Energy figures become useful only when the units and conditions agree
A table says that one imaginary energy-supply plan delivers 960 GWh in a year at an operating expenditure of S$120 million. Another plan delivers 840 GWh for S$84 million. It may seem obvious that the second is cheaper and the first produces more electricity. Both statements are true in a limited sense. But if a fictional community needs at least 900 GWh annually, the cheaper plan fails that condition; if the question asks for cost per kilowatt-hour, neither headline total is the final answer. In G3 SEC Mathematics, the correct conclusion depends on what is being compared.
This guide uses invented figures for arithmetic practice, not predictions about real nuclear plants. It follows Volume 0508’s Physics lesson on fission and fusion, but its own task is mathematical: translate energy quantities to common units, handle costs and percentages, test constraints and explain what a model can actually show. The 2027 G3 Mathematics syllabus is K310. It assesses mathematical understanding, real-world problem solving, reasoning and clear communication, not specialist reactor engineering.
- Write each figure with its unit and time period before calculating.
- Convert GWh into kWh when a cost is expressed per kilowatt-hour.
- Test the minimum demand condition before making a recommendation.
- Keep annual totals separate from monthly peaks, forecasts and measured results.
- Show working, interpret the answer, and qualify any conclusion that needs more data.
Read the official G3 Mathematics examination rules first
The official 2027 SEAB G3 Mathematics code is K310, corresponding to the former O-Level Mathematics code 4052. Paper 1 lasts two hours fifteen minutes, carries 90 marks and contributes 50% of the total assessment. It has approximately 26 compulsory short-answer questions. Paper 2 also lasts two hours fifteen minutes, carries 90 marks and contributes 50%, with 9–10 compulsory questions and a final question focused specifically on applying mathematics to a real-world situation.
The syllabus specifies that essential working matters. Non-exact numerical answers should ordinarily be given to three significant figures, or angles to one decimal place, unless the question instructs otherwise. This is why a clear conversion and a justified decision are more valuable than a calculator result without units. A proposed facility is only the setting. The assessment is whether you can reason accurately with the numbers and the conditions supplied.
Check the 2027 G3 syllabus directory and K310 Mathematics syllabus for the exact rules, current when reviewed on 1 November 2026. The separate K323 Physics rules appear in the previous volume. EMS is a study integration used on this site; English, Mathematics and Science are assessed in their respective official subjects rather than in a single combined EMS paper.
The main fictional case file
An imaginary planning team considers two alternative electricity-supply contracts. Plan A forecasts annual delivery of 960 gigawatt-hours (GWh) with annual operating expenditure of S$120 million. Plan B forecasts 840 GWh with annual operating expenditure of S$84 million. The required annual electricity supply for the exercise is at least 900 GWh. The costs are invented accounting figures, not consumer tariffs, capital investments or verified market prices. Both output numbers are forecasts rather than actual metered annual deliveries.
A brochure claims: ‘Plan B is always the sensible choice because its yearly cost is lower.’ A second leaflet says: ‘Plan A is clearly best because it delivers more.’ Neither slogan has finished the mathematical evaluation. A good answer asks which condition is compulsory, which quantities are being compared and which information is absent. We can show that A passes the supplied annual-demand threshold while B does not. We can also compare their stated expenditure per forecast unit of electricity. We cannot infer overall national policy or lifetime technology costs from these figures.
Keep the case data in one place. Plan A: 960 GWh, S$120,000,000, minimum requirement 900 GWh. Plan B: 840 GWh, S$84,000,000, same requirement. The labels A and B deliberately avoid equating fictional cost claims with actual nuclear fission or fusion performance. If a later question explicitly calls one concept a fission-based proposal and the other a future fusion-based proposal, that extra label does not make the figures verified real-world predictions.
Start by converting energy, not by comparing raw digits
One kilowatt-hour (kWh) is a unit of energy. One megawatt-hour is 1,000 kWh. One gigawatt-hour is 1,000,000 kWh. Therefore 960 GWh equals 960,000,000 kWh, and 840 GWh equals 840,000,000 kWh. Write the conversion as a multiplication before entering a calculator. The number of zeros matters when a currency total in millions is divided by a very large energy quantity.
