Work which involves an open flame or produces sparks must be specifically authorized via a Fire Hazard Work Permit or Operational Safety Procedure. Fire Hazard Work Permits are valid for a specified time not exceeding 14 days.
Jefferson Lab has a hot work program as documented in the following EH&S manual sections:
Section 6122 - Welding, Cutting, and Grinding Safety.
Section 6122-T1 - Use of Fire Hazard Work Permit.
Section 6122-T2 - Welding Safety Practices.
These sections are included for reference in See Also: Jefferson Lab Hot Work Program .
Employees are trained in making proper electrical connections to high-current leads.
Employees and outside users are trained in housekeeping procedures, especially minimizing combustible trash.
Specialized training is given to those who are authorized to investigate a fire alarm, and to those who have operational leadership roles (see See Also: Procedures for Responding to an Endstation Fire Alarm ).
Essentially all cables used in the CLAS have a fire rating. Whenever feasible, a higher rated cable was purchased over lower rated or unrated cables. An inventory of cable types and their fire ratings is given in See Also: Cables used in the CLAS and their fire safety ratings .
|
See Also: Some of these cables are rated CL2; others, purchased early in the project, are not marked with a fire rating. |
||
The only connections carrying electrical power which are not commercially manufactured are for the drift chamber low voltage power. The power connections are distinct from the signal cable connections so signal and power cables cannot be interchanged. The power connectors are equipped with positive key interlocks to prevent erroneous connections.
The hazards are monitored using highly sensitive, multiply redundant automatic systems which detect elevated temperatures, smoke, and gases related to overheated fuel. In addition to the automatic systems, remote visual verification of conditions in the e ndstation is available.
A programmable logic controller (PLC) monitors the status of all fire safety related hazard signals, makes alarm level assignments based on these inputs, and announces any alarms via annunciator boxes in the counting house.
The approach of using a PLC was chosen because of the very high degree of reliability and flexibility offered by these systems. The hardware and software components were chosen so that compatibility with the international standard for use of PLC's in safe ty systems could be achieved (IEC 1131), including the IEC compliant programming language "Structured Text". Increasing the number or type of inputs or outputs is a simple matter of adding modules; the control or monitoring points can be located anywhere a network cable can be placed, at distances of up to 6000 feet. Peer-to-peer device communication on the high-speed, dedicated network (~1 Mbyte/sec) allows near-instantaneous response to any alarm. The flexibility of having a user-accessible pr ogram means reconfiguring the system, including the alarm logic, can be easily performed. Implementing any changes requires only a few seconds of system down time for downloading the new control program into the CPU and restarting the controller. (As in a ny such system, validation procedures must test that the hardware and software are all functioning properly.)
The initial PLC configuration is indicated in See Also: Initial PLC configuration for monitoring fire safety systems and related systems. .
The intention of the PLC-based system is to provide very early warning, therefore, the alarm levels will be set with the intention that the this system will respond well before the building fire alarm system. (The building fire alarm system is triggered b y a high-level VESDA alarm, heat detection in the dome of the hall, or by a water-flow indication from a sprinkler head actuation.) The Hall B system produces `pre-alarms,' and the building fire alarm system produces `alarms,' the distinction being made i n order to define who is in charge of the response (see See Also: Procedures for Responding to an Endstation Fire Alarm ).
The basic function of the system is to alert workers in the hall or in the Hall B counting room of a pre-alarm condition as soon as it is identified, and give information on which system reported the problem, the location of the problem, and an estimate o f how serious it is (see See Also: Initial PLC alarm logic. ). The resulting actions can then be very efficiently focused on investigating the nature of the problem, such as by immediately entering the hall, or by looking at the closed circuit TV camera displays for signs of trouble. In addition to reporting to th e counting house and the hall, an autodialer allows notification of up to four telephone numbers or pagers on an equipment trouble condition, or up to four (potentially different) telephone numbers or pagers on a warning or alarm condition.
