AP Calculus optimisation problems sit at the meeting point of differentiation and applied reasoning. On the FRQ, they ask a candidate to take a word problem, write a single-variable function whose value represents what must be maximised or minimised, then defend the resulting critical point with the first derivative test, the second derivative test, or a closed-interval check. For most students reading this, the gap between a 3 and a 5 on the optimisation question lives in three places: the modelling line, the justification row, and the global-versus-local check at the end of the work.
The exam rewards students who can turn prose into algebra, then read the rubric language that accompanies each step. A correct critical value with no sign chart typically loses the justification row. A global maximum claimed from a single critical point loses the answer row. This article walks through how the optimisation FRQ is constructed on AP Calculus AB and BC, where the points are placed, and how a preparation strategy that emphasises the modelling step and the sign-change argument changes scoring outcomes.
1. Where optimisation lives in the AP Calculus exam format
Optimisation appears on the AP Calculus exam in two distinct locations: the multiple-choice section, where it shows up as a stem asking for a maximum area, a minimum surface, or a value of a derivative at an extremum, and the free-response section, where it takes the form of a multi-part FRQ. On AB, optimisation typically appears as a single FRQ worth 9 points, often in the second half of Section II. On BC, the same prompt shape appears but is usually combined with a second context, a piecewise function, or a parametric setup, and the rubric allocates an extra point for the additional context.
The MCQ versions are short and recognisable. A prompt will give a geometric figure or a written constraint, ask for the value of x that maximises something, and the candidate must translate the constraint into an expression and differentiate. These items are tight on time. The College Board allows roughly 1 minute 52 seconds per MCQ on the standard administration, and optimisation items tend to land in the moderate-difficulty band, meaning they reward fluency rather than cleverness. A candidate who cannot set up the function in under 90 seconds will run out of time before reaching the derivative step.
The FRQ version is the centre of gravity. It unfolds over three to four parts labelled (a) through (d) on AB and (a) through (c) or (d) on BC. Part (a) almost always asks for the function. Part (b) asks for a derivative or a critical value. Part (c) asks the candidate to justify that the value is a maximum or minimum. Part (d), when present, asks for the maximum or minimum value, or extends the question into a related context such as a second function or a parametric variant. The 9-point AB rubric typically allocates 2 points to setup, 3 points to differentiation and critical values, 2 points to justification, and 2 points to the final answer. The BC variant usually adds 1 to 2 points of context and an extra row for parametric reasoning.
A preparation strategy that works for one format does not always transfer. The MCQ rewards recognition of the constraint-to-function move. The FRQ rewards writing the constraint as a sentence, not just an equation, and labelling the variable clearly. For most candidates, drilling the FRQ form is the higher-leverage move because the same setup fluency shortens the MCQ time on test day.
2. The modelling line: turning prose into f(x)
The first row of points on any AP Calculus optimisation FRQ is the modelling row. The candidate must write a function whose value is the quantity to be optimised, expressed in a single variable, with units or a label that matches the prompt. Two things go wrong here most often. The candidate writes the function in two variables, or the candidate writes the function correctly but fails to declare the variable or the domain.
Consider a standard prompt: a rectangle is inscribed in a right triangle with legs of length 6 and 8. The correct function is A(x) = x(8 − (4/3)x) for x in the closed interval [0, 6], where x is the length of the rectangle's base along the leg of length 8 and the height is found by similar triangles. A candidate who writes A = xy and then solves a system of equations at the derivative step has not earned the modelling row. A candidate who writes A(x) = x(8 − (4/3)x) without stating the domain has also lost a point on most rubrics because the closed interval is part of the justification chain downstream.
The strongest habit is to write the constraint sentence first, then the function, then the domain. On a piece of paper, the work looks like: Let x be the length of the base. Then by similar triangles, the height is 8 − (4/3)x. The area is A(x) = x(8 − (4/3)x), where 0 ≤ x ≤ 6. That single block earns the modelling row, sets up the closed-interval check, and gives the grader a labelled variable to follow through the rest of the problem.
