Houston Day 2 2013 for antisurge control

Optimizing Turbomachinery Controls Symposium

Houston, TX June 2013

Agenda – Day 2 • Steam Turbine Control – –

Turbine Start-up & Shutdown Automation Speed & Extraction Control Techniques and Demo

• Application Examples – – –

LNG Liquefaction – Refrigeration Compressors NGL Fractionation Facilities & Bayu Undan Example Ammonia-Urea Unit Applications

• Process Control vs. Safety Shutdown Systems – –

System Availability and Fault Tolerance API Standards Update

• Tips for Specification Writing

Steam Turbines

Challenges of Steam Turbine Control 1. Reliability 2. Efficient Operation 3. System Integration

• Prevent unnecessary process trips and down time • Minimize the effect and duration of process disturbances • Avoid overspeed and other machine related trips

Challenges of Steam Turbine Control 1. Reliability

• Operate at efficient energy levels

2. Efficient Operation

• Accurate speed measurement

• Consistent and accurate 3. System valve positioning Integration

Challenges of Steam Turbine Control 1. Reliability 2. Efficient Operation 3. System Integration

• Speed and extraction control • Startup and shutdown automation in concert with driven equipment

Steam Turbine Startup & Shutdown Sequencing

Break Away can be Extremely Fast Bearing

Lube Oil

Shaft

High friction

RPM V1

Time

Low friction

Break Away Control Prevents Machine Damage RPM V1 RPM-SP

Time

Benefits

• Reduced overshoot during breakaway of turbine • Less mechanical stress on cold machine • Reliable and repeatable start up

Critical Speeds • Critical speed is a speed at which the turbomachinery train vibrates at a harmonic or resonant frequency • Peaks of multiple oscillating waves “add” creating constructive interference

Critical Speeds • Most turbomachinery trains have at least one and often multiple critical speeds • Operating the turbomachinery train at a critical speed for an extended period of time can result in severe damage • Critical speeds are typically below rated speed

Critical Speed Avoidance • Critical speed range low and high values are configured • RPM-SP cannot be set in this range • As soon as RPM-SP goes above Ncritical,low the controller ramps RPM-SP to Ncritical.high based on configurable ramp rate RPM-SP RPM V1 Ncritical,high

Rated Speed

Critical Speed Range

Ncritical,low

Time

Avoiding Critical Speed Damage (Lack of Steam) • • • • •

RPM-SP is ramped thru Ncritical,high Controller opens V1 to accelerate turbine to Ncritical,high With V1 100% open machine does not reach Ncritical,high within predetermined time t1 due to lack of steam pressure and/or flow Controller ramps down RPM-SP to Ncritical,low Turbine decelerates to Ncritical,low

Machine damage is avoided RPM-SP RPM Ncritical,high

t1

Critical Speed Range

Ncritical,low Time 100% V1 0%

Time

Avoiding Critical Speed Damage (Sticky Valve) • • • • •

RPM-SP is ramped down thru Ncritical,high Controller closes V1 to decelerate turbine to Ncritical,low Turbine does not reach Ncritical,low within predetermined time t1, because of a problem with V1 Controller ramps RPM-SP to Ncritical,high Turbine accelerates to Ncritical,high

Machine damage is avoided RPM-SP RPM Ncritical,high

t1

Critical Speed Range

Ncritical,low Time 100% V1 Position V1 Output 0%

Time

Start-Up Schedules for Steam Turbines • OEM provides start-up schedules for steam turbine • Machine needs to be kept for certain period on given speed • Typically there are 1 or 2 idle speeds • After start-up the machine can be loaded RPM

OEM start-up diagram

To rated speed

Idle 2 Idle 1 Start-up time 1

Start-up time 2

Time

Automatic Control of Idle Speeds • Speed controller automatically ramps turbine to and between Idle speeds and to minimum governor setting • Machine accelerates or decelerates at configurable ramp rates based on how long the turbine has been down – Normally there are (2) sets of ramp rates, one for a “hot” turbine and one for a “cold” turbine

• Ramps can be aborted and resumed at any time

Advanced Automatic Control • “Warm Start-up” • Automatically brings turbine to rated speed using calculated delays and ramp rates – Based upon the case temperature, hot and cold startup ratio, and idle time of the turbine

• Start permissive inputs – Checked both at start and during sequencing

• Configurable ramp rates

Warm Start-up Sequencing Cold Start-up To idle 2

RPM

Ramp Rate: Sc ≤ Sw ≤ Sh

Idle 1

Idle Time: tidle, h ≤ tidle, w ≤ tidle, c

Slope Sc Time tidle, c

Warm Start-up RPM

To idle 2

RPM

Hot Start-up To idle 2

Idle 1 Idle 1 Slope Sw

Slope Sh tidle, h

Time tidle, w

Time

Start-Up Sequencing

Speed Loop Tuning Techniques

Controlling Power vs. Speed

At constant speed Power Delivery

• •

=

Power Consumption

At constant speed the power consumed by the load is equal to the power delivered by the steam turbine Speed modulation is used to compensate for changes in load, however: • •

The objective of steam turbine control is to adjust the delivered power to match the current load Optimized loop tuning should take into account the relationship between rotational speed & delivered power

Fan Laws

Power = f(N3) • Power is a function of rotational-speed3 • Speed control only indirectly controls power • Constant loop tuning can work marginally well between minimum and maximum governor • The same tuning does not work well below minimum governor during start-up & shutdown Power

Minimum Governor

Maximum Governor

70%

105%

Speed

Gain Changes as a Function of Speed Digital speed controllers introduced piecewise characterizers for gain adjustment outside the normal governing range of operation Power

Speed

Variable Gain in Turbine Speed Controllers Loop gain in CCC speed controllers is as a function of the speed/power relationship over the complete speed range Power

Gain characterization function

Linear power gain for complete speed range

Speed

Benefits of Variable Gain • Allows for responsive tuning in all speed Power ranges • Provides more accurate speed control and more reliable speed limiting • Good control at low speeds is required to allow for fully automatic startup

Gain characterization function

Linear power gain for complete speed range

Speed

Integrated Antisurge Control and Turbine Speed Control Start-up • Turbine starts from zero speed and ramps to minimum speed with the recycle valve fully open

Rc

Qs,

vol

• The recycle valve starts to ramp closed & performance control is switched to auto • Should the OP touch the surge control line, the a/s controller overrides the valve ramping as needed • The turbine is brought to normal speed safely

