B513 Planning Meeting, 6th June 2002

Present:           Tony Cass, Anne Funken, Jukka Lindroos, Denis Raffourt

1.             Planning for the new Substation

  1. Anne presented the current plans for the new substation and, in particular, the bunker. In general, plans are well advanced and, as the work can be done under an existing contract, it should still be possible to commit money for the civil engineering work in 2002.
  2. Two options for the transformers are being considered: 3x2MVA or 2x3MVA. Anne considers 3x2MVA to be the best option—both in terms of redundancy and because 3MVA transformers are not standard at CERN.
  3. The major uncertainties for the substation concern the UPS systems; the background details are presented below.

1.1.                General UPS Issues

  1. CERN has a contract with Gutor for the supply of UPS systems. In the contract, the maximum rating of Gutor units is 320kVA. For the time being, only Gutor units less than 120 kVA have been installed at CERN. The actual computer centre UPS system comes from MGE (400kVA units which we have experience with). In case the CERN supply contract for UPS systems is not used a Market Survey/Tender process would be needed as the expected price is such that FC approval would  be required.
  1. It is clear that we should replace the existing system completely. Although it would be possible to integrate the existing units, they are outdated. Model production started in 1974, ended in 1994 and spare parts are guaranteed only until 2004.
  2. The UPS (and transformer) sizes depend on both the maximum load to be supported and the power factor. Tony reconfirmed the acceptibility of 2MW as a hard ceiling on the maximum power.
  3. As for the power factor, the measured value today is 0.7. Assuming a power factor of 0.8 would lead to savings as there would be no need for harmonic filters and smaller cables could be used. However,
    1. Increasing the power factor to 0.8 requires more stringent control over the power supplies in equipment that is purchased. This may not be possible and will lead to continued recurrent costs.
    2. The UPS saving is relatively small: 10 units would be required rather than 11.
    3. Computer centres we visited have sized their installations to support a  power factor of 0.7.
  4. It was agreed to assume a power factor of 0.7.
  5. Two UPS systems (2MW and 200kW) are required to allow critical equipment to be supplied by the diesels. This configuration also provides redundancy at the UPS level for critical equipment. A final decision on the battery lifetime for the 200kW UPS (2 hours or 10 minutes) has yet to be taken. One important aspect to consider is that air-conditioning equipment is not resupplied by the diesels. The main computer room is probably large enough to absorb the temperature rise from a 100kW load for up to 2 hours but the situation for the vault is less clear. In any case, intermediate solutions are not sensible—either power is restored quickly or only after a long period. Both options should be costed.
  6. Two UPS options were rejected.
    1. A special backplane that allows units to be added without taking the UPS out of service. This is unlikely to be of use as the maximum load will be at the beginning; we will not be installing units progressively as load increases over a number of years.
    2. A double bypass unit which allows UPS power to be maintained through all maintenance operations. Here, the additional expense is probably not worthwhile as the exposure is at most 2hours/year and the bypass unit does not increase the overall mean time between failures.
  7. Given the power factor of 0.7 (see above), filters will be necessary to smooth out 3rd harmonics. This needs to be studied in more detail. Anne proposes to study the possibility of simply reserving space initially, installing the filters later if necessary. This is a reasonable approach in principle, but we need to find out how much the filters cost in any case and to understand how quickly filters could be added later and with what level of disruption.
  8. A total of 11 UPS units are required—8 to cover the basic load plus 3 for redundancy. These, and the batteries, will fit into the space foreseen. However, the air conditioning may be a problem. From MGE’s figures, the UPS heat load is over 200kW and a huge airflow is required. It may not be possible to meet these requirements without increasing the area of the UPS rooms. Anne and Jukka will investigate further, concerning systems from both MGE and Gutor.

1.2.                UPS and Diesel Load Issues

  1. The HVAC load to be retaken by Diesels is limited to 500kW. 50kW is required for the tape vault (to power 4 of the 6 units) Jukka is to details other loads.
  1. Jukka and Anne must define the loads that need to be maintained during switchboard maintenance. Dual power supplies must be arranged for these loads. Similarly, Denis Raffourt must identify smoke detection systems that need a dual power supply.
  2. On this point, the “aspirateur” for the smoke detection system under the false floor in the machine room (a system of tubes each with own detector, but sharing a common “aspirateur”) is backed up by the Diesel sets although it is not on the UPS. The electrical supply to this aspirateur should be maintained in event of power cut—and, in principle, even after an emergency stop. Alternative power arrangements to cover maintenance will be arranged. No UPS cover is required as a 90s break is acceptable.
  3. In future, all non-machine room lighting and power sockets will be connected to the standard supply—i.e. no UPS coverage and  not supplied by diesel in event of power failure. As a consequence, some offices in corridors on the 1st and 2nd floor of B513 will lose their UPS protected sockets..
  4. Maintained lighting in the Machine room and Vault will be powered from the critical (200kW) UPS.
  5. As they are part of the physics load, the STK robots in B613 robots will be supplied via the 2MW UPS, not the 200kW UPS.
  6. The plan to move to a single AUG button/switch for the fire brigade (by the smoke detection control panel) with AULs for the machine rooms was reconfirmed. These systems must be arranged in such a way that the required tests can be undertaken without cutting power to computing equipment.

2.             Upgrading the existing machine room

  1. In response to a question from Anne, Tony outlined the plans, such as they are, for upgrading the existing machine room to met LHC requirements.
  1. This work has to be scheduled such that any equipment purchase from 2005 can be installed in a definitive location with the desired power connections. As the machine room cannot be emptied quickly the upgrade work will have to proceed in phases.
  2. Tony believes that the are in front of the operators box and the STK silo area can be cleared of equipment by mid-2003. Upgrade work in this area should therefore start in the second half of 2003, moving across tho the left hand side of the machine room in 2004.
  3. Jukka commented that we should reconsider the compartmentalisation between the machine room and office corridors and investigate the smoke extraction arrangements.

3.             Installation of an Additional Chiller

  1. Planning work for installation of an additional chiller is underway and should be completed in the next few weeks. Removal of the two steam humidifiers is due to start soon.
  2. Jukka confirmed that the installation of an additional chiller would be covered under existing contracts. A simple DAI is therefore sufficient and the installation could be completed within 6 weeks of order being placed.
  3. Tony was therefore keen to go ahead. However, Anne pointed out that it would be difficult to provide electrical power for a new chiller before the new substation is available. In principle, we could install the chiller but not power it up. This not sensible, though, as reception tests would be needed.

4.             AOB

  1. Next meeting: 27th June, 14:00