(190a) Value of Process Knowledge and Models in Data Analytics Applications | AIChE

(190a) Value of Process Knowledge and Models in Data Analytics Applications


Valappil, J. V. - Presenter, Bechtel Oil & Gas Inc
Messersmith, D., Bechtel
Knight, K., Bechtel National, Inc.
Value of Process Knowledge and Models in Data Analytics Applications

Jaleel Valappil and David Messersmith

Bechtel Oil, Gas and Chemicals

Andri Rizhakov and Kelly Knight

Bechtel Nuclear, Security and Environmental


A typical chemical process facility requires large capital spending to achieve full production. To realize maximum commercial benefit from these capital investments over the lifecycle of the facility, it is important to utilize these existing assets to their full potential. Process facilities have relied on traditional DCS/APC systems and enterprise performance management platforms to maximize their asset utilization. There is a great potential to further enhance the value from the existing assets using new industrial internet and data analytics technologies. Some of the key applications include (in the order of increasing complexity and value):

  1. Process and control monitoring - The data can be used to build either predictive or unsupervised models to monitor the state of the process and equipments and trigger alarms for operator to take corrective actions.
  2. Predictive Maintenance - This classic application of industrial data analytics enables identifying equipment problems beforehand and scheduling maintenance appropriately. This results in greater plant availability.
  3. Facility optimization - This application of data analytics includes giving operational guidelines to maximize capacity or efficiency of the plant operation.

To realize the maximum return from investments this emerging technology area, process industry needs to properly utilize the existing plant information and resources. Some of these include:

  • Process design and operations knowledge: This is a key input for the data analytics as this helps to establish the scope and select the right variables to be included in the data models. This will also establish the right objective functions and scope for the optimization layer.
  • Fundamental process models: Several of the operating facilities utilize steady state and dynamic models developed on fundamental thermodynamic modeling platforms.
  • Existing automation infrastructure: Existing regulatory control loops and Advanced Process Control (APC) can be leveraged to benefit the industrial internet application by working in tandem with these to establish an additional layer of facility or enterprise wide optimization.

This paper will focus on value of process knowledge and fundamental models for data analytics applications in process industry. Plantwide steady state and dynamic models are developed during engineering (EPC) stage for grassroots projects. For existing facilities, either fundamental models or causal data driven models developed for various applications could be available. These models are valuable in selecting the right model structure, boosting the data models, estimating unmeasured variables and in selecting objectives and parameters for plant optimization.

Application of the above concepts to LNG facilities will be presented here. LNG industry has grown rapidly and several new liquefaction plants have started up around the world recently. The application of process knowledge and high fidelity models in the development of process monitoring/anomaly detection, process optimization and similar data driven applications will be discussed. The methodology and results of applying big data analytics to facilitate the operation of feed gas treatment system using amines in an operating LNG facility will be presented.


This paper has an Extended Abstract file available; you must purchase the conference proceedings to access it.


Do you already own this?



AIChE Members $150.00
AIChE Graduate Student Members Free
AIChE Undergraduate Student Members Free
Non-Members $225.00