# Thinking in estimates: units, scaling and educated guessing: The big part first, then rounding, lumping and pictures

Source page: https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p3/

[All courses](https://www.materialinsights.org/learn/) / [Understanding Engineering](https://www.materialinsights.org/learn/engineering/) / [Module 2: Thinking in estimates: units, scaling and educated guessing](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/) Part 3 of 5

## The big part first, then rounding, lumping and pictures

Real situations are untidy: roofs have odd shapes, demand varies by the hour, and every input has some error. The tools in this part tame the mess. Find the effect that dominates and work it out first, replace untidy shapes by simple ones of the right size, keep only the digits you can justify, and reach for a sketch or a familiar analogy when algebra gets in the way.

### 3.1 Take out the big part first, then correct for the small ones

Most estimates contain one effect that sets the size of the answer. Work that out cleanly before anything else, then adjust for the rest with factors near one. The first step fixes the power of ten; the adjustments show how much to trust it.

Take rooftop solar. How much roof does a household need? The big part is sunlight. At noon on a clear day, about 1,000 watts of sunlight fall on each square metre facing the sun, and that is the figure at which panels are rated. A modern module turns roughly a fifth of it into electricity, so a square metre of module gives about 200 W. One kilowatt of panels therefore needs 5 m2 of module. Now the corrections. Panels in rows need gaps so one does not shade the next, and the roof needs paths for cleaning and space for parapets and tanks. A fair allowance is to double the area, so one kilowatt takes about 10 m2 of roof.

Energy comes in the same way. Noon power does not last, but adding the whole day's sunshine in much of India is equivalent to about five hours at full strength, an assumed round figure. Dust, heat, wiring and the inverter then take roughly a fifth. So 1 kW x 5 h x 0.8 gives about 4 kWh a day. A household using 250 units a month needs about 8 units a day, so around 2 kW of panels on about 20 m2. See 'Solar PV module' and 'Solar photovoltaics' on the site.

The trade-off is built in: tighter packing gives more power per roof but more shade and harder cleaning.

Data Worked estimate: roof area and daily energy for 1 kW of rooftop solar

| Step | Working | Result |
|---|---|---|
| Big part: sunlight on a square metre | Rating condition: about 1,000 W per m2 | 1 kW per m2 |
| Module area for 1 kW | Module turns about 20 per cent into power: 1,000 W / (1,000 W/m2 x 0.20) | 5 m2 |
| Correction: gaps, walkways | Roughly double the module area | about 10 m2 of roof per kW |
| Daily energy | 1 kW x 5 hours of full-strength sun x 0.8 for losses | about 4 kWh a day |
| A home using 250 units a month | 250 / 30 = 8.3 kWh a day; divided by 4 kWh per kW | about 2 kW, 20 m2 |
| Check against real values | Government consumer guidance (MNRE, 2022): about 10 to 12 m2 of shadow-free roof per kW, and 4 to 5.5 units a day on a clear sunny day. An industry source gives about 4 units a day as an all-India average | Area matches; energy matches |

The 5 sun-hours and 0.8 loss factor are assumed round values. Orientation, tilt, shade and season move the answer.
 Source: Ministry of New and Renewable Energy, FAQ on grid-connected rooftop solar (28 April 2022, published by PIB), about 10 to 12 square metres of shadow-free area per kWp; 4 to 5.5 units a day per kWp on a clear sunny day, https://static.pib.gov.in/WriteReadData/specificdocs/documents/2022/apr/doc202242850101.pdf  In the data

#### A small correction can become the limit

If module efficiency improves from 20 to 23 per cent, the module area for 1 kW falls from 5.0 to 4.3 m2, which is 13 per cent. Roof area falls less than that, because gaps and walkways account for about half of it and do not shrink with the panel. A part that looked small in the first estimate becomes the limit once the big part has been improved, which is a good guide to where design effort should go next.

- Mahajan, Street-Fighting Mathematics (MIT Press, 2010) · ch. 5, 'Taking out the big part'
- Ministry of New and Renewable Energy, FAQ on grid-connected rooftop solar (28 April 2022, published by PIB) · about 10 to 12 square metres of shadow-free area per kWp; 4 to 5.5 units a day per kWp on a clear sunny day, https://static.pib.gov.in/WriteReadData/specificdocs/documents/2022/apr/doc202242850101.pdf
- Bridgeway Power, 'Solar panel generation in India per kW' (2025) · about 4 units a day and 1,300 to 1,600 units a year per kW on average, https://bridgewaypower.in/blog/solar-panel-generation-india-per-kw-units

### 3.2 Round to one figure and lump the messy parts together

Lumping means replacing a complicated shape, distribution or process by a simple one that has the same overall size. A ragged roof becomes a rectangle, and a day of varying electricity demand becomes a steady block at the average. The lump is wrong in detail and right in scale, and scale is what an estimate wants.

Rounding goes with it. In an estimate, use numbers you can multiply in your head: 1,180 becomes 1,000, 0.27 becomes 0.3, and 9.8 becomes 10 if you are in a hurry. This feels reckless, but errors from rounding up and rounding down tend to cancel. The table shows a hostel kitchen that serves 1,180 students at 0.27 kg of rice a day over 300 days. The exact product is 95,580 kg. Rounded, the product is 90,000 kg, off by only 6 per cent, because the first rounding lowered the answer by 15 per cent and the second raised it by 11 per cent.

