Global Switch (Property) Singapore Pte Ltd v Arup Singapore Pte Ltd

• Jurisdiction: Singapore
• Judge: Quentin Loh J
• Judgment Date: 10 May 2019
• Neutral Citation: [2019] SGHC 122
• Court: High Court (Singapore)
• Docket Number: Suit No 1147 of 2014
• Published date: 06 March 2020
• Year: 2019
• Hearing Date: 10 March 2017,28 February 2017,15 December 2017,16 March 2017,02 March 2017,27 February 2017,17 August 2017,14 March 2017,15 August 2017,16 August 2017,17 March 2017,01 March 2017,15 March 2017,18 August 2017,22 August 2017,03 March 2017,06 March 2017,14 August 2017,21 February 2017,22 February 2017
• Plaintiff Counsel: Ho Chien Mien, Gregory Leong, Ng Chee Jian, Tan Li-Jie and Xu Xuekun (Allen & Gledhill LLP)
• Defendant Counsel: Daniel Chia Hsiung Wen, Tan Gim Hai Adrian, Amarjit Kaur, Annette Liu Jia Ying, Ker Yanguang, and Yeoh Jean Wern (Morgan Lewis Stamford LLC)
• Subject Matter: Building and Construction Law,Building and construction contracts,Design contracts,Mechanical and electrical design,Construction torts,Negligence,Consultant,Terms,Exclusion clauses,Incorporation,Implied terms,Fitness for purpose warranty,Scope of works,Variation
• Citation: [2019] SGHC 122

Quentin Loh J: Introduction

The plaintiff, Global Switch (Property) Singapore Pte Ltd (“GSS”), is part of the Global Switch group of companies (“Global Switch”) which owns and operates data centres in London, Paris, Amsterdam, Frankfurt, Madrid, Sydney and Singapore.¹ The defendant, Arup Singapore Pte Ltd (“Arup”), is a firm of consultant engineers who provide, inter alia, design engineering consultancy services for mechanical and electrical (“M&E”) engineering systems. GSS alleges, and Arup accepts, that Arup stated that it is experienced in providing data centre solutions, technically sophisticated and employs technically qualified people.

This suit arises out of GSS’s project (“the Project”), embarked on sometime around 2008, to construct an extension (“the Extension”) to its existing data centre facility (“the Existing Facility”) at No 2, Tai Seng Avenue, Singapore. The Existing Facility originally comprised a seven-story building and began operations in 2001.² Space was leased out to customers who required premises with a reliable and, importantly, uninterrupted electricity supply, specified cooling capacity and other services to suit the customers’ specified IT requirements for their sensitive IT equipment and computer banks (“IT equipment”). GSS’s customers included several international banks, Tier 1 telecommunications companies, as well as web-based service providers such as Microsoft.³ The Extension for new data centre space would comprise five storeys, from levels 3 to 7 (“L3 to L7”), which would extend out from the rear of the Existing Facility. Each of these storeys at L3 to L7 would have 800 m2, thereby resulting in a total additional area of 4,000 m2.

GSS appointed Arup as its M&E consultant for the Extension for an agreed fee of $595,000. Disputes arose and GSS sued Arup for damages in excess of $23.8 million, general damages, interest and costs. Arup counterclaimed for unpaid fees.

I find that GSS is only entitled to nominal damages of $1,000 for Arup’s breach of its obligations regarding the provision of additional cooling. I allow Arup’s counterclaims for a total of $71,347.60 under SCN003(a), SCN003(b), SCN005, SCN006 and SCN007.

Background Technical concepts

It is necessary at this juncture to set out some background and technical terms in relation to data centres. Data centres must provide an uninterrupted electricity supply for the proper functioning of their customers’ IT equipment. It would be deleterious for their customers if electricity supply were to cease even for very short periods. Disruption is discussed in the context of milliseconds. Equally significant is the cooling capacity of the data space, which cannot fail for any sustained period of time (since failure would cause the sensitive IT equipment to overheat).

Data centres provide uninterrupted electricity supply and cooling through alternative back-up sources for no break power (“NBP”) and short break power (“SBP”). The essential difference is that SBP ordinarily experiences a short interruption of electricity of a few seconds during faults or maintenance.⁴ Because of this, mechanical equipment (such as chillers and Computer Room Air Conditioning units (“CRACs”), collectively known as the “Mechanical Load”) is typically supplied with SBP because it can tolerate short power disruptions, unlike IT equipment (also called the “IT Load”) that cannot. NBP is thus supplied for the IT Load.