Do not confuse GWh, which is energy, with GW, which is power. GWh is a quantity of energy delivered over an interval. GW measures a rate of energy transfer or production. A question about annual GWh cannot be answered by reporting an instantaneous GW value unless the relevant time information is supplied and the intended calculation is clear. The letters differ by one character, but their meanings differ substantially.
An earlier PSLE Mathematics diagnostic guide teaches how to separate representation errors from arithmetic errors. At G3, the same discipline remains essential. If you divide S$120 million by 960 rather than by 960 million kWh, the calculator will give 125,000 instead of S$0.125 per kWh. The division is mechanically correct for the numbers entered, but the unit model is wrong.
Calculation one: unit operating expenditure
For Plan A, divide its stated annual operating expenditure by forecast annual energy delivered. S$120,000,000 ÷ 960,000,000 kWh = S$0.125 per kWh. For Plan B, S$84,000,000 ÷ 840,000,000 kWh = S$0.100 per kWh. On this narrow measure, B has the lower expenditure per forecast kilowatt-hour. The difference is S$0.025 per kWh, or 2.5 Singapore cents per kWh.
The calculation does not automatically tell you the electricity bill that a household would pay. It deliberately uses only the fictional operating-expenditure amounts supplied, and omits all unspecified factors such as financing, grid arrangements, additional fees and subsidies. In an examination answer, use the wording ‘stated operating expenditure per forecast kWh’, not ‘consumer tariff’, unless the question defines the latter.
A useful reasonableness check is to think in millions. The ratio S$120 million divided by 960 million kWh should produce a number well below one dollar per kWh, not hundreds or thousands of dollars. Canceling equal factors of one million gives 120 ÷ 960 = 0.125 directly. That is a legitimate simplification only because both quantities were expressed using consistent million factors.
Calculation two: the minimum requirement decides feasibility
The fictional annual requirement is at least 900 GWh. Plan A forecasts 960 GWh and therefore exceeds the requirement by 60 GWh. Its surplus over the threshold is 60 ÷ 900 × 100% = 6.67% to three significant figures. Plan B forecasts 840 GWh and falls short by 60 GWh. Its shortfall relative to the requirement is also 6.67%. Equal-sized deviations in opposite directions do not imply equal suitability.
If the question states that the 900 GWh threshold must be met, Plan B is not feasible on the supplied forecast, even though its expenditure per forecast kWh is lower. The correct conclusion is conditional: ‘Among these two plans, Plan A alone meets the stated minimum annual supply on the forecast values.’ Do not turn that into ‘Plan A is always best’ because the model may leave out other decision criteria and forecast uncertainty.
If the question instead asks for the less expensive option per kWh, the answer is Plan B. If it asks which has greater annual output, Plan A. If it asks which satisfies the minimum forecast, Plan A. One dataset can support different answers to different questions. The task wording is part of the mathematics, not decoration around it.
Calculation three: distinguish differences from percentage differences
Plan A’s forecast output exceeds Plan B’s by 120 GWh. Compared with Plan B as the reference, the percentage increase is 120 ÷ 840 × 100% = 14.2857…%, or 14.3% to three significant figures. Compared with Plan A as the reference, the corresponding reduction from A to B is 120 ÷ 960 × 100% = 12.5%. Both are valid percentages for different questions.
Students often lose accuracy by changing the denominator silently. The phrases ‘A is what percentage higher than B?’ and ‘B is what percentage lower than A?’ do not use the same base. Write the reference immediately next to the formula. The difference is still 120 GWh, but the percentage depends on what that difference is being measured against.
Do the same for expenditure. A costs S$36 million more annually than B under the fictional operating figures. Relative to B, that is 36 ÷ 84 × 100% = 42.857…%, or 42.9%. Relative to A, B is 30% lower because 36 ÷ 120 × 100% = 30%. A clean working line prevents a persuasive but misleading percentage claim.
