A new build starts from requirements; a retrofit starts from evidence
A new build can define a target state around owner project requirements, design criteria, capacity stages and future expansion. A retrofit must first prove current conditions, including actual loads, hidden dependencies, equipment health, configuration, space, airflow, pipework, alarms, networks and operating constraints. When drawings or asset records are unreliable, investigation and field verification become a critical work package.
- New-build risk centers on requirement drift, discipline coordination and late change.
- Retrofit risk centers on unidentified dependencies, service interruption, legacy compatibility and rollback failure.
- Both need a clear acceptance baseline, but the evidence used to establish it differs.
A new build optimizes the target state; a retrofit first protects live operations
A new build can coordinate fire, mechanical, electrical, controls and IT space together, then freeze design through mock-ups, factory tests and staged handover. A retrofit first decides what can remain, what must be replaced and what needs a temporary bypass, then checks outage windows, compatibility, spares, access and recovery. Similar replacement ratings do not prove equivalent interfaces, sequences or fault behavior.
- Retrofit procurement should include legacy-interface investigation, compatibility, temporary equipment and rollback materials.
- New-build procurement should also freeze versions, points, protocols and test ownership before site integration.
Retrofits depend more heavily on windows, isolation, temporary controls and rollback
New-build installation and prefunctional checks are normally completed before production workloads arrive, followed by staged energization and load. A retrofit operates around live services and needs an approved change, method, risk assessment, isolation boundary, supervision, temporary fire or cooling controls, communications and objective rollback triggers. Any reduction in protection should be approved by an authorized role and time-limited.
- A rollback plan should name triggers, decision authority, backup, expected duration, business impact and verification.
- A temporary measure needs capacity, monitoring, maintenance, removal criteria and ownership, not a generic coordination note.
A new build proves design intent; a retrofit also proves retained systems and recovery
New-build commissioning progresses from equipment through subsystems to integrated scenarios against owner requirements and design intent. A retrofit must also prove that retained systems were not impaired, existing configurations are controlled, business services recovered, temporary measures were removed and a new baseline exists. Both should hand over open issues, training, spares, thresholds, settings, points, record documents and conditions for retest.
- Tests should cover normal, fault, communication loss, degraded operation, recovery and operator response, not only power-up or movement.
- Performance or continuity claims require records under agreed conditions and cannot be inferred from installed quantities.
New-build and retrofit delivery comparison
Convert each comparison item into a plan, owner, approval point and acceptance record at project initiation.
| Delivery topic | New-build data center | Live-site retrofit | Common control |
|---|---|---|---|
| Starting point and baseline | Owner requirements, design criteria, capacity and future stages | Survey, assets, configuration, real load, health and hidden dependencies | Evidence sources, assumptions, open decisions and approved baseline |
| Design and interfaces | Coordinate space, capacity and new interfaces in a common model | Separate retained, replaced and temporarily bypassed systems and verify compatibility | Interface register, versions, points, command authority and fault behavior |
| Site execution | Install and inspect by area before production workload migration | Live environment, controlled windows, isolation, supervision and temporary protection | Method, risk assessment, hold points and approvals |
| Testing | Prove design intent from equipment through integrated scenarios | Prove new work, retained systems, service recovery and rollback viability | Scripts, expected results, timestamps, defects, retest and sign-off |
| Failure and rollback | Emphasize defect correction, phase isolation and readiness to enter service | Provide an executable stop, restore-legacy or temporary-system path | Objective triggers, decision authority, backup, duration and verification |
| Handover | Create the first asset, setting, record and operating baseline | Merge new and legacy records, update configuration and close temporary measures | Training, spares, open items, procedures and change controls |
New-build and retrofit conditions can coexist on one campus; select the delivery path for each system or area according to its actual live-service risk.
A common four-stage process with different controls
A common process supports governance, while inputs, approvals and rollback depth adapt to the new-build or retrofit condition.
- 01
Establish requirements or current-state baseline
Freeze owner requirements and design inputs for a new build; verify assets, configuration, load, health, dependencies and constraints for a retrofit.
Output: baseline, risk, assumptions, survey gaps and acceptance criteria - 02
Freeze design and execution controls
Coordinate systems, interfaces, procurement and tests; for retrofit also freeze windows, isolation, temporary measures and rollback.
