EPA Challenges for Thermal

Energy
Bagasse Fired Power Stations
Systems
Dr Mike Inkson, Ben Misplon and Etienne de Villiers

Abstract
Traditionally, the sugar industry has been treated leniently with respect to emissions
standards but, as it focuses more on electricity as a profitable by-product, that is changing
even though bagasse is a renewable fuel. Nowhere is that more the case than at US
Sugars Clewiston factory in Florida: even though it is not a major exporter of electricity,
when it installed a new boiler, the factory was obliged to conform with stringent EPA
standards. The challenges imposed by the standards are discussed together with the
engineering solutions and the results obtained. The outline specification was for a 500 000
lbs/h [~226 800 kg/h] boiler delivering 600 psig, 750F [~4137 kPa, ~399C] steam when
firing bagasse. The main challenges were seen in obtaining the 0.026 lbs/MMBTU [~24
mg/Nm] PM10 limit and the original 0.12 lbs/MMBTU [~70 ppmvd] NOX limit without
exceeding the 0.38 lbs/MMBTU [~363 ppmvd] CO limit or the 20 ppmvd ammonia slip limit.
Continuous monitoring was required.

Engineering was supported by fluid dynamic studies, in particular with respect to the NOx and
CO profiles in the furnace: i) maximum NOx reduction was required from the urea injection
system so location was critical; and ii) engineering down combustion NOx increases CO. In
the event, the unit comfortably passed all tests. The variable OFA nozzles which had been
installed proved particularly useful in tuning the boiler and the low uncontrolled NOx levels
[meaning there will be reduced urea charges] were pleasing to see. Whilst these stringent
requirements are unlikely to be applied to other bagasse boilers in the short term, the
lessons learned will make it possible to rise to those challenges when they arise.

Introduction
The cane sugar industry has developed export cogeneration gradually over several decades,
during which time many of the projects have been treated, from the emissions point of view,
as no different to any other sugar industry project. That has frequently meant a very
generous treatment compared to the requirements if, for instance, power utility standards
had been applied.

There have been exceptions over the years of course, most notably the stations on the
French islands [Bois Rouge, Le Gol and Le Moule] and the Florida Crystals station at
Okeelanta in Florida.

However, perhaps as an indication of what is to come, when US Sugar needed a new boiler
[Boiler #8] for its factory at Clewiston in Florida, the state authorities tried to apply
environmental requirements more strict than those at Okeelanta even though the Clewiston
site is only an ad hoc electrical exporter, primarily a sugar producer.

US Sugar has a strong environmental policy so readily agreed to the approach taken but the
requirements did pose a challenge to the engineers tasked with the boiler design.

Clewiston Boiler #8
The USSC requirement was for a 500 000 lbs/h [~226 800 kg/h] boiler delivering 600 psig,
750F [~4137 kPa, ~399C] steam when firing bagasse as primary fuel and capable of 60%
MCR output when firing oil, as a start-up and supplemental fuel. The unit was destined to
become the workhorse of the site which typically has a 200 day crop period crushing at 38
000 stcd but also operates a nominal 1500 t/d RSO refinery throughout the year. It therefore
had to operate 11 months a year without difficulty, much of that with a high turn-down ratio
[capacity as low as 40 50% MCR during off-crop].
A conventional single pass, bi-drum design was adopted from the beginning, but a quick
glance at the general arrangement immediately shows some unusual features :

Fig. 1 : Side Elevation of Clewiston Boiler #8

The most obvious are :

a continuous ash discharge [CAD] stoker instead of steam cleaned pinhole grate;
an exceptionally tall furnace for a bagasse fired boiler;
two stage superheater with inter-stage attemperation;
flue gas scrubbing and a five field electrostatic precipitator [ESP];

The superheater arrangement is a result of the need for the high turn-down ratio but the
others, as discussed later, and other less obvious features are all related to the emissions
standards imposed.

The other key aspect was the nature of the bagasse: Clewiston is on far sandier soil than
most of the Florida industry so bagasse ash levels can be very high and during certain
periods can reach 10%.

EPA Requirements
In general, the emissions standards in the USA are expressed in terms of pounds weight of a
particular emission per million BTUs [lbs/MMBTU] of gross heat input, something which
makes it difficult to express them exactly in more conventional terms or to compare them with
the standards in other countries which are typically based on emission concentrations per
normal cubic metre of flue gas referred to some standard condition. Another issue is the
difference in conditions applied around the world, a value which is critical to normalising for
comparison purposes.
Page 2 of 7
The original indications from the authorities were that they would want a NOx limit of
0.12 lbs/MMBTU and that was the basis that engineering started with. However, when the
permit was issued some 8 months later, they had agreed that such a level was unrealistic so
the permit for Clewistons Boiler #8 required the following :