Steam Extraction Control

Extraction Steam Turbine

Extraction Turbine Horsepower Relationships V1

HP section

HP horsepower

V2

LP section

LP horsepower Total developed horsepower

LOAD

LOAD Total consumed horsepower

Valve Interaction Load Change When load increases, V1 opens to supply additional power

V1

V2

LOAD

This causes the extraction flow to increase and V2 will need to open to maintain constant extraction

Valve Interaction Steam Demand Change When extraction demand increases, V2 closes to supply additional extraction steam reducing steam to the LP turbine

V1

V2

LOAD

Total power produced to drive the load drops and V1 needs to open to maintain constant rotational speed

Extraction Map Inlet Steam Flow

Maximum level of extraction

Minimum level of exhaust flow

Qin

V1

Stable zone of operation

V2

Inlet Steam flow limit

Maximum level of exhaust flow

Horsepower limit

LOAD

Qextract

Qexhaust

Minimum level of extraction Horsepower Delivered to Load

Speed and Extraction Control Loop Interactions

Inlet steam flow

B A

C

D

horsepower

LOAD

Integrating Speed & Ext. Control Load Change

Inlet steam flow

SE 3X

SIC

V1

PID

1

B A

horsepower

FT 1

XIC

PID

1

X

V2

Integrating Speed & Ext. Control

Extraction Demand Change

Inlet steam flow

SE 3X

SIC

V1

PID

1

X

B A

horsepower

FT 1

XIC

PID

1

V2

Dynamic Simulation: Extraction Steam Turbine

• • •

Compressor is controlled on Discharge Pressure by PF-1 SC-1 controls rotational speed EX-1 controls turbine extraction flow or pressure

Turbine Overspeed Protection

Overspeed Issue 1

Overspeed Issue 2

Overspeed Issue 3

API/ISO Governing and Protection Speed Requirements • Maximum Temporary Overshoot Speed – 127%

• Over-speed Trip Speed – 116%

• Max Allowable Speed Rise per NEMA D – 112%

• Maximum Continuous Operating Speed – 105%

• Rated Operating Speed – 100%

Disabled Knife Edge Trip System Mechanical Overspeed Trip Finger (Moves to Left on Overspeed) Trip Valve Actuation Lever (Moves Down Upon Trip)

Trip Valve Located within Box Knife Edge Latch (Unlatches on Overspeed)

Machine Vibration Causes Mechanical Overspeed Trip Finger (Knife Edge Latch) to Let Loose and Cause Nuisance Trips of Turbine Under Normal Running Conditions

Bricks and Metal Placed to Avoid Nuisance Trips due to Unlatched Knife Edge During Standard Operation (Dangerous Should Overspeed Need to Trip Turbine)

Overspeed Protection

Governor vs. OST

• Governor is the first line of defense for preventing over speed • Governor electronic trip acts as a backup to the primary overspeed trip device – If the turbine speed exceeds the trip speed, the governor will initiate a trip • Closes the governor valves • Initiates a trip of the turbine via T&T valve

• Primary overspeed trip system – Mechanical over speed trip system – Electronic over speed trip system

Digital Overspeed Protection • •

Turbomachinery losses among the highest paid by insurers Overspeed wreck represents one the most catastrophic accidents: – – – –

•

Endangers personnel Damages the turbomachinery train Can cause damage to other plant equipment Can result in costly interruptions of process

Mechanical overspeed trip systems are non–redundant, require overspeed testing via actual turbine run-up, are imprecise & unreliable

Speed of Response is Critical • Steam turbines can accelerate extremely quickly during process upsets • Major upsets include: – Surge on the driven compressor – Breaker trip on the generator – Fast power reduction on the local grid

• Traditional speed control can be too slow to catch these type of disturbances • Results: – Unnecessarily large process disturbance – Machine & process shutdown due to over-speed – Potential machine damage

Steam Turbine Rotor Dynamics

.619  Nrated WR Tc , rotor  6 10  hprated 2

Rotor time constant: where: NR WR2 hp

Tc,rotor

2

Rated speed (RPM) Rotor inertia (lbs-ft2) Rated horsepower

= The time it would take an instantaneous load loss to cause a doubling of rotor speed when starting from rated hp & rated speed

Steam Turbine Rotor Dynamics Steam turbine driven recycle compressor example: Recycle compressor data: NR WR2 hp

Rated speed (RPM) 13,500 Rotor inertia (lbs-ft2) 50 Rated horsepower 2,500

.619  Nrated 2  WR 2 Tc , rotor  106  hprated .619  13,5002  502 Tc , rotor  106  2 ,500 Tc , rotor  2.25seconds

• Turbine speed will be 27,000 rpm after 2.25 seconds • Overspeed trip settings (116% rated) will be reached in 337 ms • Overspeed trip system needs to react in 225 ms to prevent speed from exceeding 127%* level * Maximum Temporary Overshoot Speed

The Overspeed Avoidance Algorithm RPM-SP RPM Electronic Overspeed Overspeed Avoidance Maximum Governor

• • • • • •

Rapid load drop causes turbine to accelerate rapidly Conventional PID control starts to close the V1 valve Operating point hits overspeed avoidance line Open loop response rapidly closes the valve to avoid overspeed Speed drops below maximum governor PID control brings speed back to set point

Time V1

Time

Open Loop Control Lacks Accuracy • A fixed step change will either be too small or too big for a specific disturbance – Too small may not protect the machine – Too large may cause an unnecessary loss of speed and process disturbance

• The rate of change in speed (N/t) can be used to estimate the extent of the load loss – Calculates the appropriate size of the step change to be implemented

Improving the Accuracy of the Step Change Step = a configurable constant x

N t

• System adapts to the size of the disturbance • Bigger disturbances provoke faster closing of the valve RPM

RPM

Medium disturbance

Large disturbance

Overspeed Avoidance

V1

V1

Time

Time

Overspeed Avoidance Algorithm

Benefits: • Overspeed can be avoided for virtually any disturbance • Fewer overspeed incidents increase machine life • Process is kept on line

Application Examples

LNG

LNG Facts • Approx. 95% Methane • Cooled to - 260° F (-161° C) • 1/600th of original gas volume • http://www.youtube.com/watch?v=Ft1rHNXZozY

Installations Country

Plant

No. of Trains

Capacity (MTPA)

