Decoupling control loops using ExperTune software

Size: px

Start display at page:

Download "Decoupling control loops using ExperTune software"

Carmel Sims
6 years ago
Views:

1 Decoupling control loops using ExperTune software Theory, diagnosis, and practical considerations Bernardo Soares Torres, M.Sc. (ATAN) Lívia Camargos R. de Oliveira (ATAN) Topics to be covered in this presentation Brief ATAN s presentation PlantTriage tools to find coupled loops How to decouple control loops: Detuning / tuning tighter Static and dynamic decouplers Control strategy How to define variables pairing in a multiloop system using the RGA technique; Examples 1

Segments + AT + IT + Special Solutions 929-13-25 (Rev.

2 ATAN s PRESENTATION + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions (Rev.:3) 3 Metrics Collaborators (Base Jan./27) + Mission + Business Profile Areas of Concentration + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT Level of Education + Special Solutions (Rev.:3) 4 2

3 Customers Geographical Distribution + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions (Rev.:3) 5 + Mission Customers Market Segments + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions Food and Beverage Aluminum Cement Railroad Mining Oil and Gas Metals Petrochemicals (Rev.:3) 6 3

4 Business + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages Automation Technology + Partnership + Metric + Market Segments Information Technology + AT + IT + Special Solutions Special Solutions (Rev.:3) 7 + Mission Automation Technology Industrial Automation + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions INSTRUMENTATION, CONTROL E SUPERVISION IN ON SHORE E OFF SHORE Environments (focus in turn-key project) Systems specification and configuration (SCADA/PLC, UTR, DCS, radio frequency, data communication, etc.) Communication drivers development Start up and commissioning Training Technical assistance (Rev.:3) 8 4

5 Automation Technology PIMS Plant Information Management System + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions Production Information and Management Systems (Historian Systems PI, Infoplus.21, ihistorian, others): Interface with control / Supervision systems (PLCs, DCS and SCADA) Temporal Databank / Real-time process data Interface with Corporate Systems / Relational databank (ERP, Oracle, MS-SQL Server, etc ) Tools for data manipulation and analysis Development of customized applets Training Continuous Help Desk (Rev.:3) 9 Automation Technology Consulting + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions Master Business Plan Technical and economic feasibility study Business plan Conception, ROI assesment Variability reduction studies Basic Engineering Field information collection, Functional Specification Solution definition System configuration Execution Planning (Rev.:3) 1 5

Special Solutions Features: Process Tracking Downtime Management Production Planning Product Tracking

integration Labor accounting Production accounting 929-13-25 (Rev.

SCORE is a Supervisory and Control solution for the primary aluminum production industry: SCORE 7

SCORE 9 - Control and supervision distributed system SCORE Ultra Vision supervision system (HMI- SCORE)

6 Information Technology Production Management - MES + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions Features: Process Tracking Downtime Management Production Planning Product Tracking Production Mectrics (KPIs) Asset Management Activity Based Costing TPM Quality Management Shop floor integration Labor accounting Production accounting (Rev.:3) 11 Special Solutions SCORE + Mission + Business Profile + History + Customers Geographic Distribution + Competitive Advantages + Partnership + Metric + Market Segments + AT + IT + Special Solutions SCORE is a Supervisory and Control solution for the primary aluminum production industry: SCORE 7 Control and supervision semi distributed system SCORE 8 Control and supervision semi distributed system SCORE 9 - Control and supervision distributed system SCORE Ultra Vision supervision system (HMI- SCORE) Top-SCORE Web based information system (Rev.:3) 12 6

7 Topics to be covered in this presentation Brief ATAN s presentation PlantTriage tools to find coupled loops How to decouple control loops: Detuning / tuning tighter Static and dynamic decouplers Control strategy How to define variables pairing in a multiloop system using the RGA technique; Examples Which loops are coupled? 7

Treemap Oscillation Detection Coupled Loops generally cycle together Oscillations in the temperatures of a furnace Many loops oscillate at the same frequency 28 sec Loop name Description Oscillating

8 Treemap Oscillation Detection Coupled Loops generally cycle together Oscillations in the temperatures of a furnace Many loops oscillate at the same frequency 28 sec Loop name Description Oscillating Oscillation period 1 Oscillation period 2 Oscillation period 3 Oscillation strength 1 Oscillation strength 2 Oscillation strength 3 FIC1722 Vazão de GM da Zona 7 1* FIC1624 Vazão de AR da Zona 6 1* FIC1724 Vazão de AR da Zona 7 1* FIC1824 Vazão de AR da Zona 8 1* FIC1822 Vazão de GM da Zona 8 1* PIC114 Pressão de Ar de Combustão FIC1622 Vazão de GM da Zona FIC1522 Vazão de GM da Zona TIC1521 Temperatura da Zona PIC1424 Pressão de Ar na Zona