The initial PLC alarm logic is shown in See Also: Initial PLC alarm logic. . Four levels are defined: Normal, Warning, Alarm, and Trouble. The numerical definitions of the individual settings may depend on conditions in the hall, and will be adjusted according to experience.
In case of power failure, the PLC and its annunciator boxes have a local backup power system (UPS) which will provide 15-30 minutes of additional operation. The system has been designed to have a minimum number of components, and the components have been selected with reliability in mind. For instance, all the bulbs on the annunciator panels are neon lamps with a 25,000 hour mean operating life; no relays or switches are wired between the PLC output modules and the indicator lights or audio alarms; each a nnunciator box has two audio alarms wired in parallel, so that if one fails the other still operates. Devices such as the linear heat sensors use a monitoring current, so that a broken wire results in a `trouble' condition. Regular maintenance and testing of the system by Hall B staff will verify continued operation.
In the case that additional suppression systems are installed in Hall B, the PLC system could be used as part of the input decision as to whether to release an extinguishing agent (such as water mist).
The Very Early Warning Smoke Detection Apparatus (VESDA) is part of the installed, approved, automatic fire detection system. It provides detection of a fire in its incipient stage. The VESDA sampling tube system includes extensive arrays of sampling tube s on each level of the forward carriage, the space frame and the north and south carriages. The system also continuously monitors the return air of the central air handler unit in the hall.
The system has an array of 15 dry contact outputs which are available for use. The `or' of these is used to kill all Hall B electrical power at the same trip level that the building fire alarm system is activated.
The analog outputs of the VESDA system are monitored by the PLC from 0 to 5 volts, and are also monitored by the Hall B slow controls system. During normal operation the slow controls system plots these levels and additionally provides an audio and visual alarm, as well as archiving these voltages.
The system is configured in an approved way and is maintained by contract with a local vendor; this service is coordinated by the Jefferson Lab Fire Safety Engineer.
In addition to the commercial VESDA system, Hall B has developed a second incipient detection system referred to as the `Sniffer' because it detects gases. This system is similar in spirit to a pioneering system developed at Fermilab 12. It looks for gases which are emitted from flammables when they are heated, in principle finding the precursors to a fire before any flame has formed. This approach is partially redundant to the VESDA system, and partially complementary, in that some mat erials emit smoke when heated below the flame point, and others do not; for PVC cable jacketing, this property depends on the additives such as plasticizers and fire retarding chemicals, and so varies widely among cables which are otherwise similar.
In the Fermilab system, up to three gases are detected at the same time by point measurements, that is, each area monitored has three local detectors in it. In the Hall B system, a different approach has been taken. Here there is a single detector, and se quential sampling through long tubes is used. This approach has the disadvantage that one of the PVC signature gases (hydrogen chloride) is not readily detectable because it is deposited on the tubing wall in the presence of humidity and therefore doesn't transport well. Another disadvantage is that there is a delay in the average time it takes to sample the gas in a given room; since the sampling is sequential, a problem could go undetected in one area until other areas have been sampled.
The advantages to this approach, however, are several. It permits the use of a single detector which can be located outside the radiation area, allowing servicing at any time and removing the concern for radiation-initiated problems. It is also possible t o buy a higher-quality instrument as the detector, since only one (or one set) is required. A higher-sensitivity instrument requires less signal integration time, which means the measurement time at a given point can be short, reducing the average time de lay of the overall system. Propagation delays for transporting the gas through long tubes can be minimized by using high pumping rates and large aperture tubing.
The gas detection instrument chosen for the Hall B system is a Fourier Transform Infrared Spectrometer (FTIR), a device which measures the absorption spectrum in the mid-infrared. Since the gases of interest all absorb in the infrared, this device is sens itive to all of these simultaneously. The minimum measurement time is 4 seconds; a 10 meter folded optical cavity is used to increase the sensitivity. For 1 ppm sensitivity, multiple measurements improve the error estimate.
