Dike Propagation Near Drifts Rev 00, ICN 00 ANL-WIS-MD-000015 April 2000 1. PURPOSE The purpose of this Analysis and Model Report (AMR) supporting the Site Recommendation/License Application (SR/LA) for the Yucca Mountain Project is the development of elementary analyses of the interactions of a hypothetical dike with a repository drift (i.e., tunnel) and with the drift contents at the proposed Yucca Mountain repository. This effort is intended to support the analysis of disruptive events for Total System Performance Assessment (TSPA). This purpose is documented in the development plan Coordinate Modeling of Dike Propagation Near Drifts Consequences for TSPA-SR/LA (CRWMS M&O 2000a). These analyses are intended to provide reasonable bounds for a number of expected effects 1. Temperature changes to the waste package from exposure to magma 2. The gas flow available to degrade waste containers during the intrusion 3. Movement of the waste package as it is displaced by the gas and magma from the intruding dike (the number of packages damaged) 4. Movement of the backfill 5. The nature of the mechanics of the dike/drift interaction. These analyses serve two objectives: to provide preliminary analyses needed to support evaluation of the consequences of an intrusive event and to provide a basis for addressing some of the concerns of the Nuclear Regulatory Commission (NRC) expressed in the Igneous Activity Issue Resolution Status Report (IRSR) (Reamer 1999). The estimates of the number of waste packages at risk and the circumstances of that risk are functional inputs for two technical products: Igneous Consequence Modeling for TSPA-SR (CRWMS M&O 2000b) and Number of Waste Packages Hit by Igneous Intrusion (CRWMS M&O 2000c). 2. QUALITY ASSURANCE This document was prepared in accordance with AP-3.10Q, Analyses and Models, and the development plan Coordinate Modeling of Dike Propagation Near Drifts (CRWMS M&O 1999b), which was, in turn, prepared in accordance with AP-2.13Q, Technical Product Development Planning. A Technical Change Request (T2000-0040) was approved in accordance with AP-3.4Q, Level 3 Change Control. The development of this technical document has been evaluated (CRWMS M&O 1999a) in accordance with QAP-2-0, Conduct of Activities, and has been determined to be subject to the requirements of the Quality Assurance Requirements and Description (DOE (U.S. Department of Energy) 2000). All inputs to this document are listed in the Document Input Reference System (DIRS) for this report, in accordance with AP-3.15Q, Managing Technical Product Inputs. Accepted data as Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 8 identified in the DIRS sheets are identified in accordance with AP-SIII.2Q, Qualification of Unqualified Data and the Documentation of Rationale for Accepted Data. The conclusions presented in this AMR do not affect the repository design permanent items as discussed in QAP- 2-3, Classification of Permanent Items. In addition to the procedures cited above, the following procedures are applicable to this document: AP-2.14Q, Review of Technical Products, AP-3.14Q, Transmittal of Input, AP-6.1Q, Controlled Documents, and AP-17.1Q, Record Source Responsibilities for Inclusionary Records. 3. COMPUTER SOFTWARE AND MODEL USAGE No software subject to the requirements of the AP-SI.1Q, Software Management was used in the preparation of this document. The commercial application software Microsoft Word (Office97) and PowerPoint (Office 97) were used to generate text and to construct figures and flow charts. The software was appropriate for the applications. Two sets of analyses, labeled as the first and second conceptual models, are identified and executed to establish the basis for PA analyses of the response of waste packages to an igneous intrusion. 4. INPUTS The inputs to this document were analyses and ranges of data published in the technical literature. 4.1 PARAMETERS Specific parameters used are listed here in Table 1, Table 2, Table 3, and in Section 6.3.1 under Current Stress State, Section 6.3.2 under Design Description and Section 6.3.4 under Gas Flow. Table 1. Input Parameters and Current QA Status Definition Symbol Value References Input Status Radius of the drift a ~2.75 m DTN: SN9908T0872799.004 Accepted Thermal conductivity of solid basalt K 21.76 J/(kg sec °C) 0.0052 cal/(g sec °C) Hodgman et al. 1955, p. 2251, Table Heat Conductivity Accepted Fusion temperature of magma Tf 1046-1169 °C CRWMS M&O 2000d, Sec. 6.2.3 Technical Product Output Density of magma . 2484–2663 kg/m3 CRWMS M&O 2000d, Sec. 6.2.4 Technical Product Output Waste container skirt None ~.225 m CRWMS M&O 1999b Accepted Container separation None ~.10 m DTN: SN9908T0872799.004 Accepted Mass of Waste Package M 42.3 metric tons CRWMS M&O 2000e, p. II-2 Technical Product Output Drip shield perimeter None ~6.5 m DTN: SN9908T0872799.004 Accepted Notes: Data tracking number (DTN) Quality assurance (QA) Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 9 Table 2. List of Exsolved Gases Species Mole Fraction References Input Status H2O 73.16 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output H2 1.17 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output CO2 14.28 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output CO 0.57 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output SO2 9.45 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output S2 0.37 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output HCl 0.49 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output HF 0.06 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output H2S 0.74 CRWMS M&O 2000d ANL-MGR-GS-000002 Technical Product Output Table 3. Properties of Exsolved Gases Species Critical Temperature K Critical Pressure Atm Molecular Wt. References Input Status H2O 647.14 217.7 18.01 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted H2 32.97 12.76 2 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted CO2 304.14 72.78 44.01 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted CO 132.91 34.53 28.01 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted SO2 430.8 77.8 64.06 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted S2 1314 204.28 64.12 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted HCl 324.7 82 36.46 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted HF 461 63.95 20.01 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted H2S 373.2 88.23 34.08 Lide and Frederikse 1997 p. 4-37f, p. 6-50f Accepted Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 10 Reference information was obtained from the Technical Information Center and the Technical Data Management System, which is a controlled source. Sources are cited in the text where applicable and in Section 8, References. . 4.2 CRITERIA These analyses are investigatory and intended to provide estimates of the sizes of the effects of dike propagation. No criteria applicable to these analyses have been identified. 4.3 CODES AND STANDARDS No codes or standards were used in these analyses. 5. ASSUMPTIONS Assumptions are described here in general terms and are discussed in more detail in the sections in which they are used. 5.1 GAS FLOW As real gas is supplied to the drift by the dike or vent: • There is isothermal flow down the drift (no axial temperature gradient). • Adiabatic expansion is ignored. • Heat carried into the rock (the porous media) by the gas is ignored. • There are no chemical interactions of the gases with the rock. • The components of the engineered barrier system (EBS), including backfill, are irrelevant to the flow. Basis–Gases exsolved from magma are real gases that obey the real gas law and satisfy the theorem of corresponding states as discussed (Pirson 1977). Confirmation Status–These assumptions are shown to be conservative as a result of the analysis and do not need to be confirmed. Use in the Analysis–These assumptions are used in Section 6.3.4 in the computation of gas flow. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 11 5.2 SOLIDIFICATION OF MAGMA Estimates and calculations of the phase change behavior occur as discussed in Jaeger (1964), Kreith and Romie (1955), Soward (1980), Stewartson and Waechter (1976), Riley et al. (1974), and Pedroso and Domoto (1973). Drift wall temperature is assumed to be ~ 600°C. Latent heat of fusion (L) and the specific heat (C) are assumed to be 100 cal/gm (upper limit) and ~ 0.3, respectively (from Jaeger 1964). Basis–The literature provides a reasonable basis for magmatic properties for this analysis and data specific to Yucca Mountain has been provided (CRWMS M&O 2000d). Confirmation Status–The assumptions are taken from the literature and represent typical or best available information and do not need to be confirmed. Use in the Analysis–The cited analyses and estimates are used in Section 6.3.5 and referenced in Section 8.1. 5.3 CONTAINER A container of ~ 42.3 metric tons (CRWMS M&O 2000e, p. II-2) will be a representative package moved by the pressure pulse. Basis–Design information is taken from the current design as indicated by the citations. Confirmation Status–The description of the waste package is obtained from design data and is confirmed in those citations. Use in the Analysis–The design information is used in Section 6.3.3. 6. ANALYSIS 6.1 INTRODUCTION A long-standing problem in understanding the nature of possible volcanic disruption of a repository has been the physical details of how a dike interacts with a repository drift (and with a set of drifts) (Barr et al. 1993, p. 141). This problem of a dike interacting with a repository drift is not amenable to direct simulation—direct calculation of the interactions. Simulation would have allowed systematic examination of the most important processes. Rather, the problem is attacked piecemeal by examining end members or limiting cases of processes which can be modeled, and by interpretation, in terms of the Yucca Mountain site, of geologic observations of relic cinder cones and active volcanic systems believed to be analogous. End members (or bounding cases) are chosen by analysts because they provide an expeditious way to identify and explore physically possible dike/drift interactions for the consequences of the interactions. Some interactions can be rejected on the basis of negligible consequence and those needing more careful examination can be identified. