Strain prediction and benchmark with measurement in the Spallation Neutron Source mercury target without gas injection

The 1.4-MW Spallation Neutron Source (SNS) currently operates as the world’s highest time-average power spallation source and uses mercury as the spallation target material. The mercury circulates through a stainless-steel target vessel and heat exchanger loop that removes the heat deposited from the incident proton beam in the target. The mercury vessel is part of the fixed but replaceable target module. The target module is not safety credited in the SNS design, but reliable long life is vital for meeting operational goals and minimizing cost and waste. SNS is a short-pulse ($\sim 0.7\mu \mathrm{s}$) neutron source with a repetition rate of 60 Hz with 1.0 GeV protons onto the target. The short-pulse operation introduces two mechanisms that threaten the longevity of the mercury vessel. The pulsed target heating is essentially isochoric, which results in a distributed pressure field in the mercury with each pulse. After each pulse, the pressure propagates, interacts with, and strains the stainless-steel vessel. For intermittent times, this pressure field leads to mercury cavitation. The first mechanism of concern for the target lifetime is high-cycle fatigue to the stainless-steel vessel and the second is cavitation erosion. At the onset of the SNS project, the R&D program sought to address the need for an accurate and practical method to estimate the fatigue life of the mercury vessel. Around 2000, the R&D team conducted a series of in-beam mercury target experiments at the Weapons Neutron Research Facility at Los Alamos Neutron Science Center. These experiments successfully produced a small database of pulse response strain data. The data were used to develop and benchmark an explicit dynamic finite element analysis technique for estimating target fatigue life. SNS is now a mature facility with about 15 years of operation and 29 SNS target modules consumed. Since 2015, strain measurements have been collected from actual targets, and abundant target vessel strain data are now available in greater detail than was possible in 2000. Variances in pulse response data have been characterized using tolerance interval analysis to benchmark the simulation technique more thoroughly. The simulated results were found to match measurements generally well, but the method is always conservative. A potential adjustment to the cavitation threshold parameter value was suggested. Variances in the pulse response data are moderate, which is largely explained by the fact that data were acquired in 60-Hz burst mode and by the lasting vibration of the vessel between pulses.

Figure 1

Complete SNS target module (a) and exploded view of the mercury vessel next to the shroud (b).

Figure 2

Isometric cut view of target showing bulk flow (a) and elevation cut view of target showing window and jet flows (b).

Figure 3

Target FEA simulation model.

Figure 4

Volumetric strain and pressure relationship in Riemer model.

Figure 5

Mid-vertical plane view of initial beam pulse pressure field in SNS target.

Figure 6

Fiber sensors mounted on the surface of the mercury vessel of the SNS target for strain measurement.

Figure 7

Strain sensor positions and orientations on the mercury vessel top (a) and bottom (b) surfaces.

Figure 8

Tolerance interval calculations of sensor T strain measurement data from T26 (a), T27 (b), and T28 (c).

Figure 9

Tolerance interval calculation of sensor T strain measurement data from T26, T27, and T28.

Figure 10

Dynamic strain responses at the front center of the target in the beam direction. (a) Sensor A (center-on top). (b) Sensor M (center-on bottom) (c) Sensor T (1.8 cm from center-on top). (d) Sensor Y (1.8 cm from center-on bottom). (e) Sensor U (5.3 cm from center-on top). (f) Sensor Z (5.3 cm from center-on bottom).

Figure 11

Vessel deformation of the forward region at 0.08 ms (deformation is magnified $500\times$). (a) Side view and (b) front view with window cutoff.

Figure 12

Dynamic strain responses at the front center of the target in 45° (a) and 90° (b) to beam direction.

Figure 13

Dynamic strain responses at the front corner of the target. (a) Sensor D (left-beam direction). (b) Sensor X (right-beam direction). (c) Sensor V (left-transverse direction). (d) Sensor W (right-transverse direction).

Figure 14

Simulated deformation of the vessel near the two strain peaks of the sensor (deformation is magnified $500\times$, sensor IDs in parentheses are on the symmetry side). (a) At 0.3 ms and (b) at 0.65 ms.

Figure 15

Strain responses and deformed shape in the middle of the target. (a) Sensor G (4.6 cm from center-transverse direction). (b) Deformed shape (magnified 500×) of the target at 0.2 ms when the strain peak of sensor G occurs (left and right sensors in the same side of half symmetry model). (c) Sensor E (left-beam direction). (d) Sensor K (right-transverse direction).

Figure 16

Dynamic strain responses in the back of the target. (a) Sensor H (center-beam direction). (b) Sensor J (center-transverse direction). (c) Sensor F (left-beam direction). (d) Sensor L (right-transverse direction).

Figure 17

Deformed shape (magnified $500\times$) of the front body at the back sensor section at 0.15 ms (a) and 0.3 ms (b).

Figure 18

Relative difference of strain range values between the simulation and the measurement vs the measured strain range.

Figure 19

Strain range value comparison at the sensor locations: top view (a) and bottom view (b).

Figure 20

Strain predictions of sensor T with different mercury tensile cutoff pressures. (a) Decrease cutoff pressure. (b) Increase cutoff pressure.

Figure 21

Strain prediction of sensors K (a) and L (b) with extended simulation time.

Figure 22

Deformed shape (magnified $500\times$) of the front body at sensor K at 1.38 ms.

Figure 23

Deformed shape (magnified $500\times$) of the front body at sensor L at 2.1 ms.

Figure 24

Example strain responses of transverse-oriented sensors on the top inlet side and apparent deformation propagation speeds.

Figure 25

Strain response from example pulses at locations K (a) and L (b) for 16 ms.

Figure 26

Strain response from example pulses at locations T (a) and G (b) for 16 ms.

Figure 27

Strain noise level from mercury flow at sensors K and L (a) and strain noise level from mercury flow at sensors G, H, and T (b).

Figure 28

Strain measurement data of sensor T without and with gas injection.

Figure 29

Average and limits of ten responses from sensor T30-L acquired in burst mode (a) and individually (b).

Figure 30

Average and limits of ten strain responses from sensor T30-A acquired in burst mode (a) and individually (b).

Figure 31

Average and limits of ten strain responses from sensor T30-K acquired in burst mode (a) and individually (b).