With a system of this type it is possible to contemplate extracting `fingerprints' of each combustible, that is, one can in principle identify which combustible is being heated, from its infrared absorption pattern. To fully exploit this option requires a systematic study of the effect of temperature, humidity, and other relevant variables, a pilot study of which has been performed at Jefferson Lab. In this study, all the cables in the hall were heated to approximately 180 degrees C and the gases measured . This was done both in a dry nitrogen environment and in air of normal humidity; the study was also performed with scintillator, lightguide, and PVC pipe, representative fuels in the hall.
Even in the absence of a fuel-specific identification, it is possible to identify gases at variance with the nominal hall background gas spectrum. Hydrocarbons are identified by a characteristic feature of stretching the C-H bond, which occurs in a relat ively background-free region of the spectrum. Carbon monoxide can be quantitatively measured at the few ppm level. In addition, there are absorption peaks for many of the combustibles in background-free regions of the spectrum besides these well-defined c omponents. An advantage peculiar to this approach is that the background spectrum can be taken as normal air in a fuel-free region of the hall; in this case a sensitive measurement of the deviations from normal air is possible. Even if, for example, a veh icle is operated occasionally in the hall, as long as the vehicle combustion products are uniformly diffused throughout the hall and change concentration slowly, they can be included in the background measurement. This feature recovers some of the specifi city lost by not having the HCl signature gas.
In the Hall B system, the PLC controls the gas valves which select the area being sampled. The concentration of the gases found are transferred from the FTIR to the PLC using standard software over a dedicated network. The PLC evaluates the gas concentrat ions to determine if they are outside the acceptable range. The initial locations of the Sniffer tube inlets are indicated in See Also: Initial locations of Sniffer tube inlets..
All cable trays and cable runs are to be outfitted with linear heat sensor wire, including the captured wire in-between the drift chambers. In addition, the ceilings of the electronics rooms and other locations are instrumented as needed. These heat senso rs are the nominal 155o F type, and therefore respond at a temperature comparable to or slightly lower than the sprinklers.
The linear heat sensors report to the PLC (see See Also: PLC-Based Alarm Processor) so that fault information goes directly to the counting house staff and is incorporated into the position information reported by the PLC annunciators. The system is divided into 24 zones; a list of the zones is given in See Also: Initial zoning of linear heat sensors.. Additional zones can be trivially added as needed. These zones are intentionally much smaller in area than the typical installation of this type of product. This was done in order to pinpoint the location of the problem in a way which does not rely on calibrations or calculations.
A testing mechanism has been implemented by providing three-way switches in a junction box at the end of the wire in each zone. The three switch positions indicate 1) normal operating conditions, 2) the `broken wire' configuration, 3) the `shorted' config uration which imitates a fire condition. The testing procedure requires a technician to locate the zone, inspect the wire placement and condition in each zone, and to position the three-way switch in each of its positions; technicians at the locations of both annunciator boxes then verify by telephone that these report the correct location of the simulated problem, and report a simulated broken wire.
Linear heat sensors are adapted to monitor the extended leads on the high-current magnet supplies.
The Hall B Counting House is equipped with a closed circuit viewing system to monitor Hall B equipment areas. The experimental areas are viewed from the control console on a set of six monitors which can cycle through all the cameras or stop at a particul ar camera. This allows the operators to monitor important pieces of equipment, the operation of the experiment, and identify fires or pre-fire conditions should they develop. This system has been documented 13.
A total of approximately thirty cameras have been mounted in Hall B. One camera views the VESDA readout electronics (including a view of the bypass switches), seven cameras view the forward carriage instrument racks from both the north and south side, two cameras view the instrument racks on the side carriages, and fifteen cameras view locations on the space frame. In addition to these, there are a number of others which view the gas shed, faraday cup, tagger beam dump, access cage, as well as a number of spares for future use.
The temperature of each of the large normal-conducting magnets is interlocked to its respective power supply by a series Klixon circuit. All of these magnets are water cooled.
The cooling water flow for each of the large beamline magnet power supplies has an interlock to the supply power. The temperature of these supplies is also interlocked to the supply power.
