4. Direct Social Loss Estimation

4.1 Introduction

Another main objective of this project was to develop an MMI based methodology to estimate the number of injuries and casualties for seismic events VI – XII in Vancouver.  The steps involved with developing this methodology are: (1) Determine the spatial distribution of the commercial, industrial, and residential populations per census tract.  (2) Estimate the number of people per commercial, industrial, and residential building, based on the primary use and total area of the structure, for three distinct times of the day: 2 a.m., 2 p.m., and 5 p.m. (3) Gather population information to estimate the number of people in specialty buildings (educational, hotels, and hospitals) for the three times scenarios. (4) Multiply the number of people in each building by the adjusted casualty rates (adopted from HazUS but converted to the BC 31 prototypes) and adjusted DPM’s to get the estimated number of injuries and casualties per building.  (5) Model the spatial distribution of the results per block using a GIS. 

Injury Classification

Injury levels are classified into four levels of severity, which was developed by Durkin and Thiel (1991), Coburn (1992) and Cheu (1994) and was provided by HazUS.  Severity levels one and two are related to the non-life threatening injuries, whereas severity levels three and four represents life threatening injuries and casualties.  The following table provides a detailed description of the injury classification scale used in our analysis.

Table 4.1.1. Injury Classification Scheme

4.2 Population Distribution

The model provided by HazUS can be used to estimate the population at three separate times of the day. The three times identified in the model are 2 a.m., 2 p.m., and 5 p.m.  These time scenarios were selected because they are expected to have high residential population at 2 a.m., commercial and educational population at 2 p.m., and commuting population at 5 p.m. HazUS provided relationships for deriving these populations from census variables for the three time scenarios.

The HazUS methodology was developed for seismic risk analysis in California, using American census tract data.  This required us to classify the appropriate Statistics Canada 2001 census variables to relate to the American standards for census information used in this methodology.  In order to determine the number of people within each building at the three different time scenarios, first, some census tract variables needed to be collected: the daytime residential population, nighttime residential population, number of people commuting, number of people employed in the commercial sector,  and the number of people employed in the industrial sector (HazUS).  Nighttime residential population was substituted with the 2001 total population (Statistics Canada). Daytime residential population was inferred from five census variables, Table 4.2.2. shows the census variables used for all classes. 

Table 4.2.2. Classifying Canadian Census Variables into the US Classification Standard

In order to estimate the maximum occupancy potential for specialty buildings, we determined an occupancy rate per square foot which was multiplied by the total area of each building.   Statistics on student enrollment in grade schools, colleges, and universities came from online sources.  Using these values as benchmarks, we calculated an average rate per square foot of 0.1 for educational buildings.  The occupancy rate for hotels was based on the residential occupancy rate provided by Kate Thibert, of 0.1, which we then reduced to 0.025 (0.01 for extremely massive luxury hotels) to more accurately reflect the number of rooms within each hotel.  Kate Thibert also provided a Hospital occupancy rate of 0.1 for the UBC Hospital.  We reduced this rate to 0.05 because the Hospital buildings in our study area have particularly massive total areas, causing a value of 0.1 to produce a gross over estimate of the actual population within each Hospital building.

The calculated population values were used as a maximum occupancy value for each building, which were then adjusted to reflect the number of people expected for the three time scenarios.

Once the data was established, indoor and outdoor population distribution of the three occupancy classes (residential, commercial, and industrial), was calculated for the three time scenarios at the census tract level.  Population distribution for specialty buildings, however, was done completely on an individual building scale.

The next step was to obtain a census tract unique identifier (CTUID) for each building in the inventory.  First, we had to classify all the blocks within our study area, based on the appropriate census tract.  This required us to overlay a census tract boundary polygon on top of our block shape file, and assign the appropriate CTUID to the blocks.  Since the blocks shape file didn’t fit neatly within the census tract boundaries, 71 blocks had to be eliminated (Figure 4.2.1.).  We could then relate census tract level populations to the individual buildings in our building inventory using the CTUID.  Four of the census tracts in the south western portion of our study area had to be reduced using the half and fifth criterion* because only a fraction of these census tracts contained blocks.

Figure 4.2.1. Assignment of CTUID to the Blocks Shape file

* The half and fifth criterion were based on the percentage of spatial area, occupied by blocks within the census tract. The population values were therefore divided by two or five in order to reduce the inflated values.

