Affinity Laws in Practice: Quantifying Fan and Pump Energy Savings from VFD Speed Reduction

Every MEP engineer who has specified a chiller plant knows that variable frequency drives save energy on pumps and fans. The question is never whether to use VFDs. It is how much they save for a specific plant, what limits the theoretical savings in practice, and whether the economics justify the retrofit on an existing fixed-speed installation.

The affinity law provides the physics: power is proportional to the cube of speed. A 20% speed reduction yields approximately 50% power savings. But that statement, repeated in every VFD manufacturer's brochure, omits the constraints that determine how much of the theoretical savings you actually capture on site. Minimum stable VFD frequency, minimum evaporator flow protection, sensor placement for demand-based control, and control loop tuning quality all create a gap between the textbook number and the operating reality.

This post closes the gap between the textbook cube law and site-measured savings: the static head exception that reduces the effective exponent below 3.0, the baseline adjustment for plants that already have VFDs running unoptimized, the practical constraints that limit the achievable speed range, and a worked cost-benefit analysis at Indian tariffs. The goal is a framework an MEP engineer can use to structure the savings analysis and sanity-check the results for any pump or fan VFD retrofit.

The Cube Law: Why a 20% Speed Reduction Yields Approximately 50% Power Savings

The affinity laws describe the relationship between rotational speed and three hydraulic or aerodynamic variables for centrifugal pumps and fans:

Flow is proportional to speed: Q ∝ N

Head (pressure) is proportional to the square of speed: H ∝ N²

Power is proportional to the cube of speed: P ∝ N³

The third law follows from the first two: power is the product of flow and head (P = Q × H), so P ∝ N × N² = N³. This proportionality holds when fluid density and pump efficiency are approximately constant across the speed range. At speeds well below the best efficiency point (typically below 50 to 60% of rated), pump efficiency degrades and actual power consumption exceeds the cubic prediction.

In practical terms, the power at any reduced speed equals the rated power multiplied by the cube of the speed ratio:

P_reduced = P_rated × (N_reduced / N_rated)³

The nonlinearity of the cube function is what makes VFD savings substantial:

Speed (% of rated)	Power (% of rated)	Savings vs. full speed
100%	100.0%	0%
90%	72.9%	27.1%
80%	51.2%	48.8%
70%	34.3%	65.7%
60%	21.6%	78.4%
50%	12.5%	87.5%

The savings curve is steep between 100% and 80% and continues to steepen below. This is the "second-order effect" that makes auxiliary equipment (pumps and fans, representing 28 to 45% of total chiller plant power) the most responsive subsystem to VFD-based optimization.

The Static Head Caveat

The cube law assumes that the pump or fan operates on a system curve that passes through the origin: all system resistance is frictional, and at zero flow, the required head is zero. This assumption holds well for closed-loop chiller plant piping (both CHW and CW loops), where the system curve is predominantly friction-driven.

It does not hold where significant static head is present. In a high-rise building with an open cooling tower loop, the CW pumps must overcome the elevation difference between the chiller condenser and the tower basin. That static head component is constant regardless of flow rate. The system curve has a non-zero y-intercept, and the power reduction at reduced speed is less than cubic.

Depending on the ratio of static to frictional head, the effective exponent may be closer to 2.0 or 2.5 rather than 3.0. In these installations, the cube law overpredicts savings, and the VFD economic analysis should use the actual system curve rather than the theoretical cubic.

For the majority of Indian commercial chiller plants with closed-loop piping and cooling towers at grade or on the roof (one to two floors above the plant room), the static head is a small fraction of total head, and the cube law is a reasonable approximation. CT fans and AHU fans operate against purely dynamic (frictional) resistance with no static head component, which means the cube law applies more accurately to fan VFD savings than to pump VFD savings in systems with elevation-related static head.

The 0.614 Baseline for Already-VFD Plants

When evaluating a plant that already has VFDs on pumps or fans but operates them without automated optimization (manual speed setpoints or fixed frequency), the savings baseline is not 100% power. The calculator uses 0.85³ = 0.614, assuming the existing VFDs operate at approximately 85% average speed under manual control.