Start year

Process

Abu Dhabi

Das Island

3

5.7

1977/1994

APCI

Australia

Woodside LNG

5

16.3

1989-2008

APCI

Australia

Woodside LNG (Pluto) Bontang I-VI

1

4.8

2010

APCI

Indonesia Malaysia Nigeria

Bintulu (MLNG) Satu, Dua, Tiga Bonny Island (NLNG)

6

21.8

1977-2000

APCI

8

22.7

1995-2003

APCI

6

21.9

1999-2007

APCI APCI

Oman

Oman LNG

3

10.3

2000-2006

Qatar

QatarGas JV I-IV

7

40.9

1996-2010

APCI

Qatar

RasGas JV I-III

7

36.3

1999-2009

APCI

Egypt

Segas / Union Fenosa

1

5.0

2005

APCI

Trinidad

Atlantic LNG 1-4

4

14.8

1999-2005

ConocoPhillips

Egypt

Egypt LNG

2

7.2

2005-06

ConocoPhillips

Australia

Darwin LNG

1

3.5

2006

ConocoPhillips

Equatorial Guinea Norway

EGLNG / Marathon

1

3.4

2007

ConocoPhillips

Snohvit (Statoil)

1

4.2

2007

Linde

Russia

Sakhalin LNG

2

9.6

2009

Shell PMR

Peru

Camisea LNG

1

4.4

2010

APCI

Angola

Angola LNG

1

5.2

2009

ConocoPhillips

Indonesia

Tangguh LNG 1&2

2

7.6

2011

APCI

Algeria

Skikda LNG

1

4.5

2011

APCI

Australia

Gladstone LNG

2

7.8

2013

ConocoPhillips

Australia

Gorgon LNG

3

15

2014

APCI

Papua New Guinea Australia

PNG LNG / EOM JV

2

6.6

2013

APCI

Queensland

2

8.5

2013

ConocoPhillips

Algeria

Gassi Touil LNG

1

4.7

2013

APCI

Australia

Wheatstone LNG

1

5.0

2014

ConocoPhillips

Australia

Ichthys LNG

2

6.0

2015

APCI

USA

Cheniere Sabine Pass 2 LNG

9.0

2016

ConocoPhillips

Global LNG Production Capacity (Existing & Under Construction): 334.9 MTPA (From IGU World Report 2011) CCC Total Capacity: 312.7 MTPA

93.4%

LNG Applications • Refrigeration Compressors – – – –

Propane Ethylene Methane MR

• Boil-off Compressors • Auxiliary Compressors: – – – – – –

Feed Gas Compressor Expander Re-compressor Propane BOG Compressor Fractionation Compressor End Flash Gas Fuel Gas

LNG Economics 101 •

US Natural Gas Supply Price:

$ 3.49 per mmBTU

•

Delivered Price (Japan):

$16.00 per mmBTU

•

Spread:

$12.51 per mmBTU Revenue of 4 MTPA LNG Facility:

$6,753,055 per DAY!!

APCI MCR® Process

Phillips Optimized Cascade

Prico Process/ Single Cycle-MR

Linde N2 Refrigeration Process

Air Products N2 Refrig Process

Air Products / Modec LiBroTM Pre-Cooled N2 Process

Air Products Dual MR Process

SBM’s Proposed Tanker Modification for FLNG

APCI MCR® Process

• •

Uses propane for pre-cooling and a mixed refrigerant (nitrogen, methane, ethane, propane) for liquefaction and sub-cooling Pre-cooling is done in kettle-type exchangers while liquefaction and subcooling are done in proprietary spiral wound heat exchanger i.e. the main cryogenic heat exchanger • 61 trains in operation + 5 in construction

APCI AP-X LNG Process

Propane & Split MR Compressors Propane

MR

3

MR

MR

1

2

Propane Compressor – Primary Objectives: • Sidestream flows pose challenge to antisurge control design • Flow calculation is critical • Antisurge to antisurge decoupling Propane Compressor - Secondary Objective: • Suction pressure limiting using the antisurge valve Propane - Other Comments: • Suction conditions change continuously • GT is maintained at a constant speed • Compressor performance usually adjusted by recycle only

Propane Compressor Capacity Control Challenges

•

• •

Main control variable is suction pressure at 1st stage drum • Pressures at intermediate propane drums tied to 1st stage drum 1st stage suction must be maintained above atmospheric pressure at all times Ineffective suction pressure control results in lower refrigeration which must then be compensated by MR cycle • Control loop interactions sacrifice production

Propane Compressor Antisurge Control Challenges

• Suction conditions change continuously • Sidestream flows pose challenge to antisurge control design • Compressor design prevents direct measurement of flow in intermediate stages − Special algorithms have been developed to calculate this flow • All stages significantly influence preceding stages − Loop decoupling is critical to overall stability and performance

Propane & Split MR Compressors Propane

MR

3

MR

MR

1

2

MR Compressor – Primary Objectives: • Antisurge to Antisurge decoupling • Use of the surge control surface for IGV control (variable gas composition & temperature) • Communication between trains on shutdown or trip MR Other Comments: • GT is typically maintained at a constant speed • The cooling load can be varied by the IGV’s or the JT valves across the MCHE

Feed Gas Compressor VSDS

Primary Objectives: • Surge control under all operating scenarios • Separate antisurge application for dedicated surge detection • Suction pressure control Secondary Objective: • High discharge pressure limiting Comments: • Hp and Flow are very large • Suction Temperature and MW can vary greatly on some applications • 20-30% load changes are common!

