Review
a r t i c l e i n f o a b s t r a c t
Article history: Control loop performance monitoring (CPM) in industrial production processes is an established area
Received 6 March 2015 of research for which many methods to detect malfunctioning loops have been developed. However,
Received in revised form 6 November 2015 it is unclear which methods are successful in an industrial environment. Often, there are additional
Accepted 7 November 2015
aspects such as organizational issues, data availability and access that can compromise the use of CPM.
Available online 24 December 2015
In this paper, we are reporting on the results of a survey amongst CPM users. The survey takes stock
of existing methods and their use in industry as well as which faults are most frequent and can be
Keywords:
detected. Organizational as well as implementation issues are investigated and discussed. This paper
Control loop performance
Industrial process
aims to identify open research topics and the direction of development of CPM in industrial production
Monitoring processes.
Fault detection 2015 Elsevier Ltd. All rights reserved.
Process control
Survey
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2. Survey description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
3. CPM pervasiveness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
4. CPM methods, time trends and faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4.1. Mathematical methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4.2. Nature of time trends . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
4.3. Prevalence of faults . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5. CPM framework, workow and implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5.1. CPM workow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
5.2. CPM implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
6. Open research topics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
7. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1. Introduction
http://dx.doi.org/10.1016/j.jprocont.2015.11.002
0959-1524/ 2015 Elsevier Ltd. All rights reserved.
2 M. Bauer et al. / Journal of Process Control 38 (2016) 110
with several hundreds if not thousands of PID control loops digitally These questions were gathered into an online survey that was
implemented. distributed to control engineers at production plants in various
With the advance of automatic control came the need to assess industries and around the world. Since a survey requires a compar-
the performance of the control loops, according to the business atively large number of participants we have focused the content
stratagem If you cant measure it you cant manage it. Increas- on single control loops and particularly on PID loops.
ing computational power enabled assessments which looked, for The results of this survey amongst CPM practitioners are pre-
example, at the standard deviation of process trends. Standard devi- sented in this article. The methodology of the survey and the
ation or variance is an obvious candidate as a performance index background information of the survey respondents are described
because many engineers selling control solutions use a depiction of in Section 2. Section 3 discusses the awareness of CPM and the
data before control improvement and after where the before scope: how well are control loops performing today? Is there still
case shows a heavily uctuating process variable and the after a a need for CPM? In Section 4, the prevalence of the various meth-
near constant smoothened out time trend. ods as well as problems that can be addressed by CPM are discussed
A key challenge in the assessment of control loops is to dis- while Section 5 investigates procedures, frameworks and workow
tinguish between a controller that was performing poorly (A) that further the success of CPM in industry. Section 6 looks at future
because there was something wrong with it and (B) because of research directions and the satisfaction of CPM users with current
an outside disturbance. This was rst addressed by Harris in 1989 tools and methods.
when he published Assessment of closed loop performance
[2]. The measure, later to be named the Harris index compares
the variance of a process variable to the minimum achievable 2. Survey description
variance, which is caused by outside disturbances. This paper
attracted a signicant amount of interest in the academic world of An online survey was conducted to capture the prevalence of
control engineering and brought the problem formulation closer CPM worldwide. There is an abundance of publications on survey
to research institutions. methodologies for all survey purposes and groups of respondents
A reason for the focused attention on the assessment of control [13]. Many providers offer free platforms to easily format question-
loops was that despite the prevalence of PID controllers ensuing naires and capture results. For this survey, the authors chose Google
studies found that the controllers were not doing as well as every- forms, which is part of Google drive and does not require any soft-
one had assumed [3]. In fact, before a measure was introduced, ware installation. In addition, the results are stored in spreadsheet
controllers were just used. Now these controllers were good or form and reports are generated automatically.
bad, acceptable or poorly performing. And most studies found The design of the questionnaire is the most important aspect
that there more poor performers than expected. of the survey. When putting the questions together many pitfalls
A key requirement for the assessment of control loops is that have to be avoided. For example, the questions have to be phrased
data from routine operation and closed loop control should be objectively and clearly in a coherent order. Non-exhaustive listings
used. The initial assessment of a single control loop has since then must be avoided. To ensure a high response rate, the questions must
expanded to frameworks and procedures, and includes diagnosis, be meaningful and interesting [14].
