A 1995-built corporate headquarters in central Delhi ran ductable splits on every floor for three decades. By 2024 the indoor units were unobtainable for spare parts, refrigerant R-22 had crossed BEE phase-out, and the FM team was facing weekly breakdowns. VRF retrofit was the only realistic path. This article walks through the project: scope, refrigerant + safety considerations, electrical infrastructure findings, and the operating savings 9 months in.
The site
- Connaught Place, Delhi
- 9-storey corporate HQ, ~7,800 m² occupied
- Original 1995 system: 18 × ductable split units (8-12 TR each), R-22 refrigerant, age 27-30 years at retrofit
- Cooling load (re-calculated 2024): peak 320 TR, average 180 TR
- Operating issues: 2-3 unit failures per quarter, weekly leak-tracking, no temperature control granularity per zone (one DB per floor controlled 4-6 zones)
Why VRF (and not central chilled water)
Three retrofit paths were costed:
| Path | Capex | Opex (yr-1) | Disruption | Notes |
|---|---|---|---|---|
| A — Like-for-like ductable splits (R-32) | ₹1.4 cr | ₹78 lakh/yr | Low | No zone control upgrade; same complaints continue |
| B — VRF, 5-pipe routing | ₹2.6 cr | ₹52 lakh/yr | Medium | Multi-zone control; heat recovery (BS box) option |
| C — Chilled water, new plant + AHU | ₹4.1 cr | ₹46 lakh/yr | High | Plant room space + structural challenges in 1995 building |
Path B selected: heritage building exterior couldn’t accommodate cooling tower for Path C; Path A solved nothing structurally.
The retrofit scope
- 4 outdoor units, 8 + 12 + 14 + 18 HP, total 52 HP, R-32 refrigerant
- 64 indoor units across 9 floors: mix of ceiling-cassette (40), ductable concealed (16), wall-mount (8)
- 3-pipe heat recovery system (key feature for Delhi: cooling in summer + heating in winter from same system)
- Branch selectors at each floor riser
- 2 dedicated outdoor air units (DOAS) for fresh air (NBC Pt 8 minimum 5 L/s/person)
- Liquid + suction + discharge piping totalling ~890 m, routed through existing ductable space + new horizontal trunks
- Centralised controller + BMS gateway with BACnet
Critical investigations before retrofit
Electrical: original 1995 building MCC delivered 320 kVA total. VRF connected load: 285 kVA. Margin 35 kVA — too tight for future expansion. Upgraded MCC + main panel as part of scope (₹38 lakh additional).
Refrigerant safety (ASHRAE 15 + ASHRAE 34): R-32 is A2L (mildly flammable). Total system charge: 142 kg across 4 outdoor units. Smallest protected space (a 22 m² boardroom on level 6) volume: 60 m³. R-32 RCL: 0.30 kg/m³ × 60 m³ = 18 kg maximum charge in that space. Worst-case leak from any single indoor unit serving that space: max 2.4 kg. Safe by an order of magnitude — but we installed refrigerant leak detection (3 sensors per floor) + auto-shutoff valves per OEM recommendation anyway.
Refrigerant pipe sizing: longest run was 110 m vertical + 35 m horizontal from outdoor (roof) to ground floor reception. Standard OEM tables work to ~120 m equivalent length. We ran a ΔP check (per MEPVAULT VRF article) — liquid line ΔP at 28 kPa, suction at 78 kPa. Both within limits but tight. Built in 15 % oversize on suction line.
Heating-load check: Delhi design winter +5 °C OAT. VRF heat-pump capacity at +5 °C is 78 % of rated. We sized system to 130 % of peak heating load (200 kW design at +5 °C) — total VRF heating capacity 260 kW available, comfortable margin. Article on Heating Load for North India walks the calc.
