Industrial optimization is the science that uses mechanisms to arrive at the best or near-best decision or option for a complex business problem. Optimization methods that we use draw upon mathematical techniques as well as rule based and experience based systems. The optimization areas that we address in business and operations are further complicated due the inherent uncertainties and variability present in the real world. Some of the optimization mechanisms that we use can be broadly classified in the following areas:
Algorithms from Operations Research help us in framing models for initial static optimization. Many of the areas in supply chains, product mix, scheduling, inventory, and location benefit from static and linear optimizations. Some areas in pricing and revenue management can also be optimized using these techniques. While these are good starting points, we frequently develop hybrid or multi-stage models along with simulation to provide more accurate and realistic decision options for our customers.
Optimization in dynamic environments is a challenging task since most real-world optimization problems are changing over time. Tools like evolutionary computation and swarm intelligence are good tools to address optimization problems in dynamic environments. We look at a combination of tools along with simulation to address dynamic optimization problems especially in the operations management area of an enterprise.
Most of the real world business problems are non-linear and exhibit different degrees of variability. Our optimization models capture these behaviors through empirical models and uncertainty models. We draw upon our more than five decades of knowledge and experience base, to accurately characterize such behavior. Overlaying this with non-linear, dynamic and Bayesian methods we are able to optimize operations, supply chains and marketing decisions.
Most shop-floor controllers control machines and processes un-intelligently. Optimal control mechanisms enable the shop-floor controllers to be more intelligent and predictive resulting in increased productivity, better quality and safety. Our methods model the dynamic control behavior of a manufacturing process and optimize it for best results. These models then guide the machine controllers to operate in the best possible way.
We frequently apply rule based systems for optimization, especially for complex areas which are otherwise intractable. Techniques like matching, genetic algorithms, neural-networks have been found useful by our DBTC experts. Dastur's experience and knowledge base plays a critical role in framing these heuristic models. Most of these techniques help us for unit optimizations like furnace control, dynamic sequencing and mechatronics.