You are on page 1of 4

Feature Report

Engineering Practice

FIRED HEATERS:

Best Practices for


the Control of Fuel Gas Fuel-gas Heater High- Low-
Adopt these practices flow
(Mass
preferred)
Charge
flow
Charge
outlet-
temperature controller
burner-
temperature pressure
override
firebox-
oxygen
override Air flow
to ensure new project FI FI TI TC PC AC FI
02 06 07 01 03 05 04
success, and to achieve Maximum
fuel
safe control and reliable, ratio

efficient operation Feedforward


X K

going forward
Low-select
override
<
Allan G. Kern
Consultant

arge petroleum-refining com-

L
FC Minimum-
02 fire logic
plexes often have dozens of
fired heaters, making broad-
based competency in this area FT PT
02 03
essential to overall reliability, safety
and energy efficiency. Despite in- Fuel-gas Burner
dustry’s widespread familiarity supply header
Fuel-gas control valve
with fired heaters, they continue to (high-turndown)
pose significant challenges during Safety
valve(s)
initial control-system design and
unless properly specified, operated
and maintained, fired heaters can
hinder good control and reliable op- Minimum-fire regulator
(self-operated)
eration going forward.
All fuel-gas-fired heaters have
basic features in common, but no two FIGURE 1. This igure provides a starting point for the design of fuel-gas control
for ired heaters. The temperature-control elements and fuel-gas-low cascade are
applications are exactly the same, shown in bold. Other items are optional, depending on the heater type and perfor-
and an industry best practice for mance objectives
fuel-gas control has never emerged.
Even one of the most basic questions gas control-system design for fired high-value starting point, and in
— whether to use fuel-gas-flow con- heaters. This figure encompasses many cases a viable finishing point,
trol or pressure control — remains some areas of judgment and com- for the design of many fired-heater,
open, with little industry guidance promise, and not everyone will fuel-gas control systems.
available. A helpful approach to agree with all aspects of it. In my Essentially all heater fuel-gas
move toward industry standardiza- professional experience, I have control systems have at their core
tion is to identify the most common found that most sites tend to be a heater-outlet temperature con-
and successful practices based on strongly disposed toward the local troller that is cascaded to either a
current technology and experience, practices they are familiar with. But fuel-gas flow controller or a pres-
for use as a design starting point based on a career of sorting through sure controller. This is highlighted
and a reference point. both new project designs and ongo- in Figure 1. All other control compo-
Figure 1 provides a proposed ing operational issues, the findings nents in Figure 1, such as the over-
starting or reference point for fuel- summarized in Figure 1 provide a rides and feedforwards, are optional
60 CHEMICAL ENGINEERING WWW.CHE.COM MAY 2014
control during startup or low-firing
Feed/effluent conditions stems from old-style con-
exchanger trol valves and orifice-type flow-
Charge meters, which are notoriously and
flow
inherently unstable under low-flow
conditions. Modern high-turndown
Reactor valve designs with sturdy and pre-
FI TI cise electronic positioners, com-
06 07
bined with high-turndown flowme-
ters, such as vortex or mass meters,
make flow control as stable as pres-
sure control under low firing condi-
Fuel gas TC
f(x) 01 tions. Traditional globe-valve and
orifice-meter turndown ratios are
Feedforward 3:1, while modern designs achieve
fuel demand 10:1 or even 100:1. And because
FC flow control lends itself to more pre-
02 cise tuning, flow control is arguably
more stable than pressure control.
Pressure controllers are known to
require de-tuning for startup condi-
