The instances at hand have been used for computational testing of the methods described in 

Vogl/Braune/Doerner (2017): 
A Multi-Encoded Genetic Algorithm Approach to Scheduling Recurring Radiotherapy Treatment Activities with Alterantive Resources, Optional Activities, and Time Window Constraints, 
to be published in: Lecture Notes of Computer Science, Springer. 

The instance file contains 13 columns:
1. $index ... gives the patient index.
2. $nbDT ... gives the number of DTs that needs to be scheduled for the specific patient. 
3. $beamType ... gives the beam type the patient requires, 0...proton, 1...carbon ion.
4. $durationDT ... gives the planned duration of one DT.
5. $petct ... binary, indicates whether or not the patient requires a pet-ct.
6. $room ... indicates the required room, {12, 13, 14}.
7. $aRO ... indicated the assigned radio-oncologist, {1,2,3,4}.
8. $nbActivities ... indicates the number of activities of this specific patient .
9. $releaseTime ... gives the release day of the patient.
10. $sedation ... indicates, whether the patients needs to be sedated during treatment (not considered in this paper).
11. $setup ... gives the in-room set-up time.
12. $teardown ... gives the in-room tear-down time.
13. $deadline ... gives the deadline day of the patient.

The following list summarizes the way the instances are created:
- The name of the instance reflects the number of patients to be treated/scheduled. It varies between 25 and 200 patients.
- The number of treatments per patient is equally distributed between 8 and 12 for instances with 25 to 75 patients and equally distributed between 10 and 18 treatments for instancces including 100 and 200 patients.
- The duration of an irradiation appointment varies among the beam types: proton therapy on average takes considerably longer (~N(12,5)) than carbon beam therapy (~N(8,5)).
- Both beam-types and rooms (12, 13, 14) are equally distributed among the patients.
- 50% of the patients require PET-CTs. 
- Each patient is assigned to one radio-oncologist (RO). There are four ROs available. The ROs are assigned randomly to the patients. 
- Set-up times might take two different values: 12 minutes (probability 80%) or 22 minutes (probability 20%).
- Tear-down times might take two different values: 3 minutes (probability 70%) or 6 minutes (probability 30%). 
- Release days are equally distributed between day 0 and day 3 for smaller instances (25 to 75 patients). 
- Larger instances (100, 200 patients) have staggered release times: 25 patients start within the first week, 25 patients start within the second week, and so on. 
- Deadline days depend on the release day: deadline = release time + ~U(1,3)

Further constraints of the problem at hand:
- The minimum/maximum time lag between the DT and a PET-CT is (in minutes): [0,15]
- The minimum/maximum time lag between the DT and a WCE (if there is no PET-CT in between): [15,60]
- The minimum/maximum time lag between the PET-CT and a WCE is: [15,60]
- In between one week, the starting times of DT activities might vary +/- 30 minutes from the intra-week stable time.
(- The general (non-stable time) minimum/maximum time lag between two DT activities (assuming no WCE or PET-CT in between) is: [numberOfMinutesOfOneDay - 2*120, 2*numberOfMinutesOfOneDay + 2*120])
- Stable times of two consecutive weeks may vary by a maximum of +/- 120 minutes. 
- At least 4 DTs must be scheduled within every 5 consecutive days, starting from the day of the first daily treatment.
- Within 5 consecutive days each patient has to have at least one WCE and one PET-CT.
- Alternative resources are available for all WCE activitities. They can be performed by any RO, however, the assigned RO is preferred. 

For questions concerning the data sets please contact petra.vogl@univie.ac.at.