The DC low voltage supplies have two levels of protection. Each individual line is externally fused; both the supply and return line are individually fused, protecting against single fuse failure and other failure modes. The fuse values are chosen by meas uring the actual current draw, then increasing this number by 20% and rounding up to the next available fuse value. The second level of protection is that the power supply internal trip circuit is set to 2 A greater than the normal current draw, which is approximately 16 A for the Region I chambers and 31 A for the Regions II and III. This also protects against fuse failure and various common mode failures.
All crate supplies have overcurrent, overvoltage, and overtemperature protection. A network which monitors the status of these crates has been designed and partially implemented; this network provides notification to the counting house of the crate status . This information will eventually be included in the PLC-based notification system (see See Also: Future Upgrades).
The Hall B experimental equipment extends beyond the CLAS detector and includes apparatus along the beamline both upstream and downstream of CLAS. Most notable of this equipment are large magnets and their associated power supplies. In order along the bea m line these are the following magnet types: four Møller rasters, two Møller quads, 2 target rasters, tagger magnet, 2 sweeping magnets, 2 target rasters, mini-torus, and the pair-spectrometer. Not all of these magnets are operated simultane ously, and power supplies will be shared among the magnets.
All of these power supplies will be monitored and controlled remotely. The first line of defense against overcurrent will be the software control system which will shutdown the power supply if the read back current does not fall within some nominal range of the set current (usually 3-5%). However hardware interlocks to guard against overcurrent are also implemented in case of software failure or when the supply is operated in local mode. These hardware interlocks include overtemperature sensors on the mag nets and internal interlocks in the power supply.
During data acquisition periods the current Jefferson Lab policy permits undelayed access by counting house personnel to Hall B, under emergency conditions. The access policy for fire alarms and pre-alarms is given in See Also: Procedures for Responding to an Endstation Fire Alarm . This includes response from the MCC as discussed in the Appendix.
Emergency access to Hall B is simplified relative to the other halls in that the level of persistent radiation is very small when the beam is turned off. From a radiological control point of view, exposure to residual radiation can be controlled by limiti ng access to parts of the upstream beamline, and to the beam dump (Faraday cup) area. The details of the control of these areas are given by Radiation Control Group policy.
If the pre-alarm panel has indicated a possible fire condition, the delays involved in getting counting house personnel into the hall include the time required to decide to make an entry, notify the crew chief, and run down the stairs to the hall. If the Hall B fire alarm has sounded, access to the hall is subject to more requirements, including training and authorization of individuals who may lead a two-man investigation (see See Also: Procedures for Responding to an Endstation Fire Alarm ). If an actual fire is known to exist in the hall, no Jefferson Lab staff enter the hall, rather, information is gathered during the time the fire department drives to the site.
Rapid emergency access for firefighters is available according to laboratory policy. This includes a policy for emergency personnel entering oxygen deficiency hazard areas.
The transit time for firefighters to get to the site has been historically measured as approximately five minutes. The distances to the three local fire stations are 1.2, 1.8, and 3.8 miles.
Firefighters entering Hall B may be delayed by several actions which must take place first:
a) Shutting off the electron beam.
b) Unlocking and opening the secured gates and doors.
d) Donning SCBAs and fire fighting gear.
f) Reaching the area via the truck ramp or the labyrinth passageway.
2 Actual Hazards to Responding PersonnelThe following hazards could potentially be present during a fire in Hall B: heat, flame, smoke, oxygen deficiency, radiation, radioactive particulate, exposed lead particles, cryogenic fluids, extreme cold, toxic gases, trip and fall hazards, high voltage , magnetic fields.
Hand held extinguishers as provided in the hall are typically the 10 lb. carbon dioxide extinguisher. The BC rated CO2 extinguisher provides a clean agent that is not damaging to the equipment. They are located near the exits to each level of the four pla tforms.
Two wheeled Halon 1211 extinguishers are available on the floor level of the hall. The primary utility for these extinguishers is to put out fires which are out of reach, such as on the time-of-flight detectors mounted on a side carriage.
