It does not invent part-load curves or extrapolate beyond verified performance data.

1. Purpose and Engineering Decision

A compressor room with several machines can waste energy even when each individual compressor is efficient. The reason is dispatch: the wrong base machine, too many fixed-speed units online, a trim machine outside its stable range, or unnecessary unloaded capacity can raise total package kW at the same delivered plant flow.

Which compressor should trim? What reserve remains? How much power does the selected combination require? How does the dispatch change across an annual demand profile?

Dispatch mode	Primary output
Single demand point	Lowest-kW feasible fleet state plus next-best alternatives
Demand sweep	Recommended combination across a series of flow bands
Annual load profile	Weighted annual kWh and cost by dispatch state
Reserve-constrained	Best state that also satisfies running or standby reserve policy
Scenario compare	Before/after sequencing rule, availability or machine change

2. Comparison Gate - Common Basis First

Item	Required control
Delivered flow
Pressure duty	Use performance data at the same required discharge pressure
Power boundary	Prefer total package electrical input for every machine
Operating range	Respect verified minimum and maximum stable flow
Part-load curve	Interpolate only between verified points; never extrapolate silently
Availability	Exclude unavailable, maintenance-held or intentionally locked-out machines

A mathematically lowest-kW combination is not a valid engineering recommendation when its performance basis is inconsistent.

3. Compressor Operating-Point Model

Each compressor is represented by one or more verified flow-power points at the common pressure duty. A VSD, centrifugal or other modulating machine may have multiple verified operating points.

Machine field	Meaning
Q_min	Minimum verified stable delivered flow
Q_max	Maximum verified delivered flow
P(Q)	Package input power at each verified/interpolated flow
Type	Fixed-step / VSD / modulating / custom curve
Start/stop eligibility	Whether the dispatch engine may turn the machine on/off in the scenario
Reserve eligibility	Whether an available stopped machine counts as standby reserve

For piecewise-linear interpolation between verified points:

P(Q) = P1 + (Q - Q1) x (P2 - P1) / (Q2 - Q1)

4. Fleet Dispatch Constraints

Sum(Q_i) >= Plant Demand

Q_min,i <= Q_i <= Q_max,i for each running modulating compressor

The basic objective is to meet the specified demand with the lowest total package input power while obeying all hard constraints.

Objective: Minimize Sum(P_i(Q_i))

5. Reserve Policy

Reserve is not one concept.

Reserve type	Definition
Standby reserve	Available stopped capacity that could be started after a running-unit loss
Running reserve	Unused headroom on already-running modulating machines
N+1 check	Demand can still be met after the largest required unit is unavailable, subject to start/response assumptions
No reserve constraint	Pure energy dispatch; resilience shown separately

Standby Reserve = Sum(Q_max of eligible stopped machines)

Running Headroom = Sum(Q_max - Q_i for running machines)

6. Input Model

Input group	Field	Validation / note
Plant	Demand point(s)
Plant	Required pressure
Machine	ID / type / availability	Required
Machine	Q_min / Q_max	Verified operating envelope
Machine	Flow-power points	At least one; multiple for part-load
Machine	Power boundary	Package input preferred
Control	Fixed-step / modulating	Defines feasible dispatch states
Reserve	None / standby / running / N+1	Explicit selection
Economics	Annual hours + tariff	Optional
Scenario	Start locks / maintenance holds	Optional hard constraints

7. Search Strategy for Small Fleets

For a browser calculator with a modest fleet, the most transparent method is bounded combination search.

Enumerate allowed on/off combinations for available compressors.

Reject combinations whose total Q_max cannot meet demand.

Apply required standby or running reserve constraint.

Allocate flow among running machines using their verified part-load curves.

Calculate total package kW and surplus-flow exposure.

Rank feasible states by total kW, then by secondary criteria such as fewer starts or greater reserve if the user selected those tie-breakers.

For larger fleets, the implementation may switch to dynamic programming or constrained optimization, but the same engineering constraints and traceability must remain visible.

8. Fixed-Speed and Trim-Machine Boundary

A fixed-speed compressor that only has a verified full-load production point should not be assigned an invented 63% flow point. A VSD or other modulating machine can act as trim only inside its verified stable operating range.

Condition
Fixed-speed, one verified loaded point	Discrete output state
VSD with verified curve	Continuous/piecewise-linear allocation within Q_min-Q_max
Unverified low-flow region	Infeasible; do not extrapolate
Fixed-speed surplus above plant demand

9. Reference Fleet for Test Cases

Machine	Type	Verified flow-power data
A	Fixed-speed	20 m3/min at 110 kW
B	Fixed-speed	15 m3/min at 84 kW
C	VSD trim	5 m3/min at 34 kW; 10 at 53 kW; 15 at 75 kW

All three test machines are assumed to use the same FAD/reference basis and discharge pressure for these examples.

