First Principles for Building Services Engineers

A robust mental model for building services engineering starts from unbreakable universal laws and moves upward to the practical tools we use every day in design. When we understand this hierarchy, the calculations and decisions we make on projects stop feeling like isolated tricks. They become clear applications of deeper physics. Let us walk through the levels together, from the foundation to the final outputs.

The whole structure is easier to hold in your head as a picture. Pick something you would put on a drawing and the widget below lights every level it stands on.

1Interactive model

Trace a design task back to the axioms

Pick something you would actually put on a drawing. The widget lights every step it stands on, all the way down to the conservation laws, and sends a pulse up the chain. Nothing at the top exists without the levels beneath it.

Darcy-Weisbach gives the pressure drop and Bernoulli gives the velocity pressure. Both come out of Navier-Stokes, which is nothing more than momentum and mass written for a viscous fluid.

Base Level: The Axioms (First Principles)

These are the fundamental laws of physics. Nothing we design can ever break them. They form the ground we stand on.

Conservation of Mass

What enters a control volume must either leave or accumulate inside it. In simple terms, mass is never created or destroyed. When we look at air flowing through a duct or water moving through a pipe, the total mass flow rate coming in equals the total mass flow rate going out, plus any change stored inside the volume. This single idea lets us track every kilogram of air or water in a system.

$\frac{dm_{cv}}{dt} = \sum \dot{m}_{in} - \sum \dot{m}_{out}$

Here $m_{cv}$ is the mass held inside the control volume, and each $\dot{m}$ is a mass flow rate crossing its boundary in kilograms per second. The control volume is whatever boundary we choose to draw. A room, an air handling unit, a single coil, or the whole building. The equation does not care which one, as long as we account for every stream that crosses the line we drew. Room pressure regimes come straight out of this. If the supply fan moves more air than the extract fan, the surplus has nowhere to go, so it is stored as pressure until the leakage paths carry the difference away.

2Interactive model

What enters must leave or accumulate

Draw a boundary around one room and account for every kilogram crossing it. The supply and extract fans are yours to set. The leakage path is not, and it is what closes the balance. Time is slowed about five times, because the real room settles in under a second.

Supply fan0.25 kg/sExtract fan0.20 kg/sLeakage area200 cm²

Air at 20 °C, so a 1 Pa change in room pressure is about 2.4 g of stored air. The store is tiny, which is why the pressure moves so fast. Heat capacity of the same air, 242 kJ/K, is the slow one.

Supply beats extract by 0.05 kg/s. That surplus cannot vanish, so it is stored until the room pressure pushes the same 0.05 kg/s out through the leakage path. The room settles near 6.8 Pa positive.

Conservation of Energy (First Law of Thermodynamics)

Energy cannot be created or destroyed. It can only change form between heat, work, and internal energy. When a fan adds work to the air, or a cooling coil removes heat, the total energy balance must still close. This law forces us to account for every watt that enters or leaves a space or a piece of equipment.

$\frac{dE_{cv}}{dt} = \dot{Q} - \dot{W} + \sum \dot{m}_{in} h_{in} - \sum \dot{m}_{out} h_{out}$

$E_{cv}$ is the energy stored inside the boundary, $\dot{Q}$ is heat flowing in, $\dot{W}$ is work done by the control volume on its surroundings, and $h$ is the specific enthalpy each stream carries. A fan does work on the air rather than the other way round, so its $\dot{W}$ is negative and it adds energy.

For a steady air stream, with no change in height or speed worth counting, that reduces to the line every engineer already knows, $\dot{Q} = \dot{m} c_p \Delta T$ . The useful part is that the first law does not care where the energy came from. Take a fan that absorbs 3 kW, with its motor sitting in the air stream, moving 2.4 kg/s of air. All 3 kW ends up in the air. With $c_p$ for air at 1006 J/kg·K, the temperature of that air rises by

$\Delta T = \frac{\dot{W}}{\dot{m}\,c_p} = \frac{3000}{2.4 \times 1006} = 1.24 \text{ K}$

Nobody specified that heat and it still arrives, and on a cooling duty it is a load you then pay to remove again.

3Interactive model

Every watt has to land somewhere

Draw the boundary round the air handling unit. Heat enters at the coil and shaft work enters at the fan. The first law does not care which is which, so both raise the enthalpy of the same air stream. Turn the coil down to nothing and the fan still warms the air.

