Tues 13.30 Misbehaving PID R Hughes

Embed Size (px)

Citation preview

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    1/23

    v

    Compressor Seal Oil Control

    Compressor Seal Oil system A case study

    Richard Hughes

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    2/23

    I have a short case study to present. With hindsight the solution is pretty obvious but Ithink its interesting as a demonstration that simple problem can be hidden by a complexsimulation.

    A PID pressure controller on a seal oil system would periodically oscillate and the causewas not understood. It was not clear if it was the PID loop causing the problem or if itwas reacting to outside disturbances.

    Life is really simple, but we insist on making it complicated

    (Confucius)

    The business schools reward difficult complex behaviour more than simplebehaviour, but simple behaviour is more effective.

    (Warren Buffett)

    Introduction

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    3/23

    Compressor seal oil

    dP

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    4/23

    Flare

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    5/23

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    6/23

    Compressor Seal/Control Oil system

    Seal oil system looks very complicated

    5 high pressure and 5 low pressure oil users

    3 oil pumps ( 2 on line 1 spare ) ,

    32 pipe sections

    Two sets of filters and coolers

    Many issues

    Can not survive transition from two to one pumps [expensive!]

    System would work for a few months then pressure swings would start

    and the relief valves would feather open. Relief valves looked at , pumps looked at

    Usual fix was to work on the control valve and its positioner

    Had been a problem for > 20 years

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    7/23

    History

    Dynamic simulation carried out when plant owned by ICI ( 1983 )

    Some system changes made as a result of this

    Valve characteristics changed

    Some recommendations to change pipe lengths

    Some recent concerns about 250ms scan time on DCS PID controller

    Recent question if we should have positioner or not

    Re-visited in 2009 as many plant trips due to seal oil system

    We will see later the dynamic model is very complicated but fails to model

    the key features of the system.

    Conclusions drawn from a highly accurate model that modelled a lot of un-necessary detail but did not capture what was important

    The simulation gave no insight into the problem

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    8/23

    Oil Capacity 1

    PC

    Oil Capacity 2

    SEAL

    DPC

    210 psi

    Header pressure usually held at 430 psi

    ( Relief valves prevent us running higher )

    Five users but

    limiting one is highpressure process

    gas compressor .

    (Casing at 210psi)

    Trip

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    9/23

    SIMPLE MODEL OF OIL COMPRESSIBILITY

    dtVC

    P ..1

    B

    VoC

    P = pressure (psi)

    V = volume flow into system ( m3/sec)

    C = system capacitance

    V0 = total volume of system (m3)

    B = bulk modulus (psi) =

    D

    EBpipe

    Bpipe

    1

    Boil

    1

    B

    1d = wall thickness

    D = pipe diameter

    E = modulus of elasticity = 29500000 psi (cold rolled steel )

    000,701

    Cpsi/m3

    1 m3 oil with B = 72,500 psi

    50m of 6 pipe with 10mm wall.

    Rough Guesses but not critical !

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    10/23

    SIMPLE VALVE MODELS

    sg

    dPVKCVF ...

    F = flow (m3/hr)

    dP = pressure drop across valve (psi)

    CV = valve CV

    K = 0.227 , conversion for CV in US gallons per minute

    Sg = 0.83 = specific gravity

    V = fractional valve opening ( 0 to 1 )

    Kickback valve CV is 34.2 ; other valve modelled as sum of 5 valve CVs .

    Seal resistance modelled as valve and CV chosen to match plant oil flow

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    11/23

    Validation looks good

    plantdata

    Simulation

    Recovery timefrom pump trip

    120 secs 140 secs

    Minimumpressurereached

    160 psi 160 psi

    Valve

    movement 33 % 38 %

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    12/23

    But is it

    430 PSI

    250 PSI

    160 PSI

    Remo e inte polation

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    13/23

    Remove interpolation

    430 PSI

    250 PSI

    160 PSI

    Pressure falls in 1 scan period of historian

    Tells us nothing about the response time of the system except its < 10secs

    Low pressure ( and valve travel ) function of halving flow

    And valve CVs only

    Recovery time

    Function of controller tuning

    But is it

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    14/23

    In the simulation the PID tuning can be made much faster

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    15/23

    So what have we failed to model

    The header pressure PID loop parameters tell us a lot

    Gain Kp = 0.04 % per %, Integral time Ti = 0.03 min

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    16/23

    So what have we failed to model

    The header pressure PID loop parameters tell us a lot

    Gain Kp = 0.04 % per %, Integral time Ti = 0.03 min

    If the actual plant tuning numbers matched ZN open loop methodthen

    Ti = 3.3.Td = 0.03 mins which tells us our dead time isprobably of the order of 0.6s

    Kp = 0.9/K * ( T/Td ) = 0.04 % / %

    So low proportional gain tells us either process gain K is veryhigh or Td >> T

    Back of an envelope shows the process gain is of the order of 2

    Which implies a process time constant of 0.1 sec and a deadtime of about 0.6 s

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    17/23

    So what have we failed to model

    The header pressure PID loop parameters tell us a lot

    Gain Kp = 0.04 % per %, Integral time Ti = 0.03 min

    In fact for dominant dead time processes they are usually tuned withmuch more integral and lower gain.

    Typically Ti = Td/3 Kp = 0.18 / K

    This would imply dead time of about 5 seconds and time constantof about 0.1 seconds.

    This is pretty reasonable the hydraulic oil is virtually

    incompressible and we can easily get to a few seconds or so of deadtime

    Adding two seconds of dead time to the simulation was enough toprevent us raising the controller gain beyond that seen on the plant

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    18/23

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    19/23

    Simulation

    Added a ramp rate limit to the simulation and a couple ofseconds of dead time

    Tuning values from the plant now limiting values forsimulation

    Some comments in the original simulation report

    Transients are not accurately simulated but the behaviourof the system can be qualitatively predicted

    It is not proposed to give controller settings here pastresults would suggest they do not pass well from model toplant with a great deal of success

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    20/23

    0

    10

    20

    30

    40

    50

    60

    70

    80

    90

    100

    0 1 2 3 4 5 6 7

    dead time / time constant

    %R

    eductioninp

    eakbycontroller

    Quick simulation

    To avoid dropping below the trip setting we would need to reduce

    Dead time/time constant to about 1 i.e. dead time about 1/10 sec

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    21/23

    Simulation is below the trip setting for 70 seconds. To reduce this to the trip

    delay period of 10secs requires us to reduce our dead time/time constant to

    About 2 .

    Quick simulation

    0

    10

    20

    30

    40

    50

    60

    70

    80

    90

    100

    0 2 4 6 8 10 12 14 16 18 20

    dead time / time constant

    timebelow

    tripsetting

    (%o

    fratio=20)

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    22/23

  • 8/10/2019 Tues 13.30 Misbehaving PID R Hughes

    23/23