10. Reference Test Case A - 28 m3/min Demand

Feasible state	Delivered flow	Total kW	Comment
A + C at 8	28 m3/min	155.4	Feasible
B + C at 13	28 m3/min	150.2	Feasible - lower kW
A + B	35 m3/min	194.0	7 m3/min surplus; cycling/spill risk
C alone	<=15 m3/min	-	Insufficient capacity

A + B is not treated as equally attractive because it materially oversupplies the defined demand.

11. Reference Test Case B - Standby Reserve Constraint

At the same 28 m3/min demand, require at least 10 m3/min of stopped standby capacity. Both A + C and B + C retain one fixed-speed machine offline.

Running state	Stopped eligible reserve	Reserve result	Energy result
A + C at 8	B = 15 m3/min	PASS	155.4 kW
B + C at 13	A = 20 m3/min	PASS	150.2 kW - selected
A + B	C = 15 m3/min	PASS	194.0 kW; surplus-flow issue

At a higher demand where every compressor must run, standby reserve becomes zero.

12. Reference Test Case C - Annual Demand Profile

Use four demand bands: 10 m3/min for 2,000 h/y, 18 m3/min for 2,500 h/y, 28 m3/min for 1,500 h/y and 35 m3/min for 1,000 h/y. One practical low-kW dispatch set is shown below.

Demand band	Hours/y	Selected dispatch	Fleet kW	Note
10 m3/min	2,000	C at 10	53.0
18 m3/min	2,500	A fixed at 20	110.0
28 m3/min	1,500	B + C at 13	150.2
35 m3/min	1,000	A + C at 15	185.0

Annual Fleet Energy = Sum(Fleet kW at Demand Band x Hours at Band)

Annual energy for the illustrated dispatch is approximately 791,300 kWh/y. The 18 m3/min band is specifically flagged because a 20 m3/min fixed-speed machine cannot be assumed to deliver exactly 18 m3/min without cycling or another control mechanism.

13. Baseline Rule vs Optimized Dispatch

For example, a baseline may always start A first, then B, and use C last. At 28 m3/min that rule can force A + B at 194 kW, while the optimized state B + C at 13 m3/min requires about 150.2 kW.

Dispatch Difference at 28 m3/min = 194.0 - 150.2 = 43.8 kW

The calculator must apply the same demand hours, pressure basis and machine performance curves to both baseline and optimized cases before reporting annual savings.

14. Tie-Breakers and Operational Preferences

Lowest instantaneous kW is the primary energy objective, but real compressor rooms may need secondary rules. These should be explicit tie-breakers or hard constraints, not hidden penalties.

Preference	Possible implementation
Limit starts	Reject or penalize states exceeding user-defined start policy
Run-hour balancing	Tie-break between near-equal energy states
Preferred base machine	Soft preference only after hard constraints pass
Maintenance hold	Hard exclusion
Minimum reserve	Hard constraint
Avoid low trim point	Hard minimum Q or user-defined preferred trim band

15. Data Confidence Model

Performance data	Suggested confidence
Synchronized measured kW + flow + pressure map	High
Verified package performance curve at common pressure	High
Vendor part-load curve with clear basis	Medium-High
Mixed sources reconciled to common basis	Medium
Rated capacity + nameplate motor kW only	Preliminary
Unmatched pressure / flow / power boundaries	Optimization incomplete

16. Validation and Guardrails

Condition	Required behaviour
Plant demand <= 0	Block dispatch
Machine Q_max <= 0	Exclude / error
Q_min > Q_max	Block machine curve
Flow reference bases differ
Pressure duties differ	Flag incomplete; do not rank cleanly
Power boundaries differ	Flag incomplete; reconcile before ranking
Interpolation outside verified curve	Block extrapolation
No feasible combination meets demand	Return explicit capacity shortfall
No feasible combination meets reserve	Return reserve shortfall separately
Fixed-speed surplus materially exceeds demand

17. Output Specification

Output	Display rule
Selected dispatch	Running machines + assigned flow
Total fleet flow	On common reference basis
Total package kW	Primary energy result
Fleet specific power	Total kW / useful delivered flow
Reserve	Standby and running headroom shown separately
Next-best states	At least 2 alternatives where feasible
Annual energy	kWh/y from demand profile
Annual cost	Only when tariff supplied
Warnings	Surplus flow, cycling risk, incomplete curve, reserve failure, basis mismatch
Confidence	High / medium / preliminary / incomplete

Part Load Fleet Dispatch Calculator

Use the working tool for this page