Census tract population values were required to calculate the number of people within each individual building for the three census based occupancy classes: residential, commercial, and industrial population.  In order to relate each building to the occupancy class populations, information on the primary use of each building was used.  The primary use descriptions in our building inventory were classified into the occupancy categories.  Each building, including the specialty buildings, was classified into one of the six occupancy categories through primary use, shown in the following table.

Table 4.2.5. Classifying Primary Use into Occupancy Categories

Once the buildings were classified into the six occupancy categories, we were able to determine the prevalent occupancy type which was weighted by the total area of all buildings within a block.  The two most dominant occupancy types appear to be residential with 62% and commercial with 35% of all blocks within the study area.  Residential neighborhoods are visible in areas of the West End, Kitsilano, Shaughnessy, Mount Pleasant and parts of Chinatown and Yaletown.  Commercial blocks clearly border major roads such as Broadway, Kingsway, Main and Cambie Street.  High concentrations of commercial buildings are also located in the Vancouver core and the downtown eastside.  In general these commercial blocks are associated with concrete as the prevalent material.  Conversely, the residential blocks appear to contain wood as the prevalent material.  Twenty blocks contain educational buildings as the dominant occupancy type whereas ten blocks show locations of hotels and ten blocks show industrial areas.  The Vancouver General Hospital was the only block classified as hospital occupancy type.  The block containing St. Paul’s Hospital downtown, was classified as a commercial block because the total area of all commercial buildings outweighed the total area of the hospital (Figure 4.2.2).

Figure 4.2.2. Occupancy Type Per Block

Once the buildings were classified appropriately, we could use the population distribution relationships to estimate occupancy type populations for three time scenarios.  Dividing the census population into each individual building was weighted by the total area of the structure.  Each building’s total area was then used as a fraction of the total area of all buildings within a census tract (of a particular occupancy type).  This method was used to produce an even distribution of the population based on the size of the structures.  For example, an industrial building consisting of 10% total area of all industrial buildings in that census tract would therefore also have 10% of the total industrial population in that census tract (Figure 4.2.3.).

Figure 4.2.3. Total Population at the Three Time Scenarios

4.3 MMI Based Casualty Probabilities

Structural DPM’s are used in this methodology to determine the total number of injuries and casualties for MMI events VI - XII. The HazUS casualty model uses five structural damage states; slight, moderate, extensive, complete without collapse, and complete with collapse. For each building prototype the probability of a building being in each of the damage states is calculated given the intensity of an event.

The HazUS casualty rates were created for California (ATC-13) building types. The first step was to convert these casualty rates to BC’s 31 prototypes. This conversion allowed us to create tables providing casualty rates for the five discrete damage states (slight, moderate, extensive, complete without collapse and complete collapse), associated with each individual building prototype.

Damage probability matrices that were created for the BC 31 prototypes have seven discrete damage states. The DPM’s had to be adjusted to fit the five damage states used in the HazUS methodology. The first three damage states were combined into the slight damage state.

The adjusted MMI based DPM’s are multiplied by the appropriate indoor and outdoor casualty rates for each damage state, to get the probability of indoor and outdoor casualties. The sum of the casualty probabilities is totaled for the four severity levels. This method was used to calculate the casualty probabilities for MMI events VI – XII.

Casualty probabilities were calculated for each of the 31 prototypes. These casualty probabilities are linked to the building inventory through prototype. These probabilities are multiplied by the number of people in each building at a specific time scenario to get the number of indoor, outdoor, and total injuries and casualties per building for MMI events VI-XII.

4.4 Results

After calculating the results, distinct patterns emerged across MMI events VI - XII. At 2 a.m. the amount of injuries and casualties seem to aggregate around residential blocks. The 2 p.m. maps tend to have elevated values in blocks containing educational buildings. Commercial areas also appear to have moderate values due to the large working populations within the study area. At 5 p.m. values appear to be correlated to commercial and industrial areas. However, blocks containing colleges and universities also have slightly elevated values at this time. Certain neighborhoods, specialty buildings and major roads in the study area show higher values (Figure 4.4.1.). We will examine these as case studies to understand the cause of these results.

