In his articles in this publication, and the case studies detailed on his website, Michael Brown has for years been waving flags about the necessity of control loop optimisation in the process industry...
But his voice seems to have been 'lost in the wilderness'.
The interesting thing is that this is not just a South African phenomenon and a quick search of the Internet using the words 'control loop problems' yielded a staggering 576 000 responses.
While Michael estimates that in the average plant worldwide, over 85% of loops are operating inefficiently in automatic, this was confirmed by two articles found on the Internet. One, published by Control magazine, stated: "It has been estimated that 80% of process control loops are causing more variability running in automatic mode rather than in manual". The other article published in Control Engineering stated that three quarters of all process control loops are operating below peak efficiency. This latter article also indicated that approximately half of all loop problems relate to control valve problems, and the others are largely due to poorly tuned controllers. Both these articles and those of Michael have referred to the cost of these inefficiencies that result in effectively giving away product and often resulting in out of spec product.
In case one should think that Michael is preaching a wide off the mark message, there are many other control loop gurus in other countries who hold similar views. One of these is Michel Ruel of Canada who has also emphasised in many articles that loop optimisation before tuning is essential. Michel is the president of a company called Top Control which has as one of its objectives "...is a company specialising in process control optimisation with the objective of allowing a process company to get the maximum performance while prolonging service life". The company's website www.topcontrol.com contains many case studies and testimonials from leading companies on the benefits of using their resources or the value of their training courses.
The striking thing is that the many articles and case studies of Michael Brown have shown just how poorly many control loops are operating and how often simple changes such as the replacement of a valve, proper calibration and scaling of flowmeters and other measurement devices, correct tuning, etc, can result in dramatic changes to the stability of the loops and provide rapid response to controller changes.
What is regulatory control?
David Ender, a world-leading expert in process plant control defined the process of regulatory control as 'minimising variance' through every stage of production. It is there to keep processes at setpoints that themselves are either set manually, or by an advanced control system. These setpoints would be determined in terms of a carefully planned production strategy and if the processes cannot be kept to the setpoints, that strategy obviously cannot be properly implemented. As the regulatory control system for the average process plant costs tens of millions of rand, one would expect that the control system should really work well. If the loops are not optimised then it will underperform and the investment made by the company in the first place is wasted, as they may as well have remained operational through manual control.
Extra profits obtained with optimised control loops
Cost figures relating to poor control loop optimisation are difficult to come by, but Alpret ran a series of articles (on selecting the right control valve) in SA Instrumentation and Control during 2003 and in its concluding piece an example was given (assumed to be local) where the correct choice of valve resulted in a 1,4% improvement in process variability. This does not appear to be much but it results in a raw material saving of 45 780 litres per day and at R0,5 per litre this contributes R22 892 per day directly to profits or some R8,3 million a year. This saving was achieved on a single loop, so one can imagine what savings could arise if all control loops in a plant were optimised, some tens of millions of rand. While Alpret was making the point that the correct valves should be installed at the time of construction of the plant, a more expensive solution often paying for itself within days or weeks, the same comments are applicable in the case of non-optimisation of the control loops.
Michael Brown also commented during the same year that a study completed at that time in the USA and based on 300 control loops showed that the average plant, if only 30% of the loops were operating inefficiently in automatic, would cost the company in excess of $350 000 per year. Taking a more realistic figure that 75% of control loops are operating effectively would raise that figure to some $875 000. Chrome International South Africa (CISA) is one of the few companies who have published savings. They made use of Process+ from CSense Systems and after optimisation of just 10 loops, control engineer Erich Holmes estimated that annual savings on these alone would amount to some R900 000. Another local example is a South African gold mining company that estimates that it is saving some R20 million annually on reagents through having optimised a single control loop properly.
On the Internet one can find case studies and savings made, these usually being published by the companies that provide optimisation and tuning software. Some examples:
* Smurfit-Stone Paper Mill, savings of $280 000 per year.
* Petro-Canada, offshore oil processing, >$500 000 per year.
* Petro-Canada oil refinery, $330 000 per year.
* Alcan Aluminium, bauxite processing, $200 000 per year.
