In a data center, MEP is the building: everything else exists to protect an uninterrupted supply of power and cooling to the IT load. On a data center the mechanical and electrical systems usually make up the majority of the capital budget. This guide is for developers, colocation operators and enterprise IT owners planning data center MEP in Saudi Arabia.
Why Data Center MEP Is Different From Commercial MEP
Commercial MEP is designed around people: comfort cooling, fresh air, lighting and life safety. Data center MEP is designed around equipment that runs continuously and cannot tolerate even milliseconds of power interruption. That changes almost every design decision:
- Load density: Data halls are designed in kilowatts per rack — conventionally 5–15 kW, and far higher for AI and high-performance computing racks.
- Redundancy: Every critical component is duplicated to a defined level so that a failure or a maintenance shutdown does not drop the IT load.
- Maintainability: On higher-resilience facilities, any component must be removable for maintenance without shutting down the IT equipment.
- Testing: Every failure scenario is simulated under load before go-live.
The core principles of MEP contracting in Saudi Arabia still apply. Data centers add a resilience layer on top.
Tiers and Redundancy: N, N+1 and 2N
The most widely used framework for describing data center resilience is the Uptime Institute Tier Classification System, which defines four levels:
- Tier I — Basic Capacity: Dedicated UPS, cooling and generator, with no redundancy. Maintenance requires a shutdown.
- Tier II — Redundant Capacity Components: Spare UPS modules, chillers and similar on a single distribution path.
- Tier III — Concurrently Maintainable: Every component and distribution path can be taken out of service for planned maintenance without affecting the IT load. The most common colocation and enterprise target.
- Tier IV — Fault Tolerant: The facility can withstand any single unplanned failure without impact on the IT load, with compartmentalised systems and continuous cooling.
Tier certification is issued by Uptime Institute against design documents, the constructed facility and, separately, operations. A design that is "Tier III ready" is not the same as a certified Tier III facility. ANSI/TIA-942 is an alternative rating framework some owners specify.
Redundancy within those tiers is described using N notation, where N is the capacity needed to carry the full design load:
| Configuration | What It Means | Example (load needs 4 UPS modules) | Typical Use |
|---|---|---|---|
| N | Exactly enough capacity, no spare | 4 modules | Tier I, small server rooms |
| N+1 | One spare unit beyond the required capacity | 5 modules | Tier II–III component redundancy, cooling plant |
| N+2 | Two spare units, allowing maintenance and a failure at the same time | 6 modules | Larger plants where one spare is not enough |
| 2N | Two fully independent systems, each able to carry the whole load | 4 + 4 modules on separate A and B paths | Tier IV and premium Tier III electrical systems |
| 2(N+1) | Two independent systems, each with its own spare | 5 + 5 modules | Highest-resilience critical facilities |
The tier target and redundancy configuration should be fixed at concept stage. Changing from N+1 to 2N later is effectively a redesign.
The Power Chain
Utility Supply and Medium-Voltage Switchgear
Supply comes from the Saudi Electricity Company (SEC), typically at medium voltage for an enterprise or colocation facility, or via a dedicated high-voltage substation for a large campus. The SEC application — capacity, connection voltage, substation land — is a critical-path item from the first week. Dual feeds are requested where the network allows, but resilience is ultimately designed around on-site generation.
Transformers and LV Distribution
MV/LV transformers feed main switchboards serving UPS systems and mechanical plant. In Riyadh, transformers and switchrooms must be selected, ventilated and derated for high ambient temperatures.
Generators and Fuel Storage
Diesel generators are the long-duration backup. Data center generators are specified for continuous or prime duty rather than the standby rating common on commercial buildings, and are sized for high-ambient derating — an engine rated at 25°C will not deliver its nameplate output at 50°C. Bulk tanks, day tanks, fuel polishing and refuelling access must be planned into the site layout; many operators specify 24 to 72 hours of on-site fuel at full load.
UPS, PDUs and Busway
The UPS bridges the seconds between utility failure and generators taking load, and conditions power to the IT equipment. Static double-conversion UPS is standard, increasingly with lithium-ion batteries. Downstream, power reaches the racks through power distribution units (PDUs) or overhead busway with tap-off units, which lets racks be added without new cabling. Dual-corded IT equipment is fed from separate A and B paths; single-corded loads are served through static transfer switches or rack-level transfer switches.
Earthing, Bonding and Lightning Protection
Data centers need a low-impedance earthing system, a mesh bonding network in the data hall connecting racks, containment and raised floor pedestals, and a lightning protection system with surge protection at each level of distribution. Earthing is tested and documented before energisation.
Cooling in Riyadh's Climate
Design Conditions
Riyadh summers regularly exceed 46°C, and data center heat rejection plant is commonly selected for ambient design temperatures in the 46–50°C range. Equipment rated at 35°C can lose a significant share of capacity at 50°C, so selections must be checked at the actual design condition. Dust and sand add a second challenge: condenser coils foul, filters load quickly and outdoor plant needs coil protection and a realistic cleaning regime.
System Options
- Chilled water: Air-cooled or water-cooled chillers serving computer room air handlers (CRAH units) or fan walls. The most scalable option for larger facilities, and the natural platform for adding liquid cooling later.
- Direct expansion (DX): Computer room air conditioning (CRAC) units with their own refrigerant circuits. Simpler and modular, suited to smaller facilities and edge sites, but less efficient at scale in high ambients.
- In-row and close-coupled cooling: Cooling units placed between racks, shortening the airflow path and handling higher densities than room-level cooling.
Whichever system is chosen, hot aisle or cold aisle containment is now standard practice. It prevents hot and cold air mixing and allows higher supply temperatures within the ASHRAE thermal guidelines for IT equipment.
