Managing Risk Improves Production

Completing mandatory regulatory compliance leads to improved process cycle times and greater productivity.


A funny thing happened when GE Plastics Resin 2 plant (Resin 2, Ottawa, Ill.) completed the OSHA (U.S. Occupational Safety and Health Agency) 1910.119 - Process Safety Management (PSM) five year review regulations; reactor start-up times went from 40 minutes to 20 minutes and the time required to achieve on-spec product went from 8 hours to 2 hours.

GE Plastics (Ottawa, Ill.) is committed to achieving world class quality using the Six Sigma process, but trying to assimilate the Six Sigma process at the same time as completing the five-year OSHA PSM regulations seemed like a daunting effort until Resin 2 personnel struck on the idea to combine efforts.

Achieving Six Sigma requires attaining quality levels where only 3.4 failures occur in 1,000,000 opportunities (see CE, Jan. '99, p. 62 and Mar. '99, p. 87 ).

Like most processes, Resin 2 offered many opportunities for improvement, the challenge was identifying, quantifying, and prioritizing opportunities to ensure the 'big hitters' were addressed first.

Other GE locations had already proven the benefits of the Six Sigma process to identify process improvement opportunities, but a PSM hazardous analysis had identified nearly 400 process hazard recommendations. The challenge facing Resin 2 personnel was to complete both efforts simultaneously.

Two of the recommendations required upgrading the basic process control system (BPCS) and development of a reliable, emergency intervention method for conditions where the fault event or process upset was caused by the BPCS.

Resin 2 personnel agreed safety events could be viewed as quality or yield events gone awry making it nearly impossible to segregate hazard recommendations and improvement opportunities into safety and operational budgets.

Combining Six Sigma and PSM efforts appeared to be the best way to derive the highest value from the allocation of resources. Further, the combined program would be able to place appropriate emphasis on balancing production, quality, and safety. To prevent efforts from 'getting lost in the weeds,' Resin 2 personnel focused on safety and quality improvement.

They determined the need for three elements for an effective program: good team chemistry, meaningful team challenge, and use of the right tools.

Team chemistry
A team was assembled with representatives from technology, operations, engineering, and safety. Team members were chosen based on knowledge of the Resin 2 process. Additional personnel were identified to provide required support including maintenance and operations expertise, and drafting.

A Six-Sigma Master Black Belt was assigned to oversee the project and early in the project three of the core team members completed Six Sigma training and achieved Green Belt ratings.

Team challenge
When conducting a process review of this type, it is easy for the investigation to turn into a finger-pointing game among the disciplines represented. The team agreed this would not be allowed and the team remained focused on identifying opportunities for process improvement.

Fault Tree Analysis (FTA), developed in the 1960s by Bell Laboratories, was used during development of the Polaris missile, in evaluating probabilities of inadvertent Minuteman missile launches, and other military, space, and nuclear industry projects. FTA is an adaptable logic-based technique that can be applied to many situations, including chemical processes.

GE chose FTA, because it provides structure, forces a rigorous analysis, and produces quantified results necessary to prioritize opportunities.

Fault trees are constructed using failure logic in a struc-ture that identifies how failure of a particular base level element or set of elements triggers a 'top event.' When evaluating chemical processes, the base level elements are the instrumentation and equipment associated with production, manual operator actions, and quality control actions. Top events are things that lead directly to loss of product quality. Following a top down analysis, each new intermediate event is decomposed into its potential causes.

A complete fault tree analysis produces a greater understanding of 'failure events' within any process.

To ensure completeness and accuracy, Resin 2 contracted with Triconix (LeMarque, Tex.) to facilitate and provide FTA user knowledge to the project team.

The first team requirement was to assemble the latest revisions of documentation including:

  • Piping and Instrumentation Diagrams (P&IDs);

  • Operating procedures;

  • Reports on the kinetics and thermodynamics of the reaction;

  • Maintenance information; and

  • Failure rate data for the instrumentation and equipment.

Resin 2 production involves a number of different reactants and additives. Each reactant or additive is made up prior to use and manually added to reactant tanks according to recipe cards. Recipes correspond to different products manufactured in the process unit throughout the year. Because there are so many different products, the Resin 2 team decided to focus on the quality control of the reactant mixture and subsequent reaction steps.

Resin 2 team created a table listing all of the reactants and additives. For each reactant/additive, the following was identified:

  • Purpose of each reactant; and

  • Product quality issues with each reactant, and what steps are taken in the process to control the reactant.

Fault trees were developed based on information in the table and P&IDs.

