1. Overview

2. Magnetic Field Strengths

3. Overall Magnet Construction

4. Magnet Coils

5. Magnet Connections

6. Tests

7. Packing, Shipping, and Staging

8. Safety Systems

Appendix A. Magnet Steel

Appendix B. Copper Conductor

Appendix C. Epoxy System for Coils

Appendix D. Coil Tests/Additions to RFQ

Appendix E. Magnet System Test/Additions to RFQ

Appendix F. Danfysik Quotation/Engineering Details

Appendix G. Guidelines for Refurbishing EAG Magnets

Appendix H. Support Jacks

Appendix I. Swivel Rollers

Appendix J. DC Power Terminals

Appendix K. Stresses and Strains

This is a report describing safety issues relating to the installation and operation of the PHOBOS magnet (PHOBOS is an approved experiment to operate at the RHIC project at Brookhaven National Laboratory, located on Long Island in New York). This PHOBOS magnet is a "double dipole" located near the RHIC 10 O'clock collision hall crossing point. Powering the magnet is a power supply, located in the 10 O'clock hall service/support building. Cooling is provided by a dedicated cooling tower, located on a pad to the left of the service/support building. (See Figure 1 for the placement of buildings and areas near the collision hall.) The PHOBOS magnet straddles the beam-pipe near the collision point (see Fig. 2) and bends charged collision products for momentum analysis. One of the PHOBOS magnet dipoles is located on one side of the beam pipe, the other dipole is on the other side of the beam pipe. The two dipoles are in a series magnetic circuit.

The magnet weighs 45 tons and is of conventional copper-coil resistive design; it has modest cooling and power requirements and the magnetic field is substantially confined to the gaps. Once installed, it will not be regularly moved. The PHOBOS magnet therefore presents no unusual safety hazards.

Figure 1. The PHOBOS magnet is located near the RHIC 10 O'clock collision hall crossing point. The power supply will located in the service/support building. The cooling tower will be located on a pad to the left of the service/support building.

Figure 2. Perspective view of the PHOBOS magnets. This figure shows a perspective view of the two dipole magnets of the PHOBOS experiment, centered around the beam axis. The beam axis is along the z-axis. The y-axis is vertically up, and the x-axis is defined such that the coordinate system is right handed. The plane z=43cm contains the pole axes.

A physics package occupies the full clear space ("gap") between the pole tips. The package consists of layers of silicon with associated electronics. Some silicon layers are in the high field region between the pole faces, and other layers are in a low field region in the volume between the gap and the collision point (see Fig. 9).

The PHOBOS magnet system comprises:

1. A steel magnetic yoke, two pole support plates, four pole pieces, four support columns, four support jacks;

2. Four potted epoxy copper coils and associated busswork;

3. Water manifolds;

4. Power supply, cooled buss.

5. Cooling tower.

6. Sensors for monitoring and safety

The goal of the PHOBOS magnet is to have as large as possible and as uniform as possible magnetic field in the gap between the poles, and as small as possible magnetic field outside the gap.

Figure 3. The schematic DC power circuit. At full field, the supply provides 3600 Amperes at 95 VDC.

The magnet design started in earnest early 1993 with Leslie Rosenberg (MIT) placed in charge of the magnet subsystem. We quickly settled on a resistive coil design with near-saturated pole-pieces. (Leslie reports to Bolek Wyslouch, PHOBOS Project Manager; Bolek reports to Wit Busza, PHOBOS Spokesman). Much of the first year was spent exploring the field from various pole geometries. Harald Enge——a well respected magnet designer——was contracted by MIT to realize the magnet design and deliver assembly drawings for submission to vendors. The drawings were checked by Bob Averil and Ernie Ilhoff (MIT/Bates). The RFQ was distributed to vendors in mid-1997. The contract was awarded to Danfysik (see Appendix F) with delivery to BNL September 1998. Figure 5 shows major dimensions in a side view of the magnet (for more detail, see the attached plans). Figure 6 shows major dimensions in a top view. The magnet can be rolled into the collision hall through the cargo door without dismantling the magnet.

Figure 5. Major dimensions in a side view of the PHOBOS magnet.

Figure 6. Major dimensions in a top view of the PHOBOS magnet. The coils are circular, except for straight sections to follow the pole outline.

Figure 7 shows a plan view of the outline of the PHOBOS magnet in place at the collision hall. Notice the magnet is offset from the collision hall centerline in order to be placed symmetrically around the beamline. Figure 8 shows a side view of the outline of the PHOBOS magnet in place at the collision hall. The geometric center of the magnet is offset from the nominal interaction point by 43 cm.

