Risk or Consequence? The Concept of Consequence Based Explosion Resistant Design Sirous Yasseri IMECHE Seminar on Engineering Structures Survival: Blast and Impact Protection University of Nottingham, 7th October 2010

Goals of a Design Risk management Hazard driven measures Technology driven measures Regulatory driven measures Risk management Economic limitations Demographic limitation Environmental limitation

Elements of Design Process Exposure (Explosion load with certain probability of being exceeded) Setting the context Vulnerability (Probability of a Damage State being reached or exceeded for a given Explosion In this framework three levels are distinguished, namely the exposure, the vulnerability and the robustness. The exposure can be considered an indicator of the hazard potential for a given object or system of consideration. Considering earthquakes the exposure EX is an inherently uncertain phenomenon with probabilistic characteristics usually provided in terms of earthquake intensities and corresponding return periods. The vulnerability P(D EX) can be considered an indicator of the immediate consequences (or damages to the system) associated with a given exposure event. Considering an explosion event the vulnerability is associated with significant uncertainty and is appropriately described by a probability distribution of different damage states of structures and lifelines conditional on the exposure event, e.g. the earthquake intensity. The robustness is an indicator of the indirect consequences due to the damages of the considered system. Considering again the event of an earthquake the robustness is associated with the conditional probability of losses of various degrees conditional on the exposure and a given damage state. Consequence (probability of financial loss, injury and fatality) Remediation to keep loss below tolerable level (Risk Reduction measure and their cost effectiveness)

Expression for Expected Consequences Decision variable risk of losses DV Damage measure casualties capital loss downtime DM Engineering demand parameter displacement drift etc EDP Intensity measure hazard curve level of explosion IM

Events leading to gas explosion Events leading to gas explosion. BLEVE – Boiling Liquid Expanding Vapor Explosions The process of a gas cloud being ignited with the result of a rapid increase in pressure is defined as a gas explosion. Before an explosion is possible, there are several events that must occur. These events are illustrated in Figure. As the diagram shows, it is of course not possible to have an explosion without release of gas. Secondly, the gas must be ignited and an ignition can either lead to fire or an explosion. In this paper, it is only the explosion which is treated and not fires, even though they are more common as Figure suggest. (Gex-Con, 2006)

Example of exceedance curve There are several ways of dealing with risk analysis of gas explosions on oil platforms. One of the most accepted methods is is known as probabilistic explosion risk assessment. The method uses CFD calculations to both simulate gas dispersion and pressure from the explosion. These simulations are treated using statistics and changes the parameters which influence the explosion. Some parameters are wind speed, wind direction, size of leakage and direction. The result is a so-called exceedance curve and an example is given in Figure. Figure indicates that the frequency, and thus the probability, of a large pressure due to an explosion decrease as the presure grows. This means that if an explosion would occur it is most likely that the pressure would be relatively small. The exceedance curve is used to estimate if the probability of a certain pressure can be accepted and if the pressure is not acceptable a redesign must be considered. For example, the structural members are designed to a pressure of 2[Barg] which means that this pressure or a higher pressure occurs once every 10,000 years because the frequency is 10-4. If this frequency can be accepted, then the design pressure is correct. It is of course possible to experience a higher pressure e.g. 5[Barg] but the frequency of such a pressure is about 10-6, cf. Figure.

Possible Threats to Civil Engineered Systems

Frequency matrix Rank Frequency Category Annual Frequency Range 6 Very Often >10 Day - Month 5 Often 1 to 10 Month -Year 4 Probable 0.1 to 1 1 to 10 years 3 Moderate 0.01 to 0.1 10 to 100 years 2 Rare 0.001 to 0.01 100 to 1000 years 1 Very rare 0.0001 to 0.00001 1000 to 10,000 years Extremely rare <0.00001 >10,000 years One example of a frequency matrix is shown in Figure 3. The matrix expresses how often any given undesired event occurs. The frequency is partly given by a qualitative scale going from “very frequently” to “extremely rare”, and partly by a quantitative scale, which roughly describes the frequency and repetition periods for an event as a numeric value. It is important that both the qualitative scale and the quantitative scale are defined in a way that attains a broad agreement around the concepts. It should be noted that the quantitative scale is logarithmic.