Calculation four: compare the costs on the same energy quantity
Imagine the question asks what the two stated unit expenditure rates would imply for 900 GWh, provided the cost per kWh remains constant. Plan A’s model rate, S$0.125 per kWh, applied to 900,000,000 kWh gives S$112.5 million. Plan B’s model rate, S$0.100 per kWh, gives S$90 million. The arithmetic difference is S$22.5 million on that common quantity.
But this calculation is a hypothetical scaling exercise. The original B forecast delivers only 840 GWh, so the calculation does not prove that Plan B can supply 900 GWh. It assumes a constant unit rate and a supply quantity that B has not been shown to achieve. In a fully explained response, show the arithmetic and state the missing capacity condition. A model can calculate a hypothetical cost without establishing practical feasibility.
This distinction between ‘what would it cost if the rate applied?’ and ‘can the option actually deliver?’ is one of the most useful features of real-world Mathematics. Do not discard a valid calculation because the scenario is unrealistic, but do not let a neat result override a stated constraint. Give the calculation its proper scope.
Calculation five: average monthly output is not a guaranteed monthly output
Dividing Plan A’s 960 GWh annual forecast by twelve gives an average of 80 GWh per month. Plan B’s 840 GWh gives an average of 70 GWh per month. That average is mathematically correct when the annual total is treated as a sum over twelve months. It does not establish that every calendar month delivers exactly 80 or 70 GWh.
A community might use more electricity in one month than another, and a facility’s output might vary. The annual average says nothing about the shape of the twelve monthly figures unless further information is supplied. If a fictional report promises a minimum of 78 GWh in every month based only on Plan A’s annual average of 80, the promise is not justified. An average can be above a threshold while individual observations fall below it.
Explain the missing data rather than guessing. ‘The annual average is 80 GWh, but the report needs monthly output records to determine whether each month exceeds 78 GWh.’ This is a strong interpretation because it separates what the computation establishes from what would need to be measured or forecast at finer resolution.
Calculation six: a quarter of data cannot automatically stand for a year
A new fictional table gives Plan A’s first-quarter monthly outputs as 74, 79 and 77 GWh. Their total is 230 GWh, and their average is 230 ÷ 3 = 76.666… GWh. Multiplying that three-month average by twelve produces an annualised projection of 920 GWh if the same average continues throughout the year. It is not an observed annual total and not the original 960 GWh annual forecast.
For Plan B, let the first-quarter values be 71, 70 and 72 GWh. Their total is 213 GWh; the average is 71 GWh; the annualised projection is 852 GWh under the constant-average assumption. The fact that the quarterly annualisation differs from the original forecast is not automatically proof of fraud or calculation error. It may reflect seasonality, changing operations or uncertainty. The question needs more information.
In your answer, label the result as ‘annualised from first-quarter data’ and state the assumption. A student who writes ‘Plan A produced 920 GWh this year’ has changed a projection into a historical measurement. The numerical multiplication is right, but the interpretation is false. A full-mark response preserves both the value and its status.
Calculation seven: use percentages as multipliers
Suppose a fictitious report expects Plan A’s next-year output to rise by 7% from the 960 GWh forecast baseline. The new figure is 960 × 1.07 = 1,027.2 GWh. If the following year then falls by 7%, multiply by 0.93 to get 955.296 GWh. The two changes do not cancel, because the second percentage is applied to a different base.
This is a frequent source of mistakes. Adding 7% and subtracting 7% as if they were simple points of an unchanging total would incorrectly return to 960. The correct two-step multiplier is 1.07 × 0.93 = 0.9951. The final amount is 99.51% of the original, a net reduction of 0.49%.
If the question asks for a result to three significant figures, use 955 GWh, but retain sufficient accuracy in intermediate working. State the reference year and the assumption. Without those labels, a future output forecast can be mistaken for current measured production. Compound percentage reasoning is a mathematical idea, not a guarantee that a real plant’s output will follow this pattern.