Mass flow

2.51

kg/s dry air

Rise from the coil

11.88

Rise from the fan

1.19

Off-coil

21.1 °C

from 8.0 °C

Volume flow2.00 m³/sOn-coil temperature8 °CCoil duty30 kWFan absorbed power3.0 kW

The fan adds 1.19 K on its own, which is 9 % of the total. Small on a heating duty, and never small enough to ignore on a cooling one.

Conservation of Momentum (Newton’s Second Law)

A fluid moves only when a net force acts on it. In building services that force is almost always a pressure difference. The pressure created by a fan or pump must overcome the resistance of ducts, pipes, filters, and fittings. If the driving force is weaker than the resistance, the fluid simply will not flow at the rate we want.

Write that for a column of air filling a straight duct. If the duct has cross-sectional area $A$ and the column has length $L$ , the mass being pushed is $\rho A L$ . The net force on it is the pressure difference acting over that same area, $A(\Delta p_{fan} - \Delta p_{loss})$ . Put both into $F = ma$ and the area cancels from each side, which leaves

$\rho L \frac{dv}{dt} = \Delta p_{fan} - \Delta p_{loss}$

Steady flow is not something we assume at the start. It is the state the air reaches when friction has grown to match the fan, so there is no net force left to accelerate anything. Friction rises with the square of velocity, so it catches up quickly, and in a duct the whole transient is over in about a second.

4Interactive model

Air moves only while a net force is left over

The fan pushes on one end of the air column and the duct walls rub against it. While the push wins, the air speeds up. Friction rises with the square of velocity, so it catches the fan up quickly and the flow stops changing. Restart from rest and watch the two bars close on each other.

Fan pressure rise200 PaSystem resistance at 5 m/s150 PaDuct length30 m

The air settles at 5.77 m/s, where friction has grown to match the 200 Pa the fan supplies. Longer duct means more mass to accelerate, so the same balance takes longer to reach, but it lands in the same place.

Entropy (Second Law of Thermodynamics)

Heat naturally flows from hot to cold. Every real process generates losses. Friction in ducts, inefficiencies in compressors, and mixing of different temperature streams all increase entropy. We can never achieve a perfect system. The second law reminds us that some energy will always be lost as low-grade heat.

$\dot{S}_{gen} = \sum \dot{m}_{out} s_{out} - \sum \dot{m}_{in} s_{in} - \sum \frac{\dot{Q}}{T} \ge 0$

$s$ is the specific entropy each stream carries, and $T$ is the temperature of the boundary at the point where that heat crosses it. The line says entropy out, minus entropy in, minus whatever the heat flows account for, is never negative.

Entropy feels abstract until you price it. Multiply the entropy generated by the outside temperature and you get work, in watts, that nobody can get back. Push 2 m³/s of air down a duct that loses 500 Pa to friction and the entropy generated is 3.41 W/K. At 20 °C ambient that is 1000 W of destroyed work, which is exactly the 1000 W of air power the fan supplied to push it. Friction did not waste part of the fan power. It wasted all of it.

5Interactive model

Energy survives the process. Quality does not

Mix two air streams, or drag air down a rough duct. Run the energy balance either way and it closes to the watt. Run the entropy balance and it only ever goes one way. Multiply the entropy generated by the outside temperature and you get the power you threw away.

Mixed temperature

20.0 °C

Energy in

40.2

kW above 0 °C

Energy out

40.2

kW above 0 °C

Entropy generated

1.17

W/K

Work you can no longer get

343

Stream 130 °CStream 210 °CFlow 11.0 kg/sFlow 21.0 kg/s

Mixing generates 1.17 W/K. At 20 °C ambient that is 343 W of work you can never get back, which is why a coil that fights a mixing box is a coil doing avoidable work.

These four axioms sit at the bottom of everything. They cannot be negotiated.

Level 1: Governing Equations (The Physics Engine)

When we apply the axioms to continuous fields of fluid and heat, we obtain exact mathematical descriptions. These equations are the complete physics engine of our world.

Navier-Stokes Equations

These are the full mathematical statement of momentum conservation applied to a viscous fluid such as air or water. They link velocity, pressure, temperature, and density at every point. In theory they describe every swirl, every pressure drop, and every temperature change. In practice they are extremely difficult to solve by hand for the complex geometries we meet in buildings.

$\rho\left(\frac{\partial \mathbf{u}}{\partial t} + (\mathbf{u} \cdot \nabla)\mathbf{u}\right) = -\nabla p + \mu \nabla^2 \mathbf{u} + \rho \mathbf{g}$

$\mathbf{u}$ is the velocity at a point and $\mu$ is the dynamic viscosity. Read left to right, the terms are: how fast the velocity is changing at a fixed point, how much it changes as a parcel of air moves along, the push from the pressure gradient, the drag from viscosity, and the weight of the fluid. There is nothing about buildings in there, and nothing about buildings anywhere below Level 2. That is the point. Every tool further up is this equation with terms removed.

