3D Flow and Temperature Analysis of Filling a Plutonium MoldNicholas P. Orenstein1, William D. Peach1, Thomas A. Jachimowski1

1Manufacturing Engineering and Technology Division, Los Alamos National Laboratory2013 Flow-3D World Users Conference

UNCLASSIFIED LA-UR-13-26213

IntroductionThe plutonium foundry at Los Alamos National Laboratory casts products for

various special nuclear applications. However, plutoniums radioactivity, materialproperties, and security constraints complicate the ability to performexperimental analysis of mold behavior. The Manufacturing Engineering andTechnologies (MET-2) group previously developed a graphite mold to vacuumcast small plutonium disks to be used by the Department of Homeland Securityas point sources for radiation sensor testing. A two-stage pouring basinconsisting of a funnel and an angled cavity directs the liquid into a verticalrunner. A stack of ten disk castings connect to the runner by horizontal gates.Volumetric flow rates were implemented to limit overflow into the funnel andminimize foundry returns.

Models using Flow-3D computational fluid dynamics software are employedhere to determine liquid Pu flow paths, optimal pour regimes, temperaturechanges, and pressure variations.

SetupHardcopy drawings provided necessary information to create 3D .stl models

for import into Flow-3D (Figs. 1 and 2). The mesh was refined over severaliterations to isolate the disk cavities, runner, angled cavity, funnel, and inputpour. The final flow and mold-filling simulation utilizes a fine mesh with~5.5 million total cells. For the temperature study, the mesh contained 1/8 asmany cells to reduce computational time and set temperatures to 850 C for themolten plutonium and 500 C for the solid graphite mold components (Fig. 3).

Flow-3D solves mass continuity and Navier-Stokes momentum equationsover the structured rectangular grid model using finite difference and finitevolume numerical algorithms. The solver includes terms in the momentumequation for body and viscous accelerations and uses convective heat transfer.

Simulation settings enabled Flow-3D physics calculations for gravity at980.665 cm/s2 in the negative Z direction (top of mold to bottom); viscous,turbulent, incompressible flow using dynamically-computed Renormalized GroupModel turbulence calculations and no-slip/partial slip wall shear, and; first order,full energy equation heat transfer. Mesh boundaries were all set to symmetricboundary conditions except for the Zmin boundary set to outflow and the Zmaxboundary set to a volume flow. Vacuum casting conditions and the high reactivityof remaining air molecules with Pu validate the assumption of an initiallyfluidless void.

ResultsThe flow follows a unique three-dimensional path. The

mold fills upwards with two to three disks receiving fluid in astaggered sequence. Figures 5-9 show how the fluid fillsthe cavity, and Figure 7 includes the color scale forpressure levels in these four figures.

The narrow gate causes a high pressure region whichforces the fluid to flow down the cavity centerline. Itproceeds to splash against the far wall and then wraparound the circumference back to the gate (Figs. 5 and 6).Flow in the angled region of the pouring basin cascadesover the bottom ledge and attaches to the far wall of therunner, as seen in Figure 7. This channeling becomes lesspronounced as fluid volume levels increase.

Finally, two similar but non-uniform depressed regionsform about the centerline. These regions fill from theirperimeter and bottom until completion (Fig. 8). Such apattern is counter, for example, to a steady scenario inwhich a circle of molten Pu encompassing the entire bottomsurface rises as a growing cylinder. Cavity pressurebecomes uniform when the cavity is full. Pressure levelsbuild in the rising well section of the runner, whereimpurities were found to settle in actual casting.

Early test simulations optimized the flow as three poursso that the fluid would never overflow to the funnel, thecavities would all fill completely, and small amounts of fluidwould remain as foundry returns in the angled cavity. Theserates and durations were translated to the single 2.7s pourat 100 cm3 per second used here.

Figure 9 shows anomalous pressure fluctuations whichoccurred as the cavities became completely filled. Multiplesimulations exhibited a rapid change in pressure frompositive to negative and back within the newly-full disk andsurrounding, already-full disks.

The time required to completely fill each cavity is plottedin Figure 10. Results show negligible temperature changewithin the molten Pu during mold filling and, as seen inFigure 11, at fill completion.

ConclusionsNon-uniform cavity filling could cause crystal

microstructure irregularities. In actual casting, a technicianfound that the disks required excess flipping duringmachining and tended to bow.

The small temperature changes seen due to largedifferences in specific heat between superheated Pu andthe graphite mold occurred over a much shorter timescale than the approximately ten minutes required forsolidification in actual casting. Flow regime and nottemperature effects are therefore responsible for castingdiscontinuities.

The bottom cavity takes longer to fill because fluidmust first enter the runner and fill the well. Fill timescontinue linearly until the top two cavities. Pouring basinemptying decreases pressure at the gates which extendsfill time of the top two cavities. A technician reported apreference for working on the lower disks.

The anomalous pressure fluctuations may be due tophysical attempts by the system to reach equilibrium, butthey are more likely due to numerical errors the chosenFlow-3D solver. Unsuccessful tests were performed toremove them by halving fluid viscosity. The fine meshreduced, but didnt eliminate, the extent of the fluctuations.

Fluid Properties Plutonium, pure (Olsen, Jones)

Density (g/cm3) 16.61

Viscosity (Poise) 0.0066

Specific Heat (cal/g-C) 1.76e6

Thermal Conductivity (W/cm-C) 18.8036

Solid Properties Graphite, Carbone Lorraine 2020

Specific Heat (cal/g-C) 0.239

Density Specific Heat (cal/g-C) 0.2975

Figure 4: Actual mold and cast Pu

Figure 1: Mold drawings

Figure 3: Material properties Figure 2: Mold Assembly

0

1

2

3

4

5

6

7

0 1 2 3 4 5 6 7 8 9 10

T

i

m

e

(

s

e

c

)

Cavity Number (bottom to top)

Figure 10: Cavity fill times, from first fluid contact with pouring basin

t =0.70s(dyn/cm2)

Figure 7: Flow cascades over ledge onto runner far wall

(pressure scale for Figures 5-8)

t =0.70s

t =0.75s

t =1.00s

t =1.10s

t =1.20s

t =1.35s

t =1.45s

Figure 5: Bottom cavity fillingfrom runner

t =2.00sFigure 6: Pouring and filling

Final fluid temperaturet =6.31s

(K)

Figure 11: Fluid temperature remains

essentially constant

t = 6.00 - 6.75s by 0.05s increments (l to r, t to b)Figure 8: Edge detection of cavity fill geometry. Two similar depressed areas form

about the centerline. Top cavity shown; same pressure scale as other figures. This final fill phase occurs in each cavity and, under different temperature conditions,

could cause undesired solidification and crystallization non-uniformities.

t =2.95s t =3.00s t =3.05s

Figure 9: Cavity pressures fluctuate to negative immediately before

completely full