ASHRAE LV-11-C034-2011 Quantifying Chemical Biological Event Severity with Vulnerability-Based Performance Metrics.pdf

《ASHRAE LV-11-C034-2011 Quantifying Chemical Biological Event Severity with Vulnerability-Based Performance Metrics.pdf》由会员分享，可在线阅读，更多相关《ASHRAE LV-11-C034-2011 Quantifying Chemical Biological Event Severity with Vulnerability-Based Performance Metrics.pdf（8页珍藏版）》请在麦多课文档分享上搜索。

1、Quantifying Chemical/Biological Event Severity with Vulnerability-Based Performance Metrics Jason W. DeGraw, PhD William P. Bahnfleth, PE, PhD Member ASHRAE Fellow ASHRAE ABSTRACT Quantifying the impact of chemical or biological releases on a building and its occupants is a necessary part of risk as

2、sessment. The most informative metrics for this purpose are absolute, “threat-based” measures of lost or preserved assets that are specific to the agent of interest. Practical application of threat-based metrics may be difficult because data such as the maximum plausible release quantity and dose re

3、sponse characteristics for an agent may be unavailable. In addition, it is left to the judgment of the analyst to select appropriate agents on which to base risk management decisions. Relative metrics are an alternative to absolute metrics that compare quantities such as exposure without respect to

4、the specific agent. Such metrics may be described as “vulnerability-based“ because their values reflect the influence of the building and its systems on exposure, independent of the agent. Vulnerability-based metrics are generally easier to determine but harder to interpret than threat-based metrics

5、. The ordered, area-weighted distribution of concentration raised to a power corresponding to the toxic load exponent of an agent was selected for investigation as the basis of a vulnerability-based metric. Metric values were computed from concentration time histories generated by multizone model si

6、mulations of indoor agent releases. It was found that metrics derived from this distribution can distinguish between the severity and extent of different releases. Although easier to apply to the results of multizone modeling, the proposed metric can, in principle, also be applied to experimental da

7、ta. INTRODUCTION Security, or an acceptable level of risk, is a matter of great concern for organizations of all sizes and types. It is important that any building system that could pose a threat to the safety of occupants have a well understood set of vulnerabilities. This is particularly true of h

8、eating, ventilation and air-conditioning (HVAC) systems, since the daily life of the average person in the United States is spent primarily within the confines of buildings served by such systems. In most cases, mechanical failure of an air handling system will be more an inconvenience than a danger

9、. The systemic “failure” that turns an air handling system into a delivery system for chemical and biological (CB) weapons is potentially much more serious because CB incidents are acute events in which a high level of exposure may occur before it is possible to initiate active security measures. Th

10、ere is much interest in the development of procedures to quantitatively evaluate the security of buildings before an attack takes place. The focus of this paper is primarily on computer modeling procedures for this purpose. Much of the published literature is aimed at providing general guidance on t

11、he mitigation of risk (Bahnfleth et al., 2006), and not at the development of quantitative measures of building-specific vulnerability. Bahnfleth (2004) provides an LV-11-C034282 ASHRAE Transactions2011. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org)

12、. Published in ASHRAE Transactions, Volume 117, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAES prior written permission.overview of available guidance documents and gives a picture of the changing

13、 situation (particularly before and after the 9/11 attacks). The reviewed security guidance (from a wide range of agencies) indicates that the security must be addressed in the design and operation of HVAC systems. Some guidance literature prescribes risk assessment as part of HVAC system design. Ty

14、pically this guidance is high level in nature, and often deals only tangentially with the CB threat. For example, Jones and Singh (2008) outline a design process that takes into account risk assessment (specifically tailored to the CB threat), but which is still qualitative in nature. Other literatu

15、re specifically addresses quantitative measures of building security evaluation, and some make use of computer-based tools for security evaluation of a building (or a building design). Reddy et al. (2011a and 2011b) review and compare tools that are available for this purpose. Many of the tools desc

16、ribed lack building-specific capabilities, and those that do have such capabilities are limited. Kowalski et al. (2003) use NISTs CONTAM multizone (MZ) modeling software (Walton and Dols, 2008) to compute the spread of a contaminant and then use a dose response model to determine the severity of an

