Nak FLOW CONTROL by ELECTROMAGNETIC PUMP of the SNAP-10A SPACE NUCLEAR REACTOR

2013 International Nuclear Atlantic Conference - INAC 2013 Recife, PE, Brazil, November 24-29, 2013 ASSOCIAÇÃO BRASILEIRA DE ENERGIA NUCLEAR - ABEN ISBN: 978-85-99141-05-2 NaK FLOW CONTROL BY ELECTROMAGNETIC PUMP OF THE SNAP-10A SPACE NUCLEAR REACTOR

Eduardo M. Borges, Francisco A. Braz Filho and Lamartine N. F. Guimarães

Instituto de Estudos Avançados (IEAv - DCTA) Trevo Cel Av José Alberto Albano do Amarante n.1 12228-001 São José dos Campos, SP [eduardo, fbraz, guimarae]@ieav.cta.br

ABSTRACT

In the Institute for Advanced Studies (IEAv), were developed, successfully, the first two direct current electromagnetic pumps (DC EM pumps) of Brazil. The first was built with C-type magnet and coils; and the second, with Samarium-Cobalt permanent magnets, for magnetic field generation. Both were tested and performed quite satisfactory. The electromagnetic pump uses the Faraday principle, in which the interaction of the magnetic field and electric current generates the magneto-motive force, which produces the circulation of the fluid. This type of equipment may be used for controlling the liquid metal flow in nuclear space fast reactors. This paper shows the computer programs developed for design and evaluation of DC EM and electromagnetic thermoelectric (EMTE) pumps, the DC EM pump of Samarium-Cobalt magnets data to Mercury loop flow control, the EMTE pump of SNAP space nuclear reactor. It also compares the theoretical results to experimental data of NaK primary loop flow control by electromagnetic thermoelectric pump of the SNAP-10A space nuclear reactor, with satisfactory results, confirming the viability of the electromagnetic pumps evaluation scheme.

1. INTRODUCTION

The research and development on electromagnetic pumps at the Institute for Advanced Studies – IEAv started in 1988. Mercury was chosen as working fluid, for experimentation purpose, because it is a liquid metal at room temperature. Two direct current electromagnetic pumps (DC EM) were designed, built and tested. The first pump was built using a “C” type electromagnet for magnetic field generation. The second one was built using Samarium- Cobalt magnets for magnetic field generation. Both pumps presented satisfactory behavior with respect to the Mercury flow control [1, 2].

Without moving parts and thus with high reliability, the electromagnetic pump may control and pump an electrically and thermally conductivity fluid flow in a closed loop. That characteristic makes the electromagnetic pump highly desirable to be used in a system that will not undergo maintenance during its useful life, such as the ones intended to operate in space, such as NaK liquid flow control for the SNAP-10A space nuclear reactor application.

The working fluid of the SNAP-10A space nuclear reactor is a NaK alloy, in combination with 22% Sodium and 78% Potassium. The electromagnetic pump of SNAP project is an electromagnetic thermoelectric pump (EMTE). Its magnetic field is imposed by permanent magnets of Alnico-V. Its electric current is generated by thermoelectric converters, which change the thermal difference, imposed on their terminals in direct current. This electric current is conducted to the channel pump so that there perpendicularity to DC current,

magnetic field and fluid flow direction. The interaction field – current generates a force, witch controls the fluid flow in space reactor loop [3].

On 3 April 1965, SNAP-10A space nuclear reactor was launched from Vandenberg Air Force Base (USA) and placed into a 700 nautical mile circular polar orbit. Twelve hours after launch, the nuclear reactor was automatically brought up to operating temperature and initially produced more than 600 Watts of electrical power, or 43 KW thermal power. After 43 days of successful operation, the reactor was shut down as the result of a high voltage failure in the electrical system of Agena spacecraft [4].