Free Cooling, Water and Efficiency
In Riyadh, "free cooling" from outside air or dry coolers is limited to winter months and cooler nights, so economisers need careful analysis. Water is the other constraint: evaporative and water-cooled systems are efficient but consume water that is scarce and often desalinated. Designers balance Power Usage Effectiveness (PUE) against Water Usage Effectiveness (WUE) — both metrics developed by The Green Grid — and many Saudi projects favour air-cooled or closed-loop systems with minimal water consumption, accepting some energy penalty in exchange.
Liquid Cooling for AI and High-Density Racks
GPU-based AI and high-performance computing racks can draw several tens of kilowatts, with the latest platforms exceeding 100 kW per rack — beyond what air can remove economically. Options include rear-door heat exchangers, direct-to-chip cooling with coolant distribution units (CDUs), and immersion cooling. For new builds, the practical approach is to provide chilled water capacity, floor loading, pipework routes and electrical headroom so halls can convert to liquid cooling as demand requires.
Fire Detection and Suppression
Fire protection in a data center must detect a problem early and suppress it without destroying the equipment it protects. Typical provisions include:
- Very early warning aspirating smoke detection (VESDA or equivalent): Detects smoke at very low concentrations, giving operators time to investigate before a fire develops.
- Clean agent or inert gas suppression: Systems designed to NFPA 2001 for data halls and electrical rooms, which must be sealed and integrity-tested.
- Pre-action sprinklers: Keep pipework dry until detection confirms a fire.
- Battery room protection: Lithium-ion battery installations need specific attention to detection, ventilation and separation.
All fire systems require Saudi Civil Defence approval, and gas suppression must be coordinated early with the architecture for room sealing and pressure relief.
Monitoring: BMS, EPMS and DCIM
A data center typically runs three layers of monitoring. The building management system (BMS) controls and monitors mechanical plant. An electrical power monitoring system (EPMS) records power quality, breaker status and energy at every level of the electrical chain, with high-resolution event logging for fault analysis. Data center infrastructure management (DCIM) software sits above both, giving operators rack-level views of power, temperature and capacity. Points lists and alarm philosophy should be defined during design, not retrofitted.
Testing and Commissioning: Levels 1 to 5
Data center commissioning is usually structured in five levels:
- Level 1 — Factory witness testing: Major equipment such as generators, UPS and switchgear tested at the manufacturer before shipment.
- Level 2 — Site acceptance and installation checks: Delivered equipment inspected and installation verified against approved drawings.
- Level 3 — Pre-functional tests and start-up: Individual equipment energised and started under manufacturer supervision.
- Level 4 — Functional performance testing: Each system tested across its operating modes, often using load banks to simulate IT load.
- Level 5 — Integrated systems testing (IST): The whole facility tested together under simulated full load, including utility failure, generator start, UPS transfer and cooling failure scenarios.
IST is where design and installation weaknesses surface, and programmes should allow realistic time for it — plus time to fix what it finds. Our guide to MEP testing and commissioning in Riyadh covers the commissioning process in more detail.
Delivery Models: Turnkey and Phased Build-Out
Many data center owners in Saudi Arabia procure on an EPC or turnkey basis, with a single contractor responsible for design coordination, procurement, construction and commissioning of the building and its MEP systems. This puts interface risk between electrical, mechanical, fire and controls packages with one party.
Phased build-out is also common. The shell, utility connection and central infrastructure (MV intake, generator yard, chilled water backbone) are built for the full campus, while data halls and their UPS and cooling capacity are fitted out as customer demand materialises. Later phases are built next to live IT load, which demands careful isolation planning and strict change control.
Long-lead equipment is the main programme risk. Generators, MV switchgear, transformers, UPS and chillers can carry lead times of many months, stretched further by global hyperscale and AI demand, so early procurement is often essential.
The Saudi Context: Vision 2030 and the AI Build-Out
Digital infrastructure is a central part of Vision 2030. Global cloud providers have announced or opened cloud regions in the Kingdom, data residency expectations encourage local hosting of government and regulated data, and national AI ambitions are driving demand for high-density GPU capacity. For owners, that means competition for experienced MEP labour, commissioning specialists and equipment manufacturing slots. Dar Anan supports data center construction in Saudi Arabia alongside our wider industrial construction work.
Frequently Asked Questions
What tier should a data center in Saudi Arabia be designed to?
Most commercial colocation and enterprise facilities target Tier III (concurrently maintainable). Tier IV is reserved for applications where any downtime is unacceptable and the additional capital cost is justified. The right answer depends on service-level commitments and the cost of downtime.
What is the difference between N+1 and 2N?
N+1 adds one spare unit to the capacity needed for the load, protecting against a single component failure. 2N provides two completely independent systems, each able to carry the full load, which also protects against a failure in the distribution path itself.
Can data centers in Riyadh use free cooling?
To a limited extent. Economisers can contribute during winter months and cooler nights, but Riyadh's long, hot summers mean mechanical cooling carries the load for most of the year.
Do I need liquid cooling?
Not for conventional enterprise or colocation densities. For AI and GPU racks, liquid cooling is usually necessary.
What should I look for in a data center MEP contractor?
Critical-facility experience, in-house electrical and mechanical capability, a Level 1–5 commissioning methodology, a long-lead procurement strategy and experience working next to live systems. Our guide on how to choose an MEP contractor in Riyadh sets out the wider selection criteria.
Dar Anan has delivered construction, fit-out and MEP contracting across Saudi Arabia since 2002. If you are planning a server room, an enterprise facility or a phased colocation build, see our data center construction services or contact our team to discuss power, cooling and resilience requirements for your project.