Dry standpipes are provided in Hall B which require fire department hookup on the outside of the hall. (There are no wet standpipes with hose available.) The hookups are located near the counting house and are clearly labelled. These hookups are identifie d during fire department familiarization tours.
There is a single entry point to the site for the fire department, which is the Onnes road exit accessed by Jefferson Avenue. The guard at Post 2 directs the fire trucks from that point to the location of any fire alarm on the site. For the firefighters w ho will enter Hall B there is little chance for confusion in that the distance from Post 2 to the hall is only a few hundred feet and the Hall B truck ramp is visible from that point. If a water truck needs to get to the dry standpipes, however, the route taken is a more complicated one.
The paved area around the entrance to the truck ramp provides ample space for ground staging operations, even when a number of cars are parked in the area.
The `fire department phones' report to the MCC building. The MCC crew chief and the battalion chief can use these to communicate with firefighters within the building. Additionally, cordless emergency phones are available at the top of the truck ramp and in the counting house. These have been tested to function throughout the hall and personnel access labyrinth.
The fire department access will be either through the truck access ramp or through the personnel access labyrinth. This decision is made by the battalion chief.
A fire hydrant is located adjacent to the opening of the Hall B truck access tunnel on the east side. This is the only hydrant which could practically service the hall without making use of the dry standpipes.
The truck access tunnel is a direct route into the hall. However, it is not a well-sheltered area in case of a large fire. The rollup door is not a fire-rated door, and if the smoke exhaust fans are turned on, both rollup doors must be open. The personnel accessway is a sheltered route into the hall, since it is separated from the hall by commercial fire barriers. In a small, early-stage fire, either location is well-sheltered from smoke or heat. In the targeted scenario of very early warning coupled with rapid fire department response, either entrance point is an acceptable area of refuge.
TJNAF staff are trained in general workplace fire safety training. `Hands on' fire extinguisher training is required for "hot-work" workers, for all Hall B staff, and MCC operations staff. In addition, a number of Hall B outside users have recei ved this training. Hall B staff, MCC staff and a number of users have also received training in the use of the two wheeled Halon 1211 extinguishers located on the floor level of the hall.
The passage of smoke or fire between the experimental hall and the accelerator tunnel is prevented by a commercial fire door and the thick cement wall used as a radiation shield. The same function is performed to isolate the hall and the personnel access labyrinth with a commercial fire door and professionally installed firestops in the cable tray in this area.
The forward and large angle calorimeters have very substantial barriers which protect the scintillator material within. The sidewalls of the forward calorimeter, for instance, are 1.3" thick aluminum, while the front face is a 3" thick composite of stainless steel and structural foam.
The time-of-flight scintillators are protected on the back side with a composite backing structure similar in construction to the calorimeter box plate. The exposed side faces the target where any additional material will impact the physics program. There is, however, a layer of aluminum foil 0.002" thick which is wrapped around the scintillator, and there is a 0.005" thick layer of lead on the side facing the target. Therefore there is approximately a 0.007" thick metal layer on the face o f these detectors which provides a significant measure of protection against sparks and a marginal amount of protection from small fires.
The two gaseous detectors, the drift chambers and the Cerenkov counters, are protected on two sides either by the magnet cryostats or aluminum walls. However, they each contain entrance and exit windows which must be thin in order to meet physics requirem ents.
The drift chamber endplates are inerted in all three regions. The primary reason for doing this is for improved technical performance (limiting high voltage currents), however, it also has the desirable effect of removing oxygen from the vicinity of the o n-board electronics. As long as this condition is maintained, a fire in this location is not possible, a great advantage since there is essentially no access to these locations to extinguish a flame.
The output of the Hall B fire alarm system is monitored at the Machine Control Center (MCC). This monitor reports an alarm if fire is detected in Hall B. The alarm does not automatically shut off the electron beam to Hall B, however an operator in the MCC can, at the time of an alarm, shut down the Hall B beam line or even the entire accelerator. Other fire sensitive signals can also be made available to the MCC either through a direct line from the Hall B counting house to the MCC or through the slow con trols system via ethernet.