fault identication, isolation, and root cause and plant wide dis- To ensure the validity of the questions, interviews with industry
turbance analysis. This research area today is often called control experts were conducted and the survey questions were discussed.
loop performance monitoring (CPM). These experts were Florian Wolff at BASF, Germany, Duane Muller
CPM now forms a substantial body of research articles and at AngloAmerican, South Africa.
industry applications. Over the 25 years since the publication of For this type of survey the group of respondents are limited.
the original article, several review articles and tutorials have been The respondents were identied as lead control engineers in pro-
published. The rst reviews were written about a decade later duction companies from various industries. In order to address the
[4,5] focusing the multivariate extensions, feedforward control and target audience, several approaches were started. First, all personal
industrial aspects. Hgglund [6] as well as Jelali [7] made signi- contacts of the authors were approached. Second, what is referred
cant contributions by explaining working indices in plain terms to as snowball sampling was pursued, that is, known responding
and providing codes for industrial implementation. More recently, control experts were asked for referrals among their colleagues.
the Springer series Advances in Industrial Control includes four Thirdly, published authors in the area of CPM that now work in
monographs on control performance assessment, including valve industry were approached. The contact data was retrieved from
stiction detection [811]. Furthermore, the recent textbook [12] the journal article or conference proceedings. In addition, the sur-
explains established methods in detail and gives frameworks, vey was distributed in a Honeywell user group meeting and the
implementation guidelines, applications and tools. participants lled the results in during the meeting. All responses,
CPM was developed in close cooperation with industry. In [7], electronic or on paper, were anonymous.
a list of research articles and their industrial applications in the In total, 69 control engineering experts in production companies
chemical, petrochemical, pulp and paper and other industries in answered the survey. Fig. 1 lists the respondents by continent and
provided. The same article also lists the commercial packets that by industry. Roughly half the respondents were from Europe (33
comprise control loop monitoring tools, either as stand-alone solu- out of 69) because the authors contact were used to send out the
tions or built into automation and control software. questionnaire. The majority of respondents (64%) work in chem-
In this article, we are scrutinizing what has been achieved in icals or oil & gas. This may be partly explained because chemical
industry in the last three decades. The focus is on the production and petrochemical companies are traditional strongholds of CPM.
companies in the chemical, oil and gas, pulp and paper and other It should be noted that the answers do not always add up 69 because
industries that use CPM to manage and assess the control of their not all respondents answered all questions.
processes. The questions addressed here are: The respondents have various levels of experience in control
engineering, as indicated in Fig. 2. The total is about 1000 years of
What works in industry? Which methods are most useful? Which control engineering experience on which the survey is based on.
frameworks and processes are successful? Respondents were also asked how many loops are allocated for
Is CPM a standard or only used by leading production companies? each control expert and the results are displayed in Fig. 3. On aver-
What are the key challenges? What are the open research topics? age, a control engineer is responsible for about 450 loops. However,
M. Bauer et al. / Journal of Process Control 38 (2016) 110 3
3. CPM pervasiveness
4. CPM methods, time trends and faults Some conclusions that can be drawn from Fig. 5 are as follows:
The organization and categorizing of CPM methods depends sig- The simpler the method the higher the number of respondents
nicantly on the viewpoint of the CPM practitioner and expert.
who nd them useful. Operating mode statistics are easy to com-
For a comprehensive list of CPM methods see for example [12].
pute though not always easily accessible since they are not
Generally, CPM methods can be grouped according to three
always stored in the data historian and equally easy to interpret.
criteria:
Multivariate or advanced statistics are more difcult to compute
and to interpret.
