Exam and academic timetabling
Fix one exam clash by hand, and three more can appear. We plan exams, rooms and invigilators together, for a conflict-free exam schedule.
- Free 30-minute call
- No commitment
- You own the code
In short: An exam timetabling optimisation gives each exam a time slot, a room and supervisors, so no student has two exams at once. It also respects room sizes, individual students’ special needs and the gaps between each student’s exams.
The problem
The draft exam timetable went out on Monday. By Wednesday the inbox is full. One student has two exams on Thursday morning. A large exam is booked into a room that’s too small once the desks are spaced. You move one exam to fix the first clash, and now three other exams clash.
An exam timetable has to be right the first time, because the deadline is fixed and public. Three simple rules define a valid timetable. No student sits two exams at once. No invigilator is in two rooms at once. And no room hosts two exams at the same time. They sound easy until you apply them to a whole university.
Why a spreadsheet can’t keep up
Those rules apply to thousands of students, hundreds of exams and a fixed number of rooms and days. Every clash you fix by moving one exam can create three more elsewhere. A better spreadsheet speeds up the typing, and the puzzle stays as hard as it was. When a room floods or an examiner falls ill after publication, the cycle starts again.
What changes
Today, you build the timetable by hand and find the clashes when students complain. Our model reads your enrolment data and proposes a conflict-free exam schedule. You check it, adjust what the data can’t know and publish it. By then, the rooms, the invigilators and the gaps between each student’s exams are already settled.
Accessibility built in from the start
The model includes accessibility arrangements from the start. These are adjusted exam conditions for students with a disability. A student with extra time needs a room booked for a longer window. A separate, quiet room has to be reserved and staffed. A student using a reader, a scribe or assistive technology needs the right invigilator and equipment in place.
If you add these needs by hand after the main timetable is fixed, rooms and invigilators are suddenly missing. Then it gets hectic. If they’re in the model from the start, a timetable only counts as valid when it meets them. The timetable that comes out already meets them, so nobody has to fight for their arrangement at the last minute.
Changes after publication
Then the changes start. A room has water damage, an examiner falls ill or an exam suddenly has to be spread across two rooms. The model keeps everything that still works and re-plans only the area around the disruption. We call this local re-optimisation. It moves as few other exams as possible. So the timetable stays valid, and most people keep the dates they have already planned for.
Exam timetabling optimisation: what you get
We put your rules into the model: enrolments, rooms, available invigilators, rules for the gaps between exams and every accessibility arrangement on your list. Exam timetabling is the sibling of course timetabling, and many institutions solve them together. We test the model against your hardest exam session before you publish anything. You judge the result on figures you already know: clashes, exams per student per day and rooms used.
- A model written around your own exam and academic timetabling rules.
- Results measured in the numbers you already track.
- It slots into how you already work, and you can check every answer it gives.
How it works
Every shift, job and resource goes into one model, and your rules go in as hard limits. You get the best schedule that keeps all of them. What “best” means is your call: lowest cost, the fairest split or the fewest late orders.
A puzzle of hard rules
In optimisation, exam timetabling is called a constraint satisfaction problem. That’s a puzzle of hard rules that must all be kept at once. First the model finds any arrangement that keeps every rule. Then it searches for the one that spreads each student’s exams best and uses rooms and invigilators well. Three connected problems sit inside: student conflicts, rooms and invigilators.
Student conflict detection
Two exams clash if one student is enrolled in both. So the model builds a conflict graph from your enrolments. It’s a map of which exams share students, and how many. Every shared student becomes a hard constraint, a rule the model never violates: the two exams this student takes can’t be in the same time slot.
How many exam slots do you need?
Before you decide between a two-week and a three-week session, the conflict graph answers a sharper question. What’s the fewest slots that could ever work? In optimisation, this is called graph colouring. Each exam is a dot, each shared student a line, and each slot a colour. Two dots joined by a line can’t share a colour.
You can find a lower limit without any software. Look for the largest group of exams in which every pair shares at least one student. Say five first-year papers all overlap like that. Then no timetable fits them into fewer than five slots. Spread rules, rooms and invigilators can only add slots, never remove them, so it’s worth knowing before you fix the session dates.
Room capacity and invigilator scheduling optimisation
Room allocation runs alongside. An exam of 400 students needs a room, or a set of rooms, that seats 400 under exam spacing rules. The model assigns exams so no room is ever over capacity and no room hosts two sittings at once.
The third part is assigning the invigilators. Every room in every slot needs enough qualified invigilators. Nobody is rostered into two rooms in the same period, and everyone’s availability and workload limits are kept. The model solves all three problems at once, and you get one conflict-free exam schedule.
Spreading each student’s exams
A clash-free timetable is only the minimum, because students also care about spread. Two exams in consecutive hours are technically conflict-free. Your exam rules often forbid it, and your students dread it. So the model sets a minimum gap between any student’s exams and a cap on exams per day.
The model spreads each student’s exams as evenly as possible across the session. These rules turn a feasible timetable, one that keeps every rule, into one you can defend. These rules make the problem harder. Wider gaps between exams need more slots, and every extra slot needs rooms and invigilators of its own.
Constraint programming and MILP
The natural tool for this problem is constraint programming (CP). CP works directly with logical rules, such as “these two exams never share a slot”. Timetabling is full of them. One kind of rule gives exams that clash different time slots. Others make sure no room and no invigilator is booked twice at the same time. The model drops dead ends during the search, long before a spreadsheet would notice a clash. You get a timetable without clashes.
Other goals are numbers you can add up, such as how full the rooms are or what the invigilators cost. For those we use mixed-integer linear programming (MILP). It’s a maths model made of yes-or-no decisions and decisions about quantities. The strongest exam timetabling algorithms often combine the two. CP finds a timetable that keeps every rule, and MILP optimises the numbers on top. Your exam rules stay as they are.
Common questions.
Something else? Ask us directly.
At least as many as your largest group of exams in which every pair shares a student. If five papers all overlap that way, no timetable fits them into fewer than five slots. The true minimum can be higher, and spread rules push it up further. The conflict graph from your enrolments gives you the number before you fix the session dates.
Only if your rules allow it. You set the cap per day and the minimum gap, and the model treats them as rules or as strong preferences, whichever you choose. When the session is short, a strict cap may be impossible. Then the model tells you how many students would be affected, so you can decide with the numbers in front of you.
The model treats them like any other enrolment: it checks which exams they overlap with. If they arrive after you’ve published the timetable, the model changes it only where needed and moves as few exams as it can. You see which students and rooms a change touches before you approve it. Resit sessions can be planned as their own timetable with the same rules.
Sometimes it can, and if that’s the case, we’ll tell you. Standard software is built for an average operation, though. Yours runs on its own rules: your contracts, your exceptions, the things you know by heart. A custom model writes those rules in, and that’s where the better plan comes from.
You don’t need either to start. We build the model from the data you’ve already got, even if it lives in three spreadsheets and one person’s head. You get a system you run yourself. We train your team until they no longer need us.
You see a first result with the prototype, before you commit to anything bigger. We start with a free 30-minute call about your operation and the model that would fit it. Then we build a working prototype and run it on your numbers.
Similar pages.
Terms on this page
Every abbreviation explained in plain words, in our article “Metrics and methods”.
Tell us what you’re planning.
In a free 30-minute call with a specialist, you find out whether optimisation can help.
- Free, no commitment
- You talk to a specialist directly
- A clear go or no-go