Redox 1 is a batch process, thus it was important to include manual operator actions along with automatic control when constructing fault trees. Leaving out either event jeopardized the ability to accurately identify root causes affecting quality, yield, and/or safety.

Redox 1's initial fault tree was developed showing the top event to be incorrect make-up of the Redox 1 solution. Immediate causes of incorrect Redox 1 make-up are:

  • Incorrect number of bags of Redox 1 added;

  • Incorrect amount of acid added;

  • Incorrect metering of water into the tank;

  • Loss of mixing in the tank, and

  • Loss of solution temperature control.

With the fault tree constructed, the next step was to enter failure rate data and quantify the fault tree.

Original results for Redox 1 make-up

Basic event in fault tree

Percent contribution to the probability to fail





























Failure rate data can be obtained from plant experience or from industry published data. Resin 2 relied on industry data and, where possible, validated the industry data against Resin 2 operating experience.

Sources of industry data include:

  • OREDA: Offshore Reliability Data Handbook, 3rd Edition, Det Norske Veritas Industri Norge as DNV Technica, Norway, 1997;

  • Guidelines for Process Equipment Reliability Data, Center for Chemical Process Safety of the American Institute of Chemical Engineers, NY, NY, 1989;

  • Non-Electronic Parts Reliability Data, Reliability Analysis Center, Rome, NY, 1995;

  • Failure Mode/Mechanism Distributions, Reliability Analysis Center, Rome, NY, 1991;

  • IEEE Standard 500-1984 Guide to the Collection and Presentation of Electrical, Electronic, Sensing Component, and Mechanical Equipment Reliability Data for Nuclear-Power Generating Stations, The Institute of Electrical and Electronics Engineers, New York, NY, 1983; and

  • Lees, F.P, Loss Prevention in the Process Industries, England, 1980.

Fault tree analysis involves Boolean algebra for mathe-matical quantification, lending itself to computer modeling to obtain quantification of the fault trees. Resin 2 personnel selected a computer model capable of performing minimum cut set determination and Boolean algebra. Shortcut calculation techniques, such as adding the failure probabilities, are only valid when the probability is significantly less than 0.1. Resin 2 personnel expected the Redox 1 analysis to contain several failure probabilities greater than 0.1. If cut-set overlap had not been taken into account, failure probability would have been greatly over-predicted.

When fault tree cut sets were quantified, results were presented to all team members, also to maintenance and operations to determine if calculated results matched plant experience.

Overall probability for Redox 1 make-up to be performed incorrectly was 3.093 x E-03 and translates into Annual Failures divided by Total Annual Opportunities times 1,000,000 to yield Defects Per Million Opportunities.

Examination of the percent contribution of the cut-sets to the overall probability of failure showed loss of agitation contributed over 75%. However, Resin 2 Plant operational data did not confirm this finding. Loss of agitation had not been listed as a major contributor to any previous off-spec Redox 1 batches.

Since probability of failure and percent contributions were higher than expected, the fault tree was re-examined to determine what operational steps might have been missed in the analysis. The goal of the review was to improve the accuracy of actual probability of failure.

Careful review revealed there were operator verifications missing from the original fault tree. Manual addition of Redox 1 and acid required the operator to open the make-up tanks. While adding the chemicals, operators were able to verify agitator operation (mixing) was occurring. During the same manual loading process, operators also verified water levels were correct and heating was taking place. This new information was added, new fault trees were created, and new calculations were produced.

Estimates of the probability the operator would not conduct each of the operator checks were added and the fault tree cut-sets were again quantified. The new estimated probability of the Redox 1 make-up being incorrect decreased to 1.09E-4. The percent contribution of the elements changed dramatically.

Revised results for Redox 1 make-up

Basic event in fault tree

Percent contribution to the probability to fail





























The main contributor to new probability of failure results were verifications provided when operators added acid and Redox 1. A second review was conducted with operations and the calculated results more nearly matched plant experience.

The Resin 2 team determined further improvement in the quality control in the Redox 1 make-up step could be achieved by improving procedures for the manual addition of the Redox 1 and acid. By improving the procedure, the probability of addomg an incorrect number of bags of Redox 1 or the injection of incorrect amounts of acid would be further reduced. Since these two items accounted for nearly 100% of failure potential, reducing probability of incorrect additions would immediately have a positive impact on the quality of Redox 1 make-ups.

