Design of Ventilation & Air Conditioning Systems

An earlier version of this guide presented Course 1's envelope method as ASHRAE CLTD / CLF / SCL. That was wrong. The course teaches the Carrier Handbook of Air Conditioning System Design method — Equivalent Temperature Difference (ETD), Shading Coefficient, and Carrier solar heat gain tables. Both are legacy table methods that bury thermal lag in pre-computed values, but they are different tables, with different base conditions and different correction procedures. They are not interchangeable, and citing one when you used the other is the kind of error a reviewer will find.

The Carrier ETD tables are stated for 95 °F outdoor DB, 80 °F room, 20 °F daily range, 40° N latitude. Doha is 25.3° N at roughly 45 °C design dry bulb with a design wet bulb near 31 °C. Every one of those four table parameters is wrong for your projects. The method has correction procedures; the base tables do not transfer.

Per litre per second of ventilation air, relative to the room: Case 1 removes 11.5 g/h of moisture (a latent credit of 8.0 W per L/s), while Case 2 adds 58.3 g/h (a latent load of 40.5 W per L/s). Same building, same city, same code-required ventilation rate — and the latent duty swings by a factor of five and changes sign. Any load calculation that runs only the 45 °C condition has not seen the case that sizes the coil.

An occupant beside a large glazed elevation exchanges radiation with a hot surface and is uncomfortable while the air temperature is demonstrably within specification. Lowering the setpoint does not fix it — it overcools everyone else in the zone. It is fixed at design stage by glazing SHGC, external shading and perimeter treatment, which makes it an architectural interface item to raise at design review, not a snagging item to discover at handover. It is a recurring source of post-handover complaints that no amount of balancing will close out.

Both of the course's worked examples stop at the breathing-zone airflow. The Ventilation Rate Procedure does not stop there.

E_z accounts for how effectively supply air actually reaches the breathing zone. It is 1.0 for ceiling supply of cool air — so for a conventional office nothing changes, which is exactly why the omission goes unnoticed. Drop to 0.8 for ceiling-supplied warm air and the required zone outdoor air rises 25%. E_v corrects for the fact that in a multi-zone recirculating system the critical zone — the one with the highest outdoor-air fraction — governs the intake.

This is precisely the failure mode that produces an IAQ non-conformance after handover: the AHU outdoor-air damper is set to a total that satisfies the arithmetic sum, while one zone on the system never receives its share. It is invisible at FAT, invisible at TAB unless specifically tested, and surfaces as an occupant complaint or a CO&sub2; trend during the defects liability period. When reviewing a subcontractor's or package vendor's ventilation calculation, the presence or absence of E_z and E_v tells you immediately whether they applied 62.1 or just added two numbers together.

The course states the rule — solar heat gain is maximum on the west face, so avoid large west-facing glass. The reason matters more than the rule. West glazing takes its peak beam radiation in the late afternoon, at the same hour ambient dry bulb is at its daily maximum. Solar peak and sensible peak coincide on a west elevation. On an east elevation the solar peak lands in the morning while ambient is still low. That coincidence is why west exposure drives the building block load, not merely the zone load.

At Doha's latitude the sun sets closer to due west year-round and at a steeper terminal angle than at 40° N, so the western exposure period is more concentrated. Horizontal overhangs — which work well on a southern exposure — are largely useless against low-altitude western sun. Vertical fins, or SHGC control, are the effective responses. And note the hierarchy: external shading is thermodynamically superior to internal, because everything an exterior device absorbs or reflects is dissipated outdoors, whereas an internal blind intercepts energy that has already passed the glass and is inside the thermal envelope. This is an architectural coordination point that must be raised early; the HVAC design cannot fix it.

The course states its ETD table basis explicitly. Set it against your projects:

The Carrier method has correction procedures for the first three. It has none for the fourth that survives a 15° latitude shift intact. Answer the graded assessment with ETD and SC. Do not put either into a design basis memo without stating the source — a reviewer working to ASHRAE Fundamentals expects RTS or Heat Balance, and expects SHGC rather than SC.