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 12 The results are often simplified because the fundamental driving processes producing the intrusion, extrusion and the volcano are not well-formulated at present. The interpretations are used to produce plausibility arguments about the interaction of the intrusive dike with a repository; and, however well argued, are the matters of contention between the Project and the NRC (Reamer 1999). This AMR is intended to provide a better focus on identifying the detailed problems, why they might be important, and how they are related. The AMR is also intended to provide simple estimates for some of the effects, based only on analyses and interpretations already provided in the literature. 6.2 BASIC ISSUES This AMR addresses and attempts to provide context for several end member cases of the dike propagation problem, and to provide a framework for the dike/repository interaction problem as a whole. Because, in the absence of simulation, modeling decisions are made based on plausibility arguments constructed around their interpretations of analogue data and reasoned speculations, part of that context is presented in the form of a “decision tree.” The decision tree, Figure 1 (p. 39), presents where those decisions of what is plausible (represented by “yes” or “no” choices) are made, and how they are interrelated. A simple decision tree allows identification of the modeling choices made to reach surface release. The tree entries which receive the most interest are the final components, namely the formation of the ash plume, and its distribution and deposition. However, getting to the releases requires an assembly of components describing the details of the interactions. Some of those details are indicated on the tree as ovals, referred to as “conditionals.” These represent some of the supporting work which is required in order to make decisions about how the dike might reasonably be expected to interact with a drift. The topics of this AMR are analyses that feed these decisions. For a general discussion of dike development and properties refer to Characterize Eruptive Processes at Yucca Mountain, Nevada ANL-MGR-GS-000002 (CRWMS M&O 2000d). Interaction of a dike with a drift, based on velocity of propagation of the dike, is first a mechanical interaction and later, when connecting pathways are established, a thermal and fluid flow interaction. In order to provide an overall perspective of the interaction of a dike and a drift, it is necessary to discuss when and how current interpretations fit together. 6.3 ANALYSES There are two basic conceptual models of how a dike and drift might interact. The first presumes that a drift is a relatively insignificant heterogeneity in the rock, which is intersected and otherwise provides no interaction with the dike as it propagates to the surface. Essentially, the forces driving propagation of the dike are so large that the drift is only a minor perturbation. The result is a planar (slab-like) intersection of the dike and drift—the idealized interpretation. The dike-repository interactions are then (1) direct physical entrainment of waste in the dike and (2) flow of gas and magma, or gas and fragments, down the drift to interact with waste containers and waste. These two kinds of interactions involving an idealized intersection of drift Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 13 and dike, are pursued in Sections 6.3.3 Shock Interaction, 6.3.4 Gas Flow, and 6.3.5 Magma Flow—Thermal Environment. Since energy dissipation has been ignored, the idealization would be expected to produce stronger effects in the drift than would really occur. The second conceptual model presumes that the dike, which propagates by means of a selfgenerated crack, (Spence and Turcotte 1985; Turcotte et al. 1987) interacts strongly with the stress-altered region (which forms around the drift) and with the void space in a drift. It is currently thought that dikes are emplaced normal to least principal stress as measured regionally in the near-surface (Tsunakawa 1983; Delaney et al. 1986). For this conceptual model, the thermo-mechanical state of the mountain (and repository) is important. This is the case because the dike is interacting with a stress field which is evolving and therefore the stress-state is not fixed in time. That is, when the drift is driven, the surrounding rock first tries to relax into the drift. When waste is added, the waste heat causes thermal expansion which puts the rock around the drift into compression. This compression relaxes as the repository cools. These stress alterations are large enough so that the least principal stress (currently NNW-SSE) is rotated to vertical (Hardy and Bauer 1992); a circumstance which could alter how a propagating dike behaves as it encounters the stress-altered zone (perhaps producing a sill rather than a dike). After several thousand years, as the mountain cools, the least principal stress rotates back to its original direction. Of these conceptual models, the more complete description from current literature is the second (Watanabe 1999), which will be pursued in this analysis in Sections 6.3.1 Thermal-mechanical Evolution of the Repository and Mountain and 6.3.2 Interaction with the Drift. This second, physically more complete conceptual model functions to establish the constraints and qualitatively estimate the bounds that apply to the detailed interactions analyzed for the simple idealized model. Performance Assessment (PA) uses the first conceptual model as stated in Igneous Consequence Modeling for the TSPA-SR, ANL-WIS-MD-000017 (CRWMS M&O 2000b). PA also uses the second complex model to provide constraints on the first model. The expected dike-repository interactions are (1) propagation of the dike into and across the drifts as described by the intersection of the dike crack with the drift in an evolving stress field, (2) flow of gas and magma, or gas and fragments, down the drift to interact with waste containers and waste, and (3) direct physical entrainment of waste in the dike. The specific problem of waste entrainment has been treated in detail elsewhere (Wilson et al. 1994; CRWMS M&O 2000b (ANL-WIS-MD-000017)); CRWMS M&O 2000f (ANL-WIS-MD- 000005) and will not be further examined. The analyses are laid out as indicated in the decision tree in Figure 1, where (1) rhombuses enclose questions, (2) rectangles enclose the analyses implied by the answers, and (3) ovals are conditionals referring to supporting information. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 14 6.3.1 Thermal-Mechanical Evolution of the Repository and Mountain Analysis of the interaction of a dike with the repository requires identification of the time of occurrence and specification of the design of the repository. The initial interaction of a propagating dike with the repository is a mechanical process involving the interception of the crack propagating ahead of the dike with the void space of the drifts and the stress-relieved zone around the drifts (Brady and Brown 1985, p. 192f). After the initial interaction, the interaction becomes a fluid flow and thermal process. The expected direction of dike emplacement is controlled by the stress state and is generally normal to the least principal stress (Pollard 1987). Because the mechanical stress state of the mountain is changing as an effect of repository heat, the time dependence of the changes needs to be recognized explicitly. Accordingly, future time is separated into the thermal period, when the mountain is heated enough for the least principal stress to have rotated to vertical at the repository horizon, and the post-thermal period when the stress has returned approximately to its present orientation. This separation is honored in Figure 1, the decision tree, with the first branching. Current Stress State–According to hydraulic fracturing stress measurements at and around Yucca Mountain, it appears that the least principal stress direction is horizontal about N60W as derived from the following references: • N60W–N65W (Stock et al. 1985) • N51W–N52W (Warren and Smith 1985) • N68W (Warren and Smith 1985) • N55W–N60W (Frizzell and Zoback 1987). The minimum horizontal stresses measured were 2.6 to 3.1 MPa (Warren and Smith 1985) at 330 m and 4.2 to 5.4 MPa (Stock et al. 1985; Stock and Healy 1988) at 295, 418, and 646 m. Vertical stresses were 6.1, 8.4, and 12.9 MPa, respectively. The experimental determinations included measurements at Yucca Mountain wells (Stock et al. 1985), at G-Tunnel (Warren and Smith 1985; Smith et al. 1981), and at Hampel Wash (Frizzell and Zoback 1987) nearby on the Nevada Test Site. Expected Changes to the Stress State–Calculations of the stress state at the drifts (Hardy and Bauer 1992) were performed for design considerations, to address drift stability. Their calculations were based on the thermal loading of 57 kW/acre and an earlier layout design (Hardy and Bauer 1991, p. 5-15, p. 5-16). The analysis indicates that the three principal stresses increase, with the horizontal components exceeding the vertical component within a few decades of closure. The details of the time-dependent change of stresses around emplacement drifts and the main access drift differ somewhat, but both drift analyses lead to the conclusion that the least principal stress quickly becomes vertical. Analyses performed for the NRC (Mack et al. 1989) show similar results. Horizontal stresses at the drift alter from 3–6 MPa to 15–20 MPa. Based on this work (Hardy and Bauer 1992) it appears that, at the repository level, the vertical least principal stress will persist for the order of 2000 years (Figures 2 and 3, (p. 40)). The largest principal stress (the vertical component) is initially at about 7 MPa and the least principal stress is about 3.5 MPa (Note that the ordinate scales of the figures differ by about a factor of two.). Because of the drift alignment used, the axial stress indicated in the figures is along the Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 15 maximum principal horizontal stress. Figures 2 and 3 show that the least principal stress (horizontal at ~N60W) increases to exceed the vertical stress component and does so for about 2000 years. The increases of the stress components at the main access drift are due to thermal expansion of the rock around the emplacement drifts. Although the calculations cited (Hardy and Bauer 1992) are directed at the level of the emplacement drifts, much more rock will be involved. This rotation of least principal stress can be expected to extend several hundred meters below and out from the repository horizon (the region affected by repository heat)(Mack et al. 1989). Figure 4 (p. 41) presents a diagram that shows the extent of influence. The region of alteration in which the least principal stress has been rotated is time dependent and eventually returns to its approximate horizontal orientation. Possible Effects of Stress Rotation–Rotation of least principal stress from horizontal to vertical changes the stress field through which the dike and its leading crack must penetrate. If, as the literature indicates (Spence and Turcotte 1985; Turcotte et al. 1987, Stock et al. 1985), the dike will deflect to be normal to least principal stress, then the direction of dike propagation would be expected to be deviated. Figure 5 (p. 42) illustrates the changing stress state and the possibility of deviating the path of the dike from vertical, including possible sill formation. In Figure 5, Vo is the initial vertical principal stress, H0 is the larger horizontal principal stress and h0 is the least principal stress. These stresses are altered by the addition of thermally derived stresses from repository heat, (TS). For simplicity, Figure 5 (as modified from Anderson 1963) shows the addition to be linear with multiplying factors of a and b compared to the effect on h0 (see Figures 2 and 3 for a better estimate of growth and relaxation). At some point the line for h0 crosses that for V0 and the least principal stress is then vertical. Accordingly, it is then possible for the direction of dike to deviate from the vertical and for a sill to develop (see the analysis of Anderson 1963, pp. 50–53). This circumstance persists until cooling of the mountain allows rotation of the least principal stress back to horizontal. Since the Yucca Mountain block has bounding faults (e.g., Solitario Canyon, Bow Ridge), in addition to sill formation, a possible consequence is deflection or redirection of the dike propagation to these fault zones (Wilson et al. 1994, v. 1, pp. 2–24). These faults, which strike mostly north-south, have a strike that is only about 30 degrees from the direction of regional least principal stress. Because the way in which the thermal stress from the repository will affect these near-by fault zones is unknown, deviation of the magma flow up the fault zones to the surface is a possible alternative consequence to sill formation. As a result of these arguments on the changing stress state of Yucca Mountain, the decision tree of Figure 1 distinguishes the thermal period from the post-thermal period. There are additional, more detailed, effects on the drifts during the thermal period (Mack et al. 1989) which affect the actual dike/drift interaction, but which are not discussed because they are germane to other issues (e.g., closure of water-bearing fractures leading to the drift). Alternative Interpretations–The Igneous Activity Issue Resolution Status Report (IRSR) (Reamer 1999, p. 52f) considers the details of stress around and at Yucca Mountain on two scales, the regional scale involving the tectonic setting and extension (Reamer 1999, p. 43f), and the local scale (Reamer 1999, p. 52). These issues arise in Probability Criterion 5, the Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 16 probability of occurrence of igneous activity. The discussions under that Criterion are related to structural control of an intrusion and the ability of an existing fault to influence the propagation of a dike. An argument is presented for dike emplacement perpendicular to the least principal stress and parallel to the principal horizontal stress (Reamer 1999, p. 52). The time-dependent alteration of the local stress-state around Yucca Mountain (and possible effects on dike injection), due to repository heat, does not appear in the models discussed in the IRSR. 6.3.2 Interaction with the Drift It appears that dikes propagate by fluid-induced fracturing of the confining rock (Lister and Kerr 1991; Lister 1990a, 1990b; Spence and Turcotte 1985). A dike is preceded by a crack generated by the magma fluid pressure (Figure 4). The driving mechanism for the crack leading the dike is the buoyancy of the magma column relative to the adjacent rock column (due to the density difference) (Lister and Kerr 1991; Lister 1990a, 1990b; Pollard 1987; Spence et al. 1987; Spence and Turcotte 1985; Watanabe et al. 1999). Advance of the dike depends on the local differences between the buoyancy and the viscous pressure loss and the elastic stress in the rock. The total effective buoyancy, P, in the crack drives the fluid flow up the crack and propagates the crack in a manner which can be represented by equation 1. P = (.f - .r )gh + pe + pv (Eq. 1) where .f = fluid density .r = rock density g = acceleration of gravity h = height of the dike pe = elastic (non-hydrostatic) pressure exerted by crack walls pv = viscous pressure loss (Lister 1990b, p. 265). With insightful analyses of the fluid mechanics of viscous fluids, interpretations of emplacement of many dikes and sills, and a few data on dike injection velocities (Lister and Kerr 1991; Aoki et al. 1999), various authors have established the general behavior of dike propagation and sill formation (Lister and Kerr 1991; Lister 1990a, 1990b; Pollard 1987; Spence et al. 1987; Spence and Turcotte 1985, 1990; Watanabe et al. 1999). (Equation 1 is a simple representation for these more complicated analyses.) These analyses derive (and depend on) the interpretations that dikes propagate by means such that the rate is in the range of a few meters per second to 1 × 10–7 m/sec (Lister 1990a; Aoki et al. 1999). Once a dike has reached the surface, flow velocities, now less impeded, can be larger. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 17 Two of these interpretations will affect the analyses of the following sections: there is a fluidfilled leading crack, and the plane of the dike tends to be normal to least principal stress. Fracture/Fracture Interaction–The buoyancy-driven magma fracture eventually interacts with the drift. The drift consists of a void space—the original opening with waste packages, drip shield and backfill—and the stress-relieved region around the opening. This stress-relieved region develops as country rock relaxes into the openings. Classically, the stress-relieved region extends out to about 3 drift diameters (Brady and Brown 1985, p. 192f; Jaeger and Cook 1979). Because the thermal output of the repository has significantly altered the local stress conditions (Hardy and Bauer 1991; Mack et al. 1989), the extent of stress-relief depends on when, in repository history, the dike intrusion is presumed to occur. During the thermal period (Figure 1, left branch), there is no stress-relief, rather the drift is experiencing compression (Mack et al. 1989). Stress relief appears as strain adjustment, that is, concentric and radial fractures develop around the drift, accompanied by stoping (chimneying) in the drift. This fracture/fracture interaction is intended to provide input for each of two decisions on parallel branches on the right hand side of the tree. To establish how the dike fracture might be expected to interact with the drift, consider the case of no stress-relief, so that the interaction can be approximated as a linear crack (the dike crack) encountering a “penny-shaped” crack (the drift). This approach is extremely conservative because the drift does not provide a stress concentration comparable to that of a crack, and as will be developed, the dike can not resume propagation simply by expanding a penny-shaped crack. Further propagation of the dike across the drift is delayed until part of the drift has filled with magma and pressurized sufficiently to initiate hydraulic fracturing (conditions for hydraulic fracturing are discussed in part in Reamer 1999, p. 78f). The driving source for fluid fracturing of the rock is the magma fluid pressure in the dike (and the crack). For a steady-state linear crack, this pressure is related to material properties and the least principal stress as: Pl - s = ( 2E./pL(1 - µ2))1/2 (Eq. 2) (Daneshy 1978, p. 33). If the linear crack intersects the drift normal to the drift axis, the drift is treated as a “pennyshaped” crack. The pressure necessary to propagate the penny-shaped crack is given by PD - s = (pE./2R(1 - µ2))1/2 (Eq. 3) (Daneshy 1978, p. 33) Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 18 where Pl = pressure (linear crack) PD = pressure (penny-shaped crack) L = length of a 2-D linear crack R = radius of the “penny-shaped” crack E = Young’s modulus for the rock µ = Poisson’s ratio . = effective fracture surface energy s = in situ least principal stress The ratio of the minimum driving pressures is a measure of how easily the linear crack interacts with the “penny-shaped” crack and is Pl - s ~0.45(PD - s ) (Eq. 4) where L = 2R in equation 2. So, a linear crack, being propagated at minimum driving pressure, has only 45 percent of the pressure required to drive a penny-shaped crack (Daneshy 1978). So even for interaction with a planar segment of the drift, the dike is not able to continue propagation across the drift by converting a linear crack into a penny-shaped crack. To continue propagation, a mechanism is required for loading the drift so it can fail along a structural zone of weakness by restart of a linear crack. A possible mechanism is loading of the drift, that is, filling the drift with sufficient magma to initiate hydraulic fracturing. Loading of the drift sufficient to initiate fracturing is discussed in the IRSR (Reamer 1999, p. 78f). As a