Calculation eight: compare a forecast with an actual result
Assume a fictional facility had forecast 960 GWh but its recorded annual delivery was 924 GWh. The shortfall against forecast is 36 GWh. Relative to the forecast, the percentage error is 36 ÷ 960 × 100% = 3.75% below forecast. Relative to the actual delivery, the fraction would differ. When a question asks for percentage error against the predicted value, use that predicted value in the denominator.
Does the actual result still meet the fictional annual requirement of 900 GWh? Yes. It exceeds the threshold by 24 GWh, or 24 ÷ 900 × 100% = 2.67% to three significant figures. A shortfall against forecast and success against an independent requirement can both be true. Students must not confuse performance against a forecast with performance against a minimum.
A responsible written conclusion is: ‘The actual delivery was 3.75% lower than the forecast but remained 24 GWh above the required annual supply.’ This sentence compares two different benchmarks clearly. It is stronger than ‘the facility underperformed’, which leaves the reader uncertain about the measurement being used.
Calculation nine: distinguish fixed from variable costs
Now introduce an entirely new fictional cost model. A service has fixed annual costs of S$18 million plus a variable operating cost of S$0.09 per kWh delivered. Let E be annual energy delivered in GWh. Since one GWh is one million kWh, variable expenditure in millions of Singapore dollars is 0.09E. Total annual expenditure in millions is C = 18 + 0.09E.
At E = 800, the model predicts C = 18 + 72 = S$90 million. At E = 1,000, it predicts C = 18 + 90 = S$108 million. A student who multiplies 0.09 by 800,000,000 and then adds 18 without reconciling dollars and millions will combine incompatible units. Instead, either work entirely in dollars or entirely in millions.
This example shows why a constant unit expenditure calculated from one table is not always valid for a different output. A fixed cost changes the average cost per kWh as energy delivery changes. Before scaling up a ratio, check whether the question’s cost model includes a fixed component. The formula is an invented linear model for exam practice, not a description of actual nuclear plant finances.
Calculation ten: solve a break-even inequality
Take two fictional service prices. Model P has a fixed charge of S$30,000 plus S$0.12 per kWh. Model Q has a fixed charge of S$50,000 plus S$0.10 per kWh. Let x be the number of kWh delivered during the relevant billing period. The total costs are P = 30,000 + 0.12x and Q = 50,000 + 0.10x.
To find when Q is no more expensive than P, solve 50,000 + 0.10x ≤ 30,000 + 0.12x. Rearranging gives 20,000 ≤ 0.02x, so x ≥ 1,000,000 kWh. At exactly one million kWh the two models are equal: P = Q = S$150,000. Below that, P is cheaper; above that, Q is cheaper under the supplied formulas.
The equality point is not a universal recommendation. The model ignores all unspecified service differences and feasibility conditions. Show the inequality direction and substitute the threshold into both expressions. This check is valuable because students sometimes reverse the sign when moving terms. A real-world interpretation should say ‘under the stated cost formulas’ rather than ‘in all circumstances’.
Calculation eleven: a limit is not the same as a target
A hypothetical power-distribution contract requires annual delivery of at least 900 GWh but no more than 1,050 GWh under its specified conditions. The feasible range is 900 ≤ E ≤ 1,050. Plan A’s 960 falls inside that range. Plan B’s 840 is below it. A new hypothetical plan delivering 1,080 is above the maximum and therefore fails even though it easily exceeds the minimum.
Students who notice only ‘at least’ may think that more is always acceptable. A real-world word problem can contain a minimum, a maximum and additional conditions. Write them as separate inequalities before comparing options. The final choice must satisfy all compulsory constraints, not merely the most memorable one.
For a structured solution, state each constraint in its original units, evaluate each plan and name why it passes or fails. ‘Plan C fails because 1,080 GWh exceeds the stated 1,050 GWh upper limit’ is more informative than ‘Plan C too big’. Units and the precise comparison make the reasoning inspectable.
