Pierre Latour discusses issues with realizing the full commercial potential of proper control system performance indicators and argues that his method, Clifftent, can help mitigate problems and maximize profit. He outlines how Clifftent uses calculus and statistics to determine the optimal setpoints and limits for process variables to locate the maximum expected profit. Latour claims that adopting Clifftent's performance-based licensing approach in the 1990s could have led to a healthier and more profitable industry today.
Reliability Centered Maintenance (RCM) is a proven, logical, sensible approach that helps companies improve reliability. Yet most companies are not getting the return they expected. They see RCM as too much trouble for too little reward.
So that’s why we decided to publish this new report. Find out why RCM doesn’t work, what needs to change and how to put RCM to work at your company so it doesn’t become another Resource Consuming Monster.
Nowlan and Howard Heap published “Reliability Centered Maintenance”, the ground-breaking study that changed maintenance forever. Yet myth, mystery and confusion about RCM still abound.
So let’s begin with the basic truths. To paraphrase RCM practitioner, Doug Plucknette, of GPAllied, RCM is a structured process developed to ensure the designed safety and reliability capabilities of a process or piece of equipment. The beauty of understanding the RCM process is it can be applied to virtually any physical asset in any plant around the world.
RCM’s roots go back to the early 1960’s, when the commercial airline companies were considering buying the new jumbo jet, the Boeing 747. At the time, the airlines religiously practiced time-based preventive maintenance. Why? Because the conventional wisdom was that equipment wears out over time. So that meant taking planes out of service for maintenance every 1,000, 5,000 or 10,000 hours.
But the problem with the 747’s was the amount of maintenance specified by federal regulators was three times more than the maintenance required for Boeing 707’s. That meant more time in maintenance, more time out of service, and a huge disruption to operations.
Clearly, the airlines’ traditional approach to maintenance would not be economically feasible for the new jumbo jets.
So the airlines had two choices: Either buy a larger fleet of planes or develop a more economical approach to maintenance.
That’s why United Airlines led a task force to re-evaluate the concept of preventive maintenance and determine the most economic strategy, without compromising safety. The result was the process that we now know as RCM, which was successfully employed on the 747 and all subsequent jet aircraft.
Reliability Centered Maintenance (RCM) is a proven, logical, sensible approach that helps companies improve reliability. Yet most companies are not getting the return they expected. They see RCM as too much trouble for too little reward.
So that’s why we decided to publish this new report. Find out why RCM doesn’t work, what needs to change and how to put RCM to work at your company so it doesn’t become another Resource Consuming Monster.
Nowlan and Howard Heap published “Reliability Centered Maintenance”, the ground-breaking study that changed maintenance forever. Yet myth, mystery and confusion about RCM still abound.
So let’s begin with the basic truths. To paraphrase RCM practitioner, Doug Plucknette, of GPAllied, RCM is a structured process developed to ensure the designed safety and reliability capabilities of a process or piece of equipment. The beauty of understanding the RCM process is it can be applied to virtually any physical asset in any plant around the world.
RCM’s roots go back to the early 1960’s, when the commercial airline companies were considering buying the new jumbo jet, the Boeing 747. At the time, the airlines religiously practiced time-based preventive maintenance. Why? Because the conventional wisdom was that equipment wears out over time. So that meant taking planes out of service for maintenance every 1,000, 5,000 or 10,000 hours.
But the problem with the 747’s was the amount of maintenance specified by federal regulators was three times more than the maintenance required for Boeing 707’s. That meant more time in maintenance, more time out of service, and a huge disruption to operations.
Clearly, the airlines’ traditional approach to maintenance would not be economically feasible for the new jumbo jets.
So the airlines had two choices: Either buy a larger fleet of planes or develop a more economical approach to maintenance.
That’s why United Airlines led a task force to re-evaluate the concept of preventive maintenance and determine the most economic strategy, without compromising safety. The result was the process that we now know as RCM, which was successfully employed on the 747 and all subsequent jet aircraft.
Integrated asset model can provide a single source of the truth across the full stream for how molecules and operating conditions behave at the unit- and asset-wide level;
Thereby providing actionable insights into production activities that can drive convergence in decision-making and action across organizational silos.
