Solutions
Unplanned outages in the industrial world are always undesirable, and can sometimes even be disastrous. They cause bottlenecks, significant labor overhead, more scrap production, and even loss of business opportunities. And unplanned downtimes are very expensive. According to recent statistics, plants in the oil and gas, chemical, and power industries routinely suffer 5 to 7% unplanned downtime losses due to poor maintenance practices. And in the automotive industry, it has been shown that one minute of unplanned downtime costs an avg. $22,000, or $1.3 million per hour.
With such high price to pay for unplanned downtimes, why are most industrial...
Automated production lines are highly complex, many-step systems with increased value of product in each consecutive step. Sensors measure the performance of the machinery, and occasionally produce messages alerting to minor faults or abnormal events. Most of those are inconsequential and can safely be ignored, however some can lead to critical failures. When an error occurs or the condition of a machine declines, the cost of that error is higher the longer it takes to be detected. Additional scrap parts are produced by the failing machinery, and, if the condition becomes bad enough, unscheduled production stop...
A modern plant is highly complex production machine, ruled by predictable causes and time-delayed effects. But like any truly complex system, full grasp of everything that happens in it defies the ability of any one human. Knowing where and what the optimal running state of the complex system is at any moment, 24 hours a day, 7 days a week, is beyond even the most skilled team. It is no wonder then that sub-optimal operating procedures, poor design or improper scheduling of production have been identified as the main sources of many plant reliability problems. But...
In an automated, many-step production lines, scrap is expensive. It incurs a higher cost, the more value steps scrap parts pass on a production chain, before they are identified. Or even worse, the scrap might not be identified at all, and the customers might receive a lower quality product. And while quality control measures can usually catch most scrap parts, the process is very effort-intensive and time-consuming, and leaves too much room for human error.
With the Advanced Quality Controller (AQC) algorithmica technologies solves this issue by using advanced machine learning to replace unreliable, time-intensive manual...
Maintaining your machinery in a healthy state, and being aware when its state deteriorates, is crucial to successfully running any industrial plant. It is, however, a nontrivial task usually accomplished by employing sensors and having a fixed range of values that the measurements should be in. If the values are outside this range, then an alarm is sent out. However, this method rarely works as intended. The reality of production and aging equipment means that many false alarms are sent out, and also some unhealthy states are never alarmed. Is there a better way?
The Intelligent Health Monitor (IHM) sets...
In a plant there are some variables you want to always keep constant. An example might be the temperature of a chemical reaction. But that is less than trivial, as there are many influencing factors you cannot control, such as the weather, quality of raw materials, or outages upstream of your production. And while common advanced process control (APC) software is meant to help you solve this problem, these packages have many limitations; for instance only a very low number of independent variables, difficult and effortful implementation, no automatic adjustment to changes, etc. Most prominently, you are usually required to...
Nowadays we try to measure almost everything happening in a plant. Sensors are everywhere. But not everything you want to know can be measured with a physical sensor. What if efficiency was just a number you could look up? Some measurements are difficult and expensive enough to not be practical as a physical sensor. What if you could carry out the analysis of a chemical composition without taking a sample but merely by looking it up? That is what the Intelligent Soft Sensor (ISS) from algorithmica technologies makes possible.
ISS is not a physical sensor, but rather a computational model...
We live in the era of big data. Modern businesses, and especially modern plants, generate huge amounts of digital data from sensors and lab analyses. But while the data is there, it is too much and too complex for traditional data analyzing methods to glean any useful insights from. It is not only the sheer amount of variables, but also the complexity of the correlations, and the varying time-delay in the effects, that make industrial big data an almost insurmountable challenge for most methods of analysis to handle.
But that is where advanced mathematics, especially machine-learning...
In order for industrial processes to yield predictable and desirable outcomes, they follow very specific, carefully balanced recipes. Sometimes, however, that recipe must be changed. Perhaps one of the ingredients is no longer available or is now produced by another supplier. Or a client might be asking for a tighter specification. Often, market forces shift the pricing of ingredients such that it becomes necessary to reduce one component to a minimum.
In order to find a modified or new process recipe that will reliably produce a product with the desired specifications, plants usually perform experiments to test various possible combinations....