Product tions, while flow controllers are not
(in the author’s experience).
FIGURE 2. A strategy for hydrotreater temperature control is shown here. Feedfor- A third area of performance differ-
ward elements proactively handle changes in feed low and temperature, especially ence between flow control and pres-
those resulting from process interaction of the reactor and feed/effluent exchangers sure control is precision of tuning.
Not only is flow control easier and
and will vary with each heater ap- but by the relatively slow tempera- more accurate to tune compared to
plication, depending on the heater ture controller, and this normally pressure control, but the tempera-
type and performance objectives. takes several minutes. This can in- ture controller itself can also be
troduce a very large temperature more precisely tuned when using
Pressure versus flow control excursion, making pressure control flow control, rather than pressure
As API Recommended Practice 556 highly problematic in applications control, as the cascade secondary
(API556)* points out, one basic dif- where operation is near critical control loop. It is straightforward
ference between fuel-gas pressure metallurgical or process tempera- to analyze a historical dataset in
control and flow control is the dy- ture limits, or on processes such as a spreadsheet and arrive at a very
namic response that occurs when hydrocracking heaters, where the reliable value for the amount of fuel
adding or removing burners. With temperature spike could trigger a gas flow that is needed to raise a
flow control, when adding a burner, runaway reaction. given heater charge flowrate by a
the pressure will drop, potentially A second area of concern regard- given number of degrees. This pro-
causing a heater trip on low burner ing flow versus pressure control has vides a nearly perfect gain param-
pressure. With pressure control, to do with the stability of control eter for the temperature controller
when adding a burner, flow will in- during low-firing or startup condi- (especially if a mass flowmeter is
crease, potentially causing a high- tions. The conventional thought is used). The same cannot be said for
temperature spike. that pressure control is more stable pressure control, where the amount
But this is only half of the story. under low firing conditions. For ex- of flow for a given burner pressure
With flow control, the flow control- ample, API556 suggests that one depends upon, among other things,
ler will normally restore proper flow option is to use pressure control for the number of burners currently
in a matter of seconds. Thus, assum- startup and then switch to flow con- in service.
ing a low-pressure trip is avoided, trol after reaching operating con- By the same token, flow control
heater stability is minimally dis- ditions. But the quality of modern lends itself to the accurate applica-
turbed. But with pressure control, instrumentation has changed this tion of feedforward, which is effec-
the increase in flow can be substan- situation, so today, in most cases, tive in eliminating control variance
tial, and it is corrected not by the flow control not only performs bet- in the heater-outlet temperature
relatively fast pressure controller, ter at operating conditions, but it caused by changes in the charge
is also more stable under low-firing rate or temperature. Feedforward is
* API Recommended Practice 556, Instrumen- conditions and under load changes especially effective in breaking the
tation, Control, and Protective Systems for Gas and other disturbances. interaction that occurs in many pro-
Fired Heaters, American Petroleum Inst., Sec-
ond Ed., April 2011. Traditional instability of flow cesses, such as hydrotreaters, that
CHEMICAL ENGINEERING WWW.CHE.COM MAY 2014 61
Engineering Practice

