This problem can be addressed as above by dilution with a lower viscosity solvent but with the same problems as above. Higher temperatures cannot be used as explained above also. The use of lower injector pressures will increase the problem of high viscosity, thus emphasising the need for system optimisation. This aspect is assigned a substantial part of the budget to allow for purchase of suitable expertise with design and test facilities in order to provide the most cost effective solution of unforeseen problems. As for the char problem, as application size increases, this problem reduces as more bio-oil is required to be injected per cycle.

4.4.3.3 Alkali metals

There is a small but finite amount of alkali metals in bio-oil from the wood ash. Some is in solution as, for example, acetates, while some is bound in with the char. Char removal will, therefore, contribute to reducing the alkali metal problem although there are problems with filtration as described above. It is not clear what will happen to these contaminants in applications. If used in an engine, they may deposit in the engine or be blown out. One possibility is that a suitable lubricant will flush any deposits into the sump where they can be dealt with during routine maintenance or by continuous in-line treatment. Turbines have more exacting requirements but required and achievable tolerances have not been determined.

4.4.3.4 ____ Inhomoaeneity of the bio-oil

Flash pyrolysis oils produced at low vapour residence times may suffer from a sort of phase separation when the lignin has not been fully depolymerised resulting in a sludge which can separate out with some char on prolonged storage. This problem becomes particularly acute in filtration of the oil when the pyrolytic lignin forms a sort of gel or jelly which rapidly and completely blocks the filter medium. It is possible that this problem can be mitigated or largely resolved by careful control separation, by homogenisation of the liquid or by the use of suitable additives.

4.4.3.5 ____ Environmental problems

One of the main justifications for the development of bioenergy applications is the positive contribution made to the environment. It is therefore important that the process operates in an environmentally acceptable way and that the environmental risks are minimised. Surprisingly little attention has been paid to any environmental aspect of biomass conversion either in terms of wastes and their management or disposal (12), or as an overall audit of the environmental implication or consequence of constructing such a system (13).

A major disadvantage of hydrotreating is the significant hydrogen requirement of around 700 l/kg pyrolysis oil (62), (31 g moles/kg; 62 g/kg) for full hydrotreating, while Veba suggest 600 — 1000 l/kg bio-oil (110, 112). These should be compared to the theoretical yields of 380 !/kg for CoMo catalysts and 440 l/kg for NiMo catalysts (97). An excess of typically 100% is thus required for processing due to hydrogenation of components such as light aromatics.

The hydrogen can be generated in a number of ways including recovery and/or regeneration from the spent gases. These options are summarised in Table 6.6 which includes approximate relative costs, and depicted in the flowsheet of Figure 2. These alternatives have not been evaluated, but will significantly affect the economics. A 1000 t/d biomass processing plant will require about 50 t/d hydrogen for complete hydrotreating which, if generated from biomass by gasification and CO shifting would require up to 800 t/d additional biomass at a significant economic and efficiency penalty. A potentially attractive route is to utilise the surplus hydrogen which is available at fuel value in many refineries. Product yields and process costs are discussed below.

6.4.2 Zeolites

Early zeolite cracking research on biomass derived pyrolysis liquid was carried out on a conventional mono-functional ZSM-5 catalyst, although more, recently specialist catalysts have been developed by modification of the structure in an unspecified way (117). The reaction follows the representative equation below with a maximum stoichiometric yield of 42% on liquid, or a maximum energetic yield of about 50% wt.:

Об He O4 —> 4.6CH1.2 + 1.4C02 + 1.2H20

This is a hydrogen limited reaction with the aromatic product being limited by the availability of hydrogen for aromatics and water formation and oxygen is rejected as both CO2 and H2O.

BIOMASS

One conjectural way of overcoming the hydrogen deficit is through modification of the catalyst to include a shift component to generate in-situ hydrogen from product gases through the water-gas shift reaction. This might be carried out in a bi­functional or multi-functional catalyst that can operate in a carbon limited environment rather than a hydrogen limited one. This is depicted in the equations below which would give a maximum stoichiometric yield of 55% weight on liquid:

3.6CO+ З.6Н2О —> З.6Н2 + З.6СО2

Сб H8 O4 + З.6Н2 —> 6 CH1.2 + 4H20

This is now a carbon limited reaction with the aromatic product being limited by carbon availability assuming that there is sufficient CO available in the product gas for production of hydrogen. This second route represents an optimum but has not been attempted.