Figure 7. The plan view of the outline of the PHOBOS magnet in place at the collision hall; notice the time-of-flight walls and ductwork.

Figure 8. The side view of the outline of the PHOBOS magnet in place at the collision hall. Notice the magnet is offset from the collision hall centerline in order to be placed symmetrically around the beamline.

Fig. 9 represents the approximate field in the mid-plane of the gap as determined from a 3D TOSCA model at the nominal power consumption of 342kW. Looking at Figure 10a and 10b, the By component of the field should rise from near zero at the axis of the beam pipe (the beam pipe axis satisfies x=y=0 cm--see Fig. 2 for the coordinate system). By 35 cm inwards towards the pole tip axis from the larger circumference of the tapered poles, the By component of the field rises to > 2 T, and remains relatively constant in By as the x and z coordinates vary over the gap between pole pieces. At 15cm inwards, the field is less than 0.5 T, dropping off to zero as shown in Figures 10a and 10b.

Figure 9. The approximate magnetic field in the mid-plane in the gap of one of the dipoles as determined from a 3D Tosca model of the magnet at power consumption of 342kW. The beam-line is along the z-axis. The nominal interaction point is at x=y=z=0 in the figure.

Figure 10a (left) shows the magnetic field (in Gauss) in the mid-plane of the gap along the vertical (dotted line) trajectory (in cm) from Fig. 9; this is the straight-line trajectory leaving the beamline at right angles and continuing through the geometric center of the gap. Figure 10b (right) shows the magnetic field in the mid-plane of the gap along a straight-line trajectory starting at the nominal interaction point and continuing through the geometric center of the gap.

Figure 11 shows the magnitude of the stray magnetic field in the horizontal plane midway between the magnet pole faces; contour lines of 1 through 5000 Gauss (higher field contours are suppressed) are indicated, along with a top-view outline of the magnet. The horizontal axis is the distance, in meters, from the beam-pipe axis. The vertical axis is the distance, in meters, from the center of the magnet along the beam-pipe direction. The nominal interaction point is at horizontal coordinate 0.00 meters and vertical coordinate -0.43 meters.

Figure 11. The stray magnetic field in the horizontal plane midway between the magnet pole faces; contour lines of 1 through 5000 Gauss are indicated, along with a top-view outline of the magnet. The horizontal axis is the distance, in meters, from the beam-pipe axis. The vertical axis is the distance, in meters, from the center of the magnet along the beam-pipe direction. The nominal interaction point is at horizontal coordinate 0.00 meters and vertical coordinate -0.43 meters.

The full-field magnetic stress across a horizontal plane mid-way between the pole faces is 117,000 kg (force). The full-field magnetic stress across a vertical plane containing the beam axis is 2000 kg (force). These stresses were calculated from a 3D Tosca simulation.

The magnet mechanical construction consists of a rigid assembly of separable yoke, two pole support plates, four support columns, four pole pieces, and support jacks, which are doweled, pinned and bolted together in such a way that their alignment is reproduced upon disassembly and reassembly. Each flux return yoke is made from five pieces of steel; this segmentation into five small pieces allows for easier transportation and assembly of the magnet. The two outer pieces of the yoke run the full length of the magnet and provide the main support against deflection. The magnet gap between the poles (with coils unenergized) is 15.7cm. Under full power (nominally 342kW) and full magnetic field (2T maximum gap field) the deflection is less than 1mm. The upper and lower yokes are connected by four vertical support columns with welded-on bolt flanges.

The four pole pieces are of cylindrical "double-taper" design, with the addition of one vertical "cut" and one 12 degree "cut" (relative to the vertical) in each. The magnetic field is vertically upward-going in one gap, and vertically downward-going in the other gap (the two gaps are in a series magnetic circuit).

The magnet pole steel is XC06, which is very similar in mechanical properties to 1006 The balance of the magnet steel (for support columns, yoke, etc.) have less stingent magnetic saturation requirements and therefore are fabricated from 1018. The bolts are standard cap screws (SAE number 9 strength). The two upper poles are mounted on one support plate, the lower two poles are mounted on the other support plate. The plates are in turn mounted to the flux return yoke. Details of the steel may be found in Appendix A.