9 Finding possible couplings with PlantTriage Process Interaction Mapping Process Interaction Mapping diagnosis: No coupled loops (Metallurgical Furnace) 9

Simple strategy to decouple control loops: detune the controller Making the tuning of a loop more conservative by changing the PID parameters

10 Simple strategy to decouple control loops: detune the controller Making the tuning of a loop more conservative by changing the PID parameters robustness change. K p T i Example 1 Decoupling level and temperature loops Detuning a level loop reduces oscillation in a temperature loop K p T i 1

11 Example 2 - PIC-321 Control of the combustion air pressure in a furnace Common Oscillation periods!! What is the best tuning? The implementation of the best tuning for all loops can decrease the performance of the entire system! RRTs of the loops PIC321 and FIC 37.1 became closer after optimal tuning of these loops RRT of PIC 321 changed from 2 sec to 63 sec after tuning RRT of FIC 37.1 was 6 times bigger and became 2 times bigger only PIC 321 FIC 37.1 Coupling between loops!!! 11

PIC-321 Tuning synchronism - RRT The loops were tuned again to adjust the RRTs and reduce the coupling. RRT of the loop PIC321 became 3 times smaller than that of FIC 37.

12 PIC-321 Tuning synchronism - RRT The loops were tuned again to adjust the RRTs and reduce the coupling. RRT of the loop PIC321 became 3 times smaller than that of FIC 37.1 Decoupled Loops PIC 321 FIC 37.1 Spectral Analysis PIC-321 x FIC 37.1 Before FIC-37.1 After Spectra of PIC-321 and FIC-37.1 became different PIC-321 Coupling removed!!!!!!! 12

13 PIC-321 and FIC 37.1 after decoupling Performance after decoupling Example 3 Metallurgical Plant Initial General Statistics 13

14 Initial Assessment Couplings among loops Fan control loops: PIC14, PIC24, PIC35 and PIC45 Coke (carbon fuel) production Process Chart Coupling Coupling Coupling Coupling 14

Is making the tuning more agressive a good way

least 3 times bigger Steps of the job: Tests

15 Is making the tuning more agressive a good way to decouple loops? PIC14 PIC11 Strategy for tuning: RRT is the key!!! Goals: PIC14 Relative Response Time at least 3 times bigger Steps of the job: Tests for process identification Adjustment of the Relative Response Time (RRT) 15

16 Decoupling: smaller oscillation strength Results Response to disturbances 4 minutes 25 sec 16

17 PlantTriage Assessments Final Assessment General Statistics Before After 17

PIC14 PIC11 Decoupling However, in some cases, detuning can really harm the performance of

18 Decoupling loops based on tuning Change tuning (robustness) of the coupled loops in order to have RRT (Relative Response time) as far as possible from each other (at least 3 times). PIC14 PIC11 Decoupling However, in some cases, detuning can really harm the performance of the detuned loop to unacceptable levels. Alternative solution Decouplers design and implementation 18

Dynamic Gp D21( s) = Gp 21 22 ( s) ( s) 21 MIMO

19 Decouplers Design MIMO System Level and temperature of a tank Types of decouplers 1. Dynamic Gp D21( s) = Gp ( s) ( s) 21 MIMO System Level and temperature of a tank 2. Static (simpler) Kp21 D = Kp 22 19

configuring the extra trends, create a new loop for multiloop analysis. E.g.: Modeling couplings 2

20 Finding the models in PID Tuner Option Advanced of the Loop Setup: Configuration of extra pens allows for monitoring other variables simultaneously (Multiloop Analysis) Finding the models in PID Tuner Option Advanced of the Loop Setup: Configuration of extra loops After configuring the extra trends, create a new loop for multiloop analysis. E.g.: Modeling couplings 2

Through the off-line analysis option, it is possible to choose another PV-CO pairing and

21 Finding the models in PID Tuner Configuration of extra loops: The data acquisition of all variables is done simultaneously. Through the off-line analysis option, it is possible to choose another PV-CO pairing and then model the interaction between these variables (couplings). Example: Finding the models in PID Tuner Coupling between flow and temperature in DemoMMI 21

Temperature Example: Finding the models in PID Tuner Model

22 Example: Finding the models in PID Tuner Step in the flow loop CO while the temperature loop is in manual Flow Temperature Example: Finding the models in PID Tuner Model between the flow loop CO and the temperature Temperature Flow CO 22

23 Example: 5x5 multiloop system Block Diagram Decoupling - example Matrix of the system s transfer functions with the relevant interactions TF = 6s 1.21e.18 s.11 s.124e 85s s.16 s.192 s 5.47e 15s + 1 s.25 s.3 s e 175s + 1 s 2s.185e 692s s.59e 489s

