Real-World Mathematics: Electricity, Energy, Power and Bills

Application of Mathematics in Real-World Usage · Worked application 5 · BTT Mathematics Hub

A kettle has the largest power rating in a small learning room. It must therefore be the main source of the room’s electricity bill. That conclusion sounds reasonable, but it leaves out the quantity that changes the answer: time. A powerful device used briefly can consume less energy than a smaller load that stays on throughout the day.

Electricity is a useful place to learn the difference between a rate and an accumulated quantity. It also exposes a second distinction: the total energy used during a month is not the same question as the greatest simultaneous load. One calculation helps explain consumption. Another describes demand at a particular moment. A third attaches a price to the energy. Combining them too early produces confident but misleading answers.

This guide follows one fictional learning-room audit from a list of devices to a checked decision. It develops unit conversions, weighted averages, percentage change, functions, interval reasoning and the beginnings of integration. Every device rating, operating pattern, price and meter reading in the worked cases is invented for teaching. None is a measurement of Bukit Timah Tutor, a product recommendation or a current Singapore tariff. The mathematics is educational, not an electrical installation or circuit-safety assessment.

Start with the question the bill cannot answer by itself

A bill can tell us how much electricity was recorded over a billing period and what charges were applied. It cannot, by itself, tell us which device caused every unit of consumption. To make that attribution, we need a model, measurements or both. The first decision is therefore not which formula to use. It is what claim we are trying to support.

“How much energy did these lights use?” is a device-level question. “Why did the total rise?” requires comparing whole periods on a consistent basis. “Can two tasks be moved apart?” concerns a time profile. “Did the change save energy?” requires a comparison in which occupancy, opening hours and other important conditions have not silently changed.

Write the question in a complete sentence before collecting numbers. Then write the boundary: which room, which devices, which dates and which operating hours belong to the calculation? An audit that includes only opening hours cannot explain a meter reading that also includes overnight loads unless those loads are added separately.

Watts describe power; kilowatt-hours describe energy

The U.S. Energy Information Administration’s measurement guide distinguishes power in watts from electricity use accumulated over time in watt-hours. One kilowatt is 1,000 watts. A constant one-kilowatt load operating for one hour uses one kilowatt-hour. The multiplication by time is essential.

For constant power, energy in kWh = power in kW × operating time in hours. Equivalently, energy in kWh = power in W × time in hours ÷ 1,000. Notice that kWh is kilowatt multiplied by hour, not kilowatt divided by hour. The latter would describe a change in power per hour, which is a different quantity.

Consider a hypothetical 1,800 W kettle used for ten minutes in total. Convert 1,800 W to 1.8 kW and ten minutes to 10/60 hour. Its energy is 1.8 × 10/60 = 0.30 kWh. A 55 W fan used for six hours consumes 0.055 × 6 = 0.33 kWh. Under these assumptions, the lower-power fan consumes slightly more energy.

This does not establish a universal ranking of kettles and fans. It establishes something more useful: a ranking cannot be inferred from power alone. Change the operating hours and the comparison can reverse. The same distinction appears when comparing speed with distance or water flow with volume.

Build a quantity ledger before calculating a total

Imagine a learning room with twelve lights, three fans, three laptops, one cooling device, one kettle and one router. Treat the stated powers as constant during each on-state. For the cooling device, suppose a simple two-state model: 1.2 kW while active and zero in the off-state. Its active fraction is assumed to be 60% during a six-hour session. This is a deliberately simplified model, not a claim about a particular air-conditioner.

Load	Teaching assumption	Energy for the model day
Lights	12 × 12 W, 6 hours	0.864 kWh
Fans	3 × 55 W, 6 hours	0.990 kWh
Laptops	3 × 45 W, 4 hours	0.540 kWh
Cooling	1.2 kW, 6 hours, active fraction 0.60	4.320 kWh
Kettle	1.8 kW, 10 minutes total	0.300 kWh
Router	10 W, 24 hours	0.240 kWh

The light calculation is 12 × 0.012 × 6 = 0.864 kWh. The laptop calculation is 3 × 0.045 × 4 = 0.540 kWh. The cooling calculation is 1.2 × 6 × 0.60 = 4.320 kWh. Add the six entries to obtain 7.254 kWh for the stated day.

Keep the count, unit power and operating time in separate columns. Writing only “lights: 144” loses information about whether 144 is watts, watt-hours or a count. A useful ledger makes every multiplier inspectable. Another reader should be able to reconstruct the total without asking what an unlabeled number means.