In order to provide flexibility in studying new physics signatures and to address concerns over potentially large RHIC backgrounds, the distance between coil pairs is adjustable. Adjustment of the distance would be a rare operation, occurring at most once or twice over the life of the experiment. To allow for the unlikely event we'll need to increase the distance between coil pairs, two sets of alternate pole bolt hole circles and pole alignment holes are provided at +5cm and +10cm offsets for each pole.

All exterior surfaces of the yokes are to be delivered free of rust, primed and painted with blue machinery enamel. The pole faces will not be painted, but instead provided with an anti-rust treatment recommended by Danfysik and approved by MIT. The magnet pieces will be labeled with part numbers to aid reassembly of the magnet.

The major magnet component masses are:

Upper yoke, 14,720 kg, composed of

(2) end pieces, 3400 kg/each

(3) middle pieces, 2640 kg/each

Lower yoke, 14,720 kg, composed of

(2) end pieces, 3400 kg/each

(3) middle pieces, 2640 kg/each

Upper pole assembly, 6105 kg, composed of

(1) support plate, 1285 kg

(2) poles, 1860 kg/each

(2) coils, 550 kg/each

Lower pole assembly, 6105 kg, composed of

(1) support plate, 1285 kg

(2) poles, 1860 kg/each

(2) coils, 550 kg/each

(4) Vertical support posts, 43 kg/each

____________________________________

Total magnet weight, 41,820 kg.

Details of the magnet base support are shown in Figure 12. At four places at the magnet lower flux return steel are four "L" pads. The vertical section of the pads are affixed to the magnet with five 3/4" number 9 bolts, the horizontal section of the pads are fitted with jacks. Via the jacks, the magnet may be moved from a position resting on the floor to its nominal position with spacing about 30 cm between the lower flux return and floor. Two jacks on each side are synchronized by a common drive shaft. Details of the jacks may be found in Appendix H; the jacks are Duff Norton M9105. The jack system is designed so that the entire weight of the magnet may be supported by the working load of just two jacks. (Also shown in the figure is an extension sleeve that would be used in the unlikely instance the PHOBOS experiment would operate with poles moved away from the silicon spectrometers. This unusual reconfiguration is not covered in this report.)

Figure 12. Details of the magnet base support.

BNL has agreed to provide PHOBOS a suitable power supply; work on refurbishing the supply is in-progress. Frank Toldo indicated our supply is made by Acme Rectifier, type PS-59005, serial number 438 (or equivalent). Its input requires 440 VAC, 682 Amperes, 60 cycle in three phases. The supply weighs 9000 lbs. The supply is about 5 feet wide, 7 feet deep, and 7.5 feet tall. It is air-cooled; ambient air enters from the bottom front and exits from the top rear, with power dissipation about 40 kWatts. The power supply provides up to 3600 Amperes, and can be tapped at four output DC voltages: 75, 88, 105, and 125 VDC. PHOBOS plans to use the 105 VDC tap, for maximum output power of 378 kWatts. The output polarity is manually reversible at the power supply; we do not anticipate reversing the polarity often, if at all. Power connections are made at the lower back on a silver-plated copper buss.

Since there were no particular dimensional or material constraints on the magnet design, the support structure is "over-designed to handle the strains produced by magnetic forces and gravity" (see Appendix K for a series of structural memos). Briefly, the flux return steel is designed to deflect less than 1mm; this deflection-limited design is far from encroaching on safety margins. Any one vertical support post could withstand the entire load--static plus magnetic--without collapse. The magnet is stable: it needs to tilt17 degrees before its center of gravity is outside its base. Tilting the magnet requires a transverse force of 14 tons.

Power and cooling busswork will be done in accordance with "Guidelines For Refurbishing EAG Magnets", BNL, August 1, 1995 (see Appendix G). Connections at the back of the power supply will be with type NAR34-2N terminals (see Appendix J). There will be a short "jumper" connection (approximately 12 feet) from the power supply lugs to the cooled busswork. The "+" connection is made up from 8 parallel MCM 500 cables (each cable carrying 450 Amperes), likewise the "-" connection is 8 MCM 500 cables. The connection at the cooled buss is via NAR34-2N terminals to a bussbar welded to the cooled buss. The contact buss is silver plated, and exposed connections are protected with dielectric covers in accord with the EAG specification. The cooled buss is made of two lengths of hollow copper pipe, each about 150 feet long. The pipes penetrate the berm through a 1 foot diameter insertion, entering the collision hall near the magnet about 10 feet above the floor. The pipes cross the hall and terminate in a "distribution box" fixed to the collision hall wall near the magnet. The pipe copper cross section is 3.0 inch ID and 3.5 inch OD, surrounded by an insulating PVC sleeve. Approximately 12.5 kWatts are dissipated as Ohmic losses in the cooled buss (150 feet "+" conductor and 150 feet "-" conductor, current of 3600 Amperes, resistivity 5.7 x 10-8 Ohm-foot, inner conductor diameter 3 inches, outer diameter 3.5 inches), or 41 Watts per foot of conductor.