For most students, the modelling line is where the score is won or lost. A preparation strategy that drills ten to fifteen constraint-to-function translations, ideally across at least three prompt families, will make this row automatic. The prompt families that recur on the exam are: rectangles inscribed in triangles, rectangles under curves, cylinders inscribed in spheres, fences along a wall, and distance minimisation problems. Drilling one example from each family in the weeks before the exam builds the recognition that the constraint always hides inside a similar-triangles or Pythagorean move.
3. The differentiation row: f'(x)=0 and what the rubric reads
Once the function is in place, the next row of points is the differentiation and critical value row. The candidate must find f'(x), set it equal to zero, and solve for x. On AB rubrics, the differentiation step typically earns a point on its own, the algebraic solution earns another, and the value of x at the critical point earns a third. On BC rubrics, the same three points appear, with the possibility of a fourth point if the prompt requires a derivative at a point or a chain-rule application that involves the second function.
The differentiation step is where the chain rule, the product rule, and the quotient rule earn their keep. A cylinder inscribed in a cone of radius R and height H, with the cylinder's radius r as the variable, has volume V(r) = πr²(H − (H/R)r). Differentiating gives V'(r) = 2πr(H − (H/R)r) + πr²(−H/R). Setting this equal to zero and solving requires the candidate to factor out πr and to recognise that r = 0 is a critical value that must be discarded because the cylinder has zero volume at r = 0. The rubric rewards the factorisation, the discarded endpoint, and the surviving critical value r = (2/3)R.
A second source of error on this row is sign errors in the chain rule. When the height is expressed as a function of the radius, the inner derivative is negative. Candidates who write the height as H − (H/R)r often differentiate this as H − r when the inner derivative is one, missing the (H/R) factor entirely. The rubric reads the derivative line strictly: if the chain rule is applied incorrectly, the differentiation point is not earned, even if the final critical value happens to be correct by cancellation.
For most students reading this, the highest-leverage move is to write the derivative in unsimplified form first, then simplify. Showing the product rule explicitly, with each factor labelled, gives the grader two opportunities to award partial credit if the simplification step goes wrong. On a nine-point FRQ, partial credit on the differentiation row is the difference between a 3 and a 4 on the question, and that 1-point swing can decide the overall AP score.
4. The justification row: first derivative test versus second derivative test on the FRQ
The justification row is the row that separates a 4 from a 5 on the optimisation FRQ. The candidate must defend the claim that the critical value is a maximum or a minimum, and the rubric reads the defence literally. Two methods are accepted: the first derivative test with a sign chart, and the second derivative test with a sign on f″(c).
The first derivative test is the safer choice for most candidates. It requires writing f'(x) in factored form, picking a test point on each side of the critical value, and reading the sign of f'(x) at each test point. The rubric awards the point when the signs are recorded correctly and the conclusion ("f'(x) changes from positive to negative, so f has a local maximum at x = c") is written in words. A sign chart drawn but not interpreted loses the point. A conclusion that says "maximum" without naming the sign change loses the point.
The second derivative test is faster to write but easier to fail. The candidate computes f″(c) and reads its sign. The rubric awards the point when f″(c) is computed correctly, when the sign is stated, and when the conclusion is written in words. A candidate who writes only the numerical value of f″(c) without the sign or the conclusion loses the point. A candidate who confuses the second derivative test's direction (positive second derivative means concave up, not a maximum) loses the point. For most students, the first derivative test is the better bet because the sign chart acts as a self-check: if the sign change goes the wrong way, the candidate notices before submitting the answer.
On BC rubrics, a third option appears: the closed-interval extreme value theorem argument. The candidate evaluates f at the critical value and at both endpoints, then states that the largest of the three values is the absolute maximum. This is the only method that addresses the global-versus-local question directly, and it is the one the rubric prefers when the prompt places the variable on a closed interval. A candidate who finds a local maximum by the first derivative test and stops has not answered the prompt if the prompt asked for the absolute maximum. The closed-interval check is the move that closes the loop.