1. Mathematical method or algorithm. All methods are very useful to some CPM users, each user proba-
2. Nature of time trend analyzed. bly has his or her favorite method to use.
3. Type of fault or malfunctioning addressed. Algorithmic indices have still untapped potential and could be
promoted further. Most respondents who know about these
All methods can be grouped according to each category, how- indices nd them very useful, however, they are unknown to
ever, some are strongly associated with a particular method (e.g. about a quarter of the respondents.
PCA), time trend (e.g. oscillation) or fault (e.g. sluggish tuning such The impact which minimum variance indices, in particular the
as the Idle index [19]). It is often convenient to group the indices Harris index, had in the academic control community is not
according to their main functionality, which can be either method, matched in the industrial community. The reason may be that
time trend or fault. For example, some indices may be referred the Harris index works well to explain the idea of CPM and on
to as oscillation index while others may be labeled stiction certain data trends but is meaningless if the data contains non
index. stationary signals.
The following sections present the survey results on the useful-
ness of methods as well as the occurrence of typical time trends
4.2. Nature of time trends
and faults.
Ideally, the time trend in regulatory control is stable or constant
4.1. Mathematical methods and lies within a certain operating region. In reality, the trends are
never smooth but show certain degrees of variability. There are sev-
All control experts were asked to rank certain methods accord- eral features or disturbances in the time trends that an experienced
ing to their usefulness. The CPM methods were grouped into seven control engineer will recognize by visual inspection. Generally, dis-
categories to keep the time to answer the survey within a reason- turbances can travel through a process through physical or data
able limit. The authors felt that it would neither be feasible nor connections. Thus, a disturbance can be picked up in a number of
benecial to assess each method individually. The results are pre- control loops.
sented in Fig. 5 and the provided more detailed description of the Most CPM techniques assess the controller performance by ana-
categories are given in Table 1. lyzing the time trend. Typical time trends are shown in Fig. 6 and the
Table 1
Method categories and methods that are grouped under this category.
Category Methods
Operating mode statistics Time in manual versus time in auto, number of operator interventions,
number of tuning changes, loops in saturation
Basic statistics Minimum and maximum threshold crossings, mean, standard deviation
Algorithmic indices Idle index, integrated absolute error (IAE), scatter plots of manipulated
variable versus controller output (MV/OP plots)
Model-based techniques AR(X) models, articial intelligence (AI), stiction models, Hammerstein models
Multivariate statistics Correlation, principal component analysis (PCA), partial least squares (PLS)
Minimum variance techniques Harris index and related indices
Advanced statistics Gaussianity testing, probability density function, surrogate data analysis
M. Bauer et al. / Journal of Process Control 38 (2016) 110 5
Fig. 6. Time trends of process variable and controller output for common disturbances.
Fig. 7. Sample time trends of faults and their rating, how frequently they occur.
prevalence of these time trends according to the survey are given disturbance affects the process variable regularly and repetitively
in Fig. 7. In the following, each time trend is briey discussed. but contains higher frequency components and can therefore be
distinguished from a sinusoidal oscillation.
Saturation: Saturation manifests itself at the controller output Quantisation of process variables: All signals are digital in
which is intermittently at its maximum or minimum (usually 0% amplitude. The resolution of the signal is determined by the sen-
or 100%). It will deviate from its extreme for a short while only to sor. The sensor may have a coarse resolution so that the time
return back to it. Saturation can result from a poorly dimensioned trend shows step-like features. This often applies to temperature
actuators, poor controller tuning or may be caused by integral sensors which could have, for example, a resolution of 0.5 while
wind-up of PID controllers after a large setpoint change occurred the temperature only uctuates by 1 .
[12].
Sinusoidal oscillations: Oscillations are introduced in process
variables by a number of means such as poor tuning settings, Generally, the results shown in Fig. 7 are not new control engi-
actuator faults or external disturbances. Generally, as oscillations neers in the 1990s looked for similar features in the data then as
become sinusoidal as they travel through a process via physical today: Which controller outputs are at their limits, which loops are
connections since the process acts as a low pass lter. oscillating and which are in manual control.