There are a lot more examples but an interesting case was Blitz Weinhard Brewing Company in Ohio where it was required to optimise a distillation system with a lot of interaction between loops. A four-day software-driven optimisation exercise increased the capacity through the system by 30% with the company claiming that the cost of the exercise was returned within the first couple of weeks. Another US company estimates that 40 to 70% of cost savings from advanced process control (APC) projects are lost if the performance of regulatory control loops is not maintained. The same company believes that the frequency and severity of critical conditions are also reduced and that in the process industry critical conditions cost the US economy about $20 billion annually!
It is clear from the many case studies published by Michael Brown (on his website and in SA Instrumentation and Control) that many process managers and control engineers believe that their control loops are operating optimally and that they do not require assistance. Often the poor control performance is only shown up during factory training where attendees at the course have the opportunity to play with some simple live loops. In several cases the operator has tuned the loop in manual and has only put the controller back on automatic when everything has reached a steady state. If nothing further disturbs the loop it stays where it is, namely looking good. The problem is that they are often tuned to an unstable state and go unstable with changes in load or the setpoint.
The mass of data compiled both locally and internationally suggests that massive savings can follow optimisation of control loops in the process industry. The problem is how to convince management of this fact: A small investment in optimisation indubitably will result in massive savings and larger profits. Another benefit of optimised loops is that it frees up the time of the control staff that do not need to try and adjust manually every time something changes in the process. It is interesting to note that most people who attended these courses were very enthusiastic about the possibilities of improving plant performance, but very little has changed in their respective companies, mainly as a result of a lack of commitment by more senior personnel.
An interesting analogy
While my mandate is not to cover successful case stories, an interesting analogy was provided by the author of an article on Marathon Ashland Petroleum in the USA, a company that has a continuous improvement programme which includes optimisation of its thousands of control loops in the various refineries. He compared process optimisation to F1 motor sport racing and pit stop strategies, particularly the last one. Making the wrong choices and extra time spent in the pit cannot be made up on the track. The decisions regarding the pit stop are based on the on-board sensors and instrumentation, the accurate analysis of the data, and precise communication between driver, crew chief and the pit crew. He pointed out that while oil refining is not as exciting as motor racing it is every bit as competitive and requires teamwork, accurate instrumentation and controls that work effectively.
The lack of enthusiasm for loop optimisation
Why is the process industry in general not interested in proper optimisation of their control loops? There is no simple answer but one factor is that because the plant is generally operating in a steady state condition with manually set loops, the plant manager thinks that everything must be all right. After all they may have been operating this way for the last 10 or 20 years so why create additional work and cost. This is however an inadequate answer as the many case studies provided on the Internet and published in both local and international I&C publications show that plants through loop optimisation can not only increase throughput, but also save on material input resulting in significant extra profit. Surely the technical people read these publications and why do they not copy relevant articles (on their specific industrial sector for example) to managers further up the chain - in my experience this is a first step to motivating changes within an organisation.
An answer to this is that the industry is typically very parochial and financial people want to know up front what the cost benefits will be for them. They seem to be incapable of extrapolating from the success stories of their competitors. Their attitude is also one that says "We have already invested millions in the control system as it is; the plant is producing, so why spend more money when their own people are capable of tuning (not optimising) the loops". Unfortunately there is a shortage of good practical people everywhere who can really practically optimise loops, and poor tuning that produces product in the steady state is regarded as being adequate. It is only in the situation where there is drastic failure of a critical loop that an expert is called in to do the fire fighting. Unfortunately this exercise is usually restricted to the problem loop and even after optimising this, the plant is still operating with poor efficiency. Tuning, which occurs after optimisation, is every bit as critical but is most often done through trial and error which still often results in poor loop performance. Here plants need to invest in scientifically based software tuning packages.
A further compounding problem with control loop optimisation is that training (Technikon or University) is highly theoretical. When graduate engineers join the process industry they effectively have to learn to fly by the seat of their pants and picking up of bad habits is common. Many of these self-trained engineers may also be hesitant to inform management that they need to hire professional consultants to deal with the problems that they were hired initially to sort out. The shortage of good technical skills is compounded, as even if the production manager understands the need for optimisation he has not got the right in-house skills.