The retrofit execution
- 11-week project, weekend + after-hours work to minimize disruption
- Floor-by-floor cutover: one floor offline for 3-4 days, portable AC during cutover, ductable split removed at night, VRF indoor + ductwork installed next day, commissioning + balancing day 3-4
- Refrigerant pipe brazing: nitrogen purge + N2 hold test 30 bar × 24 hours + vacuum to 500 microns + held 12 hours
- Total project duration: 14 weeks from kick-off to full commissioning
9-month operating outcome
| Metric | Pre (2023, ductable splits) | Post (2024, VRF) | Δ |
|---|---|---|---|
| Annual cooling energy | 940,000 kWh | 580,000 kWh | -38 % |
| Annual heating energy (electric resistance previously) | 110,000 kWh | 22,000 kWh | -80 % (heat pump replacing resistance) |
| Equipment failures / quarter | 3 | 0 | -100 % |
| Per-zone temp control | Floor-level | Indoor-unit-level (64 zones) | 7× granularity |
| Tenant comfort complaints / month | 9 | 1 | -89 % |
| Refrigerant leak events | weekly | 0 (9 months) | n/a |
Total annual savings: ₹78 – ₹52 = ₹26 lakh/yr direct + reduced complaint resolution overhead. Simple payback on the VRF + electrical upgrade: ~11 years.
The intangibles dominate:
- 64-zone control transformed how the FM team responded to comfort calls
- Refrigerant leak detection added a future-proofing layer (Kigali compliance, R-32 phaseout TBD post-2030 also covered)
- Heat recovery (BS box) reclaimed ~ 8 % of cooling kWh as ‘free’ winter heating
Heritage building constraints
Two constraints that drove the design:
1. Façade — no condenser fans visible. Outdoor units installed on roof at corners with acoustic screening. Required structural verification (1995 RCC roof could carry 2.4 tonnes additional concentrated load on each corner — confirmed by IS 875 Pt 5 review).
2. Vertical shaft routing. Original ductable splits used a single vertical shaft on each floor’s NW corner. New VRF piping fit in the same shaft, but the shaft needed fire-rated re-treatment (2-hour vertical-shaft fire-stopping at every floor) — IS 9572 + NBC Pt 4 compliance.
Heritage projects always carry these hidden costs. Budget 10-15 % above visible scope.
From the Field — Engineer’s Notebook
The single most under-appreciated decision in this retrofit was choosing 3-pipe heat-recovery over 2-pipe heat-pump. Capex premium ₹22 lakh. But Delhi has a transitional shoulder season (mid-October, mid-March) when north zones need heating + south zones need cooling simultaneously. With 2-pipe, every indoor unit on a single refrigerant loop runs in one mode at a time — meaning either cool the north + waste heat, or heat the south + chill north. With 3-pipe, the BS box shifts heat from cooling zones to heating zones in real-time. We measured 12-14 % of cooling load reclaimed as heating during transitional weeks. Payback on the 3-pipe premium: 4 years. Always model the shoulder-season operation, not just summer + winter peak.
5 takeaways
1. VRF retrofit needs electrical capacity audit first. Old buildings often have inadequate MCC + main; budget the upgrade.
2. Refrigerant safety (ASHRAE 15 + 34) governs space-wise charge. R-32 A2L means leak detection + auto-shutoff for any tight space.
3. Verify OEM table ΔP for long runs. > 100 m equivalent triggers full pressure-drop check; don’t trust the chart blindly.
4. 3-pipe heat recovery worth it where transitional season is real (Delhi, Pune, Bangalore Oct + Mar). 4-5 year payback typical.
5. Heritage / occupied retrofit needs 11-14 week phased schedule. Less = disruption to operations + commissioning shortcuts that cost later.
Designer’s checklist
- [ ] Cooling + heating loads re-computed at design winter + summer (not legacy 1990s assumption)
- [ ] Electrical capacity audited; main panel + MCC sized for new VRF total connected load
- [ ] Refrigerant safety per ASHRAE 15: total charge vs RCL × space volume
- [ ] Leak detection + auto-shutoff for tight spaces
- [ ] Refrigerant pipe ΔP verified for long runs (> 100 m equivalent)
- [ ] 3-pipe vs 2-pipe decision based on transitional-season operation
- [ ] Fire-rated re-treatment of vertical shaft per IS 9572 / NBC Pt 4
- [ ] DOAS for fresh air per ASHRAE 62.1 / NBC Pt 8
- [ ] Heat-recovery zone groupings designed for transitional-season opportunity
- [ ] Nitrogen purge + 24-hr pressure hold + 12-hr vacuum-hold protocol
- [ ] BMS + centralised controller integration (BACnet preferred)
- [ ] BEE 5-star outdoor unit (subject to model availability)
Pairs with: VRF Refrigerant Pipe Sizing, Heating Load for North India