cpm Plus Loop Performance Manager 3 2

cpm. Plus Loop Performance Manager 3. 2 © ABB - 1 - Introduction to LPM

Presentation Outline n Introduction / Motivation n cpm. Plus LPM Features n n © ABB - 2 Control Performance Monitoring n n Tuning Supporting Utilities cpm. Plus LPM Plant-wide Disturbance Analysis

cpm. Plus Loop Performance Manager © ABB - 3 1. Introduction/Motivation

Why Loop Performance Monitoring? n “Does my plant run optimally? ” n If not, how much can be accounted to the process automation, especially the control loops? ” We should use available measurement data instead of just storing it. n Normal operation does not necessarily mean optimal operation n © ABB - 4 n Loop optimization saves money without new capital investments

© ABB - 5 Real world performance is suboptimal!

An investment that has to pay off! n Typical control loop as a $25, 000 asset n Half of it is lost n 50 % well tuned n 25 % uneffective control n 25 % decrease performance Half time of good performance = 6 months n 2 – 4 hours to investigate and improve control performance n Typical process contains 2000 – 4000 control loops n Only few people with appropriate know-how n Average process engineer in charge of 400 control loops n © ABB - 6 n 25 % of 4000 loops do harm, this means…

Analysts start to get the message Recent issue: “Include control loops in asset management” Les A. Kane, Editor Quotes: “…while process equipment is an integral part of AM [asset management] programs, control loops … often don’t receive the same attention. ” n “Performance of control loops … degrades slowly over time with little fanfare. ” n “Without properly tuned control loops to minimize variability, … substantial benefits are lost. ” n “… even a slight degradation in process control can result in millions of dollars in lost profitability. ” n “Identifying the high-payback control loops requires evaluating all control loops, which would be an insurmountable task without the aid of control loop performance monitoring and analysis software. ” n © ABB - 7 n “When first installed, advanced process control typically provides substantial benefits. Sustaining those benefits due to changing conditions, however, has been a problem. ” n “… it’s a good time to ensure control systems are part of your AM efforts. ”

Benefits of Tuning and Auditing n Maintains control system at its peak n Loop Tuning n n Loop Auditing n © ABB - 8 n Enables the plant engineers to reach loops optimum performance with significant time savings (vs. manual tuning) Provides timely indication of equipment/automation/process problems. In this way it easy to keep the loop at their , allowing to stay at the optimal performance Also, it provides stable foundation for multivariable/advanced control

cpm. Plus Loop Performance Manager – What is it? Loop Tuning Ø Challenge Optimal PID Tuning is critical to efficient process operation Ø Loop Tuning is a time consuming activity Ø Typically, only expert engineers can perform Tuning Ø Ø Solution Ø LPM Tuning makes definition of optimal PID parameters an easy, reliable & manageable task Loop Auditing Ø Challenge Ø Ø Ø Loop optimization is frustrating, because after few months all results seem lost due to the process variability Plant engineers have to look at hundreds of signals and among them detect possible problems Solution © ABB - 9 Ø Once Loop Optimization is performed, LPM Auditing monitors loops and allows the process engineer to immediately address problem areas

Cost of bad control Cost High Loop Tuning Execution Dream Low Time Reality Cost High Low Time © ABB - 10 Cost High Realistic dream with Auditing Low Time

cpm. Plus Loop Performance Manager © ABB - 11 2. LPM Features

LPM Tuning – Workflow n Which step to tune a Loop? Model Collect © ABB - 12 Configure Log Tune

LPM Features – Data Collection Configure database by loops n Simultaneous data collection for multiple loops n OPC connectivity n Direct connection for Infi 90/Symphony n Data collections stored as object on navigation tree for future retrieval n © ABB - 13 n Possibility to exploit auditing automatic data collection for tuning purposes

LPM Tuning - Identification n BASIC for not experts and ADVANCED with fully scalable complexity for expert control engineers n Manual or Automatic structure selection by best fit n Parameters specified - up to 4 th order n Identification also with Process in Close Loop n Validation n © ABB - 14 n Model simulated with another data set Evaluation n Ideal step response n Bode diagram BASIC ADVANCED

LPM Tuning n 5 Tuning methods available n Time domain analysis n Frequency analysis Support many vendor specific PID controller types n Ability to model, tune, and analyze Feedforward control loops. Considers feedback tuning. n Special treatment of Cascade control loops © ABB - 15 n

LPM Advanced Tuning Features New Tuning values can be assessed on model different from the ones used to obtain the tuning set (Simulate Mode) n Data pre-processing functionalities n Advanced Feedforward Loop Tuning Management n HTML-based and information-richer Tuning Logs n © ABB - 16 n Advanced Cascade Loop Tuning Management

LPM Tuning – Advantage Ø State of the art Tuning Algorithm, but with userfriendly tool to make Advanced Control Theory accessible to every Plant Engineers Ø Ready for every DCS § OPC connection § Calculated PID parameters (Kp, Ti, Td ) with the definition of your DCS Ø Identification also with Loop in normal Close Loop Mode Ø Not only basic PIDs, but also Feed. Forward and Cascade Loop © ABB - 17 Control Tuning becomes easy, fast, profitable

Performance Assessment: Tuning vs. Auditing n Tuning - Design stage Reasonable design © ABB - 18 Slightly tight design n Assessment stage ? Is this good control? If not: why?