For the really committed company (and there are several major ones in South Africa) where the need for loop optimisation is recognised, what is needed is ongoing teamwork with total commitment by management. Remember loop optimisation has a finite lifetime as valves and other control devices change characteristics with wear. Successful companies who are really committed to the total optimisation of their process also will possess a dedicated optimisation section with really skilled personnel. If the financial department is still sceptical the process engineer should encourage a limited audit of a few control loops to confirm whether these need attention or not. Such limited audits are not expensive but are very useful for motivational purposes.
While this management commitment is required there is a limited understanding at production management levels of control loops and PID controls. What they should be looking at continuously is the number of loops that are running in manual, as these are obviously problematic. Many users of CSense for example use process reports as part of the daily production meetings where problem loops can be identified and the necessary action taken with quick feedback.
The future
Is there an answer to the problem? The increasing introduction of advanced control systems (ACS) may provide a mechanism to get companies to wake up. It is well established in the literature that the introduction of ACS is a total waste of time and money, unless the underlying infrastructure, namely the regulatory control loops, have been properly optimised. Without optimisation the approach has been likened to building a skyscraper on quicksand. The smarter engineers have realised this but management needs to be aware that optimisation is ongoing and loop optimisation is regarded as having a half-life of six months. Some of the more progressive local companies that have identified loop optimisation as a priority, re-optimise all their loops on a three-month basis.
This is also a major opportunity for the many international companies, like Pretoria-based CSense Systems (the CSense being literally See Sense), who offer continuous loop monitoring systems. This software will flag poorly performing loops when a problem arises. These programs may not identify the exact problem and this will have to be investigated by a qualified control engineer, not just a technician, who could revert back to inadequate loop tuning. Another often overlooked benefit of on-line monitoring of control loop performance is that it can help a company to carry out predictive maintenance, allowing prevention of potential failures before they occur and thus eliminating the need for expensive unscheduled plant shutdowns.
Note that tuning packages and on-line continuous monitoring systems are not the be all and end all solution. They really only come into their own after the control loops have been fully optimised, particularly in terms of measuring systems and making sure that the valves being used are really capable of doing the job. Loops are also sometimes found that measure more than one parameter and these may be in conflict to proper optimisation and final tuning.
A final point for process engineers to note is that CSense offer their Process+software on an evaluation basis, the only initial charge for the company being the cost of the typical one-week installation. This then allows those responsible for the control loops to really see if it works for them and if it does this provides strong motivation for the actual purchase of the package. CSense have found this an extremely successful marketing tool with most companies seeing the benefits of the software. Although this is locally developed software it is sold throughout the world with one of the major partners being Outokumpu.
Conclusion
For a country with an extensive processing industry, including petrochemical, mining and paper, I cannot understand why the culture of continuous improvement is not widespread. If I were a stakeholder in a company and knew that a simple optimisation process with an ROI in weeks or months could significantly improve the bottom line I would make sure that something was done.
If management cannot take the initiative themselves, then maybe it should become a concern of the company's board and other stakeholders. This is an increasingly competitive environment and those companies who are not prepared to take the chance of improving efficiency, productivity and profits, have effectively made a decision that they want to go out of business. In terms of being internationally competitive, ISO 9000 certification is a requirement, and I fail to see how companies who operate so many loops in manual mode, with others being unstable, can ever be in compliance with the requirements of this standard, which effectively demands total process stability.
SA Instrumentation and Control would like to encourage the establishment of a forum on control loop optimisation, and comments on this article and those published by Michael Brown would be most welcome. Opinion pieces should be emailed to the editor, [email protected] - and any questions related to loop optimisation strategies should be e-mailed to Michael Brown: [email protected]
The author would like to thank Michael Brown (Michael Brown Control Engineering), Dr Tjaart van der Walt (CSense Systems) and Michel Ruel (Top Control) for their contribution to a better understanding of the control loop problem. It should, however, be stressed that the views expressed in this article are those of the author.
Dr Maurice McDowell has many years' experience as a technical journalist, editor, business manager and research scientist. His third party analyses of world-class companies and processes, as well as his insight into industry and technology trends are well respected.
© Technews Publishing (Pty) Ltd | All Rights Reserved
printer friendly version