Case 1 produces the larger room sensible load — 8 225 W against 6 876 W — because the 45 °C ambient drives conduction hard. Case 2 produces the larger grand total — 10 621 W against 9 503 W — because outdoor air at 23.8 g/kg carries a latent load of 2 699 W where Case 1's dry air carried only 826 W total.

So the two cases size different things: Case 1 sizes the airflow, the ductwork and the fan. Case 2 sizes the coil and the chiller. Running only the 0.4% dry-bulb condition — the habit most load calculations fall into — undersizes the cooling coil by 11.8% and leaves the building unable to hold humidity on exactly the days occupants complain about it.

Total load per person is 1.9× higher for dancing, but latent load is 2.6× higher. A high-activity space needs a coil selected for a much lower SHR — deeper dehumidification — not simply a bigger one. This is exactly the check to apply when reviewing a package vendor's AHU selection for a gym, prayer hall, canteen or assembly space: confirm the selection SHR matches the room SHR, not merely that the total kW is adequate. In a hot climate a larger share of occupant load is latent at any given room DB than the same occupancy would produce in a temperate one — one of several reasons Gulf latent duty runs proportionally higher. Note also: latent gain carries no cooling load factor, because moisture enters the air directly and instantaneously; only the radiant fraction of sensible gain is subject to storage lag.

The lecture presents Q = input × (1 − η) as general. It corresponds to the motor-losses-only case, and applied indiscriminately it substantially understates Case A — a pump or fan with both motor and driven equipment inside the conditioned space, where the entire electrical input becomes heat in that space. In-line pumps in a conditioned plant room, fan motors in a conditioned fan room, packaged equipment with integral motors: all Case A. This is a silent error — the calculation looks complete and the number is simply too small. Note also that ASHRAE applies load and use factors: a motor rarely runs at full nameplate continuously, and ignoring those overstates the load in the opposite direction.

The textbook stack diagram is upside down here. Indoor air at 24 °C is denser than outdoor air at 45 °C, so the buoyancy term changes sign:

Hot, humid outdoor air is drawn in at the upper levels and conditioned air is pushed out at grade. Upper-floor infiltration therefore carries a high latent penalty. Worse, the two drivers reinforce each other: wind speed is also higher near the top of a tall building, and in summer both wind pressure and stack effect promote infiltration there. That is why upper-floor perimeter zones in tall Gulf buildings are chronically hard to hold on humidity — and why the answer is envelope and riser sealing plus positive pressurisation, not more cooling.

Raise chilled water supply temperature far enough and the coil surface climbs above the entering air dew point. The coil then runs dry: it still satisfies the thermostat, its rated capacity is unchanged on paper, and space humidity climbs steadily with nobody able to say why. Temperature is satisfied; moisture is not being removed at all. This is a recurring finding on systems where chilled water setpoint has been raised to chase plant efficiency.

The moment fresh air is introduced, on-coil is the mixed-air condition — a mass-weighted mixture of return and outdoor air. Worked at Doha conditions, 24 °C room, 46 °C ambient, 20% outdoor air by mass:

The humidity-ratio error is far larger than the temperature error, because Doha outdoor air carries several times the moisture of the return air. The latent term dominates. Consequences compound in one direction: coil entering condition understated → coil undersized; latent capacity understated → space humidity not held even when temperature is satisfied; chilled water flow and ΔT selected against the wrong entering condition. Always compute mixed air from an actual mass and energy balance at the design outdoor condition and the design outdoor-air fraction. The error becomes severe wherever the fresh-air fraction is high — hospitals, laboratories, kitchens, any high-ACH space.