The decimal places in the arithmetic are exact consequences of the chosen inputs. They do not prove that actual energy use is known to three decimal places. This distinction between numerical calculation and measurement quality is developed in Measurement, Units, Scale and Estimation.

An active fraction is a weighted average, not an extra mystery

Suppose the cooling device is active for 60% of six hours. Its active time is 3.6 hours. Multiplying 1.2 kW by 3.6 hours gives the same 4.32 kWh as multiplying by six hours and then by 0.60. These are two representations of the same assumption.

The implied average power during the session is 0.60 × 1.2 = 0.72 kW. This does not mean the device must draw 0.72 kW at every moment. In the model it alternates between two states. The average compresses the time pattern; it does not replace it.

A more general two-state expression is average power = active fraction × active power + inactive fraction × inactive power. For example, a fictional device drawing 100 W for 40% of an interval and 10 W otherwise has average power 0.4 × 100 + 0.6 × 10 = 46 W. Treating the second state as zero would underestimate consumption.

Real equipment may have more than two states or continuously varying demand. A nameplate alone is not a record of the whole operating pattern. The educational method for combining wattage and time is also set out by The College of New Jersey’s energy-use guide. Our examples add explicit state assumptions so the estimate can be challenged rather than treated as a measurement.

Move from one day to a period without inventing a month

Suppose we study a 22-day observation window and explicitly assume that each of its days follows the model day. Total energy is then 7.254 × 22 = 159.588 kWh. Calling this a monthly total without checking the calendar would be a mistake. A calendar month may contain days with different operating patterns, and a router may continue running when the teaching room is closed.

For a 30-day month with 22 teaching days, use teaching-day energy for the teaching loads and 30 days for a router assumed to run continuously. Excluding the router, daily teaching energy is 7.014 kWh. The revised monthly model is 22 × 7.014 + 30 × 0.240 = 161.508 kWh. The extra eight router-days contribute 1.920 kWh.

This is a small numerical difference in this example, but a large conceptual distinction. The calendar is part of the model. We cannot multiply every line by the same number of days merely because one total is convenient. A weekend shutdown, a school holiday or an overnight background load changes the relevant time base.

Attach a price only after energy is established

For the 22-day observation window, use a fictional energy-only price of $0.30 per kWh. The charge is 159.588 × 0.30 = $47.8764, or $47.88 when rounded to cents at the end. This is not a complete real bill and does not include any assumed taxes, fixed charges or other services.

In Singapore, the Energy Market Authority’s regulated-tariff page is the appropriate source for the applicable regulated rate and its period. A teaching constant should never be silently presented as that current rate. For an actual calculation, use the rate that applies to the actual account and dates, and keep its inclusion or exclusion of tax explicit.

Keep energy and money as separate outputs. A larger bill can result from higher consumption, a higher unit price, a different billing interval or another charge. It does not logically prove that the devices became less efficient. Conversely, a smaller bill does not prove an energy reduction.

Separate consumption effects from price effects

Consider two fictional periods. In the first, consumption is 100 kWh and price is $0.30/kWh, giving $30. In the second, consumption is 80 kWh and price is $0.35/kWh, giving $28. Energy falls by 20%, but the charge falls by only $2, or about 6.67%.

One exact decomposition is: charge change = old price × energy change + new energy × price change. Here, 0.30 × (80 − 100) + 80 × (0.35 − 0.30) = −6 + 4 = −2. The energy reduction lowers the charge by $6 at the old price, while the price increase adds $4 to the new consumption.

Other decompositions can distribute an interaction term differently. State the convention rather than presenting one allocation as the only possible explanation. The total change, however, must reconcile exactly with the difference between the two charges. This is a useful application of algebra: a story about change should still balance.

Peak demand is a different question from accumulated energy

Assume all lights, fans and laptops are on, the cooling device is in its active state, and the router is running. Their simultaneous power is 0.144 + 0.165 + 0.135 + 1.200 + 0.010 = 1.654 kW. Turning on the kettle raises that modelled total to 3.454 kW.

Do not substitute the cooling device’s average power of 0.72 kW when the question explicitly asks about an instant during its active state. The average answers an energy question over an interval. The active-state value answers this instantaneous-load question. The EIA power-versus-energy distinction is the underlying reason these calculations must remain separate.

