HVAC Oversize Trap: 500 kW Load, 650 kW Installed, RH Fixed at 52%
Introduction #
When this guide fits: You have a calculated cooling load and a vendor proposal that adds 30–50% “spare tonnage”—and operators complain about humidity, short cycling, or high demand kW despite “plenty of capacity.”
When it is not suitable: You need refrigerant leak diagnostics, chiller bearing repair, or ISO cleanroom certification—those are service/code deliverables, not load-vs-capacity screening.
Verified: 2026-05-25 — same 12,000 m² high-bay packaging site used in our Industrial HVAC design walkthrough. Below: load, installed capacity, and post-rightsize trends.
Case snapshot — load vs what was bought #
| Quantity | Value | Source |
|---|---|---|
| Modeled coincident peak (sensible + latent) | 500 kW | Owner load schedule, Aug design day |
| Vendor proposal (nominal cooling) | 650 kW | 130% of peak “for growth” |
| Installed (two packaged DX plants) | 650 kW | As-bid |
| Measured plant cooling demand (Jul peak week) | 412–468 kW | Submeter on CHW header |
| Target RH (production) | 50–55% | QA spec |
| As-found RH (humid week) | 62–68% | BMS trend |
Rightsizing target: 95–105% of verified peak, not nameplate nostalgia from rule-of-thumb W/m².
Run peak loads in the HVAC Capacity Calculator before accepting vendor oversize factors.
Why the 650 kW install underperformed #
Oversized sensible capacity does not fix weak latent paths. Symptoms we logged:
| Symptom | Trend evidence | Interpretation |
|---|---|---|
| Compressor cycles | 11 min average runtime | Massive sensible coil, poor latent control |
| Coil ΔT | Low 4–5 °F | Short circuit through coil, little dehumid |
| RH drift | >65% 06:00–10:00 | Ventilation moisture + oversized DX |
| Demand kW | +19% vs modeled | Inrush + part-load penalty |
Load breakdown that should have set capacity #
| Zone (see design guide) | Sensible kW | Latent kW | Coincident? |
|---|---|---|---|
| Lines 1–4 packaging | 248 | 42 | Yes |
| Warehouse mezz | 52 | 8 | Partial |
| QC lab | 38 | 22 | Yes |
| Offices | 28 | 6 | No (off-peak) |
| Coincident peak total | 412 | 88 | 500 kW |
Mistake in bid: Summed zone absolute peaks (568 kW) then added spare → 650 kW. Correct approach: coincident envelope = 500 kW.
Full zoning narrative: Industrial HVAC: Design Considerations.
Psychrometrics — why “tons” lied #
Peak outdoor (design day): 32 °C DB, 24 °C MC wet bulb. Target supply dew point 11 °C for RH 52% at 22 °C space.
| Path | SHR assumed | Result |
|---|---|---|
| Bid (single DX per plant) | 0.88 sensible-heavy | Latent underserved |
| Revised (DOAS latent + staged sensible) | Split OA dehumid | RH stable |
Rule: If ventilation OA > 15% and process moisture exists, model latent on peak humid hour, not only afternoon sensible peak.
Rightsizing actions executed (Q2 2026) #
| Action | Before | After |
|---|---|---|
| Nominal installed cooling | 650 kW | 520 kW (3 stages) |
| OA handling | Mixed in DX | DOAS latent wheel |
| Minimum turndown | One compressor | VFD fan + 3-step |
| Supply setpoint | 55 °F fixed | Reset 58–60 °F when load low |
Commissioning KPIs (30-day post) #
| KPI | Before | After |
|---|---|---|
| Avg RH (production hrs) | 64% | 53% |
| Compressor runtime / hr | 11 min on | 28 min on |
| Cooling plant kW (same OAT bin) | baseline | −19% |
| Peak demand kW (site) | 18,200 | 17,100 (HVAC share of energy case) |
Selecting capacity — decision table #
| Your calculated peak | Typical mistake | Better match |
|---|---|---|
| 500 kW coincident | Buy 650 kW “spare” | 520–540 kW staged + latent path |
| High exhaust bay | Add DX tonnage only | DOAS + sensible zoning |
| Low warehouse gain | Office W/m² on bay | Zone model + destrat fans |
Electrical ripple: revisit feeders with Factory Load Calculator when compressor stacks change.
Browse HVAC calculator hub for capacity and ACH/CFM tools.
Related articles #
- Industrial HVAC: Design Considerations — zone table for this site
- HVAC Sizing Guide
- HVAC Sizing Common Mistakes
- How to Calculate HVAC Capacity
Next steps you should take #
- Compare vendor kW to your coincident load schedule—not summed zone maxima.
- Trend RH, SAT, compressor min/hr for two humid weeks before buying spare tonnage.
- Split latent OA from sensible zone loads when RH matters.
- Re-run sizing in the HVAC Capacity Calculator after zone edits.
- Request indexing after measurable fixes (updated 2026-05-25).
Is 100% of calculated peak acceptable for equipment nominal?
Often yes with staging and a verified humid-hour check—target 95–105% of coincident peak, not 130–150% legacy spare.
Does rightsizing kill N+1 redundancy?
No. N+1 means spare machines, not each machine at 150% of one zone’s peak. Stage so one unit covers the coincident envelope.
Which KPIs prove rightsizing worked?
RH stability, compressor minutes per hour, and kW/ton or kW/m² vs a baseline—not nameplate tons installed.
Why did economizer “free cooling” still run compressors hard?
Stuck OA dampers and bad enthalpy logic forced mechanical cooling—trend mixed air before blaming capacity.
When is undersized capacity the real problem?
When trends show long run times, rising space DB, and RH still high with coils at full fan—then load model or airflow (filter DP), not “add 30% tons” by default.