OSIsoft White Paper "Impacting the Bottom Line" in O>jeerd Zwijnenberg
In a new era of heightened oil-price volatility, data and technology are crucial in helping operators cut costs and maximise value; 10 real-world examples of oil and gas innovators using data for economic effect
[Oil & Gas White Paper] Liquids Pipeline Leak Detection and Simulation TrainingSchneider Electric
Increasingly, pipeline operating companies must deal with regulations that focus on environmental protection. The goal of the regulations is to minimize pipeline leaks that not only endanger the environment but also result in operator downtime and financial penalties. Identifying, verifying and responding to the abnormal conditions around a potential leak require best practices, including controller training.
A computational pipeline monitoring (CPM) system uses real-time information from the field – such as pressure, temperature, viscosity, density, flow rate, product sonic velocity and product interface locations – to estimate the hydraulic behavior of the product being transported and create a computerized simulation. With it, controllers can be alerted to actual operating conditions that are not consistent with the calculated conditions and might signal the existence of a pipeline leak. Different CPM methodologies provide different leak detection capabilities, so different methods, or a combination of methods, might be better applied to different operations.
A comprehensive CPM system also supports training best practices that help engineers and controllers develop intimate knowledge of the control system interface, alarming functions and response actions. It is an efficient way to implement refresher training to cover network modifications and expansions and to accurately document training, testing results and qualifications. Computerized simulation has demonstrated to provide more comprehensive and effective training for a specific pipeline than on-the-job training. For this reason, it is the preferred method of the U.S. Department of Transportation’s Pipeline Hazardous Materials Safety Association (DOT-PHMSA) for training controllers to recognize the abnormal conditions that might suggest a leak and to optimize the safety of the pipeline operation.
Schneider Electric’s SimSuite Pipeline solution is based on a real-time transient model that includes leak detection capabilities, crucial to safety and environmental concerns; a simulation trainer application for targeted and effective training of operational staff; and forecasting and planning functionalities that help improve business intelligence. Together, these capabilities help the pipeline operator reduce operations cost as well as comply with regulations.
Experimental evaluation of control performance of MPC as a regulatory controllerISA Interchange
Proportional integral derivative (PID) control is widely practiced as the base layer controller in the industry due to its robustness and design simplicity. However, a supervisory control layer over the base layer, namely a model predictive controller (MPC), is becoming increasingly popular with the advent of computer process control. The use of a supervisory layer has led to different control structures. In this study, we perform an objective investigation of several commonly used control structures such as “Cascaded PI controller,” “DMC cascaded to PI” and “Direct DMC.” Performance of these control structures are compared on a pilot-scale continuous stirred tank heater (CSTH) system. We used dynamic matrix control (DMC) algorithm as a representative of MPC. In the DMC cascaded to PI structure, the flow-loops are regulated by the PI controller. On top of that a DMC manipulates the set-points of the flow-loops to control the temperature and the level of water in the tank. The “Direct DMC” structure, as its name suggests, uses DMC to manipulate the valves directly. Performance of all control structures were evaluated based on the integrated squared error (ISE) values. In this empirical study, the “Direct DMC” structure showed a promise to act as regulatory controller. The selection of control frequency is critical for this structure. The effect of control frequency on controller performance of the “Direct DMC” structure was also studied.
Big Data Analytics for Commercial aviation and AerospaceSeda Eskiler
globalaviationaerospace.com
An opportunity for insight in the changing commercial aerospace business
Vision for New Applications of Analytic Insight in Commercial Aerospace
Benefit of Big Data Analytics for the Airline Operator
Modern, Mobile Experience
Big Data Analytics In Action
Predictive Analytics To Prevent Engine Events
Predictive Analytics Improves Safety and Quality
Predictive Analytics Keeps More Planes in the Air
Predictive maintenance framework for assessing health state of centrifugal pumpsIAESIJAI
Combined with advances in sensing technologies and big data analytics, critical information can be extracted from continuous production processes for predicting the health state of equipment and safeguarding upcoming failures. This research presents a methodology for applying predictive maintenance (PdM) solutions and showcases a PdM application for health state prediction and condition monitoring, increasing the safety and productivity of centrifugal pumps for a sustainable and resilient PdM ecosystem. Measurements depicting the healthy and maintenance-prone stages of two centrifugal pumps were collected on the university campus. The dataset consists of 5,118 records and includes both running and standstill values. Additionally, Spearman statistical analysis was conducted to measure the correlation of collected measurements with the predicted output of machine conditions and select the most appropriate features for model optimization. Several machine learning (ML) algorithms, namely random forest (RF), Naïve Bayes, support vector machines (SVM), and extreme gradient boosting (XGBoost) were analyzed and evaluated during the data mining process. The results indicated the effectiveness and efficiency of XGBoost for the health state prediction of centrifugal pumps. The contribution of this research is to propose an effective framework collectong multistage health data for PdM applications and showcase its effectiveness in a real-world use case.