result, if the drift is subsequently loaded, the linear crack resumes along a line of weakness, probably along a rib (drift wall). While the section of the drift is filling (which will reinitiate hydraulic fracturing), the crack leading the dike is advancing vertically in the pillars between adjacent drifts. Local delay at the drift could mean that the flow of magma is diverted preferentially to the region between drifts. In the current design, the diameter of an emplacement drift is ~5 m (DTN: SN9908T0872799.004) and drift centers are 81 m apart (CRWMS M&O 2000g), so the length of the dike (for a dike normal to the drift axis) is about six percent of the drift spacing. This suggests the possibility that a dike may tend to divert around a drift, leaving it locally filled with magma. The time to fill a section of drift can be estimated from the void space assumed for a length of the drift to be filled, the dimensions of the dike and the flow velocity in the dike (greater than the dike propagation velocity). The dike propagation velocity then allows an estimate of how far the dike has advanced before a linear fracture resumes advancement from the drift. Local Stress State–The discussion of the thermo-mechanical evolution of the mountain (Section 6.3.1) argues for a long delay (~2000 years) (Hardy and Bauer 1992) for thermo-mechanical stress relief to occur around repository openings. Stress-relief, when it occurs, is expressed in generation of fractures and rockfall and possibly as creep into the drift. In classical rock mechanics (Brady and Brown 1985, p. 192f; Jaeger and Cook 1979), an opening (drift) develops Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 19 radial and concentric fractures as strain relieves the stress, to some extent, for a distance of up to 3 drift diameters. In Figure 1, this is illustrated in the right branches of the decision tree. The propagating crack from the intruding dike encounters the stress relieved zone and the associated fractures some distance away from the original location of the drift. Effectively, the propagating crack sees a fractured rock before it reaches the drift. There are two possible end members for the dike-drift interaction. In the first, the propagating crack encounters discontinuous concentric and radial fractures. From the work of Tsunakawa (1983), propagation of magma-filled cracks becomes impeded as the angle of intersection of the dike crack and the local fractures increases. Further, Tsunakawa establishes a requirement for minimum magmatic pressures (depending on angle) for assured propagation. For the second end member, the propagating crack intersects a radial fracture that persists to the drift. Such stress-relief radial fractures, while approximately parallel to the drift axis, are short and discontinuous and would not be expected to significantly rotate the injection direction of the dike. Fragmentation History–Fragmentation is the complex process of exsolution of gas, formation of bubbles in the magma and solidification and disruption of the bubbles to form pyroclasts. For flow up an existing dike or up a conduit through the repository, fragmentation may occur below the repository depth. Figure 1, the organizing decision tree, has a pair of right-hand branches dependent on when in the course of propagating the dike that the fragmentation occurred below the repository. The issue is whether, in the course of formation, the magma propagating the dike fragmented below the repository level, or whether it propagated to the surface first and then a pressure relief wave moved back down the intrusion to allow fragmentation at the fragmentation depth. The first case is one of the two circumstances in which magma, as a liquid, is available to flow into the drift (the second is late stage degassed flow). In addition, the pressure at the drift level for the flow of ash and shards (with a density of perhaps 1000 kg/m3) can be less than that for a magma flow (with a density of 2556 kg/m3). [The value for magma density is taken from Characterize Eruptive Processes at Yucca Mountain, Nevada, ANL-MGR-GS-000002, CRWMS M&O 2000d, for 2 wt% water content.] There are two reasons for concern about whether fragmentation occurs during driving of the dike. The first is that magma has to be available at the drift in order to flow into the drift. The second is exsolution of gas and fragmentation into the drift. The drift is at ~0.1 MPa (atmospheric pressure) compared to ~7.5 MPa (~1100 psi) in a liquid magma flow and perhaps ~3.1 MPa in a fragmented flow. The fragmented flow, with a high gas content, is more likely to be susceptible to loss of pressure through fractures around the drift that are intercepted by the dike, with a reduction of flow of pyroclastics down the drift. The pressure, p, at the drift is calculated from p = .gh, where . = density of contents of the dike (magma or fragments), g = acceleration of gravity, and h is the distance to the surface. P is then the pressure at the drift necessary to support the dike. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 20 Design Description of a Drift–Current design of an emplacement drift includes the following components: ground support, invert, rails, waste packages, emplacement pallet, drip shield, and backfill, (A drain is a possibility but is not part of current design.). The dimensions are shown in Figure 6 (p. 43)(DTN: SN9908T0872799.004) and Figure 7 (p. 44) is a sketch of the general configuration. Waste containers have a skirt about 25 cm long at each end (CRWMS M&O 1999b) and adjacent containers are separated by about 10 cm along the drift (DTN: SN9908T0872799.004). Alternative Interpretations–The Igneous Activity Issue Resolution Status Report, in Section 4.2.3.3.2 (Reamer 1999, p. 78f), develops requirements for hydraulic fracturing which is initiated from a loaded drift. That is, the drift is filled with magma and two orientations for reinitiation of hydraulic fracturing are considered. These directions are normal to the drift axis and along the drift axis (in present design, normal to least principal stress and normal to maximum horizontal principal stress, respectively). The details of the mechanical interaction of the crack leading the dike with the drift have not been considered. The analyses address how a dike may be propagated by hydraulic fracturing of the rock around a drift pressurized by magma. In present repository design, the drifts are oriented roughly along the current least principal horizontal stress. Since the most likely orientation of a dike is normal to the least principal stress (Reamer 1999, p. 52), an intersection of a dike with a drift along the axis of the drift is unlikely. Therefore, this possibility has not been considered in this AMR. 6.3.3 Shock Interaction Intersection of dike and drift will result in a pressure pulse generated by the rapid exposure of the drift to the ~7.5 MPa (1100psi) pressure of the dike. This pressure pulse, which moves rapidly down the drift, interacting with the packages, drip shield and backfill. The backfill is displaced and compressed by the moving drip shield to close the openings above and around the waste packages and drip shield. The pressure pulse is followed by a flow of pyroclastics down the drift. This flow of pyroclastics reaches and reinforces (and may move) the blockage of jammed drip shield and backfill which was generated by the pressure pulse. Air in the drift is at ~0.1 MPa (atmospheric pressure) while the dike volatiles are about 7.5 MPa (~1100 psi) for a liquid magma dike and ~3.1 MPa for a fragmented flow producing the dike (pressures are derived for a dike which reaches the surface, with densities of the magma and pyroclastic flow are 2556 kg/m3 and 1000 kg/m3, respectively). A sudden connection means that leakage of gas and magma to relieve pressure through any fractures around the drift is precluded. The decision tree in Figure 1 shows two choices: either liquid magma reaches the drift with rapid exsolution and decompression or fragmented flow reaches the drift with a smaller decompression. The topic of this section is intended to allow a decision about the scope of the disruption. Figure 6 shows a number of drift components for the pressure pulse to interact with. It is expected that waste packages will be displaced and the drip shield, backfill and invert will be moved. Waste packages falling directly in the dike are not considered here, they are considered in Igneous Consequence Modeling for the TSPA-SR, ANL-WIS-MD-000017 (CRWMS M&O 2000b). Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 21 To estimate waste package movement, the pressure pulse is applied uniformly across the head (end cap) of the waste container. Figure 7 is a sketch of the pre-intrusion layout in the drift. The waste package slides along the emplacement pallet until the center of gravity of the waste package passes the emplacement pallet end (or the emplacement pallet otherwise collapses). Containers have an attached skirt about 25 cm long and are separated by 10 cm. When the first waste package moves down the drift more than 10 cm, skirts are in contact and crushed thereby moving the adjacent waste package. To estimate the velocity of a waste package it is necessary to have a mass for the waste package and a coefficient of friction. The container is taken to be a 21 assembly PWR container. The waste package mass is calculated to be ~ 42.3 metric tons (CRWMS M&O 2000e, p. II-2). The waste package rests on an emplacement pallet with its motion resisted by a coefficient of friction of 0.3 (The specific value will turn out to be unimportant because the driving pressure is so large.). The interaction is idealized in that the pressure pulse provides a constant acceleration until the shock wave reaches the end of the waste package. From Glasstone (1962, p. 121f) the shock wave velocity v in air is given by U = Co(1 + 6P/(7Po))1/2, with Co = ambient speed of sound, P = overpressure, and Po = ambient pressure. For p = 7.5 MPa, the dike pressure, U ~ 8Co. The dynamic pressure q (= 5/2 • p2/(7Po + P)) is ~17 MPa (Glasstone 1962, p. 122). Since the incident shock wave is reflected from the waste package end cap, the