HURRICANE ANDREW STORM SURVEY

by Tim Marshall

September 13, 1992

Engineers and meteorologists at Haag Engineering Company have conducted a damage survey covering southern Florida and southern Louisiana in the wake of Hurricane Andrew. Our purpose was two fold: 1) to document the wind velocities and water levels in the storm's path, and 2) to evaluate the performance of various constructed facilities.

Hurricane Andrew struck southern Florida on August 24, 1992 and entered southern Louisiana on August 26, 1992. Our storm survey team was dispatched to the affected areas on September 5 through 9, 1992. Both aerial and ground damage surveys were conducted in southern Florida and meteorological information was obtained from the National Hurricane Center. The last two days of the survey were spent in southern Louisiana. Weather information for the state was obtained from the National Weather Service in Slidell, Louisiana.

A. HURRICANE ANDREW CHRONOLOGY

Hurricane Andrew could be traced back to an easterly wave that developed near Lake Chad in North Africa during the second week in August, 1992. A tropical depression formed along the wave when it was about halfway between Africa and the easternmost islands of the Carribean (11 degrees north, 38 degrees west) on August 17, 1992. Thunderstorms organized around the center of a low pressure system and the storm strengthened to tropical storm status later that day. It was at that time the name "Andrew" was assigned to the storm.

Tropical Storm Andrew moved rapidly along a west-northwest course but fluctuated in strength over the next four days as it encountered bands of upper level wind shear. The storm exited this area and began gathering strength on August 21, 1992. At 5am, on August 22, 1992, the storm reached hurricane status; the winds increased rapidly as the central pressure dropped.

Hurricane Andrew moved steadily west along 25.4 degrees north latitude during the next day. The lowest barometric pressure (27.23 in. or 922 mb.) was recorded at 12:48pm on August 23, 1992, when the center of the storm was positioned east of the Bahamas. The hurricane had reached the upper end of category 4 status on the Saffir-Simpson scale. Maximum flight level winds (10,000 ft.) were measured as high as 195 miles per hour or 170 knots.

Around 6pm on August 23, 1992, the eye of Hurricane Andrew passed the northern end of Eleuthra Island in the Bahamas. A surface reporting station recorded a maximum wind gust of 120 mph as the eye wall passed. The height of this wind instrument above the ground was unknown. Reconnaissance reports indicated the hurricane had a "double eye" structure for a few hours. Hurricane Andrew weakened slightly as it passed through the Bahamas and the central pressure rose to 941mb.

On the morning of August 24, 1992, Hurricane Andrew struck south Florida. The eye passed over Elliot Key located on the western end of Biscayne Bay. Fowey Rocks bouy, located east of Elliot Key, reported northerly winds sustained at 141 mph (123 knots) with a peak gust of 169 mph (147 knots) at 4:00 am as the eye wall passed. Sea level pressure was 967 mb. Data transmission ceased after that time. The height of the wind instrument was 143 feet (43 meters) above sea level.

Around 4:30am, the eye of Hurricane Andrew was centered over Biscayne Bay. High storm surges occurred from near Turkey Point to as far north as Miami. Seawater inundated numerous homes along the coast. Key Biscayne was submerged during the storm. The highest storm surge was over 16 feet NGVD in Derring and Saga Bays. Many boats moored at Black Point and Coconut Grove marinas were damaged or destroyed.

Landfall of the eye occurred around 5:00am just east of Homestead Air Force Base. The eye diameter was approximately 15 miles across and extended from the Cutler Ridge area to Florida City. The National Hurricane Center, located in Coral Gables, Florida, was on the northern edge of the eye wall. They reported a maximum sustained wind of 138 mph (120 knots) with a peak gust of 164 mph (143 knots) at 4:50am before the wind instrument was destroyed. The height of the wind equipment was approximately 200 feet above ground level.

Miami International Airport reported a maximum sustained wind of 86 mph (75 knots) with a peak gust of 115mph (100 knots) from the east around 5:50 am. The height of the wind equipment was approximately 33 feet (10 meters) above the ground.

It took about three hours for Hurricane Andrew to traverse southern Florida. Towns of Homestead, Naranja, Leisure City, Goulds, Princeton, Cutler Ridge, and Florida City sustained heavy damage to buildings. Moderate building damage occurred in the communities of Perrine, Howard, and Kendall. Minor damage to buildlings occurred north of 104th Street.

By 8:00 am, the eye of Andrew was located over Big Lostman's Bay on the west coast of Florida in Everglades National Park. The storm had weakened to category 3 status, and the barometric pressure had risen to 29.91 in./951mb.