Calculation twelve: use ratios with a clearly named reference
In the original fictional table, output ratio A:B is 960:840. Divide both values by 120 to obtain 8:7. Expenditure ratio A:B is 120:84. Divide by twelve to obtain 10:7. Both simplified ratios are correct but describe different quantities. The numbers 8:7 and 10:7 do not mean that Plan A’s value is better by the same proportion in all senses.
To compare expenditure per unit of output, divide expenditure by output separately or combine ratios carefully. A has 120 ÷ 960 = 0.125 million dollars per GWh. B has 84 ÷ 840 = 0.100 million dollars per GWh. A’s unit expenditure is 25% higher than B’s unit expenditure, using B as the base: (0.125 − 0.100) ÷ 0.100 × 100 = 25%.
The earlier annual-expenditure percentage difference was 42.9% relative to B. The 25% unit-rate difference is not a contradiction. They have different denominators. Read the exact object of comparison in every sentence. This is one of the fastest ways to prevent a wrong figure from becoming a convincing explanation.
Calculation thirteen: interpret a straight-line graph
Suppose a graph plots the fictional fixed-plus-variable cost C = 18 + 0.09E, where C is measured in millions of dollars and E in GWh. The vertical intercept is 18, representing the fixed annual cost when the model quantity E is zero. The gradient is 0.09 million dollars per GWh, numerically equivalent to S$0.09 per kWh under the conversion.
At E = 600 GWh, C = 18 + 54 = S$72 million. At E = 900 GWh, C = 18 + 81 = S$99 million. The rise is S$27 million for 300 GWh, so the gradient calculated from these points is 27 ÷ 300 = 0.09 million dollars per GWh. Interpret both the gradient and intercept in words; do not report the bare numbers.
If a graph starts its vertical scale at 60 rather than zero, be cautious when comparing the visual heights of points. The numerical gradient can still be calculated from axis readings, but the picture may exaggerate the apparent separation. A well-reasoned answer gives values from the scale instead of judging by the apparent length of bars or lines.
Calculation fourteen: a table can have inconsistent definitions
A fictional report lists ‘energy produced’ as 960 GWh in one column and ‘energy delivered’ as 910 GWh in another. An editor adds the two and calls the sum 1,870 GWh of electricity available to homes. That total is not justified. The two quantities may describe different stages of the same energy flow, and delivered energy may already be part of produced energy.
Before adding quantities, ask whether the categories are mutually exclusive. Produced and delivered are not necessarily separate outputs. If the examination question defines a 50 GWh difference as transmission or other losses within its simplified model, the delivered total would be 960 − 50 = 910 GWh. That arithmetic uses an explicitly supplied relationship rather than assuming one.
This is a mathematical reading skill: know what the quantities count. The same habit applies to receipts and net sales, people enrolled and people attending, or litres collected and litres distributed. Numbers that appear in adjacent columns are not automatically additive.
Calculation fifteen: derive an efficiency from given figures
An entirely hypothetical exercise gives a system an energy input of 1,200 GWh and useful output of 900 GWh over the same period. Its efficiency, as defined in the question, is useful output divided by input × 100% = 900 ÷ 1,200 × 100% = 75%. The unused difference is 300 GWh within this simplified energy-accounting model. Do not confuse the efficiency percentage with a capacity factor or an electricity price.
If the question gives useful output as 0.75 times input and asks for the input needed to deliver 975 GWh, rearrange 0.75I = 975 to obtain I = 1,300 GWh. Check: 75% of 1,300 is 975. Reversing the ratio and multiplying 975 by 0.75 would produce 731.25, the wrong direction for a requirement that output is smaller than input.
Do not present 75% as a typical nuclear facility efficiency; it is fictional data used to teach ratios and equations. The actual physical process and performance of any facility require specific evidence. This caution is part of the subject’s mathematical reasoning, not an excuse to avoid the calculation.
Calculation sixteen: budget per household is an allocation, not a tariff
Imagine a community project assigns S$6 million in outreach and maintenance costs across 40,000 households. Simple division gives S$150 per household. It would be wrong to write that each household must pay S$150 unless the problem explicitly states that the full amount will be collected equally from households. Cost allocation and household billing are different ideas.