Boil-Off Gas (BOG) Compressors To Flare

BOG Header

Fuel Gas

LNG Tanker

• BOG Compressor: Primary Control Objectives: − Antisurge Control & Loadsharing Control • BOG Compressor: Secondary Objectives − P.O.C. On Suction Pressure • Comments: − A very tight control margin − 1.02 psig (28.25” H2O): too low (air could leak in / safety hazard) − 1.06 psig (83” H2O): too high (flare trigger)

ConocoPhillips Optimized Cascade

•

Uses three pure refrigerants (propane, ethylene, methane) for cooling and liquefaction • Pre-cooling sometimes carried out in core-in-kettle type exchanger • Plate fin heat exchangers (non-propriety) in vertical cold boxes used

LNG Plant Operating Modes 1) START-UP: Cold Start: • MCHE must be cooled to desired temperature at specified rate • Long start-up duration Warm Start: • Machines operate at minimum governor speed and full recycle • MCHE relatively cold • Shorter startup duration 2) PART-LOAD: • Plant operate at reduced production • Due to economic factors, tanker delays, upstream gas plant trips, machinery or process related problems • Compressor capacities reduced to match reduced refrigeration load • Compressors may operate with recycle valves partially open • Process may be unstable due to control loop interactions

LNG Plant Operating Modes 3) FULL-LOAD: • Plant operated at greater than design capacity • Maximize production by running machines at maximum power • Product temperatures within tight margins by adjusting refrigeration-cycle duty • Desired flow rate also maintained

4) Defrost Operation • Defrost operation happens 2 to 3 times a year for about 1 to 2 days • Transitions from defrost to C3 gas (unit does not shutdown)

5) SHUTDOWN: • Unloaded in controlled fashion during normal shutdown • Emergency conditions require safe shutdown of machines

Major Challenges in LNG System Design

Avoidance of Cascading Trips on Interdependent Turbomachinery

Introduction  The two main refrigeration compressor strings at Tangguh LNG are highly dependent on each other during operation  A cascading trip can happen within a few seconds

 This case study focuses on how to keep either string online when the other trips  Avoid surging the compressor  Avoid excessive recycle that can overload the drivers

Overview of Main LNG Refrigeration Compressors  Propane circuit cools the MR circuit and Feed Gas  4 stage compressor with sidestreams  Driven by Frame 7 GT with ST helper

 MR circuit cools Natural Gas in MCHE to produce LNG  3 stage compressor with MR HP stage on PR drive train  Driven by Frame 7 GT with ST helper

LNG Main Refrigeration Compressors MR COMPRESSOR LNG

MP

LP

ASV

ASV

NG

PROPANE COMPRESSOR LLP

LP

MP

HP

HP

ASV

Compressor Interdependency  Trip of the MR circuit  Loss of MR flow to the propane chillers will lead to the PR flow (vapor production) gradually decreasing in a relatively short time and eventually dropping off  Sudden loss of flow through the MR HP compressor due to MR MP discharge check valve closing

 Trip of the PR circuit  A trip of MR HP ASV results in sudden loss of flow through the MR LP/MP stages due to closure of MP discharge check valve

Overview of Initial Design  The original control system design was based on lessons learned from a similar LNG plant design  FFC by unloading the online compressor when the other compressor trips  Temporarily initiate the antisurge controllers’ Stop sequence to ramp open the ASVs  Duration based on the Stop ramp rate and desired ASV target opening position  Additional IGV or speed control adjustments were not necessary

Initial Configuration Settings

Mixed Refrigerant Compressor LP Stage MP Stage HP Stage Propane Unit Trip

Ramp 15%/s for 3sec

Ramp 15%/s for 3sec

Propane Compressor LLP Stage

LP Stage MP Stage HP Stage

Trip, valve Trip, valve Trip, valve Trip, valve Trip, valve steps open steps open steps open steps open steps open to 100% to 100% to 100% to 100% to 100%

Mixed Trip, valve Trip, valve Ramp Ramp Refrigerant steps open steps open 8%/s for 5s 5%/s for Unit trip to 100% to 100% 10s

Ramp 5%/s for 10s

Ramp 5%/s for 10s

Ramp 5%/s for 10s

Review and Analysis of Events  MR HP ASV was recorded going 100% open after FFC  Cause: Controller’s open loop line crossed causing it to step open output to 100% and switch to Shutdown

 PR string tripped on underspeed 11 seconds after FFC  Cause: PR HP ASV was manually opened at 55% at the time of FFC signal resulting in the ASV going to 100% open and GT high power limit being reached

 MR string trips 7 seconds after FFC signal  Cause: MP stage surge trip

 Ramp rates in the MR ASC need to be increased  ASV target positions need to be adjusted

Output

Results of Analysis

Out_final

 Ramp ASV to a fixed target position and not a fixed amount  ASC needs to remain active during FFC

Stop Ramp k(%/s)

Out_inital Duration t(s) FFC initiated

Time FFC release d

Limitations of Initial Design  Standard features of ASC Stop mode  Maximum Stop ramp rate is 16.7%/s  When the operating point crosses the controller’s open loop line, the controller immediately steps open the ASV and goes into Shutdown state  The

Rc

SLL = Surge Limit Line OLL= Open Loop Line SCL = Surge Control Line

OP

2

Q

antisurge controller’s Surge Counter/Trip functions are not active during Stop/Shutdown state

Actions Taken

 Separate Unload signal  Configurable ramp rate to 99.9%/s (LVL6)  Configurable ramp target (LVL7)

Output

 Propose ASC Software Modification

LVL 7

 Configurable hold timer (LVL8)

LVL6

 Allow ASC to override Unload sequence  Output goes to 100% if open loop line crossed put remain in Run state

LVL 8 Unload signal

Time

Actions Taken - Verification

 Run dynamic simulation  Verify increased ramp rates and ASV target openings for MR compressor  Simulate both design and off design conditions  Verify GT power stays within acceptable limits

 Site acceptance test  Verify new controller software functionality  Verify logic used to activate the Unload signal

New Configuration Settings Mixed Refrigerant Compressor LP Stage MP Stage HP Stage Propane Unit Trip

Ramp 50%/s to 50% open for 30s

Ramp 60%/s to 60% open for 30s

Propane Compressor LLP Stage

LP Stage MP Stage HP Stage

Trip, valve Trip, valve Trip, valve Trip, valve Trip, valve steps open steps open steps open steps open steps open to 100% to 100% to 100% to 100% to 100%

Mixed Trip, valve Trip, valve Ramp Refrigerant steps open steps open 80%/s to Unit Trip to 100% to 100% 70% open for 30s

Ramp 5%/s to 50% open for 30s

Ramp 5%/s to 50% open for 30s

Ramp 5%/s to 50% open for 30s

Ramp 5%/s to 40% open for 30s

Trend Results from Field MR MP trip initiated

MR LP trip initiated MR Trip

MR Trip

MR HP FFC initiated

70% Open

MR Trip

Trend Results from Field PR LLP FFC initiated

50% Open

MR Trip

PR MP FFC initiated

50% Open

MR Trip

PR LP FFC initiated

70% Open

MR Trip

PR HP FFC initiated

40% Open

MR Trip

Conclusions  No reports of cascading trips since modification  Additional benefits of software modification  Changes allow for a clearer understanding of the control system response after an event  More flexibility in configuration changes 