When Hurricane Andrew entered the Gulf of Mexico, it re-intensified to category 4 status. However, the storm never recovered its pre-Florida landfall intensity. The lowest barometric pressure recorded while the storm was over the Gulf of Mexico was 27.52/936mb at 4:00pm on August 25, 1992. Approximately two hours later, Hurricane Andrew slowed and started to curve northwestward towards the south-central Louisiana coast. Central barometric pressures continued to rise, and Hurricane Andrew gradually lost strength. The storm was downgraded to category 3 status prior to landfall on the Louisiana coast.

As Hurricane Andrew approached Louisiana, an isolated storm on one of Andrew's raindbands spawned a tornado that traveled west-northwestward through Laplace, Louisiana. The tornado damage path was 9 miles long and about 150 yards wide. The tornado was rated F3 on the Fujita damage scale. Damage to homes was more severe in the tornado than hurricane-caused damage to similarly constructed homes in Louisiana. The tornado lasted ten minutes beginning around 8:10pm.

The eye of Hurricane Andrew skirted the coast along Vermillion Bay for several hours until curving northward and coming ashore near Burns Point, Louisiana around 3 am on August 26, 1992. Slow forward movement of the storm and close proximity of the eye wall over marshland caused the hurricane to weaken; central pressures rose and the hurricane was downgraded to category 2 status just after landfall. Rising pressures in the core of the storm led to lower wind speeds and storm surges inland in comparison to to when the storm struck Florida. Consequently, the damage-causing potential of the storm was less in Louisiana than in Florida. The eye of Hurricane Andrew eventually passed over Franklin, Louisiana just after sunrise. Towns of Morgan City, Berwick, and Patterson were located east of the eye and sustained the most severe wind damage. Only minor damage occurred in towns of Lafeyette, Baton Rouge, and Houma.

The Morgan City Power Plant reported a maximum sustained wind of 92mph (80 knots) with peak gust of 108 mph (94 knots) at 3:05 am from the south. The height of the wind equipment was approximately 50 feet above the ground. The highest reported storm surge was 9 feet NGVD at the Marine Conservatory at Cocodrie, Louisiana.

Andrew continued northeastward and was downgraded to a tropical storm during the afternoon on August 26, 1992 when the center of circulation was between Baton Rouge and Lafayette.

The National Hurricane Center reported that Hurricane Andrew rated as the fifth strongest storm to strike the U.S. mainland, and the most costliest storm in history in terms of the amount of property damage. Several figures and tables are appended summarizing the meteorological data.

B. AERIAL DAMAGE SURVEY IN FLORIDA

An aerial damage survey was conducted on September 5, 1992 over southern Florida. The survey began at the Miami International Airport. We made several east-west traverses south of the Miami airport from Tamiami Airport to Coral Gables southward to Florida City. Cities of Coral Gables, Kendall, Howard, Perrine, Cutler Ridge, Naranja, Leisure City, Homestead, and Florida City were included within the aerial survey. Our final traverse extended from Ocean Reef Key northward to Key Biscayne.

We made the following observations from our aerial survey:

1) Most of the damage to residences occurred south of an east-west line extending through Tamiami airport. The damage primarily involved loss of roof coverings that were tiles and composition shingles.

2) Concrete masonry residences and warehouses performed well. We saw few failures with these buildings. Damage to these structures ranged from F0 to F1 on the Fujita scale.

3) Mobile homes were destroyed and several mobile home parks south of 104th street sustained severe damage. In contrast, adjacent conventional housing performed much better.

4) There were certain subdivisions which sustained more building damage than adjacent developments. These residences had wood-framed walls and roof structures. Homes in Country Walk, Homes by the Bay, and sustained damage ranging up to F3 on the Fujita scale.

5) Severe damage occurred to large metal buildings. Many hangers at Tamiami airport and Homestead Air Force Base were destroyed. However, smaller metal buildings fared much better, especially if their large doors faced a direction opposite of the oncoming wind.

6) Water towers remained intact in each community we visited. A few television and radio towers were downed by the storm.

7) Coconut Grove Marina and Black Point Marina sustained heavy damage and many boats were cast adrift, floundered along shore, or were driven into the mangrove swamps.

8) Many trees were uprooted throughout the survey area. Trees fell to the west-southwest in the northern one-half of the survey area extending as far south as the Kendall and Cutler Ridge area. Trees fell to the north and northeast south of this area from Naranja through Florida City.

9) The greatest wind damage to residences and mobile homes was consistent with roof level (sustained) wind velocities between 120 and 130 mph in eye wall areas.

C. GROUND SURVEY OF FLORIDA

A ground damage survey was conducted throughout south Florida south of an east-west line extending through Tamiami airport. We assessed the performance of residential structures, metal buildings, concrete masonry buildings, and tilt-up concrete buildings. We paid close attention to details that involved fastening roof coverings, decking, roof trusses, and gable ends. The types of fasteners (nails or staples) and roof connections (metal straps or hurricane clips) were noted.