The same discipline applies at the other end of a calculation. A calculator happily shows eight digits, but the digits only mean something if the inputs were known that well. Kosky and colleagues stress that the number of significant figures you quote states how well you know the number, so a result should carry no more than the weakest input supports. If attendance is known to about 10 per cent, a rice total of 95,580.0 kg claims a precision of one part in a million. 'About 90,000 kg' is the honest form, and 'close to a lakh' is honest too.

A working rule: keep one figure while estimating and at most two in the final answer.

Data Rounding costs little: rice needed by a hostel kitchen in a year

| Quantity | Exact | Rounded | Effect on the product |
|---|---|---|---|
| Students | 1,180 | 1,000 | -15 per cent |
| Rice per student per day | 0.27 kg | 0.3 kg | +11 per cent |
| Days of cooking | 300 | 300 | none |
| Total rice | 95,580 kg | 90,000 kg | -6 per cent overall |

An illustrative case with assumed numbers. The two roundings partly cancel, which is typical when rounding is done without a bias.
 Source: Kosky et al., Exploring Engineering (Academic Press, 2013), ch. 2, 'Elements of Engineering Analysis', section 2.3 on significant figures (the principle only; the example is this course's own)  Common misreading

#### 'More digits mean a better answer'

Digits are cheap; knowledge is not. A result such as 12.4817 hours looks authoritative and says nothing about how well the inputs are known. Worse, it invites a reader to treat a difference in the third digit as meaningful. Engineers keep digits only as far as the data justifies, and many write results in the form '3 x 10^8' for exactly this reason: it shows that a single figure is meant.

- Mahajan, Street-Fighting Mathematics (MIT Press, 2010) · ch. 3, 'Lumping'
- Kosky et al., Exploring Engineering (Academic Press, 2013) · ch. 2, 'Elements of Engineering Analysis' (significant figures)

### 3.3 Pictures and analogies let you reason without algebra

Some problems yield to a drawing sooner than a formula. A speed-time graph is the best everyday case. The area under the line is the distance travelled, because speed times time is distance and the area of a strip is height times width. So you can find a distance by sketching the graph and working out the area of a triangle or a trapezoid.

Picture a suburban train moving between two stations 1,500 m apart. It speeds up at 1 m/s2 for 20 s, reaching 20 m/s, then cruises, then brakes at the same rate for 20 s. The two ramps are triangles, each with area half of 20 s x 20 m/s, which is 200 m. Together they use 400 m, leaving 1,100 m of cruising at 20 m/s, which takes 55 s. Total time is 20 + 55 + 20 = 95 s, and the average speed is 1,500 / 95, about 16 m/s, or 57 km/h. An easy case follows from the picture: if the stations are only 400 m apart, the train never leaves the ramps, and a train with a higher top speed would gain nothing. That is why frequent-stop metro lines care more about acceleration than top speed.

An analogy borrows the structure of a familiar system for an unfamiliar one. Electricity is a good example. Picture voltage as the height of water in a tank, current as the flow in a pipe, and a thin pipe as a resistor: more height pushes more flow, a thinner pipe passes less. Heat in a wall behaves the same way, with temperature difference pushing heat through, and insulation acting as the thin pipe. Two layers of insulation one after the other are like two thin pipes in a row, and their resistances add.

An analogy is a guide, not a proof. Where the physics differs, the analogy fails, so use it to guess and then check with numbers.
 Worked example

#### A picture for an average-speed trap

You drive the 150 km from Pune to Mumbai at 75 km/h and come back at 50 km/h. Is the average 62.5 km/h? Sketch speed against time. The outward leg is a block 2 hours wide and 75 high; the return is a block 3 hours wide and 50 high. Both blocks have the same area, 150 km, but the slow one is wider, so the slow speed fills more of the clock. Total distance over total time is 300 km / 5 h = 60 km/h (calculated). The sketch shows at once why the average leans towards the slower speed, and why making up lost time on the way back is harder than it feels.

- Mahajan, Street-Fighting Mathematics (MIT Press, 2010) · ch. 4, 'Pictorial proofs'
- Mahajan, Street-Fighting Mathematics (MIT Press, 2010) · ch. 6, 'Analogy'

### See it in the table
[H₂O Water](https://www.materialinsights.org/commodities/water/)[Ri Rice](https://www.materialinsights.org/commodities/rice/)[PV Solar photovoltaics](https://www.materialinsights.org/technologies/e_pv/)[El Electricity](https://www.materialinsights.org/energy/electricity/)

### Check your understanding

In the solar estimate, which step is the 'big part' that is computed first?
 The dominant effect is the sunlight falling on a surface; the gaps and losses are corrections applied afterwards.

A hostel rounds 1,180 students to 1,000 and 0.27 kg to 0.3 kg. Why does the total change by only 6 per cent?
 Rounding down by 15 per cent and up by 11 per cent nearly offsets; the net error is about 6 per cent.

On a speed-time graph, what does the area under the line represent?
 Speed multiplied by time is distance, so the area under a speed-time line is the distance covered.

[← Part 2: Dimensions and easy cases: checking formulas and finding them](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p2/)[Mark read · Part 4: Scaling: why size changes everything →](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p4/)

## Other sections

- [Module overview](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/index.md)
- [Why engineers estimate before they calculate](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p1/index.md)
- [Dimensions and easy cases: checking formulas and finding them](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p2/index.md)
- [Scaling: why size changes everything](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p4/index.md)
- [When an estimate is good enough to decide](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/p5/index.md)
- [Key terms, quiz and sources](https://www.materialinsights.org/learn/engineering/thinking-in-estimates/review/index.md)
