Page 295 - Pipeline Risk Management Manual Ideas, Techniques, and Resources

13/272 Stations and Surface Facilities
Liquid stations often have several levels of leak monitoring Spill size
systems (e.g., relief device, tank overfill, tank bottom, seal pip-
ing, and sump float sensordalarms), operations systems (e.g., A spill or leak size in any scenario is a function of many factors
SCADA, flow-balancing algorithms), secondary containment such as the failure mechanism, facility design, product charac-
(e.g., seal leak piping, collection sumps, equipment pad drains, teristics, and surrounding environment. Smaller leak rates tend
tank berms, stormwater controls), and emergency response to occur due to corrosion (pinholes) or design (mechanical con-
actions. Therefore, small liquid station equipment-related leaks nections) failure modes. The most damaging leaks at station
are normally detected and corrective actions taken before they facilities may be small leaks persisting below detection levels
can progress into large leaks. If redundant safety systems fail, for long periods of time. Larger leak rates tend to occur under
larger incorrect operations-related spills are typically detected catastrophic failures such as external force (e.g., equipment
quickly and contained within station berms. In some cases, impact, ground movement) and avalanche crack failures. There
stormwater is gathered and sampled for hydrocarbon contami- may be little advantage in directly correlating a wide range of
nation prior to discharge. Note that the chronic component of possible leak sizes with specific failure modes in a risk assess-
theproduct hazardis often enhanced where a leaking liquid can ment. Up to the maximum station facility volume, almost any
accumulate under station facilities. size leak is possible in any facility component.
The potential leak volume and leak rate must both be consid-
Product hazard ered in modeling potential spill size. Certain station spill sizes
are volume dependent-more so than leak rate dependent.
As with a pipeline failure on the ROW, a station product release Spills from catastrophic vessel failures or failures of any iso-
can present several hazards. The fire hazard scenarios of con- lated station component, such as failure of an overfilled liquid
cern for all hydrocarbon product types at station facilities storage tank, reach a size dependent upon the volume ofproduct
include the following: contained in the vessel or component. Such spill events are not
appropriately measured by leak rates because the entire volume
Fireball-where a gaseous fluid is released from a high- ofa vessel can be release within seconds. Human error spills can
pressure vessel, usually engulfed in flames, and violently often involve immediate loss of limited volumes of product.
explodes, creating a large fireball with the generation of Leak rate is important since higher rates of release can cause
intense radiant heat. Also referred to as a boiling liquid more spread of hazardous product (more acute impacts),
expanding vapor explosion (BLEVE) episode. whereas lower rates are influenced by detectability (more
Liquid pool fire-where a pool of product (HVLs and chronic impacts). Leaked volume, as a function of leak rate,
liquids) forms, ignites, and creates a direct and radiant heat leak detection, reaction time, and facility capacity, adds to the
hazard. vulnerability of receptors due to normally wider spreading and
Vapor cloudfire/explosion-where a product (gases, lique- increases costs associated.
fied gases, and HVLs) vapor cloud encounters an ignition Two effective spill volumes therefore come into considera-
source and causes the entire cloud to combust as air and fuel tion. The first is the facility’s capacity-dependent leak volumes
are drawn together in a flash fire. This is not an expected fire and represents the catastrophic station spill scenario (V,J The
scenario for crude oil and most refined products that remain second is the leak-rate-dependent volume, which is based on
in a liquid state. the area under the curve of the “leak rate versus time to detect”
Flamejet-where an ignited stream of product (gases, liqui- curve (Fig 7.7). In this graph, “time to detect” includes identifi-
fied gases, HVLs, and liquids) leaving a pressurized vessel cation, recognition, reaction, and isolation times. As shown in
or pipe creates a long horizontal to vertical flame jet with Figure 7.7, depending on the equation of the curve, volume V,
associated radiant heat hazards and the possibility of a direct can quickly become the dominant consideration as product
impingement of flame on other nearby equipment. containment size increases, but volume V, becomes dominant
Contumination--can cause soil, groundwater, surface water, as smaller leaks continue for long periods. The shape of this
and environmental damages due to spilled product. curve is logically asymptotic to each axis since some leak rate
level is never detectable and because an instant release of large
As a measure of increased exposure due to increased quantities volumes approaches an infinite leak rate.
of flammable or unstable materials, an energy factor can be Because leak detection is equally valuable in smaller facility
included as part ofthe product hazard or the potential spill size. containment volumes as in larger, it is not practical to directly
This will distinguish between facilities that are storing volumes combine V, with V, for a station risk assessment. A simple
of higher energy products that could lead to more extensive combination will always point to higher-volume containment
damages. The heat of combustion, Hc (BTUllb) is a candidate as warranting more risk mitigation than smaller contain-
for measure of energy content. Another product characteristic ments-a premise that is not always correct. Some mathemati-
that can used to measure the energy content is the boiling point. cal relationship can be used to amplify the leak rate-dependent
The boiling point is a readily available property that correlates volume to provide the desired sensitivity and balance. The
reasonably well with specific heat ratios and hence burning amplification factor is used to inflate the influence of small
velocity. This allows relative consequence comparisons since leak detection since the smaller leaks tend to be more prevalent
burning velocity is related to fire size, duration, and radient and can also be very consequential. With this provision, the
heat levels (emissive power), for both pool fires and torches. model can more realistically represent the negative impact of
The energy factor can be multiplied by the Ibs of product such leaks, which far exceed the impacts predicted by a simple
contained to set up an energy-content adjustmet scale to modify proportion to leak rate. For example, a 1 gal/day leak detected
the LIE after 100 days is often far worse than a 100 gal/day leak rate