Example entry from Resin 2 improvement study





Redox 1

Redox 1 is the activator for the INITIATOR

Low flow can result in retarded reaction rate, poor conversion. High flow can cause early depletion of INITIATOR, resulting in the same effects as low flow INITIATOR. Redox 1 make-ups at improper temperatures can cause problems. At the time of make-up, visual color check is done.

Redox 1 solution is made up every 16 hours or so. As stated in the Issues category, the Redox 1 solution does have a finite batch shelf life.

The entire evaluation methodology was repeated for each reactant make-up step. When the assessment was complete, Resin 2 personnel had developed an action item list based on quantification of potential failure events. Action items included:

  • Improve start-up procedures;

  • Improve clarity of recipe procedures;

  • Update resin recipes based on input from technology;

  • Reduce reliance on operators for flow control of the reactants;

  • Improve control of reactant flows to ensure that reactant mix is correct;

  • Improve testing and maintenance frequency on several process critical components; and

  • Provide better control of the main reaction initiator to ensure appropriate reaction occurs when desired.

The action item list served as basis for overall production improvements and specific tasks were developed for each production step.

Improvement recommendations included:

  • Initiate a consistent alarm strategy for the control system-alarm for deviation after X time and alarm/shutdown after Y time;

  • Upgrade flow control based on polymer properties (pump and metering);

  • Improve Redox package delivery;

  • Validate delivered quantities against specified quanti-ties for each flow stream;

  • Utilize redundant mass-flow metering for each flow stream and take advantage of the internal self-diagnostics inherent in these devices;

  • Change piping configuration to allow on-line verification of meters;

  • Upgrade specific feed pumps to variable stroke pumps to provide reliable delivery at desired flow range; and

  • Increase the amount of line flushing to clear Redox 1 from delivery lines.

Specific tasks implemented included:

  • Development and documentation of corrective actions to be used when flow rates were confirmed to be outside of recipe ranges;

  • Conversion of chemical additions from manual to BPCS control;

  • Significant changes in how the process is started-up; and

  • Installation of a manually operated, independent emergency shutdown system that ensured the process could be shutdown independent of the BPCS.

Results delivered
Before conducting the Six Sigma and FTA evaluations, reactor start-up time averaged 40 minutes. Implementation of new start-up procedures and automation of material charging decreased reactor start-up time to 20 minutes. Associated with the reduced startup time downstream operations improved, reactant conversion improved product yield, and operators were able to control the process within safer operating boundaries.

The Resin 2 Plant study was successful due to the comprehensive approach used. The quantitative evaluations provided sufficient justification to upper management to readily approve improvement investments.

Beyond benefits already listed, this analysis helped identify opportunities to improve quality control systems and identification of BPCS improvements that could be made with minimal impact and investment. Not to be overlooked is the improvement in process reliability and safety resulting from the thorough and careful analysis required of the Six Sigma improvement process.


No comments
The Engineers' Choice Awards highlight some of the best new control, instrumentation and automation products as chosen by...
The System Integrator Giants program lists the top 100 system integrators among companies listed in CFE Media's Global System Integrator Database.
The Engineering Leaders Under 40 program identifies and gives recognition to young engineers who...
This eGuide illustrates solutions, applications and benefits of machine vision systems.
Learn how to increase device reliability in harsh environments and decrease unplanned system downtime.
This eGuide contains a series of articles and videos that considers theoretical and practical; immediate needs and a look into the future.
Choosing controllers: PLCs, PACs, IPCs, DCS? What's best for your application?; Wireless trends; Design, integration; Manufacturing Day; Product Exclusive
Variable speed drives: Smooth, efficient, electrically quite motion control; Process control upgrades; Mobile intelligence; Product finalists: Vote now; Product Exclusives
Machine design tips: Pneumatic or electric; Software upgrades; Ethernet advantages; Additive manufacturing; Engineering Leaders; Product exclusives: PLC, HMI, IO
This article collection contains the 5 most referenced articles on improving the use of PID.
Learn how Industry 4.0 adds supply chain efficiency, optimizes pricing, improves quality, and more.

Find and connect with the most suitable service provider for your unique application. Start searching the Global System Integrator Database Now!

Cyber security cost-efficient for industrial control systems; Extracting full value from operational data; Managing cyber security risks
Drilling for Big Data: Managing the flow of information; Big data drilldown series: Challenge and opportunity; OT to IT: Creating a circle of improvement; Industry loses best workers, again
Pipeline vulnerabilities? Securing hydrocarbon transit; Predictive analytics hit the mainstream; Dirty pipelines decrease flow, production—pig your line; Ensuring pipeline physical and cyber security