Cost engineering: estimating methods and estimate classes

0:00 Most overruns are estimating errors in disguise

Here's a sentence I've heard in far too many steering meetings. 'The project is thirty per cent over budget.' And very often, when you dig in, the project isn't really over budget at all. The budget was approved on a concept-level estimate as if it were a precise number. The overrun was an estimating error in disguise. In this lecture you'll learn how estimates mature through classes, the main estimating methods and when to use each one, how to scale a past project with capacity factoring, and how to build a bottom-up estimate with a proper basis of estimate. By the end you'll be able to present any estimate as a range with a stated class, which is the single habit that prevents most of those painful conversations.

0:56 Why it matters

Why does this matter to a controls professional? Because every budget is built from an estimate, and every forecast is later judged against that budget. If the original estimate was immature, every variance report afterwards is measuring against the wrong yardstick. Cost engineering is the discipline of applying engineering judgement and experience to estimating, cost control and profitability analysis. Its core outputs are credible estimates that become budgets, and the benchmarks you'll later use to challenge optimistic forecasts. Here's the principle. The precision you claim must match how well the project is defined. Early on, honesty means wide ranges.

1:39 The concept: estimate classes

Think of estimates like a weather forecast. A forecast for next month is a broad outlook. A forecast for tomorrow afternoon is quite specific. Neither is wrong; they're just made with different amounts of information. Many organisations use a class system similar to the widely referenced AACE International cost estimate classification. Class five is the least defined, used for concept screening. Class four supports feasibility studies. Class three is typically used for budget authorisation. Classes two and one are used for control, bids and check estimates, based on much more complete design. As definition improves, the accuracy range narrows. Published guidance shows typical ranges for each class, but AACE is clear that the actual range should come from risk analysis of your specific project, not from a lookup table. So state the class, and show the range.

2:39 The main methods

Now the methods. Analogous estimating scales from a similar past project. It's fast, and its accuracy is low. Parametric estimating uses a statistical relationship, such as cost per square metre, per megawatt or per kilometre. It's good when the relationship is well established in your own data. Capacity factoring recognises that cost doesn't scale linearly with size. You use C two equals C one times the ratio of capacities raised to an exponent, usually below one, which captures economies of scale. Bottom-up estimating builds cost from quantities times unit rates for labour, materials and equipment, plus indirects. And three-point estimating captures optimistic, most likely and pessimistic values, which later feed Monte Carlo analysis.

3:28 Worked example one: capacity factoring a solar plant

Let's do the simple one first. A fictional company built a fifty megawatt solar plant in Sindh for forty million dollars. It's considering a one hundred and twenty megawatt plant. Using an assumed exponent of zero point eight, which in practice you'd take from your own historical data, we calculate. One twenty divided by fifty is two point four. Two point four to the power of zero point eight is about two point zero one six. Multiply by forty million, and you get roughly eighty point six million. Notice: not ninety-six million, which a straight-line scale-up would give. That's the economy of scale. But we're not finished. You must adjust for escalation between the two dates, location factors like labour rates, logistics and import duties, and any scope differences. And you present it as a class five or class four range, never as a single figure.

4:31 Worked example two: bottom-up pipework

Now a realistic bottom-up estimate from the lesson, illustrative numbers. Work package one point four point two: eighteen hundred metres of chilled water pipework. Labour: eighteen hundred metres at nought point nine hours per metre, at twenty-eight dollars an hour, gives forty-five thousand three hundred and sixty. Materials: fifty-five dollars a metre plus eight per cent waste, one hundred and six thousand nine hundred and twenty. Equipment, twelve thousand. Insulation subcontract, thirty-two thousand four hundred. Direct cost, about one hundred and ninety-six thousand seven hundred. Add a twelve per cent share of site supervision, and we're at roughly two hundred and twenty thousand before contingency. But here's the part people skip. Every line needs its basis: which drawing revision, where the quantity came from, where the productivity and rates came from, and the date. That's the basis of estimate, and it's what makes the number defensible.

5:34 Watch me do it: an estimate sheet with a basis column

Let me show you how I lay this out in a spreadsheet so anyone can challenge it. Columns for quantity, unit, rate, waste factor, amount and, crucially, basis. The amount formula is quantity times rate times one plus waste. Indirects come from a single named cell, so if the supervision percentage changes, I change it once. Then I add two more columns: a low and a high rate for the items I'm least sure of, like productivity. That gives me a low and high total alongside the most likely. Finally, in the basis column, I type the source: drawing M-201 revision C, productivity from the last two hospital jobs, rates from the March supplier quote. Now if someone asks, 'where did this number come from?', the answer is on the same row.

6:31 Escalation, contingency and common mistakes

Two terms that must never be mixed. Escalation covers expected price changes over time: inflation, commodity prices, wage rates. It's a forecast, often based on published indices. Contingency covers uncertainty in quantities, productivity and identified risks. Merge them into one number and you hide both. In high-inflation environments, and Pakistan has experienced periods of high inflation, for example, explicit escalation modelling is essential. Now the common mistakes. No basis of estimate, so nobody knows what the number includes. A single-point estimate with no class or range. Double-counting contingency by padding every line and then adding a contingency line on top. And forgetting indirects, owner's costs, permits, commissioning and spares.

7:18 Recap and try this now

Let's recap. Estimates mature through classes, and the range you present must match the project's definition. Choose the method to fit: analogous and parametric early, capacity factoring for scale-ups, bottom-up once you have quantities, three-point wherever uncertainty matters. Keep escalation and contingency separate and visible. And document a basis of estimate for every line. One last point on AI: tools can now extract quantities from drawings and models, search historical cost data and propose parametric estimates. They're only as good as the data, and every output still needs a traceable basis reviewed by a cost engineer. Your try-this-now: prepare a bottom-up estimate for one work package using the template in the lesson, with a basis note on every line and a low and high total.