This means the plant already consumes 61.4% of rated power on those subsystems, and the optimization opportunity is the gap between 61.4% and the load-band-optimized speed profile, not the gap between 100% and the optimized profile.

This distinction matters. A plant with fixed-speed CW pumps has 20 to 30% savings potential from VFD retrofit. A plant with VFDs already installed but running at fixed setpoints has a much smaller incremental savings from optimization alone. Conflating the two overstates the case for BMS optimization on plants where the hardware already exists.

Practical Constraints That Determine How Much Theoretical Savings You Actually Capture

The cube law sets the theoretical ceiling. Five practical constraints determine the floor.

Minimum Stable VFD Frequency

Every VFD has a minimum output frequency below which the motor cannot produce reliable torque. For standard induction motors on a 50 Hz supply, this is typically 15 to 20 Hz (30 to 40% speed). Below this threshold, the motor overheats or stalls.

A pump specified for "VFD operation down to 50% speed" is realistic. Sustained operation below 30% requires an inverter-duty motor with enhanced cooling, which is not standard in Indian brownfield plants.

Minimum Evaporator Flow

In a primary-secondary chilled water system, the secondary pump speed is limited by the chiller's minimum evaporator flow requirement. If flow drops below the manufacturer's minimum, the result is laminar flow, uneven heat transfer, and potential freeze protection trips.

This is why plant optimization tools include pump staging logic at low loads: below approximately 40% load, de-staging to fewer pumps at higher individual speed is more efficient than running all pumps near their minimum stable frequency.

Sensor Placement and Control Signal Quality

VFD speed control requires a feedback signal. For CHW pumps, the standard is DP control with the sensor at the hydraulically most remote AHU coil, not at the pump header.

A header-mounted sensor overestimates available pressure at remote coils, causing the pump to run slower than required and starving distant AHUs. For CT fans, the control signal is approach temperature (CWS minus wet-bulb), measured at the tower basin. Sensor accuracy and calibration drift directly affect speed command quality.

Control Loop Tuning

A poorly tuned PID loop controlling VFD speed produces oscillation: the pump speeds up, overshoots the DP setpoint, slows down, undershoots, speeds up again.

The operators observe the oscillation, lose confidence in the automation, and switch the VFD to manual at a fixed frequency. The VFD is still installed; it just runs at constant speed. The savings disappear, and the capital cost becomes a sunk loss.

This failure mode is common enough that it should be specified against in the tender: the commissioning scope should include PID loop tuning with documented step-response tests and sign-off criteria.

Coil Valve Type for CHW Pumps

One frequently overlooked constraint: VFD speed control on CHW pumps is only effective when the AHU coils use 2-way modulating valves.

In a 3-way valve system, the bypass leg maintains near-constant system flow regardless of zone load. The pump sees a flat system curve and has limited opportunity to reduce speed.

Table F reflects this directly: CHW pump savings from VFD conversion are 20 to 30% with 2-way valves but only 5 to 10% with 3-way valves. Specifying a pump VFD without addressing the valve type is specifying the hardware without the conditions for it to deliver.

The Honest Assessment

In practice, these constraints typically limit pump operating speed to 60 to 70% as a sustained minimum, not the 50% that the cube law makes look attractive.

CT fans can often run lower (down to 30 to 40% in shoulder seasons) because they have no minimum flow constraint and the approach temperature feedback is more stable than DP. AHU fans are constrained by minimum outside air requirements and static pressure for proper air distribution.

The result: given the minimum speed floor of 60 to 70% for pumps and the control losses described above, achievable savings from VFD speed reduction are typically 70 to 85% of the theoretical cubic calculation.

A cube-law estimate of 65% savings on CW pumps becomes 45 to 55% in practice. This is still substantial, but presenting the theoretical number without the practical discount sets expectations that the site will not meet.

Fixed Speed to VFD: Working Through the Economics for a Typical Indian Commercial Plant

Plant Profile

Plant profile: 200TR commercial building. Four CW pumps at 7.5HP each, configured as two duty/standby pairs. Two pumps running at any time, at 100% speed, year-round.

