A constant-primary / variable-secondary (CP/VS) chilled-water plant was the Indian commercial default for two decades. ECBC 2017 + ASHRAE 90.1 now reward variable-primary flow (VPF) heavily — and the operating-data case for VPF has only strengthened. This article walks through a real 800 TR retrofit on a Pune office campus: scope, cost, what went wrong during commissioning, and twelve months of pump-energy savings.
The site
- Hinjawadi Phase 2, Pune — 22-acre IT campus (single tenant)
- 4 office blocks, 36,000 m² total conditioned area
- Chiller plant: 4 × 200 TR water-cooled screw chillers (2 × 250 TR design + 2 × 200 TR added for expansion)
- Original (2014) plant: constant-primary + variable-secondary, 4 + 4 pumps, decoupler bypass
- Annual chiller-plant energy: ~3.42 million kWh/yr; pump share ~22 %
The driver for retrofit
By 2023, the campus operations team had data showing:
- Plant load profile: 10-60 % of design TR for 80 % of operating hours
- Primary pumps running at full speed 24/7 (Indian commercial CP/VS standard)
- Decoupler bypass observed in BMS to be flowing wrong-way for 8-10 hours/day (low-load condition)
- ΔT degradation: design 6 °C / 12 °C ΔT was 4.2 °C actual at low-load
The decoupler bypass flowing wrong-way (a common CP/VS symptom at low load) was producing artificial chiller loading, killing part-load COP, and burning unnecessary primary pump energy.
VPF retrofit proposal: eliminate primary-secondary decoupling, run single pump set with VSD, modulate chiller flow with load via 2-way control valves at the AHU coils.
The retrofit scope
| Component | Action |
|---|---|
| Chiller evaporator side | Verify minimum-flow tolerance (2.3 m/s minimum velocity per AHRI 550/590) |
| Primary pumps | Retire 4 primary pumps |
| Secondary pumps | Become “the pump set”; add VSD if not already; controlled on plant ΔP |
| Decoupler bypass | Closed permanently (valve locked) |
| Bypass valve (chiller minimum flow) | New automatic bypass installed at chiller manifold to maintain min flow per chiller |
| 3-way coil valves | Converted to 2-way (90 % of coils were 3-way originally) |
| BMS control logic | Rewritten: chiller staging on flow + ΔT + load; pump on plant ΔP at remote AHU |
| Differential pressure sensors | Added at 3 remote AHU locations to provide plant ΔP setpoint |
| Chiller minimum flow protection | Direct evap flow sensor + variable-bypass to guarantee min |
Total retrofit capex: ₹84 lakh (~ ₹105/TR including pumps + valves + controls + ducting reroute).
What went wrong during commissioning
Two issues surfaced in the first month:
Issue 1 — Chiller flow alarm during chiller staging. When the BMS staged on a second chiller, the flow split caused the first chiller’s evap flow to dip below minimum for 90-120 seconds before the bypass valve opened sufficiently. Chiller trip alarms fired daily.
Fix: BMS logic changed to open bypass valve FIRST (anticipating staging) THEN command second chiller; pump speed boosted +5 % during transition. Stable after re-tune.
Issue 2 — Low ΔT at part load. With 2-way valves and modulating pump, ΔT at the plant should improve. Instead it dropped from 4.2 to 3.8 °C in the first six weeks.
Investigation: half the AHU coils’ 2-way valves had been converted but the BMS valve characterization curves were still set for 3-way operation (linear). With 2-way, the curve should be equal-percentage. Re-characterization at the AHU controllers fixed it; ΔT climbed to 5.4 °C within 30 days, hitting design within 60.
These are not rare; both happen on most VPF retrofits. Budget the commissioning window — minimum 60-90 days for stable operation — at the planning stage.
