California, the US Geological Survey and the media have inadvertently misled the public into thinking earthquakes just happen, and there is nothing to do but wait, suffer, and recover. Nothing is further from the truth. While humans have no control over Earth movements, much can be done to reduce financial loss, recover quicker, and save lives.
Little earthquakes are routine. The danger is small, recovery is prompt, and emergency forces cooperate. However, big quakes are few and far between (every 100 years or so), they are not routine, and one can overwhelm available emergency services and leave people solving problems they are not prepared to address.
There are several potency scales that geologists use to estimate the strength of a quake and where it is centered. For the public, a better indicator is strong shaking accompanied by loss of electrical power. If you lose power, you lose traffic signals and get traffic congestion. This delays rescue, may damage repair for all utilities, and will impact the assessment and delivery of news.
The literature is full of learned papers about the geology of Earth’s crust. The literature also contains many ramblings, masquerading as “how to’s,” which treat all big catastrophes the same. The next time you read one of these helpful books, see if it distinguishes between what is predictable and what is a surprise. Big floods, snows, heat, or winds give notice. Big earthquakes strike without warning.
There are companies in the disaster business. They provide planning help, advice, and supplies. Most focus on “life safety” issues and the first hours after the shaking stops. Also there are governmental emergency plans that guide multi-jurisdictional emergency forces in their efforts. These plans may be satisfactory for mid-range quakes, but they are not adequate for the BIG shaker.
You can’t duck, cover, and stay under your protective table for 72 hours. You must get up, brush off, get moving, assess your situation, and get on with it. If you are associated with a big computer center providing real time services, dispatch, managing financial accounts, switching communications, etc., you first need to see if your back-up site is intact and operating or whether you are facing an unplanned emergency. If there is a chance of an unplanned emergency following a big quake, read on.
1. Everyone rightly thinks about standby power.
The generators need to be started frequently (at least monthly) and run against a full load. Several earthquakes ago, a telephone central office generator powered up, ran a few minutes, and shut down. The fuel system had the usual shutoff valve between the fuel tank and the generator. The technician started the machine and failed to open the valve.
An industrial plant had standby power installed in a building that itself was at risk from a quake-caused chemical spill. If a nearby chemical tank gave way, the generator would be destroyed.
Another client had an online system in a highrise. Terminals were plugged into a special network of wall outlets used solely for emergency terminal power. But there was no standby generator to power these outlets. It probably wouldn’t have made any difference anyway, because useful work in an emergency also requires some overhead lighting and a healthy staff who don’t need the unpowered elevators.
2. Networks are very popular, and companies strive to keep the servers and the communications links online in emergencies.
Many online transactions require approval and access to status files. To continue to operate usefully, checks and balances in the online systems must survive or the system will just build queues.
A public online system (i.e. point of sale terminals) requires the remote terminals to be powered and secure during an emergency and the communication network to be intact. To do less means the central site is ready but receives only a trickle of transactions. Quake lore says A.P. Giannini, founder of Bank of America, set up a table on the sidewalk in front of his damaged bank (1906, San Francisco) and was open for business during a critical time.
3. A modern computer center has duplicate devices and software capable of switching around a failing unit.
Presumably the operational staff can recognize a damaged unit and switch it offline if it is not responding properly. Many years ago the State of California produced a guideline for computer centers that called attention to a problem at the edges of raised floors. Bracing needs to be installed at the edges of the room to take the side loads caused by a quake. Simple floor panels along the wall can jump up and allow the frame to move sideways and collapse.
I was in a computer center in Holland and a device that looked like a tool for raising floor panels was hanging on the side of a computer box. I reached over and grabbed it, and the whole room went quiet. It was the (unlabeled) master power switch. I was embarrassed. (If firemen drag a water hose into your center, your crew needs that switch.)
However, my mistake exposed the lack of a published restart procedure. The center had been running 24/7 for years, and the original management was long gone or promoted. It took two hours to power up air conditioning, lights, sequence up the electronics, test it all, and get back to work.
If your center does not exclusively occupy the building it lives in, adjacent activities can cause problems. I audited a center in a high rise and was concerned because the company cafeteria occupied the floor above, and there were no “rain coats” for the machines. A month later, a fire in the kitchen dripped water onto computer units.
4. Many modern computer centers build and maintain databases in support of online operations.
Special attention needs to be given to the boxes containing the online data. Usual practice is to keep file backups and activity files. If a failure occurs, the current copy can be rebuilt (provided the software has been tested, the surviving equipment is powered and cooled and constitutes a minimal system, and the staff knows how to reliably execute an infrequent procedure).
