Production Planning in an Engineered-to-Order Environment (GE Hydro, CAE, CMC Electronics
Commercially available planning systems have been predominantly designed for operations which make products in volume over relatively long periods of time. As such, the systems have to assist in the planning, ordering and scheduling of materials as well as the scheduling of jobs on the factory floor. For volume production, this means controlling material flow on one hand, as well as capacity and workflow on the other.
In an engineered-to-order environment, each contract brings a new requirement for materials and for the scheduling of tasks. GE Hydro (hydroelectric turbine-generators), CAE Electronics (aircraft simulators), and CMC Electronics (avionics, communications) work in such an environment. Current MRP systems (manufacturing resource planning) do not work well in an engineered-to-order environment. A new framework for the design of an MRP system has been developed which addresses this problem. Framework features include: integrated finite capacity scheduling, combined planning for engineering and manufacturing, and new scheduling algorithms for dynamic scheduling. Consulting has been done with JD Edwards, an MRP supplier, on what features are needed in their new design of an MRP system for ETO companies. Work is continuing with the companies to apply the new principles.
A study has been initiated to determine how the principles for time-based competition used in manufacturing could be adapted for use in mining by Inco Ltd. The objective is to develop a philosophy and set of paradigms which will allow mines to be operated more effectively by using time-based methods. A review has already been made of both lean manufacturing and time-based competition methods. These methods were considered from the points of view of both manufacturing production and the general business environment. To date, a new concept of "cellular mining" has been developed which holds great promise to improve the productivity of mining. Simulations of actual mining operations at Inco are underway to test the new ideas.
In-Cycle and Cycle-to-Cycle Machine Control (Industrial Materials Institute, Quality Thermoforming)
Recently, there has been a rapid advancement in the development of simulation tools and process know-how in the field of polymer part manufacturing, in particular for moulding operations. However, given the often low profit margins on commodity parts, these processes can only be profitable if they are highly optimized for quality and short cycle times. To address this challenge, a broad research and development program in model-based control of forming processes has been created with Benoit Boulet and Vince Thomson, McGill University, and Robert DiRaddo, Industrial Materials Institute, National Research Council, and Quality Thermoforming (Toronto). Control of discrete moulding systems can be performed in two different time scales, namely in-cycle and cycle-to-cycle. For in-cycle control, the sensor measurement and the corrective control action are taken within a cycle as the part is being made. For cycle-to-cycle control, the corrective control action is taken in between cycles. A controller will be developed for a thermoforming machine being developed by Quality Thermoforming.
Modeling Coordination in Design Teams
The goal of this research is to provide a comprehensive understanding of coordination, its principles and its use. It intends to show how coordination is driven through the better use of information, indicating the equivalence between coordination and integration of information. The analysis is focused on engineering processes. Mathematical models are being developed in order to be able to systematically demonstrate the superiority of certain coordination methods in given situations.
Estimating Design Projects (Astec APS, CMC Electronics, GE Hydro)
Engineering design projects have been plagued by severe schedule and budget overruns. This problem persists in spite of the significant advances that have been made in design technology over the last two decades. Techniques for estimating design effort and duration have been developed. These methods consider product related as well as people and process related factors. The methods have been validated by empirical studies based on data from historical and ongoing projects from Astec APS, BAE Systems and GE Hydro. Studies are continuing with GE Hydro to test the models further. The work is in collaboration with Hamdi Bashir, Concordia University. The new method gives an error of about 10% compared to present estimation errors of over 30%. GE Hydro has been using the methods since 1999. The method will appear shortly in a textbook on engineering design.
Modelling the Dynamics of Concurrent Engineering
A stochastic model of concurrent engineering has been built to study how the micro-activities of group work, parallelism and the structure of work are linked to the macro-effects of time compression and effort reduction. Outcomes of the research have determined how to optimize processes for new product introduction.
Determining Cause and Effect with Sparse Data (Nortel Networks)
The architecture for a process monitoring system has been designed; it is the first to permit measurement and monitoring of the performance of knowledge work processes. The critical element of the system is a reasoning function which determines cause and effect relationships for processes which have incomplete data. Artificial intelligence (AI) techniques are being used to design and develop methods for determining cause and effect relationships for comparing processes, which can have incomplete data and/or slightly different points of comparison, and for comparing performance to standard operating practices. A prototype system for testing these principles is being built using object technology which will allow the coupling of the many different functions necessary in such a system.
Developing an Integrated Process Monitoring System
This project deals with the design of a process monitoring system, which can perform realtime measurement of knowledge work processes, compares these measurements with data from process models of planned processes, and provides feedback about process performance. The research focuses on the design and testing of algorithms which can compare the data and measurements collected from real processes with corresponding items in desired process models.