If the kettle runs at a different time, the time profile may change without changing its 0.30 kWh of energy use. Moving consumption is not automatically reducing consumption. A peak calculation also does not establish whether wiring, outlets or protection devices are suitable. This guide deliberately does not convert a classroom sum into permission to connect electrical equipment.

A time profile makes the hidden assumption visible

Imagine one hour split into three intervals: 0.5 kW for the first 30 minutes, 1.5 kW for the next 15 minutes, and 0.2 kW for the last 15 minutes. Energy is 0.5 × 0.5 + 1.5 × 0.25 + 0.2 × 0.25 = 0.675 kWh. Because the full interval is one hour, average power is 0.675 kW.

The simple mean of the three listed powers is about 0.733 kW. It is wrong for this time average because it gives equal weight to intervals of unequal length. The first state lasts twice as long as either of the other states. Time supplies the weights.

On a power-time graph, each constant interval forms a rectangle. Its width is time, its height is power, and its area is energy. Adding the rectangle areas connects ordinary multiplication to accumulation. A curve can be approximated by narrower intervals; integration is the advanced version of that same idea. The units remain kW multiplied by hours.

Compare a proposed change with its whole consequence

Suppose a hypothetical operating change reduces the cooling active fraction from 0.60 to 0.45 during each six-hour session. The modelled cooling reduction over 22 days is 1.2 × 6 × (0.60 − 0.45) × 22 = 23.76 kWh. At the fictional $0.30 rate, that component alone would reduce the energy charge by $7.128.

Now suppose the change requires three fans to run for two extra hours on each teaching day. Their additional energy is 3 × 0.055 × 2 × 22 = 7.26 kWh. Net reduction becomes 23.76 − 7.26 = 16.50 kWh, corresponding to $4.95 at the same fictional price.

The second calculation is not pessimism. It is boundary control. A claimed benefit should include the compensating inputs required to obtain it. This model also says nothing about whether comfort, air quality or learning conditions remain acceptable. Those requirements need their own evidence and must not be traded away simply because one arithmetic objective improves.

A proposed change should therefore name both its objective and its preserved conditions. “Reduce kWh while meeting the same service need” is more useful than “make the energy number smaller.” The distinction echoes the objective-and-constraint reasoning in Graphs, Functions, Optimisation and Systems.

A meter comparison is a test, not an automatic verdict

Return to the original 22-day model, which predicts 159.588 kWh. Suppose fictional meter readings at the exact start and end are 20,120.4 kWh and 20,288.6 kWh. Their difference is 168.2 kWh. Observed use exceeds the model by 8.612 kWh, about 5.12% of the meter difference.

That residual deserves investigation. It does not, on its own, identify a faulty meter, an undiscovered appliance or an inaccurate cooling fraction. All are hypotheses requiring further evidence. Start with simpler boundary questions: did the meter include another room, did the observation interval match, and were all overnight loads included?

A good next observation discriminates between explanations. If the unexplained difference grows mostly overnight, investigate background loads and boundaries. If it grows during busy sessions, investigate operating assumptions. Repeating the same total calculation with more decimal places will not reveal which explanation is correct.

Sensitivity shows where better evidence is worth collecting

Suppose the cooling active fraction is uncertain between 0.45 and 0.75. The other daily loads total 2.934 kWh. At the lower fraction, model-day consumption is 2.934 + 1.2 × 6 × 0.45 = 6.174 kWh. At the upper fraction, it is 8.334 kWh.

For 22 repeated model days, this gives 135.828 to 183.348 kWh. The interval spans 47.52 kWh, much more than the apparent precision of the original three-decimal total. It is a scenario range generated by assumed input bounds, not a statistical confidence interval.

The sensitivity can be stated directly: increasing the active fraction by 0.01 adds 1.2 × 6 × 0.01 = 0.072 kWh per model day. Over 22 days that is 1.584 kWh. This identifies the input whose measurement may be most useful. It also demonstrates why uncertainty belongs beside a result rather than buried in a footnote.

A battery has an energy limit and a power limit

For a purely numerical example, assume a battery is represented by a constant nominal 12 V and a charge capacity of 20 Ah. The simplified nominal energy is 12 × 20 = 240 Wh, or 0.24 kWh. If the model allows 80% of that nominal energy to be used and assumes 90% delivery efficiency, delivered energy is 0.24 × 0.80 × 0.90 = 0.1728 kWh.