A transit and rod were used to obtain the heights of the storm surge along the coast. Debris lines and water marks in buildings were residual indicators of how high the storm surge was during the storm.

The following observations were made from our ground survey:

C1. ROOFING

1) Many concrete tile roofs sustained severe damage as the tiles were not well secured to the roofs. Tiles laid on mortar patties installed over rolled roofing performed poorly. Typically, failure initiated between the tile and the mortar patty or between the roll roofing and the deck. The result was numerous tile missiles which caused additional damage to buildings downwind.

Tiles nailed to wooden battens also performed poorly. In many instances the tiles were lifted over the fastener head leaving the nailed battens in-tact. In these instances, the fastener heads were smaller in diameter than nail holes in the tiles.

2) Lightweight three-tab shingle roofs sustained severe damage. Many homes were completely stripped of shingles. Shingle roofs that were stapled performed worse than roof shingles that were nailed. In many instances, the staples were installed incorrectly on the shingle with crowns oriented up and down (vertically) instead of along the width of the shingle (horizontally).

3) Built-up roofs over wood-fiber insulation board performed poorly when the fiber board was merely stuck down to the roof deck with asphalt. Failure initiated in the fiberboard just above the region where asphalt infiltrated the board. In some instances, little to no bond occurred between the asphalt and the fiber board.

4) Loose roof gravel became airborne projectiles and broke windows on several buildings downwind.

5) Heavy dimensional composition shingles performed better than tab type shingles, especially when nailed.

6) Overall, plywood decking performed as well as chipboard decking. Both experienced more failures when they were stapled, instead of nailed, to the rafters.

7) Pre-manufactured wooden roof trusses had little to no lateral bracing and cascaded (like falling dominoes) when the roof decking was removed. This problem was evident even when metal straps were used to secure the ends of the trusses to the top plates.

8) Wood-framed gable ends and exterior walls were lightly nailed to other framing members. Numerous failures were noted when the nailed connections pulled loose.

C2 RESIDENCES

1) Concrete masonry dwellings greatly outperformed wooden stick-type buildings. However, some of these buildings shared the wooden roof truss problems discussed previously. Few masonry wall failures were found.

2) Windows that were taped did not fare as well as windows that were boarded. Many windows were shattered by flying debris (broken tiles). Some boards were not secured well to the window frames and dislodged becoming airborne.

3) In general, mobile homes performed poorly. Many failures initiated when the stapled connections between the walls and floors as well as the roof trusses and walls had pulled loose. We did not find any metal plates between roof/wall and floor/wall connections. Exterior walls pivoted becoming horizontal as the roof covering tore loose. Several mobile homes that were not tied-down properly had rolled or flipped over on their sides.

C3 COMMERCIAL BUILDINGS

1) Many large span metal buildings collapsed. Typically the base plates on steel columns tore around the fasteners. Bay door failures were found on the windward sides of the buildings. These door failures no doubt contributed to increase interior wind pressures.

2) Several double-tee concrete roof beams cracked near mid-span and some had fallen into various buildings. Failure initiated when wind forced the top of the cambered member upward placing to tops of the beams in tension. Concrete tee failures were found in several strip shopping centers and warehouses.

3) A few tilt-up concrete buildings had collapsed when the roof along the windward wall was removed. As windward walls fell inward, the remaining walls fell outward.

D. GROUND SURVEY IN LOUISIANA

A ground damage survey was conducted throughout south-central Louisiana extending south and west of a line from Grand Isle to Houma to Lafayette. We assessed the performance of residential structures, metal buildings, and wooden commercial facilities paying close attention to connection details.

A transit and rod were used to obtain the heights of the storm surge along the coast. Debris lines and water marks in buildings were residual indicators of how high the storm surge was during the storm.

The following observations were made from our ground survey:

1) Most of the damage to residences occurred in a line from Morgan City to Franklin to Cypremort Point. Conventional wood-framed structures survived with little or no damage except on Cypremort Point where damage up to F2 intensity was observed. Failure occurred when homes were impacted by flying debris (metal boathouses). Gable ends failed in some instances, otherwise, most of the damage was confined to various roof coverings.

2) Mobile homes performed poorly. Many failures initiated when the stapled connections between the walls and floors as well as the roof trusses and walls had pulled loose. There were no metal straps between roof/wall and wall/floor members. Walls pivoted becoming horizontal as the roof covering tore loose. Mobile homes that were not tied down properly had rolled or flipped over on their sides.

3) Many trees were uprooted throughout the survey area. Trees fell to the south in the western one-half of the survey area west of Franklin. Trees fell mostly to the northwest east of Franklin.

4) Wind damage to residences and mobile homes was consistent with roof level (sustained) wind velocities less than 100 mph.