overpressure is approximately twice the incident overpressure and the total driving pressure, P = P (overpressure) + q (dynamic pressure), is ~32 MPa. In this idealized case the force F , on the head of the waste package, is given as F = (A P - µ M g) = M a (Eq.5) where A = the cross-sectional area of the container (~2.19 m2) P = the total pressure µ = the coefficient of friction (nominally 0.3) M = the mass of the waste package (represented by a 21 assembly PWR waste package of 42.3 metric tons (CRWMS M&O 2000e, p. II-2) g = the acceleration of gravity (9.8 m/sec2) a = the acceleration of the waste package Integration gives the velocity v as Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 22 v = t (A P - µ M g)/M (Eq. 6) where t is the length of time the force is applied as the waste package slides down the emplacement pallet. The driving pressure accelerates the waste package until pressure has built up on the back end cap. The time for buildup (Glasstone 1962, p. 184), t = L/U + 4R/U (L = container length, R = container radius), is t ~ 3.27 × 10-3. Substitution of this value of t into equation 6 yields a velocity v ~ 5.4 m/sec. Details of shock buildup and decay (Glasstone 1962, p. 182f), which would reduce the driving pressure (perhaps by more than a factor of 2) and the velocity, have been ignored. For small velocities, the waste package would be expected to stop moving almost immediately on contacting the invert. (Based on a train derailment analogy, rail cars of 50–100 tons each, stop fairly quickly in the track ballast even though derailment occurs typically at much higher speeds). On this basis, it appears that the first waste package will move until it drops off the emplacement pallet, presumably <2 m, when the center of gravity of the waste package passes the end. Collapse of the emplacement pallet, which would further reduce motion, is undetermined and is conservatively neglected. Since the waste package then stops quickly, its translation affects adjacent waste packages with much of the movement accommodated in crush of the skirts. The approximately 2 m of movement is accommodated by movement of 3 to possibly 4 waste packages. If the pulse is produced by a fragmented flow then the pressure and the velocity scale according to the equation for velocity, v, with the density of the flow (magma ~2556 kg/m3, fragmented flow <1000 kg/m3). (A diagram showing a final orientation of the disrupted waste packages appears in Figure 8 (p. 45)). The pressure pulse interacts with the other components of the drift as well. The interactions are not amenable to an idealized calculation and are treated qualitatively. As shown in Figure 7, the waste package sits on a emplacement pallet above a ballast invert and is covered by a drip shield loaded with backfill. If the drip shield functions as designed, diverting water from the packages, there are some few locations along a drift where the backfill is wetted by the drip. From Figure 6, the load on top of the drip shield is <10 Pa. Accordingly, it is to be expected that the pressure pulse will displace the drip shield and both the backfill and invert material as well. The drip shield (Figure 6) has a circumference of about 6.5 m (DTN: SN9908T0872799.004) and it will be made in short lengths for emplacement. The pressure pulse can be expected to displace the drip shield upward and down the drift and to jam it into the void space above the backfill (“snowplowing” the backfill). From Figure 8, the drip shield and backfill associated with one waste package appear adequate to block the opening (~13 m2 per segment); however, since three waste packages (and possibly 4) are likely to be displaced by the pressure pulse, Figure 8 shows drip shields from three waste packages involved. Figure 8 portrays displacement of the ballast, its accumulation along the damaged waste packages, and upon close inspection, tries to include mechanical damage to the waste packages (everted heads and crumpled sidewalls). The extent of damage depends on both dike flow characteristics and the age, that is to say the strength of the affected waste packages. Splitting of welds and waste package blowout are possibilities that are not addressed. The first waste packages has been treated as a rigid, non-deformable body Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 23 (except for the skirt), a conservative assessment since damage to the first waste package would reduce damage to waste packages further down the drift. Note that since failure of ground support and rockfall have not been considered here, progression of the plug of magma down the drift may be substantially overestimated. Alternative Interpretations–The Igneous Activity Issue Resolution Status Report (IRSR) (Reamer 1999, Section 4.2.3.3.1 Flow Conditions, p. 77f) discusses response to the pressure pulse and possible shock wave and is based on scoping calculations done for the NRC. The study presumes that backfill is absent from the drifts and on intersection, the pressure in the dike will fall to atmospheric pressure. The dike will then decompress and magma will have a tendency to flow into the drifts. The IRSR argues that the capacity of a drift is so large compared to the expected dike width (0.5–2m, in their case) that, locally much of the flow is diverted to the drift. Because of the large pressure drop the IRSR authors anticipate an acceleration of magma into the drift at a much larger flow rate than is anticipated. The IRSR authors expect flow speeds for magma down the drift of the order of 100 m/s. The IRSR authors also believe that a shock wave will proceed down the drift, reaching the end in about 10 s (1000 m in 10 s). For this AMR, the interaction of dike and drift starts with a pressure pulse generated by the rapid exposure of the drift to the ~7.5 MPa (1100 psi) pressure of the dike. A pressure pulse moves rapidly down the drift, interacting with the waste packages, drip shield and backfill. The backfill is displaced and compressed by the moving drip shield to close the openings above and around the waste packages and drip shield. The pressure pulse is followed by a flow of pyroclastics down the drift. This flow of pyroclastics reaches and reinforces the blockage generated by the pressure pulse. Repressurization of the drift to allow the flow from the dike as magma is not considered because the permeability of the formation is expected to be too high. An estimate is made of the number of waste packages experiencing physical damage. 6.3.4 Gas Flow An issue, which appears in the branches of Figure 1, is that of failure of the waste container. One possible cause of failure is corrosion of the container by the aggressive gases exsolving from the magma intrusion or flowing in the dike or in the vent. This section attempts to estimate the gas flow down the drift under idealized circumstances, in order to estimate the size of a source term to allow a decision about whether corrosion is a possible problem (whether sufficient gas for corrosion is available). The idealization is that the gas flow is a steady-state flow of a real gas, as controlled by leakage through the drift wall, which persists for the duration of the eruption. More specifically (as listed in Section 5.1), it is assumed that 1. Gas is supplied to the drift by the dike or vent. 2. There is isothermal flow down the drift (no axial temperature gradient). 3. Adiabatic expansion is ignored. 4. Heat carried into the rock (the porous media) by the gas is ignored. 5. There is no chemical interaction of the gases with the rock. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 24 6. The components of the EBS are irrelevant to the flow. Assumptions 2–6 are those made explicitly for this calculation. They ignore dissipation of energy to the drift wall and the EBS and are intended to be conservative because more energy is retained in the flow. The analysis based on these assumptions overestimates the gas flow. These assumptions mean that the calculation is for steady-state flow down an empty drift. The volumetric flow rate is controlled only by how fast the gas can escape by leaking through the drift wall into the country rock, and real gas properties are assumed to be more important than heat transfer into the porous medium. The flow of a real gas through a porous medium is described by the Equations 7, 8, and 9 2 . m(p) = 1 ) ( ) ( - k p c p fµ t p m . . / ) ( (Eq. 7) where m(p) is the pseudo-pressure defined as m(p) = 2 .p pm p z p pdp ) ( ) ( µ , with ) ( p µ the pressure dependent viscosity and z(p) the correction to the ideal gas law, ( . = RT M .. . .. . ) ( p z p , where . = density, M = molecular wt, R = gas constant, T = temperature, p = pressure, and z(p) = correction for real gases). In the referenced paper, a steady-state solution is presented for flow through concentric cylinders, which can be adapted for the purpose here. The equation derived there for qsc , the radial mass flux (m3/sec) is given by q = w o sc w o sc r r Tp p m p m khT ln )] ( ) ( [ - p (Eq. 8) where k = permeability (~1–100 Darcys, LeCain 1997, p. 11f) h = length of drift presumed involved T = temperature (K) Tsc = reference temperature at standard conditions (300 K) Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 25 px = pressure at the: outer boundary (x = 0), drift boundary (x = w), standard conditions (x = sc) rx = radius at the: outer boundary (x = 0), drift wall (x = w) For this problem the outer radius is allowed to become infinite and m(po) becomes zero, so the equation for q reduces to q = w sc w sc r Tp p m khT ln ) ( p (Eq. 9) (Al-Hussainy et al. 1966). To utilize this work requires values for the pseudocritical temperature and pseudocritical pressure for the volcanic gases. Those values are developed from Table 4. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 26 Table 4. Characteristics of Exsolved Gases and Their Pseudocritical Properties Species Mole Fraction Critical Temperature K Pseudo- Critical Temp K Critical Pressure atm Pseudo- Critical Pressure atm Molecular Wt. Average Molecular Wt. H2O 73.16 647.14 473.45 217.7 159.269 18.01 13.18 H2 1.17 32.97 0. 