The connections at the distribution box are the reverse of the that at the power supply side: A copper buss is welded to the cooled pipes and silver plated and again exposed connections are protected with dielectric covers in accord with the EAG specification. There are 8 MCM 500 cables at the "+" connection and 8 MCM 500 cables at the "-" connection. Here, as per Figure 3, half the current (1800 Amperes; 4 "+" MCM 500 cables and 4 "-" MCM 500 cables) connect the upper two poles in series, and the other half of the current (1800 Amperes) connects the lower two poles in series. Both upper and lower pole circuits cross from wall to magnet at the level of the top of the magnet (about 225 cm above floor level). The lower pole circuit continues down along one of the vertical support posts.

The electrical (and hydraulic) connections at each pole are supported by a thick dielectric backing plate (see Figure 13). The 16 hollow copper conductors exit each pole radially. Double pancakes are connected in series by small copper shorting tabs soldered between inter-pancake conductors. The shorting tabs are affixed to the dielectric backing with screws. The top conductor is soldered to a larger copper tab; this tab is also screwed to the backing and in addition has holes for affixing the 4 NAR34-2N terminals and MCM 500 cables. The lower conductor is soldered to a larger copper tab, as well, for fitting to 4 MCM 500 cables. At each pole, the 16 conductors, 7 shorting tabs, 2 buss tabs, and dielectric backing plate form a rigid assembly. Again exposed connections are protected with dielectric covers in accord with the EAG specification. The double pancakes within each coil are electrically connected in series, with the voltage drop across a single double pancake of 95 VDC/16= 6 VDC.

Figure 13. Electrical (and hydraulic) connections at each pole are supported by a thick dielectric backing plate. Double pancakes are connected in series by small copper shorting tabs, the tabs affixed to the dielectric backing. The top and bottom conductors are soldered to larger copper tabs; these tabs have holes for affixing power supply cables. The conductors and dielectric backing plate form a rigid assembly. Exposed connections are protected with dielectric covers.

Here, as per Figure 14, half the water flow (176 liters/minute at 4 bar differential pressure) connects the upper two poles in parallel, and the other half of the water flow connects the lower two poles in parallel. At the distribution box, each copper pipe makes a transition to two (approximagely) 2 1/2 inch I.D., flexible dielectric lines ("versa-con", or similar). As for the DC power supply lines, both flexible upper and lower coolant lines cross from wall to magnet at the level of the top of the magnet (about 225 cm above floor level). The lower cooling line continues down along one of the vertical support posts. These flexible lines carry coolant to the 8 water distribution manifolds (two manifolds per pole). The minimum length of dielectric line from distribution box to pole is 1 1/2 meter. This is much longer than EAG guidelines for 95 VDC isolation (1 to 1 1/2 inch per 10 VDC).

Figure 14. At the distribution box, each copper pipe makes a transition to two type ???? flexible dielectric lines. Half the water flow (176 liters/minute at 4 bar differential pressure) connects the upper two poles in parallel, and the other half of the water flow connects the lower two poles in parallel. Both flexible upper and lower coolant lines cross from wall to magnet at the level of the top of the magnet (about 225 cm above floor level). The lower cooling line continues down along one of the vertical support posts.

The eight manifolds are positioned as shown in Figure 15. The mainfolds are made from 2 1/2 inch metal pipe ("Tap Tube" type, Dynak Inc., or better; fabrication will be done by Danfysik) with end connections for the 2 1/2 inch flexible coolant lines and eight side connections for the flexible lines to the eight double pancakes. The upper manifold 2 1/2 inch flexible coolant inlet lines enter from the top, the lower manifold 2 1/2 inch lines enter from the bottom. All manifolds are isolated from the magnet steel with dielectric standoffs as per EAG guidelines. The connection from manifold to coil is via "Synflex 3740", 1/2 inch I.D., or similar dielectric piping. The run from manifold to coil connection is between 1/2 and 1 meter; the voltage drop between pancakes is 6 VDC, so this length of dielectric piping is more than sufficient for pancake-to-pancake isolation. There is a possibility that the copper conductor may be extended outward from the coils somewhat so that a break in the hoses will not flood the silicon; this change will have only very minor effects on safety issues.