Figure 4.4.1. Case Studies Map

Case Study: Main Street South of False Creek

In particular, Main St. stands out with high numbers of injuries and casualties occurring on almost every block.  29% of the buildings on this street are made from masonry material; however, 76% of these buildings were unreinforced prototypes which are considered the worst in terms of casualty estimation.  “The [URMLR] prototype includes buildings up to three stories in height with perimeter load bearing walls of clay brick, concrete block and hollow clay.  This was a very common form of low rise construction for commercial, institutional and industrial buildings until 1973, when the National Building Code of Canada required that all masonry be reinforced.  These buildings are easy to identify, usually showing extensive areas of red clay brick.  They behave very poorly and are considered the most hazardous form of construction in seismic areas” (Blanquera).   

It appears that commercial buildings occupy 44% of all buildings along the street.  These commercial buildings increase the number of people in this area significantly for the 2 p.m. time scenario with an estimated value of 6,881 people.  Out of those commercial buildings, 48% were unreinforced masonry prototypes.  Some of the highest injuries and casualties occurred in two educational buildings along Main St.  These buildings were also unreinforced masonry, and they are expected to contain the largest number of people (240 and 176) in an unreinforced building at 2 p.m.  In an MMI X or greater scenario, a significant number of people are expected to be instantaneously killed, mortally injured or sustain injuries that pose an immediate life threatening condition.  Other individuals may experience injuries requiring a great degree of medical care and use of medical technology.  Slight injuries require basic medical care, such as a sprain, a severe cut requiring stitches, a minor burn (first degree or second degree), or a bump on the head without loss of consciousness (HazUS, 2003).  For an MMI IX event occurring at 2 p.m., 5 casualties, 42 serious and 143 slight injuries are expected (Figure 4.4.2.).  If an MMI X event were to occur at 2 p.m., estimates reach 20 casualties, 5 life threatening injuries, 99 serious injuries and 318 slight injuries.  Conversely, at 5 p.m. the estimates are significantly lower with 6 expected casualties, 56 serious injuries and 238 slight injuries.

Figure 4.4.2. MMI IX and X on Main Street at 2 p.m.

Case Study: Yaletown

The boundaries of Yaletown are defined along Pacific Blvd., Homer St. and Smithe St.  Yaletown consists of 12 city blocks containing a total of 54 buildings in our database.  48% of them are constructed from old unreinforced masonry built prior to 1970.  For an MMI X event occurring at 2 a.m. there are 5 casualties, 2 life threatening injuries, 33 serious injuries and 103 slight injuries expected in Yaletown.  3 of these casualties and the 2 life threatening injuries occur in two large residential apartment buildings of the Un-reinforced Masonry Bearing Walls Medium Rise (URMMR) prototype.  One of the buildings was called Yaletown House, a 4 storey complex care facility for seniors, and is expected to experience 2 of these casualties and 1 of the life threatening injuries.  At 2 p.m. casualties will be occurring in commercial buildings because 55% of theses commercial buildings are unreinforced masonry prototypes.  Most of them are restaurants, bars, and offices containing high amounts of commercial population.  For an MMI X event occurring at 2 p.m. there is an expected 16 losses of life, 3 life threatening injuries, 58 serious injuries and 194 slight injuries.  An equivalent earthquake occurring at 5 p.m. has slightly lower figures with 12 casualties and 2 life threatening injuries, 59 serious injuries and 176 slight injuries.

Figure 4.4.3. MMI X in Yaletown at 2 a.m. and 2 p.m.

Case Study: Gastown

The area of Gastown consists of the city blocks located north of Pender St. and west of Main St.  to include a total of 15 blocks, containing 241 individual buildings.  This area appeared to show higher numbers of casualties and life threatening injuries for the 2 a.m. scenario.  The reason for this was because these blocks contained many large residential buildings and hotels that were constructed of un-reinforced masonry.  No casualties or life threatening injuries are expected for an MMI VIII event, however, for an MMI IX event occurring at 2 a.m., there is a total of 6 casualties two of which occur in hotels and 4 in residential buildings (Figure 4.4.4.).  These buildings also are expected to contain 13 serious injuries and 35 slight injuries for this intensity.  For an MMI X event, 17 buildings contain 18 casualties, 6 life threatening injuries, 47 serious injuries and 152 slight injuries at 2 a.m., 10 of those buildings are hotels and the remaining 7 are residential.  The hotel containing the highest number of casualties was the Balmoral Hotel, a 9 storey un-reinforced masonry building built in 1910.  This hotel was expected to contain 153 people at 2 a.m. resulting in a total of 1 casualty, 2 serious injuries and 6 slight injuries at MMI IX.  For an MMI X event occurring at 2 p.m. there is a total of 2 casualties in all of Gastown.  At 5 p.m. the results are slightly higher with a total of 2 casualties and 1 life threatening injury.

