magnussen’s data at the maximum furnace value efficiency factor (η)—0.86.

There were some minor problems during the transition period. Due to differences in the thermal expansion coefficients of the carbon and graphite electrodes of 4.5 × 10⁻⁶ K⁻¹ and 3.4 × 10⁻⁶ K⁻¹, respectively, a weakening of the pin connection between the carbon and graphite electrodes was visually observed. Therefore, it was necessary to work during the transition period until the remaining graphite electrodes were completely used up, using very great precaution in order to avoid rupturing the electrode column at the junction of the carbon and graphite parts. After completely replacing the graphite electrodes with carbon electrodes, there were no problems with the electrode column during the remaining testing period.

4. Conclusions

This test has shown that the use of carbon electrodes with a through axial hole is completely safe when smelting silicon metal in a submerged arc furnace. The possibility for the safe use of such electrodes is associated with the use of balanced smelting principles in the testing process. By the concept of balanced smelting, we mean a method of operating a submerged arc furnace where raw material is fed into the furnace in accordance with its current value of active power, taking into account the achieved (or desired to be achieved in the process of smelting optimization) level of specific energy consumption and extraction of the valuable components and preventing sharp changes in the hourly consumption of the charge (if this is not required when changing the active power). With this approach, the gas cavities around the electrodes are increased during operation and the lateral pressure of the solid charge on the electrodes is excluded, which ensures an increase in the stability of the electrodes despite their reduced mechanical strength compared to graphite electrodes. During the testing of electrodes with a through axial hole, the dependence of silicon extraction on the quartz feed was studied in parallel with a constant furnace load of 19.5 MW under conditions that ensured the principles of balanced smelting in a furnace with a capacity of 30 MVA. This dependence was processed using a mathematical method and, as a result, a corresponding mathematical model was established. The secondary dependence of the specific energy consumption on silicon extraction with a fixed furnace load of 19.5 MW was also studied. The obtained pattern was compared with the theoretical data of Magnussen T.E. The comparison showed a good correlation between theoretical and practical values, indicating that our study was a high-quality industrial experiment that led to an improvement in technological indicators. Thus, the extraction of silicon into finished products was also improved to 88–89% from an initial value of 63.5% and the specific energy consumption was reduced to 11.2–12.1 MWh/t of metallic silicon from an initial value of 14.75 MWh/t.