This updated July 2026 guide explains how and when to convert existing primary–secondary chilled‑water plants to a primary‑only, variable‑flow arrangement. It is written for facilities engineers, consultants, commissioning agents, and owners who want actionable steps, updated best practices, and the latest industry developments that affect success today: advanced controls (MPC/AI), tighter decarbonization goals, thermal storage interactions, and cyber-secure BMS requirements.

Prerequisites and context: what has changed by 2026

Primary‑only retrofits remain a proven route to lower pumping energy and simpler piping. Since the original 2024–2025 wave of projects, several trends affect design and risk management in 2026:

  • Controls evolution: Model predictive control (MPC) and machine‑learning‑enabled optimization, now widely available in commercial BMS platforms, improve DP reset and chiller sequencing stability compared with simple PID-only logic.
  • Manufacturer guidance: A growing fraction of modern centrifugal, screw, and magnetic-bearing chillers are explicitly approved for variable‑flow operation without a bypass when the chiller control package or suction modulation is present. Still, manufacturer minimum-flow rules remain binding.
  • Decarbonization and grid interactivity: Owners targeting net‑zero or participating in demand‑response/ancillary‑service programs need plant controls that can prioritize grid signals, thermal-energy‑storage (TES) charging, and chiller efficiency simultaneously.
  • Cybersecurity and telemetry: Because retrofits increasingly rely on cloud-based analytics and remote optimization, BMS cybersecurity, encrypted telemetry, and vendor contract terms are now essential procurement considerations.

Before proceeding, confirm: chiller model and firmware revision; existing pump and motor ratings; available BMS/trending and network security posture; and whether the facility runs TES or has scheduled demand events that influence sequencing.

Step 1 — Feasibility and data collection (plan before you re-pipe)

  1. Collect baseline operating data for 7–21 days: chiller kW, pump kW, pump speeds, flow (or DP), supply/return temperatures, and building load proxies (electric and/or AHU runtime). Metering shortfalls are often the single largest barrier to a clean economic case.
  2. Document chiller make/model/serial and request manufacturer confirmation (written) of allowable minimum flow, required firmware features, and any mandatory bypass hardware. Many vendors now issue model-specific variable‑flow certificates—retain these in the project record.
  3. Identify operational constraints: TES (size and control), free‑cooling sequences, freeze protection, and critical redundancy requirements (e.g., hospital life‑safety). These determine minimum redundancy and staging logic.
  4. Perform a hydraulic sketch: re-derive system curve after decoupler removal (include piping, valve authority, coil resistances). This determines pump selection and whether the plant will operate on the intended portion of the pump curve at typical loads.

Step 2 — Design and hardware changes

The physical changes in 2026 are similar to prior practice but with a few additions reflecting new hardware and code expectations.

  1. Remove the decoupler only after confirming isolation valves and providing temporary bypass for shutdown tests. If removal is staged, mark valves clearly and update drawings.
  2. Specify VFD‑rated pumps sized for the revised system curve. Use pumps with extended VFD compatibility (active front end or low‑harmonic options) for installations where multiple large VFDs produce harmonics. Reference IEEE‑519 for harmonic limits; include mitigation (filters or AFE drives) where required by the electrical engineer.
  3. Minimum‑flow protection: implement one of three accepted approaches—(a) automatic chiller bypass valve sized to manufacturer minimum, (b) plate‑and‑frame or shell‑and‑tube recirculation heat exchanger, or (c) chiller‑integrated suction modulation/anti‑surge control if vendor‑approved. In 2026 some chillers can operate safely at 8–12% of rated flow with proprietary anti‑surge features; still, require vendor documentation.
  4. Instrument upgrade: install per‑chiller flow measurement (mag or ultrasonic) and local energy meters on chiller electrical feeds. Central DP transducer should be duplicated for redundancy in critical facilities; add temperature sensors at chiller supply/return and remote coil banks for delta‑T diagnostics.
  5. Piping and valving: maintain adequate valve authority—consider PICVs at terminal coils where possible to preserve control at low flows and reduce balancing labor. Update supports and expansion compensation for altered flow-induced forces.
  6. Cyber and telemetry: specify secure BMS integration points, encryption, and a minimal set of cloud telemetry points for remote optimization. Define vendor responsibilities for remote access, updates, and data ownership in contract documents.

Step 3 — Controls and sequencing (2026 best practices)

Controls remain the most consequential element. Adopt layered logic: base PID or DP reset for reliable local control, plus higher-level optimization for energy and grid objectives.