Ramp rates and target levels can be changed independently

 Settings can be easily changed on line

Temperature (Quench) Control RV 1 Stage I

RV1 Opening was required to prevent Excursion on Stage I

Stage II Flow Stage I SCL

TE1

TE1 From Compressor Discharge

Without Decoupling With Decoupling

TV1

K1 RV1 UIC1

TV1

TIC1

Set Point Limiting to prevent energy waste

k

Antisurge Controller

Const. Temperature Lines

+ From Liquid Refrigerant Storage

Liquid Pressure

Temperature (Quench) Controller

+

Opening of TV1 is required to compensate for Opening of RV1

Liquid and Vapor Mix Enthalpy

Operation of the Quench Controller is allowed only to the right of the Calculated Low SP Clamp

Vapor

Temperature (Quench) Control Quench Control Summary – Quench Temperature set point is f(Psat) + offset – Operator set point interface • Temperature offset from saturation curve • Direct temperature set point entry with saturation curve set point clamp

– High degree of coupling between quench and antisurge • Temperature loop responds slowly • Antisurge reacts quickly • Loop decoupling for optimal response

– Start/Stop sequencing coordinated through antisurge controller

Thanks for your time & attention!

Natural Gas Liquids & Frac. Units

NGL Removal Three methods of condensate removal: • Refrigeration to remove heavy hydrocarbons • Adsorption using chemical agent that has affinity for NGL’s such as lean oils • Cryogenic expansion using turboexpanders

NGL Removal by Cryogenic Expansion

• Designed to recover ethane (C2) and heavier hydrocarbons (C3, C4, etc) from the natural gas stream • The objective is to separate more expensive products and to send methane (C1) into the pipeline. • Expander drops temperature of gas stream causing partial liquefaction of heavier components • Demethanizer separates methane from NGL

Demethanizer Example

Propane Refrigation Compressors

3 4Section Propane Refrig. Compressors Application Challenge • Poor piping lay-out design (Common suction drums) • Common antisurge and quench valves • Need for Automatic startup and SD • Existing piping layout much less than optimum for surge control and protection • Quench Temperature setpoint characterizer

4Section Propane Refrig. Compressors (Plant A)

3 4Section Propane Refrig. Compressors Quench control Setpoint Characterizer

Commissioning Findings / Recommendations • Added excessive surge trip • Added SD feature to Quench controllers when all units are SD to minimize startup time preventing high suction drums level

3 4Section Propane Refrig. Compressors (Plant B)

Propane Refrig. Compressors

Residue Gas Compressors

Turbo-Expander Re-Compressors

Overview of Turboexpanders • Turboexpanders began being used in gas processing plants around 1960 • Presently, most gas processing plants use turboexpanders • A turboexpander recovers useful work from the expansion of the gas • Turboexpander are designed to recover ethane (C2) and heavier hydrocarbons (C3, C4, etc.) from the natural gas streams. The objective is to separate more expensive products and to send methane (C1) into the pipeline. • In the process of producing work, a TX lowers the gas stream temperature. This results in partial liquefication of the gas stream.

Turboexpander Design

• Throughput of the expander part of the train is controlled by Inlet Guide Vanes • Throughput of the recompressor is typically not controlled • Turboexpander trains are equipped with a compressor recycle valve that can be used for surge control and protection • Turboexpander trains are equipped by an expander bypass valve (J-T or Joule-Thompson valve) • Turboexpander trains are either loaded to maximum capacity or are operating at set flow rate

Traditional Turboexpander Control System Design • Speed of the turboexpander typically is not controlled • During upset condition speed exceeds allowable maximum • Overspeed trip prevention is typically primitive • Trip is prevented by one of: – 1) limiting opening of IGV by position of IGV; – 2) limiting dP across the expander; or – 3) limiting speed via IGV and J-T valve in split level fashion

Advanced Turboexpander Control • • • •

Overspeed prevention by “brake control” JT Valve Prepositioning Adequate antisurge control for recompressor Loadsharing Control for parallel T-E trains

“Brake Control” for Overspeed Prevention

If speed exceeds allowable maximum: •

first, open expander compressor’s recycle valve to load up the train

• •

second, at slightly higher set point start closing IGV. J-T valve is used only when IGV is controlling speed or when IGV is 100% open Results in increased condensate production.

JT Valve Pre-Positioning • To reduce severity of tripping on the Feed Gas pressure, it is implemented a special algorithm that “pre-positions” the JT valve • Calculations are done to open the value that provides JT valve capacity equivalent to the Turbo-expander’s throughput prior to its trip • Position of the JT valve is a function of the IGV of the expander • Inlet flow of the expander relates to equivalent JT-valve stroke thus the initial output of the JT Controller • A 10-point characterizer, whose function argument is the IGV position, and the function result is the required equivalent JT valve initial opening value

Multi-Compressor Application

Case Study: Bayu-Undan Offshore Platform

• The platform processes over 1.1 billion ft3/day wet gas • Extraction of over 115,000 bpd condensate, propane, butane and produces over 950 MMSCFD dry natural gas • Phase 1 achieving production in 2004 involved wet gas processing and dry gas reinjection • Phase 2 achieving production in 2006 involved exporting dry gas to Darwin LNG

Process Overview

Flash Gas ITCS •

•

• •

• •

Provide invariant antisurge control for each stage Optimize 1st stage and inter-stage pressure control Equidistant to surge loadsharing Decoupling between antisurge and performance control loops Decoupling between antisurge control loops Limiting loops

Turboexpander ITCS • •

• • • • •

•

Maintain production separator pressure Provide invariant antisurge control for the recompressor Expander overspeed prevention control “Brake control” for overspeed prevention JT Valve Prepositioning Optimized loadsharing strategy Decoupling between antisurge and performance control loops Limiting loops

Reinjection/Export Operating Modes Pd

SLL

SCL

Stage 3 (Export) Operating Point CCL CLL

Stage 1 Operating Point

8212 rpm

190 barg Inlet Vol. Flow

• • • • •

Total 6 operating modes depending on export gas requirements and state of upstream cold process trains With one or more expanders down, gas off spec for export Operating modes clearly indicated need for antichoke control Operating modes translated to 3 defined compressor control modes Switching compressors from one mode to another needed to be bumpless with minimal upset to the process