Control loop monitoring – non-invasive! © ABB - 19 indices

LPM Auditing - General concept n based on available signals only(Set. Point, PV, CO) Info available information can be incorporated know how n performance indices, measures I 1, I 2, I 3, … n © ABB - 20 n inference engine know how Hypothesis, Diagnosis n suggest remedies

Kinds of Performance Indices in LPM Target “SP” Controller Output “CO” Measurement “PV” n Basic statistics n Nonlinearity indices n Data Validity n Property indices n Control loop modes n Housekeeping n Tuning Performance indices Special indices n Continuous indices n Oscillation indices © ABB - 21 n n Valve indices

Kinds of Diagnoses in LPM Performance indices Auditing Rules + Maintenance Diagnoses Indices plus know-how organized in a Root-Cause analysis elaborate Maintenance Suggestions n © ABB - 22 n Diagnoses dealt with problems in: Tuning, Actuators and Sensors, External disturbance

Overall Performance PRECONDITIONS Acceptable performance index Harris index Acceptable setpoint crossings index Acceptable Overall performance Setpoint crossing index (not for Level Control) Variability random Oscillation index of control error excellent Controller output within range good Saturation index Loop automatic Automatic mode index Acceptable cascade tracking Cascade tracking index (if in cascade) Acceptable response speed © ABB - 23 ACF to horizon index fair poor

Auditing workflow Loop configuration Assign • TAG connection • Signal ranges • Loop Type Auditing configuration Assign • Data collection schedule • Batch / continuous auditing Loop category configuration Assign • Sampling rate • Batch duration Report configuration Assign • Report layout Configuration file Configuration Start auditing Setpoint CO, PV Database © ABB - 24 Indices Trend Plot Process Engineer • Investigate Problem • Activate Maintenance Indices Reports Excel, HTML Maintenance Operator • Repair device • Tuning Data collection Indices calculation Diagnoses Report Excel, HTML Output Periodical reports Maintenance

Example oscillation investigation. . . cycling load static friction F FC tight tuning Diagnoses Indices Verify overall Performance Ø Oscillation details (period, amplitude…) Ø © ABB - 25 Ø Detect oscillation Ø Amount of problem for every causes Ø Decide among the 3 causes Ø Trend plot for every index

LPM Auditing - KPI Reporting & Analysis n Reporting Pre-defined report templates n © ABB - 26 n Both numerical and chart-based assessment

Advanced Auditing Features Advanced Indices & Diagnosis trend facility (on multiple even non consecutive periods) n User-defined Indices n Enhanced KPI and Diagnosis set n Server Status Monitor to supervise all the auditing functions n “What Is Changed” report to immediately eye-catch recently developed events n © ABB - 27 n Possibility to generate a “Detailed Loop” Report, with in-depth charts and numerical figures

Detailed Report Time domain view (PV, SP, CO) n Power spectrum view (PV) n Statistical view (PV, CE) n CE vs. CO, during oscillation becomes a ring. From the shape it is possible to detect stiction n Impulse response of Disturbance Rejection n Sensitivity study for Prediction horizon (good situation when lines is increasing with steps) © ABB - 28 n

And More … Operation-Sensitive Reports: allow to monitor control loops according their operating region(s) n Examples: production campaign types, loads, … n Capability to extract and utilize for Tuning purposes data automatically collected during Auditing normal operation © ABB - 29 n

Bulk Database Import for quick DB Configuration q Allows to import tag configuration details from Excel spreadsheets © ABB - 30 q Results in Relevant Manpower Savings

Infi 90/AC 800 F Bulk Import Tool © ABB - 31 Available as an add-on to standard LPM Functions

LPM auditing - Everything also by Web Facility to get and manage all LPM information from any location in the net © ABB - 32 … Reports Configuration … From the LPM Home Page it is possible to navigate to … … Reports Retrieval … … Tuning Logs

LPM Auditing: Advantages Automatic data-collection enable actual continuous loop performance assessment rather than “sporadic sampling”, maximizing the chance to identify and correct insurgent productionrelated problems n Simple, straightforward diagnostic indications are made available for the basic user or for quick assessment n Diagnostic results are based on sophisticated indices which are able to provide explanations or in depth analysis for advanced user or when needed n © ABB - 33 n Both Diagnosis and Indices are saved and stored in user-configurable Reports so to not require continuous attention from plant crew and to provide a comprehensive “plant history” track record

cpm. Plus Loop Performance Manager © ABB - 34 3. Plantwide Disturbance Analysis

Plant-wide disturbance analysis - intro n Analysis process data off-line n Searches for data pattern in time (oscillations) and frequency (specra) to identify n Oscillations n Interactions Identifies most likely root-cause (with no info on plant topology/interconnections) n © ABB - 35 n Integrated in LPM, could use auditing data or external data (e. g. plant historian)