Thus on this simplistic stoichiometry, there is little difference in yields conceptually in either the upgrading route of hydrotreating or the zeolite cracking one, although aromatics have a higher potential value than naphtha particularly as a chemical feedstock.

Organisations who have been, or are involved, in zeolite processing are listed in Table 6.7.

The processing conditions for zeolite cracking are atmospheric pressure, temperatures from 350°C up to 600°C and hourly space velocities of around 2. Earlier work focused on the lower temperatures, but as catalyst development continued, better results are understood to have been obtained with modified catalysts at higher temperatures with much lower coke formation and longer catalyst life (118). The mechanisms of cracking and reforming are not well understood, but there appears to be a combination of cracking on the catalyst "surface" followed by synthesis of aromatics in the catalyst pores. There are a number of potential advantages over hydrotreating including low pressure (atmospheric) processing; temperatures similar to those preferred for optimum yields of bio-oil; and a close coupled process. These offer significant processing and economic advantages over hydrotreating.

Actual yields of 15% aromatics in a hydrogen limited environment have been achieved, with projections of 23% if the olefinic gas by-products are utilised through alkylation (133). These correspond to 40% and 60% of the theoretical maxima respectively. Most work has been carried out in the vapour phase acting as quickly as possible on the freshly produced primary pyrolysis vapours to take advantage of the special conditions under which they are formed. Some early but limited work has been carried out on liquid phase processing, including Occidental (140), GTRI (119), INRS (122) and Mobil who mixed methanol with the condensed bio-oil before spraying it into a fluid bed (124). Liquefaction oil has also been processed in the liquid phase with aromatics yields of up to 37% when co-processed with water (155).

Related work on specific chemicals and model compounds derived from biomass including sugars and carbohydrates has been carried out by INRS (120), and later extended to pyrolysis oils (121) and liquefaction oils (122). Poor yields of hydrocarbons were obtained with high coke and tar and rapid deactivation of the catalyst. Other work carried out on model compounds includes that of Milne et al. at NREL (125, 126, 128) who have carried out fundamental R&D for about 15 years on both pyrolysis and zeolite cracking (see Table 7), and University of Laval (143).

The most extensively reported work is that by NREL (see Table 7). Milne et al. have carried out screening studies on zeolite catalysts in the NREL molecular beam mass spectrometer which is particularly well suited to this type of work (125, 126, 128, 129). Diebold et al. (131-135) have carried out more applied and larger scale work by close coupling a zeolite cracking unit to the NREL vortex flash pyrolyser. Early work was carried out on a slip stream of pyrolysis vapours from the reactor, while more recently a substantial fixed bed of catalyst was added to process all the pyrolysis product. An assessment of the resultant technology for production of gasoline has recently been completed by the lEA Bioenergy Agreement (159). Optimum catalysts and conditions have yet to be determined, but there is a trade-off between olefins and aromatics and process conditions in which catalyst life and catalyst regeneration are significant factors. Data from various researchers are insufficiently compatible to draw conclusions.

In addition to work on bio-oil or pyrolysis liquids, the process has also been applied to wastes from the wood, pulp and paper industry including tall oil (151 — 153), and biomass derived materials using synthetic clays (160).

An unusual project in the context of this review is the synthesis of zeolite catalysts from rice hulls through thermal processing at the Indian Institute of Technology although specific applications and efficacy of the resultant catalysts are unclear (161,162).

Zeolite cracking seems to be less well developed than hydrotreating and there is a much wider range of possibilities to be explored with mono — and multi-functional catalysts. In this respect the screening methods of NREL are particularly valuable. There is not only less information available from specific work carried out, but there are no directly analogous refinery processes from which data may be extrapolated, as well as an enormous variety of catalysts from which to choose. Conversely, the inherent nature of a one-step integrated process for synthesis of hydrocarbons is very attractive and offers much potential for fuels and chemicals.