12-month outcome
| Metric | Pre (CP/VS) | Post (VPF) | Δ |
|---|---|---|---|
| Annual plant energy (chiller + pumps) | 3,420,000 kWh | 2,810,000 kWh | -18 % |
| Pump-only energy | ~752,000 kWh | ~245,000 kWh | -67 % |
| Plant kW / TR at average load | 0.85 | 0.71 | -16 % |
| Plant kW / TR at design load | 0.62 | 0.58 | -6 % |
| Chiller plant ΔT (annual avg) | 4.2 °C | 5.6 °C | +33 % |
| Chiller short-cycling events | 38/year | 4/year | -89 % |
Annual savings: 610,000 kWh × ₹8.5/kWh = ₹52 lakh. Payback: ~18 months at this load profile.
Why VPF wins in Indian commercial
Three convergent factors:
1. Indian commercial load profile is long-tail low-load. ASHRAE 90.1 default load curve assumes 25 % time at 100 %; Indian commercial is closer to 5 % time at 100 %. The longer you operate below 60 % load, the more VPF saves.
2. Soft-grid power tariff doesn’t subsidize wasted pump kWh. Indian commercial tariff (₹8-12/kWh) makes pump energy a real opex line item.
3. Chiller minimum-flow technology has matured. Modern variable-speed screw chillers handle 30-50 % minimum flow without nuisance trips, given proper bypass logic.
The retrofit is now standard for any campus > 500 TR in India. Below 500 TR, the capex doesn’t quite pay back in a 5-yr horizon.
ECBC + ASHRAE 90.1 alignment
ECBC 2017 §5.2.2.10 explicitly allows variable-flow distribution in chilled water systems. ASHRAE 90.1-2022 §6.5.4.4 mandates variable-flow for plants > 90 kW (≈26 TR) cooling. Most Indian commercial plants > 100 TR are required to be VPF or VS by 90.1 reference.
For LEED EAc1 / IGBC EE / ECBC compliance, VPF + ΔT-managed plant gains 4-7 points on whole-building performance.
From the Field — Engineer’s Notebook
The single most useful diagnostic in this retrofit was a 24-hour BMS trend graph showing plant ΔT alongside chiller stage count. Before retrofit: ΔT bobbed 3.5-5.0 °C through the day, chiller staged 4-6 times. After retrofit (post-commissioning, after the equal-percentage valve fix): ΔT held 5.4-5.8 °C, chiller staged 1-2 times. The chiller staging count alone is a proxy for ΔT-managed plant health. Operations teams should plot it monthly; any month with > 15 staging events deserves investigation.
5 takeaways
1. Eliminate the decoupler if and only if you replace it with a real minimum-flow bypass. Half-retrofits cause trips.
2. 2-way valve characterization must change. Linear curves for 3-way → equal-percentage for 2-way. Don’t forget.
3. Plant ΔP sensors at remote AHUs, not at plant header. Plant-header ΔP rewards over-pressurization; remote ΔP rewards efficient distribution.
4. Budget 60-90 days for commissioning stability. Anything less and you’re shipping an un-tuned plant.
5. Trend chiller staging count as a KPI. ΔT degradation is hard to spot; staging count is binary.
Designer’s checklist
- [ ] Plant load profile analysed (BMS export, hourly load for 12+ months)
- [ ] Chiller minimum flow per AHRI 550/590 verified for all chillers
- [ ] 2-way valves planned at every AHU coil (no 3-way leftovers)
- [ ] Bypass valve sized for chiller minimum flow
- [ ] Pump VSDs verified or planned
- [ ] BMS rewrite scope + commissioning window (60-90 days) agreed with stakeholders
- [ ] DP sensors specified at remote AHU positions
- [ ] Chiller staging logic + bypass-open-first sequence written
- [ ] Equal-percentage valve characterization specified
- [ ] Annual savings target documented vs baseline kWh
- [ ] ECBC / LEED / IGBC compliance documentation prepared
Pairs with: Cooling Load Methods Compared, India Cooling Load Rules of Thumb, Cooling Load Calculator, Research Paper 021 — CLTD vs RTS vs HBM Validation