March 2006

In this issue

	Classes and seminars
	New "Top Control Advisor"
	Trade Shows
	Trick of the month
	To contact us

Coming classes and seminars

Seminar on modern control methods
March 22-24, Bangkok, Thailand
March 28-30, Malaysia, Malaysia

PID Loop Tuning
April 6, Ottawa, ON
May 8, Milwaukee, WI

Detailed Loop Analysis with ExperTune
May 9-11, Milwaukee, WI

Analytical methods for troubleshooting and Control loop tuning
May 16-18, Dalhousie, NB
May 23-25, Cincinati, OH

See our new calendar for 2006

Top Control Advisor    Unlock The Knowledge From Your Control System Database!

Find which loops have nonsense tuning parameters.
Pinpoint wrong configuration or programming.
Identify faulty or inappropriate algorithms.

Generate immediate savings and profits!
Your DCS', PLC's and controllers are used to control your plant. To perform, they must use the right amount of control action. Each one has to be adjusted to the process and to the loop objective.

Top Control Advisor Service:

1. Download the database from your system in a file and send us this file;
2. Top Control Advisor Service will analyze the data;
3. You will receive a report in which each controller will be described by a color and comments:

   	Green	numbers make sense
   	Yellow	verification suggested; numbers differ from usual
   	Red	numbers do not make sense and should be modified

Top Control Advisor Service will:

• Validate the tuning parameters in your systems
• Verify quickly if the numbers make sense
• Pinpoint and detect:
• Wrong numbers
• Wrong tuning parameters (P-I-D-F and others special parameters)
• Wrong configuration
• Inappropriate scan time
• Inappropriate filtering
• Inappropriate PID function selections
• Faulty algorithms
• Wrong spans, wrong flags, etc.
• Let you put your attention where it is really needed
• Help you determine where optimization is really needed!

Impacts of wrong tuning parameters and inappropriate selections are numerous:

• Process and process control not performing
• Equipment abuse (e.g. valves moving 6 times more than necessary)
• Operator's attention dragged by loops that have to be operated in manual mode
• Out of specification production
• Difficult and long start-ups
• High maintenance costs
• High variability leading to operating far away from the specifications; higher costs
• Too much energy usage when not performing
• High consumption of additives and raw material

How does Top Control Adviser Service work?
Top Control Advisor Service is based on:

• Tuning rules
• Process control theory and principles
• First principles
• Our team of highly experienced engineers
• Our huge database (loops from all types of plants, all over the world)
• Our special expert system developed by our team

Examples of results:

3-LIC-234	OK, all numbers make sense
TP-PIC-53	Too much filtering on PV
8-TIC-842A	Scan time is too small
42-LIC-23	Algorithm should be changed for:"P on PV"
26A-DIC-8	Loop should not use derivative
19-LIC-33	Integral time is too small for that loop
5-FFIC-19	Proportional gain should be reduced

What we need:
From the database downloaded from your system:

• Tag name, controller used, spans, units, etc.
• Tuning parameters, flags, selections, etc.

Any format will do:

• CSV, TXT, XLS, XML, etc.
• dBase, SQL, Access, Oracle, etc.

Our solution:
Send your database and Top Control Advisor Service will run an analysis of your data. Pricing: $19.95/loop; minimum $995. If you add $9.95/loop, the values will be converted to an equivalent in another algorithm or system (DCS-PLC-controller).
Example:

Top Control Advisor Analysis of:	Total cost	Unitary cost
35 loops	$995	$28.43/loop
100 loops	$1,995	$19.95/loop
1,000 loops	$19,995	$19.95/loop

Top Control Advisor Service will pinpoint wrong numbers and will let you focus on the loops that absolutely need corrections; this is a simple way to reduce costs and to do optimization work where you know for sure that there is room for improvement!

You can expect (statistics from Top Control):

	% OK	% Warning	% Wrong
Areas where loops have been optimized regularly	88	10	2
Areas where loops have been optimized a long time ago	70	15	15
Areas where personnel does not have time to optimize loops	40	25	35
Areas where tuning parameters have never been modified after the start-up	10	40	50

Extract from a report

3-FFIC1331	OK
2-LIC-023	OK
2-PIC-256	Scan time should be lower	For this loop, scan time is too long: it slows the loop down and reduces performance.
8-FIC-288	Algorithm is parallel: be cautious	With a parallel algorithm some calculations are needed before looking at the numbers. Here, everything is OK.
	Scan time should be increased	The scan time in this PLC has an impact on PID function. Here, the scan time is too close to the program scan time.
	Filter type should be changed to 1st order	A first order filter is more appropriate for that loop.
	Hi Limit is too high, should be <= 75%	Hi limit is higher than maximum CO. Performance is reduced.
8-TIC-198	Algo is bugged: switch to "D on PV";	This function is bugged and CO will sometimes go wrongly to 0.You have to change it for "D on PV".
	This PID has Eng. units on gain	No scaling: uses raw values. Hence, Kp= 0.43 corresponds to Kp=5.16 on a controller using scaling.
	Config word #9 should be "3F"	"Anti-reset wind-up" is inactive if wrong word. Performance is reduced.
	Derivative time is too high	Derivative time is too long; this will degrade performance. Loop should be optimized. Performance is reduced.
	Derivative gain should be higher	Usually around 10, this value is reduced in presence of noise, but 1.5 is really low. Performance is reduced.
	Filter value could be increased	The filter value will probably be higher than actual after optimization.
	Verify High limit: unusual CO range	Co range is 0-125; maybe the Hi limit has been left at its default value.
	Slew rate is too low	This limits the CO rate of change; it is used to protect equipment. Other techniques are more appropriate to obtain better results.