Output: approved design, interface register, method and rollback plan - 03
Install, configure and test in layers
Check equipment and point mapping before subsystem, cross-system, fault, degraded-mode and recovery scenarios.
Output: checklists, test records, defects and closure evidence - 04
Enter service, verify and hand over
Confirm performance or recovery conditions, remove temporary measures, update record and configuration baselines, and train operations.
Output: signed records, baselines, procedures, training and open items
EVIDENCE
Yabowei rack-relocation material summary
The supplied Yabowei business material contains a cross-site physical-equipment and complete-rack relocation record. Under the confirmed anonymization scope, this page retains the service type while country, client, year, exact scale and peak remain withheld.
- Scope listed
- Cross-site physical-equipment relocation and complete-rack delivery
- Project type
- Migration and retrofit support for an existing environment
- Public scope
- Country, client, year, exact scale and peak withheld
The publicly verifiable scope is limited to the cross-site physical-equipment relocation and complete-rack delivery type in Yabowei material.
The supplied material does not identify the country, client, year, exact scale, contract scope, outage window, application dependencies, acceptance report, data-integrity result, executed rollback or public client permission.
Only the supplier-stated migration service type is verifiable from the material. It does not prove a specific scale, zero interruption, zero data loss, schedule acceptance or continuity improvement.
Sources: S9
Frequently asked questions
Why should a retrofit not rely on legacy drawings alone?
Legacy drawings may omit field changes, bypasses, firmware, point mappings, loads or equipment condition. Build a verified current-state baseline through survey, configuration export, testing and operator interviews.
Can a retrofit be completed without an outage window?
Not as a general rule. Segmentation, redundancy, temporary systems or live migration can reduce impact, but isolation, capacity, fail-safe behavior and rollback still need evidence and an authorized acceptance of residual risk.
Is integrated site testing needed after a factory acceptance test?
Yes. A factory test rarely reproduces the site network, time, addresses, load, physical interfaces, retained systems and operating team. Site tests verify real dependencies and end-to-end recovery.
What is the minimum handover evidence for a retrofit?
At minimum retain approved scope, record and configuration documents, interfaces and points, tests and defect closure, service-recovery confirmation, closure of temporary measures, training, spares, open items and retest conditions.
Standards, public technical and project-material sources
ASHRAE distinguishes commissioning for new and existing systems. ISO and NIST frame facility lifecycle, continuity and systems engineering. Project material supports only the facts it explicitly lists.
- S1 · ASHRAECommissioning standards and guidance
Public process framework covering commissioning standards for new and existing systems and owner requirements.
- S2 · ASHRAEAn Introduction to ASHRAE Existing Building Commissioning Process
Supports existing-facility assessment, operational objectives and continual improvement context.
- S3 · ISOISO/IEC 22237-1:2021 Data centre facilities and infrastructures
General framework for data center facility lifecycle, availability and management.
- S4 · NISTSP 800-160 Vol. 1 Rev. 1, Engineering Trustworthy Secure Systems
Supports lifecycle, interfaces, integration, verification, risk treatment and traceable engineering.
- S5 · NISTTesting Environments
Public background for conformance, integration and peer-to-peer test environments.
- S6 · ISOISO 22301:2019 Security and resilience — Business continuity management systems
General framework for business impact, response, recovery, exercises and continual improvement.
- S7 · SecuritonSupplied Securiton ASD product and project-reference records
Supports only the fire-detection products and deployment facts listed in the material.
- S8 · GRUNERSupplied valve and actuator product material
Supports manufacturer capabilities and selection fields, not site commissioning outcomes.
- S9 · Beijing Yabowei Technology Co., Ltd.Supplied business, credential and public solution material
Supports the listed cross-site physical-equipment and complete-rack relocation service type; country, client, year, exact scale and peak are withheld, and the material does not include independent acceptance evidence.
- S10 · SecuritonSupplied Securiton data center project-reference records
Supports the nine visible Securiton data center cases; only fields explicitly listed in the material are used.
This page is a delivery-process comparison, not a commissioning plan, fire or mechanical design, method statement, outage approval, risk assessment or business-continuity plan. Actual phases, editions, windows, temporary measures, tests and rollback must reflect facility condition, business criticality, manufacturer support and local requirements.