Solar, transmission, partition and lighting together come to 6 658 Btu/h. The IT load alone is 10 230. This is the “internal loads have grown as envelopes improved” trend made concrete, and it has continued past this course's vintage. The equipment schedule is the highest-risk input in the whole calculation, and it is an information dependency on the client, not an engineering judgement. A 25% error in assumed IT load moves the total more than a wholesale change of glazing specification — which is why an undocumented internal-load assumption is one of the two most common roots of an “the AHU is undersized” change-order argument.

The Heat Load Estimation lecture defines these as room sensible components. None of them appears in the calculation:

Note the irony: a blanket 12.5% safety factor is applied while several of the specific items it nominally covers are simply absent. That is the argument against blanket factors in one paragraph — they substitute for itemised rigour rather than supplementing it. When reviewing a submitted load calculation, the presence or absence of duct gain, duct leakage and fan heat as explicit line items tells you immediately whether it was built from a proper template or copied from a teaching example.

At Case 1 the coil sensible heat ratio is 0.998 — the outdoor air is dry enough to absorb the entire occupant latent gain on its own, and almost no condensate forms. That is comfortable, but it has three consequences worth writing down:

• The apparatus dew point construction degenerates. With the process line nearly horizontal, ADP and bypass factor stop being meaningful selection quantities at this condition. Select on Case 2.
• Humidity control is unavailable when you might want it. A dry coil cannot dehumidify. Any space needing tight humidity (archives, laboratories, some clean areas) needs a separate strategy, not a bigger coil.
• Condensate drainage is still not optional. The plant runs wet for most of the year — the coil is only dry at this specific design extreme.

The lecture makes the everyday point about a chilled bottle. The design consequence is larger. Dew point is the correct criterion for condensation on any cold surface: chilled water pipework, ductwork, structural elements bridging the envelope, glazing. In Doha, summer outdoor dew points routinely exceed the surface temperature of uninsulated chilled water pipe, which means insulation thickness for condensation control is set by the dew point, not by the heat gain calculation — and the two give different answers. Vapour barrier integrity matters more than thickness here: a breached barrier lets moisture condense inside the insulation, and the failure stays invisible until the insulation is saturated and the pipe is corroding. A recurring punch item, and entirely preventable at design and installation review.

These two things are not the same, and the gap is the point of this section. A graded assessment may well mark the lecture's answer correct. A design reviewer working to ASHRAE will not. Where the two diverge below, both are given — the As taught column is what earns the mark, the Correction column is what you put in a deliverable.

Four separate defects in this course are the same defect: it is written to an Indian code and climate base. Motors are certified to IS standards (Module 2); pipe standards are cited as IS 1239 and IS 3589 (Module 5); fire dampers cite NBC 2016 and CSIR-CBRI (Module 1); and the insulation worked example uses ASHRAE climate zone 1A — very hot and humid (Module 6).

For a Qatar project the motor standard is IEC, the pipe standard is ASME B36.10M (which the same lecture also cites), the damper chain is QCDD / NFPA 90A / UL 555 / UL 555S, and Qatar is climate zone 1B — very hot, dry, so the stated R-value has no standing. That is not a defect in the course; it is a defect in using the course unchecked. Every standard, climate zone and regulatory citation in this specialization needs verifying against your project stack before it is relied on.

Module 4 states the Darcy-Weisbach equation and then gives the friction coefficient as f = 16/Re laminar and f = 0.079/Re0.25 turbulent.

Module 6 performs the same calculation correctly, using Colebrook-White with absolute roughness. Where the two modules disagree, Module 6 is right. Note why the error is invisible inside the course itself: Module 4's own worked example reads friction from charts rather than from the equation, so its numbers come out fine. It will not come out fine in a calculation that uses the equation as stated.

The guide's existing argument — that a water-cooled machine rated at 29.4 °C entering condenser water will actually see about 40.6 °C from a Doha tower — is one half of the picture. The course supplies the other half: an air-cooled machine is rated at 35 °C entering condenser air.