Reinjection/Export ITCS •

•

• • • • •

Provide invariant antisurge control for each stage Utilize “shared valve” control strategy for antisurge control Maintain suction pressure Provide integrated antichoke control Optimized loadsharing control Decoupling antisurge and performance control loops Limiting loops

Multi-Compressor Integration

Propylene Loading Compressor

Propylene Loading Compressor

FCCU Turbomachinery Control Optimization

FCCU Process Control Challenges

Wet Gas Compressor Control

• • • • •

Maintains pressure in overhead accumulator Output of PIC-1 is setpoint for SIC-1 UIC-1 & UIC-2 protect compressor sections from surge Challenges: Gas composition variations, inherently interactive recycle valves & speed control loop Flare-less startup is desirable

FCCU Wet Gas Compressor South American Refinery

Control Issues • FCCU Wet Gas Compressor operating at constant speed and continuous recycle • Compressor recycle used for suction pressure control • Suction pressure setpoint higher than desired for optimized pressure in the reactor overhead receiver

1st Stage 260 HP

Wprocess = 59.08 T/hr

2nd Stage 300 HP

Wprocess = 53.37 T/hr

Steam Turbine Extraction Map

Wet Gas Comp / Reduced Recycle • Approximate HP requirements for the current average process flow: – Stage 1: 2800 HP – Stage 2: 2800 HP – Total: 5600 HP or 4176 kW

• The estimated power requirements with improved surge control margin & resulting reduction in recycle: – Stage 1: 2540 HP – Stage 2: 2500 HP – Total: 5040 HP or 3758 kW

• Total projected power savings: – HP: 2800 – 2540 = 260 HP – HP: 2800 – 2500 = 300 HP – HP: 260 HP + 300 HP = 560 HP

Wet Gas Comp / Reduced Recycle • Convert: Horse Power to kW, 1kW = 1.34 HP results in: hp = 418 kW • Estimated Energy Savings per Annum – Steam Cost: $5/ton – Using steam curves supplied, assuming constant slope extraction lines

• From extraction map, 418 kW equates to 4T/hour of steam • Therefore, energy savings equal:

4.0 T/hour x $5/ton x 8760 hours/year=

$175,200/year

Production Increase • When blower limited, a reduction in accumulator pressure creates a potential for increasing production • At constant discharge pressure, lowering the suction pressure increases the required compression ratio • Reduced surge control margin results in the ability to meet the higher compression ratio needed • From the 1st stage compressor map, the maximum achievable compression ratio with the existing surge control line is 3.33 • The new surge control line allows the operation at a compression ratio of 3.63 at the same speed of rotation

Production Increase • Ps, min gauge = 1.69 – 1.01325 = 0.68 • Potential set point reduction – 0.89869 – 0.68 = 0.22 kg/cm2

• Lowering the pressure in the fractionator column: – Reduces resistance on the regenerator air blower – Increases mass flow of air

Power Recovery Train (PRT) Configuration Steam Turbine

Main Air Blower

Hot Gas Expander

Regenerator Air Blower Control • • • •

Mass-flow control via adjustable stator blades UIC protects compressor from surge FIC and UIC are de-coupled to avoid interaction during low load conditions and disturbances Challenge: requires responsive & stable process control

Expander Control Flue gas Cooler

Orifice Chamber

3rd stage Separators

Regenerator

Reactor Stripper

Limit control PIC 1

HSS 1

ModeSelector” selector “Soft

Hot Gas Expander

SIC 1

DPIC 1

Expander Control • Control Elements – Expander inlet valve – Expander bypass valve

• High Speed Control Loops – – – – –

Reactor/regenerator differential pressure control Regenerator pressure limiting Speed control Power & speed limiting control (as required) Breaker trip calculations & open-loop response for high speed load-shedding

Generator Breaker-Trip Problem •

20 MW is going to drive the regenerator air blower 7 MW +27 MW

7 MW

Hot Gas Expander Generator Breaker

•

Breaker trip results in loss of speed synchronization and a virtually instantaneous drop in load

•

Conventional systems rely on PID control to control speed

•

PID control is often times too slow to catch disturbance

•

The expander can trip (in a matter of seconds) on: – Overspeed – Other trip settings

Power Swapping Requirement +27MW 27 MW

• • •

Open

7 MW

• +27MW 20 MW

Close

• •

Before breaker opening there was power balance 27 MW is coming from the regenerator With bypass closed, 27 MW is going to the expander & 20 MW to the blower Generator breaker opening causes a load drop of 7 MW After breaker opening 7 MW needs to be shed through bypass valve to achieve control objectives This is achieved by simultaneous: – –

Hot Gas Expander

•

0 MW 7 MW

closing of the inlet valve opening of the bypass valve

Control objectives upon breaker opening are: – –

Keep reactor/regenerator P constant Avoid overspeed trip

CCC PRT Control Solution • p2 T2

p0 Open

In order to perform all functions the following measurements are necessary: – – – – –

T0

•

p1 T1 Close

•

The PRT system should continuously calculate required valve step-changes in anticipation of a breaker trip Upon breaker opening, the control system should: – –

Hot Gas Expander Generator Breaker Status

JT

Breaker status Inlet and bypass valve positions Power export from the generator Rotational speed, P1 , T1 , P2 , T2 Reactor-Regenerator differential pressure

– –

Initiate speed control via inlet valve control Re-direct differential pressure control to bypass valve Initiate open-loop closure of inlet valve Initiate open-loop opening of bypass valve

Calculating the Open-Loop Steps for the Inlet & Bypass Valves • Accurate step changes are critical – Speed synchronization to the electrical grid is lost – Changes in rotational speed of the regenerator air blower will result in an upset in critical air flow to the process • Air flow provides carbon burning & lift to the catalyst bed • Pressure swings can result in catalyst flow reversals and catalyst releases in some instances • PRT and process trip can occur (often part of SSD system design)

• The size of the step change is a function of the: – – – –

Amount of power being exported to the electrical grid Temperature & mass flow of the hot gas to the expander Expander characteristics as defined by the expander map Inlet and bypass valve characteristics

Reducing the Expander Maps with Dimensional Analysis Power vs. Mass Flow

Reduced Power vs. Reduced Flow

Reduced Power

jr 

J p1  ZRT 1

Reduced Flow

qr 

w  ZRT 1 po ,1   p1 p1

Calculating the Open-Loop Steps for the Inlet & Bypass Valves

• •

Monitor expander mass flow, valve positions, & power export Use simplified expander maps to calculate required reduction in expander flow related to current power being exported Calculate  valve Cv for corresponding mass flow to be shed Use valve characteristic curve to determine % of valve movement needed 120 100 80