Figure 4.4.4. MMI IX and X in Gastown at 2 a.m.

Case Study: Vancouver’s Central Business District

The boundaries of Vancouver’s Central Business District (CBD) were defined along Thurlow St., Alberni St., Cordova St., Beatty St. and Beach Ave., excluding the blocks included in the Yaletown case study.  This area contains 119 blocks with 782 buildings, 40% of which are un-reinforced masonry prototypes.  39% of the buildings in the CBD are made out of concrete materials, 13% are built out of wood materials, and 5% are constructed from steel.  The total population estimated for 2 a.m. is 27,646 people within the CBD, 62% of these people will be in residential buildings, and 35% will be in hotel buildings.  At 2 p.m. this area is estimated to have a total population of 32,843 people, 78% of which will be in commercial buildings.  In comparison to 5 p.m. the estimated total population is 34,447 people, 63% of which are in commercial buildings, 25% in residential buildings, and 9% in hotels.

Across the CBD for an MMI VIII event occurring at 2 a.m., only a single building is expected to sustain 1 casualty.  This building is called Hotel California, a 7 storey un-reinforced masonry hotel located at 1176 Granville St.  This hotel is also expected to sustain 2 serious injuries and 6 slight injuries.  At MMI IX, 1 casualty, 1 life threatening injury, 4 serious injuries and 11 slight injuries are expected.  An MMI X event results in 2 casualties, 1 life threatening injury, 6 serious injuries and 21 slight injuries.  Some of the large concrete buildings such as the Hyatt Regency Hotel and Hotel Vancouver show high values of slight and serious injuries but has consistently low values of life threatening injuries and casualties.  For example, the Hyatt Regency Hotel is the largest buildings in the CBD with 34 stories and a total area of 95,200 square feet and is the Concrete Frame with Concrete Walls High Rise (CFHR) prototype.  This building will experience 1 serious injury and 2 slight injuries in an MMI VIII scenario at 2 a.m.  These numbers rise to 2 serious injuries and 5 slight injuries for MMI IX events, and 3 serious injuries with 9 slight injuries for MMI X events occurring at the same time.  Across MMI VIII – X this building is not expected to have any casualties or life threatening injuries due to the prototype.

Within the CBD, commercial buildings make up 78% of all buildings, causing large numbers of injuries and casualties for the 2 p.m. and 5 p.m. time scenarios.  One commercial building of interest was the Vancouver Art Gallery, a 4 storey un-reinforced masonry building built in 1911, with a total area of 20,000 square feet.  This building was expected to contain 85 people at 2 p.m. and 73 at 5 p.m.  Out of those 85 people, 1 serious and 2 slight injuries are expected for MMI level VIII, 1 serious and 3 slight injuries at MMI IX and 1 casualty, 2 serious and 6 slight injuries for an MMI X event.  At 5 p.m. the number of injuries is consistent with the 2 p.m. values because the casualty probabilities are so small that a difference of 12 people is not significant.

In the entire CBD there is a total of 1 casualty, 32 serious injuries and 97 slight injuries at 2 a.m., 3 casualties, 1 life threatening injury, 25 serious injuries and 70 slight injuries at 2 p.m., and 2 casualties, 1 life threatening injury, 20 serious injuries and 73 slight injuries at 5 p.m., for an MMI VIII event (Table 4.4.1. and Figure 4.4.5.).  This table reveals that the 2 a.m. scenario produces higher numbers of non-life threatening injuries, whereas the 2 p.m. map shows the higher values for life threatening injuries and casualties.  The reason for this pattern is because the majority of the population at 2 p.m. is in commercial buildings, many of which are un-reinforced masonry prototypes whereas the majority of the population at 2 a.m. is in residential buildings and hotels mainly constructed out of concrete and wood materials.  Concrete and wood prototypes produce high values of non-life threatening injuries, and result in very minimal life threatening injuries and casualties.  The 5 p.m. results show a combined effect of commercial, residential and hotel populations causing life threatening injuries and casualties to appear higher than the 2 a.m. scenario but lower than 2 p.m., even though the CBD contains its maximum number of people at this time.