Integrated asset model can provide a single source of the truth across the full stream for how molecules and operating conditions behave at the unit- and asset-wide level;
Thereby providing actionable insights into production activities that can drive convergence in decision-making and action across organizational silos.
OSIsoft White Paper "Impacting the Bottom Line" in O>jeerd Zwijnenberg
In a new era of heightened oil-price volatility, data and technology are crucial in helping operators cut costs and maximise value; 10 real-world examples of oil and gas innovators using data for economic effect
[Oil & Gas White Paper] Liquids Pipeline Leak Detection and Simulation TrainingSchneider Electric
Increasingly, pipeline operating companies must deal with regulations that focus on environmental protection. The goal of the regulations is to minimize pipeline leaks that not only endanger the environment but also result in operator downtime and financial penalties. Identifying, verifying and responding to the abnormal conditions around a potential leak require best practices, including controller training.
A computational pipeline monitoring (CPM) system uses real-time information from the field – such as pressure, temperature, viscosity, density, flow rate, product sonic velocity and product interface locations – to estimate the hydraulic behavior of the product being transported and create a computerized simulation. With it, controllers can be alerted to actual operating conditions that are not consistent with the calculated conditions and might signal the existence of a pipeline leak. Different CPM methodologies provide different leak detection capabilities, so different methods, or a combination of methods, might be better applied to different operations.
A comprehensive CPM system also supports training best practices that help engineers and controllers develop intimate knowledge of the control system interface, alarming functions and response actions. It is an efficient way to implement refresher training to cover network modifications and expansions and to accurately document training, testing results and qualifications. Computerized simulation has demonstrated to provide more comprehensive and effective training for a specific pipeline than on-the-job training. For this reason, it is the preferred method of the U.S. Department of Transportation’s Pipeline Hazardous Materials Safety Association (DOT-PHMSA) for training controllers to recognize the abnormal conditions that might suggest a leak and to optimize the safety of the pipeline operation.
Schneider Electric’s SimSuite Pipeline solution is based on a real-time transient model that includes leak detection capabilities, crucial to safety and environmental concerns; a simulation trainer application for targeted and effective training of operational staff; and forecasting and planning functionalities that help improve business intelligence. Together, these capabilities help the pipeline operator reduce operations cost as well as comply with regulations.
Experimental evaluation of control performance of MPC as a regulatory controllerISA Interchange
Proportional integral derivative (PID) control is widely practiced as the base layer controller in the industry due to its robustness and design simplicity. However, a supervisory control layer over the base layer, namely a model predictive controller (MPC), is becoming increasingly popular with the advent of computer process control. The use of a supervisory layer has led to different control structures. In this study, we perform an objective investigation of several commonly used control structures such as “Cascaded PI controller,” “DMC cascaded to PI” and “Direct DMC.” Performance of these control structures are compared on a pilot-scale continuous stirred tank heater (CSTH) system. We used dynamic matrix control (DMC) algorithm as a representative of MPC. In the DMC cascaded to PI structure, the flow-loops are regulated by the PI controller. On top of that a DMC manipulates the set-points of the flow-loops to control the temperature and the level of water in the tank. The “Direct DMC” structure, as its name suggests, uses DMC to manipulate the valves directly. Performance of all control structures were evaluated based on the integrated squared error (ISE) values. In this empirical study, the “Direct DMC” structure showed a promise to act as regulatory controller. The selection of control frequency is critical for this structure. The effect of control frequency on controller performance of the “Direct DMC” structure was also studied.