Current Cv % maximum Cv

• •

60

 cv

New valve position

40

Current valve position

20 0 10

20

30

40

50

60

70

80

90

100

% maximum valve opening

Response from a Conventional System

Breaker Disconnect while Generating 15 MW Overspeed and Trip Speed Export Power

Rx-Rg ∆P Breaker

Breaker opens here

Response with CCC Integrated System

Breaker Disconnect while Generating 15 MW Speed Export Power

Minimal Speed Excursion Minimal Disturbance to the Regenerator

Breaker opens here Rx-Rg ∆P Breaker

Temperature (Quench) Control Location of Quench Control Line 5

2-phase region

gas region

Pressure (MPa)

liquid region

0.5

Quench Control Line

0.05 0

100

200

300

400 Enthalpy kJ/kg

500

600

700

800

Temperature (Quench) Control • Quench Control Summary – Quench Temperature set point is f(Psat) + offset – Operator set point interface • Temperature offset from saturation curve • Direct temperature set point entry with saturation curve set point clamp

– High degree of coupling between quench and antisurge • Temperature loop responds slowly • Antisurge reacts quickly • Loop decoupling for optimal response

– Start/Stop sequencing coordinated through antisurge controller

Process Control Availability and Safety Shutdown Systems

Different Platforms for SIS & Control “Regardless of the  vendors providing the  hardware and software,  how important is it for  your facility to have your  Safety Instrumented  System (SIS) hardware on  a different physical  (hardware) platform  from your Control  System?” (N=75)

While the majority of  respondents say that  having their SIS  hardware on a  different physical  platform from their  Control System, this is  even more  pronounced among  Chemical facilities  (83%) when  compared to  Upstream oil & gas  facilities (57%).

Functional Safety Safety Integrity Level

Definitions

SIS

Definitions SIF Safety Instrumented Function: A SIF is an instrument safety loop that performs a safety function which provides a defined level of protection (SIL) against a specific hazard by automatic means and which brings the process to a safe state.

Control System Availability

System Availability Analysis • Most system availability comparisons have: – Focused on “The Box”, not on the entire system – Used oversimplified models – Used a safety system mindset, focusing only on dangerous failures, and not on the total failure rate of devices and the system – Ignored controller diagnostics, common-cause failures, and other important considerations

• This approach is too simplistic and leads to invalid conclusions

Base Controller Availability • • • • •

Failure Rate: 8 Failures / 106 Hours Controller Self-Diagnostic Coverage: 90% Mean-Time-To-Repair (MTTR): 8 Hours Test Interval: 1 Year Common Cause: 2% of Failures Resulting Controller Availability: Topology

Availability (Percent)

MTTF (Years)

Annual Downtime (Hours)

2-1-0 Duplex 3-2-1-0 Triplex 3-2-0 Triplex

0.9999922551 0.9999924552 0.9999916598

117.9139 121.0418 109.4978

0.0678 0.0661 0.0731

System Boundaries The Controller is Not the Whole System! • Field devices have a huge impact on system availability, and must be considered Controllers Sensors Final Elements

Example Antisurge System • Typical complement of transmitters • I/P transducer • Air-actuated antisurge valve

FT 1

TsT 1

PsT 1

TdT 1

UIC 1

FY 1

PdT 1

Complete System Availability • Field Device Failure Rates* – – – – –

Temperature Transmitters: 31.9 Years Pressure Transmitters: 28.8 Years Flow Transmitter: 16.2 Years I/P Transducer: 15.0 Years Air Actuated Globe Valve: 20.6 Years

Resulting Control System Availability: Topology

Availability (Percent)

MTTF (Years)

Annual Downtime (Hours)

2-1-0 Duplex 3-2-1-0 Triplex 3-2-0 Triplex

0.9997100180 0.9997102181 0.9997094229

3.1484 3.1506 3.1419

2.5402 2.5385 2.5455

• There is no appreciable difference between topologies once field devices are included *Failure Rate Data From ISA TR84.0.02 and Exida

Improving System Availability • Since using a triplex controller does not improve system availability, what does? • Various techniques are used to increase controller and system availability: – – – – –

Improved Diagnostics Redundant Sensors (Transmitters) Fallback Strategies Redundant Output Transducers (I/P) Partial-Stroke Valve Testing

Controller Diagnostics Controller Diagnostic Coverage: 90%, 95%, and 99% Resulting Controller Availability: Availability (Percent)

MTTF (Years)

Annual Downtime (Hours)

95 99

0.9999922551 0.9999954783 0.9999980645

117.9139 201.9661 471.8387

0.0678 0.0396 0.0170

3-2-1-0 Triplex

90 95 99

0.9999924552 0.9999955880 0.9999980938

121.0418 206.9897 479.1004

0.0661 0.0386 0.0167

3-2-0 Triplex

90 95 99

0.9999916598 0.9999951667 0.9999979885

109.4978 188.9487 454.0041

0.0731 0.0423 0.0176

Controller Topology

Diagnostic Coverage (Percent) 90

2-1-0 Duplex

Controller Diagnostics Resulting Control System Availability (including field instruments): Diagnostic Coverage (Percent)

Availability (Percent)

MTTF (Years)

Annual Downtime (Hours)

90

0.9997100180

3.1484

2.5402

95 99

0.9997132403 0.9997158258

3.1838 3.2128

2.5120 2.4894

3-2-1-0 Triplex

90 95 99

0.9997102181 0.9997133500 0.9997158552

3.1506 3.1850 3.2131

2.5385 2.5111 2.4891

3-2-0 Triplex

90 95 99

0.9997094229 0.9997129289 0.9997157498

3.1419 3.1803 3.2119

2.5455 2.5147 2.4900

Controller Topology 2-1-0 Duplex

Adding Fallback Strategies • Statistically over 75% of control loop problems originate from field devices • Algorithms designed to provide continued operation in the event of sensor failures – Software redundancy for sensors – Cost-effective alternative to redundant sensor elements

Resulting Control System Availability: Topology 2-1-0 Duplex 3-2-1-0 Triplex 3-2-0 Triplex

Availability (Percent)

MTTF (Years)

Annual Downtime (Hours)

