Chiller Replacement Economics: Which kW/RT Are You Actually Buying?
An office owner's metered chiller-plant energy is 1.90 GWh over a 3,600-hour year at an average load of 620 RT, with electricity at S$0.26/kWh. The plant is three 400 RT water-cooled screw chillers (two duty), constant-speed primary pumps and fixed-speed cooling towers, with an implied plant performance of 0.85 kW/RT (chillers about 0.68, pumps 0.11, towers 0.06). Two proposals are on the table. Proposal A: replace both duty chillers with 2 × 450 RT oil-free VSD machines rated 0.55 kW/RT at AHRI conditions, capital S$2.6 million. Proposal B: plant controls and optimal sequencing, VSD on the primary and condenser pumps, condenser-water temperature optimisation and a 120 RT pony chiller, capital S$650,000, predicted to cut plant energy by 18 %. The owner's consultant used the bare 0.55 kW/RT figure for Proposal A, and the owner wants Proposal A endorsed because it produces the better number for the Green Mark submission. The site runs condenser water at 32 °C entering.
1. Check the Proposal A savings and compare the two options on simple payback.
Read the worked answer
Baseline check: 1,900,000 kWh / (620 RT × 3,600 h) = 0.851 kW/RT - the stated 0.85 is consistent.
Correcting Proposal A:
- Rating condition. 0.55 kW/RT is an AHRI 550/590 figure at 29.4 °C entering condenser water. At 32 °C the machine runs roughly 3 % worse per kelvin, i.e. about +8 % → 0.594 kW/RT.
- System boundary. Proposal A changes only the chillers. Pumps (0.11) and towers (0.06) stay. New plant = 0.594 + 0.17 = 0.764 kW/RT, not 0.55.
- Saving = 0.851 − 0.764 = 0.087 kW/RT → 0.087 × 620 × 3,600 = 194,000 kWh/yr → S$50,500/yr.
- Simple payback = 2,600,000 / 50,500 = about 51 years. Proposal A is uneconomic on measured performance.
Proposal B: 18 % × 1,900,000 = 342,000 kWh/yr → S$88,900/yr. Payback = 650,000 / 88,900 = 7.3 years, and it reduces both chillers and auxiliaries.
Ranking: B first, decisively. A's apparent advantage came entirely from quoting a machine rating as a plant result.
2. The owner wants the better Green Mark number and asks you to endorse Proposal A regardless. Advise, including how you would structure the decision over the plant's life.
Read the worked answer
The number and the bill are different things. Green Mark rewards design efficiency, so a 0.55 kW/RT machine scores well on paper; the owner pays for metered kilowatt-hours, and the calculation above shows Proposal A barely moves them because the auxiliaries and the part-load behaviour are unchanged. Endorsing A on the strength of its nominal rating would be giving an opinion I know to be misleading, and it would not survive measurement and verification.
What I would recommend:
- Proposal B now. It is the cheaper saving and it improves the whole-plant kW/RT, which is what an operational rating rewards. Apply for the available BCA incentive support, which materially shortens the payback.
- Stage the chiller replacement to end of life. Replacing a working chiller is a poor use of capital; replacing one at the point of failure, with the controls already in place, produces a much better incremental return. Model life-cycle cost at 3 % energy escalation and a stated discount rate rather than simple payback - and include the avoided cost of the forced replacement that would otherwise occur.
- Be explicit about the measurement basis. Report the design plant kW/RT and the corrected-to-site-conditions figure, and commit to a metered verification after twelve months.
- Address the split incentive. If the tenant pays the electricity and the owner pays the capital, the owner is being asked to fund an 18 % saving that accrues to someone else. That is often the real reason efficiency measures stall, and it is a lease problem, not an engineering one - a green lease clause or a service-charge-funded sinking fund resolves it.
I would not sign a submission that presents 0.55 kW/RT as the plant's expected performance.
3. How would you set up the measurement and verification and the commercial structure so that the savings are real and enforceable?
Read the worked answer
Measurement and verification:
- Use IPMVP Option B (retrofit isolation with continuous measurement) for a plant retrofit: permanent kWh meters on each chiller, the primary and condenser pumps and the tower fans, plus chilled-water and condenser-water flow and temperature for the independent variable.
- Establish a valid baseline first. Twelve months of metered data at the same operating hours and load is required; the existing 1.90 GWh is a single annual figure and needs to be disaggregated and normalised to cooling degree days and occupied hours.
- Normalise, don't just compare. Savings must be adjusted for weather, occupancy and load changes, otherwise a mild year looks like a successful project.
- Fix the uncertainty tolerance (typically ±10 % at 90 % confidence) and the period (twelve months, with quarterly reporting) in the contract.
Commercial structure:
- A performance guarantee with a defined shortfall remedy; the owner's risk is then the baseline, not the outcome.
- Retain a percentage of the contract sum until the first verified report.
- Agree measurement and verification protocol, meters, calibration and access before award - disputes almost always originate in a protocol written after the fact.
- Include a defined-conditions clause: the guarantee applies to the specified condenser-water temperature, load profile and operating hours; if the owner changes the building or the hours, the guarantee is re-based.
Without a defensible baseline and a normalisation method, the savings figure is an opinion rather than an asset.