Manual control: If the controller is not acting satisfactory the Somewhat surprising in Fig. 7 is that saturation is the most fre-
operator or control engineer often switches it off. This is seen in quent problem. Detection of saturation is straight forward. More
the control system as a switch from automatic to manual mode. intricate is the identication of the cause of saturation since this
Sluggish behavior: An unwanted behavior seen in the time trend is often an indication of actuator dimensioning but can also occur
is labeled as sluggish or slow. This means that the process variable when tuning settings are incorrect or no anti-windup algorithm is
only slowly follows the setpoint. There is no overshoot after a implemented.
setpoint change. The reason for sluggish behavior lies usually in A similar issue is manual control which should always be inves-
the tuning settings. tigated as it shows that there is an underlying problem with the
Nonlinear oscillations: Some control loops contain a nonlinear loop. It means that the operator, for various reasons, does not trust
element such as a faulty valve and thus cause nonlinearity in the the controller to do its job automatically. The root causes range
time trend of process variable and controller output. A nonlinear from poor tuning over interacting loops to training needs.
6 M. Bauer et al. / Journal of Process Control 38 (2016) 110
Fig. 8. Control loop elements and associated faults according to the survey results,
see Fig. 9. The height of the column reects the prevalence of the fault. 5. CPM framework, workow and implementation
Fig. 14. Satisfaction with CPM tools: How satised are you with the current tools?
0 not satised, 9 completely satised.
Fig. 13. (a) CPM as product versus CPM as service, (b) CPM as in-house development 6. Open research topics
versus commercial solution, (c) CPM used on PCS, as stand-alone solution.
After more than 25 years of active research in control perfor-
mance monitoring, it is of great interest to ask whether there are
any relevant research questions left open. Here, we address this
that CPM as a service has to be more automated and standardized question from the industrial application perspective. Fig. 14 shows
because it has to be interpreted by non-expert users. The report this level of satisfaction which appears message. Since most of the
should highlight critical loops and underlying causes and suggest respondents use CPM the message from this result is mainly: People
corrective actions. This means that either the methods are clear, the who use it see an advantage of the tools.
results have to be denite, and that there is little room for inter- Most respondents see value in applying CPM to their industrial
pretation and discussion. Fig. 13(a) shows the result of the survey: processes and are reasonably satised with the current tools. At
the split between product and service is nearly equal. This means the same time, there appears to be improvement potential because
that there is a need for both automated methods and methods that some respondents are not satised at all. This may be because of the
leave room for discussion. specic unsatisfactory tool these respondents are using but more
In the survey, the respondents indicated whether they use a likely because there is a research opportunity.
commercial packet, their own development or a combination of the In discussions with end users BASF and AnlgoAmerican areas of
two (Fig. 13 (b)). It is noteworthy that respondents using their own open research were identied and clustered. The resulting list of
development predominantly regard the solution as a service, i.e. these topics together with the survey ndings is shown in Fig. 15.
Fig. 15. List of topics for further research and development (split between what vendors and users think).
M. Bauer et al. / Journal of Process Control 38 (2016) 110 9
The clear main result is that the respondents ask for better guidance to a large amount of control loop data. Hence long computation
on corrective actions to be taken. This does not come as a surprise time still can be a drawback.2
because proper corrective action is a pre-requisite for economic
payback of the investment in CPM.
The second most important research need relates to the men- In order to address the relevant research questions in the future,
tioned prioritization of control loops. A corrective action is only it is recommended that academic institutions collaborate with the
relevant if the asset relating to it has a certain importance for the process industry, that is, with end users. This will guarantee that
overall performance or criticality of the whole plant. the relevant problems are solved, the results are feasible to be engi-
The topics of training, guidance on parameter tuning, automated neered, interpreted and used. Also, new CPM algorithms should be
data selection and graphical display relates to the implementation tested on real-life data rather than only in simulation because often
of CPM methods in industrial solutions. This result of the survey root-causes interact which can hardly be simulated realistically.
ts the current tendency to not program own CPM tools but to try
to buy commercial products or services. Commercial tools often
offer more advanced functionality for long-term monitoring, data
7. Conclusions
handling, graphical display and training material.