This paper shows the computer programs developed for design and evaluation of DC EM and EMTE pumps, the DC EM pump of Samarium-Cobalt magnets data to Mercury loop flow control, and the EMTE pump of SNAP space nuclear reactor. It also compares the theoretical results to experimental data of NaK primary loop flow control by electromagnetic thermoelectric pump of the SNAP-10A space nuclear reactor, with satisfactory results, confirming the viability of the electromagnetic pumps design and evaluation scheme.

2. ELECTROMAGNETIC PUMPS EQUATIONS

Electromagnetic pumps work based on Faraday´s principle, in which the electrical current (I) interacts with the magnetic field (B) generating the magnetic force (F). Therefore, this in turn controls the liquid metal flow (W) [5]. Figure 1 shows a schematic of the electromagnetic pump working principle. Where a is the channel height, m; b is the channel width, m; and c is the active pump length, m. All geometric characteristics are used by meters.

Figure 1: DC EM pump working principle.

Considering the magnetic field perpendicular to the main electric current and the direction of the fluid flow, the force resulting from the interaction between magnetic field and electric current is:

F = BI eb . (1)

2 Where, F is the force, N; B is the magnetic field intensity, Wb/m or Tesla; and I e is the useful electric current, A. The head pump can be defined by:

INAC 2013, Recife, PE, Brazil.

P = F/(ab) . (2)

Taking Eq. (1) into Eq. (2) becomes: P = BI e/a . (3)

Where, P is the pressure, N/m 2 , or

P = B I e / (1360 a) . (4)

Where, P is the pressure, cm Hg.

Figure 2 shows the equivalent electric circuit of a DC electromagnetic pump.

Figure 2: DC EM pump equivalent electric circuit.

The electric tension of the electromagnetic pump is calculated by:

V = I eRe+E c = IR t . (5)

Where, V is the electric tension, V; R e is the useful electric resistance (of the fluid in the channel), Ohm; R t is the equivalent electric resistance of the circuit, Ohm; I is the main current, A; and Ec is the electro-countermove force resulting of the fluid moving in the magnetic field, V. This induced voltage can be obtained by:

E c = BW/a . (6)

Where, W is the volumetric fluid flow rate, l/s.

The useful electric current Ie can be calculated by Eq. (7):

= I − E c Ie . (7)  R + R  R w R b 1+ R  w b  R + e   e R + R  R w R b  w b

INAC 2013, Recife, PE, Brazil.

Where, R b is the electric resistance of the bypass or escape, Ohm. This is related with the geometry of the pump, and it is calculated multiplying the useful electric resistance by an empiric correction factor. Hence, the useful electric current is a function of the volumetric fluid flow. In the static pressure case (without fluid flow) the last term of the Eq. (7) is zero. And in the dynamic pressure case (with fluid flow) this term is significant and not null [5].

2.1. Simulation Codes

The BEMC-1 code was written in the C++ language. It was created with the objective to perform evaluation at each development step of a direct current electromagnetic pump, allowing for important design parameter variations. One may calculate the several electric resistances involved in the DC EM pump design by defining the working fluid and establishing its properties, the channel geometry and materials. Also, one may calculate the magnetic field and static pressure as a function of the main electrical current in the DC EM pump. With the BEMC-1 code can be calculated the system operating points, in other words, the fluid flow rate and the dynamic pressure give by the pump as it operates in a closed loop. The BEMC-1 code can calculate the dynamic pressure loss as a function of the fluid flow, and equivalent diameter and length of the loop [6, 7].

The simulation of a space nuclear reactor requires the operation knowledge for the thermo electric convertors, electrical circuits, magnetic field, cooling loops, heat exchangers and heat pipe radiators, all related to the pump operation as a function of the nuclear reactor power. In this way, one may to relate the operation point for the nuclear reactor and the pump operation. The BEMTE-3 performs all types of electromagnetic pumps calculation by, interactively, solving the system equations. It takes into consideration the pressure losses in the flow loops as defined by the design and the geometry of the system (Reynolds’ number function), calculates the necessary pump flow rate and correlates that with the nuclear reactor thermal power. The BEMTE-3 program is a PC version of the BEMTE computer code (FORTRAN version) developed in the 90’s [5, 8].