0.9998306765

5.3926

1.4833

0.9998308766 0.9998300813

5.3989 5.3737

1.4815 1.4885

Automated Fallback Strategies • System monitors transmitter & MPU validity • Multiple fallback strategies should be configurable to handle transmitter failures/problems • Mode switching should be handled in a bumpless fashion

Benefits – Nuisance machine/unit trip avoidance – Latent failure alarms give time to correct – Increased machine & process availability

Partial-Stroke Valve Testing

MTTF in Years

• Position feedback from valve is compared to the controller output, any significant deviation indicates a problem • Frequent testing as compared to demand rate is necessary to achieve maximum availability improvement • Valve should be 5.00 4.75 stroked at least 4.50 15% 4.25 4.00 • Coordination is 3.75 required to 3.50 prevent process 3.25 3.00 upsets while Daily Weekly Monthly Quarterly Annually testing Test Interval Daily Dem and

Weekly

Quarterly

Annually

Monthly

Summary of Data • There is no significant system availability difference between topologies once field devices are included • Control system availability is greatly affected by issues related to field devices System Availability Improvement Technique None (Base Figure, Controllers Only) None (Base Figure, Complete System) Improved Diagnostics (99%) 1:1 Redundant Sensors Parallel Redundant Sensors (Duplex) Fallback Strategies High-Reliability Output Transducers Redundant Output Transducers Automated Final-Element Testing (Daily Test with Annual Demand)

2-1-0 Duplex 3-2-1-0 Triplex 3-2-0 Triplex MTTF (Years) MTTF (Years) MTTF (Years) 117.9139 3.1484 3.2128 6.7262 7.9197 5.3926 3.8324 3.2795 4.9329

121.0418 3.1506 3.2131 6.8518 7.9335 5.3989 3.8356 3.2818 4.9382

109.4978 3.1419 3.2119 6.4075 7.8014 5.3737 3.8228 3.2725 4.9171

API 670 Surge Detection and Control

Governing Standards & Relationships IEC 61511: Process Safety IEC 61508: Functional Safety of Electronic Systems

IEC 62061: Machinery Safety (Machinery directive)

API 670: Machinery Protection System

Surge Detection vs. Antisurge Control

Definitions

When does it take Action?

Surge control is typically defined as a method to prevent a compressor from surge

Before a Surge Cycle

Surge detection is a method that affirms surge has occurred

After a Surge Cycle is initiated

Surge Detection Purpose: – Detect and count surge cycles – Provide output for use in minimizing the number of surge cycles and output to ESD or DCS

Technical Considerations • Surge Detection shall be independent from Antisurge control • ESD shall be independent from Antisurge control • Required on all axial compressors • Vibration sensors and measurements will not be considered for Surge Detection

Surge Detection Methods FLOW

1

PRESSURE

2 TIME (sec.)

3

1

2 TIME (sec.)

TEMPERATURE

3

• Surge Detection components – Sensors – Logic solver – Sequencer • Based on field proven methods – Flow – Pressure – Temperature – Combination of above

1

2 TIME (sec.)

3

Protection Components

Integrated Protection System

Distributed Protection System

Suggestions for Project Specifications

Process Safety Design • HSE Study of 34 Industrial Accidents • Most Common Cause: Specification Errors Design and Implementation 15%

Operation and Maintenance 15%

Installation and Commissioning 6% Changes After Commissioning 21%

Specification 44%

Specification Writing • You may end up with only what you’ve specified, so review, update, and customize • Don’t just focus on the control hardware • Specify : – Overall system performance goals & criteria – System availability with safety goals & criteria – FAT, SAT, and commissioning requirements

• Insist on: – Vendor design responsibility – Vendor experience with similar applications

Sample Specification

Needs Improvement?

Process Control Requirements: Example • Goals for Process Control System: – Raw gas gathering for gas lift & export operations – Pressure control needed for both the LP and IP gas-oil separators – Glycol dehydration unit works well within a range of differential pressure / upstream & downstream should have high & low pressure limits – Precise flow control required to each gas lift injector – Gas lift will have a priority over gas export flow – Operations would like to use separate recycle valves for capacity control (rather then the antisurge valves) – Operation for extended periods of time in choke (stonewall) must be avoided

Review: Key System Design Requirements • Provide for accurate compressor mapping – Is gas composition constant? – What abnormal process conditions are present?

• Don’t sacrifice speed of response or availability – Look at the complete loop / x-mitters, valves, etc. – Is the system operating system deterministic?

• Plan for high speed inter-controller communication – Take advantage of loop decoupling algorithms – Hand-shaking on mode switching, etc. – Coordinated control between systems for loadsharing

Review: Key System Design Requirements cont… • Provide for a coordinated loadsharing scheme – Parallel, series, or compound arrangement – Important for precise process control and efficiency

• Integrate closely coupled process and machinery limiting variables for precise control & stability • Plan for high speed data trending • Maximize overall system availability – Transmitter failure fallback strategies – Redundant control hardware? – Partial valve stroke testing?

Acceptance Test Requirements • Example Test Requirements – Antisurge Control • In response to full closure of a substation suction or discharge block valve, the system must not allow any compressor to surge

– Pressure Control • Suction pressure shall be held within 0.5 % of setpoint under normal process disturbances

– Load-Sharing Control • Upon bringing a compressor on-line or taking one off-line, the control system shall re-establish steady-state operation and stable load-balancing in no more then 5 minutes from start/stop

Acceptance Test Requirements – Turbine Speed Control • In steady state, deviation of the turbine speed from its set point shall not exceed 0.5%

– Turbine Limiting Control • In response to a rise in the speed set point, the system shall not allow an increase in speed after the exhaust-gas temperature has exceeded its limiting control threshold by 0.5% of the sensor span • In response to a rise in the speed set point, the system shall not allow an increase in speed after the air-compressor discharge pressure has exceeded its limiting control threshold by 0.1% of the sensor span

Control System Considerations • Consider “purpose-built” control hardware – – – –

Hardware built specifically for turbomachinery No compromises in design solution Optimized input sampling times Optimized output update times

• Software should be application specific – Look for “deterministic” operating system / guaranteed loop execution rates – Field proven application software for each machinery configuration and process application

• Configurable, not programmable – Continuous control application programs should not be modified, only configured for each installation – Increases security, no unauthorized changes – Minimizes implementation risks – Dramatically improves system supportability

Thanks for your Attendance! Please do not hesitate to contact CCC for any of your turbomachinery control system needs…