© ABB - 36 Plant-wide disturbance analysis - intro

PDA Application – Case 1 Cascaded Distillation Columns: 1 TI TI 6 FC FC 2 Internal Condenser 1 PDI 6 Internal Condenser TI 39 2 PDI 20 2 1 Column 1 FC Steam 22 TI TI 4 3 10 FC TC Column 2 TC 32 1 3 19 7 16 TI 8 4 1 FC TI 4 5 LC 3 TI 4 3 PI FC 1 5 9 TI TC LC 3 1 Steam PI © ABB - 37 Decanter 2 TC LC 2 PI 2 FC 7

PDA Application – Case 1: Dataset Details n Primary cycle n n Cause is LC 2 valve movement problem n n Column 1 level through column 2 distillate Many variables cycling together Secondary cycle n n © ABB - 38 n Top of column 1 (distillate FC 2 and temperatures) Cause is FC 2 valve movement problem 96 hours total data, sample time = 30 sec Dataset window chosen

PDA Application – Case 1: Clustering n Three main Clusters detected: n Two Oscillation Clusters n One PCA Cluster n A few tags have been added to clusters due to process considerations © ABB - 39 Oscillation Clustering: manually added 1 related tag to grouping (primary cycle)

PDA Application – Case 1: Clustering n Three main Clusters detected: n Two Oscillation Clusters n One PCA Cluster n A few tags have been added to clusters due to process considerations © ABB - 40 Default grouping: secondary cycle, had to add ti 2. pv and ti 3. pv tags manually

PDA Application – Case 1: Clustering n Three main Clusters detected: n Two Oscillation Clusters n One PCA Cluster n A few tags have been added to clusters due to process considerations © ABB - 41 PCA cluster default grouping, manually added 2 related tags to grouping (primary cycle)

PDA Application – Case 1: Main Clustered Disturbances 1 TI TI 6 FC FC 2 Internal Condenser 1 PDI 6 Internal Condenser TI 39 2 PDI 20 2 Column 1 FC 3 22 TI TI 4 3 10 FC TC Column 2 TC 32 1 TI 7 16 TI 8 4 4 4 3 1 FC TI 5 LC 3 19 1 Steam PI FC 1 5 9 TI TC LC 3 1 Steam PI © ABB - 42 Decanter 2 TC LC 2 PI 2 FC 7

© ABB - 43 PDA Application – Case 1: Root Cause Analysis Good default results for non-linearity analysis (primary cycle) (ranks LC 2 as highest non-linearity)

© ABB - 44 PDA Application – Case 1: Root Cause Analysis FC 2 cycle (secondary cycle) analysis: non-linearity correctly identifies FC 2

PDA Application – Case 1: Disturbance Propagation 1 TI TI 6 FC FC 2 Internal Condenser 1 PDI 6 Internal Condenser TI 39 2 2 PDI 20 Column 1 FC 22 TI 10 TI FC TC 4 3 TI 7 16 TI 8 4 4 4 3 1 FC TI 5 LC 3 19 3 Steam Column 2 TC 32 1 1 PI FC 5 1 9 TI TC LC 3 1 Steam PI © ABB - 45 Decanter 2 LC 2 TC Cluster 1 Cluster 2 PI 2 FC 7

PDA Application – Case 2 Vaporizer System: Vapor Header 1 PC Steam 2 SP 5 PC PC 2 LC FC 5 LC A B Steam 7 C LC 7 FC 9 PI 9 D LC © ABB - 46 PI 9 Liquid 7

PDA Application – Case 2: Clustering n Two main Clusters detected: n One Oscillation Clusters n One PCA Cluster n A few tags have been added to clusters due to process considerations © ABB - 47 Cycle of interest

PDA Application – Case 2: Root Cause Analysis © ABB - 48 Ref. to: “Peak Performance: Root Cause Analysis of Plantwide Disturbances”, ABB Review 1/2007 Good results for non-linearity, clearly identifies LC 2 as root cause

cpm. Plus - LPM Conclusions Tuning Auditing n With LPM Process Engineers (also non expert in control theory) can optimize Loop behavior n Control Performance Monitoring is non-invasive, simple to perform and very efficient n Benefits: increase process profit, more stable working condition, more safety operations n LPM detects automatically problem at the beginning of their occurrence n Performance monitoring nowadays answers the most important questions to help the plant personnel to pinpoint and remove problems n The right information to the right people PDA n Very valuable insight on process corrrelations, oscillations and root causes with a few points and click Could use your historian data (with reasonable data compression) n © ABB - 49 n Complementary to tuning and auditing