24 Decoupling - example Simulation - Implementation of the TFs in MATLAB STEC Simulator Decoupling - example Decouplers Design 24

25 Decoupling - example The only decoupler for this system that is not feasible was: D NTA TTA ( s) =.47 e 15s e 489s + 1 5s 49s 489s + 1 =.69 e 15s s Hence, the following was adopted instead D NTA TTA ( s) =.69 Decoupling - example Implementation of the decouplers in MATLAB STEC Simulator 25

26 Decoupling - example Some results achieved by decoupling the system: Without decoupler With static decoupler Influence of a change in the TTP flow With dynamic decoupler Decoupling example - furnace Global diagram of the PID control loops of a 4x4 system and the couplings among them 26

27 Example of the use of FBD Conventional PID with decouplers and feedforward structure Implementation of dynamic decouplers τs Ke D( s) = Ts

Example of static decoupling Without decoupler With decoupler Decoupling between the differential pressure loop and the temperature in a LPG combustion chamber Other strategies Description of the

28 Example of static decoupling Without decoupler With decoupler Decoupling between the differential pressure loop and the temperature in a LPG combustion chamber Other strategies Description of the process: Pulp sump at an iron ore processing plant Control loops: Level actuates in a pump regulating the pulp outflow to flotation columns in order to keep the level constant Density actuates in a valve that adds water into the sump to reduce the density Disturbance: pulp feed to the sump (it depends on the upstream stages of the process) 28

29 Old control strategy Process behavior Disturbance in the foam level of the flotation column downstream Level of the pulp sump Foam level of the flotation column 29

30 New control strategy Results of the change in the control strategy Level of the pulp sump Foam level of flotation columns 3

What should we do if the disturbances are so strong that our decoupling strategies

31 Results of the change in the control strategy Reduction in the variability of the foam level of the flotation column B E F O R E A F T E R Control System Design What should we do if the disturbances are so strong that our decoupling strategies cannot handle them? Study the control system design and variable s pairing (CO x PV) 31

32 RGA RGA: Relative Gain Array Goal: Through the steady-state gains of the system, determine the degrees of interaction among the process variables Result: Definition of the best variables pairing RGA First Step Find the Steady State gains based on the matrix of the system s transfer functions already obtained with PID tuner TF = 6s 1.21e.18 s.11 s.124e 85s s.16 s.192 s 5.47e 15s + 1 s.25 s.3 s e 175s + 1 s 2s.185e 692s s.59e 489s

RGA The elements of the RGA can be calculated by the following expression: K ij = steady-state gain between Ci and th H = ( i, j) element of H ij M j Steps to determine the

degree of interaction grows as λ i MV j increases. λ < : The sign of the open-loop gain and of the closed-loop gain are different; thus, CVi and MV should not be paired.

33 RGA The elements of the RGA can be calculated by the following expression: K ij = steady-state gain between Ci and th H = ( i, j) element of H ij M j Steps to determine the RGA: Computation of K (gains matrix) 1 Calculation of H = ( K ) T Calculation of RGA analysis The elements in a row of column always sums to 1 λ > 1 : CV and interact and the degree of interaction grows as λ i MV j increases. λ < : The sign of the open-loop gain and of the closed-loop gain are different; thus, CVi and MV should not be paired. j λ = 1 : ideal pair there is no interaction with other loops. λ = : MV j does not affect CV ; therefore they should not be paired. i < λ < 1 : There is interaction between the loops. 33

34 Example of RGA application K (Gains matrix) FCV1 LCV2 LCV1 TCV1 ÂNG. VZ NTP NTAQ TTP TTAQ Example of RGA application Matrix Λ (RGA) VZ NTP NTAQ TTP TTAQ FCV1 LCV2 LCV1 TCV1 ÂNG

35 Conclusions Couplings among loops are harmful to process performance. This is quite common in industry; PlantTriage has some tools to help you find the coupled loops; Once detected, the loops need to be decoupled. Possible approaches: Change Tuning Design decouplers Change control strategy In order for the control system to reach a great performance level, the variables pairing needs to be carefully chosen. The RGA technique can help you to define the best variables pairing in a multiloop system and PID Tuner can be used to find the models. Bernardo Soares Torres bernardo.torres@atan.com.br Búzios Rio de Janeiro - Brazil 35

METHODOLOGY FOR REDUCING THE CONTROL LOOPS OSCILLATION AT AN IRON ORE PROCESSING PLANT

METHODOLOGY FOR REDUCING THE CONTROL LOOPS OSCILLATION AT AN IRON ORE PROCESSING PLANT Lúcio Fábio Passos 1 lucio.passos@atan.com.br Bernardo Soares Torres 1 bernardo.torres@atan.com.br Vicentino José