A Gulf design ambient of 47–50 °C is 12 to 15 K above that datum, and air-cooled capacity falls with ambient precisely when the building load peaks. On a water-cooled machine the tower is limited by wet bulb, not dry bulb — which in a 47 °C DB / 30 °C WB climate is the more favourable reference. The course states neither point. It is the single most useful thing to carry from Module 3 into an equipment comparison: a nameplate air-cooled tonnage is not a Gulf tonnage, and the derating must be quantified from vendor curves rather than assumed.

AHRI 550/590 rates chillers at 29.4 °C entering condenser water. This plant, on a Doha design day, runs at 40.6 °C. Reversed-Carnot scaling across that gap, at a fixed evaporator leaving temperature of 6.7 °C:

On this 98 TR plant that is roughly 57 kW rising to 80 kW at design. Second-law efficiency is not actually constant across the range, so treat +40% as an argument, not an answer. The argument is this: a kW/TR figure with no stated condenser condition is not a number you can evaluate a bid on. Ask for the performance point at your condenser water temperature, and ask for it in writing.

A nominal tower ton is the capacity to cool 54 mL/s of water from 95 °F (35 °C) to 85 °F (29.4 °C) at 78 °F (25.6 °C) entering wet bulb. The course also states thermal capability as 1.25 kW of heat dissipation per kW of evaporator cooling. Those two statements look unrelated. They are the same statement.

The trap is the unstated basis. 54 mL/s is per kW of evaporator duty, not per tower ton. Read it as a tower figure and you undersize the condenser circuit by a factor of 3.5. The useful form to carry is the last line: a tower ton rejects 15,000 BTU/h at 3 GPM, and a chiller ton and a tower ton are not the same ton.

In a constant primary flow scheme, constant-speed pumps deliver fixed flow at all times and supply temperature varies to meet demand. A 3-way mixing valve at each AHU lets water bypass the coil at part load.

At design, water enters the coil at 44 °F and leaves at 54 °F, all flow through the coil, no bypass. At roughly 50% load the 3-way valve throttles the coil and lets 44 °F water bypass straight into the 54 °F return line.

Chiller capacity is proportional to flow × ΔT. With flow fixed at design, ΔT must collapse as load falls — and the mixed bypass water drives the return temperature entering the chiller lower still. Since peak conditions occur perhaps 1% of operating hours, the plant spends 99% of its life in this degraded state. The course describes this scheme as obsolete and no longer in use, and the reason is structural rather than operational: the 3-way valve is the mechanism.

Coils must be piped counterflow. Chilled water supply enters at the bottom, on the opposite side to the airflow; return leaves from the top. Counterflow lets the water and air temperature ranges overlap — water 44→55 °F against air 77→54 °F in the course's example. Parallel piping cannot achieve that, and it is a drawing-review check that takes five seconds.

Pressure testing: 1.5 × design working pressure, held 24 hours, any leak repaired and the system re-tested — and joints are insulated only after the test passes. But note error E-2: Module 5 specifies a concentric reducer at pump suction to prevent air pockets. That is wrong and contradicts Module 2, which correctly requires eccentric reducers at suction — flat side up, so no vapour pocket can form. Concentric belongs at the discharge.

Fan selection must be made at the final, dirty-filter pressure drop. The gap between the clean and dirty ΔP across a multi-stage bank is routinely the single largest uncertainty in an ESP figure — and it is a one-way error. A fan sized on clean-filter resistance cannot hold design airflow before the first filter change, and no amount of commissioning recovers it. Ask which value was used, and ask to see the filter data sheet that produced it.

A vendor quoting “total static pressure” is quoting ISP + ESP. A vendor quoting “external static pressure” is quoting only the duct side. Read which one the data sheet says, because this is exactly how a fan gets selected with insufficient capability — and note the related trap from Module 2: fan selection must be made at the final (dirty) filter pressure drop, not the initial. The gap between clean and dirty across a multi-stage filter bank is frequently the largest single uncertainty in the whole ESP calculation.

