B513 Planning Meeting, 8th August 2002

Present:           Nigel Baddams, Tony Cass, Anne Funken, Jukka Lindroos, Jose Miranda-Vizuete

1.             Previous Minutes

  1. Approved with minor corrections. The Divisional Request for the UPS systems was created.

2.             Substation Planning

  1. Gutor have supplied information about the cooling requirements for their systems. Unfortunately these describe the “best case” requirements but the air conditioning system must be sized for the worst case scenario. The technical details from Gutor show that, unlike the MGE solution, filters to cope with 3rd harmonics would be required. This leads to an increasse in the space required.
  2. Anne has just received details of the power needs of the hvac systems; some time is needed to study these.
  3. Anne showed a very preliminary substation layout assuming 2MVA transformers and MGE UPS systems. This shows that a 19.4m long bunker is insufficient—extra space is required for general services switchboards.
  4. Nigel and Jose consider that a bunker longer than 19.4m is possible, but any part beyond this length must have a more restricted headroom if access to the barn door is to be preserved. At first sight, 2.8m headroom is achievable; this is less than the 2.95m desired, but the requirements could perhaps be reduced.
  5. Having analysed the loads further, Anne considers that a solution with 3 2.5MVA transformers is required in order to cover loads in B613 and B31. This solution also allows more optimal use of the existing UPS room if the physics UPS does have to be split for cooling purposes.
  6. The air conditioning arrangements for the physics UPS is still the major concern. Jukka calculated that 100 air changes/hour would be required to cool a 200kW load in a 3.5m high room with an area of 160m2. Assuming the same rate of air change (a rate Jukka considered to be at the limit of tolerability) two of the required three UPS units (each 3 modules and a spare) and their associated batteries could be housed in the existing UPS rooms. Space nearby would have to be found for the 3rd unit.
  7. Tony stated that we should assume the 3x2.5MVA solution is adopted and plan accordingly (see point 3.1). The following issues will be followed up before the next meeting.
    1. Nigel and Jose to better define the maximum headroom for the part of the bunker under the access to the barn door. Anne to check if the clearance requirements for the LV switchgear can be reduced below 2.95m.
    2. Anne to prepare a layout with UPS modules split 8+4.
    3. Jukka to study the provision of air conditioning for the UPS and for the bunker, possibly involving the use of S-401 for hvac equipment (but see 3.2b below).
    4. Jukka to establish which ducting in the S-206 corridor must be preserved and Nigel and Jose to discuss door locations for the bunker accordingly.
    5. Tony to investigate possibility of moving the PABX from S-049.

3.             Changes to support active load of 4MW

  1. The more information becomes available about the Itanium processors, the less the assumption of a constant load of 100W per box seems realistic. A better assumption seems to be a constant efficiency of 10SpecInt2000/Watt. With this assumption, the power demand in 2008 will be 2.4MW and could rise in later years. IT is therefore interested in the feasibility of supporting an active equipment load of up to 4MW.
  2. The responses were as follows
    1. In terms of electrical distribution, the density of normabarres would have to double. This is achievable and could be supported either by doubling the number of PDUs or by installing larger PDUs initially. In either case, the normabarres could be installed over time.
    2. The size of the substation (and the number of transformers) would have to double. At first sight, this could be accomodated as the bunker could have a length of up to 42m if extended to the corner of B513. Access to the extended portion would be more limited, and possibly at the cost of demolishing S-401.
    3. Phased construction of the bunker is possible if the end-wall nearest B31 is built appropriately initially. However, phased construction would probably lead to a sub-optimal layout of the substation and thus might not be possible if the space is too limited.
    4. The air conditioning capacity of the machine room could be increased above 1.5kW/m2 by blowing cold air below the false floor rather than from above. This would require a substantial modification of the air conditioning system, however.
    5. Provision of the necessary amount of chilled water would require additional (or even replacement) chillers and installation of new pipework with a greater cross section. This applies whether the chilled water is supplied to air conditioning units, directly to water cooled equipment (or racks) .
  3. Tony asked that these issues be studied further, although the priority for the moment remains the preparation of the 2.5MW solution.

4.             Machine Room Upgrade

  1. Anne has made an initial study of the possible power distribution in the machine room. A total of 18 PDUs equivalent to those in the vault would be required. When UPS arrangements are taken into account, 14 PDUs could be used completely independently of the vault, but 4 would be usable only if companion PDUs in the vault were not used.

5.             AOB

  1. Anne commented that we still need to understand how the cold water pumps are powered and the implications of this for the battery lifetime for the 200kW UPS.
  2. Anne asked for some technical information about equipment with dual power supplies. In particular, do these systems have one input labelled as the priority input? If so, this should be connected to the PDU which is supplied by diesels.
  3. Next Meeting: 14:00 on Tuesday 27th August.