Tariffs, carbon tax and Singapore's new mandatory energy audit have turned 'business as usual' cooling into a compliance problem. ECOISM cuts it — without replacing your equipment — and hands you data that survives an audit.
Four forces are pushing the cost of cooling up at the same time. None of them are optional any more.
Regulated tariff 27.27¢ / 29.72¢ per kWh (2026 Q2); SME spot rates pass S$0.30/kWh.
Source: SP Group, tariff revision Apr–Jun 2026.From S$25/t (2024–25) to S$45/t (2026–27), targeting S$50–80/t by 2030.
Source: SP Group / NCCS.Since Sept 2025, BCA's MEI rule requires energy-intensive existing buildings to audit and improve — with verifiable data.
Source: BCA / Baker McKenzie Resource Hub.80% of building stock to be Green Mark certified by 2030 — certified stock targets ~S$1.3B / 4.2B kWh saved.
Source: BCA / FOZL.Conservative, audit-ready numbers from real deployments — exactly the "verifiable data" MEI and Green Mark ask for.
Teal = ESS (panel + refrigerant), average 22.23% · grey = ESP (panel only), average 11.18%. Group averages across all PoC measurements (case simple average). Re-verified site (Site B) is shown at the case average of matched-condition 27.4% and all-condition 16.3% (=21.85%); the case card keeps 27.4%. Tested on DAIKIN & Casper equipment; client names, store names and countries withheld. Figures are corrected lower bounds.
*Site C pharmaceutical plant, metered kWh 1,693.5 → 1,253.9 (26.0%, corrected 25.3%). Outdoor temperature was slightly lower after install (favourable) and corrected via the cooling-characteristic table; humidity rose (unfavourable) — the reduction held. Daytime high-speed band reached 30.9%; nights −10.4% — reported value is the corrected lower bound.
Following the PoC results, a major supermarket chain is rolling ESS out across its stores.
"Stores" are rollout locations, distinct from the anonymous PoC "Sites" above. Client and store names withheld.
53.6°C
49.4°CTo measure savings correctly, you first have to hold these constant. Compare periods where they differ and the effect disappears into weather and occupancy noise.
Higher outdoor temperatures make heat rejection at the condenser harder, raising pipe pressure and pushing compressor run-rate — and power — up.
The more heat from occupancy, lighting and kitchens, the closer the system runs to full load. Near-empty rooms run at low speed.
Humid air takes more energy to cool (latent load). We derive relative humidity from dry-bulb and dew point and align the difference between compared periods.
ESP is a heat-exchange panel bolted to an existing condenser; ESS adds a matched refrigerant. Neither replaces the unit.
After condensing, the liquid refrigerant is sub-cooled a further 3–5 °C.
→The same cooling now needs less refrigerant mass-flow (ṁ).
→Compressor power W = ṁ × (h₂ − h₁) falls — that's the saving.
These side benefits are not counted in the reported reductions — so the figures stay on the conservative side.
A naïve before/after can't tell whether a difference is the ESP/ESS effect or just weather and crowding. Only after aligning temperature, humidity and time of day can a change in power be attributed to the technology.
Indoor heat load varies predictably by hour, so we compare before/after data from the same time slots — holding heat load roughly constant.
Among same-hour data, we accept only pairs whose outdoor-temperature difference is within about 5%.
Relative humidity from dry-bulb and dew point; we accept pairs within about 5% humidity difference.
Tiny remaining temp/humidity gaps are normalised with the DAIKIN cooling-characteristic table. Data selection and residual correction play different roles — this is not double-counting.
Favourable conditions are corrected or excluded and side-savings (fewer running units, relaxed setpoints) are not counted — so every reported figure is a transparent, reproducible, audit-ready lower bound. Measured with HIOKI temperature/humidity and current loggers.
Measurement method and conclusion for four of the PoCs above. Equipment and capacity are real; only confidential contract figures are withheld.
ESS applied to a DAIKIN VRV-X 20HP, reconciled against the customer's own electricity meter. Analysis by operating band, with same-hour data and ±5% temperature/humidity windows.
| Band | Reduction |
|---|---|
| Low-speed (night) | −10.4% |
| High-speed (day) | 30.9% |
| 12-hour total | 26.5% |
During ESP installation, a serious pre-existing fault on the customer side surfaced. The procedure was re-planned to isolate the ESP effect: a 10HP outdoor unit was run alone, with other units excluded at the distribution board.
| Condition | Reduction |
|---|---|
| Matched conditions — same temp & room-hours | 27.4% |
| All-condition average | 16.3% |
Pre-install setpoints of 18–19 °C were excessively low; after install, relaxing to 23 °C drew no complaints.
8HP (DAIKIN Ducted Split-H, ESS). Before/after compared under identical conditions, with the reduction computed directly from metered energy (kWh) — the most quantitatively rigorous case in this set.
| Item | Before | After |
|---|---|---|
| Energy (kWh) | 1,693.5 | 1,253.9 |
| Outdoor current (A) | 4.4 | 3.3 |
| Outdoor temp (°C) | 29.0 | 27.9 |
| Humidity (%) — ↑ unfavourable | 72.9 | 76.2 |
ESS (panel + matched refrigerant) on an existing 4 HP DAIKIN Homes unit; reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 24.73% |
ESS (panel + matched refrigerant) on an existing 12 HP DAIKIN VRV-Ⅲ; reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 18.53% |
ESS (panel + matched refrigerant) on an existing 10–12 HP DAIKIN VRV-Ⅲ; reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 16.45% |
ESP (panel only) on an existing 8 HP DAIKIN SkyAir unit; reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 13.03% |
ESP (panel only) on an existing 4 HP DAIKIN unit (older, R22); reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 12.50% |
ESP (panel only) on an existing 5.5 HP Casper unit (R32); reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 10.50% |
ESP (panel only) on an existing 4 HP Casper unit (R410a); reduction measured under matched conditions.
| Basis | Reduction |
|---|---|
| Matched-condition result | 8.70% |
Across different stores, capacities (10/16/20HP), countries and seasons, ESS stays in the 20% class (average 22.23%) — not tied to one environment.
Same-condition comparison plus characteristic-table residual correction make the results transparent, reproducible and audit-ready.
Favourable conditions are corrected or excluded; secondary savings from run-rate and setpoints are not counted. Based on third-party / customer data.
Better Δt means fewer running units, relaxed setpoints and CO₂ reduction. As a large electricity load, a ~20% cut has a sizeable CO₂ impact.
All verification units were existing machines in service 3+ years — a brand-new unit will not necessarily see ~20%.
Under manufacturer terms, fitting ESP/ESS counts as a modified machine, so the maker's warranty no longer applies. We therefore recommend units already past their warranty period.
On-site assessment
Scope & targets
Measured trial
Audit-ready data
Full deployment
A no-replacement retrofit suits operations where downtime and capital are both expensive.
Indicative — the exact figure is verified per site in the PoC.
Book a proof-of-concept, or request the audit-ready dataset behind every number.
From S$1,000, transparent pricing, no lock-in — built for SMEs tired of pilots that stall.
Explore AI →Premium Vietnamese bananas diversifying Singapore's import mix — proven across Sheng Siong stores in Singapore.
Explore Food →