The BEMTE-3 program models the thermo electric convertor through a calculation of the temperature distribution in each of the thermo-electric elements. For that, it solves the heat conduction differential equation, obtaining the values for the internal resistance and the Seebeck tension generated by the thermo electrical convertors as a function geometric data, conversion parameters and the temperature difference imposed over its terminals, which in turn is a function of the reactor thermal power. After that, the electric current that circulates in the EMTE pump channels is calculated. The code can to calculate the magnetic field generated as a function of the electric current, and the dynamic head produced by the EM pump, which is obtained by the magnetic field interaction with the electric current in each channel, according to the equations developed by Barnes and Blake [9, 10].

3. PUMPS EVALUATIONS

3.1. DC EM Pump

Magnetic field of 0.44 Wb/m 2 was measured for the Samarium-Cobalt magnets DC EM pump testing. A direct current electrical source is used to generate the main current until 800 A. The main electrical current interacts with the magnetic field creating the magnetic force that acts over the fluid in the channel’s pump, controlling the fluid flow. The geometry of the

INAC 2013, Recife, PE, Brazil.

channel’s pump is: height a = 10 mm, width b = 30 mm and active pump length c = 70 mm, where the letters a, b and c refers to Fig. 1. Table 1 presents the DC electromagnetic pump parameters. Table 2 shows liquid metal properties of Mercury and NaK (22% Sodium and 78% Potassium), that will be used in the simulations.

Table 1: DC EM pump parameters

Parameters data Pump channel height (a) [mm] 10 Pump channel width (b) [mm] 30 Active pump length (c) [mm] 70 Magnetic field (B) [Gauss] 4400 Loop internal diameter (Di) [mm] 12.2 Loop length (L) [m] 3.8

Table 2: Liquid metals properties

Property \ metal Mercury NaK Temperature ( oC) 20 500 Electrical Resistivity (Ohm.m) 9.3 e-7 3.1 e-7 Specific Mass (Kg/m 3) 13400 744 Dynamic Viscosity (N.s/m 2) 1.5 e-3 1.9 e-4

Figure 3 presents experimental measurements for the static pressure of the Samarium-Cobalt magnets DC EM pump, operating with Mercury as a working fluid. The curves presented are for three, four and five magnet blocks, and as a function of the main electrical current. Also it is shown the theoretical result generated with the BEMC-1 code. The comparison of theoretical curves and measured data reveals good agreement.

Figure 4 shows experimental measurement points of head pressure data to the Samarium- Cobalt magnets DC EM pump, as a function of the Mercury flow operating with a closed dynamic loop, with 3.8 m of length and 12.2 mm of internal diameter. Figure 4 also shows the theoretical pressure loss curve of dynamic loop, calculated with the BEMC-1 code.

Figures 3 and 4 present very good agreement between experimental and calculated data using the BEMC-1 code. This fact gives validation to the results produced with the BEMC-1 computational program and allows for the use of the code for other liquid metals of interest.

Figure 5 shows the theoretical system operation points, which are represented by the curves intersections. The intersections are produced by the DC EM pump theoretical performance curve and the pressure loss theoretical curves for the dynamic loop, both calculated with the BEMC-1 code, using Mercury as the working fluid, considering Table 1 and Table 2 data. DC EM pump performance curves are all parallel straight lines, that is due to the fact that the magnetic field, generated by the Samarium-Cobalt magnets, is constant equal to 0.44 Wb/m 2.

INAC 2013, Recife, PE, Brazil.

BEMC-1 5 blocks 12 4 blocks

3 blocks

8 Static pressure(cm Hg) 4

0 200 400 600 800 Main electrical current (A)

Figure 3: DC EM static pressure pump data measured and simulated to Mercury.