A constant 60 W load would then have a modelled runtime of 0.1728/0.060 = 2.88 hours. This is not a product performance promise: the constant-voltage approximation, usable fraction, conversion losses and operating conditions are assumptions. Nor does sufficient stored energy prove the battery can supply any desired power. A separate maximum-output constraint may rule out a load.

The important mathematics is the distinction between “how much is stored?” and “how quickly can it be delivered?” A water tank can contain enough water for a task but have an insufficient outlet rate. The battery problem has the same rate-versus-stock structure.

Time-dependent prices create another layer, not another kind of energy

Imagine an entirely fictional two-price system: $0.20/kWh in one interval and $0.40/kWh in another. Using 60 kWh in the first and 40 kWh in the second costs $28. Shifting 20 kWh from the second interval to the first changes the allocation to 80 and 20 kWh, costing $24.

Total energy remains 100 kWh. The $4 reduction comes from the timing of consumption, not a reduction in consumption. Whether such a price structure actually applies is an account-specific question. These invented numbers demonstrate weighted sums, not a claim about the reader’s electricity contract.

This distinction matters when evaluating a claim such as “the bill fell, so efficiency improved.” A lower charge can result from a changed schedule even when every device consumes the same energy. Keep quantity, time allocation and price as separate columns until the interpretation is clear.

Compare like with like without hiding the total

If a room consumes 160 kWh during 80 occupied hours and later 180 kWh during 120 occupied hours, total use rises by 20 kWh. Yet energy per occupied hour falls from 2.0 to 1.5 kWh/hour. Both statements are true. They answer different questions.

Reporting only the total hides the growth in activity. Reporting only the normalised value hides the additional total consumption. A useful comparison shows both. More importantly, dividing by occupied hours does not by itself prove improved technical efficiency. Different weather, activities or occupancy density may still explain the change.

This is the role of the denominator: it makes comparisons more interpretable but does not magically create a controlled experiment. Use the ratios and rates guide when the arithmetic is secure but the comparison base is unclear.

A complete worked conclusion

A defensible conclusion from the fictional audit would read: “Under the stated device powers and time assumptions, the 22-day model predicts 159.588 kWh. The meter difference is 168.2 kWh, leaving an unexplained 8.612 kWh. Cooling is the largest modelled component, but its active fraction is uncertain. The proposed operating change gives a net modelled reduction of 16.50 kWh after including extra fan use. This is a prediction to test under comparable conditions, not an observed saving.”

Notice the sequence. The conclusion names the model, compares it with an observation, identifies a limitation and separates predicted change from measured change. It does not select the largest-looking device and invent a story. It provides enough information for another reader to disagree constructively.

Practice: twenty questions that require interpretation

Use only the assumptions stated in each question. All prices and equipment values remain fictional. Give units and keep at least one checking sentence for questions involving a decision.

1. A 900 W device runs for 20 minutes. Find its energy in kWh.
2. Eight 9 W lights operate for five hours. Find total energy.
3. A 40 W load runs continuously for 30 days. Find its energy.
4. A meter rises from 4,521.7 to 4,678.2 kWh. Find recorded use.
5. Price the use in question 4 at a fictional $0.28/kWh, with no other charges.
6. A 1.5 kW device is active for 40% of eight hours and otherwise draws zero. Find energy and interval-average power.
7. Calculate energy for 0.4 kW over two hours followed by 1.0 kW over half an hour.
8. Find the time-weighted average power for question 7.
9. Consumption falls from 250 to 210 kWh. Find the percentage reduction.
10. A 1 kW task and a 2 kW task each last one hour. Compare energy and peak if they run together versus consecutively, with no other load.
11. A change saves 18 kWh but adds another load consuming 5 kWh. Find net saving and its value at $0.32/kWh.
12. For 20 model days, an uncertain daily use lies between 4.2 and 5.1 kWh. Find the scenario range.
13. A model predicts 120 kWh while a matching meter interval records 132 kWh. Find the residual and its percentage of recorded use.
14. A fictional battery contains 500 Wh nominally; usable fraction is 0.8 and delivery efficiency 0.9. Estimate runtime for a constant 60 W load.
15. Ten kWh moves from a $0.45 interval to a $0.25 interval. Find the charge change and energy change.
16. Use is 150 kWh over 75 occupied hours. Find energy per occupied hour.
17. An adapter label gives a maximum rating. Does that establish its exact month-long energy use? Explain what is missing.
18. A 10 W router runs for a 31-day month. Calculate its energy without using the number of teaching days.
19. A device draws 80 W for 25% of an interval and 8 W otherwise. Find average power.
20. A bill falls while occupancy also falls. What comparison would help test whether the operation actually became more energy-efficient?