Figure 15. Top view of magnet showing position of upper manifolds. The lower manifolds are placed similarly on the lower flux return.

Danfysik has agreed to notify MIT within fifteen working days prior to conducting any tests, and MIT has the option to witness any and all tests. Also, Danfysik is to provide all reasonable support to MIT personnel while witnessing tests. All tools, equipment, and accessories necessary to perform the tests are to be supplied by the Danfysik. The equipment required for measuring and testing is to be certified by Danfysik to be sufficiently accurate for the type of work to which it is applied and to have been calibrated to traceable to industry standards before and after use. MIT reserves the right to reject test results obtained with improper or substandard equipment. MIT representatives shall upon request be allowed reasonable access to the production and test areas of Danfysik and subcontractor facilities at any time during the progress of the work called for by this specification.

"Factory Acceptance" will be granted for the magnet upon satisfactory completion and written approval by MIT of all tests. "Final Acceptance" will be granted upon satisfactory installation and testing of the magnet by MIT at BNL upon MIT determination that the performance is satisfactory. "Final acceptance" means satisfactory installation and testing of the magnet at BNL and determination by MIT that the magnet performance is satisfactory. Danfysik is responsible for assembly and tests of the magnet at the Vendor's factory. MIT is responsible for assembly and Final Acceptance tests of the magnet at BNL.

The magnet coils will be tested both prior to and subsequent to epoxy impregnation. Each coil will be permanently identified with a unique serial number.

These are pre-impregnation coil tests as specified by MIT and agreed to by Danfysik: Prior to impregnation, Danfysik will conduct the following tests on the coils:

a. They will successfully blow a polished steel ball of a diameter not less than 0.6 times the diameter of the cooling channel (that is, 0.6 times 0.363 inch = about 1/2 cm) through each formed coil to check for obstructions to the coolant flow.

b. A test of the turn-to-turn insulation by means of a switched-type ringing tester developing a pulse of approximately 10 Volts per turn. A photograph of the resultant oscilloscope trace showing a slowly decaying waveform will be submitted to MIT as demonstration of the absence of any shorts. Each photograph will be clearly marked with the serial number of the coil being tested. A photograph of a shorted turn will be included for comparison.

d. The coil will be filled with water and pressurized to a minimum of 250 PSIG. The coil will then be isolated from the pressure source. The internal pressure and ambient temperature shall be measured at the beginning and end of a one-hour minimum test interval. The coil will show no change in pressure due to water leakage after any necessary correction for temperature change during the test interval.

These are post-impregnation coil tests as specified by MIT and agreed to by Danfysik

Subsequent to impregnation, the following tests will be conducted on the coils:

b. A test of the turn-to-turn insulation will be conducted by means of a switched-type ringing tester developing a pulse of approximately 10 Volts per turn. A photograph of the resultant oscilloscope trace showing a slowly decaying waveform will be submitted to MIT as demonstration of the absence of any shorts. Each photograph will be clearly marked with the serial number of the coil being tested. A photograph of a shorted turn will be included for comparison. (This is similar to the pre-impregnation test.)

In addition to these minimum MIT-specified tests, Danfysik has minimum standards the magnet must satisfy by their ISO 9000 certification (see Appendices D and E). In particular, materials must be certifiably up to standards, MIT receives voluminous amounts of quality control documentation, the coils and steel must be within dimensional tolerance at several stages in production, and the magnet and coils are thermally cycled.

Following satisfactory completion of inspection and tests at the Danfysik's factory, and after receiving written factory acceptance from MIT, the magnet system will be crated and shipped to BNL. The agreement with Danfysik calls for particular care to be exercised to protect machined surfaces from rusting, scratches, and nicks during shipment or packing for shipment. All parts will be numbered and directions will be given, referencing these part numbers, in Danfysik's supplied manuals as to the proper assembly. The agreement calls for the components to be crated in such a manner as to protect the contents from damage while in transit. In particular, the coils will be prevented from shifting or chafing due to jolts received in transit. The upper two coils, the two poles, and the upper support plate will be shipped as one complete pinned, bolted and aligned assembly at the minimum pole separation setting. The lower two coils, poles, and the lower support plate will be shipped as another complete pinned and bolted assembly at the minimum pole separation setting. The packing crates will be placed on pallets suitable for movement using fork lift trucks. All water circuits will be blown free of any water prior to shipping or storage. Danfysik will notify MIT prior to making the shipment to BNL. The notice will include the purchase contract number, origin, date, routing of shipment, approximate date of arrival, and shipping weights.