However, the length of time all this takes deserves discussion. The simplest recovery software is a batch update where the data file is passed and the activity file reflected in it. This could take a couple of hours. A more sophisticated rebuild would screen current transactions against the activity file, queue the ones that match for later attention, and process those that do not match.
An earthquake sends out two types of earthwaves when the earth shifts. The P-wave is first, causes little damage, and is followed by the more powerful S-wave. These two types of waves can be sensed (recognized) by off-the-shelf electronic equipment. If the P-wave is recognized, it may provide enough time for an orderly shutdown of the computer file system.
The problem occurs whenever the updated records are queued before being reflected in the database. If a quake happens and these queues are lost, the database is corrupted. However, if it is known that a shock is coming, the queues can be recorded in a scratch file for use during restart. This could greatly reduce the blackout period when the online system is unavailable.
Some CPUs had an optional feature called external interrupt. If triggered by a coming quake, this could interrupt routine processing and initiate the save.
5. There are earthquakes, and there are EARTHQUAKES.
There are small earthquakes every day. Occasionally, there are big earthquakes that cause big damage. These are not common and in the day-to-day business of life tend to fade and reside only in the recesses of the mind. However, six big quakes are notable. Their locations and dates of occurrence include the following:
- New Madrid, Mo. – 1812, 7.7 mag (Mississippi River ran backward for two days)
- Fort Tejon, Calif. – 1857, 7.9 mag (rent in Earth was 220 miles long)
- Charleston, S.C. – 1886, 7.3 mag (heavy building damage)
- San Francisco, Calif. – 1906, 7.8 mag (burned half of the city)
- Prince William Sound, Alaska – 1964, 9.2 mag (devastated Anchorage)
- Louisa Co, Va. – 2011, 5.8 mag (damaged Washington Monument)
Quakes are hard to forecast. The US Geological Survey (USGS) is trying, but good predictions are not yet possible. The best that can be done is study the past and rely on USGS to forecast the future.
The San Andreas Fault runs the length of California. The last big quake on this fault damaged Northern California (1906, San Francisco). They lacked a good municipal plan, and after the quake the city burned up. The last big quake on the southern segment (there are seven segments) of the San Andreas Fault occurred in 1857. It started in the middle of California (Parkfield), and the rent in the earth ran south for 220 miles. In an effort to communicate with the ordinary citizen, the USGS published a scenario describing what might happen to the LA area when this fault moves again. This paper is available on the Internet at http://pubs.usgs.gov/circ/1324/c1324.pdf.
If the San Andreas moves massively in the next 25-30 years (as forecast), several Southern California counties will shut down with long recovery times.
6. If your computer center lies right on top of the epicenter of a big quake, you have a lot more to worry about than the survival of your files.
But if you are some distance away, both you and your center may survive. The immediate damage to a specific building will vary according to eight attributes:
- magnitude and the depth of the quake’s epicenter
- distance the building is from the epicenter
- type of soil under the building
- building code in effect when plans were approved
- type of construction used
- height of the building
- orientation of the building to the quake waves
- damage to the buildings immediately adjacent to the building
After any initial damage, additional damage can be caused by
- further shocks
- fire
- water
- reconstruction
7. Building codes supposedly guarantee safe construction.
In California, those high standards haven’t quite been achieved due to politics. The master building code is updated every three years (so new construction could be three or four years behind best practices when the building permit is approved). After state approval, the new code must be reviewed and adopted by each jurisdiction (add another year). Some cities adopt the new code in total. Others find some sections to delay or omit entirely. It is possible that a set of plans approved by one city would require modification in another.
The biggest deficiency in current practice is non-compliant existing buildings. If a set of plans complied when the building was built, and then the code was amended, subsequent updates to the code are not automatically called to the attention of the building’s owner until the owner wants to make a major change to the building and seeks a new permit.
Thus, an existing building that has been consistently used for its intended purpose can be at risk. A big computer center that has had several equipment upgrades and system replacements may just be waiting for the next quake. If a family residence has been used continuously by the same family for a number of years, a strong quake could cause severe damage. This happened in 1989 in Loma Prieta.
In 1971 in San Fernando, reinforced concrete tilt-up buildings failed at the corners and walls spread enough that roofs collapsed inward.
In 1994 Northridge, they discovered an unexpected phenomenon: reinforced concrete buildings could fail under expected conditions.
For those operating in an old building, a comparison between the code used when building plans were approved and the current master code would be prudent. Any practicing industrial architect should have all the data needed for comparison.
8. About six months after each quake, the quake fraternity describes what happened in a published field report.
While each site is different, there is evidence the value of the contents of a building can approximate the value of the building itself. To be quake-ready, a computer organization must avoid internal damage. Not only must the equipment in the computer center be tied down, the rest of the site deserves attention, too.