Table 1 : Boiler Flue Gas Permit Levels at Clewiston

Air Pollutant Maximum Units Approximate

Concentratio Alternative Units
Particulate matter [PM10] 0.026 lbs/MMBTU ~24 mg/Nm
Opacity 20 % n/a
Sulfur dioxide 0.06 lbs/MMBTU ~32 ppmvd*
Nitrogen dioxide 0.14 lbs/MMBTU ~81 ppmvd*
Carbon monoxide 0.38 lbs/MMBTU ~363 ppmvd*
Volatile organics [as C2H6] 0.05 lbs/MMBTU ~100 ppmvd*
Ammonia Slip 20 ppmvd* n/a
-5
Lead 3.8 x 10 lbs/MMBTU
-6
Mercury 3.0 x 10 lbs/MMBTU
-4
Metals (sum of 8) 1.0 x 10 lbs/MMBTU
Hydrogen chloride 0.02 lbs/MMBTU
* at 7% O2

The permit enforcement for NOx and CO was to be with a Continuous Emissions Monitoring
System [CEMS] so there was no question of being out of permit as is often the case
elsewhere. The only concessions were that the limits were on a rolling average basis. It was
accepted that no chemical scrubbing would be used and, therefore, that the fuel-related
pollutants [SOx, heavy metals and HCl] would be whatever they would be. There was,
however, still a need to be sure that the permit levels were not exceeded, so bagasse
samples were analysed over an extended period in order to verify this and, coincidentally, to
provide design data for the boiler.

Particulate matter limits are very typical of the environmental standards imposed around the
world but the permit level at Clewiston is remarkably low :
World Bank
Clewiston

Europe

Mexico

Bolivia

ESP Hi g scrubber Conventional Scrubbers

0 50 100 mg/Nm 150 200 250

Fig. 2 : Boiler Flue Gas PM10 Limits

PM10, sometimes referred to as the thoracic limit, refers to particles equal to or smaller than
10 m. The limits shown have all been normalised to 12% CO2 by volume dry.

Page 3 of 7
Similarly, the NOx limit not a typical requirement is remarkably low and lower even than
that at Okeelanta, only a few miles south :

Okeelanta
Clewiston
Original
Final
Urea Injection Natural levels

0 0.1 0.2 #/MMBTU 0.3 0.4 0.5

Fig. 3 : Boiler Flue Gas NOx Limit

The CO limit, on the other hand, was not considered too onerous if taken on its own. The
problem is, however, that trying to drive down NOx by careful furnace design tends to drive
up the CO content of the flue gas.

Engineering
USSC wanted a boiler which was efficient [in order to maximise the bagasse available for off-
crop use], easy to operate and maintain, able to operate at a high turndown ratio and, of
course, be permit compliant under all operating conditions.

Experience at Okeelanta showed that any form of intermittent discharge grate would result in
permit excursions during cleaning. The ash content of the bagasse indicated that this would
be quite frequent at times. It was, therefore, decided from the outset that the unit would be
equipped with a CAD stoker in order to avoid that problem and assist with ease of operation.
That decision also gave USSC the option of burning other fibrous fuels during off-crop after
the surplus bagasse supply was exhausted. The Thermal Energy Systems stoker is
engineered to minimise maintenance: it has larger diameter shafts than comparable units,
very robust catenary tensioned chains and high temperature castings throughout. In order to
deliver the 227 t/h of steam, it would have to be one of the largest stokers in the world.

Having established the grate concept, the focus of the engineering for Boiler #8 moved to the
furnace and the need to control NOx, ammonia slip and CO. Selective Catalytic Reduction
[SCR] of NOx was considered inappropriate and therefore the only practical alternative, non-
catalytic reduction [SNCR], was adopted. The significance of that decision was that SNCR,
which involves injecting ammonia or, more conveniently, urea into the furnace in order to
reduce the NOx content to molecular nitrogen [N2], is only 50% effective at best so untreated
NOx levels had to be no more than 0.28 #/MMBTU, a low target.

One of the other issues with SNCR is that it is most effective within tight temperature ranges,
so it was necessary to predict the temperature profiles in the furnace for the full range of
possible operating conditions and to design the urea injection system to cover that range. In
addition, over-treating with urea leads to ammonia slip: surplus ammonia being released to
the atmosphere in one form or another. When SNCR is adopted, a limit on ammonia slip of
20 ppmvd [at 7% O2] is imposed.

SNCR is a highly specialised technology so the supplier was selected from an early stage
and the furnace engineering developed in conjunction with him. It was clear that
computerised fluid dynamics would be needed to determine the shape of the furnace, the
optimum way to inject the secondary combustion air to approximate perfect mixing within the
furnace volume and the optimum injection points for the urea injection. For the latter two
challenges, the systems had to be capable of working within the full range of possible
operating conditions, whether that be steam generating capacity or fuel characteristics.

Page 4 of 7
The boiler was modelled using the Furnace computerised fluid dynamics [CFD] code
developed in Australia using over million individual cells. In the end, because of the
relationship between NOx and CO, it was decided to use CO concentration in addition to gas
velocities and particle trajectories to analyse the mixing within the furnace.