Because an uninterrupted electricity supply is so crucial to the powering and cooling of their customers’ equipment, data centres install back-up electrical systems, in other words standby items of equipment that seamlessly “kick-in” and supply electricity the moment there is an interruption or anomaly in electricity supply coming off the power grid. Some of these back-up power systems can, to varying extents, “clean up” surges or drops in voltage, which occasionally occur in relation to the electricity coming off the power grid. Such anomalies are undesirable as they can interrupt the smooth functioning of IT equipment. I should mention that the frequency of alternating current from the grid (measured in Hertz (“Hz”)) can also fluctuate, although I am told that does not affect IT equipment.⁵

The two relevant back-up systems in this case, each with their relative advantages and disadvantages, are the Static Uninterruptible Power Supply (“SUPS”); and the Diesel Rotary Uninterruptible Power Supply (“DRUPS”).

A SUPS system essentially comprises a battery and a generator. If electricity supply is interrupted, the SUPS units “kick-in” and supply autonomous power to support the critical IT Load for around five to ten minutes. This buys time for the emergency generator to come online. SUPS systems can also condition the voltage input through a wide range of voltage deviation and act as a frequency converter.⁶ Some disadvantages of SUPS systems include their weight and loading on floor as well as the space taken up to locate the batteries; the batteries also degrade and require regular replacement, in GSS’s case, typically every two years or so.⁷

A DRUPS system, considered the more modern equipment, is also designed to provide uninterrupted, conditioned power to the critical IT Load if electrical supply is interrupted.⁸ It does so without the use of batteries. Instead, there is one compact inline assembly, which can be containerised, comprising a kinetic energy accumulator (akin to a flywheel) which is coupled through induction to an inner rotor which drives the alternator, all of which are kept constantly spinning. That assembly is also connected to the diesel engine. The kinetic energy accumulator, in this case the outer rotor, spins continuously at 1,500 rpm. If there is any interruption or anomaly in the electrical supply, the diesel engine is activated, but meanwhile the interruption is momentarily taken up by the kinetic energy in the constantly spinning outer rotor which continues to drive the inner rotor and alternator thereby supplying the electricity (this being analogous to the battery function in the SUPS system).⁹ When the diesel engine comes up to speed, it then engages and drives the alternator.

The DRUPS unit is connected to the power mains. Under normal circumstances the IT equipment is powered by the electricity from the mains that flows through the DRUPS unit, which act as a power conditioner to some extent to improve the quality of the electrical supply to the IT equipment.¹⁰ DRUPS units have power conditioning capability and can supply some frequency conditioning, but cannot condition output frequency in the manner that a double converter can.¹¹ DRUPS units can supply both NBP as well as SBP.¹² SUPS units cannot.¹³

The backup system in GSS’s Existing Facility’s was a SUPS system. There were initially (from October 2001) 6 MTU diesel generator sets with 2000 kVA alternators and batteries in the Existing Facility as the take-up of space by customers was gradual. As at August 2007,¹⁴ only 50% of the Existing Facility’s space was occupied, however with Google and Merrill Lynch’s commitment to lease space by the end of 2007, thereby bringing the occupancy level to about 84%, GSS increased its capacity with the addition of another 6 MTU diesel generator sets (and batteries).¹⁵ A DRUPS system was to be used for the Extension for both NBP and SBP,¹⁶ given that it could deliver SBP concurrently.

Next, I come to the concept of redundancy. Essentially this means providing back-up to the back-up, because an uninterrupted power supply is so critical to data centres. Data centres vary in terms of the level of redundancy provided. For instance: In an “N+1” configuration, “N” is the number of items of standby equipment (eg, generators or batteries combined with generators) required for emergency use when there is an interruption in electrical supply and “+1” represents an additional standby item of equipment for the redundancy capacity. If any of the “N” units fail, there is an additional unit to take its place. The “+1” is also useful when one of the “N” units have to be taken offline for maintenance or repairs. In an “N+N” configuration, each generator is backed up by another generator, thereby ensuring that there is an equal number of backup generators available to take over the entire IT Load should all the main generators fail.¹⁷ The higher the specifications, power requirements, capacity and reliability, the higher the capital expenditure to provide the same. A data centre operator would therefore have to make a commercial decision as to what segment of the market it would like to occupy, considering also the kind of customers in the market. If the market consisted mainly of tenants who required medium to lower-end specifications, then it would not make economic sense to incur the higher capital expenditure because such tenants would not be prepared to pay the higher rent that high-end customers would be prepared to.

GSS describes the SUPS units in the Existing Facility as being configured in an “enhanced” “N+1” arrangement,¹⁸ where each floor receives an “N+N” power feed from separate generator switchboards, under the criteria of no two generators simultaneously failing, which gives 8 MW of standby capacity.¹⁹

Finally, brief mention should be made of two different kinds of cooling systems. IT equipment in a data centre generates a significant amount of heat and therefore requires adequate cooling to prevent overheating and failure.²⁰ It may therefore be cooled by: A supply air system where the room will be divided into hot and cold aisles.²¹ Cold air from the CRACs will be pushed into the cold aisles through a pressurised floor void and then over the IT equipment. After the cold air passes over the IT equipment, it absorbs the IT equipment’s heat and turns into hot air. This hot air is then ejected into a hot aisle before being extracted from the aisle by the...