This program was initiated in 1979 with the main objective of establishing conditions for maximising liquid yields from biomass, particularly from forest materials. The University of Waterloo can be credited with the foundation of modern fast pyrolysis and their research is probably the most extensively published and publicised in this area. Two fluid bed reactors are used of 100 g/h and 3 kg/h. The results from these units were used to design a 200 kg/h unit for Union Electrica Fenosa in Spain as described below. Temperatures of 400-700*C have been examined with a wide range of feedstocks, pretreatment methods, various inert and reactive gases and some work on in-bed catalysts. The fluid beds were designed in a blow-through mode to entrain the char from the bed while retaining the sand hence there is no need for sand circulation or replacement. The highest yields of total liquid product are around 70 wt% on dry feed and are typically obtained at temperatures around 500‘C.

From cellulose, the maximum liquid yield from the RTP III reactor system approaches 90% by mass while gas yields are about 10% and char yields are negligible. The overall liquid yield is up to 83% by weight on dry feed basis and has typical characteristics as listed in Table 5.9. Hydroxyacetaldehyde (HA) yields from the RTP III reactor have been reported of up to 12% from wood while from other biomass materials, HA yields of up to 30% can be achieved.

The first commercial application of the Ensyn Technology was at Red Arrow Products in Wisconsin, USA which produce chemicals with the residual oil used as a boiler fuel (30). Two plants are now installed there.

Preliminary combustion tests performed have shown that fast pyrolysis bio-oils could be used in place of heavy fuels oils in industrial boiler applications (30, 31) where bio-oil has been shown to have a similar heat release rate and flame length as Number 6 fuel oil. Special combustion chambers would, therefore, not be required and a fossil fuel fired boiler or furnace could be easily converted to use this oil. Carbon monoxide emissions would be comparable to those from fossil fuel oil combustion, particulate emissions would be greater and NOx emissions would be lower. SO2 emissions from bio-oil combustion are less than 2% of those from the combustion of Number 6 oil.

Currently Ensyn is working with a number of companies in Europe to test fire Ensyn oil in an internal combustion engine. Ensyn are also involved in a large project with a gas turbine manufacturer to test fire bio-oil in a gas turbine. However, the oil can contain up to 0.2% by mass of inorganics which is derived from the ash in the feed. This which may present a problem in gas turbine operation and ash removal is being investigated by both hot gas filtration and liquid filtration.

There is an extensive literature relating mostly to the pioneering work on higher temperature fast pyrolysis for gases (e. g. 32, 33, 34, 35) with more recent publications on the commercialisation of the technology and investigations into applications (e. g. 36, 37).

The pyrolysis of the three main components of most biomass types have been reviewed above. It could therefore be expected that the pyrolysis of wood would exhibit similar characteristics to the pyrolysis of its components. From TGA DTA and DSC work, rt has been concluded that the mechanism of wood pyrolysis is a linear combination of these three components (1, 9, 13, 15, 18, 20, 40, 41,42, 43, 44, 45,46). The focus of this section is fast/flash pyrolysis of wood which has been accounted for in three major proposed reaction schemes.

1. Fast Pyrolysis of Biomass Workshop, October 1980

The first comprehensive reaction scheme to account for the fast/flash pyrolysis of wood was proposed by Diebold (47). The reaction scheme was the general consensus of pyrolysis specialists attending a workshop on the Fast Pyrolysis of

Biomass, at the Solar Energy Research Institute (now National Renewable Energy Laboratory, NREL), in October 1980. This is depicted in Figure 2.3.

This was the first model to take into account the influence of heating rate, temperature and pressure with regards to biomass pyrolysis. The focus of the model is the formation of hydrocarbons, СО, C02, H2 and H2O as the interest at the time was in the production of olefins. Diebold Proposed Modified Reaction Scheme

Diebold then proposed a more simple reaction scheme in that the biomass initially decomposed to a viscous primary precursor with an elemental structure similar to wood (48). This is shown in Figure 2.4.