Cost comparison:

Price to analyse a loop

Based on bump tests (analysis by hand)

Time to collect data and to run tests:	1h
Time to analyze data:	2h
Time to test new values:	1h
Time to document tests:	1h
Total:	5h * $80/h = $400

Using specialized tools (e.g. PIDTuner by ExperTune) and trained personnel to do it effectively

Time to connect, collect data and run tests:	60 min
Time to analyze data:	60 min
Time to test new results:	45 min
Time to document tests:	45 min
Total:	3.5h * $80/h = $280

Service done by Top Control specialist using specialized tools

Time to connect, collect data and run tests:	20 min
Time to analyze data:	10 min
Time to test new results:	15 min
Time to document tests:	15 min
Total:	1h * $175/h = $175

Costs comparison

Method	Unitary cost
Bump tests, by hand	$400.00/loop
Tests + software + trained personnel	$280.00/loop
Top Control specialist + software	$175.00/loop
Top Control Advisor Analysis	$19.95/loop	Loop not optimized

Trade Shows

Exfor 2006 in Montreal (Quebec):
This year again we were at Exfor 2006 in montreal, we thank those of you who came to our booth.

ISA Edmonton:
Top Control will be at ISA Edmonton April 5-6 2006(Booth 78). Come and see us.
Conferences
Wednesday, April 5th, 2006, 10:30 AM - 11:15 AM
            Managing Assets Using Performance Supervision
            Michel Ruel - Top Control

Thursday, April 6th, 2006, 11:15 PM - 12:00 PM
            Quick Start-up Loop Tuning & Subsequent Remote Monitoring
            Jean-François Dubé - Top Control

For more informations about ISA Edmonton please click here.

Trick of the month : Numbers can talk!

Your control system database is a goldmine of information. Values, names and selections can reveal errors, bad values and problems.

1- Database

2- Analysis

3- Results

An example: Consider a flow loop that is part of a temperature cascade scheme:

• Derivative should not be used
• Filter value should be small
• PID values range can be estimated
• Best algorithms to use are known (e.g. no filter on SP)
• Decoupling is easy since FFIC is a lot faster than TIC

From :

loop type: process: product: goal: values: choices:

simple, cascade, ratio, override, feed forward flow, pressure, level, temperature, etc. gas, liquid, pulp, etc. reject disturbance, follow closely SP changes, etc. P, I, D, F, scan time, ranges, limits, etc. algorithms, functions, flags, types, etc.

You can get the following results:

• everything is OK;
• warnings and suggestions;
• errors that needs to be corrected.

A specialist will need minutes to hours to evaluate everything. If one decides to optimize the loop (including verifying all the values), he will probably take hours to do it.

Be careful: one size does not fit all!

Algorithms of different controllers are different: for instance, their units are different.

The following controllers have different numbers but equivalent results (same behavior)

Controller	P	I	D	F
Brand A	33%	6 Min	1 Min	0.14 Min
Brand B, Algo 6	3.53	0.50 RPM	3 Min	8.4 s
Brand B, Algo 3	3	0.17 RPM	1 Min	8.4 s
Brand C, Algo 1	3	360 s	60 s	8.4 s
Brand C, Algo 2	3.53	420 s	51 s	8.4 s
Brand D	3	6 Min	1 Min	0.02 Hz
Brand E	28.3%	420 s	51 s	16.8 s

Examples

Downloaded from DCS's(not all lines are shown)

	3-FFIC1331	2-LIC-023	2-PIC-256	8-FIC-288	8-TIC-198
	Acid flow	Boiler drum level	Water pump discharge	Feedstock flow	Outlet temperature
Scan	0.5 s	250 ms	2 s	20 ms	1 s
Proportional	800%	3.6	145%	0.09	0.43
Integral	0.08 min	235 s	9 s	2.25 min	3.30 min
Derivative	0.00 min	0 s	0 s	0.00 min	4.17 min
Filter	0.01 min	3 s	1 s	0.02 min	0.10 min
Filter type	2nd order Butt	1st order	1st order	Moy. average	2nd order
Controller	xxx	xxx	xxx	xxx	xxx
Algorithm	Basic	P on PV only	PID extended	Independent	D on error
Range PV	0-10 USGPM	-30 in/40 in	0-300 PSI	0-300 T/d	1-1500 °C
Range CO	0-100%	0-100%	0-100%	0-75%	0-125 USGPM
CO limits	0-100%	0-100%	0-100%	0-100%	0-100
	Not all lines are displayed
Slew rate					0.4%/s
Filter SP	SPLL=1		9 s
Others				Beta=0.6	Der Gain=1.5

Question: What can you deduct from the loops in the table?

From the tag and the name, we can determine loop type, process and goal. To analyze the values, it is essential to know the algorithm type. Since we hid the controller names, let's assume that all the controller algorithms are ideal or series type; only 8-FIC-288 is a parallel one.

Also, 8-TIC-198 has some distinctive features, including:

• PV and CO are not scaled: controller compute raw values;
• This algorithm is bugged when using D on Error;
• Actual configuration word does not activate anti reset wind-up mechanism.

What can you find?

• Tuning parameters?
• Limits?
• Choices?
• Selections?

To find what "Top Control Advisor" has found, click here

Conclusion
This new approach generates important savings. Immediate results generate economies and improve profitability.

To contact us

	Lévis's office	+1 (418)834-4321
	Pointe-Claire's office	+1 (514)695-3492
	Hartland's office	(877)867-6473
	Toll free	(877)867-6473
	Email	info@topcontrol.com