17、attack. The model includes the possible effects of specific countermeasures (e.g. ultraviolet air disinfection). A related approach is Kowalskis Building Protection Factor metric (Kowalski, 2006). This approach overcomes many of the difficulties described in the Kowalski et al. (2003) study to compu

18、te a single number representing the degree to which a building is protected. Bahnfleth et al. (2006) conduct a threat-based assessment using MZ modeling in which the effects of varying model parameters (including release characteristics) are studied. The authors display their results in a useful gra

19、phical form in which the dose is plotted as a function of the fraction of occupants. This demonstrates that a metric need not be a single number (e.g. fraction of occupants infected), but can be a distribution over some parameter of interest population, area, or time, for example. Firrantello et al.

20、 (2007) use a similar MZ-based approach. The study focuses on aerosols and no agent-specific data is used. A dormitory is modeled, and the results are used to draw conclusions with respect to the process of vulnerability reduction. Bem (2008) investigate the role of cost, with an economic assessment

21、 that includes the cost of casualties. Persily et al. (2007) report on a NIST study of the effects of building retrofits upon contaminant transport using CONTAM. Both particulate and gaseous contaminants are considered and relative comparisons of contaminant impact are made. A relatively large numbe

22、r of parameters are varied, including air filtration, envelope tightening, and shelter-in-place emergency response. The cost of retrofits is considered, and retrofit guidance is offered. SECURITY METRICS The objective of the investigation is to identify a security metric, i.e., a quantitative measur

23、e of the severity of a CB event. In the literature, two approaches have been taken. One is to avoid the issue of identifying particular agents of concern and consider only relative effects such as the ratio of average building concentration to average concentration resulting from a reference event.

24、The other is to identify an agent, a threshold exposure, and a plausible release quantity and then to determine either the fraction of area or fraction of occupants exposed at or above the threshold. In both cases, there are multiple options to be tested different time intervals over which the metri

25、c is calculated, spatially or temporally averaged values, and others. These two types of metrics are referred to as “vulnerability based“ and “threat based“ respectively. Threat-based metrics are specific to an agent and are generally stated as an absolute number (e.g. number of casualties). When su

26、fficient information is available, it is relatively straightforward to compute any of a number of severity measures. However, a major drawback of this approach is the sheer volume of possibilities there are limitless possible event scenarios. A vulnerability-based metric is not specific to an agent,

27、 and generally results in a metric that measures severity relative to some baseline event. Nevertheless, relative metrics have been widely used due to the low requirements on the specifics of a release, and it is this approach that is adopted. A proper security metric must be a function of the chara

28、cteristics of a system. While the metric need not be continuous in the mathematical sense, it must be unambiguously computable from the characteristics of the system. Such a metric may be positive in the sense that the metric expresses freedom from risk or vulnerability to risk, respectively. For so

29、me metrics, particularly the relative type, complementary positive and negative metrics have been defined. For others, particularly absolute metrics, the availability of a positive metric does not guarantee the existence of a negative metric. The National Research Councils (2007) report on the prote

30、ction of building occupants identifies three classes of protection metrics: those 2011 ASHRAE 283that measure the fraction of occupants exposed, those that measure the fraction of a building exposed, and those that measure lives saved. All three of these metric classes admit a complementary metric (

31、e.g. lives lost), and only the lives saved class is necessarily an agent-specific, threat-based metric. Depending on the information available on occupancy, the data necessary to compute a fraction of building exposed metric may also be used to also calculate a fraction of occupants exposed metric.