5. The global-versus-local trap: why endpoints decide the answer row
AP Calculus optimisation problems often hide a global-versus-local trap. The candidate finds a local maximum at a critical value and reports it as the answer, but the absolute maximum is actually at an endpoint. The trap appears most often in fence-along-a-wall problems and in surface-area-of-a-box problems, where the geometric constraint forces the variable to a closed interval and the function is monotone over part of that interval.
Consider a prompt: a rectangular pen is built against a barn using 100 feet of fencing for the other three sides. The width perpendicular to the barn is x and the length parallel to the barn is 100 − 2x. The area is A(x) = x(100 − 2x) on the closed interval [0, 50]. The derivative is A'(x) = 100 − 4x, which equals zero at x = 25. By the first derivative test, A'(x) is positive on (0, 25) and negative on (25, 50), so x = 25 is a local maximum. The area at x = 25 is 1250 square feet. The areas at the endpoints are zero. So 1250 is the absolute maximum, and the closed-interval check confirms it.
Now consider a variant: the same pen, but the side parallel to the barn must be at least twice the side perpendicular to the barn. The constraint is now 100 − 2x ≥ 2x, which gives x ≤ 16.67. The function is still A(x) = x(100 − 2x) but the domain is now [0, 50] intersect [0, 16.67], which is [0, 16.67]. The critical point x = 25 is no longer in the domain, so the absolute maximum is at the endpoint x = 16.67, where the area is approximately 1111 square feet. A candidate who finds x = 25 and reports the area as 1250 has lost the answer row because the candidate failed to apply the additional constraint.
For most students reading this, the move is to write the domain explicitly, before the derivative step, and to check the critical value against the domain before drawing conclusions. If the critical value is outside the domain, the answer is at an endpoint. If the critical value is inside the domain, the closed-interval check is still the cleanest defence because it covers the global-versus-local question in one stroke.
6. The endpoint family: when the answer is a corner of the domain
A surprising number of AP Calculus optimisation FRQs have their absolute extrema at an endpoint rather than at an interior critical point. The pattern shows up in two forms. In the first form, the function is monotone on the entire domain, so the derivative never vanishes, and the extreme value theorem forces the answer to an endpoint. In the second form, the function has a critical point, but the critical point falls outside the domain because of an additional geometric or economic constraint.
The first form is the easier one to recognise. If the function is linear, or if the derivative reduces to a non-vanishing constant, the candidate should not search for a critical value. The candidate should evaluate the function at both endpoints and pick the larger or smaller value as the answer. The rubric awards the setup point, the endpoint evaluation points, and the conclusion point. The differentiation row in this case is a single line: f'(x) = k ≠ 0, so there are no critical values, so the extrema are at the endpoints.
The second form is the one that costs students a point per FRQ. A prompt might say "a cylindrical can must hold at least 500 cubic centimetres and use the minimum amount of material." The variable is the radius r, and the constraint is πr²h ≥ 500 with surface area S = 2πr² + 2πrh to be minimised. Solving h from the constraint gives h = 500/(πr²), so S(r) = 2πr² + 1000/r. The derivative is S'(r) = 4πr − 1000/r², which equals zero when r³ = 250/π, so r ≈ 4.30. The function S(r) is convex for r > 0, so this critical point is a global minimum on the open interval (0, ∞). The endpoint behaviour is that S(r) → ∞ as r → 0 and as r → ∞, so the global minimum is at the critical point. No endpoint trap here, but the candidate must still write the domain as r > 0 and reason that the global minimum exists because of the second derivative test or the convexity argument.
For most students, the practical move is to draw the function's graph mentally before computing the derivative. If the function is U-shaped or ∩-shaped, the critical point is the answer. If the function is monotone, the endpoint is the answer. If the function has a critical point that is not in the domain, the endpoint is the answer. This triage takes 30 seconds and prevents the most common endpoint error.
7. Common pitfalls and how to avoid them
Across the optimisation FRQs on recent administrations, the same five pitfalls account for most lost points. None of them is about the calculus itself; all of them are about the chain of reasoning that the rubric reads.
