This paper summarizes a methodology used by the Underground Analysis and Planning System (UGAPS) at Lawrence Livermore National Laboratory (LLNL) for the derivation of probabilistic damage curves for US Strategic Command (USSTRATCOM). UGAPS uses high fidelity finite element and discrete element codes on the massively parallel supercomputers to predict damage to underground structures from military interdiction scenarios. These deterministic calculations can be riddled with uncertainty,

more »

... y when intelligence, the basis for this modeling, is uncertain. The technique presented here attempts to account for this uncertainty by bounding the problem with reasonable cases and using those bounding cases as a statistical sample. Probability of damage curves are computed and represented that account for uncertainty within the sample and enable the war planner to make informed decisions. This work is flexible enough to incorporate any desired damage mechanism and can utilize the variety of finite element and discrete element codes within the national laboratory and government contractor community. 1.0 1/13 probabilistic war planning system, UGAPS must develop a methodology to translate deterministic analysis into probabilistic curves. The methodology presented here is complementary to the work being performed for the Improved Groundshock Vulnerability Number (IGVN) program, where the damage metrics are peak strain and percent (%) tunnel closure. CAVEATS Presented here is a methodology to populate a statistical model with computational and experimental data. Much like the Monte Carlo technique, the model will improve with every additional data point that can be included. Data points include computational simulations as well as experimental data and lessons learned from the Tunnel Target Defeat ACTD. As such, this model accommodates computational uncertainty, intelligence uncertainty, and experimental validation. Conversely, there exists a level of uncertainty from professional engineering judgment that is inserted during intelligence production and interpretation, as well as with the building of computational models. This uncertainty is difficult to bound, especially since UGAPS is the end user -not the producer -of intelligence data. Consequently, this uncertainty is not accounted for in the model presented here. Further, this methodology assumes a "direct hit", thus, ignoring the Circular Error Probable (CEP) of a weapon. CEP uncertainty is a spatial phenomenon and is handled separately. Therefore, the STRATCOM planning metric, Damage Expectancy (DE) is the product of Probability of Severe Damage (PSD) as described in this paper, probability of getting within the CEP, and Probability of Arrival (PA) as defined by the weapon/delivery system reliability. 2/13