Everything about a Gulf summer invites you to size insulation on the 45–50 °C design dry bulb. That is the wrong hour. A hot, dry afternoon gives the surface film a large temperature difference to work with and the required thickness comes out small. The condensation case is the humid coastal night — a lower dry bulb sitting only a few kelvin above a very high dew point — and the required thickness there is two to three times larger. Size on the annual design dry bulb and the line will sweat in August at three in the morning.

The second conclusion is harder. ASHRAE recommends designing outdoor insulation to withstand 90% relative humidity — and at that humidity the +2 K surface margin is mathematically unreachable at any thickness, because the film cannot reject heat to air barely 2 K warmer than the surface. So the honest engineering position for an outdoor chilled water line in Doha is not a thickness. It is: state the design ambient and RH you have actually sized for, accept that brief surface wetting is possible outside it, and put the defence in the vapour barrier and the jacketing rather than in more insulation. A submittal quoting 25 mm with no stated design condition has not made that argument — and 25 mm does not satisfy any of the four rows above except the plant room.

Breakout is sound escaping between the duct walls and the silencer casing — a function of silencer form and size, casing gauge, and the acoustic media in the walls. The mitigation is positional, not material: put the silencer as near the source as possible, so the sound is attenuated before it travels to where breakout matters.

Which produces a drawing-review rule with real teeth. For airborne noise from plant room equipment, the best silencer position is in the plant room wall itself — it removes the sound as it leaves the room and solves breakout in the occupied space at the same time. But fire dampers are usually positioned on that same wall. The two cannot occupy the same penetration. When they clash, the silencer moves before the wall, inside the plant room; and only if there is no space there does it go outside. Check this on any plant room wall section: a silencer and a fire damper drawn at the same penetration is a coordination clash that will be resolved on site, badly, by whoever gets there first.

Module 1 is taught to the National Building Code of India (NBC) 2016; Module 2 to NBC 2016 Vol. 1 Part 4 and the ASHRAE Handbook — HVAC Applications, Ch. 52. Every criterion the course states — 50 Pa, 25–30 Pa, 6 ACPH, 12 ACPH, 250 °C for 120 minutes, 29 mg/m³ CO — is an NBC figure.

On a Qatar project the governing stack is the Qatar Civil Defence (QCDD) regulations, the project technical specification, QCS, and the NFPA references they invoke. The physics and the method transfer. The numbers do not. Re-derive every criterion from the governing documents before it reaches a deliverable — and note the course's own 50 Pa staircase figure happens to coincide with common practice, which makes it easy to carry across without noticing that its authority does not.

In the compressor room the code minimum is irrelevant — an engineer sizing from the ACPH table installs a fan at a quarter of the required duty, and the room runs roughly 21 K above ambient instead of 5 K. In the car park the opposite: the code minimum governs by a factor of 6.7 and the contaminant calculation is not binding at all.

Neither method can be assumed to control. Run both, take the larger, and read the sequence in the right order — compute by physics, then check the result is not below the code floor. ACPH is a floor, never an answer for a heat- or contaminant-driven space. Note the course leaves two gaps here: the CO emission rate E is never given (and cold-start emission factors have fallen by an order of magnitude over three decades, making a stale E the largest single error source), and the car park fire-mode rate is stated to be higher but never quantified — yet it is usually what actually selects the fan.

The fan is selected for a duty set by the five-open-door scenario. Ask the obvious question the course never asks. With all 17 doors closed, total leakage area collapses to 0.183 + 17×0.0213 = 0.545 m², and rearranging the orifice equation at rated flow:

Applying the door-force equation at 945 Pa gives approximately 1 460 N against a 133 N limit — an 11-fold exceedance. Nobody opens that door. The system designed to save lives becomes the thing that traps people in the stairwell.