Big Data Analytics for Commercial aviation and AerospaceSeda Eskiler
globalaviationaerospace.com
An opportunity for insight in the changing commercial aerospace business
Vision for New Applications of Analytic Insight in Commercial Aerospace
Benefit of Big Data Analytics for the Airline Operator
Modern, Mobile Experience
Big Data Analytics In Action
Predictive Analytics To Prevent Engine Events
Predictive Analytics Improves Safety and Quality
Predictive Analytics Keeps More Planes in the Air
Predictive maintenance framework for assessing health state of centrifugal pumpsIAESIJAI
Combined with advances in sensing technologies and big data analytics, critical information can be extracted from continuous production processes for predicting the health state of equipment and safeguarding upcoming failures. This research presents a methodology for applying predictive maintenance (PdM) solutions and showcases a PdM application for health state prediction and condition monitoring, increasing the safety and productivity of centrifugal pumps for a sustainable and resilient PdM ecosystem. Measurements depicting the healthy and maintenance-prone stages of two centrifugal pumps were collected on the university campus. The dataset consists of 5,118 records and includes both running and standstill values. Additionally, Spearman statistical analysis was conducted to measure the correlation of collected measurements with the predicted output of machine conditions and select the most appropriate features for model optimization. Several machine learning (ML) algorithms, namely random forest (RF), Naïve Bayes, support vector machines (SVM), and extreme gradient boosting (XGBoost) were analyzed and evaluated during the data mining process. The results indicated the effectiveness and efficiency of XGBoost for the health state prediction of centrifugal pumps. The contribution of this research is to propose an effective framework collectong multistage health data for PdM applications and showcase its effectiveness in a real-world use case.
1. HPIN CONTROL
PIERRE R. LATOUR, GUEST COLUMNIST
clifftent@hotmail.com
APC for min maintenance or max profit?—Part 2
While I have promoted proper control system performance
indicators'""^'' since 1964, like Dr. Y. Zak Friedman,' 1 also am
,iware of some deeper problems with realizing its full commercial
potential.-*'"""*''^'^'^' Friedman once piiblically claimed Clifftcnt"^'^
w,is no panacea,"' without seeing ii in action. Moreover, he
LÍi.irged, without any basis, that C'liHtenr, a mathematical pro-
cedure, could lead to burning down a furnace. My experience is
proper deployment of CÜiflrent to operate the HPI would mitigate
Micb events''' because ii mandates caretul modeling ofthe financial
consequences of exceeding properly set limits like burning down
liirnace.s, so operators would be using better information and
models to make appropriate .sctpoint decisions for risky tradeoffs.
I >etermining .sucb effects rather than ignoring them is the main
idea of science, mathematics and engineering. I refrain from blam-
ing MVC^, I'll), IP. SQC. feed-forward, inferentials, six-sigma,
.statistics, talculu.s or tiialhematics for burning down furnact'.s.
I have published practical examples ol using Clifftent to calcu-
late APC benefits many times for real applications'"'''''*'"^^'^*' since
[be tow-MiIlur fuel oil example^'* in HP, December 1996. I bavc
.1 library ot commercial oil refmery, olefln, aromatics, polymer,
synluels, chemicals and gas processing applications now. Every
single one is a success because it determines a profit improvement
(.sometimes small) and bow to achieve it. I recently sbowed^^ how to
(¡uaiitify the value of Friedmans alkylation control approach."^''
As for how Clifftent works, 1 have published that
too^"''''^''^*^'*'^^''^* and will repeat it here. Every controlled vari-
-ible iCV) has a risky profit tradeoff. Calculus and statistics teach
integration ofthe product of a data frequency di.stribution with
its associated profit function gives its average (expected value)
profit. Repeated integrations witb incremented data means give
ihe average profit profile vs. data mean; a smooth hill (for any
sianilard deviation >Ü). It's easy to locate the hilltop max profit
.md corresponding optimum data mean. Thats how all HI'l
sctpoints, limits, targets and specs are set now, although witb less
rigor. That's also bow C.VIKPi profit meters can be built. ' 1 he
HPI received the panacea in 1996."'
A Canadian University ChE Department included deter-
mining dynamic system financial performance, ClifTtent, in
it.s process control course in September 2009. Hopefully this
rekindled academic interest in process control and research on
building Clifftent models for integrated alarm management,
process maintenance and safety. Once a set of candidate CKs
;ind related manipulated variables (Af Ks) are established and
I be pracess operation and économies are known, good process
control engineering practicc^'^'''''^'^'^'"^^'^'* calls for determining
tbe economic sensitivity of eacb CV, their Clifftents and a vari-
,nice reduction claim for each control system design, providing
ihf appropriate sctpoint determination method and benefit for
ihat variance reduction. 7 hen the process control engineer is in a
•>olid position to design, m;untain and improve the Instruments
and control system, the process operator is in a sound posiiuin to
use it and operating company management understands it.s role
and value. They finally have a chemical engineering method for
operation that is pragmatic, prudent and profitable.