LOOP 3.8 m

12 I= 600 A

I= 500 A

I= 400 A Pressure (cm Hg) I= 300 A

4 I= 200 A

1 2 3 4 5 6 Mercury flow (l/min)

Figure 4: DC EM head pump data measured and simulated to Mercury flow.

Figure 6 shows the DC EM pump performance curves and the pressure loss in the dynamic loop, both calculated with the BEMC-1 code using NaK (22% Sodium and 78% Potassium) as working fluid, considering data of the Table 1 and Table 2, and magnetic field, generated by the Samarium-Cobalt magnets, is constant and equal to 0.44 Wb/m 2.

INAC 2013, Recife, PE, Brazil.

2.5E+4

I= 800 A 2.0E+4 LOOP 3.8 m ) 2 2 I= 600 A 1.5E+4

1.0E+4 I= 400 A Head pump Head (N/m

5.0E+3 I= 200 A

0.0E+0

0 4 8 12 16 20 Mercury flow (l/min)

Figure 5: DC EM head pump data simulated by BEMC-1 to Mercury flow.

3.0E+4

I= 800 A

2.0E+4 ) 2

I= 600 A

I= 400 A

Headpump(N/m 1.0E+4

I= 200 A

circ 3.8 m

0.0E+0

0 4 8 12 16 20 NaK volumetric flow (l/min)

Figure 6: DC EM head pump data simulated by BEMC-1 to NaK flow.

Table 3 presents the DC EM pump dynamic operation points, using as given parameters: the main electrical current (I) of 600 A and a magnetic field (B) of 0.44 Wb/m 2, considering Mercury and NaK (22% Sodium and 78% Potassium) as working fluids, as calculated by the BEMC-1 code, considering data of the Table 1 and Table 2.

INAC 2013, Recife, PE, Brazil.

Table 3: DC EM system operation points

Parameter\Fluid Mercury NaK Flow (l/min) 4.3 16.5 Pressure (N/m 2) 1.7 e 4 1.2 e 4

3.2. SNAP EMTE Pump

Figure 7 shows a photo of the SNAP-10A space nuclear reactor and Agena spacecraft mockups [4], notes the small nuclear reactor, its core and nuclear control systems. The core must be cooled. The heat generates must be change to space, by cooling system with loops, pumps, heat exchangers and heat pipe radiators. Electrical DC to Agena spacecraft is generates by thermo electrical convertors as function geometric data, conversion parameters and the temperature difference imposed over its terminals, which in turn is a function of the SNAP-10A reactor thermal power.

Table 4 presents the main geometric data and operational characteristics of electromagnetic thermoelectric pump (EMTE) of the SNAP-10A space nuclear reactor [3].

Figure 7: SNAP-10A space nuclear reactor photo.

Table 4: SNAP EMTE pump parameters

Parameters data Pump channel height (a) [mm] 12.7 Pump channel width (b) [mm] 25 Active pump length (c) [mm] 63.5 Magnetic field (B) [Gauss] 2350 Electrical current (I) [A] 1800

INAC 2013, Recife, PE, Brazil.

Figure 8 shows performance curves of the EMTE SNAP pump to primary loop of the space nuclear reactor SNAP-10A. The black continuous line is the experimental head pump as function of the NaK mass flow rate [3]. The blue trash line is head pump as function of the NaK mass flow calculated with the BEMTE-3 code, considering SNAP EMTE pump parameters and operations data of the Table 3 (the magnetic field, generated by the Alnico-V magnets is constant and equal to 0.235 Wb/m 2, and main electrical current is 1800A), and NaK (22% Sodium and 78% Potassium) properties of the Table 2. Figure 8 presents satisfactory results, confirming the viability for design and evaluation of direct current electromagnetic (DC EM) and electromagnetic thermoelectric (EMTE) pumps.

3.0E+4

SNAP BEMTE SNAP Exp.

2.0E+4 ) 2

Head pump (N/m Head 1.0E+4

0.0E+0

0.0 1.0 2.0 3.0 NaK mass flow (Kg/s)

Figure 8: SNAP EMTE head pump data to NaK mass flow rate.