Primary pump control (differential‑pressure reset)

  1. Place DP transducer(s) where representative of the longest run and least dynamic location—commonly across the plant header. Duplicate transducers for critical applications.
  2. Implement DP setpoint reset keyed to a load proxy (building cooling load, chiller inlet temp, or chilled‑water energy meter). In 2026, MPC-based reset that forecasts load for 15–30 minutes reduces hunting compared with aggressive short-horizon resets; consider MPC where BMS supports it.
  3. Use minimum‑speed locks tied to manufacturer minimum‑flow requirements and anti‑windup logic in the VFD input. Include soft‑start or flywheel options for large motors to limit inrush and reduce nuisance trips.

Chiller sequencing

  1. Sequence chillers using a blend of flow permissives and efficiency curves: enable a chiller only when its inlet/outlet temperatures and measured flow meet manufacturer minimums and projected run-time threshold (to avoid short cycles).
  2. Where TES or demand-response events occur, include a supervisory mode that can (a) prioritize TES charging/discharging, (b) constrain maximum plant electrical demand, or (c) direct chillers to follow a schedule for grid services while preserving minimum flow protections.
  3. Ride‑through and warm‑start routines are mandatory—use short deadbands and staged warm-up to avoid compressor surge or liquid slugging when returning from low‑flow conditions.

Minimum‑flow enforcement

Combine hard and soft protections:

  • Hard: automatic bypass or recirculation loop controlled by a valve or pump to guarantee measured flow meets the named minimum.
  • Soft: supervisory interlocks that prevent a chiller from starting if a flowmeter reading is below threshold, and that de‑energize a chiller if its measured flow falls unexpectedly during operation.

Step 4 — Commissioning and acceptance (detailed sequence)

Commissioning is the single most important risk-mitigation activity. Expand acceptance beyond start/stop checks to dynamic, data-driven validation.

  1. Baseline archival: ensure the pre‑retrofit baseline dataset is archived in a retrievable format. If pre‑retrofit data are missing, conduct a 2–4 week baseline before major changes.
  2. Hydronic verification: measure pump curves in situ (at minimum three speeds) and correlate with calculated system curve. Confirm NPSH margin at worst-case conditions and verify no cavitation signatures during speed changes.
  3. Minimum‑flow validation: at minimum VFD speed, confirm measured flow per chiller meets manufacturer minima. Exercise bypass and ensure control response is predictable and repeatable.
  4. Delta‑T testing: perform staged load tests (25%, 50%, 75%, 100% typical load) and document delta‑T across chillers and representative coil banks. Use these results to calibrate DP reset and flow allocation logic.
  5. FDD and transient testing: enable fault detection algorithms where available. Create controlled faults (sensor disconnect, pump trip, chiller disable) and verify safe degradation modes and operator alerts.
  6. Energy verification: after a stabilization period (commonly 30–90 days to cover variable weather and operational schedules), compare metered energy and kW/ton to baseline. Recent retrofit datasets typically show pump energy savings of 20–35%; document variance drivers (occupancy, controls tuning, TES dispatch).
  7. Operator training and documentation: deliver hands‑on training and updated O&M manuals that include minimum‑flow diagrams, override procedures, and contact points for manufacturer support.

Common mistakes (and how to avoid them)

  • Assuming all chillers are equal: Do not assume older chillers tolerate low flow. Always obtain written manufacturer approval for variable‑flow operation and minimum‑flow values.
  • Poor sensor strategy: Under‑instrumentation (no per‑chiller flow) undermines control and commissioning. Invest in flow metering and redundant DP sensing in critical plants.
  • Ignoring TES interactions: When TES is present, sequencing must coordinate TES and chiller operation—otherwise, chargers/dischargers can cause unexpected low flows or rapid setpoint changes.
  • No cybersecurity requirements: If remote optimization or cloud analytics are used, define secure access, data ownership, patching responsibilities, and vendor SLAs up front.
  • Skipping transient tests: Not testing step changes in load and pump speed often leaves instabilities undiscovered until occupant complaints or faults occur.

Pro tips — advanced advice for better outcomes

  • Use model predictive control (MPC) where available to smooth DP reset actions and anticipate load swings—this reduces hunting and chiller staging events.
  • Deploy per‑chiller electrical metering and kW/ton trending to spot degraded heat‑transfer or fouling that may be mistaken for control issues.
  • Consider PICVs at terminal units in variable‑flow retrofits to preserve control authority and reduce balancing labor; they also make delta‑T recovery more robust.
  • Include harmonic mitigation and AFE drives for large multi‑VFD plants to comply with IEEE‑519 and minimize nuisance trips or power‑quality issues.
  • For critical facilities (hospitals, data centers), retain a small decoupler or bypass as an operational contingency if redundancy or extremely low flows are expected during maintenance.