ln hct, 1 claim had one or iwo computer-integrated manufaciur-
ing (CIM) solution suppliers adopted pcrtormancc-based licensing
when it was commercialized in the mid-199üs,^"'^'^'*'^^ the land-
scape of the APC, and (^IM business in the HPI today would he
healthier, more significant and more profitable lor those suppliers
and their operating company partners. And much misguided effort
and cxpcase would have heen avoided. fViedman wouid certainly not
he proposing in a 2009 HP editorial forgoing .ÍO'^íi of APC'.perfor-
mance, worth >$ü.3()/bbl crtidere-fined,to reduce AI'C. maintetiatice
cxjsts. It would not he pragmatic, prudent or profitable.
Dr Y. Zak Friedman's HPln Control editorials continue to
serve the HPI. 1 hope I can add to them on occasion. W*
LITERATURE CITED
'" l^tour. V. R.. "Why Invesi in PRt}CRSS CONIROL?". CONTROL Vol.
XV. xS. May 2002. pp. 41-46.
''' l^aiiiur. P R., "Why tune control loops? Wby mdte control loops?", ediio-
rial guest columnist, Hydrocarbon Processing, V8I. n9. September 2(102. pp.
'" McMahon.T K., (& P. R. Latour). "CLIFFTENT For Process Optimization."
CONTROL VI7. nl2. December 2004, p. 66.
'' l^tdur. P R., "Detisions about risk reduction." litter to Editor, Hydrocarhon
Processing, V80. n6, June 2001, p. 3').
^^ Latoiit. I'. R.. "Quantifying financial values", HP In Control Guest Columnist.
Hydrucarhm lhvcrssir,g.V%{i, n7. July 2001. pp. 13-t4.
'' ljtoiir. P. R., "Align alkybiion separaiiiin to economics." HPln Cx)ntrol
liditorial, Hydrociirbiiti Processing, VHS. i i ! , lanuary 2009. p. 98.
^'' Latour. ['. R.. "Process control: t'l.ll'i'TIÍN I siiows its more profitable than
ex|TecEed," Hydnittirhon froressing, V7'i, nl2. December 1')%. pp. 7S-H().
Repiibliühed in Kane. [.es. VA, Aduamrii Process Control ami Informniion
Sysifim for the l^ocess ¡ndustrirs. Gulf Piiblisliing, Co. I')')'>. pp. .ÍI-37.
^'' Ldoiir. P. R.. "CLIFHIHNT: Determining I-ull Financial Benefit from
Improved Dynamic Performance." PajierCOI, ihirJ Intcrnarioual Conference
on l-dundaiions of C)mpurer-Aided Process Operations. Snowbird, Utah.
)uly "i—IO, l')')8. Proceedings published in AlChK Symposium Series No.
320. V94. 1998. pp. 297-302.
^'' Baker. J. A. et ul. "The Rc]X)ri ofllie BP U.S. Refineries Independent Safety
Review Panel." January 2007.
^' Friedman, Y. Z., (& G. D. Martin. P R. Latour). "APC Survey," Exchange of
Ixiiters to the Fldiior, Hydrocarbon Prficessirig., V85. nlO. OttoUr 2006. pp.
45-46 and V85, n i l . November 2006. pp. 45-52.
'" l^tour. lî R.. "Align Olefin Operations to Lxonomics - Clifiicnt optimizes
scipoints." pn-senteil at 2007 Spring AlChF. Meetinj^ Ethylene Producers
Conference, Houston. Tex;«, April 2^. 2007. Published in t^onfcrcnce
Proceedir^s CD.
''' Fricdnian. Y. Z., "Alkylatioti prcKluci separation control," HPIn Conttol
editorial. Hydrocarbon Processing, V87, n9. September 2008, p. 178.
The author, president of CLIFFTENT Inc., is an independent conr.ulting chemic.il
engineei specializing in identifying, capturing and sustaining measurable financial
value from HPI dynamic process control, IT and CIM solutions (CLIFFTENT) using
performance-based shared risk-shared reward (SR2) technology licensing
HYDROCARBON PROCESSING NOVEMBER 2009 13