3857 12.76 0.14929 2 0.0234 CO2 14.28 304.14 43.43 72.78 10.39 44.01 6.2846 CO 0.57 132.91 0.7576 34.53 0.19682 28.01 0.1597 SO2 9.45 430.8 40.71 77.8 7. 3521 64.06 6.0537 S2 0. 37 1314 4. 8618 204.28 0.7558 64.12 0.2372 HCl 0.49 324.7 1.5910 82 0.4018 36.46 0.17865 HF 0.06 461 0.2766 63.95 0.03837 20.01 0.0120 H2S 0.74 373.2 2.76168 88.23 0.652876 34.08 0.2522 Tpc = 568.22 Ppc = 179.21 NOTE: Critical temperatures and critical pressures, p. 6-50f, and Molecular weights, p. 4-37f, Handbook of Chemistry and Physics—Lide and Frederikse 1997; Mole fractions from CRWMS M&O 2000d) In order to utilize the tables provided by Al-Hussainy and co-workers (their Table 1) for m(p) in terms of measured properties of real gases, a pseudo reduced pressure and a pseudo reduced temperature are extracted from Table 4, above. For P = PprPpc, with P = 75 atm (7.5 MPa) at the drift and, Ppr = 0.38, Ppc = 179.2 atm, and T = TprTpc, with T = 1100 0C (1373 K, magma temperature (CRWMS M&O 2000d), Tpc = 568.22 K, Tpr = 2.42. For a viscosity of 0.47 cp (linear extrapolation of table values for H20: Lide and Frederikse 1997, p. F43f), a pressure at the drift wall of 1100 psi (7.5MPa), m(p) is found to be 7.97 x 103 (atm)2/cp. With these numbers, the values for q are presented in Table 5. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 27 Table 5. Volumetric Flow Rates for Two Different Wall Lengths and Two Different Permeabilities of the Drift Wall h (length of drift involved) Air Permeability k = 1 Darcy Air Permeability k = 10 Darcy 5. 5 m (1 container) ~2 m3/sec ~20 m3/sec 1 km (~1 drift length) ~3.5 ×102 m3/sec ~3.5 × 103 m3/sec The volumetric flow rate for the entire drift at k = 10 Darcy implies an unreasonably high velocity for the gas at the effective dike face (location of intersection of dike and drift). The unreasonably high velocity suggests that in this approximation, gas flow is over estimated and that the model should include cooling of gas and heating of the country rock through which the gas is leaking. It appears that the volume of gas arriving at a container is not directly a limiting factor in corrosion. (For a discussion, definition of, and use of pseudo properties see Pirson 1977, p. 341f). Alternative Interpretations–The Igneous Activity Issue Resolution Status Report (IRSR) (Reamer 1999), in Section 4.2.3.3.1, discusses flow conditions and the possibility that decompression of the magma produces a shock wave and a vesicular flow. It may be inferred from the discussion in the cited section that the drift is expected to pressurize. For this AMR, Section 6.3.4 discusses the constraints on gas flow down the drift for flow controlled by leakage through the surrounding rock. While the intent is to establish whether sufficient gas is available for container corrosion, as a result it appears that pressurization sufficient to prevent fragmentation is expected to be unlikely. 6.3.5 Magma Flow—Thermal Environment If the fragmentation depth is below the repository, then liquid magma can enter a drift either during the initial intersection of dike and drift, as an intermittent stage, or as a late stage, degassed flow. The first, initial intersection, is associated with the entire dike, and as the dike flow fragments below the repository, the unchilled portion can flow back to the dike from the drift because the dike pressure is reduced, leaving covered (or possibly partially covered) waste packages. The second, the intermittent stage, is related to formation of a vent and conduit, which reduces the population of exposed drifts, and can allow accumulation of fragments (ash) in the drift. The late stage is flow of degassed liquid magma into the drift. Magma flow is the presumption of one of the right hand branches of Figure 1, and is intended to support decisions about waste package failure and entrainment of contaminants. The magma flow is presumed to follow the decompression and gas flow into the drift, so the state of the drift and waste packages is that suggested in Figure 8. The first, and possibly second, waste packages should have experienced considerable damage to their heads (endcaps), with Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 28 eversion and cracking as strong possibilities, and may be collapsed and fractured along the sides; waste packages further down the drift are damaged to a lesser extent. Rapid flow of magma (presumed to be approximately at its liquidus) into the drift, as suggested by flow velocities in the dike, would fill the drift more rapidly than it, the magma, could chill, forming a plug in the void space around the waste packages as indicated in Figure 9 (p. 46). The plug space drawn in Figures 7, 8, and 9 is about 5.5 m (dia.) × 11m (to ~22 m) (length) and if the cross-section of the drift and dike represents the source, would be expected to fill in minutes. The case of a continuing flow of magma, as considered in Reamer (1999), requires an exit to the surface from the drift, a circumstance that is not considered here. An exit would be possible only through paths occurring as a result of reestablishment of dike propagation or opening of existing bounding faults. Further, flow of magma, down an open drift to its end, would be expected to be rapid compared to the chilling (solidification) time for the magma. There are certain peculiarities of the waste package, which affect the magma–container interaction. The waste package, as a large, cool (relative to the magma) mass with high thermal conductivity and heat capacity, will rapidly chill the magma making initial contact, and as is observed for some objects in lava flows, will tend to form an insulating crust about the waste package. This crust is unstable (not necessarily well-bonded to the waste package). Correspondingly, cracks in the waste package will admit limited volumes of magma that will tend to chill and plug. Large cracks and tears might allow the void spaces in a waste package to fill with magma that chills. The interaction of waste package and magma is too non-specific and variable to be amenable to a direct calculation without a more detailed description of the part of the interaction to be modeled. As an approximation, the magma fills a cylindrical chamber (the drift) and solidifies to form a plug. Because, waste can only be dissolved or mobilized while the magma is liquid, what is of interest is how long liquid magma could be in contact with a waste package and its contents. Magma, which is presumed to be approximately at the liquidus temperature, as it solidifies (chills), releases the latent heat of fusion (e.g., Jaeger 1964, p. 451f) of the order of 3.348-4.185 × 105 J/kg (80–100 cal/g). For the cylindrical plug in contact with the drift wall and ground support, the solidification surface gradually moves inward. This problem is a “Stefan” problem, a so-called moving boundary problem. There is substantial literature describing solidification of a spherical and a cylindrical molten mass (e.g., Soward 1980; Stewartson and Waechter 1976; Riley et al. 1974; Pedroso and Domoto 1973; Kreith and Romie 1955), which has provided series solutions for the temperature history and for the time of final solidification. The particular problem of a magma plug of finite length is not among the solutions mentioned or among the problems tried. The problem for magma flow solidifying in an infinitely long cylindrical drift is described in Soward (1980), and the general approach to such problems is given in Carslaw and Jaeger (1959, p. 295f). The problem of actual interest, a finite plug imbedded in a medium, at a constant temperature at one end and cooling at the other and cooling along a cylindrical surface, has not Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 29 been addressed. However, enough can be gleaned from the existing solutions to make preliminary assessments about the waste package environment. Several of the analyses (Kreith and Romie 1955; Riley et al. 1974) provide simple expressions for the time for solidification of a cylinder filled with chilling magma, in terms of the diameter of the drift, the drift wall temperature, and the physical properties of the magma. Riley et al. (1974, p. 1514) estimate a time to chill, tf, as tf = a2(ß + 1)/(4k) (Eq. 10) where ß = L/(C(Tf -Tw)) a = radius of the drift (~2.75m) k = thermal conductivity of solid—basalt at 21.76 J/(kg sec C) (0.0052 cal/(gm sec C) (Hodgman et al. 1955, p. 2251, Table Heat Conductivity) L = Latent heat of fusion (3.348-4.185 x 105 J/kg (80–100 cal/gm) )(Jaeger 1964) C = Specific heat (~0.3) (Jaeger 1964) Tf = Fusion temperature of magma (~1100 °C, selected as a representative value from the range reported in CRWMS M&O 2000d, also see Table 1) Tw = Drift wall temperature (~600 °C) With these numbers, the time for solidification to reach the center of the magma filled drift is approximately 70 days. Estimates using other reference calculations suggest that to the order of accuracy of the expansions that the time is 70< tf < 82 days. These estimates depend on the drift wall temperature, which is a function of the thermal properties of the surrounding rock; increasing the drift wall temperature increases the length of time to chill. As an upper estimate of temperature, part of the waste package has seen Tf for most of the time for solidification. The actual temperature distribution for the chilling magma cylinder is given by Soward in series form (1980, p. 143f). Alternative Interpretations–The Igneous Activity Issue Resolution Status Report, in Section 4.2.4.3.1 (Reamer 1999, p. 83f), develops a heat transfer model in order to establish a temperature history for waste packages being enveloped in magma. They assume that the waste packages are located centrally in a convecting magma cylinder, the temperature is only a function of radius, the magma represents an infinite heat source, and the drift wall is approximately at the magma temperature for 20 days. Magma displaced by the waste package and waste package thermal properties are accounted for by treating the magma intrusion as a cylinder the diameter of the drift. The calculation allows for the possibility of a chilled rind of Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 30 basalt forming on the drift wall and reducing the heat