Updated real‑world example (anonymized, 2025 retrofit)

A 300,000 ft² mixed‑use building retrofitted its primary‑secondary plant in 2025. Actions: removed decoupler, installed redundant primary pumps with AFE VFDs, added per‑chiller ultrasonic flowmeters, implemented a 12% manufacturer‑approved bypass, and deployed an MPC layer for DP reset and chiller staging. Commissioning included 30 days of baseline and 90 days of post‑retrofit trending. Results: primary pump energy down 28% vs baseline, chilled‑water delta‑T improved from 6°F at part load to 9–10°F, and measured chiller kW/ton improved by ~12% at typical mid‑season loads. The project also enrolled in a demand‑response program, generating small additional revenue and enabling controlled load curtailment without comfort complaints.

Post‑retrofit monitoring and continuous optimization

Monitor for 3–6 months and then evaluate quarterly. Key trending items:

  • Chiller power (kW) and kW/ton at representative load bands
  • Primary pump kW, speed, and system DP
  • Supply/return temperatures and delta‑T across chillers and major coil banks
  • Frequency and duration of chiller staging and short cycles
  • FDD alarms, sensor health, and data gaps

Use this data to refine DP reset curves, adjust minimum speed locks, and tune chiller staging hysteresis. Where cloud analytics or vendor optimization is used, define a quarterly report and decision gate before auto-applying major strategy changes.

Final checklist before you flip the switch

  1. Written chiller manufacturer approval for variable flow and documented minimum‑flow strategy.
  2. Per‑chiller flow measurement and DP sensing installed and calibrated.
  3. VFDs specified with harmonic mitigation appropriate to the electrical distribution and utility requirements.
  4. BMS logic validated in factory/bench tests and field tested for transient conditions.
  5. Commissioning plan includes dynamic tests, FDD activation, and at least 30–90 days of post‑retrofit trending.
  6. Operator training completed; O&M docs updated to include restart and bypass procedures and vendor support contacts.
  7. Cybersecurity and data‑ownership terms signed if remote optimization or cloud analytics will be used.

Conclusion

Primary‑only chilled‑water retrofits still deliver measurable energy and operational benefits in 2026—but success depends more than ever on controls sophistication, instrumentation, and integration with grid and TES objectives. Respect chiller minimum‑flow requirements, specify adequate measurement and cybersecurity, and invest in commissioning and data‑driven optimization. With those elements in place owners can realize lower pumping energy, cleaner hydronic layouts, and more stable delta‑T—while positioning the plant to participate in grid programs and net‑zero strategies.

FAQ

Is primary‑only always the best retrofit option?

No. Primary‑only is attractive when pump energy is significant and chillers are manufacturer‑approved for variable flow. For small plants, heavily constrained piping, or older chillers that require constant bypass, a staged or hybrid approach (retaining a small decoupler or adding per‑chiller recirculation) may be better. Evaluate case-by-case with hydraulic modeling and vendor input.

How low can chiller flow go safely?

Minimum‑flow requirements are chiller‑specific. Many modern chillers can operate safely at flows down to roughly 8–15% of rated flow when equipped with anti‑surge/suction modulation features; older models commonly specify 15–25%. Always secure written manufacturer guidance and implement a hardware bypass or software interlock to guarantee compliance.

What instrumentation is essential for a successful retrofit?

At minimum: per‑chiller flow measurement, plant DP transducer(s), chiller supply/return temperature sensors, and energy meters on chillers and primary pumps. Redundant DP sensing and remote trending capability are recommended for critical facilities.

Can I use cloud‑based optimization for DP reset and chiller staging?

Yes—with caveats. Cloud analytics and vendor‑hosted MPC provide improved optimization but require robust cybersecurity, clear data ownership, and SLAs for updates and fault handling. Always retain local safe modes and fail‑safe logic in the plant controls so the system operates predictably if cloud services are unavailable.

How long before I see payback from a primary‑only retrofit?

Payback varies widely: expected pump energy savings typically range from 20–35% on similar retrofit projects; when combined with reduced maintenance and operational improvements, many projects see simple paybacks of 2–6 years. Exact payback depends on operating hours, utility rates, incentive programs, and any added value from demand‑response or reduced chiller runtime.