If the question later says that only 60% of the project cost is recovered through equal household contributions, the amount to recover is 0.60 × S$6 million = S$3.6 million. Dividing by 40,000 gives S$90 per household. Now the payment inference follows from a stated rule. Naming the missing allocation rule is often the difference between valid and invalid interpretation.
Students should watch for the verbs ‘costs’, ‘allocates’, ‘charges’ and ‘pays’. A maths problem may deliberately distinguish them. Clear reading prevents a very ordinary division from supporting the wrong real-world conclusion.
Calculation seventeen: estimate the result before calculating
Suppose an annual output of 910 GWh is shared across twelve months. A quick estimate is around 900 ÷ 12 = 75 GWh a month. The exact mean is 75.833… GWh. If a calculator result shows 758.33, something is wrong: most likely a misplaced decimal point, a wrongly entered number or a unit mix-up. Estimation helps catch errors before they become final answers.
For money, a rough check works the same way. S$90 million divided by 900 million kWh is about ten cents per kWh. If a worked solution announces S$100 per kWh, the result should be challenged immediately. The estimate does not replace exact working; it is a guardrail. High-performing students do not rely exclusively on calculators to tell them whether a result is plausible.
The primary-to-secondary transition matters here. The Primary 6 Mathematics learning hub develops percentage, ratio and problem-model foundations. At G3, the context and the amount of information grow, but the habit of estimating before committing to a numerical answer still pays off.
Calculation eighteen: distinguish a rate from a total
A fictitious pump transfers 18,000 litres of cooling water in 15 minutes under an approved classroom word problem. Its average transfer rate is 18,000 ÷ 15 = 1,200 litres per minute. At the same constant rate, 20 minutes would move 24,000 litres. The rate is per minute; the total is litres. Write both clearly, because simply copying 1,200 without its unit does not answer either question fully.
This is a neutral rate exercise. The numbers are not a real reactor design specification and do not authorize students to experiment with nuclear equipment. The mathematics is transferable to school water tanks, transport systems and ordinary pumps. We use a wider science setting only to practise selecting the appropriate rate and checking whether constant-rate scaling has been permitted by the question.
If the pump’s rate varies during the period, the average rate can still be computed from total volume and time, but it need not represent the rate at every moment. A conclusion about a precise minute requires further information. Again, the key is to keep an average from being mistaken for an instantaneous measurement.
Calculation nineteen: conditional rankings can be different and both correct
From the original fictional data, Plan B has the lower unit operating expenditure. Plan A has the higher annual forecast output and is the only one that clears 900 GWh. A table ranking the plans by cost per kWh places B first; a table ranking them by compliance with the minimum output places A first. Those rankings do not contradict each other because they use different criteria.
A decision question must say how to handle multiple criteria. If the minimum is compulsory, first remove plans that fail it, then compare costs or other allowed criteria among the feasible plans. If the minimum is merely a preference, the choice may require a trade-off specified by the question. Never invent an unstated weighting system and report it as objectively correct.
In your final explanation, name the criterion immediately: ‘Plan B is cheaper per forecast kWh, but Plan A is the only plan meeting the stipulated 900 GWh annual supply requirement.’ That is an accurate two-part conclusion. It leaves no need for a dramatic statement that one number has defeated all others.
Calculation twenty: integer decisions cannot be solved with fractional projects
A hypothetical district requires at least 250 GWh from modular supply units. Each fictional module can deliver 72 GWh annually under the exercise assumptions. If n is the number of modules, solve 72n ≥ 250. Division gives n ≥ 3.4722… . Because n represents a whole count of modules, the minimum feasible number is four, giving 288 GWh.
Rounding to three would supply only 216 GWh and fail the minimum. Rounding to 3.47 modules is not meaningful for a question that explicitly requires complete units. This integer constraint echoes the earlier nuclear-emission counting in Volume 0505, but the owner here is energy-supply planning and minimum feasible quantities, not the composition of radioactive nuclei.
