11.4 Data Center Assessments: Optimization Studies, Current-State Evaluation, and the Findings Report

11.4 Data Center Assessments: Optimization Studies, Current-State Evaluation, and the Findings Report

The Data Center Operations and Maintenance Assessment area (10% of the exam) contains 13 task statements, more than any other area. They read like a checklist for a single engagement: a client asks the DCDC to evaluate an existing data center and recommend improvements. Sections 11.1 to 12.3 teach the operational topics themselves. This section shows how they come together in an assessment.

---

The 13 Tasks as One Workflow

---

Step 1: Objectives and Data Collection

Begin by agreeing the client's goals. Typical goals are adding capacity without a new building, cutting energy cost, confirming resilience after years of changes, planning a technology refresh, or deciding whether to consolidate into colocation or cloud. The goals decide how deep each part of the assessment goes.

Then request documents before the site visit: record drawings and single-line diagrams, original commissioning reports, maintenance logs and incident reports, 12 to 36 months of utility bills, BMS and DCIM trend data, the asset register with installation dates, SOPs, MOPs and EOPs, security policies, and the disaster recovery plan.

Step 2: Site Walkdown, Interviews, and Measurements

Compare the documents with what is actually installed. Interview the operations staff, who usually know where the weak points are. Take measurements where data is missing or doubtful: UPS output and utility input power, rack inlet temperatures, airflow and pressure, and infrared scans of electrical connections.

Step 3: Evaluate Each Discipline

• System state and resilience: Trace every path from utility to rack and from chiller to rack. Years of changes often add hidden single points of failure: a single fuel transfer pump serving all day tanks, a common controls panel for "redundant" chillers, or single-corded equipment on an A-only circuit. A facility designed as Class F3 may no longer be concurrently maintainable.
• Efficiency: Use measured data. A PUE should cover 12 months and state its measurement category (Section 4.3). Look for common opportunities: raising supply air temperature within the ASHRAE recommended range, containment, blanking panels, variable-speed fans and pumps, economizer hours, and UPS operating modes.
• Capacity: Separate installed, available, and stranded capacity in space, power, cooling, and network (Section 12.1).
• Aging and refresh: Compare each asset's age with its expected service life (Section 12.2) and its manufacturer support status. Refrigerants matter too: R-22 production and import ended in the United States in 2020, which makes old R-22 equipment costly to keep running.
• Utilities: Review rate structures, demand charges, and power factor penalties; utility reliability history; water availability and cost; and carrier diversity (Section 10.5).
• Physical security: Assess electronic systems, architectural security, and vehicle barriers against the threat assessment (Chapters 13 and 14).
• Operational performance: Check staffing levels and shift coverage, training, change management, preventive maintenance completion rates and backlog, and whether procedures are current.
• Suitability against IT requirements: Compare the facility with the IT roadmap: rack densities, liquid cooling plans, network and carrier requirements, and whether the disaster recovery plan's RTO and RPO are achievable.
• Documentation: Check that historical commissioning reports, logs, and DCIM records match the field. Inaccurate records are a finding in their own right.
• Safety: Confirm a current lockout/tagout program, an arc flash risk assessment reviewed at least every five years as NFPA 70E requires, current equipment labels, PPE, and training records.

---

Step 4: The Optimization Study

An optimization study compares alternatives against the client's goal. Define the objective (for example, "maximize usable IT capacity within the existing utility service" or "cut energy cost with payback under three years"), list the constraints (budget, downtime windows, availability class), and evaluate each option's cost, benefit, risk, and disruption.

Worked Example: Airflow and Setpoint Optimization

A 2 MW (IT) data hall has a measured annual PUE of 1.8. Containment, blanking panels, raised supply air temperature, and fan speed control are estimated to bring PUE to 1.55 for a cost of 900,000 USD. Electricity costs 0.10 USD/kWh.

$\text{Energy saved} = 2{,}000\text{ kW} \times (1.80 - 1.55) \times 8{,}760\text{ h} = 4{,}380{,}000\text{ kWh/year}$

$\text{Annual saving} = 4{,}380{,}000 \times 0.10 = 438{,}000\text{ USD}$

$\text{Simple payback} = \frac{900{,}000}{438{,}000} \approx 2.1\text{ years}$

Over 10 years at an 8% discount rate, the present value of the savings is $438{,}000 \times 6.7101 \approx 2{,}939{,}000$ USD, so the net present value is about 2.04 million USD. The study should also note the operational effects: higher supply temperatures reduce the time available to respond to a cooling failure, so ride-through and alarm settings must be reviewed with the change.

---

Step 5: The Findings Report

The report turns the evaluation into decisions. A practical structure:

1. Executive summary with the most serious risks and a one-page roadmap.
2. Scope, objectives, and method, including documents reviewed and measurements taken.
3. Findings by discipline, each with evidence (photos, data, references), a severity rating (critical, high, medium, low), and the risk it creates.
4. Recommendations with order-of-magnitude or budget cost estimates, expected benefit, payback where relevant, and the downtime or risk involved in implementing them.
5. Prioritized roadmap grouping work into immediate safety and resilience fixes, short-term improvements, and long-term refresh or expansion.
6. Appendices: data, calculations, single-line markups, and photographs.

---

Common Exam Traps & DCDC Pitfalls

• Trusting record drawings: Field verification often finds changes that add single points of failure.
• Spot PUE readings: Efficiency findings need measured data over a full year and a stated measurement method.
• Recommendations without costs: A client cannot fund a list of problems; give estimates, benefits, and priorities.
• Ignoring operations: Staffing, training, procedures, and maintenance backlog cause many outages and belong in the assessment.
• Forgetting safety: An out-of-date arc flash study (older than five years) or missing LOTO procedures is a priority finding.