transfer coefficient. These assumptions lead to the conclusion that there is ample time to heat the waste packages to failure during a typical basaltic eruption. This AMR analyzes a magma plug, which has entered a drift from the intersecting dike, as an infinite cylinder of solidifying magma. The magma has filled the drift rapidly and there is no further flow along the drift axis (there is no exit for flow to continue). Details of the heat transfer, such as orientation and location of the waste package relative to the center of the drift, and the fact that a waste package is a large metal object with good thermal conductivity relative to the rock, are treated by replacing the waste package with magma. While these details are important to the actual temperature distribution in the waste package, they are relatively unimportant to the issue of waste package failure because the length of time the waste package is subject to heat soak is so long. The calculation is for the length of time it takes for the magma filled drift to solidify. This time represents the length of time that a typical waste package will be exposed to a temperature near the magma liquidus temperature (1100 0C). The time calculated for solidification for the magma around the embedded waste packages is long compared to the time estimated in the IRSR. The case of a continuing flow of magma, as considered in Reamer (1999) requires an exit to the surface from the drift, a circumstance that is not considered here. An exit would be possible only through paths occurring as a result of reestablishment of dike propagation or perhaps opening of existing bounding faults. Further, flow of magma, down an open drift to its end, would be expected to be rapid compared to the chilling (solidification) time for the magma. 7. CONCLUSIONS The topics (end members) examined in this AMR are: (1) waste package temperature due to flow of magma down a blind drift, (2) steady-state gas flow down drift to interact with containers, (3) physical interaction of pressure pulse from the dike to displace waste packages and drift contents, and (4) qualitatively, the interaction of the self-generated crack leading the dike with the stress-altered region around the drift. Based on these analyses the following conclusions are physically possible and not excluded by data relevant to the mountain or to a repository. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 31 1. The thermally altered stress state of the mountain may cause propagating dikes to deviate for the order of 2000 years (addresses the nature of the mechanics of the dike drift interaction). 2. Disruption of waste packages caused by flow from the dike extends down the drift from the dike edge to 3 or possibly 4 waste packages (addresses movement of the waste packages as they are pushed down the drift). 3. Magma flow down the drift is limited to a few waste package lengths (11 to 22 m) by plugging from the crumpled drip shield and displaced backfill (addresses disruption of the drip shield and backfill as they are pushed down the drift). 4. The temperatures inside a magma plug (and approximately in embedded waste packages) are given by Soward (1980) with a solidification time of 70 to 82 days (addresses temperature changes to the waste package from exposure to magma). 5. Gas flow down an idealized drift is about 3.5× 102—3.5 × 103 m3/sec, and suggests an isothermal model is inadequate and overestimates the flow rate (addresses the amount of gas available to degrade waste containers). Data and assumptions used in these analyses are derived from the published literature as indicated in Table 1. No corresponding “qualified” data (data acquired under the controls of current QA procedures) have been developed specifically for Yucca Mountain and vicinity. This document may be affected by technical product input information that requires confirmation. Any changes to the document or its conclusions that may occur as a result of completing the confirmation activities will be reflected in subsequent revisions. The status of the input information quality may be confirmed by review of the Document Input Reference System database. 7.1 ALTERNATIVE INTERPRETATIONS In order to provide the basis for the PMR addressing the IRSR criteria, this section develops observations about the relevance of the analyses done in this AMR to the NRC – IRSR criteria, by comparison to the intent of the NRC – IRSR criteria and the analyses presented in the IRSR. The summary observation is that a more complete level of detail than is included in the IRSR is required to address the NRC – IRSR criteria. The NRC has developed the following acceptance criteria related to the Igneous Activity Key Technical Issue and the subissues of consequences (Reamer 1999, pp. 76 and 82). The first acceptance criterion related to the subissue of consequences is (Reamer 1999), Section 4.2.3.1. Acceptance Criterion: “Estimate of the dose consequences of igneous activity on the proposed Yucca Mountain high-level radioactive waste repository will be acceptable provided that: The models adequately account for changes in magma ascent characteristics and magma/rock interactions brought about by repository construction.” Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 32 The IRSR analysis assumes that an idealized interception of the drift by the dike has occurred. The analysis is qualitatively concerned about propagation of shock waves and pyroclastic flow down the drift. It explores quantitatively pressurization of the drift by magma and the conditions for re-establishment of dike propagation by hydraulic fracturing. This criterion contains several implicit questions which are only partly addressed by the IRSR interpretation of the physical processes. These questions are addressed in somewhat more detail by the following analyses (developed in Section 6.3): 1. Stress state evolution around the repository 2. Interaction of a linear crack with a penny-shaped crack 3. Consequences of shock wave movement (pressure pulse) down a drift. These implicit questions concern how the stress state affects the assent of magma, both in direction and in the detailed interaction with the drifts, and the consequences of the intersection, including pressure pulse propagation down the drift. The questions are mixed, in the sense, that they actually involve both conceptual models introduced in Section 6.3, the simple idealized intersection model, and the stress-field dependent model. Analyses 1. and 2. involve the second conceptual model—a model in which the stress state of the mountain and its evolution are important. Analysis 3. involves the idealized, first model, in which the stress state of the repository itself is ignored. The results of this AMR differ from those of the IRSR in part because of alternative interpretations of how much physical detail is required. The second acceptance criterion related to the subissue of consequences is Reamer 1999, Section 4.2.4.1. Acceptance Criterion: “Estimate of the dose consequences of igneous activity on the proposed Yucca Mountain high-level radioactive waste repository will be acceptable provided that: The models account for the interactions of basaltic magma with engineered barriers and waste forms.” This IRSR analysis assumes that the waste packages are enveloped by a convecting magma flow which fills the drift. On this basis, the analysis tries to establish whether waste packages will see temperatures sufficiently high and persistent to compromise package integrity. This criterion contains several implicit questions which are addressed in part by the following analyses (developed in Section 6.3): 1. Interaction of a linear crack with a penny-shaped crack 2. Steady-state flow of a real gas down a drift 3. Chilling of a magma plug enveloped within waste packages. Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 33 These implicit questions relate to the details of dike propagation in a space with an already existing circular opening, the flow of corrosive gas down a drift (amount available), and the temperature history of waste packages that are enveloped by magma. As for the previous Criterion, the questions involve both conceptual models (Section 6.3), the simple idealized intersection model, and the stress-field dependent model. In this case, analysis 1. depends on the second, stress-dependent conceptual model and analyses 2. and 3. on the first, simple idealized, conceptual model. The results of these analyses differ from those of the IRSR in part because of an alternative interpretation of how much physical detail is required. 8. INPUTS AND REFERENCES 8.1 DOCUMENTS CITED Al-Hussainy, R.; Ramey, H.J., Jr.; and Crawford, P.B. 1966. "The Flow of Real Gases through Porous Media." Journal of Petroleum Technology, 18, 624-636. Dallas, Texas: Society of Petroleum Engineers. TIC: 237975. Anderson, E.M. 1963. The Dynamics of Faulting, and Dyke Formation with Applications to Britain. London, United Kingdom: Oliver and Boyd. TIC: 246728. Aoki, Y.; Segall, P.; Kato, T.; Cervilli, P.; and Shimada, S. 1999. "Imaging Magma Transport During the 1997 Seismic Swarm Off the Izu Peninsula, Japan." Science, 286, (5441), 927-930. Washington, D.C.: American Association for the Advancement of Science. TIC: 246714. Barr, G.E.; Dunn, E.; Dockery, H.; Barnard, R.; Valentine, G.; and Crowe, B. 1993. Scenarios Constructed for Basaltic Igneous Activity at Yucca Mountain and Vicinity. SAND91-1653. Albuquerque, New Mexico: Sandia National Laboratories. ACC: NNA.19930811.0013. Brady, B.H.G. and Brown, E.T. 1985. Rock Mechanics for Underground Mining. London, United Kingdom: George Allen and Unwin. TIC: 226226. Carslaw, H.S. and Jaeger, J.C. 1959. Conduction of Heat in Solids. 