Equations of State

These equations connect pressure, volume, and temperature for a fluid. The Ideal Gas Law is the most familiar example. From equations of state we build the thermodynamic property tables that tell us enthalpy, density, and specific heat for air, water, and refrigerants. Without them we could not convert between temperature readings and energy quantities.

$p = \rho R T$

$R$ is the specific gas constant, 287 J/kg·K for dry air, and $T$ has to be in kelvin rather than degrees Celsius. For air at 101325 Pa and 20 °C that gives 1.204 kg/m³. Now take a fan handling 2 m³/s and move it to a station pressure of 90 kPa, roughly 1000 m above sea level. The density falls to 1.070 kg/m³. The volume flow has not changed at all. The mass flow has fallen by 11 %, and so has every duty, every fan pressure, and every energy figure that depends on it. This is why 101325 Pa is a default and not an assumption.

6Interactive model

From p, v and T to the numbers on the drawing

Fix the volume flow at 2 m³/s and move the air's pressure and temperature. The ideal gas law sets the density, and the density is what decides how much air you actually have. Duty, mass flow and fan pressure all move with it. Volume flow on its own tells you nothing about energy.

Density

1.204

kg/m³

Mass flow

2.41

kg/s at 2 m³/s

Duty for 10 K

24.2

Altitude

m, ISA

Temperature20 °CAbsolute pressure101.3 kPa

At sea level and 20 °C the density is 1.204 kg/m³. Every other line on this widget is that number multiplied by something.

Together these governing equations give us the complete, exact picture. They are powerful, but they are also too heavy for everyday design calculations.

Level 2: Engineering Approximations (The Working Tools)

Because the full Navier-Stokes equations are impractical for routine duct and pipe work, we simplify them. We remove certain terms under carefully chosen assumptions and obtain tools we can actually use at a desk or on site.

It is worth seeing that removal happen. Each assumption deletes a term, and each deleted term costs a prediction we can no longer make.

7Interactive model

Delete a term, gain a tool, lose a prediction

Navier-Stokes is complete and unusable. Every equation you can actually work with is this one with terms crossed out. Switch the assumptions and watch which equation you land on, and what the surviving model can no longer tell you about a plain 20 m duct.

ρ ∂u/∂t+ ρ(u·∇)u= −∇p+ μ∇²u+ ρg

Kept as a field equation: every quantity still varies in three directions and you need a computer.

What you are left with

Navier-Stokes, complete

Reduced form

No closed form. This is what a CFD solver integrates, cell by cell.

Predicted loss over 20 m

20.2 Pa

actual 20.2 Pa

Solvable at a desk

With the friction term in, this duct loses 20.2 Pa, or 1.01 Pa per metre. That is a number you can size a fan on.

Bernoulli’s Equation

This is an extreme simplification of Navier-Stokes. We assume the fluid is frictionless (inviscid) and incompressible, and that it moves along a single streamline. Under those conditions the equation shows that static pressure and dynamic pressure trade places with each other. That relationship is the reason a Pitot tube can measure velocity, and it also explains the basic head that fans and pumps must supply.

$p + \tfrac{1}{2}\rho v^2 + \rho g z = \text{constant}$

The three terms are the static pressure, the velocity pressure, and the height term, with $z$ measured from any datum you choose. For air the height term is small enough to drop across the depth of a plant room. Take air at 1.2 kg/m³ moving at 10 m/s:

$p_v = \tfrac{1}{2}\rho v^2 = \tfrac{1}{2} \times 1.2 \times 10^2 = 60 \text{ Pa}$

Turn it around, and a Pitot tube reading 60 Pa is telling you the velocity:

$v = \sqrt{\frac{2 p_v}{\rho}} = \sqrt{\frac{2 \times 60}{1.2}} = 10 \text{ m/s}$

That square root is the entire theory behind a velocity traverse.

8Interactive model

Static and velocity pressure trade places

Bernoulli says total pressure is held along the streamline, so anything the air gains in velocity pressure it gives up in static pressure. Slide the probe into the throat and watch the swap. Squeeze the throat far enough and the static pressure goes below the pressure outside the duct.

Local diameter

250 mm

Local velocity

15.28

m/s

Velocity pressure

140

Static pressure

110

Volume flow0.75 m³/sThroat diameter250 mmProbe position50 %

The probe reads 140 Pa of velocity pressure, and √(2·140/1.2) gives 15.28 m/s. That square root is the entire theory behind a Pitot traverse.

