Informs Annual Meeting 2017

SC42

INFORMS Houston – 2017

2 - Modelling Delay Dissatisfactions in Appointment Scheduling Problems

novel porosity prediction method on morphological features and the temperature distribution of the top surface of the melt pool as the LBAM part is being built. Self-organizing maps are then used to further analyze the 2D melt pool dataset to identify similar and dissimilar melt pools. To experimentally locate porosity within the Ti-6Al-4V thin wall specimen X-Ray tomography is used. 3 - High Dimensional Functional Regression for a Manufacturing Process Mostafa Reisi Gahrooei, Georgia Institute of Technology, Atlanta, GA, United States, mrg9@gatech.edu Many of manufacturing processes involve a set of process variables measured over time or space that directly affect the geometric shape of the output. Understanding the effects of the input variables on the output shape requires a model that combines the functional inputs to predict a surface. The main goal of this paper is to develop a high dimensional regression model that efficiently estimates a high dimensional output based on a set of functional inputs. To enable the proposed method for fast implementation, we exploit regularized high- dimensional regression, to decompose the output surface. With several simulation and case studies, we evaluate the performance of the proposed method. 4 - A Vision-based Approach for Real-time Quality Monitoring for Cyber-manufacturing Processes Zimo Wang, Texas A&M.University, College Station, TX, 77843, United States, zimowang@tamu.edu, Ashif S.Iquebal, Satish Bukkapatnam Laser kirigami processes that use sheet precursors offer considerable advantages over traditional powder based additive manufacturing for fast realization of functional complex shapes for custom manufacturing. An in-process monitoring tool that captures shape transformation and process thermomechanics in real time is necessary to assure cost and quality parity with mass production. We present a spatial regression model to combine information from multiple low resolution cameras for real-time monitoring of core kirigami processes. Experimental investigations on our test bed suggest that the present approach can provide fast (fps~5) and accurate (2% error) geometric feature estimates. Chair: Michael Johnson, University of Massachusetts-Boston, University of Massachusetts-Boston, Boston, MA, United States, michael.johnson@umb.edu 1 - Improving Decision-making Skills of Nonprofit Professionals Michael P.Johnson, University of Massachusetts-Boston, Department of Public Policy & Public Aff, 100 Morrissey Boulevard, Boston, MA, 02125-3393, United States, michael.johnson@umb.edu, Jason Wright Nonprofits face gaps in organizational capacity, including program design and evaluation, but previous research suggests that capacity building exercises have a positive effect. We describe a data analytics training workshop with staff from Boston-area nonprofits reflecting a wide range of sectors. Through analysis of participant work on case studies provided by the instructor, we examine how participants made sense of training materials, the various strategies employed by participants to solve three case study problems, and participant feedback about the session. Our findings provide a basis for novel interventions in community- based operations research. 2 - Refugee Resettlement via Integer Programming Andrew C.Trapp, Worcester Polytechnic Institute, Foisie Business School, 100 Institute Rd., Worcester, MA, 01609, United States, atrapp@wpi.edu, Alex Teytelboym Hundreds of thousands of refugees are resettled yearly from refugee camps to host countries. Local areas that host refugees are reluctant to open capacity, and most impose tight restrictions on the refugee family types they accept. We model this matching challenge as a 0-1 knapsack variant and explore various objectives in the search toward an optimal allocation. We tackle several new theoretical and computational obstacles, and using actual historical resettlement data to simulate realistic scenarios, we address fairness and welfare criteria that play an important role in the current process. SC43 360B Community-based Operations Research Sponsored: Public Sector OR Sponsored Session

Shuming Wang, Associate Professor, University of Chinese Academy of Sciences, No. 80 Zhong-Guan-Cun East Rd, Beijing, 100190, China, wangshuming@ucas.ac.cn, Marcus Teck Meng Ang, Tsan Sheng NG

We consider a problem of appointment sequencing with uncertain service times and no-shows. A Tolerance Aware Delay (TAD) index is proposed to quantify the user’s dissatisfaction by incorporating their delay tolerance level. We develop two appointment optimization models using the TAD index, one based on sample average approximation, and the other based on the limited distributional information of the service times. The corresponding optimization models are shown to be in the formats of a mixed-integer LP and mixed-integer SOCP, respectively, which can be evaluated efficiently using off-the-shelf solvers. Numerical experiments demonstrate the performance and insights of using our TAD models. 3 - Real Time Scheduling of Regular and Emergency Patients with Waiting Time Targets Jing Wen, Shanghai Jiao Tong University, Shanghai, China, janewen@sjtu.edu.cn, Na Geng, Xiaolan Xie To reduce the waiting time of appointed patients without affecting the treatment quality of emergency patients, this paper studies the real-time scheduling between regular and emergency patients with waiting time targets. A Markov Decision Process model is proposed to minimize the total waiting cost and structural properties of the optimal control policies are established. Several heuristic policies are proposed to solve the large size problem. Numerical experiments are performed to show the influence of parameters and compare the performances of heuristic policies. Key words: emergency patients; waiting time target; regular patients; Markov decision process; heuristic policies 4 - Two-stage Stochastic Surgery Downstream Capacity Planning with Chance Constraint and Equity Metric Shanshan Wang, PhD Candidate, Beijing Institute of Technology, 5 Southstreet Zhongguancun, Haidian District, Beijing, 100081, China, shshwang_bit@163.com, Jinlin Li This paper investigates surgery planning on a given planning period, taking downstream resources (e.g. beds in surgical intensive care unit) into consideration. Surgery durations and length of stay are uncertain. The equity of access which can be thought of as a match between patients and providers is defined as constraint of the same access levels among different types of patients. Overtime for ORs captured by chance constraint to guarantee the probability of overtime is no more than a given risk parameter. We propose a two-stage stochastic model with recourse decision, and design an enhanced decomposition scheme. Finally, a practical case study is performed by real data to test our model. 360A Manufacturing Informatics Invited: Advanced Manufacturing Invited Session Chair: Li Zeng, Texas A&M University, College Station, TX, 77843, United States, lizeng@tamu.edu Co-Chair: Qiang Zhou, The University of Arizona, Tucson, AZ, 85721, United States, q.zhou@arizona.edu 1 - Modeling of Hydrogel Swelling in 3D Printing of Artificial Meniscus Qian Wu, Texas A&M.University, 1501 Harvey Road, Apt 839, College Station, TX, 77840, United States, hi_qianwu@tamu.edu Hydrogel is a class of biomaterials that has great potential as materials in 3D printing of artificial meniscus (3DPAM). Swelling of hydrogel is critical to geometric fidelity and long-term stability of 3D printed meniscus. On the other hand, hydrogel designed for the 3DPAM application has complex structure and its swelling property is unknown. This study proposes a data-driven modeling framework for hydrogel swelling based on an integration of varying-coefficients models and engineering knowledge. In the case study, the proposed method is applied to data from a novel 3DPAM process and insight that helps understand the underlying swelling mechanism is obtained. 2 - In-situ Monitoring and Modeling of Melt Pools for Porosity Prediction of Laser Based Additive Manufacturing Processes SC42

Mojtaba Khanzadehdaghalian, Mississippi State University, Mississippi State, MS, United States, mk1349@msstate.edu

LBAM is still not widely accepted and is often considered inconsistent and unreliable for many industrial applications. The objective of this research is to characterize and use the underlying thermo-physical dynamics of LBAM captured by melt pool signals to predict porosity during the build. Herein, we propose a

82