An interesting point is the second last statement that CPM meth-
This paper presented results from a survey on industrial appli-
ods relates to condition monitoring of rotating equipment. As a
cation of CPM in the process industries. It has become evident
matter of fact, there is a certain disconnect between CPM on the one
that there is considerable knowledge about control performance
side and traditional condition monitoring Condition monitoring
present in many process industries. The topic of monitoring and
refers to fault detection for mainly mechanical process equipment.
assessing control performance has been important in the past and
If applied to the same process at the same time, both are often
will remain to be important also in the future.
dealt with in different software tools that are managed by different
We conclude that after 25 years of intensive research there
people. Nowadays, there is much more information available for
are still relevant research questions to be solved in the area of
analysis, such as production data, condition monitoring data, oper-
CPM. We will here disregard all indirect aspects such as software
ating conditions, plant parameters, etc. Such data will increasingly
implementation, usability, presentation of results, data access, and
be accessible also for CPM1 . Hence, methods should be developed
integration into automation systems. They may lead to research
that make use of that data fusion. The benet of integrating differ-
effort, however this is hardly specic for CPM application.
ent data sources should be an active eld of research. This is often
Staff responsible for control performance in process plants are
the result of recent trends referred to as data analytics and big
increasingly given more and more complex tasks so that extensive
data.
tool support for CPM has become a standard. The application of
Such gaps can be bridged by research, which brings both
such tools, however is still quite heterogeneous. One reason is the
technologies closer together, or by organizational structures. The
different experiences companies have gathered over the years. A
latter can be realized by specic staff that share the responsibil-
second reason is the different maturity of commercial tools, they
ity of both aspects in the same plant. Chemical company BASF
differ in philosophy (product versus expert tool versus service) and
in Germany which were interviewed initially for the purpose of
in price. There is not yet any standard which methods to use and
the survey, has deployed so-call Asset Managers for each of their
how to present the results to the users, though attempts have been
plants. These asset managers connect plant data from the automa-
made [21].
tion system, the maintenance management system and other ERP
CPM tools are mainly needed to prioritize maintenance actions
data.
related to deteriorated control performance. Guidance for correc-
Other areas such as automatic, data-driven detection of stiction
tive action is a vital requirement that is not yet solved sufciently
behavior in process control valves has been extensively researched.
well. For this, very often simple methods seem to be sufcient. Root-
There are more than 15 different methods published already. Also,
causes diagnosing is still an open problem, but, at the present state,
efforts have been made to integrate them into an index-fusion in
one can detect loop malfunctions with a good level of condence
order to increase the reliability of detection [11]. While it may be
in many situations.
worthwhile to further improve the reliability of existing methods,
One key nding is that CPM is more that the application of
it does not appear to be fruitful to develop yet another detection
numerical algorithms to plant data. Companies that successfully
algorithm that will outperform the existing ones.
apply CPM have undertaken thorough efforts to integrate CPM
A conclusion of the answers about further research direction
into their daily plant operation and asset management. This also
from an industrial perspective is that simple methods work best
includes different work procedures.
in most cases. Simplicity is here referring to three aspects:
A nal question of interest is if the control performance in the
process industry has been improved during the last 25 years (or if it
would have deteriorated more without CPM). This question is dif-
1. Simplicity of parameterization: Methods that need to individ-
cult to answer. The companies that see value in CPM obviously seem
ually be tuned, parameterized, trained etc. will most likely not
to apply it successfully. In those companies, it can be expected that
survive in the industrial CPM practice.
control is performing on a better level than earlier. There are, how-
2. Simplicity in interpretation of the results: Methods that require
ever, still many companies neglecting the advances in CPM such
signicant training, experience of interpretation, do not resolve
that the industry average of non-optimal control performance can
ambiguities and interpretation guidelines will most likely not
be expected to be much lower than desirable. There is hence still
survive in industrial CPM practice.
signicant potential to improve control in the process industry.