Communicate and Consult High-level framework Communicate and Consult Objectives Stakeholders Criteria Key issues Establish the Context What risks? How might they occur? Identify risks Controls Probability Consequence Rank Evaluate risks Identify options Assess options Set priorities Evaluate treatment options Develop and implement future improvement plan Treat risks Monitor and Review Source: Developed from the International Infrastructure Management Manual, 2006 and the AS/NZS 4360: 2004

Defence in Depth Robustness works as Russian Matryoshka dolls. Outside each layer of protection, there is another layer containing it, and monitoring and maintaining it… but is there a larger doll outside that doll, protecting it? If yes what is that layer

Philosophy of Seveso II Directive I N S P E C T I O N S Land-Use Planning Safe Management Technology Information to the Public Emergency Planning Demonstrate safety in the Safety Report

The Bow Tie Analysis

What do we mean by performance? Building What is the cost? P = RISK Fatality/Injury (safety) £ Money (damage) Downtime (loss of use)

Peak reflected overpressures Pr (in MPa) with different W-R combinations

Quantitative range of performance Force parameter Displacement parameter The damage evaluation procedures are performance-based; that is, they measure acceptability (and changes in acceptability) on the basis of the degree to which a structure achieves one or more performance levels for the hazard posed by one or more hypothetical future earthquakes. A performance level typically is defined by a particular damage state for the components of a building. Commonly-used performance levels, in order of decreasing amounts of damage, are Collapse Prevention, Life Safety, and Immediate Occupancy. Hazards associated with future hypothetical earthquakes commonly are defined in terms of ground shaking intensity with a certain likelihood of being exceeded over a defined time period, or in terms of a characteristic earthquake likely to occur on a given fault. The combination of a performance level and a hazard defines a Performance Objective. For example, a common Performance Objective for a building is that it maintain Life Safety for ground motion with a ten percent chance of exceedance in fifty years. $, % replacement 25% 50% 100% 25% 50% 100% 25% 50% 100% $, % replacement $, % replacement 0.0 0.0 0.0001 0.001 0.01 0.25 0.0001 0.001 0.01 0.25 Casualty rate Casualty rate 1 7 30 180 1 7 30 180 Downtime, days Downtime, days

The Connected World System of interdependent critical infrastructures All elements are linked by arrows indicating characteristic relations between them.

Initial & Lifecycle Costs The cost of a facility should be seen in terms of both the initial cost as well as the life-cycle cost. In some projects, initial cost of setting up the structure may seem low, but in the long-run, cost of maintaining and upgrading it may be high. To capture such financial burdens during the planning stage itself, life-cycle cost is relied upon to get a more meaningful reflection of the viability and sustainability of the facility. Determining the initial cost of constructing and protecting a facility involves a number of factors. But, one factor that seems to directly affect protection cost is the stand-off distance of the facility from the un-protected access point nearest to it. All protection strategies that are affected by the level of threat, say by the amount of charge used in the explosive (in terms of equivalent tons of TNT), are seen as variable costs, implying that they depend on stand-off distance. On the other hand, protection strategies or measures that do not depend on the level of threat, e.g., the hardware required at the security surveillance control room or the space required to house the facility, are seen as fixed costs. Since designers do not have control on the level of threat, they often attempt to reduce the initial cost of the structure by increasing the stand-off distance, because the peak blast pressure is inversely proportional to the square of the stand-off distance. But, increasing the stand-off distance has the downside of increased perimeter from the facility to be protected and increased land cost. Thus, one can identify an optimal stand-off distance from the partial initial cost, which is the sum of cost of protection (i.e., hardening cost) and cost of stand-off (i.e., land cost and perimeter protection cost). The fixed costs need to be added to this partial initial cost to obtain the initial cost (Figure 1.2). Contributors to initial cost of fatality (Source: FEMA 427)

Vulnerability DM

Vulnerability Functions - Definition Collapse Prevention Damage Control Serviceability Probability of Reaching a Limit Response Blast Overpressure Severity

Vulnerability Functions – Wider Applications Effect on other infrastructure Economic Consequence Societal Reaction Population of Buildings Horizontal Axis – Blast Overpressure Vertical Axis – Effect on a System or a Region