Darcy-Weisbach Equation

Bernoulli’s equation ignores friction. Real ducts and pipes have friction. The Darcy-Weisbach equation puts friction back in by calculating the pressure drop caused by viscous shear against the walls. Almost every duct-sizing and pipe-sizing calculation we perform rests on this equation. It is the bridge between ideal flow and the real losses we measure on site.

$\Delta p = f \frac{L}{D} \cdot \frac{\rho v^2}{2}$

$L$ is the length of the run, $D$ is the diameter, and $f$ is the friction factor, which depends on the Reynolds number and on how rough the duct wall is. Twenty metres of 400 mm galvanised duct running at 6 m/s carries 0.75 m³/s. The Reynolds number is about 159 000, the Colebrook-White friction factor works out at 0.0186, and the pressure drop is 20 Pa, or 1.0 Pa per metre. Double the velocity to 12 m/s and the drop rises to 75 Pa. That is a factor of 3.7 rather than a clean 4, because $f$ falls slightly as the flow becomes more turbulent, but the square is the part worth remembering.

9Interactive model

The square on the velocity is the one that costs money

This is the equation almost every duct and pipe on a project is sized with. Move the flow and watch the drop climb far faster than the flow does. Then move the diameter one step and watch most of it disappear again. The dashed curve is the next size up.

Velocity

5.97

m/s

Reynolds number

158803

turbulent

Friction factor

0.0186

Colebrook-White

Pressure drop

20.0

Pa over 20 m

Per metre

1.00

Pa/m

Diameter400 mmVolume flow0.75 m³/sLength20 m

Double the flow through this duct and the drop goes from 20 Pa to 75 Pa, a factor of 3.73. It falls just short of four because the friction factor drops a little as the Reynolds number rises. Go one size up instead, to 450 mm, and the same flow costs 11 Pa.

Psychrometric Principles

These principles apply mass conservation and energy conservation directly to mixtures of dry air and water vapour. By tracking the dry-air mass flow and the water-vapour mass flow separately, and by tracking enthalpy, we can calculate exactly how much cooling, heating, or dehumidification a coil must provide. The psychrometric chart is simply a visual map of these two conservation laws working together.

Written out, that is three sums. Dry air is conserved, water is conserved, and energy is conserved.

$\dot{m}_{a,in} = \dot{m}_{a,out} = \dot{m}_a$

$\dot{m}_a W_{in} = \dot{m}_a W_{out} + \dot{m}_w$

$\dot{Q} = \dot{m}_a \left[ (h_{in} - h_{out}) - (W_{in} - W_{out}) h_f \right]$

$\dot{m}_a$ is the dry-air mass flow, $W$ is the humidity ratio in kilograms of water per kilogram of dry air, $\dot{m}_w$ is the condensate running to the drain, $h$ is the enthalpy of the moist air per kilogram of dry air, and $h_f$ is the enthalpy of that condensate as it leaves. Take 2 m³/s of air arriving at a coil at 28 °C and 50 % relative humidity, and leaving at 13 °C and 95 %. The humidity ratio falls from 0.0118 to 0.0089 kg per kg of dry air, and the enthalpy from 58.4 to 35.4 kJ/kg. The specific volume on-coil is 0.869 m³/kg, so the dry-air mass flow is 2.30 kg/s. The total duty comes out at 52.3 kW. Of that, 35.5 kW is sensible and 16.8 kW is latent, a sensible heat ratio of 0.68, and 24.6 kg of water an hour runs to the drain. No chart was read to get any of those numbers.

10Interactive model

A cooling coil is three sums, not a chart lookup

Dry air in equals dry air out. Water in equals water out plus what runs to the drain. Energy in equals energy out. Move the on-coil condition and the off-coil temperature, and the sensible and latent split appears on its own. The plot on the right is only a picture of those sums.

Dry air

2.30

kg/s

Total duty

52.3

Sensible

35.5

Latent

16.9

Sensible heat ratio

0.68

wet coil

Condensate

24.6

kg/h

On-coil temperature28 °COn-coil humidity50 %Off-coil temperature13 °CVolume flow2.0 m³/s

The coil is below the on-coil dew point, so 24.6 kg of water an hour leaves the air stream. That is 16.9 kW of latent duty on top of 35.5 kW of sensible, giving a sensible heat ratio of 0.68.

These approximations are not perfect, but they are accurate enough for design and they remain firmly rooted in the original axioms.

