PORTFOLIO

Unit Stability

A unit with more than 30 years of operation was analyzed to determine sources of instability in reactor feed concentration.

First, the impact of reactor feed concentration on product quality and unit operability was quantified through use of statistical based modeling. Because of the severity of the instability, a model was required to deconvolute the impact vs. time.

An extension of the model was used to predict and track concentrations throughout the unit including the reactor feed. Through the use of Monte Carlo based modeling, it was proven that the method and feed location of a key raw material was responsible for the instability. A project was scoped and completed. Once online, the unit had direct control over this concentration enabling another entire generation of possible optimization.

Compressor

A large integrated refrigeration system that included a 5-stage compressor was operating outside of design parameters and caused severe restrictions to operating rates

The problem was convoluted because of multiple feedback points in the process. Each variable in the process was mapped and modeled to ensure fundamental relationships that should exist had not changed. More than 20 hypotheses were generated from this analysis. Ultimately, by mapping the compressor performance using thermal imaging against the Mollier diagram for the refrigerant, it was found the 4th stage had moved out of range. This work allowed justification for the high cost of a rebuild. After disassembly, it was found the compressor had fouling present on the 4th stage likely caused by a prior refrigerant contamination event (1 of the 20 hypotheses).

Centrifuge Optimization

A unit was restricted due to reliability of two centrifuges operated in parallel. A Six Sigma project was chartered to increase reliability of the centrifuges and prevent associated unplanned events.

The first action completed was to analyze the system to understand why two centrifuges were needed to run near maximum capacity. The result was eye opening. Each centrifuge had a theoretical capacity that exceeded the average required instantaneous run rate, however because of upstream instability, the effective required run rate was significantly higher. An innovative process control scheme was designed and implemented that eliminated the instability therefore making it possible to achieve maximum rates on one centrifuge.

Several operations schemes were designed and modeled to address operation of the 'now spared' centrifuge. Furthermore, the reliability of each individual centrifuge was increased in multiple ways through redesign of a washing system, redesign of the oil system, and custom design of gaskets to increase mean time to failure.

The actions taken in this project eliminated the centrifuges as a bottleneck in the process.

Tower High Pressure

High pressure repeatedly occurred in a distillation column on startup threatening to overpressure the column. The source of the problem could not be found.

Through use of ASPEN modeling, it was determined the most likely cause was contamination by water. It was proven that the VLE of a water/product mix would exactly match that of the column. Personnel had already sampled the column and determined there was no issue with contamination. Tests for this hypothesis were redesigned to reduce probability of a false negative and the cause was confirmed.

An FMEA was conducted to uncover all possible ways that this contamination could occur. Two major routes, previously unknown, were found. One was deemed the root cause, however both were mitigated with preventative actions to prevent recurrence.

Tower Pressure Drop

A distillation column periodically exhibited extremely high pressure drop causing an inability to run the unit. Some prior efforts were successful at temporarily resolving the problem, however, the issue continued to recur roughly once every 1-2 years.

The tower exhibited symptoms of a flood. A deep dive into the tower and old documentation revealed this problem had been recurring on and off for roughly 30 years. Cleaning seemed to correct the problem, but it was unclear if the cleaning or complete restart of the equipment could be credited.

Because of the opened ended nature of the problem, the hypothesis matrix constructed was extensive and included foaming as one hypothesis. Ultimately it was found the material exhibited a fairly rare form of foaming via Ross mechanism where otherwise stable ternary systems will foam as they approach the plait point. This was confirmed by lab experimentation. Two improvement plans were devised. The selected plan was implemented more than 10 years ago and has successfully eliminated the issue.

Variable Margin Velocity

Because of the size, depth, and interconnectivity of a portfolio of products, knowledge of how each contributed to the overall business performance was difficult to determine. This issue impeded optimization of both total cost to serve and portfolio reliability.

The variables required to understand and map network flow of products are typically scattered throughout different functions and formats. The exercise to map one product may be trivial, but when thousands are considered, the task because untenable.

To address this problem, code was used to first harmonize and collect inputs for a variety of various sources. Once this was complete, values for cost of technology and intensity of production were investigated and assigned to each product culminating in a overall value stream map of the business. The output enabled line-of-sight to the importance and value of each product throughout this extensive portfolio and allowed for the entire operation to be lifted through subsequent optimization.

Pipeline Reliability

Pipelines carrying a specific hazardous product were prone to failure resulting in extreme cost of repair and high EHS risk.

The pipeline failures could be attributed to a unique fundamental property of the material they carried. While this was well known by operations, the design was not sufficient to allow adequate protection against failure.

The project kicked off with dynamic FEA modeling of prior events that allowed deep understanding of how those failures occurred. 'Most-Effective Technology' was developed that provided engineering guidelines for retrofit and operation of existing pipelines along with construction of new pipelines. In context of this effort, new technology was scaled for real time monitoring that allowed unprecedented access into the health and operation of these lines. To date, no pipeline has failed following completion of this project.