Honest caveat, because the objection is obvious: a real fan rides up its curve and will not deliver rated flow against 945 Pa. Take a pessimistic-but-realistic 50% flow turndown and ΔP falls to 236 Pa — giving a door force of 395 N, still three times the limit. The conclusion is robust across the entire plausible range of fan behaviour. A barometric relief damper, a relief bypass, or VFD control on a stairwell pressure sensor is not an optimisation — it is the difference between a compliant system and a lethal one, and the lecture does not mention it once.

1. The design fire size, and who approved it. Everything scales from it. A 5 MW assumption presented without authority agreement is an unaccepted risk, not a design.
2. Make-up air. You cannot extract 96.8 m³/s from a sealed atrium. Make-up must be provided at low level below 1.0 m/s so it does not disturb the smoke layer — and it must be shown on the drawings, not assumed.
3. Door-force calculation at the maximum pressure, not just at the design pressure. See E2 above.
4. Fan and damper response times, and the cause-and-effect matrix that drives them, cross-checked against the fire alarm submittal.
5. The witnessed test procedure, agreed before installation. Smoke-control commissioning is a pass/fail demonstration in front of the authority; discovering the criteria at that point is far too late.

Lobby pressurization costs roughly 2.2 times the air — because the open lobby doors alone account for 6.3 m³/s, about 80% of the ground-floor duty, and they are governed by a velocity criterion rather than a pressure one. The course still prefers lobby pressurization, because it protects the stair as well as the well. That trade — double the fan duty in exchange for protecting two cores instead of one — is the argument to have ready whenever a chosen scheme has to be justified.

The 0.25 m² of smoke vents per lift well exists for one case: a fire originating inside the well itself. In that event the pressurization fan must not start — running it would distribute smoke to every floor of the building through the very shaft it was meant to protect. The vents discharge instead.

This is the provision practitioners most often omit, because it requires recognising that the system has a failure mode in which its normal operation is the hazard. The same logic drives the pressure relief damper: the calculation assumes a defined set of doors is open, and in the event those doors may be closed. The fan keeps delivering full design volume into a system whose leakage area has just collapsed, pressure rises, and the door-opening force exceeds the limit. It is the same failure as Worked example E2's 945 Pa — and it is most likely to be found at a witnessed test, because it only appears in the most favourable leakage case, which is not the case anybody calculates.

Hot combustion products rise by buoyancy, are trapped by the ceiling, and spread radially as a descending hot layer. The design objective is stated precisely and is worth memorising: control the rate of growth of the smoke layer so the clear layer stays above head height long enough for occupants to escape — not to remove all the smoke. The governing inequality is simply exhaust rate > generation rate. Watch also for the Coandă effect: in tall buildings hot gases escaping a compartment cling to the façade, travel vertically, and can re-enter at a higher level — which is why fresh-air intake location is a smoke-control decision, not just an IAQ one.

The example assumes an ambient of 303 K (30 °C). At a Gulf summer ambient near 320 K (47 °C) the same fire yields a smaller temperature ratio T_s/T_o, and therefore a different extract volume — the plume is less buoyant relative to a hotter ambient. This is the same class of error as the density assumption in the pressurization calculations, and it points the same way: every number in a temperate-climate worked example that depends on air properties needs re-deriving before it crosses the Gulf.

First: this zone's design condition corresponds to 1 387 ppm, not the 1 000 ppm that gets copied into specifications as though it were a code limit. Setting the controller to 1 000 ppm here would force more outdoor air than ASHRAE 62.1 requires, all day, every day — a demand control system that increases energy consumption.

Second: halving occupancy from 10 to 5 does not halve the outdoor air. It falls from 53.8 to 41.3 L/s — a 23% reduction, not 50% — because the area component R_a × A_z = 28.8 L/s does not depend on people and must keep flowing. In this zone the area term is the larger half of the requirement. DCV vendors who quote savings proportional to occupancy have not read the ventilation rate procedure.