3. Simplicity in computational complexity. Even though computa-
tional power still increases, CPM is typically applied often and
2
This comment will in the future possibly be less important when CPM calcula-
tions largely are done in the cloud. Some companies already offer the experiment
1
This unlimited access to data is called Industrie 4.0 in Europe and Industrial with CPM-as-a-Service where the algorithms are performed in a powerful (local or
Internet in North-America. private) cloud infrastructure.
10 M. Bauer et al. / Journal of Process Control 38 (2016) 110
Acknowledgements [8] B. Huang, S.L. Shah, Performance Assessment of Control Loops, Springer, Berlin,
1999.
[9] A.W. Ordys, D. Uduehi, M.A. Johnson, Process Control Performance Assessment:
The authors would like to thank Florian Wolff at BASF and Duane From Theory to Implementation, Springer, Berlin, 2007.
Muller at Anglo American for their insight as well as Richard Sal- [10] M.A.A.S. Choudhury, S.L. Shah, N.F. Thornill, Diagnosis of Process Nonlinearities
liss at Honeywell for distributing the questionnaire at a user group and Valve Stiction Data-Driven Approaches, Springer, Berlin, 2008.
[11] M. Jelali, B. Huang, Detection and Diagnosis of Stiction in Control Loops: State
meeting. This paper is a result of many discussions with senior of the Art and Advanced Methods, Springer, Berlin, 2010.
control practitioners and academics. We would like to thank Russ [12] M. Jelali, Control Performance Management in Industrial Automation, Springer,
Rhinehart, Kevin Brooks, Tom Edgar, Biao Huang, Mike Grimble and Berlin, 2013.
[13] P.V. Marsden, J.D. Wright, Handbook of Survey Research, 2nd ed., Emerald
Gerrit van der Molen for their extremely valuable contributions.
Publishing Group, Bingley, UK, 2010.
[14] D.A. Dillman, Mail and Internet Surveys: The Tailored Design Method, 2nd ed.,
Wiley, New York, 1999.
References [15] M. Bauer, I.K. Craig, Economic assessment of advanced process control a
survey and framework, J. Process Control 18 (2008) 218.
[1] S. Bennett, A brief history of automatic control, IEEE Control Syst. Mag. 16 (3) [16] M. Bauer, K.S. Brooks, C. Sandrock, Industry expectations and academic practice
(1996) 1725. in control engineering education a South African survey, in: IFAC World
[2] T.J. Harris, Assessment of control loop performance, Can. J. Chem. Eng. 67 (1989) Congress, Cape Town, 24-20 August 2014, 2014.
856861. [17] D. Wei, I.K. Craig, Grinding mill circuits a survey of control and economic
[3] D. Ender, Process control performance: not as good as you think, Control Eng. concerns, Int. J. Miner. Process. 90 (2009) 5666.
40 (1993) 180190. [18] L. Desborough, R. Miller, Increasing customer value of industrial control perfor-
[4] S.J. Qin, Control performance monitoring a review and assessment, Comput. mance monitoring Honeywells experience, AlChE Symp. Ser. 98 (326) (2002)
Chem. Eng. 23 (1998) 173186. 153186.
[5] T.J. Harris, C.T. Seppala, L.D. Desborough, A review of performance monitoring [19] T. Hgglund, Automatic detection of sluggish control loops the single-loop
and assessment techniques for univariate and multivariate control systems, J. case, Control Eng. Pract. 7 (12) (1999) 15051511.
Process Control 9 (1999) 117. [20] F. Wolff, S. Kramer, Regelguetemanagement. Atp, 3/2014 (2014) 5664.
[6] T. Hagglund, Industrial implementation of on-line performance monitoring [21] NAMUR User Association of Automation Technology in Process Industries, Con-
tools, Control Eng. Pract. 13 (2005) 13831390. troller Performance Management: Monitoring and Optimisation of Regulatory
[7] M. Jelali, An overview of control performance assessment technology and Control in Production Plants. NAMUR Recommendation NE 152, 2014 www.
industrial applications, Control Eng. Pract. 14 (2006) 441466. namur.de.