Performance based design Demand for specific hazard level Building Damage States Force parameter Immediate occupancy Life safety Collapse prevention The damage evaluation procedures are performance-based; that is, they measure acceptability (and changes in acceptability) on the basis of the degree to which a structure achieves one or more performance levels for the hazard posed by one or more hypothetical future earthquakes. A performance level typically is defined by a particular damage state for the components of a building. Commonly-used performance levels, in order of decreasing amounts of damage, are Collapse Prevention, Life Safety, and Immediate Occupancy. Hazards associated with future hypothetical earthquakes commonly are defined in terms of ground shaking intensity with a certain likelihood of being exceeded over a defined time period, or in terms of a characteristic earthquake likely to occur on a given fault. The combination of a performance level and a hazard defines a Performance Objective. For example, a common Performance Objective for a building is that it maintain Life Safety for ground motion with a ten percent chance of exceedance in fifty years. Displacement parameter Performance Levels

Qualitative range of performance Performance level Damage Downtime Collapse Prevention • Severe structural damage Incipient Collapse Probable falling hazards Possible restricted egress Probable total loss Demand for specific hazard level Life Safety • Probable structural damage No Collapse No falling hazards Adequate emergency egress Possible total loss Damage Control • Slight structural damage Life safety attainable Essential systems repairable Moderate overall damage 2 to 3 weeks Immediate Occupancy • Negligible structural damage Life safety maintained Essential systems operational Minor overall damage 24 hours

Expected losses for an event Inelastic analysis results slight moderate extensive complete slight moderate extensive complete Fragility relationships Probability Displacement Earthquake force Earthquake force on building on building Expected losses for event Casualties Repair/replacement costs Downtime Pushover curve Pushover curve Global displacement Global displacement of building of building

Traditional code-based design Building Hazard Assessment . structural model Are consequences acceptable? Yes: Design goals are satisfied No: Propose suitable remediation

Consequence based Blast Resistant Design DV Compute the loses. Consequence based Blast Resistant Design DM Characterize the damage. EDP Analyze the structure. IM Determine the hazard.

Vulnerability – Measures of Input and Response Random Variables Strength Parameters Geometric Parameters Correlation & Distribution Response Limit States Serviceability Damage Control Collapse Prevention Input Motion Captures Response Conservatively Includes a Measure of Variability Requires Scaling

Structural analysis Nonlinear model Multiple response history analyses EDP Nonlinear model Multiple response history analyses Sets of demands for 10 individual records Statistically generated sets of correlated demands for 200 realizations

Consequence function for repair costs DM

Vulnerability Minimization Existing Vulnerability Measures of Intervention

Consequence-Based Engineering Start 4. Decision Making Done confident that consequences will be acceptable yes Are Consequences Acceptable? Damage Analysis Rapid Assessment 1. System Definition define system of interest define hazard define characteristics Should Acceptable Consequences be Redefined? no 5. Refine Hazard Estimate further refine hazard for more precise loss assessment yes no Should Parameters be Refined? 6. Refine Inventory Estimate further refine inventory of built environment for more precise loss assessment 2. Rapid Estimate of Consequences quick assessment of likely consequences yes Are System Interventions Possible? no 7. Refine Fragility Relations further refine vulnerability of built environment with refined hazard and more precise response analyses 3. Define Acceptable Consequences define stakeholder needs no Allows mitigation action plans to be prescribed in an optimal way that minimizes risk across an entire system of interest. Can continually be improved with research using advanced technologies in risk assessment and visualization. Steps defined in four categories: rapid assessment decision making damage synthesis consequence minimization yes Consequence Minimization 10. Prescribe System Interventions to Minimize Consequences rehabilitate or demolish vulnerable structures, construct new structures, re-route network flows, re-manage land use, etc. 8. Re-Assess Social Impact assess social and economic consequences of event in terms of refined hazard and inventory estimates and fragility relations Inappropriate Solution consequences will NOT be acceptable 9. Re-Visualize Consequences examine effects of system alterations on reducing consequences

Take Home Lesson Not always possible to calculate risk and hence loads associated with a given risk. It is better to design for consequences that we know we can tolerate. A consequence based Design methodology was outlined for Situations where risk cannot be well defined Never Forget the connected world when determining consequences; or take credit when no Layer of Protection is provided For After Sale Care write to Sirous.Yasseri@gmail.com