TY - JOUR
T1 - Optimal resource investment and scheduling of tests for new product development
AU - Maravelias, Christos T.
AU - Grossmann, Ignacio E.
N1 - Funding Information:
The authors would like to gratefully acknowledge financial support from the National Science Foundation under Grant ACI-0121497.
PY - 2004/6/15
Y1 - 2004/6/15
N2 - Previous optimization models for the scheduling of tests for pharmaceuticals and agrochemicals assume that the resources available for testing such as laboratories and scientists are constant throughout the testing horizon, and that all testing tasks have fixed cost, duration and resource requirements. In order to be able to handle more effectively a number of potential products in the R&D pipeline a company may consider the option to install/hire additional resources. Also, the type and amount of resources can be chosen for some tests to reduce their duration. In this paper we propose a new scheduling MILP model that addresses these issues to optimize the overall costs. More specifically, the model (a) allows for installation of new resources during the course of testing, and (b) handles cost and duration of tests as functions of the type and amount of the resources assigned to each test. To enhance the solution of the model we develop: (a) a reformulation with disaggregated variables, (b) a procedure for the tightening of the big-M constraints, (c) a procedure that fixes some of the sequencing binary variables, and (d) logic cuts. The solution of a single large-scale problem is avoided with a heuristic decomposition algorithm that relies on solving a reduced mixed-integer program that embeds the optimal schedules obtained for the individual products. The proposed algorithm is shown to be one to two orders of magnitude faster than the full space method yielding solutions that are optimal or near optimal.
AB - Previous optimization models for the scheduling of tests for pharmaceuticals and agrochemicals assume that the resources available for testing such as laboratories and scientists are constant throughout the testing horizon, and that all testing tasks have fixed cost, duration and resource requirements. In order to be able to handle more effectively a number of potential products in the R&D pipeline a company may consider the option to install/hire additional resources. Also, the type and amount of resources can be chosen for some tests to reduce their duration. In this paper we propose a new scheduling MILP model that addresses these issues to optimize the overall costs. More specifically, the model (a) allows for installation of new resources during the course of testing, and (b) handles cost and duration of tests as functions of the type and amount of the resources assigned to each test. To enhance the solution of the model we develop: (a) a reformulation with disaggregated variables, (b) a procedure for the tightening of the big-M constraints, (c) a procedure that fixes some of the sequencing binary variables, and (d) logic cuts. The solution of a single large-scale problem is avoided with a heuristic decomposition algorithm that relies on solving a reduced mixed-integer program that embeds the optimal schedules obtained for the individual products. The proposed algorithm is shown to be one to two orders of magnitude faster than the full space method yielding solutions that are optimal or near optimal.
KW - New product development
KW - Optimal resource investment
KW - Scheduling
UR - http://www.scopus.com/inward/record.url?scp=1942532283&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=1942532283&partnerID=8YFLogxK
U2 - 10.1016/j.compchemeng.2003.09.019
DO - 10.1016/j.compchemeng.2003.09.019
M3 - Article
AN - SCOPUS:1942532283
SN - 0098-1354
VL - 28
SP - 1021
EP - 1038
JO - Computers and Chemical Engineering
JF - Computers and Chemical Engineering
IS - 6-7
ER -