32、BUILDING MODELS AND SCENARIOS The investigation employs a simplified version of the CONTAM model of a barracks used by Firrantello et al. (2007) in an earlier study. It has been simplified, in the sense that a number of the more complicated modeling elements are not used. Chief among these are ducts

33、, controls, and occupancy scheduling. All of these elements are still usable with the proposed methodology, but omitting them simplifies both model development and model analysis. The building is comprised of four levels with a total floor area of approximately 22,600 ft2(2,100 m2). The first three

34、levels are occupied and the fourth level is an unpartitioned storage area. Each occupied level has a central connecting corridor and is served by a single air handling system at a rate of 1.1 scfm/ft2(5.6 sL/(s m2) with 15% outdoor air. Two stairwells and one elevator shaft connect the levels. Each

35、of the occupied levels has two bathrooms, each with an exhaust fan. The first level of the building is shown in Figure 1. Figure 1 First level of the barracks model with leakage path, diffuser, return, and zone icons displayed. The primary interest here is in acute events of sudden onset. A number o

36、f options are available to simulate the onset of such events with CONTAM. One is to use a source to release the appropriate mass over a short period of time, and another is to instantaneously introduce the appropriate mass via an initial condition or a burst source. The former option has the defect

37、that it requires selection of a duration (or a rate of release). This adds another parameter that may affect the results, so the appropriate mass is introduced instantaneously. The implicit assumption in this choice is that the release is a sudden event (e.g. an explosive dispersion) that causes the

38、 contaminant concentration in the release zone to very quickly reach a well-mixed state. This assumption is not very compatible with large spaces. For the sake of implementation simplicity, the initial concentration for a release zone is computed from the mass and the zonal volume. Multiple release

39、events are simulated and compared to one another. The initial concentration is adjusted to give the same mass in each release zone in order to make releases in different zones directly comparable. Because the mass released is the same for every event, the initial building-average concentrations will

40、 be the same even when the initial zonal concentrations are different. Note that the actual numerical value of the mass is unimportant as long as the same value is used for each event. AN EXPOSURE DOSE METRIC As indicated above, the security assessment approach taken here is to develop a relative, v

41、ulnerability-based metric. A practical metric for risk should be computable from available data, unambiguous, and sufficiently accurate to be useful. Measures based on actual uptake of an agent by exposed individuals require many assumptions to be made about the exposed population and its activity l

42、evel that make their application difficult. Averaging of concentration, whether spatial or temporal, 284 ASHRAE Transactionsgives misleading results. A method based on cumulative, local exposure (“exposure dose”) may offer the best balance of practicality and accuracy. Exposure dose, D, is defined f

43、or the purposes of this discussion as nullnullnullnullnullnullnullnullnullnullnullnull, (1) where C is the time-varying contaminant concentration, T is the duration of an event starting at time = 0, and n is the so-called toxic load exponent, which reflects the dose response characteristics of a par

44、ticular agent. So defined, D is representative of the effects of exposure to a time-varying concentration of a harmful agent (ten Berge, et al., 1986). Use of similar expressions dates to the 1930s (e.g. Busvine, 1938) and these expressions have been applied in a variety of contexts (e.g. Bennett, 2

45、009). Typical values of n fall in a range from 1 to 3. When n = 1, effects of concentration and duration are interchangeable, which reasonably approximates the likelihood of infection from relatively short term exposure to a biological agent. Many chemicals, however, exhibit a non-linear relationshi

46、p between concentration and duration of exposure such that effects of increasing concentration are more severe than proportionate increases in duration and are better modeled by n = 2 or even higher. While it appears that the need to select an appropriate value of n makes this method inherently agen

47、t-specific, that is not really the case. Without specifying an agent, analysis could be performed for a range of toxic load exponent values to represent typical biological and chemical agents. To illustrate the application of exposure dose in a relative exposure metric, a value of n = 1 is used for

48、the remainder of the analysis. The methodology outlined here is independent of the choice of n. For a given event, the exposure dose is evaluated for each zone and paired with the floor area of the zone. Arranging these results in order of increasing exposure dose and plotting as a function of the c

49、umulative area (of zones with lesser exposure dose) gives a distribution such as that shown in Figure 2, which describes the consequences of a release in a single zone for inhabitable/occupied areas of a building. The vertical axis is a logarithmic scale so that the large exposure dose in the release zone does not hide the results elsewhere. Note that the release zone exposure dose is many times the exposure dose of the rest of the building in this case, 98% of the total floor area experiences an exposure dose an order of magnitude smaller than the release zone experiences.