Many approaches and methods related to manufacturing and supply chain management were developed by statisticians and scientists in operations research, with limited knowledge about the practical context in which their models are supposed to be used. These methods primarily focus on the optimization problem and are based on a number of assumptions that usually do not hold in real-world manufacturing and supply chain environments. They are mostly based on stochastic models that are too complex for operators to understand and cannot properly handle the uncertainties of the actual operations settings.
Overall these methods and operations management software based on these methods are quite difficult to build and implement. First, they require a complete and accurate data set to properly function, and that level of accuracy in operational data usually comes with significantly increased costs. Second, they require good fundamental understanding of statistical methods, queuing theory and mathematical optimization techniques by the operators. There is no single algorithm that can be applied generically; each is defined for a specific case and with certain assumptions. This creates a huge gap between theory and practice. That is precisely the reason why most organizations build their manufacturing and supply chain policies based on a combination of ball-park estimations, simple spreadsheet based arithmetic, and "gut feeling" of planners and managers, which often yield inferior results.
The KBIA technology was developed to overcome these challenges and bridge the gap between theory and practice. It accomplishes this by;
- Eliminating the user-interface implications related to the complexity of the algorithms.
- Presenting data in cognitively efficient visualizations that are delivered within the context and domain knowledge of the users.
- Significantly reducing the data requirements by its native ability to function with imprecise and missing data elements.
- Adapting itself to decision making patterns of users over time, thus eliminating the tedious data maintenance work.
For example the traditional solutions to multi-echelon, multi-site inventory optimization problem require a good understanding of statistics to decompose the supply chain into single nodes with specific demand distributions. These solutions use statistics to calculate mean, variance and other statistical attributes of the distributions of back-orders and associated delays at each node. Once this stochastic model is built, a linear or Lagrangian optimization algorithm is applied to trade-off between total inventory carrying costs vs. cost of backorders.
Contrarily KBIA powered webOps TIS constructs the decision making model based on user decisions. It collects this information with context driven scenarios by obtaining decisions from users in the form similar to "If my demand rate is around 500units/week, and my lead time is around 4 weeks, I want to keep 6 weeks of supply at my depot and 2 weeks of supply at my VMI location". While the users provide these inputs through few "clicks" and "drag-and-drops" for each of these system generated scenarios that are based on actual data, they are presented with the implications of their decision rules in terms of Key Performance Measures, such as "Inventory Turn", "Committed Service Time", "Delivery Performance to Customer Request Date". These implications are provided by the webOps TIS using state-of-the-art scientific methods through KBIA technology. Now the users can understand the end results that are based on theory, and adjust their decision making rules to explore rule sets with improved performance. This brings the best of theory and practice together to significantly improve operational performance.