It is possible to interlock the fire protection system into the accelerator fast shutdown system. This system takes as input a 24 VDC signal, and if this 24 V level transitions to ground the accelerator will be shut down. Hall B has requested that another level be created on the fast shutdown system that will terminate beam to Hall B alone and not shut down the entire accelerator. Such a hall specific shutdown system is more appropriate for an automated beam shutdown due to a fire alarm in the hall. This option will be taken advantage of at a future date.
Personnel entering the hall to investigate a fire alarm or pre-alarm are protected from beam exposure by the Personal Safety System (PSS). The PSS will not allow anyone to enter Hall B without first terminating the presence of the electron beam; it interl ocks access doors with the electron injector to the accelerator using redundant systems. The PSS does not protect personnel from radiation exposure due to activated elements in the Hall. Due to the low beam current in Hall B it is expected that the activa ted elements will be limited to the beamline, especially the Faraday cup.
A detailed description of this system and its associated safety features exists. 14 The cryotarget was constructed by a group at Centre d'Etudes de Saclay in France, in cooperation with Hall B engineering staff. This system was reviewed for safety by an external committee. The target is used both for photon and electron beam experiments. A brief summary is included here for completeness; reference to the original document should be made in the case of detailed questions.
All routine target operations are completely controlled by an industrial process controller (brand SEGELEC-GEFANUC). All pressures, temperatures, and levels are stored on a regular basis on hard disk and are histogrammed.
The cryogenic target is designed to operate with liquid hydrogen, deuterium, 3He or 4He. The operating parameters are different for these four liquids; for the flammable gases, the operating temperature is 20 K, and the pressure in the holding cell is 1.0 5 bar. The volume of the gas is 1600 liters for H2 and 1700 liters for D2, of which only 600 liters is within Hall B; the rest is contained in the storage system outside. In the Jefferson Lab Flammable Gas Standard categorization scheme, the outdoor stora ge area is at risk class 0.
A passive feature of the system in the case of target cell rupture is the large vacuum volume of the system. In the case of a target cell rupture the vacuum chamber pressure rises to less than 1 atmosphere (absolute pressure). In the event of a rupture of the vacuum system, the hydrogen warms up and is released back into the storage vessel by a pressure relief valve. Every electropneumatic valve in the system is protected by a relief valve.
In each electronics area there is a power kill switch for the clean power and for the utility power. These are implemented using shunt-trip breakers. A testing program for these devices exists (see See Also: Elements of Hall B Maintenance for Safety-related Systems ).
In the counting house there is an array of kill switches which duplicate the function of the buttons in the hall. In addition there is a master kill switch which shuts down all of the electronics rooms, both the clean power and utility power. There is als o a connection to the VESDA dry contact relays which automatically kills power if any of the VESDA circuits reach the same level that activates the building fire alarm system. The PLC can optionally also automatically kill power.
At this time there are not power kill switches for the large magnet supplies. When operating in local mode these can be turned off with the normal switches on the power supply. When operating in remote mode these can be turned off using their software con trols.
Protection systems are designed to ensure that a fire will be successfully controlled until such time that emergency response forces arrive to complete extinguishment. Since there are some fuels which cannot be protected by conventional suppression system s without compromising the functioning of the detector, the detection, notification, and response aspects of the overall scheme have been emphasized. Sprinkler systems have been provided for the fuels which are able to be protected by conventional systems .
Automatic sprinklers have been installed by contract with qualified service personnel in all levels of all four platforms which have electronics or fuels. These are wet pipe sprinklers requiring only a fused or melted sprinkler head before the discharge o f suppression water. This configuration minimizes the delays before automatic suppression can take place and relies on a minimal number of components. Given the relatively low initial energies of fires that can be expected with the electronics in the thes e areas, the effectiveness of the system is limited to preventing spread of fire in an advanced stage.