If a maximum of five modules is also given, four or five meet the requirement, and a separate cost rule may determine which is preferable. A number satisfying one inequality is not automatically the best complete solution. Use all the constraints in the problem.
Calculation twenty-one: read a threshold on a graph
Consider a fictional graph of cumulative annual electricity delivery. At month eight the curve reaches 610 GWh, at month nine 705 GWh, and at month ten 795 GWh. If the question asks when the cumulative figure first exceeds 700 GWh, the answer is by the end of month nine among these reported monthly endpoints. Do not interpolate to a specific day unless a line model and within-month behavior are provided.
A graph point at 705 is five above the threshold, but the line’s apparent position does not tell you exactly what happened between the monthly measurements. A careful student distinguishes the recorded endpoint from a continuous process assumed for convenience. If the graph explicitly provides a linear interpolation instruction, use it and label the result as an estimate.
Threshold language matters. ‘At least 700’ includes equality. ‘More than 700’ excludes equality. ‘Below 700’ excludes 700. Such small phrases influence the inequality and the final conclusion. Underline them before solving.
Calculation twenty-two: an uncertain forecast needs a range
Suppose the imaginary Plan A forecast is 960 GWh with a planning range of 8% below to 5% above the central estimate. The lower bound is 960 × 0.92 = 883.2 GWh. The upper bound is 960 × 1.05 = 1,008 GWh. The result is a forecast range from 883.2 to 1,008 GWh, not a guarantee that all values are equally likely.
The lower bound falls below the required 900 GWh. Therefore the central forecast meets the threshold, but the supplied planning range does not guarantee that it will be met. That is a different conclusion from the one reached using the central number alone. The question may ask for both. This is not a probability calculation unless a probability distribution or a frequency interpretation is given.
Do not call the range a confidence interval or formal error bar merely because it has upper and lower limits. It is an invented planning interval under the stated assumptions. The correct label matters. Mathematical precision includes accurate description of what numbers represent.
Calculation twenty-three: test a proposed graph scale
A student proposes using a graph with the vertical axis starting at 800 GWh to display 840 and 960 GWh as two bars. The graph can be drawn that way if the axis is clearly labelled, but the difference may look visually exaggerated. A bar chart usually communicates magnitude through length, so omitting the lower part can mislead a casual reader. The mathematical values themselves must remain clear.
For examination interpretation, read the numerical scale rather than trusting apparent bar height. If a question asks for the absolute difference, subtract 960 − 840 = 120 GWh. If it asks for the percentage difference, select the specified reference. Do not infer ‘twice as much’ from bars that appear twice as tall on a truncated axis.
This connects to English visual-text reading without turning the Mathematics problem into a media studies lesson. A transparent graph helps other people reach the right numerical comparison; a misleading scale makes the same numbers harder to judge fairly.
Calculation twenty-four: distinguish total cost from average cost
Two plans may have different fixed and variable components. Suppose Plan R costs S$40 million plus S$0.08 per kWh, while Plan S costs S$10 million plus S$0.12 per kWh. At 500 million kWh, R costs S$80 million and S costs S$70 million. At one billion kWh, R costs S$120 million and S costs S$130 million. The cheaper plan changes as the quantity changes.
Set the totals equal: 40,000,000 + 0.08x = 10,000,000 + 0.12x. This gives 30,000,000 = 0.04x, so x = 750,000,000 kWh. Verify at that quantity: R = S$100 million and S = S$100 million. Below the break-even amount, S is cheaper under the formulas; above it, R is cheaper. These are fictional service equations, not genuine plant price forecasts.
Why does this matter? A headline that one plan ‘has a lower rate’ may refer to a variable component while ignoring fixed costs. The total answer changes with the production amount. When a question gives a cost function, use the whole function, not a single coefficient lifted out of context.
Calculation twenty-five: keep significant figures under control
Suppose the final comparison yields 119.847 GWh of additional supply. To three significant figures, report 120 GWh. But if the problem asks whether the figure is strictly greater than 120 GWh, you cannot first round it to 120 and then conclude that the original value is equal to or greater than the threshold. The unrounded value, 119.847, is below 120. Rounding for presentation must not change a decision at a boundary.