Worked answers and checking notes

1. 0.30 kWh. Convert to 0.9 kW and one-third hour; multiply. Treating 20 minutes as 0.20 hour would undercount time.
2. 0.36 kWh. Total power is 72 W, or 0.072 kW; multiply by five hours. The count belongs inside the calculation.
3. 28.8 kWh. The duration is 30 × 24 = 720 hours, so 0.040 × 720 = 28.8. A small continuous load can accumulate appreciable energy.
4. 156.5 kWh. Subtract the earlier cumulative reading from the later one. The individual readings are not each a period’s consumption.
5. $43.82. Multiply 156.5 by 0.28. This is the specified energy-only calculation, not a statement about a real tariff.
6. 4.8 kWh and 0.6 kW. Active time is 3.2 hours; average power is 1.5 × 0.4. Dividing energy by eight hours gives the same average.
7. 1.3 kWh. Add 0.4 × 2 and 1.0 × 0.5. Each rectangle must use its own duration.
8. 0.52 kW. Divide 1.3 kWh by 2.5 hours. The unweighted mean, 0.7 kW, ignores unequal durations.
9. 16%. The reduction is 40 kWh and the base is the original 250 kWh: 40/250 × 100%.
10. Both use 3 kWh. Together, peak power is 3 kW and completion takes one hour. Consecutively, peak is 2 kW and completion takes two hours. Timing changes the profile.
11. 13 kWh and $4.16. Subtract the additional input before multiplying by the price. Gross saving is not net saving.
12. 84 to 102 kWh. Multiply both bounds by 20. This is a scenario interval; no probability level has been established.
13. 12 kWh, about 9.09%. Use 12/132 because the question specifies recorded use as the denominator. The residual does not identify its cause.
14. 6 hours. Delivered energy is 500 × 0.8 × 0.9 = 360 Wh; divide by 60 W. Actual product runtime requires actual specifications and conditions.
15. A $2 decrease; no energy decrease. The same 10 kWh is priced $0.20 lower per unit. This is a timing benefit.
16. 2 kWh per occupied hour. This normalises activity but does not by itself control all conditions affecting consumption.
17. No. We need operating time, actual power through time and the relevant account boundary. A maximum rating is not a complete load history.
18. 7.44 kWh. Calculate 0.010 × 24 × 31. Non-teaching days still belong to a continuous-load calculation.
19. 26 W. Use 0.25 × 80 + 0.75 × 8 = 20 + 6. The inactive state contributes energy.
20. Compare total kWh and an appropriate activity-normalised measure under comparable conditions. Also inspect price, period length and important operating differences. A falling charge alone is insufficient evidence.

Teaching the transfer, not merely the formula

Begin with the kettle-and-fan comparison and ask the learner to predict before calculating. The useful misconception is visible immediately: power is being treated as energy. Next change only one duration. Ask what must remain the same and what can reverse. This trains sensitivity to structure rather than memorisation of which appliance supposedly consumes more.

Then provide the ledger with one missing unit, one mismatched time base and one nonzero standby state. Require the learner to repair the representation before calculating. For a more advanced learner, replace constant states with a time profile, ask for a weighted average and then connect rectangular areas to an integral. The arithmetic is a route to a larger skill: knowing what the calculation actually measures.

A useful final assessment is a short written conclusion that separates assumption, prediction and observation. A student who calculates 159.588 correctly but calls it a measured monthly bill has not yet finished the applied task. A student who identifies the boundary and explains the residual has begun to exercise mathematical judgement.

Sources, scope and the next useful route

The external sources support unit definitions, the energy-estimation method and the place to check an actual Singapore regulated tariff: EIA: Measuring electricity; TCNJ: Estimate Your Energy Use at Home; EMA: Buying at Regulated Tariff. All worked datasets, exercises and comparisons here are original teaching constructions, not figures supplied by those organisations.

Continue with Water, Flow, Tanks and Rainfall to see the same rate-and-accumulation structure with storage and overflow; Packaging, Cutting and Material Waste for physical capacity and whole-number constraints; or Scheduling, Critical Paths and Resources when the order of activities changes what is possible. Return to the BTT Mathematics Hub for the wider learning route.