250

Percent change in fuel-gas heating value


have feed-effluent heat exchangers.

or oxygen demand relative to methane


In this case, the temperature con- 200
troller is tuned for heater response
in the “minute” domain, but the 150
feed-effluent feedback loop can be
in the “hour” domain, which often 100
results in large sustained tem-
50
perature oscillations after an ini-
tial disturbance, such as an abrupt 0
change in charge rate or to a more H2 CH4 C2 H6 C3H8 C4H10
exothermic hydrotreater feed. Fig- -50
ure 2 depicts a hydrotreater process
with feedforward control action to -100
Fuel composition
eliminate this type of temperature
feedback cycle. Effect of fuel-gas composition on heating value when flow
is metered in volume, std. ft3/h
The main advantage of pressure
Effect of fuel-gas composition on heating value when flow
control over flow control is that, is metered in mass, lb/h
with pressure control, setpoint lim- Effect of fuel-gas composition on oxygen demand when flow
its can be directly configured to help is metered in volume, std. ft3/h
prevent heater trips caused by low- Effect of fuel-gas composition on oxygen demand when flow
or high-pressure conditions in the is metered in mass, lb/h
burner. This is a potentially impor- FIGURE 3. The effect of varying fuel gas composition on heating value (akin to tem-
tant advantage, but overall it prob- perature control) and on oxygen demand (akin to fuel/air limiting) is shown here for
ably does not outweigh the several mass versus volumetric low. The curves are nearly completely lat (not sloped) when
ongoing performance advantages of fuel gas is metered in mass units, although large changes in hydrogen content can
still pose a non-linearity
flow control. A compromise, increas-
ingly popular today as a result of the gas heating value to understand setting can be effective over the full
ongoing emphasis on safety systems, the problem, but such analyzers are operating range of fuel gas condi-
is to use a combination of flow control too slow to be used in actual heater tions. This complication (the lack
with pressure overrides. This is also temperature control, so the diffi- of a single valid setting when fuel
mentioned in API556, but in practice culty has persisted. gas is metered volumetrically) has
this approach has some drawbacks. A good solution based on today’s traditionally limited and usually
In particular, it is inherently hazard- technology is to measure fuel gas completely defeated the use of this
ous to put a low-pressure override flow with a mass flowmeter, such essential safeguard on balanced
(high selector) on fuel gas. Also, this as a Coriolis meter, as mentioned or induced-draft heaters. But with
solution is problematic from a DCS in API556. Modeling fuel gas as mass metering of the fuel gas, a sin-
configuration and operational stand- a mixture of hydrogen, methane, gle “set and forget” ratio setting can
point (see the discussion of “override ethane and propane, Figure 3 il- be employed, resulting in a highly
placement” below). Moreover, a low- lustrates that heating value varies reliable and available safeguard. As
pressure override is unnecessary if a enormously with composition when Figure 3 shows, where sudden high
minimum fire regulator is used (see metered volumetrically, relative to levels of hydrogen are possible, ar-
“minimum fire” below). methane. But the heating value is riving at an optimum universal set-
essentially flat for variations in hy- ting may still pose a concern, but for
Mass versus volumetric flow drocarbons when metered in mass, variations due to hydrocarbon spe-
Traditionally, fuel-gas flow is mea- although there is still a strong non- cies and normal or small amounts
sured with an orifice plate, using linearity due to hydrogen. (Volu- of hydrogen, mass metering over-
volumetric units of standard cubic metrically speaking, for an ideal comes this traditional limitation to
feet per hour (std. ft3/h). This has gas, atoms of each type are equiva- this important safeguard.
also been a traditional limitation, lent, while from a mass standpoint, Readers should note that limiting
because the heating value of fuel each atom’s heating value is roughly the maximum fuel-to-air ratio is the
gas commonly varies by as much proportional to its molecular “important half ” of full-blown air/
as 700–1,500 Btu/std. ft3, due to weight; hydrogen has a dispropor- fuel cross-limiting, which is common
varying amounts of hydrogen and tionately high heating value for its on fired boilers and in the power-
heavier hydrocarbons, such as pro- small weight.) generation industry, but is usually
pane and butane, which often enter The dashed lines in Figure 3 show considered unnecessary and im-
the fuel gas system in bursts, some- the effect on oxygen demand (as op- practical on simpler fired heaters,
times upsetting every heater in posed to heating value). For fuel- and in applications that may have
the refinery. Many refineries have to-air ratio limiting, when fuel gas dozens of such heaters, as opposed
installed analyzers to monitor fuel- is metered in mass, a single ratio to a single power boiler.
62 CHEMICAL ENGINEERING WWW.CHE.COM MAY 2014
Overrides (and their placement) point of creating potential haz- threshold. This replaces the need
Figure 1 shows three of the most ards. What happens too often is for a low-pressure override control
common and high-utility fuel-gas- that overrides are placed below and its attendant problems, while
control overrides. The low-oxygen the base-layer controller, or some at the same time physically limit-
override responds to fuel-gas flood are placed above and others below, ing the amount of flow the regu-
