Fire Tests and Analyses of a Rail Cask-Sized Calorimeter

Three large open pool fire experiments involving a calorimeter the size of a spent fuel rail cask were conducted at Sandia National Laboratories’ Lurance Canyon Burn Site. These experiments were performed to study the heat transfer between a very large fire and a large cask-like object. In all of the tests, the calorimeter was located above the center of a 7.93m diameter fuel pan, elevated 1m above the fuel pool. The relative pool size and positioning of the calorimeter conformed to the required positioning of a package undergoing certification fire testing. Approximately 2000 gallons of JP-8 aviation fuel were used in each test. The first two tests had relatively light winds and lasted 40 minutes, while the third had stronger winds and consumed the fuel in 25 minutes. Wind speed and direction, calorimeter temperature, fire envelop temperature, vertical gas plume speed, and radiant heat flux near the calorimeter were measured at several locations during each test. Fuel regression rate data was also acquired. The experimental setup and observations pertaining to fire characteristics are described in this paper. Results from three-dimensional fire simulations performed with the Cask Analysis Fire Environment (CAFE) fire code are also presented. Comparisons of the thermal response of the calorimeter to the results obtained from the CAFE simulations are discussed. In general, CAFE underestimated the average internal surface temperature near the top of the calorimeter, while it over estimated the average internal surface temperature on all other sides of the calorimeter. Thus, results showed that CAFE slightly over estimated the overall average temperature of the surface of the calorimeter. INTRODUCTION Large, fully-engulfed objects, such as rail-cask-type spent fuel packages, have a great impact on the surrounding fire environment. To adequately predict incident heat flux to rail-cask-type spent fuel packages, computational fluid dynamics (CFD) models have been employed at Sandia National Laboratories (SNL). Because of the impact that this massive objects have on fires, CFD models must be benchmark against experimental data from tests that have similar size objects [1] to adequately assess the predictive capabilities of the CDF models. Three very large open pool fire experiments were conducted at SNL to gather heat flux and temperature data from pool fires using a calorimeter the size of a spent fuel rail cask. These data were used to benchmark temperature response predicted by the Container Analysis Fire Environment (CAFE), the CFD code used at SNL to analyzed 10 CFR 71.73 regulatory fire cask scenarios. In all tests, the calorimeter was located above the center of a 7.93m (26ft) diameter fuel pan which had approximately 7.58m (2000 gallons) of JP8 per test. The total burn time for each test was greater than 25 minutes. All tests were conducted in relatively low wind conditions (<5m/s) to assure the calorimeter was fully or partially engulfed. This paper presents the pool fire experimental setup and data that was collected. Due to the large amount of data collected, only the data for Test 1 is presented here. A more complete description of the test data is included in Greiner et al [2]. This data includes wind data, fuel pool regression rate, and calorimeter temperatures. The calorimeter internally measured surface temperatures are then compared to results obtained from the CAFE/P-Thermal benchmark run. BACKGROUND CAFE was design to calculate the thermal insult to a spent fuel transportation package using computationally fast and proven numerical methods. To achieve computational efficiency, CAFE relies on some physics-based empirical models to predict the fire environment enveloping the spent fuel transportation package. CAFE uses the finite volume approach with orthogonal Cartesian discretization to solve: (1) the three momentum equations, (2) the mass continuity equation, (3) the energy equation, (4) the equation of state, (5) a number of scalar transport equations for tracking the flow of species, and (6) participating media equations to solve diffusive radiation inside the flame zone and view factor radiation outside the flame zone [3]. CAFE uses a variable density Pressure-Implicit SplitOperator (PISO) algorithm to obtain a velocity field which satisfies both the momentum and continuity equations. CAFE has a number of turbulence models, but for this study a large eddy simulation formulation was used. CAFE only generates the fire conditions outside the external surfaces of the calorimeter. To adequately model the effects of the calorimeter on the fire, CAFE was coupled to P-Thermal to obtain the thermal response of the calorimeter and the subsequent heat flux feedback to the fire. P-Thermal uses CAFE-predicted external cask temperatures, convection coefficients and fluid temperatures to calculate the spatial temperature distribution inside the calorimeter. A specialized mapping scheme is used to transfer this data to the external surfaces of the PThermal, finite element model [3]. The subsequent outer-surface spatial temperature distribution is then used by the CAFE code to adjust the fire response to the calorimeter. The advantage of using this method is that CAFE is able to adjust the fire environment as a result of the threedimensional thermal response of the large calorimeter. Since the development of the CAFE code, there has been a continuing effort to benchmark and fine-tune this fire model by making use of relevant empirical data obtained from experiments that used various size calorimeters and different calorimeter-pool configurations [4-8]. In this study, CAFE was benchmarked against experimental data obtained from a fire test series conducted at SNL Lurance Canyon Burn Site during the summer of 2007. The unique feature of this benchmark effort is that the calorimeter was close to the actual size of a rail cask, and the experiment setup was closely matched to the regulatory hypothetical fire accident scenario outlined in 10CFR71.73 for the certification of nuclear spent fuel transportation casks. In addition, unlike similar calorimeter size benchmark efforts [7, 8] results presented here assesses predictions of CAFE/P-Thermal coupled code using a three-dimensional, calorimeter thermal response model. EXPERIMENTAL SETUP The calorimeter was a carbon steel cylindrical pipe approximately 2.43m (96in) in diameter,4.6m (180in) in length, and a nominal 2.54cm (1in) thickness wall, and had bolted lids on each end [see Figure 1(a)]. The calorimeter was placed on two stands at the center of a 7.93m (26ft) diameter fuel pool. The stands maintained the calorimeter 1m (39.4in) above the fuel surface. Approximately 7.58m (2000 gallons) of JP8 were used per test. Total burn time varied with each tests, but was at least 25 minutes long. All the tests were conducted in relatively low wind conditions (<5m/s) to assure the calorimeter was fully or partially engulfed [see Figure 1(b)]. (a) (b) Figure 1. Large calorimeter fire test: (a) test setup and (b) fire fully engulfing the calorimeter. Thermocouples (TCs) were installed on the interior walls of the calorimeter to measure interior surface temperatures. All TCs on the round walls were installed in a ring configuration as shown in Figure 2. Heat flux gages were placed just outside the round walls of the calorimeter in a ring configuration and outside the lids to obtain incident heat flux measurements close to the outer walls of the calorimeter. Fuel burn rates were measured using a linear array of TCs traversing the depth of the fuel layer at known distance intervals. Directional flow probes were installed just outside of the calorimeter walls to measure the flow speed of the hot gases near the calorimeter walls. Finally, ultrasonic sensors were placed on four towers to measure wind speed and wind direction: (1) two sensor towers aligned with the calorimeter lids and (2) the other two sensor towers perpendicular to the cylindrical section of the calorimeter, on opposite sides. Each tower was approximately 24.4m (80ft) from the center of the pool and had three ultrasonic sensors 2, 8, and 10m (6.5, 26.2, and 32.8ft) from the ground. Fuel Pool Instrumentation Heat Flux Gages Fuel Surface Insulated Wires 39" 72" Heat Flux Gages Thermocouple Rings