Current CW Pump Energy

Running pump kW  = 2 pumps × 7.5 HP × 0.746 kW/HP = 11.19 kW
Annual kWh       = 11.19 × 4,500 hours             = 50,355 kWh
Annual cost      = 50,355 × ₹9/kWh                 = ₹4.53 lakh

With VFD at 70% Average Speed

The 70% average assumes the plant operates at 50 to 80% cooling load for most of its hours, with the CW pump speed tracking condenser water flow demand. Applying the cube law:

VFD pump kW      = 11.19 × 0.7³ = 11.19 × 0.343   = 3.84 kW
Annual kWh       = 3.84 × 4,500                     = 17,280 kWh
Annual cost      = 17,280 × ₹9                      = ₹1.56 lakh

Savings

Annual kWh saved  = 50,355 − 17,280                  = 33,075 kWh
Annual cost saved = ₹4.53 − ₹1.56                   = ₹2.97 lakh
Savings %         = 65.7% (= 1 − 0.343)

Applying the practical discount: At 75 to 85% of theoretical (accounting for minimum speed constraints, control losses, and periods when pumps must run at higher speed), the realistic savings range is ₹2.2 to 2.5 lakh per year.

VFD Cost Estimate

Four 7.5HP (5.6 kW) VFDs at current Indian market pricing: approximately ₹8,000 to ₹12,000 per kW of drive capacity, depending on brand and features.

Total drive cost: approximately ₹1.8 to 2.7 lakh. Installation (panels, wiring, commissioning) adds approximately ₹1.0 to 1.5 lakh.

If existing motors are not rated for inverter duty, sustained operation below 40% speed may require motor replacement or auxiliary cooling, adding to the retrofit cost. Harmonic filtering (line reactors or active filters) may also be required when multiple VFDs share a common electrical bus.

Total installed cost      = ₹2.8 to 4.2 lakh
Annual savings (realistic) = ₹2.2 to 2.5 lakh
Simple payback             = 13 to 19 months

Assumptions stated: 200TR plant, 2 CW pumps running at any time, 70% average VFD speed, 4,500 annual operating hours, ₹9/kWh tariff, practical savings discount of 75 to 85% of theoretical. VFD pricing is indicative and varies by brand, IP rating, and harmonic filter requirements. All four pumps require VFDs even though only two run at a time, because either pair may be on duty.

This is the arithmetic for CW pumps alone. The same cube-law economics apply to CHW secondary pumps (20 to 30% savings from fixed-to-VFD conversion with 2-way valves), CT fans (30 to 65% depending on climate zone), and AHU fans (25 to 40% for on/off to VFD conversion).

When all four subsystems are converted from fixed speed to VFD with demand-based control, the combined savings on auxiliary equipment typically represent 8 to 15% of total plant energy, with payback periods in the 12 to 24 month range depending on plant size and operating hours.

From Component Savings to Whole-Plant Optimization

Pump and fan speed reduction is one axis of the total plant kW optimization discussed in the Week 4 post on sweet-spot curves.

Reducing CW pump speed saves pump kW but also reduces condenser water flow, raising CW return temperature and increasing chiller lift. The cube law tells you what the pump saves in isolation. The sweet-spot curve tells you what the plant saves in total.

The difference is the reason that VFD speed commands in an optimized plant are load-band targets that shift with building load, wet-bulb temperature, and chiller staging, not fixed setpoints.

Tor Shield's Chiller Plant Manager uses the load band table to set speed targets for each subsystem at each load level.

At 70% load, the CW pump midpoint is 82.5%; at 50% load, it drops to 72.5%. CT fans follow a parallel schedule, from 75% at 70% load to 57.5% at 50% load.

These are not arbitrary setpoints. They are the output of the total-plant kW minimization that balances pump savings against chiller lift impact.

The HVAC Savings Calculator quantifies the subsystem-level savings from VFD conversion using Table F, with separate rows for each pump and fan type, climate-adjusted CT fan savings, and the coil valve type's effect on CHW pump savings.

Enter your plant configuration to see where the cube law applies to your equipment.

P ∝ N³ is the most expensive equation your facilities team may never have applied.