conditions or potentially to other and all are shown in careless ar- lator could potentially deliver in
fuel or air supply faults, depending rangement on the DCS graphic. the event of regulator failure (in
on the type of heater draft, and is For the operator, this can result in addition to the pressure setting,
especially important where other a confusing puzzle that is prone to the regulator orifice size and maxi-
excess air controls are absent or misunderstanding and mis-opera- mum flow are sized similarly to the
may not always be in normal mode. tion, rather than as an intuitively traditional restriction orifice). Op-
(The low-oxygen override should obvious heater-temperature-con- tionally, the minimum-fire regula-
not be mistaken with excess oxy- trol standard practice. Operator tor can be supplied by natural gas,
gen control, as this latter option, interviews during heater-inci- rather than by fuel gas, to assure
when used, is applied to air flow or dent investigations frequently clean and reliable operation This
heater draft controls, rather than reveal feedback about such “con- concern is mentioned in API556. It
the fuel-gas controls.) The high- fusing” control configurations is also worth noting that under on-
burner-pressure override, like low- and graphics. going operation, the regulator will
pressure protection, is finding in- The DCS control configuration remain fully closed, so that rela-
creased use due to modern focus on itself can also create override po- tively expensive natural gas is not
safety systems. tential hazards when multiple se- continuously consumed.
As important as which overrides lectors are introduced into the con- A third purpose of the regula-
are employed in each application, trol design. Correct functionality tor is to facilitate easy implemen-
is where they are placed, in terms becomes dependent on a number tation of minimum-fire trip logic.
of both the control strategy design of DCS configuration parameters Minimum-fire trip logic is initiated
and the DCS graphics presenta- that historically are unsecured — when process heat needs to be re-
tion. For instance, the low-select that is, are not addressed in man- moved, but there is no integrity
override in Figure 1 is purposefully agement-of-change procedures and issue with the heater itself. By trip-
placed above the gas flow control- may be routinely changed by con- ping to minimum fire, rather than
ler. Moreover, there is no compel- trol engineers or even operators. completely tripping the heater,
ling reason to show most overrides, Not only does this present the DCS burners will be maintained and re-
feedforwards, or other control en- console operator with an array of starting is greatly simplified. This
hancements, on the Level 2 (oper- choices for taking direct control of can have substantial economic and
ating) DCS graphic. In this way, the fuel-gas valve, but each choice safety benefits and is a widely un-
the core concept of temperature may function differently (or not at der-utilized technique in industry
cascaded to flow remains intuitive all) depending on the current state to avoid full heater trips. With a
on the DCS graphic and operators of the others. This also creates a minimum-fire regulator in place,
see clearly how to take direct con- hazardous situation. Intuitively ob- tripping to minimum fire is easily
trol of the valve when necessary, vious and operable controls should accomplished by soft-tripping the
without the need for heater-specific be a high-priority criteria in any control valve, which is a built-in
training or the potential for mis- control design, especially some- function in most of today’s DCS
operation. The override setpoints thing as critical as fuel-gas control control systems. n
can be connected to their alarm for fired heaters. Edited by Suzanne Shelley
or alert settings so that operators
are informed whenever an override Preventing burner trips Author
becomes active, and to navigate to In Figure 1, the minimum-fire self- Allan Kern is founder of,
the Level 3 (detail) heater display, operated regulator should not be and principal consultant
with, APC Performance
which would include the full con- mistaken with the past practice LLC (Email: Allan.Kern@
trol strategy representation, simi- of installing a restriction orifice APCperformance.com). He
has 35 years of process con-
lar to Figure 1 or Figure 2. around the control valve to provide trol experience and has au-
The use of a framework such the right amount of stable flow for thored numerous papers on
topics ranging from field
as this — in terms of appropri- initial burner light-off, although the instrumentation, safety sys-
tems and loop tuning, to mul-
ate override placement within the regulator does serve this purpose, tivariable control, inferential
control strategy design and DCS as well as two others, in a simple control and expert systems, always with empha-
sis on practical solutions and effectiveness in a
graphics presentation — is often and reliable manner. plant operation context. Kern is the inventor of
missing, but it is a much-needed The minimum-fire regulator also XMCTM, a model-less multivariable control and
optimization solution technology. He holds pro-
industry best practice. Current protects against trips resulting fessional engineering licenses in control systems
habits across industry tend to be from low burner pressure, by hav- and chemical engineering, is a senior member
of ISA, and is a graduate of the University of
much more random, often to the ing a setting slightly above the trip Wyoming.

CHEMICAL ENGINEERING WWW.CHE.COM MAY 2014 63

You might also like