A massive chain of property damaging explosions involving an ammunition train occurred at the railroad yard, Roseville, California. The train had pulled into the yard after a night trip of some 100 miles across Donner Summit and down the extended Norden-Roseville grade. Physical evidence confirmed that first explosions were centered at a DODX type boxcar loaded with 250 lb. bombs. Further, bomb cook-off detonation tests established that the triggering bomb blast was not a result of shock loads but rather derived from an engulfing fire initiated in the boxcar wood plank floor under influence of extended heavy braking action on the mountain grade. It was also suspected that high friction composition brake shoes were fitted on the car as replacements for cast iron shoes but the brake mechanical linkage lever ratios had not been modified as required. Results of a comprehensive research program are presented within context of the explosion event, and include analytical computer simulation of train descent profiles on mountain grades through full scale dynamometer tests with actual rail wheels and ultimately more scientific scaled wood floor ignition experiments in the laboratory. The thermal response of a simulated DODX boxcar wood floor was studied through experiments, full scale at a rail wheel dynamometer test facility, and in the laboratory. Certain input data for the wood floor ignition test program were measured on an actual boxcar joined with a freight train consist in transit down the Norden-Roseville grade. Two series of scaled wood ignition experiments were conducted on simulated DODX boxcar floors. Objectives of these tests were to determine: Influence of a cooling air stream on the ignition behavior of radiantly heated wood surfaces, and effectiveness of DODX (stand-off) and AAR (flush) type spark shields in preventing ignition of wood surfaces under radiant heating. It was found that for radiant heat flux levels representative of high friction composition shoes under severe train braking conditions, low speed airflow (wind) exerts a dramatic influence on the wood ignition time. For example, average ignition time for a simulated boxcar floor at a heat flux level of 0.4 cal/cm2sec was determined to be 15.6 min. with a 5 mph wind as compared to 3.6 min. with no wind. In the spark shield effectiveness tests, conducted at heat flux levels representative of cast iron shoes under severe braking conditions, the DODX (stand-off) type spark shield failed to prevent spontaneous flaming ignition of a wood surface directly above it. Under identical conditions, no flaming ignition was encountered with the AAR (flush) type spark shield.
Embedded flame heat flux method for simulation of quasi-steady state vertical flame spread