Figure 2.4 Diebold Modified Reaction Scheme 1985 (48)

A key advantage of production of liquids is that fuel production can be de-coupled from power generation. Peak power provision is thus possible with a much smaller pyrolysis plant, or liquids can be readily transported to a central power plant using engines or turbine. There are additional benefits from potentially higher plant availability from the intermediate fuel storage. The economics of power generation suggest that a niche of up to 10 MWe is available for exploitation which is illustrated later.

Bio-oil has been successfully fired in a diesel test engine where it behaves very similar to diesel in terms of engine parameters and emissions (43). A diesel pilot fuel is needed, typically 5% in larger engines, and no significant problems are foreseen in power generation up to 15 MWe per engine. Gas turbine applications are also considered to be feasible although no work has been carried out in the last 10 years nor on fast pyrolysis liquids. However, a new project has recently been implemented to investigate this and this is more likely to be viable for larger scale applications (44).

5.24.1 Summary

The objective of this process (Oil From Sludge, OFS) is the pyrolysis of raw or digested sewage sludge to produce synthetic oil, gas and char (103., 104). Two liquid fractions are produced; a light fraction and a heavy fraction. The process is based on a continuous laboratory scale plant (5 kg/day) for the conversion of sewage sludge to synthetic oil and a solid fuel developed at Tubingen University, Germany for which the Wastewater Technology Centre (WTC) have a license to use the patented technology in combination with the WTC process (104, 105).

Bench scale research work commenced in 1982 (106) leading to the first 1 tonne/day pilot scale plant in 1986 (106). Bench scale results were confirmed and various parameters such as operating temperature, pressure and solid and gas retention times were optimised (106). A second 1 tonne/day pilot plant was built by Campbell Environmental Ltd. (CEL) of Perth, Australia in 1988 (106). Two one tonne/day pilot scale plants have been constructed (Canada and Australia) while a commercial scale plant consisting of three 24 tonne/day process trains is planned for Highland Creek, Toronto, Canada in 1993 (105). Several plans to commercialise the process in Canada have been announced but no plants have yet been built (107).

5.24.2 Description

Sewage sludge consists of a mixture of organic materials and inorganics. The sludge (either raw or digested) needs to be dried from approximately 35% solids to approximately 90 to 95% solids to minimise the production of water which has to be burnt (105). Dried sewage sludge is fed from a live bottomed feed hopper mounted on load cells to the pyrolysis reactor by a variable speed (plug ended to seal the reactor system) screw feeder. If the process is not used continuously, this plug may partially pyrolyse and forming a solid mass which could block the feeding system.

The pilot reactor has an internal diameter of 254mm and is constructed from stainless steel. The reactor operates at a pressure of approximately 2.5 cm water and is heated by hot gases (at 700-800°C) produced by the combustion of propane in an external burner which pass through a jacket surrounding the reactor. Heat is directed to the first third of the reactor by a single inlet pipe and to the second two thirds of the reactor by two inlet pipes. The flow of hot gases to the reactor and hence the heating rate is controlled by baffles. The best oil yields are obtained using a reactor temperature of 450°C and a solids retention time of 20 minutes. The solids retention time in the reactor can be controlled by varying both the sludge feed rate and the reactor operating conditions. It is aimed to quickly heat the feedstock to 450°C in the first third of the reactor and to maintain this temperature along the whole length of the reactor. Waste flue gases from the reactor exit the reactor heating jacket through two flues and are discharged to atmosphere.

Sewage sludge moves through the stainless steel reactor on two conveyors. A screw conveyor moves feed through the first third of the reactor and a series of paddles move the feed through the final two thirds of the reactor. In the first stage of the reactor, the sludge is heated in the absence of oxygen (105). Approximately 50% of the sludge is vaporised in the first stage. In the second stage of the reactor, the pyrolysis gases are contacted with the residual char in either a counter-current or co-current mode to catalyse vapour phase reactions to convert lipids and proteins to hydrocarbons (105, 108).

The char product leaves the end of the reactor through a chute and enters a water cooled plug end screw. The screw conveyor transfers the char into an ash collection skip which is purged with nitrogen to avoid fires.