2nd Edition. Oxford, Great Britain: Oxford University Press. TIC: 206085. CRWMS M&O 1999a. Conduct of Performance Assessment. Activity Evaluation, September 30, 1999. Las Vegas, Nevada: CRWMS M&O. ACC: MOL.19991028.0092. CRWMS M&O 1999b. Enhanced Design Alternative (EDA) II Repository Estimated Waste Package Types and Quantities. Design Input Transmittal PA-WP-99184.T. Las Vegas, Nevada: CRWMS M&O. ACC: MOL.19990526.0188. CRWMS M&O 2000a. Coordinate Modeling of Dike Propagation Near Drifts Consequences for TSPA-SR/LA. TDP-WIS-MD-000027 REV 01. Las Vegas, Nevada: CRWMS M&O. ACC: Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 35 Hardy, M.P. and Bauer, S.J. 1992. "Invited Paper: Rock Mechanics Considerations in Designing a Nuclear Waste Repository in Hard Rock." Rock Mechanics, Proceedings of the 33rd U.S. Symposium, Santa Fe, New Mexico, 3-5 June, 1992. Tillerson, J.R. and Wawersik, W.R., eds. Pages 1041-1050. Rotterdam, The Netherlands: A.A. Balkema. TIC: 239168. Hodgman, C.D.; Weast, R.C.; and Selby, S.M. eds. 1955. Handbook of Chemistry and Physics, A Ready-Reference Book of Chemical and Physical Data. 37th Edition. Cleveland, Ohio: Chemical Rubber Publishing Co. TIC: 247067. Jaeger, J.C. 1964. "Thermal Effects of Intrusions." Reviews of Geophysics, 2, (3), 443-466. Washington, D.C.: American Geophysical Union. TIC: 246686. Jaeger, J.C. and Cook, N.G.W. 1979. Fundamentals of Rock Mechanics. 3rd Edition. New York, New York: Chapman and Hall. TIC: 218325. Kreith, F. and Romie, F.E. 1955. "A Study of the Thermal Diffusion Equation with Boundary Conditions Corresponding to Solidification or Melting of Materials Initially at the Fusion Temperature." Proceedings of the Physical Society, Section B, 68, 277-291. London, United Kingdom: The Physical Society. TIC: 246910. LeCain, G.D. 1997. Air-Injection Testing in Vertical Boreholes in Welded and Nonwelded Tuff, Yucca Mountain, Nevada. Water-Resources Investigations Report 96-4262. Denver, Colorado: U.S. Geological Survey. ACC: MOL.19980310.0148. Lide, D.R. and Frederikse, H.P.R., eds. 1997. CRC Handbook of Chemistry and Physics. 78th Edition. Boca Raton, Florida: CRC Press. TIC: 243741. Lister, J.R. 1990a. "Buoyancy-Driven Fluid Fracture: Similarity Solutions for the Horizontal and Vertical Propagation of Fluid-Filled Cracks." Journal of Fluid Mechanics, 217, 213-239. Cambridge, United Kingdom: Cambridge University Press. TIC: 225065. Lister, J.R. 1990b. "Buoyancy-Driven Fluid Fracture: The Effects of Material Toughness and of Low-Viscosity Precursors." Journal of Fluid Mechanics, 210, 263-280. Cambridge, United Kingdom: Cambridge University Press. TIC: 246674. Lister, J.R. and Kerr, R.C. 1991. "Fluid-Mechanical Models of Crack Propagation and Their Application to Magma Transport in Dykes." Journal of Geophysical Research, 96, (B6), 10,049- 10,077. Washington, D.C.: American Geophysical Union. TIC: 225066. Mack, M.G.; Brandshaug, T.; and Brady, B.H. 1989. Rock Mass Modification Around a Nuclear Waste Repository in Welded Tuff. NUREG/CR-5390. Washington, D.C.: U.S. Nuclear Regulatory Commission. TIC: 229166. Pedroso, R.I. and Domoto, G.A. 1973. "Inward Spherical Solidification - Solution by the Method of Strained Coordinates." International Journal of Heat and Mass Transfer, 16, 1037- Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 36 1043. New York, New York: Pergamon Press. TIC: 246685. Pirson, S.J. 1977. Oil Reservoir Engineering. Huntington, New York: Robert E. Krieger Publishing. TIC: 246776. Pollard, D.D. 1987. "Elementary Fracture Mechanics Applied to the Structural Interpretation of Dykes." Mafic Dyke Swarms, A Collection of Papers Based on the Proceedings of an International Conference held at Erindale College, University of Toronto, Ontario, Canada, June 4 to 7, 1985. Halls, H.C. and Fahrig, W.F., eds. Special Paper 34, 5-24. St. John's, Newfoundland, Canada: Geological Association of Canada. TIC: 247071. Reamer, C.W. 1999. "Issue Resolution Status Report (Key Technical Issue: Igneous Activity, Revision 2)." Letter from C.W. Reamer (NRC) to Dr. S. Brocoum (DOE), July 16, 1999, with enclosure. ACC: MOL.19990810.0639. Riley, D.S.; Smith, F.T.; and Poots, G. 1974. "The Inward Solidification of Spheres and Circular Cylinders." International Journal of Heat and Mass Transfer, 17, 1507-1515. New York, New York: Pergamon Press. TIC: 246684. Smith, C.; Vollendorf, W.C.; and Warren, W.E. 1981. In-Situ Stress from Hydraulic Fracture Measurements in G Tunnel, Nevada Test Site. SAND80-1138. Albuquerque, New Mexico: Sandia National Laboratories. TIC: 207978. Soward, A.M. 1980. "A Unified Approach to Stefan's Problem for Spheres and Cylinders." Proceedings of the Royal Society of London, A 373, 131-147. London, United Kingdom: The Royal Society of London. TIC: 246671. Spence, D.A. and Turcotte, D.L. 1985. "Magma-Driven Propagation of Cracks." Journal of Geophysical Research, 90, (B1), 575-580. Washington, D.C.: American Geophysical Union. TIC: 225148. Spence, D.A. and Turcotte, D.L. 1990. "Buoyancy-Driven Magma Fracture: A Mechanism for Ascent Through the Lithosphere and the Emplacement of Diamonds." Journal of Geophysical Research, 95, (B4), 5133-5139. Washington, D.C.: American Geophysical Union. TIC: 246860. Spence, D.A.; Sharp, P.W.; and Turcotte, D.L. 1987. "Buoyancy-Driven Crack Propagation: A Mechanism for Magma Migration." Journal of Fluid Mechanics, 174, 135-153. Cambridge, United Kingdom: Cambridge University Press. TIC: 225149. Stewartson, K. and Waechter, R.T. 1976. "On Stefan's Problem for Spheres." Proceedings of the Royal Society of London, A 348, (1655), 415-426. London, United Kingdom: Royal Society of London. TIC: 246670. Stock, J.M. and Healy, J.H. 1988. "Stress Field at Yucca Mountain, Nevada." Chapter 6 of Geologic and Hydrologic Investigations of a Potential Nuclear Waste Disposal Site at Yucca ANL-WIS-MD-000015 Rev 00 39 April 2000 Figure 1. Decision Tree for Perspective View of Modeling Release in ash plume? Conduit forms through drift? Container failure? Contaminant entrainment? Drift size and growth, conduit erosion Drift entry by dike flow? Yes Yes Yes Yes No No No No Fracture/ fracture interaction with stress-relieved zone? Dike propagates to surface? Yes Yes No No Dike terminates Conduit forms between drifts and outside repository Shockwave disrupts DS, backfill and containers? No "Knife" cut of dike across drift? Yes No Dike compromises cotainers? Yes No Yes Lateral entrainment to conduit between drifts? Yes No ?? Release in ash plume Yes No Release in local deposition Mechanical failure (shock, abrasion/erosion of container/fuel pellet) Gas corrosion Elastic collision Inelastic collision (sticky particles) Respirables Plume deposition Fracture/ fracture interaction with stress-relieved zone? Dike propagates to surface? Yes No Release in ash plume Conduit forms through drift? Container failure? Contaminant entrainment? Stress interaction controls location; Conduit surge and growth Drift entry by dike flow? Yes Yes Yes Yes No No No No Yes Conduit forms between drifts and outside repository Gas flows down drift Mechanical and corrosion failures Fragmentation occurs below repository? Fragmentation occurs above repository; Magma reaches drifts No Yes No Dike terminates Shockwave disrupts DS, backfill and containers? No "Knife" cut of dike across drift? Yes No Dike compromises containers? Yes No Yes Lateral entrainment to conduit between drifts? Yes No Release in ash plume ?? 3 2 4 No release Conduit forms Yes No No release Release in ash plume 4 Yes No adverse interaction Dike reaches repository Encapsulated waste/containers Thermal period intrusion Dike propagates to surface? Thermal period stress rotation Sill forms below repository? Magma flow exits to surface along fault zone? Containers in magma, fail? Sill forms at repository Intrusion only Waste dissolved and entrained? Flow system alteration with aqueous transport Distributed/ mixed encapsulated waste Flow system alteration with aqueous transport Encapsulated waste 1 No No No Yes No No 2 Yes Yes Yes Yes Intrusion occurs during post-thermal period? No Dike deviated by repository? Dike intercepts repository ~vertically? ?? No Yes Dike exits via fault zones? (SCF,DHW) Yes Yes No No Thermal and gas effects Fracture/ fracture interaction Yes No 2 3 Lava flow SCF = Solitario Canyon Fault DHW= Drill Hole Wash DS = Drip Shield ??= To be investigated Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 40 Figure 2. Thermo-Mechanical Response of the Stress State—Main Drift Figure 3. Thermo-Mechanical Response of the Stress State Vertical stress Horizontal stress Tuff Main Access Drift 10 9 8 7 6 5 4 3 2 10 100 1000 10000 Years After Start of Waste Emplacement Stress (MPs) Axial stress Vertical stress Horizontal stress Emplacement Drift 18 10 100 1000 10000 Years After Start of Waste Emplacement Stress (MPs) Axial stress 16 14 12 10 8 6 4 2 0 Figure 4. Generic Location of Transition from Horizontal to Vertical Least Principal Stress Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 41 N s3 s2 s1 s1 s2 s3 ~ 1 m (Dike not to scale) Repository Ghost Dance fault Solitario Canyon fault Bow Ridge fault Dike Regional least principal (NNW/SSE) stress Vertical stress = Least horizontal stress Region of vertical least principal stress Propagating crack Magma s1, s2, s3 Principal stresses s Least principal stresses Figure 5. Idealized Stress Trajectory for the Mountain. Principal Stresses: Vo = Initial Vertical, Ho = Larger Horizontal, and ho = Smaller Horizontal (Least Principal Stress), TS = Thermal Addition to Stress, TSm = Maximum Additional Stress H0 h0 H= H0 +aTS V= V0 +bTS h = h0 +TS TSm TS(Temperature) V0 Increasing stress Initial Stress Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 42 Figure 6. Design Cross-Section 0.606 Sketch not to scale Springline 0.5 1.339 1 2.75 1.495 0.396 0.835 26 deg 1.231 backfill invert NOTE: Sketch corresponding to In-Drift Data for Drift-Scale Models for TSPA-SR (Rev 01). Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 43 DTN: SN9908T087279.004 Figure 7. Pre-Intrusion Layout 5.5 m Backfill Drip shield Spacing between containers ~10 cm Pallet Invert Container Drip shield Backfill Skirt ~.25 m Dike Propagation Near Drifts ANL-WIS-MD-000015 Rev 00 April 2000 44 Sketch Not to Scale DTN: SN990870872799.004 CRWMS M&O 2000f ANL-WIS-MD-000015 Rev 00 45 April 2000 Figure 8. Post-Intrusion Disruption Jammed drip shields Displaced backfill Distorted drip shield Displaced invert Dike ANL-WIS-MD-000015 Rev 00 46 April 2000 Figure 9. Post-Dike Emplacement, Magma Plug Superimposed on Figure 8 Dike Magma plug Jammed drip shield Distorted drip shield Displaced backfill Displaced invert