Before 1994 Northridge, the USGS placed motion recorders in strategic buildings. During the quake, one recorded 1g horizontal force on the tenth floor of a high rise. That’s enough to dump desktop computers and roll wheeled furniture around.
For years a standard quake preparedness recommendation has been to tie everything top heavy or dangerous to the wall so it won’t fall over and hurt someone. Except for swivel chairs, all casters should have locks. Special attention should be given to evacuation routes so cabinets and/or equipment doesn’t shift and block doors. Don’t forget to run this obstacle course with no illumination but emergency lights.
Next time building maintenance is replacing tubes in an overhead light fixture – which is flush mounted in the dropped ceiling – climb up the ladder, lift a ceiling panel, and see if the fixtures are tied to the structure so they don’t swing free if the dropped ceiling would fail. There is a classic quake photograph of a four-light fluorescent fixture swinging from its electrical wires and just clearing a desk by a few inches.
9. Emergency response will be a problem after a big quake.
Commercial power is likely to fail, and if one is close to the strongest shaking (near the epicenter) or on soft soil, that person may lose communications. First find a working cell phone or a phone line with all the handsets securely in their cradles to get a dial tone. Given an active phone one may call for emergency service. By the time that person is connected with dispatch, all emergency crews may be out on calls. If they are fighting fires or freeing trapped individuals, a computer center may not rank very high on their priority list.
In California, a cooperative practice called mutual aid encourages those jurisdictions whose emergency resources are not overloaded to send their excess to any location where the local crews are overwhelmed. A quake as big as 1906 San Francisco can result in many jurisdictions needing help but no one having any to spare.
If your center deals with financial information, it is of special interest to thieves. That’s why one renders old storage media unreadable before being discarded. However, following a big quake it may be wide open to those who would steal information, saturate a system to deny service to others, or just perform mischief which could benefit competitors. With law enforcement saturated, and most of the staff working on recovery and reestablishing service, beware of vulnerability to those who would cause purposeful harm.
For those who have damage, plan on providing one’s own perimeter security. For those with an emergency generator, a couple of outside construction lights and some long outdoor extension cords would be valuable.
10. Availability of key staff may be one of the biggest problems.
As mentioned in the USGS scenario, a big quake will damage roadways and bridges, disrupt utilities, and demand instant response to free the trapped and contain the fires.
If the quake occurs during normal business hours, the personnel necessary to assess damage and recover in stages (according to one’s quake recovery plan) may already be on site. However, if they have not prepared at home, they may leave to look after their families.
In Southern California two on-duty highway patrolmen left their assigned stations and went to help another patrolman’s family. After the Katrina hurricane several policeman abandoned their stations to go home. To reduce the risk that key people will leave a center just when they are needed most, “urge” them to prepare at home so they can help at work.
The news media acts like quakes happen, and there is nothing that can be done about them. While the timing and the severity of a quake are not under human control, many of the preparations to reduce losses and protect lives can be undertaken with a (relatively) small investment of dollars and talent. In frustration with the risks being unconsciously taken, I wrote a series of quake readiness essays. They are free and available on the Internet at https: //sites.google.com/site/socalquakeessays/.
It is imperative that key staff prepare at home and have emergency training and occasional exercises.
Has an automatic door closer been installed on the vault door that will close the door if the commercial power goes off? It is unfair to expect some brave person to close the vault when visibility is limited by smoke and dust or he or she smells fire.
11. By now it has likely been deduced that big quakes are different.
They have many dimensions, occur with no warning, and are so infrequent and their reoccurrence period is so long that no one can remember the last one. For those near an earthquake epicenter, transportation, utilities, and communications will suffer; weak buildings can break; and a building’s contents can be severely damaged.
After a big quake, business will suffer for weeks, months, and even years, and shortages of food and fuel will occur. For those in California, be sure to be prepared. Likely there will be many who are woefully unprepared. Further, overwork will affect all emergency personnel, saturate medical facilities and make prescription drugs hard to obtain.
Management will be challenged to serve surviving terminals and provide access to uncorrupted files. Recovery will impact the bottom line, and the municipal tax base, and without some new initiative at the national level, insurance selloffs will drive the stock market down. Finally, federal recovery funds will be required.
Robert L. Patrick is a retired computer consultant, student of earthquake effects, sees opportunity to reduce loss and save some lives from a (infrequent) big earthquake. Quake preparations can either be done slowly and cheaply or much more can be spent in recovery. You can reach him at bobpatrick@mac.com.