The vapour phase products enter a condenser which extracts a heavy oil fraction from the product vapours. The condenser is cooled using electrically heated air enabling a small temperature gradient to be maintained permitting control of the heavy oil yield and quality. Vapours leaving the first condenser enter a water scrubber (spray tower) which extracts further oils. The removed oils and water are separated using a disc centrifuge and the water is cooled and recycled back to the spray tower. Following a run, the dirty water is recycled to the water processing plant on site. The non-condensable gases from the system leaving the water scrubber are flared, A flow sheet is given in Figure 5.22.

Flowsheet of 1 tonne/day Oil From Sludge Pilot Plant 5-1 02

The total yield of oil from the pyrolysis of digested sewage sludge is shown in Figure 5.23. The yield of oil from raw sewage sludge is reported to be 27% of the feed input (basis not reported) while the oil yield from digested sludge is reported to be 11.5% of the mass feed input (105). The exit temperature of the product oil from the system is not reported but it is reported to be cool (105).

The pilot scale oil from sludge process produces two oil products; a light oil fraction and an heavy oil fraction. Both oil products are reported to have very low oxygen contents. The heavy oil product is removed from the first condenser following the reactor. This oil is thick and viscous and flows very slowly at 15°C. The water content of the heavy oil is approximately 1% by mass. The light oil product is removed from the disc centrifuge. The light oil has a caramel colour, has a low viscosity and contains approximately 5% water by mass. The heavy oil does not distil while the light oil is entirely volatile. The as-received (including water) higher heating value of the light and heavy oils is between 35 and 40 MJ/kg. The heavy oil product accounts for approximately 80% of the liquid yield while the light oil product accounts for the remainder. Oil char separation is reported to be efficient — char levels in the heavy oil are not measurable.

The heavy oil product has been found to be suitable as an anti-stripping compound in asphalt as the properties which make the oil a poor quality fuel make it a good anti stripping compound. The use of the product oil in asphalt production would also prevent CO2 emissions from the use of the product oils. Pilot scale experiments are now, therefore, being optimised towards specific high value non­fuel products. Canmet are currently investigating possible end uses for the products from the oil from sludge process. Liquid composition data for the heavy and light oil fractions is not currently available. The composition data shown in Table 5.22 shows the composition of a single liquid oil product produced by the oil from sludge system before the two separate oil stages were installed. Some heavy metals are carried over from the sludge to the oil although the majority of the heavy metais remain with the char (108).

Estimated mass and energy balances for a 180 tonne/day pyrolysis plant are shown in Figure 5.23. The process thermal efficiency (ratio of oil chemical energy to the chemical energy of the dry sludge) calculated using the figures shown in Figure 5.23 is 52.2%. Although the yield of oil varies with the type of sludge processed, the effect of reaction temperature on oil yield is the same for all sludges (108). Peak oil yield is obtained at a reaction temperature between 425 and 450°C. Accurate reactor temperature control is, therefore, required (108). Above 450°C, conditions favour the production of product gas. The data indicate that the heating value of oil produced from digested sludge is higher than the heating value of oil produced from raw sludge (105).

5.25 WORTHING INDUSTRIES INC.

A 200 kg/h fluid bed reactor derived from the University of Waterloo work was originally constructed in 1986 by Encon for wood and peat pyrolysis. The unit was constructed as a self contained plant on the back of a trailer for transportation to test sites for demonstration. This is now used for processing old telegraph poles. These have their outer skin of treated wood removed and which is then pyrolysed in a 50 kg/h fluid bed pyrolyser for recovery of chemicals including creosote and PCP as well as bio-oil. Up to 30,000 poles per year are processed. The skimmed

poles are re-treated and recycled (109). No information is available on performance or costs.

Although fast pyrolysis of biomass has achieved commercial status, there are still many aspects of the process which are largely empirical and require further study to improve reliability, performance, product consistency, product characteristics and scale-up. This section summarises these topics.

3.3.1 Reactor configuration

A variety of reactor configurations have been investigated as listed in Table 3.1. Pyrolysis, perhaps more than any other conversion technology, has received considerable creativity and innovation in devising reactor systems that provide the essential ingredients of high heating rates, moderate temperatures and short vapour product residence times for liquids.