- Pitfall 1 — Two-variable function with no substitution. The candidate writes A = xy or V = πr²h and tries to differentiate both variables independently. The rubric awards the modelling point only when the function is in a single variable with a declared domain. The fix is to solve the constraint for one variable and substitute before writing f(x).
- Pitfall 2 — Critical value with no justification. The candidate finds x = 4 and writes "maximum" without a sign chart, a second derivative sign, or a closed-interval check. The justification row is the most common place to lose a point, and the fix is to write the test explicitly with a one-sentence conclusion.
- Pitfall 3 — Local maximum reported as absolute maximum. The candidate finds a local maximum at a critical point and reports the corresponding function value as the answer, ignoring the endpoints of the domain. The fix is the closed-interval extreme value theorem: evaluate f at the critical point and at both endpoints, then pick the largest or smallest.
- Pitfall 4 — Sign error in the chain rule on the height expression. The candidate differentiates h = H − (H/R)r as if the inner derivative were 1, missing the (H/R) factor. The fix is to write h as H(1 − r/R) and differentiate with the inner derivative (1/R) written explicitly.
- Pitfall 5 — Domain not declared or declared incorrectly. The candidate writes f(x) without a domain, or writes the domain as the positive reals when the geometric constraint forces a closed interval. The fix is to write the domain on the same line as the function, derived from the geometric constraint, and to check the critical value against the domain before concluding.
A preparation strategy that focuses on these five pitfalls, with one timed FRQ per pitfall, raises the optimisation question score by an average of 1 to 2 points across a six-week cycle. The drill is more effective than re-reading the textbook chapter because the pitfalls are behaviour patterns, not concept gaps.
8. BC versus AB: where the extra point lives
On AP Calculus BC, the optimisation FRQ carries one to two more points than on AB, and the additional points are not in the calculus. They are in the context. A typical BC optimisation FRQ adds a second function, a parametric representation, or a related-rates flavour that requires the candidate to read the prompt as a multi-stage problem rather than a single optimisation move.
The first extra point usually appears at the modelling row, where the BC prompt asks for two related functions instead of one. A prompt might say "a particle moves along a curve, and the position is given parametrically; find the time at which the speed is maximised." The candidate must find the speed function as a single-variable function of t, differentiate, and locate the critical time. The modelling row on the BC rubric awards a point for the speed function and another point for the domain, which is often a closed interval derived from the parametric range.
The second extra point, when present, appears at the justification row. The BC rubric often requires the candidate to verify that the critical point is a global extremum on the parametric interval, not just a local one. The move is the closed-interval check applied to the speed function over the parametric range, with the largest speed value reported as the answer. A candidate who finds a local maximum on the speed function and stops loses the BC point.
The table below summarises how the rubric allocates points across the two courses.
| Rubric row | AB allocation | BC allocation | What the grader is reading |
|---|---|---|---|
| Setup / function in one variable | 2 points | 2–3 points | Variable named, constraint written as a sentence, domain declared |
| Differentiation and critical value | 3 points | 3 points | f'(x) computed correctly, set to zero, solved |
| Justification | 2 points | 2–3 points | First or second derivative test, sign chart or sign on f″(c), conclusion in words |
| Answer | 2 points | 2 points | Absolute maximum or minimum value, with units, on the closed interval |
For most students, the BC optimisation question is the higher-leverage place to spend preparation time because the additional context forces the modelling row to be airtight. A candidate who has drilled AB-level modelling on a closed interval will find the BC variant more comfortable than a candidate who has only seen open-domain optimisation prompts.
9. A six-week preparation strategy for the optimisation question
The optimisation question rewards a specific kind of preparation: constraint-to-function translation, sign chart fluency, and closed-interval discipline. A six-week plan that targets these three skills in order produces a 5-ready level of fluency by exam day.
In week one, the candidate should drill ten constraint-to-function translations across at least four prompt families: rectangles inscribed in triangles, rectangles under curves, cylinders inscribed in spheres or cones, and fence-along-a-wall prompts. The drill is timed at 4 minutes per prompt, with the function and domain written on a single line. The goal is recognition: when the candidate sees the prompt family, the variable name and the constraint equation should appear in the candidate's mind within 30 seconds.