Delivery of the magnet pieces will be to the AGS by flat-bed truck. The heaviest single piece is the pole assembly at 6105 kg. Rigging the magnet will be done by BNL crews under the direction of Joe Scaduto. (There is a possibility that a commercial rigging company will move the magnet from the AGS to RHIC, this, too, would be under Scaduto's direction.) A forklift of suitable capacity will offload the magnet pieces from the truck and deliver them to a staging area at the AGS within reach of the overhead AGS crane. (Alternatively, the flatbed truck could position its cargo platform reach of the overhead AGS crane.) All delivered pieces have rigging holes. The magnet will be assembled starting by aligning the five pieces of flux return steel and bolting them together. The lower pole plate assembly (consisting of plate, two poles, and two coils) will then be placed onto the flux return steel and bolted into place. The four vertical support posts will then be bolted into the lower flux return steel. The upper plate assembly will be placed pole face down on two pieces of hardwood (the wood protects the pole face from scratching; any relatively soft material will suffice). The five pieces of upper flux return steel are arranged on top of the pole assembly and bolted together and to the upper pole plate assembly. Finally, the unit consisting of the upper pole plate assembly and upper flux return steel is lifted unto the four vertical posts and bolted into place. This completes the rough assembly of the magnet.

The only crucial intra-magnet alignment is the alignment of the upper pole axes with the lower pole axes; this alignment is done after rough assembly. For alignment, the bolts attaching the lower pole plate are loosened from the flux return steel. A special fixture, shown in Figure 16 is placed over a steel plate edge and loosened attachment bolt. By tightening the fixture bolt, the lower pole plate may be repositioned slightly within the bolt hole clearance to allow alignment of upper and lower pole axes.

Figure 16. Fixture for adjusting relative alignment of the upper and lower poles.

After assembly and pole alignment at the AGS, the present schedule calls for moving the magnet to the collision hall and cabling it up. However, if the 10 O'clock hall utilities are not in place, we will likely defer the move to the collision hall and cable up the magnet whilst at the AGS. I will assume the former scheduling; the later scenario has little effect on magnet safety issues.

After the magnet is assembled and aligned and the coils pass basic integrity and continuity checks, the assembled magnet will be moved to the 10 O'clock hall at RHIC. A flatbed truck will transport the magnet to RHIC. There are two possibilities for loading the magnet on the truck: (1) The AGS crane lifts the magnet unto the truck bed. The magnet weighs 44 tons, the AGS crane capacity is 40 tons. Periodically, the AGS crane is tested with an overload, and it is possible the 10% overload represented by the magnet can be accommodated as part of the normal crane overload tests. (2) If the overhead crane cannot be used, the magnet can be rolled to where it can be lifted by a suitably sized mobile crane. The magnet is rolled by jacking the magnet up about 6 inches on all four sides. The jacking is done by the permanent jack supports shown in Figure 12 (the jacks are described in Appendix H). Heavy duty rollers are then slipped under the magnet (the rollers are described in Appendix I), the magnet lowered unto the rollers, the magnet then pulled by forklift to where it can be lifted onto the truck. The lifting points are the four lower "L bracket" attachments to the jacks. Appropriately sized spreader bars are part of the rigging.

At the 10 O'clock hall at RHIC, the magnet will be offloaded from the truck, a reversal of the loading procedure: the mobile crane lowers the magnet onto the concrete pad, the magnet is jacked up, rollers placed under the magnet, and magnet lowered onto rollers. The magnet is then pulled by forklift into the collision hall through the utility tunnel access. A full-size mock-up of the magnet has been brought into the collision hall in this way and the stay-clear area for magnet entry marked on the floor. In the collision hall, the magnet is positioned transversely into place by the surveyors (while still on rollers), then the rollers are removed and the magnet vertically positioned by adjusting the jacks. The transverse and vertical positioning may have to iterate until the alignment is satisfactory. The magnet is then ready to be cabled up (cooling, DC power, and safety systems).

Fire protection is that of the standard RHIC system in the collision hall: temperature rate-of-rise sensors, smoke detectors, and sprinklers.

Occupants will be warned of "magnet on" status by a lighted enunciator affixed to the collision hall wall near the magnet. The enunciator is wired to the power supply and will light when the power supply is energized.

[Editors Note: Appendices not included in this HTML document]