Advanced Operational Analytics make it simpler for specialty chemical companies to quickly and continuously understand, assess and act on the noisy, consumption patterns of energy. The increased understanding and timeliness of assessments can lead to significant reductions in energy consumption. Energy reduction is an increasingly critical lever in operational excellence because it delivers desirable impacts to both the environment and the bottom line.
Effective energy reduction programs benefit from using a working energy model. A specialty chemical batch operation may have thousands of instruments, each generating data at its own pace using its own sampling algorithms. Operational analytics packages can easily line up the data on the same time axis and determine the various relationships between the points. Such a seemingly simple action had been historically complex and a discouraging activity. The discovered relationships are defined by energy models which can be part of an effective low to no capital energy reduction program that consistently delivers results. The steps to successfully building and running such a program are outlined below.
1. Develop energy models, which predict energy consumption using utility instrument data and process instrument data generated by the control systems. Development of multi-variate models is quite simple with new operational analytical tools.
2. Look for areas where the model performs poorly. Such areas normally indicate a missed modeling instrument, poor controller performance or a gap in standardizing operating procedures. In the steam model example, the variation increase in one weekend over another looked strange. The investigation revealed that poor control tuning was wasting significant amounts of energy.
3. Look for the model parameters that are not statistically significant but should be, or have values that do not follow the laws of thermodynamics. For example, The parameters in the steam model in graph 1 indicated that one of the steam jet valves had no impact on overall steam demand. The valve was tested and was not closing properly leading to significant waste. Another parameter in the model was the high intercept, which shows a higher than expected fixed energy load.
4. Update the model after resolving the initial findings and engage operations with routine daily reviews of model deviations. Only statistically significant deviations are to be discussed and investigated. Investigations will lead to improved models or improved operations. Such discussion can lead to failed steam traps and other equipment faults.
How and Why Energy Models Work:
With thousands of instruments, failure of one or multiple instruments is probable. Many times, failures are not detected because control systems can compensate and hide them. Fortunately, the laws of physics and thermodynamics are well understood and consistently hold true. For example, when a steam valve opens, there are multiple responses in the system defined by scientific laws:
1. The process will respond by absorbing the energy supplied by the steam
2. The steam generating system will respond to added demand
3. The cooling system will respond as needed to the additional heat (e.g. condense process vapors).
When the responses of these systems do not follow the laws of nature, there is a defect in the system: a failed meter, valve, trap, or exchanger. For example, when a valve feeding a steam jet opens from 0 percent to 100 percent, the plant steam generation should on average increase by the designed amount and the heat load on the cooling tower should increase by a similar amount. When the response is not consistent with expectations, there is an indication that equipment is malfunctioning and not fulfilling its purpose. There are many more operations in a site occurring simultaneously. The beauty of a comprehensive model is that it can factor out all of the other operations and isolate the impact of the valve alone. Only a handful of energy models are needed to monitor thousands of instruments on a site. When the process deviates from a model, an instrument has failed and an investigation is required.
The technique of energy modeling has been a tool for improvement for quite some time. Their use has allowed for continuous, low to no capital improvement for years. Unfortunately, until the existence of operational analytical packages, models were very difficult to develop, quite simplistic, rarely updated and would eventually become obsolete. Today’s analytical packages resolve all of those previous modeling issues. Current energy models are now very easy to develop and update, have multi-variate complexity, and consistently identify energy reduction opportunities. They were the cornerstone of the improving trends below. It is expected that the early findings of operational analytics packages will increase the slope of improvement and extend it for quite some time.