4. CONCLUSION

Two direct current electromagnetic pumps (DC EM) were designed, built and tested at IEAv. The first pump was built using a “C” type electromagnet. The second one was built using Samarium-Cobalt magnets for magnetic field generation, 0.44 Wb/m 2 was experimentally measured for this pump design. Both pumps presented satisfactory behavior with respect to the Mercury flow control, the operation points for the system pump-loop were experimentally obtained. Very good agreement between experimental and calculated data using the BEMC-1 code was get. This fact gives validation to the results produced with the BEMC-1 computational programs, this and BEMTE-3 code can to evaluate the DC EM and EMTE pump theoretical performance, operating with liquid metal fluids of interest.

INAC 2013, Recife, PE, Brazil.

will not undergo maintenance during its useful life, such as the ones intended to operate in space, such as NaK liquid flow control for the SNAP-10A space nuclear reactor application.

This paper shows compares the theoretical results to experimental data of NaK primary loop flow control by electromagnetic thermoelectric pump of the SNAP-10A space nuclear reactor, with satisfactory results, confirming the viability of the electromagnetic pumps design and evaluation scheme.

ACKNOWLEDGMENTS

The authors are grateful for the assistance provided by the TERRA project. In addition, they are thankful for the CNPq under the grant “MCT/CNPq/AEB nº 33/2010 - Formação, Qualificação e Capacitação de RH em Áreas Estratégicas do Setor Espacial” no. 559928/2010-6. And the authors wish to acknowledge the “Fundação de Amparo à Pesquisa do Estado de São Paulo – FAPESP” for sponsoring the participation of some of the authors in this scientific event.

REFERENCES

1. E. M. Borges, et al., “Ensaios de Pressão Estática de Bomba Eletromagnética de Corrente Contínua”, XIII Congresso Brasileiro de Engenharia Mecânica , Belo Horizonte, MG, Brasil, (1995). 2. E. M. Borges, et al., “Rare-Earth Magnets Applied to Liquid Metal Flow” , XIV International Workshop on Rare-Earth Magnets and Their Applications , São Paulo, SP, Brazil, (1996). 3. J. L. Johnson, “Space Nuclear System Thermoeletric NaK Pump Development – Summary Report”, NASA-CR-121231, USA, (1973). 4. G. L. Schimidt, “SNAP 10A Test Program”, Energy Technologic Engineering Center, USA, (1988). 5. E. M. Borges, “Desenvolvimento e Simulação Computacional de Bombas Eletromagnéticas Termoelétricas para o Controle do Escoamento em Reatores Nucleares Espaciais Refrigerados a Metal Líquido”. Tese de Doutorado ITA , São José dos Campos, SP, Brasil, (1991). 6. E. M. Borges, et al., “Software for CD Electromagnetic Pump Simulation – BEMC-1”, XVII Congresso Brasileiro de Engenharia Mecânica , São Paulo, SP, Brasil, (2003). 7. E. M. Borges, et al., “A Parametric Study and Performance Evaluation of DC Electromagnetic Pumps with BEMC-1 Code”, Engenharia Térmica , ISSN: 1676-1790, Vol. 5, N. 2 , pp. 16-23, (2006). 8. E. M. Borges, et al., “Programa Computacional BEMTE-3”, Atividades de Pesquisa e Desenvolvimento – Instituto de Estudos Avançados – IEAv – Anai do 9th Workshop Anual de Pesquisa e Desenvolvimento do Instituto de Estudos Avançados. ISSN 1983- 1544, Vol. 2, pp. 140, (2009). 9. A. H. Barnes, “Direct-Current Electromagnetic Pumps”, Nucleonics , Vol. 11, N. 1, pp. 16-21, (1953). 10. L. R. Blake, “Conduction and Induction Pumps for Liquid Metals”, Proceedings of IEEE, paper n° 2111U, (1956).

INAC 2013, Recife, PE, Brazil.