Carry adequate precision through intermediate steps, and round at the final stage unless instructed otherwise. If the stated input data are already rounded, do not pretend the final answer has unlimited physical precision. In G3 Mathematics, the official general presentation rule gives a default; the exact question may override it. Read that instruction, and use more accurate intermediate values when a proof of a specified rounded answer is required.
One good habit is to write a separate conclusion after the numerical working. It can say: ‘Using the unrounded value, the option is 0.153 GWh below the strict 120 GWh threshold and therefore fails.’ This is the sort of contextual interpretation that earns credibility even when the arithmetic is simple.
Paper 2 real-world strategy: read the whole constraint set
The final K310 Paper 2 question explicitly focuses on a real-world application. Before calculating, read all subparts and mark the known values, required outputs, units and constraints. Sketch a compact model, such as a table of Plan A and Plan B, before writing a page of calculations. This prevents you from answering a question that the later information changes or qualifies.
Proceed in layers. Layer one: convert all comparable quantities into common units. Layer two: compute direct differences, ratios and rates. Layer three: solve equations or inequalities. Layer four: test feasible outcomes against every stated condition. Layer five: write a conclusion that names the metric and its limitations. If you are uncertain whether an extra assumption is allowed, flag it in words rather than hiding it inside a calculation.
Work must be visible. The examiner cannot reward a method that does not appear on the page simply because your final number happens to be right. Write a formula, substitute the figures and keep the units near the result. Cross out a wrong line cleanly rather than obscuring the correct path with unexplained changes.
A ten-minute self-check
Use the original fictional dataset without looking at the worked solutions. Question one: convert 960 GWh into kWh. Question two: compute Plan A’s operating expenditure per kWh. Question three: state which plan meets 900 GWh. Question four: find Plan A’s output percentage increase relative to B. Question five: explain why an annual average cannot prove a monthly minimum. Question six: distinguish a forecast from an actual outcome.
The answer key is: 960,000,000 kWh; S$0.125 per kWh; only Plan A meets the supplied forecast threshold; 14.3% to three significant figures relative to B; an annual average does not fix individual monthly values; and a forecast is a prediction, not a measurement of what happened. Write each answer with its reason. A bare figure often hides whether the question was understood.
For a second round, change the annual requirement to 980 GWh. Neither of the original central forecasts then meets the threshold. The correct recommendation becomes ‘neither is feasible on the stated output figures’, even though A still has greater output. Changing one condition while keeping all calculations visible is an effective test of genuine understanding.
What this guide owns, and where to go next
This volume owns mathematical planning comparisons: GWh-to-kWh conversion, expenditure per unit, percentage differences, threshold decisions, linear costs, cumulative graphs and limits on what forecasts prove. It does not explain the nuclear reaction itself; Volume 0508 owns the Physics foundation. Nor does this lesson turn invented values into factual claims about Singapore’s future electricity supply.
For deeper mathematics support, the site has a secondary Mathematics guide to choosing equations, tables, graphs and diagrams and a PSLE Mathematics unit-and-model checking lesson. These are separate foundational routes, not duplicate nuclear-energy pages.
The next English guide will examine how a fictional information graphic can describe an annual energy figure accurately in one sentence and exaggerate it in another. The integrated EMS guide after that will ask for a responsible community-forum explanation using three independent checks: scientific mechanism, mathematical feasibility and careful public language. In each case, the best answer names what is known, what has been calculated and what cannot yet be concluded.
Official sources for the 2027 examinations
Use the SEAB SEC overview for the qualification, the SEAB 2027 G3 subject index for current codes, and the 2027 K310 G3 Mathematics specification for assessment rules and content. For a complete context behind the subject matter, the separate K323 Physics syllabus covers nuclear fission and fusion at its prescribed level. These sources were checked on 1 November 2026. Fictional exercises are identified as such and are not official specimen questions.