Table 3.1 Pyrolysis Reactors, Heating Methods and Heating Rates

NOTES

* Used for solid waste processing, not liquids production 1 Used for gas production

§ Operational mode unclear

There are two important requirements for heat transfer in a pyrolysis reactor:

1 to the reactor heat transfer medium (solid in ablative reactor, gas and solid in fluid and transport bed reactors, gas in entrained flow reactors),

2 to the pyrolysing biomass.

Two main ways of heating biomass particles in a fast pyrolysis system can be considered: gas-solid heat transfer as in an entrained flow reactor where heat is transferred from the hot gas to the pyrolysing biomass particle by primarily convection (for example the Egemin process), and solid-solid heat transfer with mostly conductive heat transfer. Fluid bed pyrolysis utilises the inherently good solids mixing to transfer approximately 90% of the heat to the biomass by solid — solid heat transfer with a probable small contribution from gas-solid convective heat transfer of up to 10%. Circulating fluid bed and transport reactors also rely on both gas-solid convective heat transfer from the fluidising gas and solid-solid heat transfer from the hot fluidising solid although the latter may be less significant than fluid beds due to the lower solids bulk density. Some radiation effects occur in all reactors.

The important feature of ablative heat transfer is that the contact of the biomass and the hot solid abrades the product char off the particle exposing fresh biomass for reaction. This removes particle size limitations in certain ablative reactors (e. g. the NREL vortex reactor). Attrition of the char from the pyrolysing particle can also occur in both fluid and circulating fluid beds, due to contact of the biomass with in­bed solids where solids mixing occurs. In fluid bed reactors however, attrition of the product char is relatively low and it has been observed that the char particles have the original particle shape, but are slightly reduced in size by char layer shrinkage and attrition.

Char removal is an essential requirement for large particles (> 2 mm) to avoid slow pyrolysis reactions from the low thermal conductivity of biomass giving low heating rates through larger particles which leads to increased char formation. Hot char is known to be catalytically active. It cracks organic vapours to secondary char, water and gas both during primary vapour formation and in the reactor gas environment, therefore it’s rapid removal from the hot reactor environment and minimal contact with the pyrolysis vapour products is essential.

Since the thermal conductivity of biomass is very poor (0.1 W/mK along the grain, ca 0.05 W/mK cross grain), reliance on gas-solid heat transfer means that biomass particles have to be very small to fulfil the requirements of rapid heating to achieve high liquid yields. Claimed temperature increases of 10,000’C/s may be achieved at the thin reaction layer but the low thermal conductivity of wood will prevent such temperature gradients throughout the whole particle. As particle size increases, liquid yields reduce as secondary reactions within the particle become increasingly significant (2). Union Fenosa are using particle sizes of smallest dimension 2 mm in their 200 kg/h fluid bed to achieve total liquid yields of around 55% wt. on dry feed with a 15% water content (3). Ensyn claim that particle sizes of up to 5-6 mm

in their reactor will still give total liquid yields of up to 75% wt. on feed with 10% moisture which is equivalent to 83% wt. on dry feed basis. Based on the original vortactor work at the University of Western Ontario, it seems likely that attrition of particles at the base of the Ensyn RTP reactor would be a feature of their system to remove char from the particle surface with some degree of solid-soiid heat transfer. Egemin found with their entrained reactor that particle sizes of 6 mm caused a large proportion to be expelled from the reactor substantially unreacted due to poor heat transfer and no char removal from ablation. This resulted in total liquids yields of less than 40% wt on dry feed (4), while GTRI claimed yields of over 60% total liquids with an up flowing entrained flow reactor using feed sizes up to 6 mm (5).

A consistent method of expressing product yields is required to remove ambiguities in the comparison of product yields. It is recommended that the water in the feed should be discounted in the final pyrolysis products with only the water of pyrolysis being quoted and the product yields expressed on a dry feed basis. As a rule of thumb, the water of pyrolysis is typically 11 wt% of dry feed.

This chapter contains a description of the significant research, demonstration and commercial processes that are based on fast pyrolysis for production of liquids. The list is alphabetical and is summarised in Table 5.1. Dormant or dismantled or abandoned processes are only included if there is useful information available.