Table 4.4.1. Total Injuries and Casualties within the CBD for MMI VIII - X

Figure 4.4.5. MMI IX in Vancouver’s Central Business District at 2 a.m. and 2 p.m.

Case Study: Hospitals

The Vancouver General Hospital block contains the highest values in the study area across all earthquake scenarios.  This is mainly due to the effect of the Heather Pavillion, a massive un-reinforced masonry building containing between 1,960 - 2,520 people depending on the time of day.  For an MMI VIII event occurring at 2 p.m. this building will contain the maximum number of people, resulting in 5 casualties, 3 life threatening injuries, 20 serious injuries and 50 slight injuries.  An MMI IX event might result in 10 casualties, 5 life threatening injuries, 33 serious injuries and 96 slight injuries.  Finally, an MMI X event would produce 18 casualties, 10 life threatening injuries, 58 serious injuries and 192 slight injuries.  This increase in MMI appears to scale the results by a factor of approximately 2.  Within this block there are 7 additional hospital buildings which are constructed out of concrete materials.  Because these buildings are concrete, they are not expected to experience any casualties or life threatening injuries for MMI events X and below; however, they are expected to have some serious injuries.  In particular, the Laurel Pavillion is estimated to contain 3,780 people at 2 p.m., and is expected to have the second highest number of serious injuries, with 4 for MMI VIII, 8 for MMI IX and 11 for MMI X. 

St. Paul’s Hospital is a good example of how un-reinforced masonry compares to concrete.  St. Paul’s Hospital is made up of two buildings in our database, existing on adjacent blocks along Burrard St. downtown.  The larger 9 storey building has a total area of 7,200 square feet and is the CFHR prototype.  This building is estimated to contain 324 people at 2 p.m.  The other building has 7 storeys and a total area of 5,250 square feet and is the URMMR prototype.  This building has an estimated 236 people at 2 p.m.  For an MMI VIII event occurring at 2 p.m. this building will have no casualties or life threatening injuries; however it is expected to sustain 2 serious injuries and 5 slight injuries, whereas the concrete building only expects 1 slight injury.  As the MMI level increases the effect of the building prototypes is made evident in the slight and serious injury values.  An MMI IX event occurring at 2 p.m. is expected to experience a single casualty with no life threatening injuries for the un-reinforced masonry building.  This building is also expected to have 3 serious injuries and 9 slight injuries.  At this intensity the concrete building only expects 1 serious injury and 2 slight injuries.  Lastly, for an MMI X event occurring at 2 p.m. the concrete building is expected to have 1 serious injury and 3 slight injuries, whereas the un-reinforced masonry building is expected to sustain 2 casualties, 1 life threatening injury, 5 serious injuries and 18 slight injuries.

Case Study: Educational Buildings

In our database there are 55 educational buildings, 45% of them are constructed out of un-reinforced masonry material.  This poses a great risk because educational buildings are generally large and contain many students at 2 p.m.  For example, the Kitsilano Secondary School is URMMR prototype and has a total area of 12,900 square feet, and contains an estimated 1,032 people at 2 p.m.  Starting at MMI VIII, this school already expects 2 casualties, 1 life threatening injury, 8 serious injuries and 21 slight injuries.  At MMI IX, this school expects 4 casualties, 2 life threatening injuries, 13 serious injuries and 39 slight injuries.  Lastly, at MMI X, this school expects 7 casualties, 4 life threatening injuries, 24 serious injuries and 78 slight injuries.  Besides the Heather Pavillion at VGH, this school has the highest expected values at 2 p.m.

Another educational building that displayed some interesting results was the Native Education Centre located at 285 E. 5th Avenue.  This building is 2 stories tall with a total area of 2,000 square feet and is the Wood Post and Beam (WPB) prototype.  At 2 p.m. this building is expected to contain 160 people.  Due to the low population of this building, for any injuries to occur, a minimum MMI level of IX must be reached.  At this intensity there is 1 serious injury and 3 slight injuries expected.  However, at MMI X, 1 casualty is expected, along with 2 serious injuries and 4 slight injuries.  The WPB building prototype is expected to perform poorly, as it is the worst prototype aside from the un-reinforced masonry prototypes.  Many of the highest values of injuries and casualties occurred in educational buildings around the study area.  The following table lists some of the highest values determined on an individual school building basis (Table 4.4.2.).

Table 4.4.2. Educational Buildings With The Highest Number of Injuries and Casualties