The output from the modelling was portrayed as a series of elevation and plan sections
showing predicted values for the gas velocity vectors, temperatures, particle trajectories, O2
concentrations and CO concentrations. Figure 4 shows examples of two such portrayals :

Fig. 4 : Example Portrayals from CFD Analysis

The left hand image shows the predicted gas velocities and directions for a particular
configuration. [Dark blue is up to 3 m/s and the green the highest velocity in this instance
is 12 to 15 m/s.] The right hand image shows the predicted gas temperatures for a different
configuration. [Dark blue is up to 100C and the darkest reds are in excess of 1300C.]
Other portrayals showed the same parameters in plan sections and other parameters
included particle tracks, oxygen levels and CO levels.

Once the basic model had been established it was possible to evaluate various
configurations, in particular of the overfire air [OFA]: both the percentage of total air applied
as OFA and the number and arrangement of OFA nozzles. The ultimate objective was to
approximate a perfectly mixed furnace and hence minimum CO production with an
understanding of the temperature profile and hence the locations for urea injection under
various load conditions.

Part of the solution ultimately adopted was to use variable throat OFA nozzles specifically
developed for the project so that the injection profile could be changed during the operation
of the boiler and the urea injection optimised for different steam output conditions.

Having finalised the furnace, it was possible to proceed with the remaining thermal design
and the design of the boiler ancillaries. Allowance had to be made for the additional gas
burden imposed on the system by the injection of the urea solution: the resultant ammonia
[and reaction products] plus the carrier water. The aspect relevant to this paper was the flue
gas treatment in order to achieve the required particulates levels.

Page 5 of 7
It was clear that an electrostatic precipitator [ESP] was going to be an essential part of the
solution required but it would only be part of the total solution. The main issue with ESPs on
fibrous fuel boilers is that the discards, containing activated carbon from semi-combusted [in
this case] bagasse, are highly flammable so the ESP should be operated under positive
pressure to minimise the gas oxygen content and avoid fires. That places the ID fan before
the ESP and hence exposes it to the erosive grit content of the flue gas.

In addition, ESPs are not particularly good at collecting carbon particles : there is a tendency
for them to break free again from the plates and pass up the stack.

USSC, as part of its ongoing environmental impact abatement policy, had recently retro-fitted
one of its existing boilers with a non-saturating scrubber system, the water uptake rate being
carefully controlled to avoid saturating the gas which would result in condensation and
consequently corrosion within the ESP. What was not understood at the time was that the
ash content of the Clewiston fuel would essentially be calcined in the furnace and form a
concrete in such scrubbers.

The system adopted for Boiler #8 was therefore a pair of non-saturating Thermal Energy
Systems cyclonic scrubbers before the ID fan, followed by a five field ESP after it. The use
of five fields was to ensure that permit levels would still be achieved even when rapping one
of the five. The cyclonic scrubbers use a highly turbulent venturi inlet section where the gas
is sprayed, followed by a high-g cyclonic section to separate the water again, together with
the scrubbed solids.

Results
The engineering of this complex boiler started in March 2003 and commissioning started in
late January 2005 when the first oil was fired to warm up and dry out the unit. The formal
testing of the unit took place just before Easter in that year, very much towards the end of
crop and not long after first firing bagasse.

The environmental tests were conducted by specialist consultants on behalf of Florida State
using their own portable instruments which were also used to calibrate the CEMS. The first
values, compared to the permit values, are presented in Table 2.

Table 2 : Measured Values of Pollutants in the Clewiston Boiler Flue Gas

Air Pollutant Maximum Measured Units

Permitted Value
Particulate matter [PM10] 0.026 0.0039 lbs/MMBT
U
Opacity 20 n/a %
Nitrogen dioxide 0.14 0.131 lbs/MMBT
U
Ammonia Slip 20 26.2 ppmvd
Carbon monoxide 0.38 0.354 lbs/MMBT
U
lbsMMBT
Volatile organics 0.06 0.012
U
The measured values were each the average of three runs and the boiler was generating an
average of 500 853 lbs/h of steam at the specified conditions. Opacity, an alternative
measurement to the PM10 was not measured because PM10 was. It can be seen that, even
at this early stage, the boiler was running well : only the ammonia slip was outside of permit
value, something which was put down to incomplete commissioning of the SNCR system. It
was soon brought under control and within permit once the SNCR system had been fully
commissioned and tuned.

Page 6 of 7
The variable OFA nozzles which had been installed proved particularly useful in tuning the
boiler and the low uncontrolled NOx levels [meaning there will be reduced urea charges]
were pleasing to see.

The issue of the formation of a sort of concrete in the scrubbers, as referred to early, was
unexpected and caused blockages in the scrubbers outlets from time to time. In the end, the
factory engineers came up with an ingenious solution by designing and installing macerators
in the bases of the scrubbers. These are operated typically once a shift to break up any
lumps of this concrete which might have broken lose from the walls so that they are flushed
away by the scrubber water.

Conclusions
It is accepted that the stringent requirements of the Florida DEP and Federal EPA are
unlikely to be applied to other sugar industry boilers in the short term not even in other
states of the USA but the important point is that, with the lessons learned at Clewiston, it
will be possible to rise to those challenges when they arise.

Page 7 of 7