In week two, the candidate should drill ten differentiation steps on the functions produced in week one. The drill is timed at 3 minutes per derivative, with the product rule, quotient rule, and chain rule written explicitly. The goal is to make the differentiation row automatic so the candidate has time for the justification row on test day.
In week three, the candidate should drill ten first derivative test sign charts on the derivatives from week two. The drill is timed at 3 minutes per sign chart, with the sign change written as a one-sentence conclusion. The goal is to make the justification row automatic.
In week four, the candidate should take five full 9-point AB FRQs under timed conditions (15 minutes each) and grade the work against the rubric. The candidate should mark which rubric row lost points and identify the pattern. The goal is to surface the personal pitfall set.
In week five, the candidate should drill the personal pitfall set with five additional prompts targeting the weakest row. The goal is to convert the pitfall into a habit.
In week six, the candidate should take two full AB and two full BC FRQ sections under timed conditions, with the optimisation question timed at 15 minutes on AB and 18 minutes on BC. The goal is to lock in the pacing and to confirm that the modelling, differentiation, justification, and answer rows are all under 4 minutes each on AB, leaving 1 minute for the review pass.
For most students reading this, the plan raises the optimisation question score by 2 to 3 points across the six-week cycle. The plan is also transferrable to other applied FRQ types, including related rates and accumulation, because the underlying skill — turning prose into a single-variable function with a declared domain — is the same.
10. Reading the prompt: a worked example end to end
To close the loop, here is a worked example that exercises every row of the rubric. The prompt is: a rectangular box with a square base and an open top must have a volume of 500 cubic centimetres. Find the dimensions that minimise the surface area of the box, and find the minimum surface area.
The modelling row: let x be the side length of the square base in centimetres, and let h be the height in centimetres. The volume constraint is x²h = 500, so h = 500/x². The surface area is S = x² + 4xh, where the x² is the base and the 4xh is the four sides (no top). Substituting h gives S(x) = x² + 4x(500/x²) = x² + 2000/x. The domain is x > 0, derived from the geometric constraint that the base side length must be positive.
The differentiation row: S'(x) = 2x − 2000/x². Setting S'(x) = 0 gives 2x = 2000/x², so 2x³ = 2000, and x³ = 1000, so x = 10. The second derivative is S″(x) = 2 + 4000/x³, which is positive for all x > 0, so S is convex and the critical point is a global minimum.
The justification row: by the second derivative test, S″(10) = 2 + 4000/1000 = 6 > 0, so S has a local minimum at x = 10. Because S(x) → ∞ as x → 0⁺ and as x → ∞, the local minimum is the global minimum on the domain x > 0.
The answer row: at x = 10, h = 500/100 = 5, and S(10) = 100 + 2000/10 = 100 + 200 = 300. The dimensions are 10 cm by 10 cm by 5 cm, and the minimum surface area is 300 square centimetres.
This example exercises the modelling row with a substitution, the differentiation row with a rational function derivative, the justification row with the second derivative test, and the answer row with both the variable value and the function value reported. A candidate who can produce this work in 12 to 15 minutes is at a 5-ready level on the AB optimisation question. The BC variant would add a parametric context or a related-rates flavour, but the underlying structure is the same.
Optimisation on the AP Calculus exam is a skill that compounds with practice. The modelling row is the highest-leverage place to spend time, the justification row is the most common place to lose points, and the closed-interval check is the move that closes the global-versus-local question. Candidates who drill the prompt families, the sign charts, and the closed-interval discipline in the six weeks before the exam consistently score higher on the optimisation question than candidates who re-read the textbook chapter.
AP Courses' AP Calculus AB and BC tutoring programmes assign a student's first optimisation FRQ as a diagnostic, then build a six-week preparation plan around the modelling row, the sign chart, and the closed-interval check. The plan targets the specific prompt family that costs the student the most points, and it converts the 5 target into a measurable preparation routine.