Level 3: Building Services Applications (The Output)

At the top of the hierarchy we combine the tools into the deliverables that appear on drawings and in energy models.

Aeraulic and Hydronic Design

We balance the performance curve of a fan or pump against the pressure-drop curve of the duct or pipe system. The intersection of the two curves gives the actual operating point. Everything we learned about Bernoulli, Darcy-Weisbach, and conservation of mass comes together here so that the right volume of air or water reaches every terminal unit.

$\Delta p_{sys} = k_{sys} \dot{V}^2$

The system curve is Darcy-Weisbach rewritten in terms of volume flow, with everything fixed about the ductwork rolled into the one coefficient $k_{sys}$ . The fan curve is not a formula at all. It is a line the manufacturer measured on a test rig, so we work with it as a curve.

Take a system that needs 500 Pa at 2 m³/s, so $k_{sys}$ is 125 Pa per (m³/s)². Put it with a fan whose measured curve fits $\Delta p_{fan} = 800 - 75\dot{V}^2$ , in pascals with $\dot{V}$ in m³/s. Setting the two equal gives $200\dot{V}^2 = 800$ , so the flow is 2 m³/s at 500 Pa, as intended. Now let a filter load up until $k_{sys}$ doubles to 250. The intersection moves to 1.57 m³/s at 615 Pa. The flow fell by 22 %, and the air power went from 1000 W to 965 W. Throttling bought us less air at almost the same cost, which is the whole argument for slowing the fan instead.

11Interactive model

Only the intersection is real

The rising curve is the system, which is Darcy-Weisbach plotted out. The falling curve is the fan, measured on a test rig. Neither one is an operating point on its own. Close the damper and the flow slides back up the fan curve, without anyone asking it to.

Volume flow

2.00

m³/s

Pressure

500

Air power

1.00

Against design

100 %

of 2.00 m³/s

Fan speed100 %System resistance1.00 ×

At 100 % speed the fan and the system agree on 2.00 m³/s and 500 Pa. Move either curve and the point moves with it. There is no third option.

Thermal Dynamics

We apply the heat-transfer laws (Fourier’s law for conduction, Newton’s law of cooling for convection) together with the energy balance to predict temperatures, overheating risk, and energy use. Daylighting calculations and dynamic energy models are further extensions of the same conservation principles.

$\dot{q} = -\lambda \frac{dT}{dx} \qquad \text{and} \qquad \dot{q} = h_c (T_s - T_\infty)$

$\dot{q}$ is a heat flux in watts per square metre. In the first, $\lambda$ is the thermal conductivity of the material and $dT/dx$ is the temperature gradient through it. In the second, $h_c$ is the surface heat transfer coefficient, $T_s$ is the surface temperature, and $T_\infty$ is the air temperature away from the surface. It is $h_c$ here and not $h$ , because $h$ is already the enthalpy above.

Those two set the heat crossing the fabric. The energy balance then turns them into a rate of change of stored heat, which is the only thing a dynamic model is really doing.

$C \frac{dT}{dt} = \dot{Q}_{gain} - (UA)(T - T_{out})$

$C$ is the heat capacity of everything inside the boundary, in joules per kelvin, and $UA$ is the total heat loss in watts per kelvin, fabric and ventilation together. A 50 m² office with 45 W/K of fabric loss and 2 air changes an hour has a $UA$ of 146 W/K. Give it a medium-weight construction at 80 kJ per m² of floor per kelvin and $C$ is 4 MJ/K, so the time constant $C/UA$ is 7.6 hours. That number is why the warmest hour inside is not the warmest hour outside, and why thermal mass changes when a room overheats as much as whether it overheats.

12Interactive model

Run the same balance forward in time

Fourier and Newton give the heat crossing the fabric. The energy balance turns that into a rate of change of stored heat, and stored heat is what thermal mass is. Change the construction and the peak does not just shrink, it arrives later. This is the whole content of a dynamic thermal model.

Peak indoor

28.6 °C

Peak at

17:58

3.0 h after outside

Hours above 26 °C

6.9

per day

Time constant

7.6

Total heat loss

146

W/K, 101 of it ventilation

Internal gains1200 WVentilation2.0 achFabric heat loss45 W/KOutdoor mean19 °COutdoor swing± 7 K

Medium construction gives a time constant of 7.6 hours, so the indoor peak lands 3.0 hours after the outdoor one and reaches 28.6 °C. The heat did not go anywhere. It went into the structure and came back out later.

When we design a system this way, we are never inventing rules from thin air. We are simply choosing the right level of simplification for the problem in front of us, always knowing that the deeper laws remain in force.